Summary
Background Precision medicine and drug repurposing are attractive strategies, especially for tumors with worse prognosis. Glioblastoma is a highly malignant brain tumor with limited treatment options and short survival times. We identified novel BRAF (47-438del) and PIK3R1 (G376R) mutations in a glioblastoma patient by RNA-sequencing. Methods The protein expression of BRAF and PIK3R1 as well as the lack of EGFR expression as analyzed by immunohistochemistry corroborated RNA-sequencing data. The expression of additional markers (AKT, SRC, mTOR, NF-κB, Ki-67) emphasized the aggressiveness of the tumor. Then, we screened a chemical library of > 1500 FDA-approved drugs and > 25,000 novel compounds in the ZINC database to find established drugs targeting BRAF47-438del and PIK3R1-G376R mutated proteins. Results Several compounds (including anthracyclines) bound with higher affinities than the control drugs (sorafenib and vemurafenib for BRAF and PI-103 and LY-294,002 for PIK3R1). Subsequent cytotoxicity analyses showed that anthracyclines might be suitable drug candidates. Aclarubicin revealed higher cytotoxicity than both sorafenib and vemurafenib, whereas idarubicin and daunorubicin revealed higher cytotoxicity than LY-294,002. Liposomal formulations of anthracyclines may be suitable to cross the blood brain barrier. Conclusions In conclusion, we identified novel small molecules via a drug repurposing approach that could be effectively used for personalized glioblastoma therapy especially for patients carrying BRAF47-438del and PIK3R1-G376R mutations.
Similar content being viewed by others
Background
Half of the brain tumors represent diffuse gliomas, and the World Health Organization (WHO) provided classification criteria to predict the clinical behavior of these neoplasms [1]. Glioblastoma is classified as grade IV/IV gliomas with specific characteristics, including high cellularity, cellular pleomorphism, nuclear atypia and necrosis [2]. Glioblastoma has a dismal prognosis despite aggressive treatment [2, 3]. Various genetic mutations and aberration have been identified in glioblastoma patients, such as MGMT methylation [3, 4], BRAFG596A [5], BRAFV600E [5,6,7], PIK3R1G376R, PIK3R1D560Y, PIK3R1N564K mutations [8].
Mutations in proteins critical for cancer biology usually lead to uncontrolled cell growth, resistance to conventional chemotherapy and subsequently, therapy failure. In the past decade, the demand for effective novel agents to combat cancer has drawn the attention to specific molecular alterations in cancer cells as targets for therapy [9]. The concept of precision medicine implies the application of targeted drugs coupled with specific diagnostic assays in order to determine whether patients are likely to benefit from targeted therapy. The shift from classical, non-selective, cytotoxic chemotherapy to molecular targeted cancer drugs resulted in increased tumor response and patients’ survival rates [10,11,12].
Therapeutic monoclonal antibodies are considered a successful strategy for targeted cancer therapy. Antibodies have, however, several limitations, including targeting only cellular surface epitopes, high immunogenicity and high production costs [13]. Small molecule inhibitors possess advantages compared to antibodies, as they address both intracellular and surface proteins. A showcase example is the BCR-ABL inhibitor imatinib, which resulted in dramatic improvements in the survival of chronic myeloid leukemia patients. The same applies to small molecule inhibitors directed against targets in solid tumors, e.g. the epidermal growth factor receptor (EGFR) kinase inhibitors gefitinib and erlotinib to treat non-small cell lung cancer and the vascular endothelial growth factor receptor (VEGFR) kinase inhibitor sorafenib against renal cancer [14,15,16]. In the case of EGFR mutations in the kinase domain of the receptor, it happens that erlotinib is no longer therapeutically effective as cancer cells develop resistance to erlotinib. Numerous mutations appear during tumor progression and cause resistance [17, 18]. Therefore, it is essential to find novel inhibitors, which target and inhibit tumor-related mutant proteins.
The sequencing of tumor genomes and transcriptomes is more and more routinely applied in clinical oncology. The concept of precision medicine by mutations identified by sequence analyses may serve as a basis to select treatment options specifically addressing these mutations. Since such mutations would exclusively occur in tumors but not in normal tissues, it is expected that targeted treatments ought to provoke no or only minimal side effects. However, the vast majority of the data acquired cannot be used for therapeutic purposes yet, as we still lack drugs addressing all relevant mutations. Also, tumors frequently consist of heterogeneous subpopulations with mutational profiles different from the main cell population. Resistance to a targeted drug develop if subpopulations without the treatment-specific mutation appear. This may foster the fatal outcome of malignant diseases. Hence, one wishes to have a larger arsenal of drugs at hand to combat otherwise drug-resistant subpopulations of tumors with mutations, for which no targeted drugs are available currently.
One approach to address this latter problem is to develop individualized tumor vaccination to target mutated tumor surface proteins of individual patients. With the advent of advanced vaccination technologies, it is possible to generate individual vaccines to treat each patient according to the tumor’s individual mutational profile [19, 20]. This approach is difficult to achieve for intracellularly located proteins with tumor-specific mutations since antibodies mainly address extracellular rather than intracellular epitopes in living cells (although endocytic antibody internalization may take place). Therefore, therapeutic strategies are required to address intracellular tumor proteins, which constitute a considerable – if not the largest – portion of the tumor proteome.
In an endeavor to device new strategies for precision medicine based on small molecules to attack mutated intracellular proteins, we developed a concept, which integrates transcriptomic data with virtual drug screening of chemical libraries.
A central element in this context may be the concept of drug repurposing. Drug repurposing is a promising strategy and based on the successful usage of already approved drugs, which were initially developed for other indications than cancer [21]. One of the advantages of these drugs is that they already passed clinical safety testing in clinical trials. Considerable investments in terms of money and manpower are being spent to generate clinical evidence of safety and tolerability of a new drug to fulfill the high standards of the drug-approving authorities. Hence, drug repurposing is attractive from economic as well as medical points of view. The identification of thalidomide against severe erythema nodosum leprosum and retinoic acid against acute promyelocytic leukemia are two successful examples of drug repositioning [22, 23]. Plerixafor was initially developed as HIV drug to block viral entry into the cell, but clinical trials failed. However, leukocytosis in peripheral blood CD34 hematopoietic stem cells led to repurposing of plerixafor as stem cell mobilizing drug [24]. Our group has recently reported on drug repurposing of the anti-malarial artesunate for cancer therapy [25], e.g. prostate carcinoma [26] and colorectal cancer [27] as well as for other diseases such as viral infections [28] and schistosomiasis [29]. In addition, we found that the anti-fungal niclosamide exerted cytotoxic activity towards multidrug-resistant leukemia [30].
In the present study, we identified novel mutations in a glioblastoma patient and first screened a chemical library of more than 1500 FDA-approved drugs to find established drugs, which target novel BRAF and PIK3R1 mutations. Furthermore, we screened more than 25,000 novel compounds from the ZINC database. Then, the cytotoxicity of the selected compounds was evaluated. Furthermore, the protein expression of BRAF, PIK3R1 and other cancer biomarker proteins in the brain tumor tissue obtained from the patient was analyzed by immunohistochemistry. Finally, we identified novel small molecules that effectively targeted cells carrying mutant BRAF and PIK3R1, implying their potential for personalized glioblastoma therapy.
Methods
Patient characteristics
The patient characteristics were recently described [31]. In brief, a 65-year old patient had been diagnosed with glioblastoma WHO grade IV on May 26th, 2014. Informed written consent had been given by the patient that the data are allowed to be scientifically used and published (dated: January 26, 2016). The patient was enrolled in the INTRAGO trial [32], which was registered at clinictrials.gov, no. NCT02104882 (registration date: 03/26/2014) (https://clinicaltrials.gov/ct2/show/NCT02104882). INTRAGO was approved by the local ethics committee (Medical Ethics Commission II of the Faculty of Mannheim, University of Heidelberg 2013-548S-MA) and the Federal Office of Radiation Protection (Z 5-2246/2-2013-063). Temozolomide-based radiochemotherapy had been applied according to the current standard of care [32]. Treatment led to stable disease until May 2017, when surgery of the progressive tumor became necessary.
Transcriptome sequencing of tumor and normal tissue was performed at the National Center for Tumor Diseases (NCT, Heidelberg, Germany), which confirmed the histological diagnosis of glioblastoma. In total, 65 nucleotide substitutions, three focal and 17 larger copy number changes, as well as MGMT promoter methylation was found.
Magnetic resonance imaging (MRI) and computer tomography (CT)
Tumor development at the glioblastoma patient within 1 year (before, during and after temozolomide treatment) can be observed from the MRI scans depicted in Fig. 1. Tumor shrinkage occurred after temozolomide treatment.
In silico analyses
A library of FDA-approved drugs (> 1,500 compounds) (http://zinc15.docking.org/substances/subsets/fda/) was screened for established drugs binding with high affinity towards wildtype BRAF, BRAF47-438del and PIK3R1G376R. The PyRx software [33] was used for initial virtual drug screening. The 10 top-ranked compounds with the highest affinities were selected for further analyses. As a second step, the ZINC database library [34] was used (> 25,000 compounds from the “all clean subset with pH 7”) to identify novel non-approved investigational compounds. Again, the 10 top-ranked compounds with the highest affinities towards BRAF47-438del and the 10 top-ranked compounds with the highest affinities towards PIK3R1G376R were selected for further analyses. The selected candidate compounds were further subjected to molecular docking with AutoDock 4 [35]. Whole protein surfaces were taken into account for virtual screening and molecular docking. Three independent docking calculations were conducted, with 25,000,000 energy evaluations and 250 runs by using the Lamarckian Genetic Algorithm. The corresponding lowest binding energies (LBE) were obtained from the docking log files (dlg), and mean values ± SD were calculated. For visualization of docking results, AutoDock Tools and Visual Molecular Dynamics (VMD) were used (Theoretical and Computational Biophysics group at the Beckman Institute, University of Illinois at Urbana-Champaign) (http://www.ks.uiuc.edu/Research/vmd/). Two known BRAF inhibitors (sorafenib [36] and vemurafenib [37]), two known PIK3R1 inhibitors (PI-103 [38] and LY-294,002 [39]) were used as control drugs.
Immunohistochemistry
For immunohistochemical protein detection, we applied the UltraVision polymer detection method (kit from Thermo Fisher Scientific GmbH, Dreieich, Germany). The method was previously described in detail [26]. Briefly, formalin-fixed, and paraffin-embedded tumor tissue was incubated in a humidified chamber for 1 h at room temperature with primary antibodies against BRAF (1:100, polyclonal, Life Technologies GmbH, Darmstadt, Germany), PI3KR1 (1:100, clone U5, Life Technologies GmbH, Darmstadt, Germany), EGFR (1:50, clone H11, Life Technologies GmbH, Darmstadt, Germany), SRC (1:200, clone 1F11, Life Technologies GmbH, Darmstadt, Germany), AKT (1:100, clone 9Q7, Life Technologies GmbH, Darmstadt, Germany), mTOR (1:100, clone F11, Life Technologies Europe BV, Bleiswijk, Netherlands), NF-κB (1:100, polyclonal, Life Technologies GmbH, Darmstadt, Germany), or Ki-67 (ab16667, dilution 1:100, Abcam, Cambridge, UK). Afterwards, Primary Antibody Amplifier Quanto (Thermo Scientific) was applied for 10 min at room temperature, and HRP Polymer Quanto (Thermo Scientific) was applied for another 10 min. Protein expression was visualized by diaminobenzidine (DAB). Quanto chromogen (Thermo Scientific) was mixed with 1 ml DAB Quanto. The tissues were counterstained in hemalaun solution (Merck KGaA, Darmstadt, Germany). Standard histochemical staining was performed with hematoxylin-eosin (HE).
The immunostained slides were scanned by Panoramic Desk (3D Histotech Pannoramic digital slide scanner, Budapest, Ungary) and quantified by panoramic viewer software (NuclearQuant and MembraneQuant, 3D HISTECH) as previously described [26].
NCI cell lines
The panel of cell lines of the Developmental Therapeutics Program (National Cancer Institute, USA) consists of 7 brain tumor cell lines (SF-295, SF-539, SNB-19, SNB-75, SNB-78, U251, XF498) and of cell lines from other tumor origins (leukemia, melanoma as well as carcinoma of the breast, ovary, prostate, lung, colon, and kidney). The origin of the cell lines has been described [40]. Up to 70 cell lines have been tested per drug using a sulforhodamine assay [41]. Some of these drugs identified by our virtual drug screening approach have been tested in this panel of cell lines, and the log10IC50 values have been deposited at the NCI website (https://dtp.cancer.gov/) [42].
Cytotoxicity
Fifty per cent inhibitory concentrations (IC50) values for the selected compounds from virtual screening of FDA approved drugs towards the brain cancer cell line panel of the NCI cell lines were selected from the National Cancer Institute (NCI) database (http://dtp.nci.nih.gov). The cytotoxicity for the NCI cell line panel was assayed by the sulforhodamine B test, as previously described [43].
Results
Genotyping
The copy number variation plot with selected mutations and variations obtained from tumor genome sequencing of a glioblastoma patient is depicted in Fig. 2. Specific mutations have been identified in BRAF (BRAF47-438del, BRAF-TTYH3 fusion), PIK3R1, MAP3K4, MAP3K5, and PTK2. The promoter of MGMT was methylated. Additionally, several genes were completely deleted, i.e. ABCA13, ABCB4, CDK13, EGFR, EIF4H, and HGF).
Immunohistochemistry
In order to confirm the relevance of the results of RNA-sequencing, we performed immunohistochemistry for selected markers, i.e. BRAF, PIK3R1, and EGFR. In addition, signaling molecules in the EGFR downstream signaling cascade have been investigated, e.g. kinases (AKT, SRC) and transcription factors (mTOR, NF-κB). Furthermore, general tumor features have been investigated by hematoxilin-eosin staining to monitor the typical glioblastoma multiforme morphology and by the immunohistochemical detection of Ki-67 to estimate the proliferation rate of the tumor. As shown in Fig. 3, all tumor markers except EGFR revealed strong immunopositivity. The genotyping data revealed EGFR whole gene deletion and this is confirmed by EGFR staining. The protein expression of BRAF and PIK3R1, as well as the lack of EGFR expression, fit well to the data obtained by RNA-sequencing. The expression of the additional markers (AKT, SRC, mTOR, NF-κB, Ki-67) are clues for the aggressiveness of the tumor.
In silico analyses
In order to search for novel treatment options based on individual mutational profiles of glioblastoma genomes, we used the mutations of this patient for further bioinformatical analyses. We focused on the BRAF47-438del and PIK3R1G376R mutations because genetic alterations in these two genes are frequently assumed as driver mutations with relevance for tumor development and progression. The other mutations in the MAP3K4, MSP3K5 and PTK2 genes were not further considered.
The BRAF47-438del is a novel mutation found in this patient for the first time. Therefore, we compared this novel mutation with the most common BRAFV600E mutation. The BRAF-TTYH3 fusion protein could not be used for virtual drug screening, as no crystal structure of this protein was available in the Protein Data Bank (https://www.rcsb.org/). Therefore, a reliable homology model of this fusion protein could not be generated on the basis of the single wildtype BRAF and TTYH3 proteins.
Using PyRx and AutoDock 4.2, virtual screening of the library of FDA-approved drugs revealed drugs with high affinities towards BRAF47-438del and PIK3R1G376R (Table 1). The docking poses of top 10 compounds as well as of vemurafenib and sorafenib (control drugs for BRAF), LY-294,002 and PI-103 (control drugs for PIK3R1) are visualized in Fig. 4. With regard to the PyRx screening results, all 10 compounds revealed a stronger affinity to BRAF47-438del than vemurafenib and sorafenib. Concerning PIK3R1G376R, the top 10 compounds revealed stronger interaction than LY-294,002. All top 10 compounds except tubocurarine, metocurine and idarubicin possess higher affinity than PI-103. Concerning the BRAF47-438del AutoDock results, STK396645, pimozide, folidan, acetophenone, LS-194,959, danazol revealed stronger interaction than sorafenib. STK396645, pimozide, folidan, acetophenone, LS-194,959 also have stronger affinity than vemurafenib. Within the compounds revealed by PIK3R1G376R docking results, all top 10 compounds except albamycin, daunorubicin, tubocurarine had higher affinities than PI-103, whereas all compounds except albamycin, daunorubicin revealed stronger binding than LY-294,002. The established anti-BRAF drug sorafenib bound with slightly less affinity to BRAF47-438del (-9.39 ± 0.09 vs. -9.57 ± 0.22 kcal/mol) than to the corresponding wildtype protein as observed in molecular docking analyses. The established anti-PIK3R1 inhibitors; LY-294,002 and PI-103 revealed stronger binding to PIK3R1G376R mutant protein than wildtype protein (-6.47 ± < 0.01 vs. -5.19 ± 0.02 for LY-294,002 and − 7.22 ± 0.01 vs -5.76 ± 0.06 kcal/mol for PI-103). Accordingly, it was possible to identify drugs that bind stronger to BRAF47-438del than sorafenib and vemurafenib and drugs that stronger bind to PIK3R1G376R than LY-294,002 and PI-103, all of those can specifically target mutant proteins.
Furthermore, 10 investigational non-approved compounds from the ZINC database were identified for BRAF47-438del and PIK3R1G376R mutated proteins (Table 2). Residues forming hydrogen bonds were labeled in bold. Docking poses are depicted in Fig. 5. With regards to the PyRx results, all compounds revealed stronger interaction than the control compounds. BRAF47-438del AutoDock results pointed out that all compounds interact stronger than sorafenib. ZINC09339473, ZINC22798105, ZINC33068262, ZINC00691692, ZINC08667624, ZINC33068261, ZINC09219428, ZINC31840966 interact stronger than both vemurafenib and sorafenib. PIK3R1G376R AutoDock results revealed that all compounds except ZINC02690584 interact stronger than both PI-103 and LY-294,002 (Table 3).
Cytotoxicity
Virtual drug screening revealed that anthracyclines, among other drugs, might target the mutated BRAF and PIK3R1 proteins. Since these drugs are not clinically used for the treatment of glioblastoma multiforme, we explored whether these compounds are able to kill glioblastoma cells in vitro with comparable efficacy than cell lines from other tumor origin. For this reason, we screened the NCI database for test results of the compounds by our virtual drug screening approach. The log10IC50 values of several brain tumor cell lines for the corresponding compounds included into this database are shown in Fig. 6. For comparison, the log10IC50 mean values for each of the test compounds of 46–63 other tumor cell lines have been calculated. The following observations were made: (1) the anthracyclines (daunorubicin, idarubicin, aclacinomycin) were the most cytotoxic compounds in this panel of drugs; (2) Pimazole also showed considerable cytotoxicity against tumor cells, although this is a psychoactive drug used for the treatment of chronic schizophrenic psychoses; (3) The cytotoxicity of the compounds tested were not significantly different between glioblastoma cell lines and all other tumor cell lines, indicating that these compounds might be effective in glioblastoma patients, given that appropriate formulations will be chosen to cross the blood-brain barrier. Aclarubicin revealed cytotoxicity (log10IC50) in the range of -7.2 to -7.6, higher than both sorafenib (in the range of -5.6 to -5.8) and vemurafenib (in the range of -5.3 to -5.6) (Fig. 6a). Idarubicin and daunorubicin (in the range of -7.4 to -7.9) revealed higher cytotoxicity than LY-294,002 (in the range of -4.7 to -5.6) (Fig. 6b).
Discussion
Since many years, MRI/CT imaging is routinely used for monitoring treatment success. However, imaging cannot deliver hints, which salvage therapy could be applied if standard therapy fails. A novel concept in precision medicine is to sequence tumor transcriptomes to identify patient-specific mutations, which can be then addressed by targeted therapeutics [44].
In the present investigation, we explored the possibility to use transcriptomic information for virtual drug screening, in order to indicate novel treatment options for individual patients. For this purpose, we used the mutational profile of a glioblastoma patient as an example to prove the feasibility of this approach. We first focused on screening of FDA-approved drugs, since repurposing of drugs initially approved for other diseases than cancer provided interesting treatment alternatives in many cases. The mutation profile obtained by genotyping of the patient’s tumor has several implications for therapy:
-
1.
The BRAF inhibitors vemurafenib and sorafenib may be less efficient to inhibit mutated BRAF proteins than wildtype BRAF.
-
2.
Promoter-methylation of the MGMT gene indicates that MGMT protein expression may be downregulated. As MGMT represents a resistance factor for temozolomide [45], this drug may be exquisitely efficient to treat this patient. Indeed, the treatment history of the patient demonstrated a very good response to the temozolomide-based radiochemotherapy protocol.
-
3.
The EGFR deletion indicated that EGFR-directed drugs (such as erlotinib, gefitinib and others) might not be effective.
-
4.
Small molecules addressing the other mutations are not available yet, which limits the therapeutic options in this patient.
-
5.
The HGF gene deletion might be relevant for toxic side effects. HGF (hepatocyte growth factor) is known to protect from drug-induced hepatotoxicity [46,47,48]. In fact, this patient experienced hepatotoxicity by radiochemotherapy with temozolomide and compassionate use of artesunate and Chinese herbs [31]. It can be speculated that the observed hepatotoxicity might be explained at least in part by the HGF deletion.
To validate the RNA-sequencing data, we performed immunohistochemical analyses. The strong immunostaining of BRAF and PI3KR1 indicated that the primary antibodies used for immunohistochemistry detect not only wildtype but also mutated forms of these proteins and that both proteins were expressed at high levels in this glioblastoma case. Accordingly, immuno-positive staining of tissue slices by these antibodies is probably not the suitable indicator for the usage of the compounds newly identified, whereas sequencing of tissue is more appropriate for the detection of the new mutations.
Furthermore, the sequencing results showed that the EGFR gene was deleted, which was confirmed by immunohistochemistry since the tumor was EGFR-negative in immunohistochemical staining. Even if EGFR is not expressed in this tumor, the question arises, whether downstream signaling pathways are still operative, which might then not be linked to EGFR but to other signaling molecules. We observed strong expressions of signaling kinases (SRC, AKT) and transcription factors (mTOR, NF-κB). Hence, these signaling molecules may still be susceptible to specific small molecule inhibitors against SRC, AKT, mTOR, or NF-κB) [49,50,51,52].
A limitation of the tumor sequencing approach is that therapeutic antibodies and small molecules are approved only for some of the major tumor-related proteins, but the vast majority of mutations cannot be therapeutically addressed as of yet. A premise to exploit the full potential of precision medicine would be that the therapeutic options keep pace with the diagnostic power and that ideally hundreds of drugs are available to inhibit the numerous tumor-related mutated proteins. An approach to reach this goal may be for instance tumor vaccination based on mutanomic profiling [20]. Comparable approaches are difficult to realize for small molecules, because the clinical approval of novel drugs is laborious and cost-intensive, and it can take years to bring a new drug to the market.
Therefore, it may be feasible to search for drugs, which have been already approved for other diseases and conditions and which also show activity against cancer by inhibiting tumor-related proteins. This approach has been termed drug repurposing and recently led to astonishing treatment successes [53,54,55].
In the present investigation, we attempted to utilize the wealth of information of RNA-sequencing data obtained from tumor genotyping to identify already FDA-approved drugs that bound with high affinity to mutated proteins in a glioblastoma patient. A database of > 1,500 FDA-approved drugs was first screened by PyRx. The results obtained were then refined by molecular docking with AutoDock 4.2. Furthermore, we used the novel BRAF47-438del mutation and the known PIK3R1G376R mutation as models to prove the feasibility of this approach. It is reasonable to search for drugs that stronger bind to BRAF47-438del than sorafenib and vemurafenib and to search for drugs that stronger bind to PIK3R1G376R than LY-294,002 and PI-103 to identify compounds that can specifically target mutant proteins. The BRAF47-438del docking results pointed out that STK396645, pimozide, folidan, acetophenone, LS-194,959, danazol bound stronger than sorafenib. STK396645, pimozide, folidan, acetophenone, LS-194,959 also had stronger affinities than vemurafenib. Considering the PIK3R1G376R results, all compounds except albamycin, daunorubicin, tubocurarine had higher affinities than PI-103, whereas all compounds except albamycin, daunorubicin revealed stronger binding than LY-294,002.
Virtual drug screening unraveled a panel of drugs that bound with high affinity to BRAF47-438del or PI3KR1G376R. These drugs were from diverse chemical classes and had different pharmacological activities. Strong poisons such as tubocurarine appeared in the screening, which cannot be considered for treatment. Remarkably, anthracyclines were also among the top-ranking drugs, which are known for their strong anticancer activity (daunorubicin, idarubicin, aclarubicin, carminomycin). Further drugs identified by virtual drug screening revealed anti-inflammatory (celecoxib), psycho-active (pimozide, acetophenone), muscle relaxant (fazadon, metocurine), diuretic (triamterene) antidiabetic (gliquidone), antibiotic (novobiocin), anti-endometriotic (danazol) or other pharmacological activities. The wide variety of drugs can be taken as a clue for the potential of repurposing drugs with diverse pharmacological modes of action for cancer therapy. The question arises, which of these drugs may be useful to treat the glioblastoma of the patient presented in our study. The decision may depend not only on the known side effects of these drugs but also on whether they are able to cross the blood-brain barrier (BBB). While this is well known for psychoactive drugs, anthracyclines are known substrates of the ATP-binding cassette (ABC) transporter, P-glycoprotein, which is an important constituent of the BBB. Anthracyclines such as daunorubicin, idarubicin and others are usually not used for the treatment of glioblastoma, since sufficient amounts cannot cross the BBB to exert considerable therapeutic anticancer effects in the brain [56]. This problem may, however, be tackled by novel nanotechnologically engineered anthracycline-releasing devices to overcome the BBB [57,58,59]. Since several anthracyclines appeared as top-ranked drugs to bind with high affinity to BRAF47-438del and PI3KR1G376R, it is worth reconsidering them for glioblastoma therapy, if appropriate nanotechnological formulations allow BBB-crossing. This problem may, however, be overcome, if appropriate nanotechnological formulations will be applied. There are daunorubicin liposomes preparations available on the market [60], and anthracycline liposomes have been shown to cross the blood-brain barrier [61,62,63].
Of course, anthracyclines cannot be viewed as mono-specific drugs addressing the two mutations in BRAF and PI3KR1 in an exclusive manner. As classical drugs of natural origin, they can be reconsidered to act in a multi-specific fashion. Anthracyclines are known to intercalate into DNA, to inhibit DNA topoisomerase II, and to generate cytotoxic superoxide- and hydroxyl radicals. Several other on- and off-target effects can be assumed, and binding to mutated BRAF and PI3KR1 proteins may belong to these still unknown modes of action of anthracyclines.
Having the multi-specific nature of many drugs in mind, it is reasonable to investigate their cytotoxic potential in cell lines with diverse mutations. Therefore, we compared the cytotoxicity of the drugs identified by our virtual drug screening approach in the panel of brain tumor cell lines of the Developmental Therapeutics Program of the National Cancer Institute (NCI, USA) and compared their activity with those of cell lines from other tumor origins. These results may deliver additional information, whether these drugs - initially developed for the treatment of other diseases - are suitable for drug repurposing in glioblastoma therapy. The data obtained with the NCI panel of cell lines pointed out aclarubicin, daunorubicin and idarubicin for glioblastoma treatment since they revealed higher cytotoxicity than the known inhibitors (sorafenib, vemurafenib and LY-294,002). Although the standard drug for glioblastoma treatment is temozolomide, anthracyclines also revealed strong cytotoxicity in vitro against brain tumor cell lines.
Pimozide is a psychoactive drug that does cross the BBB. Since this drug also revealed cytotoxicity towards brain tumor cell lines in vitro, its repurposing for clinical glioblastoma treatment may also be considered.
At the time point, when we obtained these results, the glioblastoma was already at a progressive state and the patient decided not to undergo any further drug treatment anymore. Unfortunately, the patient died before we could recommend an individualized treatment with a liposome-based anthracycline. Hence, a final proof of the clinical success of the concept presented in his paper could not be given. Nevertheless, our study represents a preclinical feasibility approach and a method to identify possible treatment candidates in cases that lack therapeutic options, which is a scenario frequently occurring in clinical practice.
We suggest that further investigations using the tumor genomes of patients should be made to predict active drugs and to observe whether treatment of patients with these drugs really leads to clinically significant response rates.
In general, repurposing of already approved drugs may not deliver more drugs for the individual treatment of cancer patients due to the multiplicity of tumor-related mutations. Therefore, we applied a second approach and used virtual drug screening to screen a chemical library of > 25,000 non-approved investigational compounds. We analyzed whether novel compounds can be identified that also bind to the BRAF and PI3KR1 mutations with high affinities. Indeed, most of the selected compounds revealed stronger binding than the known inhibitors. For instance, all of the selected compounds bound stronger to BRAF47-438del mutant protein than sorafenib and selected compounds except ZINC21793973 bound stronger than vemurafenib. The identified substances may be used as a starting point for further drug development to improve glioblastoma therapy in the future.
Despite the attractiveness of the virtual drug screening approach based on tumor sequencing data, there are also some limitations of this approach that have to be discussed:
-
1.
Virtual drug screening of both FDA-approved drug as well as non-approved investigational compounds may result in the identification of a certain rate of false positive candidates [64, 65]. Therefore, results from virtual drug screening should be verified by in vitro or in vivo experiments.
-
2.
The sequencing of tumor genomes and transcriptomes delivers information on both relevant driver as well as irrelevant passenger mutations regarding tumor development and progression. Hence, careful visual inspection of the sequencing data is advisable to make rational decisions, which mutated proteins are most suitable for virtual drug screening to find suitable candidate drugs with the therapeutic potential to significantly and sustainably improve tumor response to treatment and to prolong the survival time of patients.
Conclusions
Systematic genome-dependent repurposing may reveal putative drug candidates for the further development of precision medicine in cancer and other gene-dependent diseases. A combined concept consisting of multiple techniques, i.e. tumor transcriptomics, diverse virtual drug screening methods, chemical libraries for drug repurposing and in vitro testing may help to make beneficial predictions on small molecules to inhibit distinct mutated proteins and to improve cancer therapy for each individual patient. We term the concept presented here “ReSeqtion” (Repurposing of drugs by genome Sequencing and bioinformatic calculation).
Data availability
The data are available upon reasonable request.
Abbreviations
- AKT:
-
AKT serine/threonine kinase 1
- BBB:
-
blood brain barrier
- BRAF:
-
B-Raf proto-oncogene, serine/threonine kinase
- CD34:
-
cluster of differentiation marker 34
- CRISPR/Cas9:
-
Clustered regularly interspaced short palindromic repeats/CRISPR-associated 3
- CT:
-
computer tomography
- EGFR:
-
epidermal growth factor receptor
- FDA:
-
Food and Drug Administration
- HGF:
-
hepatocyte growth factor
- LBE:
-
lowest binding energy
- MAP3K4/5:
-
Mitogen-activated protein kinase kinase kinase 4/5
- MGMT:
-
O-6-methylguanine-DNA methyltransferase
- MRI:
-
magnetic resonance imaging
- NCI:
-
National Cancer Institute
- NF-κB:
-
nuclear factor-kappa B
- PIK3R1:
-
Phosphoinositide-3-kinase regulatory subunit 1
- PTK2:
-
Protein tyrosine kinase 2
- RMSD:
-
root mean square deviation
- SRC:
-
SRC proto-oncogene, non-receptor tyrosine kinase
- TTYH1:
-
Tweety family member 1
- VEGFR:
-
vascular endothelial growth factor receptor
- VMD:
-
Visual Molecular Dynamics
- WHO:
-
World Health Organization
References
Thon N, Kreth S, Kreth FW (2013) Personalized treatment strategies in glioblastoma: MGMT promoter methylation status. Onco Targets Ther 6:1363–1372
Louis DN et al (2007) The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol 114(2):97–109
Rivera AL et al (2010) MGMT promoter methylation is predictive of response to radiotherapy and prognostic in the absence of adjuvant alkylating chemotherapy for glioblastoma. Neuro Oncol 12(2):116–121
Esteller M, Herman JG (2004) Generating mutations but providing chemosensitivity: the role of O6-methylguanine DNA methyltransferase in human cancer. Oncogene 23(1):1–8
Hatae R et al (2017) A comprehensive analysis identifies BRAF hotspot mutations associated with gliomas with peculiar epithelial morphology. Neuropathology 37(3):191–199
Nakata S, Horiguchi K, Ishiuchi S, Yoshimoto Y, Yamada S, Nobusawa S, Ikota H, Hirato J, Yokoo H (2017) A case of high-grade astrocytoma with BRAF and ATRX mutations following a long-standing course over two decades. Neuropathology 37(4):351–357. https://doi.org/10.1111/neup.12370
Behling F et al (2016) Frequency of BRAF V600E mutations in 969 central nervous system neoplasms. Diagn Pathol 11(1):55
Cancer Genome Atlas Research, N (2008) Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455(7216):1061–1068
Schmitt MW, Loeb LA, Salk JJ (2016) The influence of subclonal resistance mutations on targeted cancer therapy. Nat Rev Clin Oncol 13(6):335–347
Gish RG, Finn RS, Marrero JA (2013) Extending survival with the use of targeted therapy in the treatment of hepatocellular carcinoma. Gastroenterol Hepatol (N Y) 9(4 Suppl 2):1–24
Rolfo C et al (2015) Improvement in lung cancer outcomes with targeted therapies: an update for family physicians. J Am Board Fam Med 28(1):124–133
Motzer RJ et al (2009) Overall survival and updated results for sunitinib compared with interferon alfa in patients with metastatic renal cell carcinoma. J Clin Oncol 27(22):3584–3590
Chames P et al (2009) Therapeutic antibodies: successes, limitations and hopes for the future. Br J Pharmacol 157(2):220–233
O’Brien SG et al (2003) Imatinib compared with interferon and low-dose cytarabine for newly diagnosed chronic-phase chronic myeloid leukemia. N Engl J Med 348(11):994–1004
Druker BJ et al (2006) Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia. N Engl J Med 355(23):2408–2417
Yap TA, Workman P (2012) Exploiting the cancer genome: strategies for the discovery and clinical development of targeted molecular therapeutics. Annu Rev Pharmacol Toxicol 52:549–573
DiMasi JA, Grabowski HG (2007) Economics of new oncology drug development. J Clin Oncol 25(2):209–216
Hoelder S, Clarke PA, Workman P (2012) Discovery of small molecule cancer drugs: successes, challenges and opportunities. Mol Oncol 6(2):155–176
Tureci O et al (2016) Targeting the heterogeneity of cancer with individualized neoepitope vaccines. Clin Cancer Res 22(8):1885–1896
Diken M et al (2017) Discovery and subtyping of neo-epitope specific T-cell responses for cancer immunotherapy: Addressing the mutanome. Methods Mol Biol 1499:223–236
Kadioglu O, Efferth T (2014) Contributions from emerging transcriptomics technologies and computational strategies for drug discovery. Invest New Drugs 32(6):1316–1319
Sirota M et al (2011) Discovery and preclinical validation of drug indications using compendia of public gene expression data. Sci Transl Med 3(96):96ra77
Aronson JK (2007) Old drugs–new uses. Br J Clin Pharmacol 64(5):563–565
Devine SM et al (2008) Rapid mobilization of functional donor hematopoietic cells without G-CSF using AMD3100, an antagonist of the CXCR4/SDF-1 interaction. Blood 112(4):990–998
Efferth T (2017) From ancient herb to modern drug: Artemisia annua and artemisinin for cancer therapy. Semin Cancer Biol 46:65–83
Michaelsen FW et al (2015) Activity of Artemisia annua and artemisinin derivatives, in prostate carcinoma. Phytomedicine 22(14):1223–1231
Krishna S et al (2015) A randomised, double blind, placebo-controlled pilot study of oral artesunate therapy for colorectal cancer. EBioMedicine 2(1):82–90
Efferth T et al (2008) The antiviral activities of artemisinin and artesunate. Clin Infect Dis 47(6):804–811
Saeed MEM et al (2016) Antischistosomal activity of artemisinin derivatives in vivo and in patients. Pharmacol Res 110:216–226
Hamdoun S, Jung P, Efferth T (2017) Drug repurposing of the anthelmintic niclosamide to treat multidrug-resistant leukemia. Front Pharmacol 8:110
Efferth T et al (2017) Hepatotoxicity by combination treatment of temozolomide, artesunate and Chinese herbs in a glioblastoma multiforme patient: case report review of the literature. Arch Toxicol 91(4):1833–1846
Giordano FA et al (2014) INTRAGO: intraoperative radiotherapy in glioblastoma multiforme-a phase I/II dose escalation study. BMC Cancer 14:992
Dallakyan S, Olson AJ (2015) Small-molecule library screening by docking with PyRx. Methods Mol Biol 1263:243–250
Irwin JJ et al (2012) ZINC: a free tool to discover chemistry for biology. J Chem Inf Model 52(7):1757–1768
Morris GM et al (2009) AutoDock4 and AutoDockTools4: Automated docking with selective receptor flexibility. J Comput Chem 30(16):2785–2791
Casadei Gardini A et al (2016) Efficacy of sorafenib in BRAF-mutated non-small-cell lung cancer (NSCLC) and no response in synchronous BRAF wild type-hepatocellular carcinoma: a case report. BMC Cancer 16:429
Shelledy L, Roman D (2015) Vemurafenib: First-in-class BRAF-mutated inhibitor for the treatment of unresectable or metastatic melanoma. J Adv Pract Oncol 6(4):361–365
Gao CL et al (2016) [A dual PI3K/mTOR inhibitor, PI-103, cooperates with TRAIL in laryngeal squamous carcinoma cells in vitro]. Zhonghua Yi Xue Za Zhi 96(27):2187–2191
Shi XY et al (2008) [Reversal effect of PI-3K/Akt pathway inhibitor LY294002 on multidrug resistance of ovarian cancer cell line A2780/Taxol]. Ai Zheng 27(4):343–347
Alley MC et al (1988) Feasibility of drug screening with panels of human tumor cell lines using a microculture tetrazolium assay. Cancer Res 48(3):589–601
Monks A et al (1991) Feasibility of a high-flux anticancer drug screen using a diverse panel of cultured human tumor cell lines. J Natl Cancer Inst 83(11):757–766
Amundson SA et al (2008) Integrating global gene expression and radiation survival parameters across the 60 cell lines of the National Cancer Institute Anticancer Drug Screen. Cancer Res 68(2):415–424
Rubinstein LV et al (1990) Comparison of in vitro anticancer-drug-screening data generated with a tetrazolium assay versus a protein assay against a diverse panel of human tumor cell lines. J Natl Cancer Inst 82(13):1113–1118
Horak P et al (2017) Precision oncology based on omics data: The NCT Heidelberg experience. Int J Cancer 141(5):877–886
Cabrini G et al (2015) Regulation of expression of O6-methylguanine-DNA methyltransferase and the treatment of glioblastoma (Review). Int J Oncol 47(2):417–428
Roos F et al (1992) Reduction of alpha-naphthylisothiocyanate-induced hepatotoxicity by recombinant human hepatocyte growth factor. Endocrinology 131(6):2540–2544
Enriquez-Cortina C et al (2013) Hepatocyte growth factor protects against isoniazid/rifampicin-induced oxidative liver damage. Toxicol Sci 135(1):26–36
Bai P et al (2016) A synthetic biology-based device prevents liver injury in mice. J Hepatol 65(1):84–94
Arepalli SK et al (2015) Novel NF-kappaB inhibitors: a patent review (2011–2014). Expert Opin Ther Pat 25(3):319–334
Olgen S (2016) Design strategies, structures and molecular interactions of small molecule Src inhibitors. Anticancer Agents Med Chem 16(8):992–100
Calimeri T, Ferreri AJM (2017) m-TOR inhibitors and their potential role in haematological malignancies. Br J Haematol 177(5):684–702
Ebrahimi S et al (2017) Targeting the Akt/PI3K signaling pathway as a potential therapeutic strategy for the treatment of pancreatic cancer. Curr Med Chem 24(13):1321–1331
Bertolini F, Sukhatme VP, Bouche G (2015) Drug repurposing in oncology–patient and health systems opportunities. Nat Rev Clin Oncol 12(12):732–742
Cheng F et al (2016) Individualized network-based drug repositioning infrastructure for precision oncology in the panomics era. Brief Bioinform 18(4):bbw051
Chaudhari R et al (2017) Computational polypharmacology: a new paradigm for drug discovery. Expert Opin Drug Discov 12(3):279–291
da Ros M et al (2015) The use of anthracyclines for therapy of CNS tumors. Anticancer Agents Med Chem 15(6):721–727
Gulati K, Aw MS, Losic D (2012) Nanoengineered drug-releasing Ti wires as an alternative for local delivery of chemotherapeutics in the brain. Int J Nanomedicine 7:2069–2076
Mo J et al (2016) Tailoring particle size of mesoporous silica nanosystem to antagonize glioblastoma and overcome blood-brain barrier. ACS Appl Mater Interfaces 8(11):6811–6825
Su CH et al (2016) Evaluation of blood-brain barrier-stealth nanocomposites for in situ glioblastoma theranostics applications. Nanoscale 8(15):7866–7870
Weissig V, Pettinger TK, Murdock N (2014) Nanopharmaceuticals (part 1): products on the market. Int J Nanomedicine 9:4357–4373
Gong W et al (2011) Improving efficiency of adriamycin crossing blood brain barrier by combination of thermosensitive liposomes and hyperthermia. Biol Pharm Bull 34(7):1058–1064
Iqbal U, Abulrob A, Stanimirovic DB (2011) Integrated platform for brain imaging and drug delivery across the blood-brain barrier. Methods Mol Biol 686:465–481
Ying X et al (2010) Dual-targeting daunorubicin liposomes improve the therapeutic efficacy of brain glioma in animals. J Control Release 141(2):183–192
Deng XQ et al (2008) Pharmacophore modelling and virtual screening for identification of new Aurora-A kinase inhibitors. Chem Biol Drug Des 71(6):533–539
Sink R et al (2010) False positives in the early stages of drug discovery. Curr Med Chem 17(34):4231–4255
Funding
Open Access funding enabled and organized by Projekt DEAL.
Author information
Authors and Affiliations
Contributions
M.E.M.S., O.K. A.Y. and K.M. performed the bioinformatical in silico analyses. M.E.M.S. performed the immunohistochemical staining. F.W. and F.A.G. were the treating physicians. H.J.G. read and revised the manuscript. T.E. designed the concept and wrote the manuscript.
Corresponding author
Ethics declarations
Ethical approval
The patient was enrolled in the INTRAGO trial [32], which was registered at clinictrials.gov, no. NCT02104882 (registration date: 03/26/2014) (https://clinicaltrials.gov/ct2/show/NCT02104882). INTRAGO was approved by the local ethics committee (Medical Ethics Commission II of the Faculty of Mannheim, University of Heidelberg 2013-548S-MA) and the Federal Office of Radiation Protection (Z 5-2246/2-2013-063).
Informed consent
Informed written consent had been given by the patient that the data are allowed to be scientifically used and published (dated: January 26, 2016).
Conflict of interest
All authors declare there is no conflict of interest. However, patent application #EP18211231 “A method to determine agents for personalized use” is disclosed.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Saeed, M.E.M., Kadioglu, O., Greten, H.J. et al. Drug repurposing using transcriptome sequencing and virtual drug screening in a patient with glioblastoma. Invest New Drugs 39, 670–685 (2021). https://doi.org/10.1007/s10637-020-01037-7
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10637-020-01037-7