Integration of Bioinformatics Approaches and Experimental Validations to Understand the Role of Notch Signaling in Ovarian Cancer

Summary

Bioinformatics is a useful way to process large-scale datasets. Through the implementation of bioinformatics approaches, researchers can quickly, reliably, and efficiently obtain insightful applications and scientific discoveries. This article demonstrates the utilization of bioinformatics in ovarian cancer research. It also successfully validates bioinformatics findings through experimentation.

Abstract

Notch signaling is a highly conserved regulatory pathway involved in many cellular processes. Dysregulation of this signaling pathway often leads to interference with proper development and may even result in initiation or progression of cancers in certain cases. Because this pathway serves complex and versatile functions, it can be studied extensively through many different approaches. Of these, bioinformatics provides an undeniably cost-efficient, approachable, and user-friendly method of study. Bioinformatics is a useful way to extract smaller pieces of information from large-scale datasets. Through the implementation of various bioinformatics approaches, researchers can quickly, reliably, and efficiently interpret these large datasets, yielding insightful applications and scientific discoveries. Here, a protocol is presented for integration of bioinformatics approaches to investigate the role of Notch signaling in ovarian cancer. Furthermore, bioinformatics findings are validated through experimentation.

Introduction

The Notch signaling pathway is a highly conserved pathway that is important for many developmental processes within biological organisms. Notch signaling has been shown to play a significant role in cell proliferation and self-renewal, and defects in the Notch signaling pathway can lead to many types of cancers^1,2,3,4,5,6. In some circumstances, the Notch signaling pathway has been linked to both tissue growth and cancer as well as cell death and tumor suppression⁷. Multiple Notch receptors (NOTCH 1−4) and co‒activator Mastermind (MAML 1−3), all with diverse functions, add an additional level of complexity. While the Notch signaling pathway is sophisticated in terms of functions, its core pathway is simple on a molecular basis⁸. Notch receptors act as transmembrane proteins composed of extracellular and intracellular regions⁹. A ligand binding to the extracellular region of Notch receptors facilitates proteolytic cleavage, which allows the Notch intracellular domain (NICD) to be released into the nucleus. NICD then binds to co‒activator Mastermind to activate downstream gene expression¹⁰.

In recent years, Notch signaling has been shown to play a variety of roles in the initiation and progression of several types of cancers across different species^6,11. For instance, Notch signaling has been linked to tumorigenesis involving the human NOTCH1 gene¹². Recently, the NOTCH2, NOTCH3, Delta-like 3 (DLL3), Mastermind‒like protein 1 (MAML1), and a disintegrin and metalloproteinase domain‒containing protein 17 (ADAM17) genes were shown to be strongly associated with ovarian cancer, especially with the poor overall survival of patients¹³.

As the amount of experimental and patient-associated data continuously increases, the demand for analysis of the available data increases as well. The available data are scattered across publications, and they may deliver inconsistent or even contradictory findings. With the development of new technology in recent decades, such as next-generation sequencing, the amount of available data has grown exponentially. Although this represents rapid advancements in science and opportunities for continued biological research, assessing the meaning of publicly available data to solve research questions is a great challenge¹⁴. We believe bioinformatics is a useful way to extract smaller pieces of information from large-scale datasets. Through the implementation of various bioinformatics approaches, researchers can quickly, reliably, and efficiently interpret these large datasets, yielding insightful discoveries. These discoveries may range from the identification of potential new drug therapy targets or disease biomarkers, to personalized patient treatments^15,16.

Bioinformatics itself is rapidly evolving, and approaches are constantly changing as technological advances sweep medical and biological science. Currently, common bioinformatics approaches include the utilization of publicly accessible databases and software programs to analyze DNA or protein sequences, identify genes of particular relevance or importance, and determine the relevance of genes and gene products through functional genomics¹⁶. Although the field of bioinformatics is certainly not limited to these approaches, these are significant in helping clinicians and researchers manage biological data for the benefit of patients as a whole.

This study aims to highlight several important databases and their use for research about the Notch signaling pathway. NOTCH2, NOTCH3, and their co‒activator MAML1 were used as examples for the database study. These genes were used because the importance of the Notch signaling pathway in ovarian cancer has been validated. Systematic analyses of retrieved data confirmed the importance of Notch signaling in ovarian cancer. In addition, because Notch signaling is well conserved across species, it was confirmed that overexpression of Drosophila melanogaster NICD and Mastermind together can induce tumors in Drosophila ovaries, supporting the database findings and the significant and conserved role of Notch signaling in ovarian cancer.

Protocol

1. Prediction of Clinical Outcomes from Genomic Profiles (PRECOG)

NOTE: The PRECOG portal (precog.stanford.edu) accesses publicly available data from 165 cancer expression datasets, including gene expression levels and patient clinical outcomes¹⁷. It specifically provides the Meta‒Z analysis, which incorporates large datasets to provide Z‒scores of different genes in 39 cancer types to indicate patient overall survival. Poor and good survival rates are indicated by positive and negative Z‒score values, respectively.

1. Create an account with an academic affiliated email to access this database. Enter the email address and password associated with the account.
2. Click on the View Details button located underneath the Meta-Z analysis heading.
3. Input the gene of interest into the Search bar.
4. Use the scroll bar located on the bottom of the screen to obtain the survival Z-score for the specific cancer type of interest.

2. CSIOVDB

NOTE: CSIOVDB (csibio.nus.edu.sg/CSIOVDB/CSIOVDB.html) is a microarray database developed by the Cancer Science Institute of Singapore to study ovarian cancer¹⁸. This database contains data of carcinomas from different tumor sites as well as normal ovary tissue data. In addition, CSIOVDB provides Kaplan‒Meier survival plots to assess patient survival with differential gene expression levels. CSIOVDB can be applied to investigate the association between gene expression levels and ovarian cancer stages/grades.

1. Input gene of interest, then click the Search button.
2. Click on the Disease State tab.
   NOTE: This tab provides summary statistics of gene expression of the target gene of interest in ovarian cancer disease states.
3. Click on the Histology tab.
   NOTE: This tab provides summary statistics of gene expression of the target gene of interest in major ovarian cancer histologies.
4. Click on the Clinico-pathological Parameters tab.
   NOTE: This tab provides a comparison of the gene expression levels among different ovarian cancer stages, grades, and clinical responses with Mann-Whitney tests.
5. Click on the Survival tab.
   NOTE: This tab provides Kaplan-Meier plots associated with Overall Survival and Disease-Free Survival. For this database, disease-free survival is considered progression- and recurrence-free survival¹⁸. Multivariate analyses for Overall Survival and Disease-Free Survival are also found under this tab. The multivariate analyses compare features that relate to ovarian cancer prognoses (stage, grade, surgical debulking, histology, age) and the gene of interest.
6. Click on the Subtype tab.
   NOTE: This tab provides summary statistics and Mann-Whitney tests for the expression level of the gene of interest in molecular subtypes of ovarian cancer. This tab also provides Kaplan-Meier plots associated with Overall Survival and Disease-Free Survival of the gene of interest in molecular subtypes of ovarian cancer.

3. Gene Expression across Normal and Tumor tissue (GENT)

NOTE: The GENT portal (medical‒genome.kribb.re.kr/GENT) is developed and maintained by the Korea Research Institute of Bioscience and Biotechnology (KRIBB)¹⁹. It collects 16,400 (U133A; 241 datasets) and 24,300 (U133plus2; 306 datasets) publicly available samples. After standardization, GENT offers gene expression data across diverse tissues, which are further divided into tumor and normal tissues.

1. Click on the Search tab at the top of the screen.
2. In the section labeled 1. Keyword, select the Gene symbol for the Terms from the dropdown menu, input the gene symbol of the gene of interest in the blank area of the Keyword section, and select Tissue for the Type option.
3. Click the Search button at the bottom of the 1. Keyword section. It shows the summary graphs of gene expression in normal and tumor tissues of different cancer types based on the U133A and U122Plus2 platforms.
   NOTE: It is optional to select the Data Filtering option on the top of the summary graph to single out a particular database to study.
4. Click the link next to Result Data Download to access the detailed information about the gene expression values, tissue types, and data sources.

4. Broad Institute Cancer Cell Line Encyclopedia (CCLE)

NOTE: CCLE (portals.broadinstitute.org/ccle) was created by the Broad Institute and provides genomic profiles and mutations of 947 human cancer cell lines²⁰.

1. Input the desired genes into the search bar and then click the Search button.
2. In the section labeled Select Dataset, click the mRNA expression (RNAseq) option from the dropdown menu.
   NOTE: Other options include mRNA expression (Affy), Achilles shRNA knockdown, and Copy Number.
3. Click on the Toggle All Traces button. Select the tissue type of interest from the gray box on the right. Scroll down to the bottom of the screen and click the Download mRNA expression button.
4. Open the downloaded text document. Copy and paste all the text into Sheet 1. Copy all the text in the Sheet 1.
5. Click on the sheet in the spreadsheet software Sheet 2 tab on the bottom of the spreadsheet. Right click on the A column, select Paste Special, and then select the Transpose option in Sheet 2.
6. Once the text is transposed into two columns on Sheet 2, click the dropdown arrow for the Sort & Filter option heading and then select the Filter option. An arrow will appear in the heading area labeled Gene. Click on the arrow and type in the tissue type of interest.
   NOTE: This step will filter all the data and only display gene expression levels for the tissue type of interest.

5. cBioPortal

NOTE: cBioPortal (www.cioportal.org) was developed at the Memorial Sloan Kettering Cancer Center (MSK), and accesses, analyzes, and visualizes large scale cancer genomic data^21,22. Specifically, this portal allows researchers to search for genetic alterations and signaling networks.

1. Using the query on the landing page, click the organs/tissues of interest under the section labeled Select Studies. Select the particular study of interest, then hit the Query By Gene button.
2. In the section labeled Select Genomic Profiles, select from the three options: Mutations, Putative copy-number alterations from GISTIC, or mRNA Expression. Further select corresponding data from the dropdown menu for Select Patient/Case Set.
3. Enter the target gene symbol(s) in the query box of Enter Genes. Click the Submit Query button.
4. Click on the Network tab at the top of the page to retrieve the desired gene network.
   NOTE: The signaling network is color-coded. The inputted genes are indicated by seed nodes with a thick border. Each gene is represented by a red circle, and the color intensity of the red circle reflects its mutation frequency. Genes are connected by differently colored lines. Brown lines mean "In Same Component", indicating the involvement in the same biological component. Blue lines mean "Reacts With", indicating gene reactions. Green lines mean "State Change", suggesting that one gene might cause a state change of another gene.
5. Click on the File tab at the top of the image to choose Save as Image (PNG) for network image downloading.

6. Dissection of Drosophila with desired genotypes and DAPI staining

NOTE: Collect the female Drosophila with the desired genotypes, then dissect the fly ovaries to undergo the procedures of DAPI staining for imaging.

1. Prepare fly stocks tj-Gal4, Gal80ts/CyO; UAS-NICD-GFP/TM6B, w*; UAS-mam.A; and w[1118] to create flies with NICD-overexpression (tj-Gal4, Gal80ts/+; UAS-NICD-GFP/+) and NICD and mam-overexpression (tj-Gal4, Gal80ts/UAS-mam.A; UAS-NICD-GFP/+) capability.
2. Apply the temporal and regional gene expression targeting (TARGET) technique to control spatiotemporal gene expression²³. Raise flies at 18 °C until adulthood, then shift to 29 °C for 48 h with yeast before dissection.
   NOTE: tj-Gal4 can only drive UAS expression under higher temperatures, when the inhibition by Gal80ts is relieved. The addition of yeast prior to dissection enlarges the ovaries for harvesting.
3. Place 3 mL of 1x phosphate-buffered saline (PBS) (137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, 1.8 mM KH₂PO₄) in an embryo collection dish. Use a CO₂ pad to anesthetize the flies.
4. Choose a female fly, then carefully grab the lower thorax of the fly using a pair of dissecting forceps and submerge it into the 1x PBS solution in an embryo collection dish. Use a second pair of forceps to pinch the lower abdomen and pull gently to release the internal organs.
5. Identify and detach the pair of ovaries from the fly body. Break the muscular sheath located at the posterior end of the ovaries and separate the ovarioles.
   NOTE: Separating the ovarioles and breaking the muscular sheath is required in order to achieve higher quality staining results.
6. Place the ovaries in a 1.5 mL centrifuge tube that contains 500 μL of 1x PBS. The tube should remain on ice until all ovaries are collected.
7. Remove the 1x PBS and place 0.5 mL of fix solution (4% formaldehyde) into the tube. Place the tube on the nutator for 10 min.
8. Remove the fix solution from the tube and dispose of it in a suitable waste container. Use 1 mL of 1x PBT (1x PBS supplemented with 0.4% Triton™ X-100) to wash the ovaries 3x for 15 min.
9. Discard the final PBT wash and add 1 mL of PBTG (0.2% bovine serum albumin, 5% normal goat serum in 1x PBT) to prevent nonspecific binding.
   NOTE: This step could be skipped for DAPI staining, but it is essential for antibody staining. Detailed immunohistochemistry staining can be found in Jia et al.²⁴.
10. Place 150 μL of DAPI (10 μg/mL) in the tube for 10−15 minutes nutation. Discard the DAPI and wash the ovaries 1x for 10 min using 1 mL of 1x PBT. Remove the PBT and wash 2x for 10 min using 1x PBS.
11. Remove excess PBS until approximately 300 μL of PBS remains in the tube with the ovaries. Pipet the ovaries up and down several times using a 200 μL pipette, in order to free the egg chambers.
12. Gently spin down the tube and carefully remove as much 1x PBS solution as possible without removing the ovaries. Place 120 μL of mounting solution (1 g n-propyl gallate, 5 ml of 10X PBS, 40 ml of glycerol and 5 ml of dH₂O) into the tube.
    NOTE: Mounting solution is sticky, so it is difficult to transfer exactly 120 μL of mounting solution into a tube. To alleviate this issue, a 1,000 μL pipette tip can be used to add three drops of mounting solution into the tube.
13. Remove approximately 0.33 mm from a 200 μL pipette tip and use the newly cut pipette tip to place the mounting solution on a microscope glass slide.
14. Gently place the coverslip glass on the mounting solution and seal the edges of the cover slip with transparent nail polish.
    NOTE: Sealing the edges of the cover glass is needed to prevent the egg chambers from flowing inside of the mounting solution when taking confocal images.
15. Acquire images with a confocal microscope using the following settings: objective lens = 10x magnification; numerical aperture = 0.8; DAPI emission wavelength = 410−513 nm.

Results

Using the procedure mentioned in step 1 using the PRECOG portal, the Z-scores of NOTCH2, NOTCH3, and MAML1 in ovarian cancer were obtained (1.3, 2.32, 1.62, respectively). The negative Z‒score values indicate the poor overall survival of patients with high expression levels of the three genes. Using Conditional Formatting of the spreadsheet software, the Z‒score values are shown in a colored bar graph in Figure 1.

The CS...

Discussion

As there are countless approaches and methods for the utilization of bioinformatics, there are numerous databases available online to the general public. An abundance of information can be extracted from each of these databases, but some are best suited for particular purposes, such as assessing patient survival based on certain inputs. Systematic analyses of retrieved data from different individual databases can convincingly yield important scientific findings.

The current analysis focuses on...

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
DAPI (4',6-Diamidino-2-Phenylindole, Dihydrochloride)	Invitrogen	D1306	1:1000 Dilution
PBS, Phosphate Buffered Saline, 10X Powder, pH 7.4	ThermoFisher	FLBP6651	Dissolved with ddH2O to make 1X PBS
Goat serum	Gibco	16210064	Serum
Embryo dish	Electron Microscopy Sciences	70543-45	Dissection Dish
Nutating mixers	Fisherbrand	88861041	Nutator
tj-Gal4, Gal80ts/ CyO; UAS-NICD-GFP/ TM6B	Dr. Wu-Min Deng at Florida State University	N/A	Fly stock
w*; UAS-mam.A	Bloomington Drosophila Stock Center	#27743	Fly stock
w[1118]	Bloomington Drosophila Stock Center	#5905	Fly stock
The PRECOG portal	Stanford University	precog.stanford.edu	Publicly accessible database of cancer expression datasets
CSIOVDB	Cancer Science Institute of Singapore	csibio.nus.edu.sg/CSIOVDB/CSIOVDB.html	Microarray database used to study ovarian cancer
The Gene Expression across Normal and Tumor tissue (GENT) Portal	Korea Research Institute of Bioscience and Biotechnology (KRIBB)	medical–genome.kribb.re.kr/GENT	Publicly accessible database of gene expression data across diverse tissues, divided into tumor and normal tissues.
Broad Institute Cancer Cell Line Encyclopedia (CCLE)	Broad Institute and The Novartis Institutes for BioMedical Research	portals.broadinstitute.org/ccle	Provides genomic profiles and mutations of human cancer cell lines
cBioPortal	Memorial Sloan Kettering Cancer Center (MSK)	cioportal.org	Portal that allows researchers to search for genetic alterations and signaling networks
Zeiss 710 Inverted confocal microscope	Carl Zeiss	ID #M 210491	Examination and image collection of fluorescently labeled specimens