Metabolic Analysis of Tumours

Enrichment of Genomics with Clinically Informative Molecular Phenotyping

Cancer is one of the most devastating human diseases, and it causes an enormous number of mortalities worldwide every year. Over past decades, there has been many astounding breakthroughs and contributions made in this field. In terms of taking a clinical decision in oncology, genomics-based therapeutic targeting represents the dominant approach. Worldwide, different cancer centers are working on sequencing a whole genome of tumours to identify mutations which inform the clinician on potential drug targets. However, this approach to personalizing cancer care is limited. Nevertheless, Multiple processes modify genetic information such as epigenetic events, transcription of various splice variants, expression of non-coding RNA and miRNA, and post-translational modifications of proteins. Therefore, genotype cannot consider as an accurate reflection of tumour phenotype. Drug sensitivity is one of the example of tumour phenotype. To make predictions of biological behaviour and drug susceptibility entirely based on genotype may not work effectively because of its constantly modifying nature. Conversely, it is well established that cancer metabolism differs from the normal metabolism and an important postulation published in 1920’s by Otto Warburg. The statement was cancer cells utilize anaerobic metabolism as a primary source of energy under physiological oxygen levels. Therefore, central cancer metabolism has been studied comprehensively, and an enormous range of metabolic processes or alterations found in cancer. Thus, additional information derived from metabolic profile of cancer patients might be useful to take a therapeutic decision. Metabolomics is relatively fast and accurate technique that can be used with either a specific application or in a global manner to reveal novel information about biological systems. My overall goal is to develop the capability to identify what drug(s) a tumour is susceptible to, based on the molecular features of the tumour. My hypothesis is that information related to the metabolomic profile of the tumour enriches the information that is available from the genomic sequence and transcriptional profiler because the metabolome is the closest molecular representation of phenotype. My project involves the creation of a workflow to define the metabolomic features of a tumour and to integrate the data from this metabolomic analysis with data derived from the whole genome sequence and the transcriptome. In the future, this work will allow testing the hypothesis formally.

Background

 

Cancer can describe as a large group of disease and result of the cellular malfunction. Hundred types of cancer have identified so far, and the characterization of these different types of cancer distinguish by the origin of tissues. For example, cancer that occurs in the epithelial tissues are known as Carcinomas; Adenocarcinomas are known to occur in glandular tissues [1]. As cancer cells have the potential to invade and spread different parts of the body, it is essential to differentiate between a benign tumour and malignant tumour. Though benign tumour is not life-threatening, it could be dangerous due to their location such as a brain tumour. Cancer primarily considered a genetic disease as evidence suggests that most of the cancer agents cause alterations to the DNA sequence. These DNA alterations can occur in a single base pair to large chromosomal variations and accumulation of various mutations over the time shows a multi-step process which is the key to Carcinogenesis [2]. There are different types of cancer exhibit different mutational signatures. Over the years, scientists revealed that some specific genes are mutated in cancer cells frequently than others, for example – RAS, HER2, Tp53. In 2013, Vogelstein et al. published an article where it stated that melanomas and lung tumours contain more than 200 nonsynonymous mutations per tumours [3]. As we know that cancer is complicated, but one of the significant complexity that makes it very challenging to treat patients is heterogeneity. Heterogeneity refers to the state where tumour cells exhibit different molecular characteristics for example – different gene expression, metabolism. Although heterogeneity is a broad term in tumours, it can describe in two simple terms. First, inter-tumour heterogeneity which refers to differences between tumours and second is intra-tumour heterogeneity which is within the tumour. Studies show that different types of heterogeneity can occur from both genetic and non-genetic variability. Example of genetic heterogeneity is impaired DNA mechanism which leads to considerable replication errors as well as defects in mitosis that can cause gain or loss of chromosomes [4].

It is a pervasive question in cancer that what are the risk factors or what causes this deadly disease. Due to the complex nature, it is tough to answer this simple question. However, some specific types of cancer have a single known cause. Some of the known causes are age, smoking, alcohol, different environmental factors (chemical explosion), different types of infections, etc. There are some cancer cases has been filed where patients do not have these risk factors. Finally, in 2000, Hanahan and Weinberg described six hallmarks of cancer (applicable to most of the cancer perhaps not all). After a year, they proposed two other emerging hallmarks and their characteristics. After an excellent analysis of different oncogenes and tumour suppressors, it has been suggested by the researchers that all these factors play a vital role in cellular metabolism. In 1920’s Otto Warburg proposed an interesting statement where he stated that cancer cells showed a distinct metabolic phenotype by consuming an enormous amount of glucose compared to the healthy cells. Afterward cancer metabolism has studied broadly, and researchers performed different types of experiments to understand the metabolism and effect of altered metabolic pathways in cancer patients. With the advanced technology known as “Metabolomics” reveals cancer as more likely to be a metabolic disorder. Similar to oncogenes, researchers found oncometabolites. Oncometabolite refers to the term where an accumulation of small molecules or metabolites has the potential to initiate or sustain tumour growth and metastasis [5]. 2-hydroxyglutarate which is a result of the reduction in -ketoglutarate, was the first oncometabolite that discovered. [6]. Afterward, several oncometabolites has been identified in different cancers for example – succinate, fumarate, sarcosine, etc. There is some controversial statement available for some of the oncometabolites and researchers are still working on that to figure it out. However, the key point to focus here is that most of these oncometabolites are the result of or needed for aerobic glycolysis, glutaminolysis or one-carbon metabolism [5]. Association of metabolic reprogramming with tumourigenesis and constitutive cell proliferation suggests that inhibition of altered metabolic function might impede the tumour progression. Thus it is necessary to do a systemic characterization of metabolism.

Figure 1: Diagram of different hallmarks of cancer.

In 1923, German scientist Otto Heinrich Warburg and Seigo Minami first observed changes in tumour metabolism [7]. They used rat liver carcinoma and healthy liver tissue to measure the lactic acid production and oxygen consumption. Surprisingly they observed an increased amount of lactate production in the presence of oxygen in tumour tissue while the healthy tissue maintained the Pasteur effect [8]. Pasteur effect refers to the condition where lactate production inhibited in the presence of oxygen. Afterward, Warburg and his colleagues tested kidney and heart tissue where they found the same higher amount of lactate production (nearly 70-fold higher than normal liver tissue). Thus, Warburg proposed that lactate production did not dependant on oxygen, and cancer cells used glycolysis followed by lactic acid fermentation to produce high rates of energy in the presence of oxygen. As it was first observed and described by Otto Warburg, this process is known as “Warburg Effect.”

Since the Warburg effect discovered, researchers performed an enormous amount of experiments to understand the mechanism of responsible factors that contribute to cancer by altering the metabolism. Almost all critical oncogenes and tumour suppressor genes have the potential to affect the metabolic pathways. However, it is yet unknown or part of ongoing research that by which molecular mechanisms cancer metabolic phenotype is accomplished via oncogenes and tumour suppressors [9]. In 2016, Ralph et al. published an article explaining how and why metabolic reprogramming occurs in tumour cells as well as how the mechanism is related to altered metabolism that results in tumorigenesis and metastasis. Some of the breakthrough discoveries after Warburg effect or similar to it and implication to cancer therapeutics described briefly here.

Uncontrolled tissue

                                                     

Glucose                                Pyruvate                                          Lactate

                                                                                                 O₂

Figure 2 : Schematic representation of Warburg effect in cancer cells or proliferative tissues.

Glycolysis is a process where glucose breaks down to pyruvate and continue producing energy. Besides glycolysis, there is another alternative pathway available known as pentose phosphate pathway (PPP) by which sugar is produced that make up DNA and RNA. However, no  ATP produced or used up in this process. The first step in glycolysis is glucose transformed with an additional phosphate group and form glucose-6-phosphate. This glucose-6-phosphate enter the PPP, and there is two-phase present in the PPP; Oxidative phase and non-oxidative phase [10]. When a molecule loses a minimum of one electron, the process is known as oxidation. Thus, in the oxidative phase of PPP glucose-6-phosphate gets oxidized and form a new 5-carbon molecule, ribulose-5-phosphate. This ribulose-5-phosphate act as a precursor to the sugars that makes up DNA and RNA. In the non-oxidative phase, ribulose-5-phosphate converts to two different molecules – ribose-5-phosphate and xylulose-5-phosphate. These molecules get converted into glyceraldehyde-3-phosphate and fructose-6-phosphate which are the product of glycolysis [11].

One of the most common and popular tumour suppressors is p53. Evidence suggests that p53 inhibits the gene expression by binding to the promoter region of GLUT1 and GLUT4 [12]. GLUT1 and GLUT4 are the glucose transporter genes. Different studies also show that p53 can inhibit the oxidative phase of PPP indirectly by hindering the expression of the glycolytic enzyme [13]. Thus, p53 plays a vital role in the cancer metabolism as losing of this tumour suppressor initiate glucose uptake in cancer cells resulting in glycolysis and PPP. In 2012, exciting findings published by Ying et al. that – in pancreatic ductal adenocarcinoma, KRAS (oncogene) enhance glycolysis and drives the intermediate of glycolysis into the non-oxidative phase of PPP. Inactivation of KRAS decreases expression of the different glycolytic enzyme, glycolysis activity and reduction of the non-oxidative phase of PPP [14].

PI3K/Akt is one of the critical signaling pathways. Disruption in this pathway is related to different types of cancer. One of the well-known oncoprotein, phosphoinositide 3- kinase (PI3K) can control glucose consumption using PI3K/Akt signaling pathway [15]. Thus, it can increase the expression of glucose transporters and phosphofructose kinase activity (supports glycolysis). Evidence suggests that in most of the cancers, a series of cellular proteins are phosphorylated by Akt which is activated by the mammalian target of rapamycin complex 2 (mTOR2) [16]. Thus it leads to the cell survival.

Figure 3 : An overview of PPP in cancer cells.

One of the critical regulators of cancer cell proliferation is the hypoxia-inducible factor (HIF). These are the transcription factors that respond to the limited presence of oxygen by forming new blood vessels. HIF has a structure of helix-loop-helix which is made of a heterodimeric complex of two subunits –  and  [17]. Subunit  contains three isoforms – HIF-1, HIF-2 and HIF-3. Many research has been carried out to understand the mechanism of these transcription factors, yet the mechanism is not fully understood. HIF-1 degradation is accelerated by oxygen and -ketoglutarate which are the intermediates of the tricarboxylic acid cycle (TCA). Studies show that in different cancer cell line, HIF activates as a result of cellular mutations. Thus these findings suggest that either different tumour suppressors become inactivated or different oncogenes become activated in different growth factor pathways [18]. As mentioned earlier the primary function of GLUT transporters is to facilitate the entry of glucose into the cell. HIF regulates the up-regulation of glucose transporters in cancer cells. Different studies reveal HIF-1 induced increased activity of different glycolytic enzymes such as hexokinase I & II, phosphofructokinase-L, lactate dehydrogenase-A, etc. [19].

Furthermore, HIF stimulates a transcription factor named pyruvate dehydrogenase kinase 1 which potentially blocks the production of 2-carbon unit acetyl CoA. Acetyl CoA is essential for fatty acid synthesis. In 2014 Rabinowitz et al. found an exciting source of acetyl CoA production in different cancer cell lines in hypoxic condition and that source was acetate (500M U-¹³C acetate caused 50%-86% increase in acetyl CoA labeling) [20]. Another study demonstrates that under the hypoxic stress, mitochondrial enzyme SHMT2 is formed. SHMT2 is crucial for sustaining the production of NADPH and redox balance to support the tumour cell survival and growth [21]. With the advanced technology, it is now possible to determine how cells use different metabolic pathway to produce energy particularly cancer cells as they pose a diverse activity to adapt to the microenvironment to survive. For example – a recent study shows that in normal condition NADPH is produced mostly by the malic enzyme in mouse adipocytes whereas hypoxia perturbs the metabolic pathway and switch the primary NADPH source to the oxidative phosphorylation [22,23]. There are numerous amount of experiment reveals that HIF-1 plays a vital role in cancer metabolism as it is a master regulator of various proteins and enzymes associated with energy metabolism.

__

Figure 4: Role of HIF- 1 in different metabolic pathways.

Glutaminolysis refers to the term where glutamine converts to glutamate and some other molecules. Glutamine is one of the most abundant metabolites that present in the blood. In normal cells, glutamine converts to glutamate and glutaminase enzyme catalyze the reaction. Furthermore, glutamate dehydrogenase converts glutamate into -ketoglutarate which is an intermediate of TCA cycle. In tumour cells, conversion of glutamate to -ketoglutarate occurs through glutamate oxaloacetate transaminase as a result of low glutamate dehydrogenase. Evidence suggests that different oncogenic alterations in cancer cells reprogram glutamine metabolism for example – c-Myc binding to glutamine importers results in higher glutamine uptake [24]. Tumour suppressors have potential ability to regulate glutamine metabolism such as retinoblastoma protein and liver kinase B1 suppress the expression of ASCT2 by E2F transcription factor 3 and result in higher glutamine uptake [25]. Numerous amount of experiments refers that glutamine driven oxidative phosphorylation is a significant means of ATP production in hypoxic cancer cells.

Furthermore, glutamine serves as a nitrogen donor. In different types of cancer such as ovarian, breast, prostate – overexpression of GLS correlates with poor prognosis. Thus, demonstrates donation of nitrogen by glutamine in cancer. Another enzyme asparagine synthetase transfers amide group from glutamine to aspartic acid to form asparagine, and it is well known that asparagine is an essential metabolite for the survival of leukemic cells [26]. Not only nitrogen also glutamine serves as a carbon donor. Evidence suggests that under hypoxic condition glutamine delivers the carbon for citrate and fatty acid synthesis due to increased reductive carboxylation [10]. Glutamine can shuffle the intracellular signaling to support tumour growth as well as hinder apoptosis to induce drug resistance [27].

Figure adapted from – Yang, L., Venneti, S. & Nagrath, D. Glutaminolysis: A Hallmark of Cancer Metabolism. Annual Review of Biomedical Engineering 19, 163-194 (2017) [25].

Figure 5: Glutamine anaplerosis in the TCA cycle.

Mitochondria are essential for maintaining different physiological functions of cells thus it is called the powerhouse of cells. In the presence of oxygen (normoxic condition), cells depend on aerobic respiration; specific transporters help glucose to enter the cell and get oxidized to pyruvate in the cytosol. Pyruvate enters into mitochondria to get solely oxidized via TCA cycle (tricarboxylic acid) and beta-oxidation where the final products are guanosine-5′-triphosphate (GTP) or adenosine triphosphate (ATP), the reduced form of nicotinamide adenine dinucleotide (NADH), fumarate ubiquinol and carbon dioxide molecules [28]. Researchers have found that mutations in mitochondrial DNA and dysregulation of mitochondrial metabolism play a vital role in tumourigenesis, but there is limited knowledge about the role of mitochondrial dynamics in cancer metastasis (metastasis- is one of the hallmarks of cancer and the most dominant cause of cancer death). Though the primary role of mitochondria is to generate ATP (energy) following the metabolism of pyruvate derived from glycolysis but they are also responsible for producing reactive oxygen species (ROS), and ROS contributes to cell death and proliferation [29]. In carcinogenesis, the role of oxidative stress is known as a result of mitochondrial DNA mutations or due to altered mitochondrial function [30]. One of the major causes of activation of apoptosis is reactive oxygen species (ROS); because of decreased mitochondrial activities or increased metabolic activities, different types of cancer are fundamentally related with prominent ROS generation as well as fragmentation of mitochondria [31,32,33].

Figure 6 : Schematic diagram of mitochondrial function in cell.

Likewise, mitochondria has the potential to communicate with altered or changed metabolic states to the nucleus via retrograde signalling utilizing different mediators such as Ca²⁺ or calcineurin (Cn) signaling. Various oncogenic factors for instance AKT/PI3k also upregulated glucose transporters GLUT1/GLUT4 and several glycolytic enzymes can activate by the Cn signaling. Resulting shift cell metabolism to glycolysis [34]. In 2017, Peeters,K. et al. found that fructose-1,6-bisphosphate activates RAS during high glycolytic activity [35]. As a result, RAS can stimulate glycolysis by forming a positive feedback loop [34,35]. Evidence also suggest that KRAS mutation occurs under metabolic stress or glucose deprivation and these conditions cause upregulation of glucose transporters, thus supporting the survival of cancer cells [35,36]. In the field of oncology, among all the metabolic pathways in mitochondria, TCA has been the spotlight. From all the mitochondrial experiments of cancer it is well established that several enzymes of the TCA get mutated in the sporadic and hereditary forms of cancer [37]. In human cancers, mutation in isocitrate dehydrogenase-2, succinate dehydrogenase and fumarate dehydratase are commonly found [38]. Studies reveal that different oncometabolite such as 2-hydroxyglutarate (2-HG) driven alterations in epigenome consequences suppression in differentiation as well as promoting cell proliferation [37,38,39]. Moreover, 2-hydroxuglutaret plays a critical role in the degradation of hypoxic inducible factors by inhibiting the prolyl hydroxylases. Thus, it is well-defined that mutation in enzymes that are involved in mitochondrial TCA cycle contribute to cancer by yielding oncometabolites.

Another important fact is that during starvation, cells can support metabolism through clearing the damaged protein and organelles by themselves and this is known as autophagy. Since autophagy is a relatively new term, in cancer, the role of autophagy is debatable. Some evidence shows it helps cancer cell to proliferate and survive, on the contrary it helps cell proliferation to stop by acting as tumour suppressor. Some of the events such as oxidative stress, chromatin instability, DNA damage etc. are provoked due to deficiencies in autophagy which leads to the accumulation of damaged macromolecules and organelles. Role of autophagy as tumour suppressor found mostly in the studies involved Bcl2-interacting protein, Beclin1 such as deletion of Beclin1 has frequently found in breast, ovarian, prostate cancer [40,41,42,43] also alteration in ATG6 gene found in experimental mice with lymphoma, lung and liver cancer [40,44].

Metabolic targeting for cancer therapy is an emerging field which is currently under investigation to identify the small molecules which might specifically block the fundamental metabolic steps linked to tumour growth. As glycolysis plays an essential role in the development of cancer, attenuation or inhibition of glycolysis may be useful for the prevention of abnormal cell proliferation, and invasion, as well as metastasis. Several enzymes are the primary drivers of glycolysis such as HK, OFK, and pyruvate kinase. Hexokinase is a potential target for cancer metabolism as several types of cancer exhibit high levels of HK II [45]. For example- in lung, breast or brain cancers, the deletion of HKII is beneficial. Phosphofructokinase 1 found increased in many types of cancer. There are some clinical trials currently underway with small molecule PFKFB3 inhibitors. Another critical step is at which glycolysis derived pyruvate can either can be imported into the mitochondria to be either oxidized in the TCA cycle or converted to lactate in the cytosol. In the mitochondria, pyruvate converts to Acetyl CoA and the enzyme responsible for regulating this crucial junction in pyruvate metabolism are the pyruvate dehydrogenase complex. Pyruvate dehydrogenase kinase performs inhibitory phosphorylation which reduces the activity of PDH resulting decrease in pyruvate flux and enhanced lactate production. Expression of PDHK in various types of cancer makes it a potential therapeutic target [46]. Lactate dehydrogenase complex also plays a crucial role in regulating the fate of pyruvate in cancer. Different cell line experiments reveal that the inhibition of IDHA by small molecule inhibitors or genetic approaches results in slowed cancer cell growth and increased cell death in a variety of cancer settings including hepatocellular carcinoma and breast cancer [47]. Several early-stage clinical trials are under the process to assess the efficacy of some LDH inhibitors. The pre-clinical development of inhibitors that have more specificity for LDHa is currently ongoing [48]. An enormous amount of research is going on to explore the inhibition of glutamine transporters of cancer cells. Another promising area of research is to inhibit the GDH to limit the glutamine production to enter to the TCA cycle via -ketoglutarate. Currently, there are no small molecules available to block the production of GDH, but the initiative to develop some specific GDH inhibitors may allow more effective targeting of glutamine flux into the TCA cycle.

Targeting small metabolites present in the TCA has become one of the greatest successes to date in therapies attacking cancer metabolism. Recently there has been a success in preclinical and clinical settings with novel compounds that inhibit the gain of function activity of mutant IDH. This mutant IDH was shown to dramatically reduce the production of 2HG and cause cancerous cells to differentiate towards a normal phenotype [49]. There is an early phase trial, ongoing with the small inhibitor of mutant IDH2, AG-221.

Aims and Hypothesis

Each year a significant proportion of patients are diagnosed with cancer which may or may not develop into an aggressive phenotype. Many patients undergo invasive procedures for diagnosis and prophylactic treatment that may be unwarranted. Thus, there is a need for new non-invasive methods that allow for better screening strategies as well as new biomarkers that can distinguish between indolent and aggressive metastatic forms of cancers. The consistent appearance of metabolic changes suggests that there may be a connection between metabolism and clinical progression. My research focuses on finding molecular characteristics that will improve the selection of cancer treatment. Overall goal of this research is to develop a new metabolomics-based method that will enhance the genomic-based methods of identifying effective therapeutic targets in human cancers. Metabolites are the end result of cellular processes, and their levels can consider as an essential response of biological systems to genetic or environmental changes which leads to my primary hypothesis that the metabolomic profile of tumours is an accurate molecular representation of tumour phenotype. Thus, the metabolic profile can provide crucial supplementary information to genomic, transcriptomic and proteomic information. The primary challenges in addressing this question are the need to comprehensively analyze metabolism of tumour tissues and my need to associate large numbers of clinical outcome data with specific metabolic phenomena. To accomplish the goal, I will utilize two of the most exceptional facilities in Calgary of their field. One of them is the University of Calgary Hepatobiliary and Gastrointestinal Tumour Bank which houses over 22,000 samples from almost 4,000 patients. The other facility is the Calgary Metabolomics Research Facility (CMRF) which is a state-of-the-art mass spectrometry and computational laboratory explicitly built for unraveling the complex metabolic interactions that occur during human infections or diseases. I will use these resources to conduct the first comprehensive analysis of tumour metabolic determinants of clinical progression.

Specific aims for the research are:

1. Extensive transcriptional analysis of different metabolic subtypes of breast cancer to determine the biological functions of differentially abundant gene expressions.
2. Comprehensive survey of metabolic diversity in a retrospective collection of pancreas cancer patients with high-resolution mass spectrometry.
3. Development of a comprehensive workflow to integrate metabolomic information with information derived from transcriptomic platform of pancreas cancer patients.

Aim 1: Transcriptional analysis of different metabolic subtypes of breast cancer

Over decades many studies have used DNA microarrays to reveal the signature gene expressions of different types of human cancers. Unfortunately, the signatures contain complex or critical features that makes them enigmatic. Likewise, evidence suggests that changes in gene expression cause significant metabolic reprogramming in cells [50]. Though it is not necessary that all the altered metabolic activities contribute equally to cancer. Researches have compared tumour tissue gene expression profiles and normal tissue to find the potential genes in cancer [51]. Different studies show interesting insights such as – comparisons between tumour and normal tissues express numerous numbers of significant genes that are expressed in a different manner in various types of cancer; comparison also exhibits numerous transcriptional dysregulation of different metabolic genes [50,51]. Since gene expression has convincing association between the oncogenic drivers and metabolic phenotypes, it is essential to look into the gene expression profile of different cancer patients to get a better understanding of cancer metabolism. However, distinguishing these significantly expressed genes is challenging for example – how to classify patients based on their significant metabolic gene expressions more precisely is questionable. Furthermore, what could be the best approach to utilize different tumour subtypes clinically is a major part of dilemma [52].

As a part of my research, I am going to do a comprehensive transcriptional analysis of breast cancer patients. The gene expression data set has taken from The Cancer Genome Atlas (TCGA). TCGA provides a multidimensional map of the most crucial genomic alterations in 33 different types of cancer and it is available publicly. From Bathe lab a numerous number of samples has submitted to TCGA in between 2012-2015. All the samples passed the quality checks that includes pathology review, examination of tumour for cellularity and an extensive molecular analysis. The molecular data for each tumour samples contains different copy number variations (CNVs), gene and miRNA expression, promoter methylation and DNA sequence/mutation analysis. For the metabolomic analysis, I am going to use different tumour samples that has submitted to the TCGA. The method and workflow will be developed using data derived from these samples. Method development and creation of workflow will be explained in aim 2 and 3 section. In Bathe lab, different types of breast cancer patients were classified into four different clusters (cluster 1 & 2 – mostly Basal-like, cluster 3 – healthy control and cluster 4 – mixed Basal-like, LumA, LumB, HER2) based on their significant metabolic genes. Hierarchical clustering figure is attached in the preliminary data section. A brief description of the study parameters is given below-

In distinction to TCGA database, 1192 samples from female patients were collected where 1081 were tumour and 111 matched healthy adjacent breast tissue. Out of 14,735 genes 1439 metabolic genes were identified using Reactome Pathway Database. 1375 metabolic genes were acknowledged as significant when comparing cluster 3 (healthy controls) to at least one of the metabolic subtypes 1, 2, and/or 4. In my aim, I am going to characterize these metabolic expression subtypes such as which metabolic pathways are perturbed and how they affect other hallmark functions of cancer as well as identifying master regulators for different metabolic subtypes. Initially, I performed gene set enrichment analysis (GSEA) using the entire gene expression datasets of 1192 patients. GSEA is a functional enrichment analysis method to identify genes that are over represented in a large set of genes and linked with disease phenotypes. Analytical process of GSEA includes- calculation of enrichments score (ES – represents to which degree a gene is over presented at the top or bottom of the list similar); estimation of statistical significance of ES which is done by a phenotypic based permutation test; normalization of ES for each gene set and calculation of False Discovery Rate (FDR). From the analysis it shows that in cluster 1, three gene sets are significantly upregulated at FDR< 25% (Nucleotide, TCA cycle and Amino acid metabolism). On the contrary, in cluster 2, three gene sets are significantly downregulated at FDR<25% (Energy, lipid and Vitamin & Co-factor). And in cluster 4, one gene sets are significantly upregulated and three gene sets downregulated respectively at FDR>25% (Upregulated – Nucleotide; Down-regulated- Energy, Lipid and Vitamin & Co-factor). To determine the biological relevance of different metabolic expression subtypes, I assessed their relation with other cellular pathways. For this analysis I used seven functional cancer hallmarks including DNA repair, Angiogenesis, Apoptosis, G2M checkpoint, Epithelial mesenchymal transition (EMT), Inflammatory response and Invasion & Metastasis. GSEA has performed based on mRNA expression at FDR<0.01. Individually in cluster 1, DNA repair pathway was significantly upregulated whereas in cluster 2, G2M checkpoint; no significant changes observed in cluster 4. GSEA result summary-

• Cluster 1 subtypes has upregulated nucleotide, TCA cycle and amino acid pathways which is related to increased DNA repair and decreased angiogenesis, inflammatory response and invasion & metastasis.
• Cluster 2 subtypes has downregulated energy, lipid and vitamin & cofactor pathway that are associated with increased G2M checkpoint and decreased apoptosis, angiogenesis, EMT, inflammatory response and invasion & metastasis.

Based on GSEA analysis, it clearly illustrates that there is a strong connection present between the metabolic activity and cancer hallmark pathways. To have a better understanding of these result and look deep into the pathways, I am going to perform a broad pathway analysis using one of the advanced bioinformatic tool Ingenuity Pathway Analysis (IPA). Primary functions of this application are that it identifies key regulators and activities that provide a better explanation of differentially expression patterns such as relationships, mechanisms, functions and pathway relevant to changes in assigned dataset. From this analysis I will be able to find the regulators or molecules that are involved in the significant pathways and I will create a list of perturbed regulators associated with each metabolic subtype. This list will be merged later with the perturbed metabolite list of the patients and will be used to create the workflow.

Gene Set Enrichment Analysis

Breast Cancer Samples-1192                                              Breast Cancer Samples-1192

Gene list – 14,735                                                            Gene list – 1375 metabolic gene

Seven Hallmarks of Cancer                                                     Seven Super Metabolic Pathway

1. DNA repair                                                                   1. Amino Acid
2. Epithelial Mesenchymal Transition (EMT)                 2. Carbohydrate
3. Angiogenesis                                                                 3. TCA Cycle
4. Apoptosis                                                                      4. Nucleotide
5. Inflammatory Response                                                5. Energy
6. G2M Checkpoint                                                          6. Lipid
7. Invasion & Metastasis                                                   7. Vitamin &Cofactor

Figure 7: Workflow of Gene Set Enrichment Analysis

Aim 2 : Comprehensive survey of metabolic diversity in different pancreas cancer patients

Since reprogramming of energy metabolism is one of the hallmarks of cancer and my primary hypothesis is that metabolites reflect the tumour phenotype more accurately, I must identify scope and diversity of metabolic phenotypes in clinical populations. In order to perform that task firstly, I will assemble a collection of tumour samples (22 cases of Hepatocellular carcinoma, 1 case of cholangiocarcinoma and 7 cases of pancreatic cancer) and healthy tissue samples (Subcutaneous Adipose Tissue, Skeletal Muscle, and Liver tissue). To execute the metabolic experiment, I am going to use liquid chromatography-mass spectrometry (LC-MS) as it provides a wide range of metabolites. For this experiment I will use two different chromatographic methods of LC-MS, one is Hydrophilic interactions liquid ionization chromatography (HILIC), and another one is Reverse phase ion-pairing (RPIP). However, I am going to work with frozen samples, and I established the most appropriate method for metabolomic analysis of frozen tissues. The method is given below-

Human Tissue Sample Preparation Protocol

1.    Take samples from -80C and keep them in ice always.

2.    Weight 50mg of the tissue sample and put it in a 2ml tube.

3.    Add 200l of 50% cold MeOH and homogenize at least one minute keeping the tube in ice.

4.    Rinse the pestle with 800l (normalized volume 1:20) of 50% cold MeOH.

5.    Vortex for 20s and centrifuge for 2 minutes at room temperate.

6.    Transfer 200l of supernatant into a mass spec vials and store rest of the samples at -80C.

Prior to submitting tumour samples to metabolomic analysis, I will establish some controls. These controls will be used to compare which metabolites are perturbed as different metabolites are expected to be differentially abundant in different tissues under several circumstances. Thus a set of virtual controls will be established to define the range of relative concentrations of metabolites. To determine the relative concentrations, I am going to test ten adipose, ten muscle, and ten liver tissues. Once the relative concentrations of different metabolites detected by the LC-MS will be measured, I will create a pooled sample from each normal tissue, and this will use as a control for the rest of the experiment. Analysis of a tumour with LC-MS will allow me to identify approximately 150 metabolites in aqueous extracts and 100 metabolites in the organic extracts. After identification of different metabolites, I will compare the abundance of individual metabolites to each control. T-test or ANOVA will be used to determine the abundance of the metabolites. Moreover, to evaluate the metabolic differences, multivariate approaches including principal component analysis (PCA) and Orthogonal partial least squares discriminant analysis (OPLS-DA) will be performed. To reveal the perturbed metabolic pathways, I will analyze most abundant metabolites using Ingenuity pathway analysis and Pathview. This analysis will allow me to get an overview of altered biochemical pathways and their functions. Furthermore, KEGG pathways will simplify the identification of perturbed pathways, and Small Molecule Database (SMDB) could be useful for the better dataset. Finally, a Metabolite Set Enrichment Analysis (MSEA) will be performed to identify other dysregulated metabolites that will assist to identify different biomarkers of known conditions.

 

 

 

 

Tumour Sample List- Hepatocellular Carcinoma (HCC; n=22)

Pancreatic Adenocarcinoma (PDAC; n=11)

Metabolic Analysis of Tumours

Figure 8: Workflow of metabolic analysis of different tumour samples.

 

Aim 3: Comprehensive workflow to integrate metabolomic information with information derived from transcriptomic platform

 

In multicellular organisms, every cell contains the same genome thus same genes, but it is not necessary that every gene transcriptionally active in every cell [43]. Different cells exhibit different arrangements of gene expression. In various cells and tissues, these variations emphasize a wide range of physical, biochemical and developmental differences which might play a vital role in the difference between health and disease. Therefore, it is possible to acquire a more in-depth understanding of what constitutes a specific cell type as well as how changes in transcriptional activity may reflect or contribute to the disease.

A transcriptome signifies a small percentage of the genetic code that transcribed into RNA molecules (estimated to be less than 5% of the genome in humans) [44]. In a complex organism, the proportion of transcribed sequences are enormous. Because of alternative splicing, RNA editing or alternative transcription initiation and termination sites, each gene may yield more than one variant of messenger RNA (mRNA). By analyzing transcriptomic data, it might be possible to determine when and where genes are turned on or off in different types of cells or tissues. Likewise, it might be possible to create a broad picture of which genes are active at various stages of development.

The primary purpose of transcriptomic is to identify genes differentially expressed among different conditions which lead to new understanding of genes or pathways associated with the conditions [45]. To explore the association between expression of metabolic enzymes and levels of the metabolites participating in reactions that they catalyze, it is worth trying to analyze metabolomic and transcriptomic profiles cooperatively. As my primary hypothesis is to characterize tumour tissues based on their metabolome and target those abundant small molecules with potential drugs, I will look into the genes and transcripts that primarily function to modulate metabolism such as encoding enzymes involved in a metabolic pathway as well as genes and transcripts that possess secondary metabolic effects e.g. KRAS, P13K.

I will submit tumour samples for metabolomic analysis that previously submitted to whole genome sequencing and RNA-Seq. Tumour samples will be analyzed using the methods developed in Aim 2 when compared to normal tissues; differentially abundant metabolites will be identified. Using Ingenuity Pathway Analysis and other metabolomics specific pathway software, I will determine which biochemical pathways are perturbed. This metabolomics-based information will be enriched using transcriptomic and genomic data, using a workflow already defined in the Bathe lab.

Progress to Date

 

 

 

 

 

 

 

 

 

 

 

 

Significance

 

Chemotherapy is an essential mainstay of treatment. However, it is toxic, and it is not guaranteed whether chemotherapy will efficiently work for any individual. Due to its toxicity and adverse side effects, only a proportion of individuals respond to chemotherapy. Thus, a proportion of patients have to endure the toxicities of chemotherapy without benefits. It is for this reason that substantial effort is being made to individualize treatments for cancer.

With the advanced technology, it is now possible to do whole genome sequencing of a tumour. Therefore, it is possible to identify all of the mutations contained within a tumour. Researchers/ oncologists have utilized this information to predict the types of drugs that might be effective in an individual. For example, a specific mutation present in a gene might cause accelerated glycolysis (use of glucose to generate energy), a drug that inhibits glycolysis may be useful in that patient. Indeed, some notable individual successes accomplished by using this approach. However, this approach is not helpful for all individuals which gives a clear indication that genomic information may not be sufficiently informative to make critical decisions on therapy.

Additional information obtained from the metabolic profile of cancer patients might reveal potential insights into each case. For instance, this would help to identify biochemical processes that are deranged more accurately. Better decisions on the best type of therapy could be derived. This novel approach to enhance clinical decisions using molecular information may alter the way of individualizing cancer care in the future, minimizing drug exposure to patients who will not benefit and also enhancing the likelihood that they will benefit from any particular drug.

Future Directions

1. Demonstrate that metabolomic workflow works to identify drugs: cell lines
2. Devise workflow to integrate highly dimensional genomic, proteomic and metabolomic data
3. Clinical trials

 

 

 

 

 

Preliminary Data

 

 

Figure 9: Hierarchal clustering of metabolic subtypes in breast cancer.

Pathways	NES	Nom p-Value	FDR	Hallmarks	NES	Nom p-Value	FDR
Nucleotide	1.74	0.004	0.012	DNA repair	1.52	0.006	0.156
TCA	1.38	0.158	0.134	G2M Checkpoint	0.73	0.695	1.000
Amino Acid	1.31	0.174	0.139	Apoptosis	0.72	0.943	0.781
Energy	-1.34	0.100	0.449	Angiogenesis	-1.12	0.357	1.000
Lipid	-1.16	0.275	0.553	EMT	-0.92	0.546	0.703
Vitamin & Cofactor	-1.10	0.326	0.472	Inflammatory Response	-0.61	0.871	0.892
Carbohydrate	-0.77	0.900	0.800	Invasion & Metastasis	-0.94	0.549	1.000

Figure 10: Gene Set Enrichment Analysis of Cluster 1 vs Cluster 3 (significantly upregulated pathways and hallmark functions are highlighted in Red).

Pathways	NES	Nom p-Value	FDR	Hallmarks	NES	Nom p-Value	FDR
Nucleotide	1.36	0.061	0.315	DNA repair	1.18	0.241	0.605
TCA	0.79	0.647	1.000	G2M Checkpoint	1.76	0.008	0.032
Amino Acid	0.77	0.774	0.815	Apoptosis	-0.78	0.862	1.000
Energy	-1.39	0.075	0.205	Angiogenesis	0.54	0.959	0.951
Lipid	-1.24	0.158	0.200	EMT	-0.62	0.835	0.901
Vitamin & Cofactor	-1.30	0.075	0.201	Inflammatory Response	1.13	0.335	0.475
Carbohydrate	1.13	0.213	0.603	Invasion & Metastasis	0.95	0.515	0.576

Figure 11: Gene Set Enrichment Analysis of Cluster 2 vs Cluster 3 (significantly upregulated and downregulated hallmark functions and pathways are highlighted in Red and Green respectively).

Pathways	NES	Nom p-Value	FDR	Hallmarks	NES	Nom p-Value	FDR
Nucleotide	1.41	0.054	0.122	DNA repair	1.33	0.114	0.269
TCA	0.67	0.820	0.945	G2M Checkpoint	1.27	0.259	0.174
Amino Acid	-0.87	0.619	0.683	Apoptosis	-0.84	0.720	0.758
Energy	-1.77	0.002	0.012	Angiogenesis	-0.94	0.517	0.786
Lipid	-1.56	0.004	0.049	EMT	-1.04	0.456	1.000
Vitamin & Cofactor	-1.79	0.000	0.022	Inflammatory Response	-0.73	0.729	0.758
Carbohydrate	1.13	0.213	0.603	Invasion & Metastasis	-1.03	0.400	0.939

Figure 12: Gene Set Enrichment Analysis of Cluster 4 vs Cluster 3 (significantly upregulated and downregulated hallmark functions and pathways are highlighted in Red and Green respectively).

References

 

1.Pecorino, L. Molecular biology of cancer. (Oxford Univ. Press, 2016).

2.Stratton, M., Campbell, P. & Futreal, P. The cancer genome. Nature (2009). doi:doi:10.1038/nature07943.

3.Vogelstein, B. et al. Cancer Genome Landscapes. Science 339, 1546-1558 (2013).

4.Alderton, G. Epigenetic and genetic heterogeneity in metastasis. Nature Reviews Cancer17, 141-141 (2017).

5.Wishart, D. Is Cancer a Genetic Disease or a Metabolic Disease?. EBioMedicine 2, 478-479 (2015).

6.Fan, J. et al. Human Phosphoglycerate Dehydrogenase Produces the Oncometabolite d-2-Hydroxyglutarate. ACS Chemical Biology 10, 510-516 (2014).

7.Otto, A. Warburg effect(s)—a biographical sketch of Otto Warburg and his impacts on tumor metabolism. Cancer & Metabolism 4, (2016).

8.Potter, M., Newport, E. & Morten, K. The Warburg effect: 80 years on. Biochemical Society Transactions 44, 1499-1505 (2016).

9.Bensinger, S. & Christofk, H. New aspects of the Warburg effect in cancer cell biology. Seminars in Cell & Developmental Biology 23, 352-361 (2012).

10.Pelley, J. & Pelley, J. Elsevier’s integrated review biochemistry. (Elsevier/Saunders, 2012).

11.Pelley, J. Elsevier’s Integrated Biochemistry. (Mosby, 2007).

12.Schwartzenberg-Bar-Yoseph, F., Armoni, M. & Karnieli, E. The Tumor Suppressor p53 Down-Regulates Glucose TransportersGLUT1andGLUT4Gene Expression. Cancer Research 64, 2627-2633 (2004).

13.Patra, K. & Hay, N. The pentose phosphate pathway and cancer. Trends in Biochemical Sciences 39, 347-354 (2014).

14.Ying, H. et al. Oncogenic Kras Maintains Pancreatic Tumors through Regulation of Anabolic Glucose Metabolism. Cell 149, 656-670 (2012).

15.DeBerardinis, R., Lum, J., Hatzivassiliou, G. & Thompson, C. The Biology of Cancer: Metabolic Reprogramming Fuels Cell Growth and Proliferation. Cell Metabolism 7, 11-20 (2008).

16.Shaw, R. & Cantley, L. Ras, PI(3)K and mTOR signalling controls tumour cell growth. Nature 441, 424-430 (2006).

17.Wu, D., Potluri, N., Lu, J., Kim, Y. & Rastinejad, F. Structural integration in hypoxia-inducible factors. Nature 524, 303-308 (2015).

18.Maxwell, P. Activation of the HIF pathway in cancer. Current Opinion in Genetics & Development 11, 293-299 (2001).

19.Chen, J. & Russo, J. Dysregulation of glucose transport, glycolysis, TCA cycle and glutaminolysis by oncogenes and tumor suppressors in cancer cells. Biochimica et Biophysica Acta (BBA) – Reviews on Cancer 1826, 370-384 (2012).

20.Courtnay, R. et al. Cancer metabolism and the Warburg effect: the role of HIF-1 and PI3K. Molecular Biology Reports 42, 841-851 (2015).

21.Kamphorst, J., Chung, M., Fan, J. & Rabinowitz, J. Quantitative analysis of acetyl-CoA production in hypoxic cancer cells reveals substantial contribution from acetate. Cancer & Metabolism 2, 23 (2014).

22.Ye, J. et al. Serine Catabolism Regulates Mitochondrial Redox Control during Hypoxia. Cancer Discovery 4, 1406-1417 (2014).

23.Liu, L. et al. Malic enzyme tracers reveal hypoxia-induced switch in adipocyte NADPH pathway usage. Nature Chemical Biology 12, 345-352 (2016).

24.Wise, D. et al. Myc regulates a transcriptional program that stimulates mitochondrial glutaminolysis and leads to glutamine addiction. Proceedings of the National Academy of Sciences 105, 18782-18787 (2008).

25.Yang, L., Venneti, S. & Nagrath, D. Glutaminolysis: A Hallmark of Cancer Metabolism. Annual Review of Biomedical Engineering 19, 163-194 (2017).

26.Mullen, A. et al. Oxidation of Alpha-Ketoglutarate Is Required for Reductive Carboxylation in Cancer Cells with Mitochondrial Defects. Cell Reports 7, 1679-1690 (2014).

27.Kalyanaraman, B. Teaching the basics of cancer metabolism: Developing antitumor strategies by exploiting the differences between normal and cancer cell metabolism. Redox Biology 12, 833-842 (2017).

28. Osellame, L., Blacker, T. & Duchen, M. Cellular and molecular mechanisms of mitochondrial function. Best Practice & Research Clinical Endocrinology & Metabolism 26, 711-723 (2012).

29.

30.Chen, E. Mitochondrial dysfunction and cancer metastasis. Journal of Bioenergetics and Biomembranes 44, 619-622 (2012).

31.Ježek, J., Cooper, K. & Strich, R. Reactive Oxygen Species and Mitochondrial Dynamics: The Yin and Yang of Mitochondrial Dysfunction and Cancer Progression. Antioxidants 7, 13 (2018).

32.Rodic, S. & Vincent, M. Reactive oxygen species (ROS) are a key determinant of cancer’s metabolic phenotype. International Journal of Cancer 142, 440-448 (2017).

33.The Interplay between Oncogenic Signaling Networks and Mitochondrial Dynamics. Antioxidants 6, 33 (2017).

34.Jia, D., Park, J., Jung, K., Levine, H. & Kaipparettu, B. Elucidating the Metabolic Plasticity of Cancer: Mitochondrial Reprogramming and Hybrid Metabolic States. Cells 7, 21 (2018).

35.Peeters, K. et al. Fructose-1,6-bisphosphate couples glycolytic flux to activation of Ras. Nature Communications 8, (2017).

36.Yun, J. et al. Glucose Deprivation Contributes to the Development of KRAS Pathway Mutations in Tumor Cells. Science 325, 1555-1559 (2009).

37.Zong, W., Rabinowitz, J. & White, E. Mitochondria and Cancer. CellPress (2018). doi:http://dx.doi.org/10.1016/j.molcel.2016.02.011

38.Gaude, E. & Frezza, C. Defects in mitochondrial metabolism and cancer. Cancer & Metabolism 2, 10 (2014).

39.Parker, S. & Metallo, C. Metabolic consequences of oncogenic IDH mutations. Pharmacology & Therapeutics 152, 54-62 (2015).

40. Choi, K. Autophagy and cancer. EXPERIMENTAL and MOLECULAR MEDICINE 44, 109-120 (2012).

41. Futreal PA , Söderkvist P , Marks JR , Iglehart JD , Cochran C , Barrett JC , Wiseman RW . Detection of frequent allelic loss on proximal chromosome 17q in sporadic breast carcinoma using microsatellite length polymorphisms . Cancer Res 1992 ; 52 : 2624 – 2627

42. Aita, V. et al. Cloning and Genomic Organization of Beclin 1, a Candidate Tumor Suppressor Gene on Chromosome 17q21. Genomics 59, 59-65 (1999).

43. Zuhlke, K. Truncating BRCA1 Mutations Are Uncommon in a Cohort of Hereditary Prostate Cancer Families with Evidence of Linkage to 17q Markers. Clinical Cancer Research 10, 5975-5980 (2004).

44.Qu, X. Promotion of tumorigenesis by heterozygous disruption of the beclin 1 autophagy gene. Journal of Clinical Investigation 112, 1809-1820 (2003).

45.Keijer, J. & van Dartel, D. Reprogrammed Metabolism of Cancer Cells as a Potential Therapeutic Target. Current Pharmaceutical Design 20, 2580-2594 (2014).

46.Dang, C., Hamaker, M., Sun, P., Le, A. & Gao, P. Therapeutic targeting of cancer cell metabolism. Journal of Molecular Medicine 89, 205-212 (2011).

47.Ward, P. et al. The Common Feature of Leukemia-Associated IDH1 and IDH2 Mutations Is a Neomorphic Enzyme Activity Converting α-Ketoglutarate to 2-Hydroxyglutarate. Cancer Cell 17, 225-234 (2010).

48.Miao, P., Sheng, S., Sun, X., Liu, J. & Huang, G. Lactate dehydrogenase a in cancer: A promising target for diagnosis and therapy. IUBMB Life 65, 904-910 (2013).

49.Unterlass, J. et al. Validating and enabling phosphoglycerate dehydrogenase (PHGDH) as a target for fragment-based drug discovery in PHGDH-amplified breast cancer. Oncotarget

50.Vander Heiden, M. & DeBerardinis, R. Understanding the Intersections between Metabolism and Cancer Biology. Cell 168, 657-669 (2017).

51.Rhodes, D. et al. Large-scale meta-analysis of cancer microarray data identifies common transcriptional profiles of neoplastic transformation and progression. Proceedings of the National Academy of Sciences 101, 9309-9314 (2004).

52.Peng, X et al. Molecular Characterization and Clinical Relevance of Metabolic Expression Subtypes in Human Cancers. Cell Reports 23, 255-269 (2018).