CARDIAC HYPERTROPHY IN HUMAN STEM CELLS-DERIVED CARDIOMYOCYTES

(1)

CARDIAC HYPERTROPHY IN HUMAN

STEM CELLS-DERIVED

CARDIOMYOCYTES

Biomarker identification and pathway analysis of endotheline-1 induced cardiac hypertrophy in human-induced pluripotent stem cells-derived cardiomyocytes

Bachelor Degree Project in Bioscience

G2E 30 ECTS

Spring term 2020

Author:

Benyapa Tangruksa

a17benta@student.his.se

Supervisor:

Jane Synnergren

jane.synnergren@his.se

Examiner:

Zelmina Lubovac

(2)

2

Abstract

(3)

(4)

4

Introduction

Cardiac hypertrophy and human stem cell-based model

Cardiac hypertrophy is a condition in which heart muscles thicken as an adaptive response to several stimuli. When stimulated, cardiomyocytes (CMs) increase in size and cause an enlargement of the muscle (Frey, Katus, Olson & Hill, 2004), which can be further categorized into physiological and pathological cardiac hypertrophy. Physiological hypertrophy occurs naturally by exercise (Pluim, Zwinderman, van der Laarse & van der Wall, 2000), pregnancy (Eghbali et al., 2005), or puberty (Janz, Dawson & Mahoney, 2000). It is generally harmless where the hypertrophic response is reversible. On the other hand, prolonged pathological cardiac hypertrophy can lead to heart failure and severe cardiovascular disease. It is associated with stimuli and conditions, such as aortic stenosis (Badiani et al., 2016), gene mutations, chronic hypertension, neurohormonal stimulation, and physical stretching (Marian, 2008). The heart increases in size and is unable to pump blood effectively due to changes in several pathways that affect the function of the heart. The characteristics of the pathological cardiac hypertrophy include an increase in glucose consumption by the cardiomyocytes (Carlson et al., 2013), an increase in natriuretic peptides protein production (Du, Zhang & Liu, 2019), and an aberrant gene expression profile (Frey & Olson, 2003). One limitation in comprehensively studying cardiac hypertrophy is the lack of human cardiomyocytes available for research. Recently, a hypertrophic model using human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) was introduced (Kattman, Koonce, Swanson & Anson, 2010). The stem cells have the unique capability of differentiating into any cell types and self-renewal. Studies have shown several benefits of utilizing hiPSC-CMs since the model allows scientists to perform research in vitro (Carlson et al., 2013 and Lundin et al., 2018) that otherwise would not have been possible.

Endothelin 1 (ET-1) is commonly used to create hormone-induced cardiac hypertrophy in vitro (Carlson et al., 2013). ET-1 acts as a neurohormone that contributes to the thickening of heart muscles by stimulating receptors that couple to the heterotrimeric G-protein. The simulation results in the activation of downstream signaling pathways that lead to an increase in catalytic activities and trigger hypertrophic conditions (Bupha-Intr, Haizlip & Janssen, 2012). The ET-1 stimulation on hiPSC-CMs results in an increase in CMs size and an increase in lactate levels due to higher glucose consumption, as well as increases in A- and B-type natriuretic peptides (Johansson et al., 2019), which are characteristics of pathological cardiac hypertrophy. In Johansson et al. (2019), the concentrations and culturing time affect the gene expression of the hiPSC-CMs model. The study also found a large number of DEGs after the first eight hours of ET-1 stimulation; the number of DEGs decreases rapidly by the extended culturing time. The exponential decreases of DEGs suggest that the acute cardiac response is highly different from how cardiac hypertrophy develops over time.

(7)

7 profile of ET-1 induced hypertrophy represents the pathological or the physiological type of cardiac hypertrophy to a greater extent.

Representative in vitro model can be extremely useful for studying pathological cardiac hypertrophy. However, to study physiological cardiac hypertrophy and understand the molecular differences between the two types, the animal-based model is still essential. This is because pregnancy-, exercise- or post-natal growth-induced cardiac hypertrophy cannot be created in the hiPSC-CMs model. Many studies have used animal models; however, the cardiovascular systems in animals and humans are different in certain aspects, and it is questionable to what extent the results from animal models are translatable to the human system (Lin et al., 2014). The rodent-model is suitable for studying gene expression regulations, and pathway analysis since the animal models do not have to mimic the syndromes that occur in human very closely (Milani-Nejad & Janssen, 2014). There is a considerable amount of non-protein coding and coding RNA expression differences between humans and mice, but many similarities are also highlighted by multiple studies (Guttman et al., 2009; Washietl, Kellis & Garber, 2014). Also, the network and pathway analysis are usually universal, and therefore, it is expected that the biological pathway of cardiac hypertrophy is not species-specific. A transcriptional profiling study in mice suggests that pregnancy-induced cardiac hypertrophy is more complex than exercise-induced cardiac hypertrophy (Chung, Heimiller & Leinwand, 2012). Several genes, such as Car3, Mt2, Nppb, and Ralgapa1, were regulated during pregnancy-induced hypertrophic conditions but not during exercise-induced hypertrophy. The regression of hypertrophic conditions started early in the pregnant mice, only 12 hours post-natal. This raises the questions about differences between physiological and pathological cardiac hypertrophy, and what makes the latter type of hypertrophic response irreversible. Taken together, the studies mentioned above showed that hiPSC-CM hypertrophic models have great potentials for modeling cardiovascular diseases involving hypertrophic response. However, further comprehensive investigations are required. It is uncertain whether ET-1 induced cardiac hypertrophy has the same gene expression profile as cardiac hypertrophy in patients even though the cells have phenotypes that represent the characteristics of pathological cardiac hypertrophy. Different types of cardiac hypertrophy also differ in complexity and are shown to have several unique genes involved. Moreover, many genes have been identified as cardiac hypertrophy biomarkers (Zhu, Li, Liu, Xu & Zhou, 2019). These biomarkers not only potentially may assist in cardiovascular disease diagnostic but also aids disease prediction for preventative medical approaches.

Biomarkers and clinical uses

Biomarkers or biological markers are molecules that can be used to indicate a biological phenomenon such as a disease or drug response. They must be detectable as present or showing unique characteristics in a specific medical state (Mayeux, 2004). There are several applications of biomarkers. They can be used to diagnose a disease, measure drug efficacy, measure prognosis of a disease, predict the risk of developing a disease, or determine an approach for treating a disease. A molecular approach for medical assessment utilizing biomarkers allows preventive practices.

(8)

8 The downregulated muscle-specific microRNA-1 (miR-1) has shown to be an essential element that activates growth-related genes and induces hypertrophy in a mouse model (Sayed, Hong, Chen, Lypowy & Abdellatif, 2007). These findings suggest that transcriptome analysis can and should be used to carry out comprehensive studies of the molecular mechanism of cardiac hypertrophy. The discoveries of novel biomarkers can be useful for the early diagnosis of diseases that involve in cardiac hypertrophic response from an early stage. Multiple growth factors and glycoprotein were recently discovered as molecular biomarkers of cardiac hypertrophy (Zhu, Li, Liu, Xu & Zhou, 2019). However, not every identified marker is suitable for clinical diagnosis due to laborious or harmful procedures during sampling. It is, therefore, important to identify molecular markers for the hypertrophic condition that is easy to use for detection, and that is possible to measure without invasive procedures. Examples of these are proteins secreted in the blood since they are likely to be detectable with a simple blood test.

A good biomarker candidate for clinical use measures a medical state in a patient with high accuracy, high reproducibility, and high efficiency (Selleck, Senthil & Wall, 2017). To achieve these criteria and develop a useful biomarker for clinical used, analytical validation and qualification processes with evidentiary assessment are crucial. One example of a recent success was the use of gene expression profiles in cancer prognosis that identified cancer patients who need aggressive treatment and those who do not. The success led to commercial gene signatures that are wildly available for testing breast cancer. However, the development of the assays involves extensive and intensive analytical validation and clinical validation with many collections of participant samples for a decade long (Holland, 2016).

Transcriptional analysis

Gene expression profiling performed by using high-throughput methods such as RNA-seq or DNA microarrays provides a comprehensive measurement of the gene expression level that can be used to predict and determine the pathways and biomarkers of biological functions (Chung, Heimiller & Leinwand, 2012). The analysis of these expression profiles can be used to determine relevant, differentially expressed genes (DEGs) and distinguish different cellular conditions. Transcriptome analysis of the expression profiles offers an instant, global view of cell activity and possible cell products even though the analysis does not always provide information about the exact products that were made because of alternative splicing and post-translational modifications. Differentially expressed genes can also be identified as biomarkers that indicate a pathogenic condition or cellular responses to pharmaceuticals as well as aid the diagnostic procedure.

Aim and research questions

(9)

9 The research questions are as follows: 1. Which genes are responders of ET-1 stimulation in hiPSC-CMs and can serve as putative biomarkers of induced cardiac hypertrophy? 2. Are any of the genes identified in research question 1 annotated as coding for secreted proteins that may be usable in clinical diagnosis? 3. Which biological pathways are affected by ET-1- induced cardiac hypertrophy in hiPSC-CMs? 4. Are there any known pharmaceuticals that are targeting the identified genes identified in research question 1 above?

Materials and methods

Data

The RNA-seq data were obtained from an in vitro differential expression analysis on human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) from Johansson et al. (2019). In the study, the analysis was carried out with the quasi-likelihood F-test using the edgeR package. The quasi-likelihood F-test applies negative binomial generalized linear models with a stricter error rate control than other edgeR pipelines (Lund, Nettleton, McCarthy & Smyth, 2012). The false discovery rate calculated using the Benjamini-Hochberg method was used as a significant cutoff since the analysis was a multiple testing analysis. The list of DEGs with at least the absolute value of fold change (FC) of 2 and a false discovery rate (FDR) of ≤ 0.05 was included in their result. From Johansson’s experiment, the hypertrophic response was induced in hiPSC-CMs by ET-1 stimulation (10nM). Every 24 hours, fresh medium containing ET-1 was added to the CMs that were incubated for more than 24 hours. RNA-seq was carried out on the stimulated cells after 8, 24, 48, 72, and 96 hours respectively, in triplicates. This experiment was also repeated three times. The raw and processed RNA-seq datasets from Johansson et al. (2019) is accessible at ArrayExpress (https://www.ebi.ac.uk/arrayexpress/) accession number: E-MTAB-8548 To investigate whether there are correlations between the in vitro hypertrophy model and the in vivo situation, a microarray dataset of transcription profiles of myocardial left ventricular biopsy samples from Aggarwal et al. (2014) was used. There were two sample groups in this experiment, an experimental group and a control group. The experimental group consists of heart samples from male patients (n=6) with aortic stenosis and noticeable left ventricular hypertrophy. The patients’ hearts either have normal ejection fraction (EF, >50%) or low EF (<30%), which represents different levels of severity of the valve disease and heart failure progression (Gronda et al., 2019). The control biopsy samples were harvested from male patients with coronary disease, but no cardiac hypertrophic response and have a normal EF (>60%). The RNA isolated from biopsies was analyzed using Affymetrix HG-U133A and U133B GeneChip sets. The microarray data from Aggarwal et al. (2014) is accessible at ArrayExpress (https://www.ebi.ac.uk/arrayexpress/) accession number: E-MEXP-2296.

Quality control and expression analysis of the heart biopsy microarray data

(10)

10 generalized linear modeling that was carried out using the limma R package (Smyth, 2005). The R script used for this step can be found in Appendix A. The genes with an absolute log2 FC of > 1 and FDR ≤0.05 were considered as statistically significant. The DEGs were then uploaded to IPA to identify the genes from the probe identifiers and remove duplicates and ambiguous probes that represent multiple loci or genes. The overlapping DEGs between the in vivo heart biopsies data and the DEGs of the cardiac hypertrophy stem cell-based model were then identified.

Canonical pathway selection

The Core Analysis function in Ingenuity Pathway Analysis (IPA) software was carried out to allow the in vivo DEGs to be interpreted under the context of biological pathways. IPA’s Canonical pathways are information manually curated by scientists based on literature review. The software overlaid the gene data input on to their gene product present in relevant pathways. The statistical significance of identified canonical pathways was then assessed using the right-tailed Fisher’s Exact Test with the significance threshold of p ≤0.05. This analysis also generated z-score values for prediction of whether the pathways were activated or inhibited base on the correlation between the datasets and the activated state of the pathway. The Comparison Analysis function was then performed to compare the significant pathways across the samples. This function allows several pathways relevant to DEGs in each sample to be compared to each other. Numbers of pathways were then selected to be used as parameter for filtering biomarker candidates in the later steps. These pathways were assumed to have an association with cardiac hypertrophy based on the literature review by the author and suggestions from scientists in this research area. In addition, the selected pathways must have a significance value of p≤0.05 in at least one of the samples.

Putative biomarker and identification of known pharmaceuticals

(11)

(12)

12 The second approach was similar to the first one with an extra filtering step that discriminated molecules that are not involved in any of the canonical pathways associated with cardiac hypertrophy. The pathways with p-value≤0.05 for early-progression of cardiac hypertrophy model were selected. The list of genes overlaid in these pathways was listed. The list of genes was then compared to the biomarker candidates identified in the first approach. The genes were then discriminated, and the overlapping genes with association to cardiac hypertrophy pathways were identified as biomarker candidates. The late-progression cardiac hypertrophy biomarker candidates were also filtered in the same manner. Biomarker Filter also revealed several additional information about the identified biomarker candidates, such as targeting pharmaceuticals and detection in body fluids. Information on the use of these genes in any present biomarker applications was also derived using the QIAGEN’s database. The finding reference number starting with IEF (Ingenuity Expert Findings) or NCT (ClinicalTrials.gov identifier) was stated if applicable.

The third approach relied on the identification of overlapped DEGs of the in vivo and in vitro samples. The DEGs from transcription profiles of patients with aortic stenosis and left ventricular hypertrophy were compared to the DEGs of the ET-1 induced cardiac hypertrophy model in hiPSC-CMs. The patients here either have normal EF (>50%) or low EF (<30%). The overlapping genes were then identified as biomarker candidates for cardiac hypertrophy.

Molecular activity prediction (MAP)

(13)

13

Results

Quality control and differentially expression analysis

The microarray in vivo data was normalized using the RMA algorithm. The distribution plots can be seen in Appendix B. The plots show a normal distribution with no significant outliers different from other samples, and therefore, no further filtering steps were needed. The heatmaps (Appendix C) generated showed samples of the same condition (Normal EF, Low EF or control) were assigned to the same cluster, the clusters were also seen in PCA plots (Appendix D). The expression data with probe IDs representing DEGs of both in vitro and in vivo datasets were uploaded on to IPA. For the in vitro data, genes were represented by their Ensembl ID. For the in vivo data, the Affymetrix-probe ID were used. Gene symbol and Entrez gene names were mapped onto these IDs. The unmapped IDs were duplicates or have ambiguous probes that represent multiple loci or genes. The number of totally identified DEG IDs and mapped genes were shown in Table 1.

Table 1. All ensemble IDs identified by R and mapped DEGs of the in vitro hiPSC-CMs based model (Left) and all Affymetrix-probe IDs identified using the Miodin package and mapped DEGs of the in vivo-heart biopsy data (Right).

Canonical pathway associated with cardiac hypertrophy

Eleven pathways associated with cardiac hypertrophy were identified (Figure 2). Among these, four were selected as relevant pathways for early-progression cardiac hypertrophy since they are statistically significant in both 8h and 24h samples (p≤ 0.05). These include apelin cardiac fibroblast signaling pathway, cAMP-mediated signaling, G-protein coupled receptor signaling, and ILK signaling. The DEGs from the investigated dataset and involved in these pathways were used to filter biomarkers candidates of early-progression cardiac hypertrophy. In the late progression of cardiac hypertrophy in hiPSC-CMs models, the cardiomyocyte differentiation via the BMP receptors pathway was statistically significant. The genes from the dataset associated with that pathway were used to filter biomarker candidates for the late-progression cardiac hypertrophy.

DEGs of the in vitro hiPSC-CMs based model

Samples All IDs Mapped IDs 8h 1017 978 24h 699 680 48h 152 148 72h 157 151 96h 45 44 DEGs of the in vivo-heart biopsy data

Samples All IDs Mapped IDs Normal EF 106 105

(14)

14 Figure 2. Heatmap of signaling pathways associated with cardiac hypertrophy. Left; Z-score of samples at different time points; orange color represents positive z-score (activation), blue represents negative z-score (inactivation), white represents z-score around 0, and grey represents that no prediction can be made. Right; P-values of samples at different time points, the dots indicate pathways that do not pass the cut off significant value of P≤0.05.

Common biomarkers for early- and late- progression cardiac hypertrophy

Thirty genes were initially identified as putative biomarkers for the early progression ET-1 induced cardiac hypertrophy in hiPSC (Appendix E). These genes encode proteins of different families, including enzymes, G-protein coupled receptors, growth factors, ion channels, kinases, phosphatases, transcription regulators, transmembrane receptors, transporters, and other families. For the late progression of cardiac hypertrophy in the hiPSC model, eight biomarkers were initially identified (Appendix F). Out of the identified biomarkers, the two stages have three biomarkers in common, which are GRM1, NPPA, and STC1. Notably, these three genes are detectible in blood. The expression values and the FDR for these three biomarker candidates from both early and late-progression cardiac hypertrophy, as well as their biomarker applications, are shown in table 2.

Table 2. The expression and previous applications of three overlapped biomarker candidates for ET-1 induced cardiac hypertrophy model in hiPSC-CMs at different stages (early & late-progression) using biomarker identification approach 1. The identified reference numbers starting with IEF (Ingenuity Expert Findings) or NCT (ClinicalTrials.gov identifier) were also stated if applicable. The arrows indicate whether the genes were upregulated (↑) or downregulated (↓) in the expression data. Putative biomarker LogFC (8h,24h, 72h,96h) FDR (8h,24h, 72h,96h) Biomarker application (s)

(15)

15 NPPA↑ Natriuretic peptide A 1.048 1.754 1.025 1.208 2.79E-08 1.96E-28 5.28E-17 2.34E-03

diagnosis heart failure, ischemic stroke, atrial fibrillation NCT0053762 8, NCT0021378 5, NCT0054265 9 Yes efficacy Cardiomyopat hy treatment using diltiazem NCT0031998 2 unspecified correlation to mild congestive heart failure IEF# 11530095 STC1↑ Stanniocalc in 1 2.24 2.47 2.29 2.81 1.98E-41 2.11E-41 4.65E-35 2.90E-06

diagnosis Ovarian cancer American Association for Cancer Research (AACR) Meeting Abstracts Yes

Biomarker candidates for different hypertrophic stages in hiPSC-CMs

(16)

16 Table 3. Biomarkers candidates of early progression of cardiac hypertrophy associated with signaling pathways relevant to cardiac hypertrophy. The table shows expression levels and FDR of the genes, as well as their known biomarker applications. The identified reference numbers starting with IEF (Ingenuity Expert Findings) or NCT (ClinicalTrials.gov identifier) were stated if applicable. The arrow indicates whether the biomarker is upregulated (↑) or downregulated (↓) in early-progression cardiac hypertrophy. Symbol LogFC (8h,24h) FDR (8h,24h) Biomarker application (s) Content Findings ID/ Source Detect-able in Blood GRM1 ↓ Glutamate metabotropic receptor 1 -2.720 -3.390 6.07E-03 1.00E-06 efficacy melanoma treatment using riluzole NCT00667901 Yes MAP2K6 ↓ Mitogen-activated protein kinase 6 -2.055 -2.763 3.03E-12 3.69E-19

prognosis bladder cancer AACR Meeting Abstracts No DUSP6 ↑ Dual specificity phosphatase6 1.433 2.110 8.01E-14 1.34E-36 diagnosis non-small cell lung cancer AACR Meeting Abstracts Yes DUSP4 ↑ Dual specificity phosphatase4 1.054 1.844 1.87E-14 5.43E-07

(17)

(18)

18 pulmonary arterial hypertension disease progression benign prostatic hyperplasia, melanoma NCT00407329, AACR meeting abstracts efficacy ezetimibe in atherosclerosis treatment, spironolactone in the treatment of atrial fibrillation NCT00485121 NCT00141778 prognosis acute respiratory distress syndrome, colorectal cancer, non-small cell lung cancer NCT00673517, NCT00362882, AACR meeting abstrac safety candesartan in the treatment of hypertension, orlistat in the treatment of obesity, NCT00644475, NCT00422058 For the late-progression cardiac hypertrophy sample, eight genes were initially identified using Biomarker Filter were further discriminated based on the genes that present in the cardiomyocyte differentiation via the BMP receptors pathway. The pathway was enriched in the late-progression cardiac hypertrophy (P≤0.05). One biomarker candidate associated with cardiac hypertrophy pathways was identified, which was NPPA (Table 4).

(19)

19

Comparison to the expression profile of aortic stenosis induced cardiac hypertrophy

The expression analysis of the biopsies data of the hypertrophied heart with normal- and low-ejection fraction (EF) revealed 106 and 88 DEGs, respectively. Sixteen different overlapped genes were identified. These genes were presented in both the DEGs of the hypertrophied heart biopsies data (either with Normal EF or low EF) and at least one of the samples of in vitro ET-1-induced cardiac hypertrophy model in hiPSC-CMs. These genes include ACE2, CCN2, COLA1, CORIN, DDH1, FAM155B, HSPA2, IER3, IRS2, KCNIP2, NES, NPPA, NPPB, PPP1R1A, RASL11B, and THBS1. When identifying the overlapping DEGs between the high-EF heart profile and the hiPSC hypertrophic model, the analysis identified 7, 5, 2, 1, and 1 gene at 8h, 24h, 48h, 72h, and 96h, respectively. For the overlapping DEGs between the low-EF heart profile and the hiPSC hypertrophic model, the analysis identified 8, 11, 3, 3, and 2 genes at 8h, 24h, 48h, 72h, and 96h, respectively. Both up- and downregulated overlapped genes identified from this analysis can be seen in Figure 3. There were 16 DEGs from the different time points of ET-1 hiPSC-CMs cardiac hypertrophy model that overlapped with the DEGs of aortic stenosis induced cardiac hypertrophy (Figure 3).

Figure 3. The overlapping genes between the in vivo hypertrophic heart profile and the hiPSC hypertrophic model. The boxes highlighted in red represent upregulated overlapping DEGs and in green for down-regulated DEGs.

Among these overlapped DEGs, three genes were unique to the normal expression profile of the heart with normal EF and not present in low EF; these genes were THBS1, COL12A1, and FAM155B. On the other hand, nine genes were unique to the profile of the low EF heart; these were NPPB, NPPA, IER3, RASL11B, ACE2, IRS2, NES, KCNIP2, and CORIN. Out of the 16 overlapping DEGs, four

(20)

20 genes were also identified by IPA’s Biomarker Filtering tool as lenient biomarker candidates for early-progression cardiac hypertrophy. These include IER3, NPPA, NPPB, and THBS1.

Putative target pharmaceuticals

Two known drugs were identified by the IPA database to target biomarker candidates of the early-progression cardiac hypertrophy model in hiPSC-CMs. The first drug is fasoracetam, which is known to target GRM1. This gene was downregulated in the hiPSC dataset. The MAP tool revealed that the drug activates target gene GRM1, as well as some other glutamate metabotropic receptors, which have an indirect effect in G-coupled protein receptor activation. Another putative target drug was drotrecogin alfa, which targets SERPINE1. The effect of the drug was predicted by MAP. Drotrecogin alfa indirectly inhibits cardiac fibrosis and myofibroblasts differentiation in the apelin cardiac fibroblast signaling pathway (Figure 4). Figure 4. Effects of drotrecogin alfa on apelin cardiac fibroblast signaling predicted by MAP. Red represent activated elements, and blue represents inhibited elements. Solid lines represent direct effects, and dotted lines represent indirect effects. More information about the symbol legends can be seen in appendix G.

Apelin cardiac fibroblast signaling pathway

The 8h sample was predicted to activate cardiac fibrosis and myofibroblast due to the upregulation of SERPIN1, CTGF, and SPHK1 in the dataset (Figure 5A). In contrast, cardiac fibrosis and myofibroblast activation were inhibited in the 96h sample. Upregulations of Angiotensin II complex and ACE2 peptidase were presented in the data. These two interact, but there are some inconsistencies with these findings and the state of downstream molecules. As seen in figure 5B, the upregulated APLN complex leads to the inactivation of TGFβ and CTGF, which, in turn, leads to the inactivation of cardiac fibrosis. However, the activation of SERPINE1 caused by AGTR1 activation and upregulated angiotensin II usually correlates to cardiac fibrosis activation. The upstream molecules result in the inconsistent in these findings, and therefore, the activity prediction of cardiac fibrosis should be considered with less confidence.

(21)

21

Figure 5. Apelin cardiac fibroblast signaling in (A) 8h- and (B) 96h- ET-1 induced hypertrophy stem cell sample. More information about the symbol legends can be seen in appendix G.

Discussions and conclusions

In vitro cardiac hypertrophy biomarker candidates

(22)

22 cell migration, and regulation of cardiac muscle cell contraction. The gene was upregulated in this study, therefore suggested a decrease in cellular calcium concentration and inactivation of cAMP signaling. This is in opposition to the literature suggesting that cAMP and calcium ion influx has been associated with cardiac remodeling and cardiac hypertrophy (Houser & Carnell, 2009). In cAMP signaling, the activation of G-protein coupled signaling triggers the signal that catalyzes ATP into cAMP. The increase in cAMP concentration leads to an activation of protein kinase A, and activation of voltage-dependent calcium channels, causing calcium ion influx (Zaccolo, 2009). In a study using rats, it was found that when using triac to induce myocardia disarray and hypertrophy, there was an increase in intracellular calcium concentrations (Pearce, Hawkey, Symons & Olsen, 1985).

(23)

23 and heart failure. Lastly, an upregulation of SERPINE1 was identified as biomarker candidates for the early-progression cardiac hypertrophy induced by ET-1. Existing evidence suggests that the gene is indirectly involved in cardiac fibrosis activation (Flevaris et al., 2017). Myocardial fibrosis contributes to heart failure and is one of the hallmarks of a manifestation of hypertrophic cardiomyopathy, a disease in which, the thickening of heart muscle walls is one of the symptoms (Ho et al., 2010).

Biomarkers candidates from in vivo-in vitro data comparisons

The third approach for identifying biomarker candidates was to identify overlapped DEGs between in vitro and in vivo data of cardiac hypertrophy. The DEGs obtain from hypertrophied heart profiles were also used as an early validation method of the biomarker candidates identified earlier. Sixteen overlapping genes were identified. These genes were present in both the DEGs of the hypertrophied heart biopsies data (either with Normal EF or low EF) and at least one of the samples of in vitro ET-1-induced cardiac hypertrophy model in hiPSC-CMs. Out of these genes, four were also identified by the Biomarker filtering approaches using IPA including, IER3, NPPA, NPPB, and THBS1. This pointed out one weakness of using the Biomarker-Filter function to identify biomarker candidates since it relies on existing biomarker applications of the genes that have been curated into the Ingenuity Knowledge Base. Collagen type 12 alpha- chain 1 (COL12A1) was one of the protein-coding genes that were upregulated in all time points in the in vitro data. Still, there was no information on its application as a biomarker in IPA’s database. In the heart biopsy data, the gene was overexpressed in a hypertrophied heart with normal EF. The gene encodes for a chain of type 12 collagen. Excessive collagen deposition has been found to play roles in cardiac hypertrophy and heart failure progression (Maulik & Mishra, 2015). Another gene that was overexpressed in all in vitro samples but only in the low EF hypertrophied heart was NPPA. The result suggests that the expression profiles of the in vitro and in vivo have underlying differences. The hypertrophied heart profiles also revealed the differences between hypertrophied heart with normal EF and low EF in terms of their expression profile. The EF of the heart had been to assess the functions of the hearts and track heart failure. A high ejection fraction was a sign of hypertrophic cardiomyopathy at the later stages (Olivotto, Cecchi, Poggesi & Yacoub, 2012). The cause of heart disease patients to also develop low EF is still unclear. This study showed hypertrophied hearts with normal EF, and low EF may have different mechanisms since they yielded different expression profiles.

Strengths and limitations of biomarker identification approaches

Biomarker candidates were identified using three different approaches; (1) by identifying common genes across all sample with previous biomarker applications, (2) by identifying genes with previous biomarker applications and association to the cardiac hypertrophy associated pathways, (3) By identifying overlapped DEGs between in vitro and in vivo data of cardiac hypertrophy.

(24)

24 By filtering the genes and considering only those associated with cardiac hypertrophy-related pathways in the second biomarker identification approach, the identified genes are more likely to be relevant to the hypertrophic response. One limitation of using IPA’s Biomarker Filter function was the limitation of the Ingenuity Knowledge Base (IKB) database developed by Qiagen. There must be existing biomarker application information about the biomarker candidates for them to be considered in the filtering step. This can be an advantage where the curated information about the molecules suggests the relevancy of the biomarker candidates. However, those molecules that have not been previously utilized as biomarkers and those that have not yet been curated would not be considered as putative biomarkers by Biomarker Filter function. The risk of making a type II error increases.

The third approach relied on the identification of overlapped DEGs of the in vivo and in vitro samples. One advantage of this in vitro-in vivo comparison approach is that the putative biomarkers can be identified without the need for previous information about the biomarker characteristics and applications. This approach was also an attempt of early-phase validation of the putative biomarkers to see whether the biomarker genes identified in the in vitro data also present in the in vivo data. The limitation of this method is that the ET-1 induced cardiac hypertrophy might not be a suitable model for cardiac hypertrophic response induced by other stimuli. The disparity between hiPSC-CMs and CMs in the heart also plays a role in the difference in their expression profiles. Furthermore, there might not be information about previous biomarker characteristics and applications to support the biomarker candidates identified.

Putative pharmaceuticals targeting cardiac hypertrophy biomarker candidates

Another objective of the study was to identify any known pharmaceuticals that are targeting the biomarker candidates identified in the research. The two drug candidates identified were fasoracetam targeting GRM1 and drotrecogin alfa targeting SERPINE1.

(25)

25 and indirectly inhibit cardiac fibrosis through the apelin cardiac fibroblast signaling pathway. Cardiac fibrosis is one of the signs of hypertrophic cardiomyopathy and causes heart failure. Drotrecogin alfa may be further investigated for its effect on cardiac hypertrophy. However, it is important to be aware of the adverse bleeding risks causing by drotrecogin alfa and thus, the potential clinical use remains questionable.

Apelin cardiac fibroblast signaling pathway in early- and late-progression cardiac

hypertrophy

Another objective of this study was to identify the biological pathways that are affected by ET-1- induced cardiac hypertrophy in hiPSC-CMs. Eleven pathways were associated with cardiac hypertrophy identified based on literature reviews by the author and suggestions from scientists in this research area. The pathways relevant to the datasets used in this study were then selected based on their statistical relevance (p≤ 0.05). The study considered the Fisher’s exact test p-value instead of Benjamin-Hochberg (B-H) adjusted calculated by IPA, despite the multiple comparisons being carried out. This was to avoid the false negatives caused by the multiple comparison corrections in the B-H procedure. The identified pathways were biologically relevant, with evidence supported by the literature. Therefore, the statistically significant was not prioritized. One of the interesting pathways in this study was the apelin cardiac fibroblast signaling pathway. This was because it was the only pathway that the MAP tool predicts an opposite cellular activity in the early- and late- progression cardiac hypertrophy (8h and 96h sample, in the case). In the other pathways, the cellular function and activity were predicted to be similar, or there was not enough information (too few DEGs) for the prediction to be carried out. Activation of cardiac fibrosis, which is one of the hallmarks of cardiac hypertrophy, was predicted in the 8h sample. On the other hand, in the 96h sample, cardiac fibrosis was inactivated due to the upregulated apelin peptide encoded by the APLN gene from the dataset. The activity prediction in the 96h sample contained some inconsistency. Cardiac fibrosis is an abnormal disposition of extracellular matrix or the thickening of heart valves caused by abnormal cardiac fibroblast-proliferation (Gourdie, Dimmeler & Kohl, 2016). This condition is found in patients with heart failure. In Johansson’s study, the size of cardiomyocytes stops increasing after 48 hours. Certain genes present in the apelin cardiac fibroblast signaling pathway may be correlated to the regulation of cell size in Johansson’s study. However, the result of this study was limited, and it was not possible to conclude the cause for the cell size to stop increasing.

Ethical issues and impacts on society

(26)

26 The in vivo data of the transcriptional profile of human heart biopsies from the study by Aggarwal was also used in this project. The participants of Aggarwal’s study were patients who were undergoing heart surgery when collecting heart samples. Participants were informed about the procedure, and written consents were given. The use of in vivo samples is crucial in comprehensive research that may lead to the understanding of hypertrophic response and the development of in vitro disease models in the future.

Both datasets used in this study are available in the public database ArrayExpress (https://www.ebi.ac.uk/arrayexpress/). The data are accessible to perform different analysis to make the most use of resources. Furthermore, the stem cell-based cardiac hypertrophy model has the potential to replace the need for animal-based cardiac hypertrophy research.

The most common leading cause of death in Sweden was cardiovascular diseases, which caused over 30,000 deaths in 2018, according to the National Board of Health and Welfare (Socialstyrelsen, 2019). Pathological cardiac hypertrophy is an important risk factor that leads to heart failure; it is therefore improved knowledge is therefore important to better understand the mechanism involved. One big challenge in understanding the development of cardiac hypertrophy was the lack of human samples. Scientists have been developing an in vitro model that functions as a cardiac hypertrophy model and can potentially be used as a drug screening platform. This study attempted to identify biomarker candidates and target pharmaceuticals, as well as provide insights into the relevant mechanism of the in vitro model. These findings could potentially be helpful for the early diagnosis of cardiac hypertrophy in the future. Several identified biomarkers candidates in this study helps support the association between the genes and cardiovascular system and cardiac hypertrophy previously found in other literature. The study also revealed the underlying differences between the in vitro model and cardiac hypertrophy of the heart. However, it was unclear whether these differences are significant biologically. More research is needed to determine to what extent the in vitro model represents the in vivo hypertrophic response of the heart.

Conclusions

In this study, Ingenuity Pathway Analysis (IPA) and three approaches of biomarker candidate identification were used to investigate underlying mechanisms and molecules involved in the cardiac hypertrophy model in hiPSC-CMs. The three biomarker identification methods offer different advantages and limitations; however, all of them successfully identified numbers of putative biomarkers. These biomarker candidates require further validation steps in order to develop a biomarker with potential clinical applications. Several relevant pathways in hiPSC-CMs cardiac hypertrophy model were different from the known activity in the literature, for example, cAMP-mediated-, MAPK-, and Proteinase A-signaling were inhibited in the early-progression cardiac hypertrophy model in this study. Thus, the stem-cell-based model does not fully represent the cardiac hypertrophy of a heart in vivo. However, the model might still be suitable for exploring the cardiac hypertrophy condition due to identified molecular and phenotypic similarities. More research is needed to assess the extent of the applications of the ET-1 induced hiPSC-CM cardiac hypertrophy model.

(27)

27 and future analysis should also consider using proteomic data. Furthermore, the in vitro cardiac hypertrophy model used in this study was generated by only one cell-line, cardiomyocytes. This model does not fully represent the environment of a human heart, and it would be interesting to analyze a model with multiple hPSC-lines and their interactions.

Acknowledgments

First and foremost, I would like to express my sincere gratitude to my supervisor, Associate Professor Jane Synnergren – for the continuous support throughout the project. Assoc. Prof. Synnergren has helped me to focus on the relevance implementation and results and also gave interesting insights. Her guidance and mentoring were the important factor that this project was carried out successfully and smoothly, even during the coronavirus pandemic.

(28)

28

References

1. Aggarwal, P., Turner, A., Matter, A., Kattman, S. J., Stoddard, A., Lorier, R., … Broeckel, U. (2014). RNA Expression Profiling of Human iPSC-Derived Cardiomyocytes in a Cardiac Hypertrophy Model. PLoS ONE, 9(9). doi: 10.1371/journal.pone.0108051

2. Badiani, S., van Zalen, J., Treibel, T., Bhattacharyya, S., Moon, J., & Lloyd, G. (2016). Aortic Stenosis, a Left Ventricular Disease: Insights from Advanced Imaging. Current Cardiology

Reports, 18(8). doi: 10.1007/s11886-016-0753-6

3. Bupha-Intr, T., Haizlip, K., & Janssen, P. (2012). Role of Endothelin in the Induction of Cardiac Hypertrophy In Vitro. Plos ONE, 7(8), e43179. doi: 10.1371/journal.pone.0043179

4. Carlson, C., Koonce, C., Aoyama, N., Einhorn, S., Fiene, S., Thompson, A., … Kattman, S. (2013). Phenotypic Screening with Human iPS Cell–Derived Cardiomyocytes. Journal of Biomolecular Screening, 18(10), 1203–1211. doi: 10.1177/1087057113500812

5. Chung, E., Heimiller, J., & Leinwand, L. (2012). Distinct Cardiac Transcriptional Profiles Defining Pregnancy and Exercise. Plos ONE, 7(7), e42297. doi: 10.1371/journal.pone.0042297

6. Du, X., Zhang, S., & Liu, L. (2019). The association of B-type natriuretic peptide with Left ventricular hypertrophy. International Journal of Cardiology, 297, 143. doi: 10.1016/j.ijcard.2019.06.022

7. Eghbali, M., Deva, R., Alioua, A., Minosyan, T. Y., Ruan, H., Wang, Y., … Stefani, E. (2005). Molecular and Functional Signature of Heart Hypertrophy During Pregnancy. Circulation Research, 96(11), 1208–1216. doi: 10.1161/01.res.0000170652.71414.16

8. Elia, J., Ungal, G., Kao, C., Ambrosini, A., De Jesus-Rosario, N., & Larsen, L. et al. (2018). Fasoracetam in adolescents with ADHD and glutamatergic gene network variants disrupting mGluR neurotransmitter signaling. Nature Communications, 9(1). doi: 10.1038/s41467-017-02244-2

9. European Medicines Agency. (2011). Xigris (drotrecogin alfa (activated)) to be withdrawn

due to lack of efficacy. Retrieved from

https://web.archive.org/web/20140308033604/http://www.ema.europa.eu/docs/en_ GB/document_library/Press_release/2011/10/WC500116970.pdf

10. Flevaris, P., Khan, S., Eren, M., Schuldt, A., Shah, S., & Lee, D. et al. (2017). Plasminogen Activator Inhibitor Type I Controls Cardiomyocyte Transforming Growth Factor-β and Cardiac Fibrosis. Circulation, 136(7), 664-679. doi: 10.1161/circulationaha.117.028145

11. Frey, N., & Olson, E. (2003). Cardiac Hypertrophy: The Good, the Bad, and the Ugly. Annual Review Of Physiology, 65(1), 45-79. doi: 10.1146/annurev.physiol.65.092101.142243

12. Frey, N., Katus, H., Olson, E., & Hill, J. (2004). Hypertrophy of the Heart. Circulation, 109(13), 1580-1589. doi: 10.1161/01.cir.0000120390.68287.bb

(29)

29 14. Gronda, E., Vanoli, E., Sacchi, S., Grassi, G., Ambrosio, G., & Napoli, C. (2019). Risk of heart failure progression in patients with reduced ejection fraction: mechanisms and therapeutic options. Heart Failure Reviews, 25(2), 295-303. doi: 10.1007/s10741-019-09823-z

15. Guttman, M., Amit, I., Garber, M., French, C., Lin, M. F., Feldser, D., … Lander, E. S. (2009). Chromatin signature reveals over a thousand highly conserved large noncoding RNAs in mammals. Nature, 458(7235), 223–227. doi: 10.1038/nature07672

16. Ho, C., López, B., Coelho-Filho, O., Lakdawala, N., Cirino, A., & Jarolim, P. et al. (2010). Myocardial Fibrosis as an Early Manifestation of Hypertrophic Cardiomyopathy. New England Journal Of Medicine, 363(6), 552-563. doi: 10.1056/nejmoa1002659

17. Holland, R. (2016). What makes a good biomarker?. Advances In Precision Medicine, 1(1),

66. doi: 10.18063/apm.2016.01.007

18. Hosac, A. (2002). Drotrecogin Alfa (Activated): The First Fda-Approved Treatment for Severe Sepsis. Baylor University Medical Center Proceedings, 15(2), 224-227. doi: 10.1080/08998280.2002.11927844

19. Houser, S., & Carnell, L. (2009). Ca2+ Signaling Domains Responsible For Cardiac Hypertrophy and Arrhythmias. Circulation Research, 104(4), 413-415. doi: 10.1161/circresaha.109.193821

20. Janz, K., Dawson, J., & Mahoney, L. (2000). Predicting Heart Growth During Puberty: The Muscatine Study. PEDIATRICS, 105(5), e63-e63. doi: 10.1542/peds.105.5.e63

21. Johansson, M., Ulfenborg, B., X Andersson, C., Hagvall, S., Jeppsson, A., Sartipy, P., & Synnergren, J. (2019). Cardiac Hypertrophy in a Dish: A Human Stem Cell Based Model.

doi: 10.20944/preprints201911.0395.v1

22. Kattman, S., Koonce, C., Swanson, B., & Anson, B. (2010). Stem Cells and Their Derivatives: A Renaissance in Cardiovascular Translational Research. Journal Of Cardiovascular Translational Research, 4(1), 66-72. doi: 10.1007/s12265-010-9235-1

23. Keys, J., Greene, E., Koch, W., & Eckhart, A. (2002). Gq-Coupled Receptor Agonists Mediate Cardiac Hypertrophy Via the Vasculature. Hypertension, 40(5), 660-666. doi: 10.1161/01.hyp.0000035397.73223.ce

24. Lafeber, F., Flik, G., Wendelaar Bonga, S., & Perry, S. (1988). Hypocalcin from Stannius corpuscles inhibits gill calcium uptake in trout. American Journal Of Physiology-Regulatory, Integrative And Comparative Physiology, 254(6), R891-R896. doi: 10.1152/ajpregu.1988.254.6.r891

25. Lin, S., Lin, Y., Nery, J. R., Urich, M. A., Breschi, A., Davis, C. A., … Snyder, M. P. (2014). Comparison of the transcriptional landscapes between human and mouse tissues. Proceedings of the National Academy of Sciences, 111(48), 17224–17229. doi: 10.1073/pnas.1413624111

26. Liu, R., & Molkentin, J. (2016). Regulation of cardiac hypertrophy and remodeling through the dual-specificity MAPK phosphatases (DUSPs). Journal Of Molecular And Cellular Cardiology, 101, 44-49. doi: 10.1016/j.yjmcc.2016.08.018

(30)

30 27. Lund, S., Nettleton, D., McCarthy, D., & Smyth, G. (2012). Detecting Differential Expression in RNA-sequence Data Using Quasi-likelihood with Shrunken Dispersion Estimates. Statistical Applications In Genetics And Molecular Biology, 11(5). doi: 10.1515/1544-6115.1826

28. Lundin, A., Delsing, L., Clausen, M., Ricchiuto, P., Sanchez, J., Sabirsh, A., … Falk, A. (2018). Human iPS-Derived Astroglia from a Stable Neural Precursor State Show Improved Functionality Compared with Conventional Astrocytic Models. Stem Cell Reports, 10(3),

1030–1045. doi: 10.1016/j.stemcr.2018.01.021

29. Marian, A. (2008). Genetic determinants of cardiac hypertrophy. Current Opinion In Cardiology, 23(3), 199-205. doi: 10.1097/hco.0b013e3282fc27d9

30. Martino, J., Wall, B., Mastrantoni, E., Wilimczyk, B., La Cava, S., & Degenhardt, K. et al. (2012). Metabotropic glutamate receptor 1 (Grm1) is an oncogene in epithelial cells. Oncogene, 32(37), 4366-4376. doi: 10.1038/onc.2012.471

31. Maulik, S., & Mishra, S. (2015). Hypertrophy to failure: What goes wrong with the fibers of the heart?. Indian Heart Journal, 67(1), 66-69. doi: 10.1016/j.ihj.2015.02.012

32. Mayeux, R. (2004). Biomarkers: Potential uses and limitations. Neurorx, 1(2), 182-188.

doi: 10.1602/neurorx.1.2.182

33. Milani-Nejad, N., & Janssen, P. (2014). Small and large animal models in cardiac contraction research: Advantages and disadvantages. Pharmacology & Therapeutics, 141(3), 235-249. doi: 10.1016/j.pharmthera.2013.10.007

34. Muslin, A. (2008). MAPK signaling in cardiovascular health and disease: molecular mechanisms and therapeutic targets. Clinical Science, 115(7), 203-218. doi: 10.1042/cs20070430

35. Olivotto, I., Cecchi, F., Poggesi, C., & Yacoub, M. (2012). Patterns of Disease Progression in Hypertrophic Cardiomyopathy: An Individualized Approach to Clinical Staging. Circulation: Heart Failure, 5(4), 535-546. doi: 10.1161/circheartfailure.112.967026

36. Ottaviani, L., & da Costa Martins, P. (2017). Noncoding RNAs in cardiac hypertrophy. The Journal Of Physiology, 595(12), 4037-4050. doi: 10.1113/jp273129

37. Papathanasiou, S., Rickelt, S., Soriano, M., Schips, T., Maier, H., & Davos, C. et al. (2015). Tumor necrosis factor-α confers cardioprotection through ectopic expression of keratins K8 and K18. Nature Medicine, 21(9), 1076-1084. doi: 10.1038/nm.3925

38. Pearce, P., Hawkey, C., Symons, C., & Olsen, E. (1985). Role of calcium in the induction of cardiac hypertrophy and myofibrillar disarray. Experimental studies of a possible cause of hypertrophic cardiomyopathy. Heart, 54(4), 420-427. doi: 10.1136/hrt.54.4.420

39. Pluim, B., Zwinderman, A., van der Laarse, A., & van der Wall, E. (2000). The Athlete’s Heart. Circulation, 101(3), 336-344. doi: 10.1161/01.cir.101.3.336

40. Selleck, M., Senthil, M., & Wall, N. (2017). Making Meaningful Clinical Use of Biomarkers. Biomarker Insights, 12, 117727191771523. doi: 10.1177/1177271917715236

(31)

31 relevance to human heart failure. American Journal Of Physiology-Heart And Circulatory Physiology, 285(1), H442-H448. doi: 10.1152/ajpheart.01071.2002

42. Smyth, G. (2005). limma: Linear Models for Microarray Data. Bioinformatics And Computational Biology Solutions Using R And Bioconductor, 397-420. doi: 10.1007/0-387-29362-0_23

43. Socialstyrelsen. (2019). Statistics on Causes of Death 2018 [Ebook]. Official statistics of Sweden. Retrieved from https://www.socialstyrelsen.se/globalassets/sharepoint-dokument/artikelkatalog/statistik/2019-9-6300.pdf

44. Sun, L., Zhang, Y., Zhang, Y., Gu, Y., Xuan, L., Liu, S., … Yang, B. (2014). Expression profile of long noncoding RNAs in a mouse model of cardiac hypertrophy. International Journal of Cardiology, 177(1), 73–75. doi: 10.1016/j.ijcard.2014.09.032 45. Ulfenborg, B. (2019). Vertical and horizontal integration of multi-omics data with miodin. BMC Bioinformatics, 20(1). doi: 10.1186/s12859-019-3224-4 46. UniProt: a worldwide hub of protein knowledge. (2018). Nucleic Acids Research, 47(D1), D506-D515. doi: 10.1093/nar/gky1049 47. Wang, Y. (2007). Mitogen-Activated Protein Kinases in Heart Development and Diseases. Circulation, 116(12), 1413-1423. doi: 10.1161/circulationaha.106.679589

48. Washietl, S., Kellis, M., & Garber, M. (2014). Evolutionary dynamics and tissue specificity of human long noncoding RNAs in six mammals. Genome Research, 24(4), 616-628. doi: 10.1101/gr.165035.113

49. Wilson, B. (2018). Developmental Genetics (1st ed., p. 119). London: ETP.

50. Zaccolo, M. (2009). cAMP signal transduction in the heart: understanding spatial control for the development of novel therapeutic strategies. British Journal Of Pharmacology,

158(1), 50-60. doi: 10.1111/j.1476-5381.2009.00185.x 51. Zhang, J., Hu, S., & Li, Y. (2019). KRT18 is correlated with the malignant status and acts as an oncogene in colorectal cancer. Bioscience Reports, 39(8). doi: 10.1042/bsr20190884 52. Zhang, M., Jiang, Y., Guo, X., Zhang, B., Wu, J., Sun, J., … Xu, C. (2019). Long noncoding RNA cardiac hypertrophy-associated regulator governs cardiac hypertrophy via regulating miR-20b and the downstream PTEN/AKT pathway. Journal of Cellular and Molecular

Medicine, 23(11), 7685–7698. doi: 10.1111/jcmm.14641

53. Zhang, W., & Liu, H. (2002). MAPK signal pathways in the regulation of cell proliferation in mammalian cells. Cell Research, 12(1), 9-18. doi: 10.1038/sj.cr.7290105

54. Zhou, W., Gong, L., Li, X., Wan, Y., Wang, X., Li, H., & Jiang, B. (2018). Screening key candidate genes and pathways involved in insulinoma by microarray analysis. Medicine, 97(22),

e10826. doi: 10.1097/md.0000000000010826

55. Zhu, L., Li, C., Liu, Q., Xu, W., & Zhou, X. (2019). Molecular biomarkers in cardiac hypertrophy. Journal Of Cellular And Molecular Medicine, 23(3), 1671-1677. doi: 10.1111/jcmm.14129

(32)

(33)

33

Appendices

Appendix A – R script

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52

#process HG-U133A genechip library("miodin")

mp <- MiodinProject( name = "Thesis",

author = "Benyapa Tangruksa", path ="."

)

mshow(mp)

sampleTable <- data.frame(

SampleName = paste0("S", 1:9),

SamplingPoint = paste0("sp", rep(1,9)), DiseaseState = rep(c("Hyper_EF.50.","Hyper_lowEF.30.","Control_EF.60."), each = 3), stringsAsFactors = FALSE ) assayTable <- data.frame( SampleName =paste0("S", 1:9), DataFile= paste0("datasets/E-MEXP-2296/",

c("12_A.CEL", "14_A.CEL", "16_A.CEL", "11_A.CEL", "7_A.CEL", "8_A.CEL", "10_A.CEL", "18_A.CEL", "19_A.CEL" ) ) ) ms <- studyDesignMultipleGroups( studyName = "E-MEXP-2296", factorName = "DiseaseState", levelNames = c("Control_EF.60.","Hyper_EF.50.","Hyper_lowEF.30."), numReplicates = rep(3, 3), sampleTable = sampleTable, assayTable = assayTable, assayTableName = "RNA" )

insert(ms, mp, overwrite = TRUE)

mw <- MiodinWorkflow(name="Microarray workflow") mw <- mw +

downloadRepositoryData( name = "RNA downloader", repository = "ArrayExpress", accession = "E-MEXP-2296", path = "datasets/E-MEXP-2296")+ workflowConfiguration( affyUser = getPass::getPass("User"), affyPass = getPass::getPass("Password"))+ importMicroarrayData(

(34)

34 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 chipName = "HumanGenomeU133A",

#unzip annotationfile in downloads manually #probeAnnotationFile = ""

)%>%

processMicroarrayData(name ="RNA processor",

contrastName = "Hyper_EF.50.VsControl_EF.60.", datasetName =

"Hyper_EF.50.VsControl_EF.60.")%>% performOmicsModeling(

name = "RNA analyser",

contrastName = "Hyper_EF.50.VsControl_EF.60.") mw <- mw +

processMicroarrayData(name ="RNA processor 2",

upstreamModule = "RNA importer", contrastName = "Hyper_lowEF.30.VsControl_EF.60.", filterLowExp = TRUE, filterLowExpThresh = 10, datasetName = "Hyper_lowEF.30.VsControl_EF.60.")%>% performOmicsModeling(

name = "RNA analyser 2",

contrastName = "Hyper_lowEF.30.VsControl_EF.60.") mshow(mw) mw <- insert(mw, mp) mw <- execute(mw) saveDataFile(mp) exportAll(mp, "analysisResult") mshow(mp)

#process HG-U133B genechip library("miodin")

mp <- MiodinProject( name = "Thesis",

author = "Benyapa Tangruksa", path ="."

)

mshow(mp)

sampleTable <- data.frame(

SampleName = paste0("S", 1:9),

SamplingPoint = paste0("sp", rep(1,9)), DiseaseState = rep(c("Hyper_EF.50.","Hyper_lowEF.30.","Control_EF.60."), each = 3), stringsAsFactors = FALSE ) assayTable <- data.frame( SampleName =paste0("S", 1:9), DataFile= paste0("datasets/E-MEXP-2296/",

c("12_133B.CEL", "14_133B.CEL", "16_133B.CEL", "11_133B.CEL", "7_133B.CEL", "8_133B.CEL", "10_133B.CEL", "18_133B.CEL", "19_133B.CEL" )

) )

(35)

35 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 studyName = "E-MEXP-2296", factorName = "DiseaseState", levelNames = c("Control_EF.60.","Hyper_EF.50.","Hyper_lowEF.30."), numReplicates = rep(3, 3), sampleTable = sampleTable, assayTable = assayTable, assayTableName = "RNA" )

insert(ms, mp, overwrite = TRUE)

mw <- MiodinWorkflow(name="Microarray workflow") mw <- mw +

downloadRepositoryData( name = "RNA downloader", repository = "ArrayExpress", accession = "E-MEXP-2296", path = "datasets/E-MEXP-2296")+ workflowConfiguration( affyUser = getPass::getPass("User"), affyPass = getPass::getPass("Password"))+ importMicroarrayData(

name = "RNA importer", platform = "affymetrix", dataType = "rna", studyName= "E-MEXP-2296", assayName = "RNA", datasetName = "E-MEXP-2296", contrastName = "Hyper_EF.50.VsControl_EF.60.", chipName = "HumanGenomeU133B")%>%

processMicroarrayData(name ="RNA processor",

contrastName = "Hyper_EF.50.VsControl_EF.60.", datasetName =

"Hyper_EF.50.VsControl_EF.60.")%>% performOmicsModeling(

name = "RNA analyser",

contrastName = "Hyper_EF.50.VsControl_EF.60.") mw <- mw +

processMicroarrayData(name ="RNA processor 2",

upstreamModule = "RNA importer", contrastName = "Hyper_lowEF.30.VsControl_EF.60.", filterLowExp = TRUE, filterLowExpThresh = 10, datasetName = "Hyper_lowEF.30.VsControl_EF.60.")%>% performOmicsModeling(

name = "RNA analyser 2",

(36)

36

Appendix B – distribution plots of microarray heart biopsy data

Figure B1. Distribution plot of the normalized mRNA microarray heart biopsies data

Appendix C – Heatmaps of microarray heart biopsy data

Figure C1. Heatmap showing sample clusters of the normalised mRNA microarray heart biopsies data

(37)

37

Appendix D – PCA plot of microarray heart biopsy data

(38)

Appendix E – Biomarker candidates: lenient parameter (early-progression)

Table 1C. lenient biomarker candidates involved in early progression cardiac hypertrophy categorised by protein family

Symbol Entrez Gene Name Location Drug(s) Blood Urine Biomarker Application(s) Enzyme

TGM3 transglutaminase 3 Cytoplasm N/A x x unspecified

CP ceruloplasmin Extracellular Space

N/A x x efficacy

G-protein coupled receptor CXCR4 C-X-C motif

chemokine receptor 4

Plasma Membrane

USL311, PF-06747143, burixafor, LY-2510924, mavorixafor,

cladribine/cytarabine/filgrastim/idarubicin/plerixafo r, [68Ga]pentixafor, BL-8040, GMI-1359,

ulocuplumab, plerixafor, balixafortide, filgrastim/plerixafor x diagnosis GRM1 glutamate metabotropic receptor 1 Plasma Membrane fasoracetam x efficacy Growth factor EREG epiregulin Extracellular

Space

N/A prognosis,response to therapy

PDGFB platelet derived growth factor subunit B

Extracellular Space

(39)

39 Ion channel CFTR CF transmembrane conductance regulator Plasma Membrane

ivacaftor/tezacaftor, tezacaftor, lumacaftor, ivacaftor/lumacaftor, crofelemer, ivacaftor

x diagnosis,disease progression

Kinase LRRK2 leucine rich repeat

kinase 2

Cytoplasm N/A x diagnosis

MAP2K6 mitogen-activated

protein kinase kinase 6

Cytoplasm N/A prognosis

STC1 stanniocalcin 1 Extracellular Space

N/A x x diagnosis

phosphatase DUSP6 dual specificity

phosphatase 6

Cytoplasm N/A x diagnosis

DUSP4 dual specificity phosphatase 4

Nucleus N/A diagnosis

transcription Regulator ANKRD2 ankyrin repeat

domain 2

Nucleus N/A unspecified application

ETV4 ETS variant transcription factor 4

(40)

40

RUNX1 RUNX family transcription factor 1

Nucleus N/A x unspecified application

SOX9 SRY-box

transcription factor 9

Nucleus N/A disease progression,prognosis

Transmembrane receptor SFRP2 secreted frizzled related protein 2 Plasma Membrane N/A diagnosis SFRP5 secreted frizzled related protein 5 Plasma Membrane N/A diagnosis Transporter PDYN prodynorphin Extracellular

Space

N/A unspecified application

SLC7A5 solute carrier family 7 member 5

Plasma Membrane

N/A x diagnosis

Other IER3 immediate early

response 3

Cytoplasm N/A unspecified application

KRT8 keratin 8 Cytoplasm N/A x prognosis

KRT18 keratin 18 Cytoplasm N/A x efficacy

(41)

41 CCN1 cellular communication network factor 1 Extracellular Space

N/A x x unspecified application

NPPA natriuretic peptide A Extracellular Space

N/A x diagnosis,efficacy,prognosis,unspecified application

NPPB natriuretic peptide B Extracellular Space N/A x diagnosis,efficacy,prognosis,safety,unspecifie d application SERPINE 1 serpin family E member 1 Extracellular Space

drotrecogin alfa x diagnosis,disease

progression,efficacy,prognosis,safety

THBS1 thrombospondin 1 Extracellular Space

N/A x x diagnosis,efficacy

DNAJA4 DnaJ heat shock protein family (Hsp40) member A4

(42)

42

Appendix F – Biomarker candidates: lenient parameter (late-progression)

Table 1D. lenient biomarker candidates involved in the late-progression cardiac hypertrophy categorised by protein family

Symbol Entrez Gene Name Location Drug(s) Blood Cardiomyocytes Biomarker Application(s) LDHA lactate dehydrogenase

A

Cytoplasm N/A x efficacy,unspecified application

LPL lipoprotein lipase Cytoplasm lovastatin/niacin, nicotinic acid/pioglitazone, atorvastatin/niacin, nicotinic acid, tyloxapol x x efficacy,unspecified application GRM1 glutamate metabotropic receptor 1

Plasma Membrane fasoracetam x efficacy

STC1 stanniocalcin 1 Extracellular Space N/A x diagnosis

APLN apelin Extracellular Space N/A x unspecified application

CASQ2 calsequestrin 2 Cytoplasm N/A unspecified application

NPPA natriuretic peptide A Extracellular Space N/A x x diagnosis,efficacy,prognosis,unspecified application

PRSS12 serine protease 12 Extracellular Space N/A diagnosis

(43)

CARDIAC HYPERTROPHY IN HUMAN STEM CELLS-DERIVED CARDIOMYOCYTES