Patent Application
20200032347
The present invention relates to the field of medicine and molecular diagnostics. In particular, it relates to a novel RNA profiling assay allowing simultaneous detection of inter alia transcripts and alternative splice variants thereof and mutations therein, from genes involved in disease, including genes involved in metabolism, resulting in a guidance for personalized treatment with drugs targeting disease-associated molecular aberrations, optionally in combination with dietary compounds, food supplements or inhibitors of metabolism.
Malfunctioning cells are typically distinct from healthy cells in that they have altered metabolism. As an example, cells in diabetic patients have adapted to cope with lack of glycogenesis and high extracellular glucose concentrations. In another example, aberrantly growing cells such as in hyperplasia and in cancer need to process excessive amounts of nutrients to produce nucleotides, amino acids and fatty acids for DNA/RNA synthesis, protein synthesis and membrane synthesis. To accommodate this demand, growing cells have adapted by an altered metabolism. There is a number of compounds that can serve as fuel for malfunctioning cells. These include glucose, fatty acids and amino acids, such as glutamine and glutamate. A selection of genes that are involved in cell metabolism are presented in Table I here below. It should be noted that genes involved in a metabolic pathway may also be involved in another metabolic pathway; the person skilled in the art is aware of this.
Altered metabolism may be a result of cancerous transformation of cells, but for a number of cancer types it is also a cause of cancer. A well-known example of metabolic alterations (alterations within one or more metabolic pathway) resulting from cancer growth is the alterations that are a consequence of hypoxia, the lack of oxygen that occurs in growing tissues that have outgrown the vascular blood supply. Under oxygenated conditions, the transcription factors Hypoxia Inducible Factors HIF1α and HIF2α are hydroxylated by the oxygen-dependent enzyme proline hydroxylase (PHD). Proline-hydroxylated HIFs have binding sites for the VHL-E3 ubiquitin complex, resulting in HIF ubiquitinylation and proteasomal breakdown. This pathway is an important regulator of HIF-levels in cells. Under normoxic conditions glucose will be converted to pyruvate that will be processed to acetyl-CoA which enters the mitochondria for processing in the tricarboxylic acid (TCA) cycle. The TCA cycle is directly coupled to oxidative phosphorylation and yields for every mole of glucose the energy equivalent of 36 moles of ATP, CO2 and H2O. Full processing of glucose via this pathway does not yield carbon building blocks.
Under hypoxic conditions PHDs are inactive and as a result unhydroxylated HIF1/2a will accumulate in cells, heterodimerize with HIF-13 (ARNT) and activate genes that are needed to survive hypoxia. These genes (Table I, steps 1-38) regulate different processing of glucose, using pyruvate for lactate instead of acetyl CoA production. Conversion of pyruvate to lactate yields only 2 moles of ATP for every mole of glucose. The inefficiency of this process in terms of energy production requires extra intake of glucose, which is accomplished by increased expression of glucose transporters (GLUT1, GLUT3). Genes involved in the next steps of glucose processing are also activated (especially hexokinase 2). HIF accumulation also results in activation of the gene encoding vascular endothelial growth factor (VEGF-A), resulting in an angiogenic response.
Glycolysis in cancer is not restricted to hypoxic areas but can also occur under normoxic conditions. There is a number of causes for glycolysis in normoxic cancers, for example elevated expression of the Myc oncogene [resulting in activation of PDK (Table I, step 33) and preventing influx of acetyl-CoA into the TCA cycle], decreased function of tumor suppressors (VHL, PTEN) and elevated activity of oncogenic pathways (e.g. PI3K, AKT) all leading to increased HIF activity. Aerobic glycolysis in cancers is known as the Warburg effect.
Whereas aberrations in oncogenes and tumor suppressor genes in a cancer and conditions such as hypoxia induce metabolic alterations, these alterations depend on the specific nature of the molecular aberrations. As a consequence each tumor has its own specific metabolic demands.
Adding an extra level of complexity, instead of being a consequence of carcinogenesis, altered metabolism may also drive carcinogenesis. Hotspot mutations in isocitrate dehydrogenase 1 and 2 (Table I, step 22) in substantial percentages of diffuse gliomas of the brain, acute myeloid leukemia, chondrosarcomas and hepatic cholangiocarcinomas result in consumption of alpha ketoglutarate (α-KG) and NADPH to produce the oncometabolite D-2-hydroxyglutarate (D-2HG) that can accumulate to milliMolar concentrations. The small difference between the chemical structures of α-KG and D-2HG in combination with the high concentrations of the latter, results in competitive displacement of α-KG from α-KG-dependent enzymes that subsequently cannot function properly. Important examples are the Ten Eleven Translocation (TET)-family of enzymes that are involved in demethylation of CpG islands in the DNA, and JmJ proteins that are involved in histone demethylation. Consequently IDH-mutated cancers present with hypermethylated CpG islands and histones, resulting in deranged gene transcription profiles. Of importance, the consumption of NADPH in IDH-mutated cancer cells results in low levels of reduced glutathione and decreased resistance to reactive oxygen species (ROS). The increased activity of ROS in IDH mutated cancer cells may increase the chance of second hits in oncogenes and tumor suppressor genes, and result in cancer. Except for the known hotspot mutation that leads to 2-HG production, IDH-mutants have been described that are defective in NADPH production, but do not produce D-2HG (1).
Other examples of cancers in which mutations in metabolic enzymes are cancer drivers are phaeochromocytomas and paragangliomas, that carry inactivating mutations in one of the SDH subunits A-D (Table I, step 30) or in SDH assembly factor SDHAF2 (2). These mutations can be hereditary, leading to the HPGL/PCC syndrome, or somatic. Other mutations in metabolic genes that cause cancer occur in FH (Table I, step 31). Mutations in FH are associated with leyomyomatosis and papillary renal cell cancer (3). Mutations in VHL protein occur in clear cell renal cell cancers, and directly result in glycolysis via a defect in HIF breakdown, as described above. The IDH, SDH and FH genes can therefore be considered tumor suppressor genes.
Dietary compounds, food supplements or safe to use drugs exist that can inhibit metabolic pathways, and the use of such drugs have been considered as potentially beneficial for the treatment of cancer. Examples are deoxyglucose, inhibiting glucose uptake by the cell and preventing glycolysis (Table I, step 1 and following) (4), 3-bromopyruvate, blocking the activity of hexokinases (Table I, step 2 and following) (5, 6), 6-amino-niocotinamide (6-AN, blocking G6PD in the pentose phosphate pathway, Table I, step 3 (7)), metformin, blocking OXPHOS (8), bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl)ethyl sulfide (BPTES, blocking glutaminase, Table I, step 28), epigallo-3-catechin gallate (EGCG, blocking glutamate dehydrogenases and other NADPH-generating enzymes (Table I, step 41) and fatty acid synthase Table I, step 36 (9)) (10), cerulenin, inhibiting fatty acid synthase (Table I, step 48) (9). These inhibitors have been shown to have anti-tumor effects in different models of cancers.
Although these compounds have also been tested in humans, anti-cancer effects may require systemic concentrations that are not tolerated by healthy tissues. At non-toxic concentrations, treatment with metabolic inhibitors below maximal tolerated doses can however augment the activity of other treatments, such as radiotherapy, chemotherapy or targeted therapy.
Current treatment protocols for patients with cancers who cannot be cured by surgery are ineffective in that cancers generally develop resistance to treatment (11-14).
There is therefore a great need for safe and adjuvant treatments that increase the efficacy of the state of the art therapies. These state of the art therapies are applied according to guidelines that are based on the outcomes of phase III clinical trials. For some cancer types, the effects of treatment can be predicted (e.g. colon cancers with KRAS mutations do not respond to EGFR inhibitors, cancers with the BRAF-V600E mutation develop resistance to the BRAF inhibitor vemurafenib by upregulating signaling from EGFR, gliomas with hypermethylation of the DNA repair gene MGMT respond better to the DNA alkylating chemotherapy Temozolomide). Although some cancer types are now routinely analyzed for so called companion biomarkers-biomarkers based on which a personalized treatment can be initiated—analyses on such markers cannot be performed if tissue cannot be made available, such is e.g. the case in patients with inoperable cancer.
Therefore there is an urgent need for a test that measures parameters in a patient (subject) that are relevant for treatment decision making, with treatment protocols consisting of the most appropriate metabolic inhibitors, combined with the most appropriate available targeted drugs or radiotherapy or chemotherapy. Currently available tests to investigate metabolism in cancer include magnetic resonance spectroscopic imaging (MRSI), but this is a technically challenging technique that can only be performed in specialized centers and requires concomitant in depth knowledge of MR principles and cell biology. Furthermore, MRSI can only measure a limited number of metabolites. An alternative method to investigate metabolism in cancer is to make extracts of metabolites of a cancer and perform mass spectroscopy. In this case, results will be influenced by the fact that these assays are performed on tissues that have seen hypoxia after surgery.
Molecular diagnosis of cancer is currently performed by analysis of tumor DNA and aims for detection of actionable mutations. DNA analyses allow the identification of mutations and variations in metabolic enzymes such as FH, SDH and IDH, and actionable mutations and amplifications in oncogenes and tumor suppressor genes. DNA analyses can be performed using whole genome analyses or whole exome analyses but can also be performed with Molecular Inversion Probes as described in the literature (15). The technique is depicted in
DNA analyses cannot measure gene activity, e.g. activity of metabolic genes, which is regulated by epigenetic processes and the presence of transcription factors and transcription repressors (17).
A) Heat map of the individual gene profiles. Unsupervised agglomerative clustering of log-transformed expression levels of the targeted genes of interest was performed. Agglomerative clustering was performed according to WardD2 method by calculating Manhattan distance between individual profiles using bio-informatic R-software scripts.
B) Kaplan-Meier curve displaying the overall survival data of the computer-generated groups A and B of the heat-map in panel a). The results show that groups A and B have different survival with high significance (Fisher's exact test; p<0.0001), demonstrating that this test has high prognostic value in gliomas. Groups A and B are here annotated as IDH-MT and IDH-WT.
C) heterozygous IDH1R132H detection in one of the samples, in this case with 38% of transcripts being from the mutant allele and 62% of transcripts from the wt allele
D) Subgroup analysis of IDH-wild-type patients with very poor survival (OS<12 months) versus IDH-wild-type patients with better prognosis (OS>14 months) showed that high expression levels of carbonic anhydrase 12 are associated with poor prognosis (p<0.001; Fisher's exact test, see Kaplan-Meier curve in D).
The present inventors have used multiplex profiling of RNA transcripts to determine which genes that are involved in metabolism are active, and which genes that are involved in pathologies are active. The inventors have found that from the combined information in the RNA profiles, the metabolic pathways that are most prominent in the pathological tissue can be deduced, and the genes that are actively involved in pathologies can be identified. This information can result in a personalized advice to treat an individual suffering from a disease with e.g. drugs that target the product of the gene that is aberrantly expressed and is involved in disease progression. These drugs (pharmaceutical compounds) include but are not limited to drugs that are approved by the United States Food and Drug Administration (FDA) and/or the European Medicines Agency (EMA) and are known as targeted drugs to the person skilled in the art. Such treatment advice can be combined with an advice to treat the disease further with a compound that inhibits the most essential metabolic pathways in the pathological tissue. This concept is known as synthetic lethality to the person skilled in the art. Added value of the test is generated by concomitant information on mutation status of metabolic genes.
The test requires a small aliquot of an RNA of interest that may be derived from solid tissue, isolated cells or bodily fluids, including, but not limited to, saliva, urine, sperm, blood, blood platelets and cerebrospinal fluid. The sample RNA can be converted to copy-DNA (cDNA) using a method known in the art, such as using oligo-dT primers or a mixture of random hexamer oligonucleotide primers. These techniques are standard techniques and are known to the person skilled in the art ((see e.g. Green and Sambrook (2012) Molecular Cloning: A Laboratory Manual, Fourth Edition, Cold Spring Harbor Laboratory Press, NY).
The RNA of interest may be from human genes but may also be from genes of pathogens such as DNA viruses and RNA viruses, including but not limited to human immune deficiency virus (HIV); human papilloma viruses, including but not limited to the subtypes HPV16 and HPV18; hepatitis A virus; hepatitis B virus; hepatitis C virus; hepatitis E virus; Ebola virus; Epstein Bar Virus (EBV); influenza viruses; West-Nile virus, chikungunya virus, polyoma virus; cytomegalovirus; rhinovirus, but also genes from the category of oncolytic viruses that are known to persons skilled in the art to treat cancers. The RNA of interest may also be from genes of parasites, including but not limited to Plasmodium falciparum and Plasmodium vivax, parasites causing malaria, and trypanosoma. In addition, the RNA of interest may be from (pathogenic) fungi, including but not limited to Aspergillus. The RNA of interest may also be from (pathogenic) bacteria, such as Listeria, Legionella, Staphylococcus, Streptococcus, Mycobacterium and/or Yersinia.
The subject (interchangeably also referred to as patient) may be a human or an animal. Accordingly, the RNA of interest may also be from genes from domesticated, wild and farm animals and/or from genes that are present in pathogens such as the pathogens listed here above or their counterparts that cause disease in animals.
The present invention further provides for a set of single molecule molecular inversion probes (smMIPs) to detect the RNAs of interest that carry the information that is needed to formulate a treatment advice. A preferred set is selected from the group listed in Table II.
A preferred method of generating RNA profiles is by using smMIPs that can be designed with the published MIPGEN protocol (18) that selects optimal ligation and extension probe sequences that are predicted to hybridize against a cDNA of interest while leaving a gap between the ligation and extension parts of the probe. The ligation and extension parts of the probes may hybridize to any part of the cDNA, including sequences that are protein encoding and untranslated regions. Extension and ligation parts of the probes can be located in the same exon.
A preferred method is to locate the ligation and extension parts of the probes in different exons of a cDNA, which allows detection of specific splice variants.
A preferred method according to the invention is to contact a library of designed smMIPs according to the invention, that may consist of any number of smMIPs, with a population of cDNA molecules. After an initial heating and denaturation step followed by cooling, each smMIP will hybridize to its target cDNA sequence. By incubating the mixture with a DNA polymerase enzyme, all four deoxynucleotides and DNA ligase in an appropriate buffer, the extension probe part of the MIP will be extended until the 5′ end of the ligation probe is reached. The DNA ligase will then covalently link the 3′ end of the extended extension probe part to the ligation probe part, producing a circular smMIP molecule.
In the next step, a method known to the person skilled in the art, is used to remove unreacted, linear smMIPs and cDNA from the reaction mixture by exonuclease treatment, leaving a purified library of circular smMIPs.
Using a forward and a reverse oligonucleotide primer that specifically anneal to the backbone sequence that connects the ligation and extension probes parts of the MIP, a PCR amplification of the gap sequence is performed. Preferably, one of the oligonucleotide primers that are used in this PCR is equipped with a barcode, allowing easy selection of all PCR products that are obtained from a specific sample. In a next step, the library of PCR amplicons are preferably analyzed on a next generation sequencing platform that yields FASTQ files containing information on nucleotide sequences of all PCR amplicons in the sample. Using an algorithm all PCR amplicons with the same barcode are grouped, producing a list of sequences for each individual cDNA sample.
Next, using another algorithm that uses the UMI, all identical PCR products will be considered to be derived from one originating smMIP. In this manner for each original RNA sample a list can be created that contains values that represent the original number of circularized smMIPs in the original library. This number is proportional to the number of cDNAs in the original sample.
In a preferred method of interpretation, the values obtained for each individual smMIP are divided by the summated values of all smMIPs for each sample, followed by multiplying with a factor of one million, thus yielding a fragments per million value for each smMIP.
In a preferred method of interpretation, the mean FPM values of all different smMIPs that correspond to one transcript, are considered to be proportional to the number of transcripts that were present in the initial RNA sample of the analysis.
In another preferred method of interpretation, mean FPM values of individual transcripts are divided by mean FPM values of so-called house-keeping genes, to yield a relative abundance value of a transcript of interest.
In another preferred method, mean FPM values for transcripts from genes that are involved in metabolic pathways are used to deduce the predominant metabolic pathways in a tissue.
A preferred method to analyze the FASTQ files further is to detect mutations in the next generation sequencing data. Preferably, mutations are considered as relevant if they are detected in more than two reads. The sequence information as provided in the FASTQ files should not be so narrowly construed as to require inclusion of erroneously identified bases. The skilled person is capable of identifying such erroneously identified bases and knows how to correct for such errors. A list of relevant mutations in a sample can be included in a database, preferably a standard query language (SQL)-based database that allows statistical analyses, for example by multivariate analysis.
A preferred method of analysis of the database results in a list of metabolic pathways that are active in a tissue or in a person with a disease and that can be used to give a dietary advice to relieve the symptoms of the disease and to improve the efficacy of other therapies.
Another preferred method of analysis of the database results in a list of aberrancies that can be treated with available pharmacological drugs.
Yet another preferred method of the invention uses a software algorithm that translates RNA profiles of diseased tissues directly to a treatment advice that can be given via an application that can be installed on a personal computer or a mobile device.
The method according to the invention can be readily implemented in routine patient care in case RNA from diseased tissue or blood platelets is available.
Accordingly, in a first aspect, the present invention provides for a method for in vitro determination of the susceptibility and/or resistance of a subject suffering from or at risk of a disease or condition for a drug to treat the disease or condition, comprising:
Said method is herein referred to as the method according to the invention. “RNA profiling” is herein also referred to as targeted RNA sequencing of transcripts. An aberrant level of a transcript is a level of transcription that can either be higher or lower than the transcript level as compared to a reference sample and/or as compared to the level of transcript in a healthy subject.
Preferably, in a method according to the invention, RNA profiling is performed by multiplex mRNA sequencing, targeting multiple regions of interest. The sample RNA of interest may first be converted to copy-DNA (cDNA) using a method known in the art, such as using oligo-dT primers or a mixture of random hexamer oligonucleotide primers. The RNA of interest may be from human genes but may also be from genes of pathogens such as DNA viruses and RNA viruses, including but not limited to human immune deficiency virus (HIV); human papilloma viruses, including but not limited to the subtypes HPV16 and HPV18; hepatitis A virus; hepatitis B virus; hepatitis C virus; hepatitis E virus; Ebola virus; Epstein Bar Virus (EBV); influenza viruses; West-Nile virus, chikungunya virus, polyoma virus; cytomegalovirus; rhinovirus, but also genes from the category of oncolytic viruses that are known to persons skilled in the art to treat cancers. The RNA of interest may also be from genes of parasites, including but not limited to Plasmodium falciparum and Plasmodium vivax, parasites causing malaria, and trypanosoma. In addition, the RNA of interest may be from (pathogenic) fungi, including but not limited to Aspergillus. The RNA of interest may also be from (pathogenic) bacteria, such as Listeria, Legionella, Staphylococcus, Streptococcus, Mycobacterium and/or Yersinia.
Preferably, in a method according to the invention, the multiplex mRNA sequencing is performed using molecular inversion probes (MIPs), preferably MIPs comprising a detectable moiety, preferably a unique identifier sequence of a string of 3 to 10 random nucleotides (depicted as “N” in a sequence listing), more preferably a string of 3, more preferably 4, more preferably 5, more preferably 6, more preferably 7, most preferably 8, or preferably more than 8 random nucleotides (N) adjacent to the ligation part of the MIP or to the extension part of the MIP sequence (smMIPs).
Preferably, in a method according to the invention, the aberrant level of a transcript, an alternative splice variant and/or a mutation is linked to a an aberrance in a metabolic pathway which is in turn linked to the susceptibility and/or resistance of a subject suffering from or at risk of a disease or condition for a drug. In all embodiments of the invention, a drug is as meant in the art, a pharmaceutical compound. Such pharmaceutical compound may be comprised in a pharmaceutical composition. In all embodiments of the invention, a subject is a human or an animal, preferably a human. An animal may be any animal, preferably a domestic, wild or farm animals.
Preferably, in a method according to the invention, the disease or condition is at least one selected from the group consisting of a cancer, a viral infection, a bacterial infection, an autoimmune disease and a genetic disease.
In a method according to the invention, the sample may be any appropriate sample known to the person skilled in the art, preferably selected from the group consisting of a tissue, a tumor tissue, urine, sperm, saliva, blood, blood plasma, cerebrospinal fluid, blood platelets, and/or exosomes, more preferably selected from tumor tissue and blood platelets.
In a method according to the invention, the metabolic pathway is preferably selected from the group consisting of a glucose processing pathway, a glutamine processing pathway and/or a fatty acid pathway.
Preferably, in a method according to the invention, the multiple regions of interest are within the mRNA of—glucose processing genes, glutamine processing genes, fatty acid anabolism genes, transporter genes, redox homeostasis genes, genes with potential involvement in cancer, such as oncogenes, genes involved in angiogenesis, genes involved in immune suppression, and viral genes.
Preferably, in a method according to the invention, the multiple regions of interest are within the mRNA of at least one, two, three, four, five or at least six genes selected from the group consisting of: ABAT, ACACA, ACACB, ACLY, ACO2, ACSS2, ADPGK, ALDOA, ARHGAP26, ATG4A. ATP5A1, CBR1, CBS, CHKA, CKB, CPT1A, CYCS, EGLN1, ENO1, G6PC, GAD1, GCLC, GCLM, GFPT1, GLDC, GSS, HK1, HK2, HK3, GLY1, G6PD, RGN, PGD, RPIA, RPE, TKT, PGI, ALDOA, GAPDH, PGAM1/2, ENO, PKM1/2, PDHA1, PDK1, PFKB1, PFKMb, PGAM1, PGD, PGK1, PKM, PRDX1, PRKAA1, RPIA, PC, CS, ACO1, IDH1, IDH2, IDH3A, IDH3B, IDH3G, OGDH, SUCLA2, SDHA/B/C/D, FH, MDH1, MDH2, PDK, LDHA, LDHB, SLC16A1, SLC16A3, CA9, CA12, SLC4A10, VHL, SDH, SDHAF2, HPGL/PCC, FH, CS, D-2HGDH, L-2HGDH, FH, IDH1-3A-G, MDH1-2, MYC, OGDH, SDHA-D, VHL, PHD, HIF1a, EPAS2 PDCD1, SLC1A5, ASCT2, GLS, GLUD1/2, GOT, GPI, GS, BCAT1, BCAT2, SLC1A2, SLC7A11, SLC25A1, ACLY, ACACA, ACACB, FASN, CPT1, SLC5A7, CHKA, CPT2, VLCAD, HADHA/B, SCAD, MCAD, LCAD, SCHA-D, 2-Enoyl-VoA hydratase, MCKAT, SLC16A1, SLC16A7, SLC2A1, SLC2A3, SLC5A1, SLC5A5, SLC7A1, SLC9A1, SLCA12, redox homeostasis genes: NAMPT, NAPRT1, NOX1, NOX3, NOX4A, NQO1, SOD, SOD2, CAT, TAL, TIGAR, TRX, PARP1, ALK, AXL, BRAF, KRAS, TP53, MAPK8, MYC, TP5313, FGFR1, FGFR2, IGF1-R, KDR, NTRK1, NTRK2, PDGFRA, PDGFRB, EGFR, EGFRvIII, ERBB2, ERBB3, ERBB4, MERTK, PLXND1, RET, Androgen receptor (AR), AR variant 7, AR variant 12, FOLH1, KLK3, MET, METdelta14, METdelta7-8, KIT, RON PTEN, VEGF-A121, VEGF-A144, VEGF-A165, VEGF-A189, CD274, CTLA4, HPV-E2, HPV-E6, and HPV-E7.
Preferably, in a method according to the invention, the multiple regions of interest are within the mRNA of:
Preferably, in a method according to the invention, the presence of an aberrant level of a transcript, an alternative splice variant and/or a mutation also provides an indication for treatment with dietary compounds or phytochemicals, optionally in combination with a drug. The person skilled in the art knows that drug treatment can beneficially be combined with treatment with dietary compounds or phytochemicals.
The method according to the invention can conveniently be used for guiding treatment in a subject (personalized medicine). Accordingly, in a further aspect, the invention provides for a method of treatment of a subject suffering from or at risk of a disease or condition, comprising:
Preferably, in a method of treatment according to the invention, the disease or condition is at least one selected from the group consisting of: a cancer, including but not limited to glioma, meningioma, ependymoma, pilocytic astrocytoma, adenocarcinomas, sarcomas, hemangioma, head and neck cancer, breast cancer, lung cancer, prostate cancer, kidney cancer, ovarian cancer, endometrial cancer, cervical cancer, colon cancer, rectal cancer, pancreatic cancer, esophagus cancer, basal cell cancer, penile cancer, vulva cancer, melanoma, uveal melanoma, lymphoma, acute myeloid leukemia, acute lymphoblastic leukemia, cholangiocarcinoma, hepatocellular carcinoma, soft tissue sarcoma, and osteosarcoma; a viral infection; a bacterial infection; an autoimmune disease and a genetic disease.
Preferably, in a method of treatment according to the invention, the drug treatment is supplemented with treatment with dietary compounds or phytochemicals.
The invention further provides for a medicament (drug) for use in the treatment of a subject suffering from or at risk of a disease or condition, wherein:
Preferably, in the medicament (drug) for use according to the invention, the disease or condition is at least one selected from the group consisting of a cancer, a viral infection, a bacterial infection, an autoimmune disease and a genetic disease.
Preferably, in the medicament (drug) for use according to the invention, the drug treatment is supplemented with treatment with dietary compounds or phytochemicals.
The invention further provides for method for the production of a medicament (drug) for the treatment of a subject suffering from or at risk of a disease or condition, comprising:
Preferably, in the method for the production of a medicament (drug) for the treatment according to the invention, the disease or condition is at least one selected from the group consisting of a cancer, a viral infection, a bacterial infection, an autoimmune disease and a genetic disease.
Preferably, in the method for the production of a medicament (drug) for the treatment according to the invention, the drug treatment is supplemented with treatment with dietary compounds or phytochemicals.
The invention further provides for a molecular inversion probe selected from the group as set forward in Table II. The invention further provides for a set of molecular inversion probes of at least two, three, four, five, six or more selected from the group as set forward in Table II.
The invention further provides for a library of circularized molecular inversion probes obtainable by a method according to the first or second aspect of the invention.
In this document and in its claims, the verb “to comprise” and its conjugations is used in its non-limiting sense to mean that items following the word are included, but items not specifically mentioned are not excluded. In addition, reference to an element by the indefinite article “a” or “an” does not exclude the possibility that more than one of the element is present, unless the context clearly requires that there be one and only one of the elements. The indefinite article “a” or “an” thus usually means “at least one”.
The word “about” or “approximately” when used in association with a numerical value (e.g. about 10) preferably means that the value may be the given value (of 10) more or less 5% of the value.
The sequence information as provided herein should not be so narrowly construed as to require inclusion of erroneously identified bases. The skilled person is capable of identifying such erroneously identified bases and knows how to correct for such errors. In case of sequence errors, the sequence of the polypeptides obtainable by expression of the genes present in SEQ ID NO: 1 containing the nucleic acid sequences coding for the polypeptides should prevail.
All patent and literature references cited in the present specification are hereby incorporated by reference in their entirety.
In the past four decades an overwhelming amount of data has become available on the molecular events that underlie carcinogenesis. Research has mainly focused on molecular alterations and their consequences for among others the PI3K/pAKT/mTOR pathway(19-22) and cell cycle control, apoptosis (23, 24) and DNA repair pathways (25, 26). Currently, numerous FDA-approved drugs are available that target cancer cells based on these genetic defects with a level of specificity that is not attainable with conventional chemotherapies (27, 28), permitting personalized medicine. Whereas targeted cancer therapies may prolong survival, it is now widely recognized that inherent genetic instability ultimately leads to therapy resistance of most cancers (11-14).
For proliferation, cancer cells need to generate ATP to maintain energy balance and ion homeostasis, import carbon and nitrogen sources for synthesis of amino acids, nucleotides and lipids (29, 30) and maintain redox potential to protect cells against oxidative stress (31). Blocking one or more of these processes may prohibit proliferation and/or sensitize cells to toxic therapy in a synthetic lethality approach. As an example, increasing oxidative stress in a cancer with metabolic inhibitors may enhance the efficacy of radiotherapy (32) or chemotherapy (33). With the increasing knowledge of deranged metabolic pathways in cancer (34-37), (adjuvant) targeting of cancer-specific metabolic pathways may be a highly interesting addition to current treatment protocols. The best-known example of cancer-specific metabolic adaptation is aerobic glycolysis, also known as the Warburg effect (38). As glycolysis is inefficient in terms of ATP production, cancer cells characteristically upregulate glucose transporters GLUT1 and/or GLUT3. Besides glucose, glutamine and fatty acids are recognized as important fuels for cancer cells (39, 40) (41, 42).
While metabolic adaptations are mostly seen as a consequence of carcinogenesis, it has been unequivocally established that metabolic alterations can also cause cancer, examples being mutations in genes encoding mitochondrial (e.g. IDH2, FH, SDH) and cytosolic (e.g. IDH1) metabolic enzymes (43, 44) (45, 46) (1, 47, 48), the latter being prominent in among others low grade gliomas and secondary glioblastomas (1, 48). Clear cell renal cell carcinoma (ccRCC) is now considered a metabolic disease with metabolic alterations resulting indirectly from inactivating mutations in or epigenetic silencing of VHL, found in ˜80% of clear cell renal cell cancers (ccRCC) (49). pVHL is a major regulator of ubiquitination and breakdown of transcription factor hypoxia inducible factors HIF-1α and HIF-2α (49). Mutations in the aforementioned metabolic enzymes and in VHL have been shown to induce epigenetic alterations that affect expression of other metabolic enzymes in an unpredictable fashion (17, 50-52).
To apply metabolic inhibitors as potential additions to the current anti-tumor armamentarium, it is of high importance to identify which metabolic pathways are active in a specific cancer in a personalized fashion. Here we applied a novel next generation-sequencing based method using single molecule molecular inversion probes (smMIPs(15)), to detect expression levels of 104 genes involved in metabolism, and concomitantly identify variants therein. As a proof of concept, we applied smMIPs to map part of the metabolic transcriptome of a VHL-defective ccRCC cell line and a corresponding VHL-rescued isogenic derivative, as well as in patient derived glioma xenograft models. We validated the technique by correlating results with whole transcriptome RNAseq data (as gold standard for transcriptome analysis) and protein expression. We further verified the ability of the assay to detect oncogenic mutations in cell lines and patient tumor tissue.
Our data show that targeted RNA sequencing of transcripts encoding metabolic enzymes using smMIPs predict the predominant metabolic pathways that are operational in cancer (53) and simultaneously allows variant detection in the targeted transcripts.
The cell line SKRC7 is derived from a primary human ccRCC and has been described before (54). Cells were cultured in RPMI 1640 (Lonza Group, Switzerland) supplemented with 10% fetal calf serum (FCS) (Gibco, Thermo Fisher Scientific, Waltham, Mass., USA) and 40 μg/ml gentamycin (Centrafarm, Etten-Leur, The Netherlands). An isogenic SKRC7 cell line expressing a functional haemagglutinine (HA)-tagged VHL (SKRC7-VHLHA) was created by transfection with pcDNA3.1-VHLHA followed by selection of stable transfectants in the same medium with 400 μg/ml geneticin (Gibco, Thermo Fisher Scientific, Waltham, Mass., USA). The patient-derived glioma xenograft models E478 and E98 have been described before (55, 56).
Use of patient material was according to the guidelines of the local ethical committee for use of patient material and was performed with informed consent. Surgically obtained tissue from a male patient with a grade III astrocytoma was snap frozen frozen in liquid nitrogen.
RNA and cDNA Preparation—
Total RNA was isolated from sections of snap-frozen E478 xenograft tissue, human tumor tissue and from 80% confluent SKRC7 and SKRC7-VHLHA cells using TRIzol reagent (Life Technologies, ThermoFisher Scientific, Waltham, Mass., USA) according to the manufacturers' instructions. RNA quality was estimated based on relative levels of 28S, 18S and 5S rRNA bands on agarose gel and with Bioanalyzer assays (Agilent Technologies, Amstelveen, The Netherlands). RNA was reverse transcribed to cDNA using Superscript II reverse transcriptase (Invitrogen, ThermoFisher Scientific, Waltham, Mass., USA) and random hexamer primers (Promega, Madison, Wis., USA) according to standard protocols. Next, cDNA was purified using the NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel, DOren, Germany). For quality control, cDNA was subjected to PCR for reference gene hydroxymethylbilane Synthase (HBMS) with forward primer HMBSFw (5′-CTGGTAACGGCAATGCGGCT-3′) and reverse primer HMBSRv (5′-TTCTTCTCCAGGGCATGTTC-3′) using AmpliTaq Gold 360 master mix (Applied Biosystems, ThermoFisher Scientific, Waltham, Mass. USA).
High quality RNA with RIN scores >8 was subjected to whole transcriptome RNAseq according to standard protocols. Sequencing was performed on an Illumina Hiseq and yielded 30-50 million reads per sample (paired end sequencing protocol). The dataset was analyzed using the ‘Tuxedo’ protocol; reads were mapped against the RefSeq human genome (hg19) with TopHat and final transcript assembly was done with the Cufflinks package (57). Normalization was done with both Cuffquant and the calculation of fragments as transcript per million mapped reads (TPM) to obtain relative expression values. Occurrence of single nucleotide variants was visualized in the Integrated Genomics Viewer browser (IGV, the Broadinstitute).
smMIP Design
The technique of targeted RNAseq using smMIPs is depicted in
642 smMIPs (IDT, Leuven, Belgium) were pooled at 100 μM/smMIP. The smMIP pool was phosphorylated using T4 Polynucleotide Kinase (New England Biolabs, NEB, Ipswich, Mass., USA) in T4 DNA ligase buffer (NEB) at 37° C. for 45 min, followed by inactivation for 20 min at 65° C. The capture reaction was performed with 50 ng of cDNA and an estimated 8000-fold molar excess of the phosphorylated smMIP pool (16) in a 25 μL reaction mixture containing Ampligase buffer (Epicentre, Madison, Wis., USA), dNTPs, Hemo KlenTaq enzyme (New England Biolabs, NEB, Ipswich, Mass., USA) and thermostable DNA ligase (Ampligase, Epicentre). The capture mix was incubated for 10 min at 95° C. (denaturation), followed by incubation for 18 h at 60° C., during which hybridization and concomitant primer extension and ligation occurs. Directly after this step non-circularized smMIPs, RNA and cDNA were removed by treatment with 10 U Exonuclease I and 50 U of Exonuclease III (both NEB) for 45 min at 37° C., followed by heat inactivation (95° C., 2 min). The circularized smMIP library was subjected to standard PCR with 2× iProof High-Fidelity DNA Polymerase master Mix (Bio-Rad, Hercules, Calif.) with a primer set containing a unique barcoded reverse primer for each sample. Generation of PCR products of correct size (266 bp) was validated on agarose gel electrophoresis, and PCR-libraries from different samples were pooled based on relative band intensity. The pool was then purified using AMPureXP beads (Beckman Coulter Genomics, High Wycombe, UK) according to manufacturers' instructions. The purified library was run on a TapeStation 2200 (Agilent Technologies, Santa Clara, Calif., USA) and quantified via Qubit (Life Technologies, ThermoFisher Scientific, Waltham, Mass. USA) to assess quality of the library. Reproducibility of the technique was tested by preparing biological replica libraries, using different RNA preparations from the same cell lines.
Libraries were sequenced on the Illumina NextSeq platform (Illumina, San Diego, Calif.) at the Radboudumc sequencing facility to produce 2×150 bp paired-end reads. Reads were mapped to the reference transcriptome (hg19) using the SeqNext module of JSI SequencePilot version 4.2.2 build 502 (JSI Medical Systems, Ettenheim, Germany). The random 8 nt sequence flanking the ligation probe was used to reduce PCR amplicates to one smMIP (unique reads).
All single nucleotide variants (SNVs) called with a minimal variant percentage of 5% detected in at least 5 unique reads (forward and reverse) were selected for further analysis. Variants were annotated and classified into synonymous or non-synonymous. Next, they were validated in whole transcriptome RNAseq data, generated from different RNA isolations from the same cell lines.
Individual read counts for each smMIP were divided by the total read count within a sample and multiplied by 106 resulting in a fragment per million (FPM) value for each smMIP in a sample. We choose for this normalization procedure instead of normalization against housekeeping genes because perfect housekeeping genes do not exist. E.g. expression of metabolic genes is subject to variation, dependent on cell cycle, and the same is true for expression of genes such as actin and tubulin.
Cell extracts were prepared from SKRC7 and SKRC7-VHLHA cells by solubilizing in RIPA buffer (Cell Signaling) and protein concentrations were determined using BCA assays. 20 μg of protein was separated on 12% SDS-PAGE gels and electroblotted on nitrocellulose. After blocking in Odyssey blocking buffer (1:1 in PBS) membranes were incubated overnight in Odyssey blocking buffer containing antibodies against HK-2 (2867S, Cell signaling technology), CA9 (M75, Dr. Oosterwijk) or γ-tubulin (C20, Santa Cruz Biotechnology, Dallas, Tex.) as loading control. Antibodies were detected with secondary antibodies conjugated with Alexa680 or DyLight800, and signal was visualized with the Odyssey scanner (LI-COR).
FPM values for each transcript (mean FPM values from all smMIPs targeting one transcript) were correlated with TPM values (transcripts per million values for the same transcript obtained from whole RNAseq data from the same cell lines. For three samples replicate assays were performed. Correlation analyses were performed using GraphPad Prism v.5.03 (GraphPad, San Diego, Calif., USA).
SmMIP-based next generation sequencing (NGS) of genomic DNA was recently introduced in routine diagnostics in our institute to detect tumor-associated mutations in DNA (16). To investigate whether smMIPs can also be used for multiplex determination of gene expression levels, concomitant with variant detection, we designed a smMIP set for targeted detection and sequencing of transcripts encoding metabolic enzymes. To establish the strength of the technique we used the SKRC7 and SKRC7-VHLHA isogenic cell line pair as prototypical cell lines in which different metabolic pathways prevail. Like 80% of ccRCCs, the SKRC7 cell line carries a defective VHL gene resulting in constitutive stabilization of HIF-1α and HIF-2α and a pseudohypoxic response (53, 54, 58). Re-introduction of VHL was expected to result in rapid HIF1/2 breakdown and repair of this metabolic aberration.
Whole RNAseq-derived gene expression data of SKRC7 cells confirmed the presence of a nonsense and functionally inactivating Q132-stop mutation in 100% of VHL transcripts (
Using an initial set of 642 smMIPs, covering 104 transcripts of interest for this study (see Table I), we tested our protocol of library preparation with 50 ng of hexamer-primed cDNA generated from 13 different RNA samples (cell line- and xenograft derived) of which also whole RNAseq datesets were available. A 25-cycle PCR with barcoded primers on the circularized smMIP library yielded PCR fragments of the expected size of 266 bp (not shown).
Based on initial experiments we also tested the procedure on 10 and 25 ng cDNA. Both conditions yielded less PCR fragments and less unique reads compared to 50 ng. We therefore continued with 50 ng cDNA input in subsequent experiments. Illumina NextSeq sequencing of the libraries generated of SKRC7 and SKRC7-VHL cells, yielded 286,000 and 69,000 annotated unique reads respectively (corrected for PCR-amplicates based on the random 8N sequence in the smMIP), which is in the range of other samples run with the same smMIP panel (not shown). For most transcripts performance of individual smMIPs was variable (see example in Table II, showing FPM values for 10 different smMIPs designed against the VHL transcript in both cell lines), a known phenomenon also in DNA smMIP NGS (16)). This was a priori reason to include at least 5 smMIPs per gene transcript in our panel, allowing transcriptome analysis using mean normalized smMIP values for each transcript. This number was a trade-off between generating expensive, large panels which would yield in part futile and irrelevant data, and too small panels resulting in under- or overestimation of transcript levels.
First we compared the targeted smMIP RNAseq dataset, generated with a 864 smMIP panel, to a whole transcriptome RNAseq dataset (considered as gold standard), performed on different RNA isolates from the same cell lines. The whole RNAseq dataset consisted of 3.2×107 and 3.4×107 reads, assigned to 44,503 different transcripts for SKRC7 and SKRC7-VHLHA, respectively. For each transcript of interest, TPM (transcripts per million) values from the whole RNAseq dataset were plotted against mean FPM values from smMIP analyses. Such analysis for metabolic transcripts and tyrosine kinase transcripts separately, gave correlation coefficients of 0.903 and 0.974, respectively, for SKRC7 (
One of the appealing characteristics of targeted RNAseq using smMIPs is that panels can be expanded to detect novel transcripts of interest. To test how this affects the outcome of the assay, we added 222 smMIPs for detection and targeted sequencing of other transcripts of interest to our initial panel and re-performed the assay using newly isolated RNA from the same cell line. Relative levels of transcripts within samples correlated well between assays with the initial and the expanded smMIP set (SKRC7: r=0.903, SKRC7-VHLHA: r=0.876).
Functional Validation of Targeted smMIP Data
Having confirmed the validity of the smMIP dataset, we analyzed expression levels of genes involved in metabolism in SKRC7 and SKRC7-VHLHA cells.
To investigate whether smMIP based RNAseq allows efficient detection of single nucleotide variants (SNVs), we performed variant calling of the smMIP library in SeqNext. Several heterozygous and homozygous variants were detected that could be validated in the whole RNAseq dataset (see VHL example in
Here we present a novel approach of library generation for targeted RNA next generation sequencing using smMIPs and show that the technique yields reproducible and biologically relevant information which is qualitatively comparable to whole transcriptome RNAseq. Especially for the evaluation of relative contributions of metabolic pathways in cancer, that are amenable to epigenetic and transcription-factor based regulation, DNA sequencing may not always yield relevant information. Using an isogenic pair of cell lines differing only in VHL expression as a test case, we here show that targeted RNA seq of transcripts involved in metabolism with our smMIP panel yields relevant information on metabolic pathways with relative abundancies that are similar to that of whole transcriptome RNAseq.
Although the generation of smMIP libraries for targeted RNAseq obviously yields only a fraction of the data compared with whole transcriptome RNAseq, it has distinct advantages: 1) costs of the technique are approximately 5-10% of the cost of whole transcriptome RNAseq; 2) by designing smMIPs with ligation and extension probes localized on neighbouring exons, the library is expected not to be contaminated with heteronuclear RNA, transfer RNAs and ribosomal RNAs that may account for a large number of reads in whole transcriptome RNAseq, dependent on preprocessing; 3) the choice of extension and ligation target sequences allows the design of smMIPs that specifically detect splice variants; 4) the technique allows detection of SNVs and indels with high efficiency, once smMIPs are chosen to cover the mutated region; 5) coverage per target sequence of transcript of interest is higher than with whole transcriptome RNAseq; 6) smMIP sets may be extended with novel smMIPs of interest without affecting performance; 7) data sets are much smaller and easier to handle.
The technique was validated with an isogenic ccRCC cell line pair, differing only in expression of VHL. Our targeted smMIP analysis revealed that, as expected, expression levels of HIF target genes HK-2, CA9, CA12 and LDH-A were downregulated upon rescue of VHL, a known regulator of HIF proteolysis. This suggests that rescue of VHL function reduces flux of glucose into the glycolytic pathway.
Because altered metabolic activity in cancer can be a consequence of general oncogenic stress but also of mutations in genes encoding metabolic enzymes and conditions relating to the microenvironement (oxygen tension, access to stromal cell-derived metabolites (59)) cancer metabolism is in an extremely complex landscape (60). Nevertheless, targeting of cancer-specific metabolic pathways is gaining importance in cancer research. Since decades glycolysis has been considered the predominant metabolic pathway in cancer, but it is increasingly clear that tumors can also thrive using glutamine as carbon and nitrogen donor. Identification of the fuel-processing pathways that represent metabolic Achilles heels in cancer is important to apply metabolic inhibition in a personalized fashion. Approaches to identify metabolic pathways in clinical cancers currently include carbon tracing using ex vivo mass spectroscopy (59, 61, 62) and in vivo 1H-, 13C carbon- or 31P-based magnetic resonance spectroscopic imaging (40, 63) but these approaches are not suitable for implementation in routine patient care. SmMIP-based transcript profiling may be a highly relevant alternative with added value in the field of cancer diagnostics as it can identify metabolic Achilles heels by simultaneously measuring smart combinations of relative gene expression levels and variants. When combined with smMIP sets that detect actionable mutations in oncogenes or tumor suppressor genes, personalized treatment protocols may be further optimized by including inhibitors of the most predominant metabolic pathways such as glycolysis (e.g. 3-bromopyruvate, dichloroacetate(64-66)), pentose phosphate pathway (e.g. 6-aminonicotinamide (33), glutaminolysis (e.g. epigallocathechin-3-gallate(67, 68)), mitochondrial oxidative phosphorylation (e.g. metformin(69-71)), fatty acid oxidation and lipid synthesis (e.g. cerulenin (72)).
Clear cell renal cell carcinoma (ccRCC) are relatively resistant to radiotherapy and chemotherapy. Due to dysfunctional von Hippel-Lindau (VHL) protein these tumors accumulate hypoxia-inducible factors HIF-1α and HIF-2α resulting in pseudohypoxic responses that accompanying aberrant metabolism (49, 73). This translates in expression of a set of transporters and enzymes that increase glucose uptake for use in aerobic glycolysis and lactate production, instead of oxidative phosphorylation (73). Increased uptake of glucose and its conversion to glucose-6-phosphate by the hexokinase family of enzymes leads to an increased flux through the pentose phosphate pathway (PPP), providing the cell with NADPH, the most important form of reducing power in cells, and ribose-5-phosphate (R5P), a precursor of nucleotide synthesis (33). Whereas in normal cells oxidative glucose metabolism yields mitochondrial citrate as a major carbon source for lipid biosynthesis, this pathway is blocked in VHL-deficient cells that process pyruvate towards lactate instead of acetyl-CoA for TCA cycle feeding (74). Cells with a VHL defect therefore use glutamine as a metabolic rescue pathway. During glutaminolysis, glutamine is converted to glutamate and α-KG via the sequential activities of glutaminase and glutamate dehydrogenase. Subsequently cells employ reductive carboxylation of α-ketoglutarate (α-KG) in the cytoplasm to produce isocitrate (reverse reaction of IDH1) that is converted to citrate (aconitase) and acetyl-CoA (ATP-citrate lyase) that, together with NADPH, is processed to fatty acids (75). In VHL-mutated cancers high expression levels of enzymes of the pentose phosphate pathway (PPP), combined with low levels of TCA enzymes correlates with poor survival (76).
In ccRCC differential expression of HIF-1α and HIF-2α is observed, with tumors expressing either both subtypes or exclusively HIF-2a. Part of the effects of HIF-1α and HIF-2α are overlapping, but they also have distinct effects on cell metabolism. HIF-1α causes glycolytic enzyme expression (77) and limits mitochondrial pyruvate consumption (78, 79), thereby blocking anabolic biosynthesis via this pathway. Furthermore HIF-1α inhibits cell cycle progression via inhibition of c-myc (80). HIF-2α however, does not regulate glycolysis and stimulates cell-cycle progression (81). The exact metabolic pathways may therefore differ between different VHL-mutated cancers. Unraveling the metabolic pathways of cancer cells that facilitate malignant behavior is of high importance, since these may be amenable for targeting with the aim to inhibit cell growth and tumor progression, or induce oxidative stress sensitizing cells to radiotherapy or chemotherapy.
Here we investigated the metabolic pathways in two VHL impaired ccRCC cell lines, SKRC-17 (expressing only HIF-2) and SKRC-7 (expressing both HIF-1 and HIF2) (54, 58). Expression of metabolic enzymes was explored with smMIP sequencing, and carbon sources that are essential for proliferation of these cells were identified. Results show that SKRC-7 cells use glucose for lactate production (high levels of enzymes for glucose-to-pyruvate-to-lactate). Gene expression profiles of SKRC-17 suggest that cells use glucose mainly for the pentose phosphate pathway. Both cell types have high levels of glutaminase and glutamate dehydrogenase, suggesting sensitivity to the glutamate dehydrogenase inhibitor EGCG. The high levels of PPP in SKRC17 suggest additional activity of 6-aminonicotinamide (6-AN).
SKRC-7, derived from primary human RCC, and SKRC-17, derived from a soft tissue metastatic lesion of human RCC (82) both carry a non-sense mutation in VHL (Q132X in SKRC-7 and S65X in SKRC-17) and therefore lack functional pVHL. In SKRC-7 this leads to high levels of HIF-1α and HIF-2α, whereas SKRC-17 presents with high levels of HIF-2α only (54, 58). Unless stated otherwise, cells were cultured in RPMI 1640 (Lonza Group, Switzerland) supplemented with 10% fetal calf serum (FCS) (Gibco, Thermo Fisher Scientific, Waltham, Mass., USA) and 40 μg/ml gentamycin (Centrafarm, Etten-Leur, The Netherlands).
642 smMIPs (IDT, Leuven, Belgium) were pooled at 100 μM/smMIP. The smMIP pool was phosphorylated using T4 Polynucleotide Kinase (New England Biolabs, NEB, Ipswich, Mass., USA) in T4 DNA ligase buffer (NEB) at 37° C. for 45 min, followed by inactivation for 20 min at 65° C. The capture reaction was performed with 50 ng of cDNA and an estimated 8000-fold molar excess of the phosphorylated smMIP pool (16) in a 25 μL reaction mixture containing Ampligase buffer (Epicentre, Madison, Wis., USA), dNTPs, Hemo KlenTaq enzyme (New England Biolabs, NEB, Ipswich, Mass., USA) and thermostable DNA ligase (Ampligase, Epicentre). The capture mix was incubated for 10 min at 95° C. (denaturation), followed by incubation for 18 h at 60° C., during which hybridization and concomitant primer extension and ligation occurs. Directly after this step, non-circularized smMIPs, RNA and cDNA were removed by treatment with 10 U Exonuclease I and 50 U of Exonuclease III (both NEB) for 45 min at 37° C., followed by heat inactivation (95° C., 2 min). The circularized smMIP library was subjected to standard PCR with 2× iProof High-Fidelity DNA Polymerase master Mix (Bio-Rad, Hercules, Calif.) with a primer set containing a unique barcoded reverse primer for each sample. Generation of PCR products of correct size (266 bp) was validated on agarose gel electrophoresis, and PCR-libraries from different samples were pooled based on relative band intensity. The pool was then purified using AMPureXP beads (Beckman Coulter Genomics, High Wycombe, UK) according to manufacturers' instructions. The purified library was run on a TapeStation 2200 (Agilent Technologies, Santa Clara, Calif., USA) and quantified via Qubit (Life Technologies, ThermoFisher Scientific, Waltham, Mass. USA) to assess quality of the library.
Reproducibility of the technique was tested by preparing biological replica libraries, using different RNA preparations from the same cell lines.
To verify transcript levels observed in the smMIP sequencing data on protein level, western blots for some of the interesting enzymes active in glycolysis (HK2, PKM2, GAPDH), glutaminolysis (GLUD) and reverse carboxylation (IDH1) were performed. Cells were cultured in 6 well plates till 80% confluency, then cells were harvested and processed to cell lysates by scraping in 100 μl 10 mM Tris-HCL pH 7.5 and 0.32 M sucrose and sonicating on ice (3 cycles of 30 sec max power and 30 sec off, Bioruptor, Diagenode). Lysates were centrifuged (14000 rpm, 10 min, 4° C.) and supernatants were subjected to BCA assays (Pierce, Thermo Fisher Scientific, Waltham, Mass., USA) for protein concentration measurements. 20 μg of total cytosolic protein was subjected to SDS-PAGE and electroblotted onto a nitrocellulose membrane (Whatman Optitran BA-S85, GE healthcare, Little Chalfont, UK). After blocking in Odyssey Blocking buffer (LI-COR biosciences, Licoln, Nebr., USA) in PBS (1:1) the membrane was incubated with mouse-anti-HA (1:500, Sc-7932, Santa cruz, Dallas, Tex., USA), rabbit-anti-GLUD (1:400, GTX105765, GeneTex Inc, San Antonio, Tex., USA), rabbit-anti-IDH1 (1:1000, HPA035248, Sigma Aldrich, St. Louis, Mo., USA), mouse-anti-GAPDH (1:10,000, ab8245, Abcam, Cambridge, US), rabbit-anti-HK2 (1:1000, 2867S, Cell Signaling Technology, Danvers, Mass., USA), rabbit-anti-PKM2 (1:1000, D78A4, Cell Signaling Technology, Danvers, Mass., USA) and goat-anti-γtubulin (1:5000, sc-7396, Santa Cruz, Dallas, Tex., USA) in blocking buffer, followed by incubation with goat-anti-mouseDyLight800 (1:10.000, Thermo Fisher Scientific, Waltham, Mass., USA), goat-anti-rabbitAlexa680 (1:10.000, Invitrogen, Waltham, Mass., USA), or donkey-anti-goatAlexa680 (1:10,000, Invitrogen, Waltham, Mass., USA) in blocking buffer. After washing, blots were analyzed on the Odyssey scanner (LI-COR biotechnology, Lincoln, Nebr., USA). Signals were corrected for γ-tubulin and the mean of 3 independent experiments is plotted. Statistical significance was determined with an unpaired Student's T-test.
Cellular protein content was determined as a measure of cell proliferation, using suforhodamine B (SRB) assays as described in (83). Sensitivity to EGCG was determined by adding a concentration range of EGCG (0-50 μM) or DMSO solvent one day after seeding 1,000 cells per well in 96-wells plates (Nunc, Roskilde, Denmark). An SRB assay was performed after 3 days, and IC50 values were determined in GraphPad Prism using the sigmoidal dose response with variable slope nonlinear regression analysis.
Furthermore cell proliferation over 8 days in presence or absence of EGCG was determined. Cells were seeded at 1,000 cells per well and at day 1 and day 5 after seeding the medium was changed for medium with or without 10 μM EGCG. Controls were incubated with DMSO. Protein content was determined every 2 days. Experiments were performed in triplicate and statistical significance was determined using one-way ANOVA with bonferroni correction.
To determine the sensitivity of the cells to glutamine or glucose deprivation, the regular medium was changed for either D-glucose depleted (0 g/L D-glucose and 4 g/L L-glutamine), L-glutamine depleted (1 g/L L-glutamine and 5 g/L D-glucose) or regular RPMI medium supplemented with 10% FCS and antibiotics with or without 10 μM EGCG one day after seeding the cells. Again protein content was determined every 2 days. Experiments were performed in triplicate and statistical significance was determined using one-way ANOVA with bonferroni correction.
Cells were grown till 80% confluency in culture flasks, and after trypsinization 1.5*106 cells were resuspended in culture medium and transferred to the thermostated (37° C.) chamber of an Oxygraph-2k equipped with Datlab recording and analysis software (Oroboros Instruments, Innsbruck, Austria). The basal respiration was measured and then the remaining mitochondrial respiration was inhibited with 2.5 μM complex V inhibitor oligomycin. Then maximal respiration was measured by sequential addition of 0.5 μM mitochondrial uncoupler FCCP. Subsequently 0.5 μM complex I inhibitor Rotenone and 2.5 μM complex III inhibitor Antimycin A were added to completely shut down the electron transport chain. The remaining oxygen consumption is due to non-mitochondrial respiration. Two separate experiments were performed and significance was determined with an unpaired Student's T test.
Since SKRC-7 and SKRC-17 cells are unable to grow as colonies, sensitivity to radiotherapy was analyzed by monitoring cell proliferation with the xCELLigence. This method has been shown to measure effects of radiotherapy that correlate with the conventional clonogenic assays (84). Cells were plated at 1,000 cells per well and left to adhere overnight. Then cells were treated with 10 μM EGCG for 24 hrs, after which they were irradiated with 4 Gy (IR, 3.1Gy/min; XRAD 320 ix, Precision XRT; N. Brandford, Conn., USA). The cell index was measured from the moment of seeding for 200 hrs in real time with intervals of 15 min, fresh medium supplemented with 10 μM EGCG was added every 72 hours. Cell index was normalized to the moment of applying radiotherapy, and cell growth was calculated by performing linear regression in GraphPad Prism. The experiment was performed with two internal duplicates, and statistical significance was determined with an unpaired Student's T test.
SKRC7 and SKRC17 present with different expression profiles (
The melanoma cell line Mel57 expresses low levels of vascular endothelial growth factor (VEGF-A). VEGF-A consists of different splice variants, consisting of exons 1-5,8 (VEGF-A121), exons 1-5,7,8 (VEGF-165) and exons 1-8 (VEGF-189). These variants have differential activities, ranging from vessel dilatation to full neo-angiogenesis (85). We designed smMIP165 that has its ligation and extension probes in exon 5 and 7, smMIP121 that has its ligation and extension probes in exon 5 and 8, and smMIP189, that has its ligation and extension probes in exon 5 and 6. We performed the smMIP assay with a panel including smMIP121, smMIP165, smMIP189 and 5 smMIPs located in exons 1-5, recognizing all isoforms of VEGF on RNA isolated from the Mel57 cell line and from cell lines expressing the different VEGF isoforms, as described in (85).
Cancer cells can induce changes in RNA splicing events if these give the cells a growth advantage. These changes may be an inherent characteristic of a cancer, but may also be selected under pressure of treatment. An example is EGFR variant III that results from an intragenic deletion in the EGFR gene that results in loss of exons 2 to 7 in the mature transcript. Whereas 50% of glioblastomas are characterized by amplification of the EGFR oncogene, in 50% of this population expression of EGFRvIII is found. By placing extension and ligation probes of an individual smMIP in exons 1 and 8, respectively, only the exon 1-8 fusion product is detected, because the backbone sequence of 40 nucleotides is physically not able to bridge exons 2-7 in the wild-type transcript (
Another example is androgen receptor in prostate cancer. Patients with castration-resistant prostate cancer are treated with enzalutamide. However, a change in splicing that results in loss of exon 7 (ARv7) results in resistance to enzalutamide. By designing a smMIP with extension and ligation probe arms in exons 6 and 8, respectively, Arv7 is readily detected in cell lines derived from enzalutamide—resistant cancers, while it is not detected in any other cancer type (FPM=277 and 640 in VCAP and DuCAP prostate cancer lines, respectively vs mean FPM=0.01 in 130 other cancers).
A sample of brain tumor tissue, obtained from patient N16-10 who signed informed consent for the study, was snap-frozen directly after surgery. RNA was isolated from the tissue via the Trizol protocol, followed by cDNA synthesis and preparation of the smMIP library. After Illumina next generation sequencing a mutation in the IDH1 gene was identified that corresponds to the hotspot IDH1R132H mutation. The same analysis revealed low levels of carbonic anhydrase 9 and hexokinase 2, indicating lack of hypoxic responses. Furthermore the sample shows low ratios of glutaminase/glutamate dehydrogenase, suggesting that the tumor was using glutamate for metabolism, and therefore suggesting sensitivity to glutamate dehydrogenase inhibitors such as EGCG and chloroquine. Furthermore, the data show high expression levels of TrkB and, to a lesser extent, PDGFRα. The absence of hypoxia suggests that the tumor was not of a World Health Organization guidelines-defined grade IV type. The presence of high levels of tyrosine kinases suggests astrocytoma, and based on the data a diagnosis of IDH1-mutated grade III astrocytoma was made, concordant with the original diagnosis that was set on histopathology.
The data of example 4 were expanded with 74 additional samples of brain tumor tissues, obtained from patients who all signed informed consent for the study. The samples were snap-frozen directly after surgery and treated similarly to what has been described in example 4: RNA was isolated from the tissues via the Trizol protocol, followed by cDNA synthesis, preparation of the smMIP libraries and barcoded PCR. After Illumina next generation sequencing FASTQ files were processed by SeqNext software (JSI SequencePilot version 4.2.2 build 502 [JSI Medical Systems, Ettenheim, Germany]). All reads were mapped against the human genome (version hg19) and against manually added variant transcripts (e.g. EGFRvIII, Arv7, METd7/8, METd14). Thus, for every tumor sample a list of targeted transcript levels was generated and a list of all detected mutations/variations. From all 75 patients fully documented clinical follow-up was available.
In a first step, we performed unsupervised agglomerative clustering of log-transformed expression levels of the targeted genes of interest. Agglomerative clustering was performed according to Ward's method by calculating Manhattan distance between individual profiles using bio-informatic R-software scripts. The profiles were translated in a heat map which is represented in
In a next step, potential associations of the clusters with overall survival was investigated by now including survival data for the patients (overall survival, counted from first diagnosis) and generating a Kaplan-Meyer curve. The results in
In a third step, associations between groups and mutations were analyzed by including the list of all detected mutations in a sample. Groups A and B were distinguished by mutations in the isocitrate dehydrogenase genes (IDH1 R132 and IDH2R172) with high significance (p<10E-11).
In a fourth step we performed a subgroup analysis to further refine prognosis. Analyzing IDH-wild-type patients with very poor survival (OS<12 months) versus IDH-wild-type patients with better prognosis (group B in the Kaplan-Meyer curves) in such subgroup analysis showed that high expression levels of carbonic anhydrase 12 are associated with extremely poor prognosis (p<0.001; Fisher's exact test,
In a fifth step we retrospectively analyzed all data with respect to molecular information that was obtained during routine patient care. All mutations that we observed on the RNA level and that are routinely tested for in glioma patient care, were confirmed with DNA sequencing technology (Table III)
The profiles also reveal expression of genes in brain tumors that are associated with other cancers. An example is the androgen receptor that is often expressed at high levels in prostate carcinoma, and prostate specific membrane protein (PSMA). Other groups have described expression of this target on blood vessels in malignant tumors, including glioma (87). To investigate this further we analyzed tumors with high and low PSMA transcript levels using immunhistochemisty. Results indeed revealed blood vessel expression of PSMA protein in blood vessels from tumors with high transcript levels, and not in tumors with low transcript levels (see three examples in
Detailed analysis of expression of metabolic genes revealed that in the group of long survivors low levels of glucose importers, carbonic anhydrase and hexokinase 2 were expressed, indicating lack of hypoxia. Furthermore in this group high levels of glutamate dehydrogenase and aminobutyric acid aminotranferase (ABAT) RNA were observed, suggesting that these tumors use neurotransmitters (glutamate and GABA) for their catabolism, inducing sensitivity to glutamate dehydrogenase- and ABAT inhibitors (epigallocatechin-3-gallate, vigabatrine).
In the group of IDH-mutated gliomas high expression levels of TrkB (mean FPM 15833 vs 4708 in IDHmut vs IDHwt, p=2×10E-11) were detected, suggesting sensitivity to the TrkB inhibitor entrectinib. EGFR expression was higher in IDHwt tumor (609 vs 116 in IDHwt vs IDHmut cancers, p=0.004), whereas EGFRvIII was exclusively expressed in this group (201 vs 0.16 in IDHwt vs IDHmut, p=0.0005, see also
Whereas in the group of IDH wild type cancers EGFR/EGFRvIII expression was observed in 52% of samples, the tyrosine kinase MET is observed in 9/75 brain tumors (12%). Of interest, profiles showed that tumors that expressed relatively high levels of MET, were low in EGFR and vice versa (correlation coefficient r=−0.95). An interesting further finding was occurrence of a mutation in BRAF (BRAF-V600E) in one glioma. Wild-type BRAF is a crucial signaling intermediate that processes signals from activated membrane tyrosine kinases (e.g. EGFR, MET) to the nucleus, thereby inducing cell proliferation. BRAF-V600E is an auto-active molecule that signals to the nucleus without input from receptortyroinse kinases in an uncontrolled fashion. This BRAF mutation occurs in 50% of melanoma cancers and can be inhibited by the targeted drug vemurafenib. The glioma containing this mutation did not express MET nor EGFR. Although anecdotal, this case suggests susceptibility to vemurafenib and unsusceptibility to EGFR or MET inhibitors.
The astrocytoma cell line E98 carries an auto-activating mutation in c-MET (22) and is highly sensitive to the multispecific VEGFR2/MET kinase inhibitor cabozantinib (86) and the MET-selective inhibitor Compound A (88). This sensitivity is reflected in decreased MET phosphorylation on western blot and decreased proliferation rates in vitro and delayed tumor development in vivo. The renal cancer cell line SKRC17 expresses similar levels of MET, phosphorylation of which is effectively inhibited by Compound A (
The Netherlands is one of the first countries implementing detection of high risk human papilloma viruses (hrHPV) in cervical swabs in a population-wide screening program, to allow early detection and preventive treatment of cervical cancers. The life-time risk of an HPV infection is 80%, and in the group of participating women 8% will test positive in this assay. It is known on forehand that 90% of these women are overtreated because HPV infections may resolve spontaneously. Furthermore, sex with an HPV positive partner will result in positivity in the sensitive PCR-based HPV DNA detection tests, but does not mean that the virus will actually infects cervical epithelials cells.
To discriminate between contamination and actual cellular infection, we designed smMIPs for detection of hrHPV transcripts E2, E6 and E7, based on available knowledge that loss of E2 gene expression is associated with chromosomal integration in infected cells and overexpression of the HPV-E6 and E7 oncoproteins. With the entire panel of smMIP probes, supplemented with hrHPV-detecting smMIP probes, we profiled a series of 29 gynecological tissues, ranging from normal uterus extirpations to ovarian cancer, endometrial cancers and cervix carcinomas (
In a second step we investigated the sensitivity of HPV-testing by profiling 1, 10, 100, 1000, and 10,000 Hela cells, derived 490 years ago from a woman with an HPV-18 positive cervix carcinoma. Profiling of only 1 cell already detected 69 unique HPV18 reads, increasing to 168, 1419, 36767 reads in 10, 100 and 1000 cells respectively.
Tumor cells can follow distinct evolutionary paths to become resistant to epidermal growth factor receptor inhibition. Nat Med. 2016; 22(3):262-9.
Oncogene. 2008; 27(41):5527-41.
Development of the tumor vascular bed in response to hypoxia-induced VEGF-A differs from that in tumors with constitutive VEGF-A expression. Int J Cancer. 2006; 119(9):2054-62.
FEBS Lett. 2014; 588(16):2704-11.
Nat Cell Biol. 2015; 17(12):1556-68.
IDH1 R132H mutation generates a distinct phospholipid metabolite profile in glioma. Cancer Res. 2014; 74(17):4898-907.
Pharmacol Rep. 2016; 68(5):918-25.
| Number | Date | Country | Kind |
|---|---|---|---|
| 17159630.7 | Mar 2017 | EP | regional |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2018/055548 | 3/7/2018 | WO | 00 |
