Supplementary Material - University of Georgia



Supplementary Material

A. Data used in this study

ATP consumption data for turtle brain cortical cells and hepatocytes, frog skeletal muscle cell, naked mole rat cell, and Wistar rat thymocyte cells are collected from the published literature [pic]1-5. We have also collected the matching gene-expression data of these cells used in the same studies on proteins of the following biological processes: translation (GO0006412), sodium-potassium exchange ATPase activity (GO0005391), calcium transporting ATPase activity (GO0005388), gluconeogenesis (GO0006094), actin-activated ATPase activity (GO00030898) and urea genesis from the GO and KEGG databases [pic]6, 7. In addition, we have collected gene-expression data for the same six groups of proteins in the human breast epithelial cell, renal proximal tubule epithelial cell, B lymphocyte, endothelial cell, smooth muscle cell, breast cancer cell line, rat hippocampus cell and mouse brain, heart, lung and muscle cells as well as RNA-seq data of naked mole rats under both hypoxia and matching normoxia from the GEO database. Also we have collected gene-expression data for human breast tissues with and without BRCA mutations, inflammatory and normal colon tissue, precancerous colon tissue and TCGA RNA-seq data of colon carcinoma. The following table provides the detailed information about these datasets.

Supplementary Table 1: Gene expression data used in this study

|GEO ID |Species |Number of |Conditions |Description | |

| | |samples | | | |

|GSE3537 |Homo sapiens |69 |Hypoxia vs normoxia |Cell lines of human breast epithelial cell, renal proximal tubule epithelial| |

| | | | |cell, endothelial cell, and smooth muscle cell | |

|GSE4086 |Homo sapiens |4 |Hypoxia vs normoxia |Cell lines of human B lymphocyte | |

|GSE480 |Mus musculus |20 |Hypoxia vs normoxia |Mouse brain, heart, lung and muscle cells | |

|GSE3763 |Blind mole rats |12 |Hypoxia vs normoxia |Muscle tissue | |

|GSE1357 |Rattus norvegicus |24 |Hypoxia by ischemia |Hippocampus cell of hypoxia-sensitive and hypoxia-tolerant rat tissue | |

|GSE13671 |Homo sapiens |14 |BRCA1 mutation |Normal breast tissue with and without BRCA mutations | |

|GSE17072 |Homo sapiens |20 |BRCA1 mutation |Normal breast tissue with BRCA mutation from breast cancer patients | |

|GSE19338 |Mus musculus |24 |Normoxia |Mouse colon tissue with and without APC mutation | |

|GSE7307 |Homo sapiens |677 |Normoxia |Collection of 70 human cell types | |

|GSE30337 |Naked mole rats |13 |Hypoxia vs normoxia |Transcriptome sequencing of naked mole rat tissue | |

|GSE9649 |Homo sapiens |30 |Control vs Lactate/hypoxia |Human mammalian epithelial cells | |

|GSE29406 |Homo sapiens |12 |Control vs Lactate/hypoxia |MCF7 cells | |

|GSE11341 |Homo sapiens |23 |Hypoxia for extended hours vs |Human Lung endothelial and cardiac cells | |

| | | |normoxia | | |

|GSE4183 |Homo sapiens |15 |Precancerous vs normal |Colon inflammatory bowel diseases vs normal colon tissue | |

|GSE4183 |Homo sapiens |15 |Precancerous vs normal |Colon adenoma vs normal colon tissue | |

|GSE12391 |Homo sapiens |114 |Precancerouse vs cancer and |Dysplastic nevus, radial and vertical growth phase melanoma vs common | |

| | | |normal |melanocytic nevus | |

|GSE11223 |Homo sapiens |202 |Inflammatory vs normal |Ulcerative colitis inflamed colon tissue vs uninflamed colon tissue | |

|TCGA data archive |Homo sapiens |254 |Cancer vs normal |Colon carcionoma vs normal colon tissue | |

B. Definition of substantially reduced expression levels

We have used a non-parametric method introduced in [pic]8 to assess if a specified set of genes are differentially expressed under hypoxic versus normoxic conditions. Consider a data set D of expression levels of N genes collected on a set of samples. We rank its genes {g1, …, gN} in the increasing order of their expression levels averaged over all the samples. In the following analysis in this section and Supplementary Section C, we assess if a subset of genes S of D is differentially expressed under the above two conditions based on the overall change in the rankings of the genes of S in the whole ranking lists of D’s genes under the two conditions. Previous studies have shown that such a non-parametric method is highly stable and reliable for assessing changes in gene-expressions under different conditions and even across different organisms 9, 10.

For a target gene set S of D, define

[pic]

where [pic]. The expression level score (ELS) of S is defined as

[pic].

The ELS(S) value reflects the overall rankings of the expression levels of the genes in gene set S among all the genes in D. Let ELS(S, C) represent the expression-level score of S in the data set of condition C; we use the value

[pic]

to assess the altered overall expression levels of S under a hypoxic versus the matching normoxic condition.

C. A regression model for ATP-consumption reduction versus reduced expression levels

We have derived a linear regression model between the reduction percentages in ATP consumption in hypoxic condition, defined by

[pic],

and the reduced gene expressions ([pic]) of the six groups of proteins defined in Supplementary Section A using publicly available data for naked mole rats and hypoxia-tolerant rats measured using the [pic] values. From the available data, we noted that the ATP consumption reduction percentage [pic] is smaller and the [pic] value is larger for hypoxia-tolerant rats than those for naked mole rats. For the linear regression model, we assume [pic] when [pic]=0, namely the ATP consumption will not have significant changes when the expression level of ATP consumption genes remain at the same level. Then we estimated the parameters a and b in the linear model: [pic], which optimally match the collected [pic] and [pic] values for the two organisms and the assumption of [pic] when [pic]=0. The following gives the model parameters we derived:

Supplementary Table 2: Regression coefficient estimates

|[pic] |[pic] |[pic] |

|Translation |-5.056 |1.009 |

|Sodium-potassium exchange ATPase activity |-0.6026 |1.110 |

|Calcium transporting ATPase activity |-0.8246 |0.997 |

|Gluconeogenesis |-1.919 |1.002 |

|Urea genesis |-1.012 |0.995 |

|Actin-activated ATPase activity |-0.979 |1.004 |

We then applied this model to the [pic] values estimated for human, mouse and rats based on their gene expression data under hypoxic versus normoxic conditions as in Supplementary Section B. Figure 1 of the main text shows the estimated reduction percentages in ATP consumption values for these organisms. In the analysis, we assumed that the proportions of the ATP consumption by each relevant process are comparable across all the species examined.

Based on our literature search, turtle and frogs are known to substantially repress their ATP consumptions during hypoxia. It is also known that blind mole rats have a larger reduction than that of hypoxia-tolerant rats and naked mole rats and further than those of human, mouse and hypoxia-sensitive rats. These qualitative data supports the data shown in Figure 1 in the main text.

D. Validation of the ATP reduction regression model

Here we provide a rationale as well as supporting evidence for our regression model for prediction of reduced ATP consumption based on the reduced gene expression levels of the relevant proteins.

1. The foundation for our regression model to be meaningful is that there is a strong correlation between the reduced gene expression levels of the six groups of proteins (see Supplementary Section A) and the reduced activity levels of these proteins. Only when such correlation exists will the derived regression model be statistically meaningful. To see this, note that it has been well established that there is a strong correlation between changes in protein abundance and in the mRNA level of the corresponding gene 11 in general. For these six groups of proteins, we note that the majority of their mRNA and proteins have relatively long half-lives, which is one essential factor for highly correlated mRNA expression and protein abundance 12, confirming that the above general observation on mRNA and protein abundance levels applies to these six groups of proteins. In addition, Michaelis–Menten kinetics directly indicates that increased enzyme abundance will result in increased enzymatic activities 13.

2. The relative order among the eight organisms under consideration in terms of their reduction percentage in ATP consumptions predicted by our model is consistent with a number of published data, hence providing supporting evidence that our model is meaningful. For example, the predicted reduction in ATP demand by naked mole rats is smaller than that by blind mole rats, which is smaller than those by frog and turtle. It is known that the naked mole rats typically require a minimum level of 10% oxygen in its habitat while blind mole rats can stay viable at 3% oxygen while turtles and frogs can survive months of anoxia ( ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download