# The genetic relationship of SIRT Sequences with Tumor Growth and Longevity

By Samantha Leano, Stephanie Frohrip, Christopher Wilson

[TOC]

## Introduction
In recent times, from yeast to humans, sirtuins (Silent Information Regulator Transcription) are known to either prolong life spans or cause cancer. Despite being homologous, these seven SIRT genes (SIRT:1,2,3,4,5,6,7), each play a specific role in aging, transcription, apoptosis, inflammation, and stress resistance (Polito, L, et al., 2010, 216). SIRT is highly conserved in the mammalian genome, and is a transcription factor for several different processes. It is a nicotinamide adenine dinucleotide (NAD+) dependent histone deacetylase, and also is a target for several proteins. It is expressed virtually everywhere in the body, including yet not limited to the brain, heart, kidney, and liver. Even though these genes are commonly known as longevity genes, they are also known to affect a variety of other processes, as sirtuin plays roles in a multitude of diseases. By regulating downstream pathways, SIRT affects a range of age-related, metabolic, and cardiovascular diseases. In such diseases (Parkinsons, diabetes, etc), SIRT is found in decreased levels, compared to standard. In comparison, an overexpression of SIRT results in increased cell viability and decreased cell apoptosis.
Figure 1. A representation of the a SIRT protein.
https://en.wikipedia.org/wiki/Sirtuin#/media/File:1SZD.png

### SIRTs Associated with Longevity
Sirtuins were first investigated for their involvement in calorie restriction (CR), which involved the reduction of daily food intake. Researchers had tested SIRT1, and found out that mice that didn’t have the gene had a shorter lifespan than their wild-type (Polito, L, et al., 2010, 215). The experiment suggested SIRT prolonged life expectancy and delayed age-associated disorders. It was soon later discovered that SIRT1,3,4, and 5, are all associated with longevity. Further testing has been done but not yet concluded in the scientific community on whether common allelic variations of SIRT genes are associated with longevity in humans.

### SIRTs Relationship with Tumor Growth
After a thorough investigation of all SIRT proteins, it was noticed that SIRT, nicknamed a "Master Metabolic Regulator", controlled many of the same traits that influence cancer. SIRT affects DNA repair, transcriptional regulation, metabolism, aging, and senescence, which all heavily affects cancer and its severity. By searching thorough several databases, we discovered some that correlated cancer expression with SIRT genes thorough the Human Protein Atlas. One is a comprehensive list of all the SIRT genes (SIRT.tsv) that contains gene ID, gene description, chromosome position, molecular function, and RNA cancer specificity score. The other contains pathology information (pathology.tsv.zip) about SIRT genes, which includes Gene ID, mRNA expression, and patient mortality.

### Our Study
For this project, we will look into the roles of each SIRT gene and the possible genetic roles they may play in cancerous tumor formation. In addition to understanding the function of all SIRT genes, specifically, we will focus more on the variants that have an association to both tumor growth and longevity. By focusing our understanding on sirtuins, we may be able to quickly identify the mutations of different SNPs associated with different cancers to understand why SIRT genes are associated with both longevity and cancerous diseases.

## Methods

### Databases We’ve used

### *SIRT_rfn.tsv*
#### Data Acquisition
Our first database can be found in the human protein atlas website. In this database in particular, we are looking at the general description of all SIRT genes (SIRT.tsv). This dataset consists of: the gene name, gene description, location of the chromosome and the position of that gene, the protein class, biological class, and evidence indicating which disease it’s involved with.
Using R, we modified the dataset to filter out the columns that are not useful to us, and remove the RPS19BP1 protein. This will give us our first modified dataset (SIRT_rfn.tsv). Once we have our modified dataset, we could see that there is a lot of unfilled data written as (NA). Since these genes are still being studied and researched, we expected to not have some of the data.

# Opening up tidyverse package 
library(tidyverse)

# Reading the SIRT.tsv file as SIRT_data
SIRT_data <- read_tsv(file='SIRT.tsv')

# Selecting important columns: 1,3, and 4 are all the name and gene IDs for all SIRT genes. 
# 6 though 12 are the locations and brief functions of SIRT genes, 16 and 17 are cancerous tissues that are detected on an RNA level   
beta_filter <- select(SIRT_data,1,3,4,6:12,16,17)

# Removing the bottom row to just having SIRT genes
beta_filter <- filter(beta_filter,Gene %in% c('SIRT1','SIRT2','SIRT3','SIRT4','SIRT5','SIRT6','SIRT7'))

# Double check we have everything
view(beta_filter)

# Saving beta_filter as the information for SIRT reference
write_tsv(beta_filter,file = 'SIRT_rfn.tsv')

Preview of SIRT_rfn.tsv

#### Data Integration
However, there is a lot we can look at based off of our SIRT_rfn data. This dataset will be useful as we can utilize UCSC Genome Browser to locate the positions of each SIRT gene and see different variations within the human genome. What’s interesting to point out is that even though this table suggests that SIRT3 is not involved with any disease, we know based off of other research done with this gene, that SIRT3 is associated with both longevity and cancer.

### *SIRT_path.tsv*
#### Data Acquisition
Our next database can also be found in the human protein atlas website. Pathology data (pathology.tsv) is a dataset that delves into the staining profiles for proteins in human tumor tissues and survival rate of those patients. The dataset contains: the gene name and ID, the type of cancer it produces, the level of proteins present with the amount of stains done on the sample, and p-values for patient survival and mRNA correlation.
We modified it to SIRT genes and to where the disease has been diagnosed and could either be cured or not cured. The reason why we took out the unfavorable columns are due to the uncertainty if that sample was diagnosed with a specific cancer. We want the SIRT genes that have p-values in categories where a patient that does have a known cancer could survive or not.

# Opening up tidyverse
library(tidyverse)

# Reading the pathology file
pathology_expression <- read_tsv(file= 'pathology.tsv')

# Manipulating pathology_expression
alpha_filter = pathology_expression

alpha_filter <- select(pathology_expression,1:8,10)

# Format the column names to be a singular string
colnames(alpha_filter)
colnames(alpha_filter)[2] <- 'Gene_name'
colnames(alpha_filter)[7] <- 'Not_Detected'


# Filtering out Gene names to just have SIRT genes
alpha_filter <- filter(alpha_filter,Gene_name %in% c('SIRT1','SIRT2', 'SIRT3', 'SIRT4', 'SIRT5', 'SIRT6', 'SIRT7'))

# Double checking
View(alpha_filter)

# Saving alpha_filter as the pathology data of SIRT
write_tsv(alpha_filter,file = 'SIRT_path.tsv')

Preview of SIRT_path.tsv
#### Data Integration
After manipulating the data, we could see that we could fill in missing data into our first dataset (SIRT_rfn). For example, SIRT3 is associated with kidney, uteral, and pancreatic cancer. This information has become useful and we will integrate this into our SIRT_rfn data and also other information that was gleaned from the second database (SIRT_rfn_2). Essentially, we have merged the two datasets into one.

# Inserting new data into the columns
beta_filter$`Disease involvement`<-c('Cancer-realted genes head liver pancreatic prostate','Cancer-related genes Neurodegeneration','Cancer-related genes Tumor suppressor','Cancer-realated thyroid breast glioma','Cancer-related skin ovarian lung','Renal,Carcinoid,Pancreatic','Disease variant')

# Double check we have everything
view(beta_filter)

# Saving beta_filter as the information for SIRT reference
write_tsv(beta_filter,file = 'SIRT_rfn_2.tsv')

With such a small p-value, we believe that there is some significance in the data that shows us that SIRT 3,4,5, and 6 have some favorable prognostics while SIRT 2 and 7 does not. This p-value agrees with other published research done on SIRT genes as SIRT2 and 7 are commonly associated with cancer genes (Polito, L., et al, 2010, 2015 table 1).

### *SIRT_location.tsv*
#### Data Acquisition
The final database we will be using to understand the function of SIRT genes is found also in the Human protein atlas dataset. Subcellular_location.tsv is a dataset that has: gene name, supported locations, single cell variation intensity, and GO ID.

# Opening tidyverse
library(tidyverse)

# Reading the tsv file as SIRT_location
SIRT_location <- read_tsv(file='subcellular_location.tsv')

#Manipulating the data
omega_filter = SIRT_location

# Selecting only the columns needed
omega_filter <- select(SIRT_location,1:4,8,14)

# Editing names
colnames(omega_filter)[2] <- 'Gene_name'

# Filtering only SIRT genes
omega_filter <- filter(omega_filter,Gene_name %in% c('SIRT1','SIRT2', 'SIRT3', 'SIRT4', 'SIRT5', 'SIRT6', 'SIRT7'))

# Double checking
view(omega_filter)

# Wiriting it as a tsv file
write_tsv(omega_filter,file = 'SIRT_location.tsv')

#### Data Integration
We modified the data to only have SIRT genes, main location, and the GO ID (SIRT_location.tsv). These columns are useful as we could use their GO IDs and run them into cytoscape. From there, this could help us find connections within the cell and see which SIRT genes are most commonly associated with a specific location within the cell. Since SIRT_location tells us the location on a subcellular basis, this can tie us back to SIRT_path and give us additional information as to how that cancer is formed. In addition, SIRT_path also ties us back to SIRT_rfn as we could check within those specific locations what type of variants formed and any nuances found there.

Preview of SIRT_location.tsv

#### All datasets mentioned can be found here: https://drive.google.com/drive/folders/1qcswEmAL8PblVyygKQecCZKrnjmskkhx?usp=sharing

## Results

### *UCSC Genome Browser (Part1)*
SIRT 1
SIRT 2
SIRT 3

SIRT 4
SIRT 5
SIRT 6
SIRT 7

For each of the seven SIRT genes, we intiated a search on the UCSC Genome Browser. By using the Human GRCh38/hg38 genome, and turning the setting to "show" under Cancer Gene Expression, the above graphs were produced.

Each SIRT gene has two associated graphs. The top graph is where each gene is expressed, and the level of expression. Several overlaps are seen. The second graph is the different cancers and their relativity for each gene.

### *UCSC Genome Browser (Cancer Gene Expression Track Part 2)*

We used the UCSC Genome Browser to garner more information about the relationship between the SIRT genes and cancer. Using the Cancer Gene Expression track, we were able to determine which types of cancer each SIRT gene was most highly expressed in if there was a significant margin. This showed us that none of the SIRT genes were more significantly expressed in any of the 33 cancer types of which data was available for. However, we were able to find that certain transcripts from each SIRT gene were found in higher abundance in particular cancer types, with the exceptions of SIRT 4 and 5.

UCSC Genome Browser with "squish" forms of the Cancer Gene Expression tracks. Shows 5 transcripts of SIRT 6 are more highly expressed in specific cancers.

### *BLAST and NCBI Orthologs*

We used both the BLAST feature the orthologs feature from NCBI. We used BLAST to search for other species in which the SIRT genes or similar genes are found. We had mixed results with BLAST however and did not learn anything helpful to our project. Primarily our BLAST of SIRT 3 only had predicted results, no actual ones. We then used the orthologs feature to see if we could obtain more useful data. When searching for orthologs of SIRT 3 we wound up with over a thousand results, including SIRT 1 and 2. These results varied more than expected, and surprisingly did not only appear in animals, as a few orthologs were found in different fungi as well.

### *Cytoscape*
Using cytoscape, SIRT_location.tsv was imported and uploaded to a network. "Gene" was selected as source node, "Main Location" as target node, and "GO ID" as interaction type. The resulting network is below.

As demonstrated in the network, the genes ENG00000077463, ENSG00000187531, and ENSG00000096717 (SIRT6, SIRT7, and SIRT1, respectively) are associated with the nucleoplasm. ENSG00000124523 (SIRT5) are associated with the mitochondria. ENSG00000068903 (SIRT2) is associated with cytosol and the nuceloli.

## Discussion

### UCSC Genome Browser part 1
In UCSC Genome Browser part 1, we searched each gene one by one and observed the results. As seen in the overlaps in the tracks in the topmost graphs for each SIRT, some genes are expressed multiple times. The higher the bar, the higher the median expression at that point. The bottom graphs for each SIRT gene indictates the levels of cancer expression for that gene. None of the graphs are particularly noteworthy, as each has around the same levels as expression. However, it can be seen that each varies wildy with each cancer type.

### UCSC Genome Browser part 2
We used the NCBI's BLAST on SIRT 3, as it is one of the SIRTs that research has shown to be associated with longevity, in an attempt to see if we could find similar genes in any animals known for their longevity. Unfortunately, this BLAST search did not turn up any useful information as our search excluded humans, and all of the results were predicted, and not actual published results.

### Cytoscape
Our Cytoscape results are interesting because we know that the nucleoplasm is associated with cell growth and proliferation. Which makes sense from our SIRT_path dataset, if mutated, SIRT 1, 6, and 7 could be the likely cause of most common cancers. The same thing can be said about SIRT5, if mutated, will cause the mitochondria to be dysfunctional and cause cancer. In addition, SIRT2, which causes tumors to grow and form–also that can be traced back to our SIRT_path dataset.

### Where We Can go From Here
For future expansion, we hope to investigate the function of specific SIRT genes in better detail. As per our results, it seems that SIRT genes are associated with cancer, especially SIRT 1, 2, 3, 6, and 7. However, the cancer connection with SIRT 4 and 5 bears further investigation. There are several reasons why the results on SIRT 4 and 5 may be unclear. The first, and most likely, is that SIRT 4 and 5 are connected to several other genes that may work in concert to perform function. To investigate this further, it might be beneficial to try knocking out SIRT 4 and 5 and observing how other genes perhaps might be affected. It is also possible that SIRT 4 and 5, just due to the nature of their function, simply might just not affect cancer in any way. SIRT 4 had no association in cytoscape, and SIRT 5 was the only gene expressed in the mitochondria. It’s possible that due to location (or lack of) there might not be any expression of the gene in the correct places to affect cancer.

Additionally, by understanding the potential SNPs that SIRT genes may have, and how they may affect cancer and longevity, we could hope to positively affect cancer patient prognosis and let them live longer. In the future, perhaps by using CRIPSR to alter certain parts of the gene, cancer treatments could be advanced significantly and people can live longer and healthier lives.

#### *Source*
Polito L, Kehoe PG, Forloni G, Albani D. The molecular genetics of sirtuins: association with human longevity and age-related diseases. Int J Mol Epidemiol Genet. 2010 Jun 20;1(3):214-25. PMID: 21537393; PMCID: PMC3076766.