Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Meta-Analysis
. 2020 Nov 9:9:e59980.
doi: 10.7554/eLife.59980.

Integration of genomics and transcriptomics predicts diabetic retinopathy susceptibility genes

Andrew D Skol  1 , Segun C Jung  2 , Ana Marija Sokovic  3 , Siquan Chen  4 , Sarah Fazal  4 , Olukayode Sosina  5   6 , Poulami P Borkar  3 , Amy Lin  3 , Maria Sverdlov  3 , Dingcai Cao  3 , Anand Swaroop  6 , Ionut Bebu  7 ; DCCT/EDIC Study group; Barbara E Stranger  8 , Michael A Grassi  3
Affiliations
Meta-Analysis

Integration of genomics and transcriptomics predicts diabetic retinopathy susceptibility genes

Andrew D Skol et al. Elife. .

Abstract

We determined differential gene expression in response to high glucose in lymphoblastoid cell lines derived from matched individuals with type 1 diabetes with and without retinopathy. Those genes exhibiting the largest difference in glucose response were assessed for association with diabetic retinopathy in a genome-wide association study meta-analysis. Expression quantitative trait loci (eQTLs) of the glucose response genes were tested for association with diabetic retinopathy. We detected an enrichment of the eQTLs from the glucose response genes among small association p-values and identified folliculin (FLCN) as a susceptibility gene for diabetic retinopathy. Expression of FLCN in response to glucose was greater in individuals with diabetic retinopathy. Independent cohorts of individuals with diabetes revealed an association of FLCN eQTLs with diabetic retinopathy. Mendelian randomization confirmed a direct positive effect of increased FLCN expression on retinopathy. Integrating genetic association with gene expression implicated FLCN as a disease gene for diabetic retinopathy.

Keywords: LCLs; diabetic retinopathy; eQTL; folliculin; gene expression; genetics; genomics; human; mendelian randomization.

Plain language summary

One of the side effects of diabetes is loss of vision from diabetic retinopathy, which is caused by injury to the light sensing tissue in the eye, the retina. Almost all individuals with diabetes develop diabetic retinopathy to some extent, and it is the leading cause of irreversible vision loss in working-age adults in the United States. How long a person has been living with diabetes, the extent of increased blood sugars and genetics all contribute to the risk and severity of diabetic retinopathy. Unfortunately, virtually no genes associated with diabetic retinopathy have yet been identified. When a gene is activated, it produces messenger molecules known as mRNA that are used by cells as instructions to produce proteins. The analysis of mRNA molecules, as well as genes themselves, can reveal the role of certain genes in disease. The studies of all genes and their associated mRNAs are respectively called genomics and transcriptomics. Genomics reveals what genes are present, while transcriptomics shows how active genes are in different cells. Skol et al. developed methods to study genomics and transcriptomics together to help discover genes that cause diabetic retinopathy. Genes involved in how cells respond to high blood sugar were first identified using cells grown in the lab. By comparing the activity of these genes in people with and without retinopathy the study identified genes associated with an increased risk of retinopathy in diabetes. In people with retinopathy, the activity of the folliculin gene (FLCN) increased more in response to high blood sugar. This was further verified with independent groups of people and using computer models to estimate the effect of different versions of the folliculin gene. The methods used here could be applied to understand complex genetics in other diseases. The results provide new understanding of the effects of diabetes. They may also help in the development of new treatments for diabetic retinopathy, which are likely to improve on the current approach of using laser surgery or injections into the eye.

PubMed Disclaimer

Conflict of interest statement

AS, SJ, AS, SC, SF, OS, PB, AL, MS, DC, AS, IB, BS, MG No competing interests declared

Figures

Figure 1.
Figure 1.. Response to glucose.
(a) Volcano plot summarizing transcriptional response to glucose for all 22 individuals (RGAll consisting of nDM, nDR, and PDR individuals). Each point represents a single gene. Red indicates genes showing a differential response (FDR < 0.05; log10 >1.3 represented by the dotted line) and an absolute log2FC >0.17. Adj p-value is false discovery rate (FDR). FC indicates expression fold change with positive values indicating higher expression in the high glucose condition relative to the standard condition. Source and code files for this plot are available in Figure 1—source data 1 and Source code 1. An additional source file can be found in Gene Expression Omnibus (GEO) at https://www.ncbi.nlm.nih.gov/geo/ under accession code GSE146615. (b) QQ (quantile-quantile) plot plot summarizing GSEA of transcriptional response to glucose in all 22 individuals. Pathways are classified as upregulated (red) or downregulated (blue) in response to glucose. Only significant GO categories (FDR < 0.1%) are labeled. Red line indicates the null expectation. Source and code files for this plot are available in Figure 1—source data 2, 3, 4, and 5 and Source code 2.
Figure 1—figure supplement 1.
Figure 1—figure supplement 1.. p-value distribution for transcriptional response to glucose in all 22 individuals (RGAll) (no diabetes, nDR, and PDR).
Plotted are limma-derived differential expression p-values for 11,548 genes. The dashed line represents the expected null distribution. Source and code files for this plot are available in Figure 1—source data 6 and Source code 3.
Figure 1—figure supplement 2.
Figure 1—figure supplement 2.. Intra- and interindividual transcriptome variation in high glucose treatment.
Intraindividual variation in the transcriptome was quantified among biological replicates and compared to interindividual variation, for individual genes (2,3), and comparing the two distributions of all genes (3,4). Source and code files for these plots are available in Figure 1—source data 7 and 8, and Source code 4.
Figure 1—figure supplement 3.
Figure 1—figure supplement 3.. QQ plot demonstrating a significant shift away from the null represented by the red line of no difference between individuals as determined by assessing intra-individual variance from the biologic replicate samples.
Figure 1—figure supplement 4.
Figure 1—figure supplement 4.. Plot showing distribution of inter vs intra-individual variance.
Figure 2.
Figure 2.. Differential transcriptional response to glucose among individuals with diabetes with and without retinopathy.
Volcano plot summarizing genes exhibiting a differential response to glucose between individuals with diabetes with and without retinopathy (RGPDR–nDR). The difference in FC between groups is represented on the X-axis and p-value of this difference on the Y-axis. Red indicates the 103 genes showing the most differential expression between individuals with and without retinopathy (p<0.01). FC, fold change. Source and code files for this plot are available in Figure 2—source data 1 and Source code 5. An additional source file for this plot can be found in Gene Expression Omnibus (GEO) at https://www.ncbi.nlm.nih.gov/geo/ under accession code GSE146615.
Figure 2—figure supplement 1.
Figure 2—figure supplement 1.. Multidimensional scaling based on differential response to glucose (rg).
Each point represents a single study subject; individuals with proliferative diabetic retinopathy (PDR, red, n = 8) and individuals with diabetes without retinopathy (nDR, blue, n = 7). The first coordinate (Dim.1, x-axis) is correlated with subject retinopathy status (RGpdr–ndr, p=3×10−6). Source and code files for this plot are available in Figure 2—source data 2, 3, 4, and 5 and Source code 6.
Figure 2—figure supplement 2.
Figure 2—figure supplement 2.. Gene set enrichment analysis (GSEA) of genes with differential response to glucose between individuals with diabetes with and without diabetic retinopathy.
Normalized enrichment score (NES). Source files for this plot can be found in Gene Expression Omnibus (GEO) at https://www.ncbi.nlm.nih.gov/geo/ under accession code GSE146615.
Figure 3.
Figure 3.. Association of glucose differential response genes (RGpdr–ndr) with diabetic retinopathy.
(a) Workflow of analytical steps integrating glucose differential response genes with genetic association with diabetic retinopathy. Flow chart showing key experimental steps based on stepwise findings. (b) QQ plot revealing a skew away from the null and above the FDR 0.05 threshold suggests that expression of some of the glucose response genes may be causally related to diabetic retinopathy. 7253 GTEx eSNPs were generated from the 103 differential response genes and tested for their association with diabetic retinopathy in a GWAS. Observed vs. expected p-values are plotted. The null hypothesis of no difference between the observed and expected p-values is represented by the red line. No influence of population structure or other design factors was observed (genomic control inflation estimate λGC = 1.005) (Devlin and Roeder, 1999). Source and code files for this plot are available in Figure 3—source data 1 and Source code 7. (c) Bar plot comparing frequency of p-values <0.05 in diabetic retinopathy GWAS of: all eSNPs, all SNPs, and eSNPs from the 103 differential response genes. An excess of GWAS p-values of <0.05 is observed in the eSNPs from the glucose differential response genes (p=0.0012 vs. all eSNPs and p=0.0023 vs. all SNPs). The proportion of SNPs with p<0.05 in the all SNPs, all eSNPs, and 103 differential response gene eSNPs are 0.0505, 0.0499, and 0.0571, respectively. Source and code files for this plot are available in Figure 3—source data 2 and Source code 8.
Figure 3—figure supplement 1.
Figure 3—figure supplement 1.. Enrichment of eGenes in glucose differential response genes.
The proportion of eGenes is plotted to compare all 12,503 genes assessed on the microarray to the 103 glucose response genes. An eGene is defined as any gene with a GTEx eSNP (q-value <0.05) in any tissue. GTEx (version 7). Source files are available with open access at https://www.gtexportal.org/home/datasets, and code in Source code 9.
Figure 3—figure supplement 2.
Figure 3—figure supplement 2.. Histogram of frequency in which permutations of eSNPs generated from random sets of 103 genes revealed similar p-values to those generated from the set of 103 differential response genes to glucose (red dot).
For each resampling, 103 genes were chosen at random from the genome. eSNPs for each gene were generated from GTEx (version 7) using all 48 tissues. p-Values for each eSNP were determined in our prior meta-GWAS for diabetic retinopathy (Grassi et al., 2011). The x-axis shows the proportion of eSNPs in each set with an FDR < 0.05 in the diabetic retinopathy meta-GWAS. The figure reveals a significant shift to the right (represented by the red dot) for the glucose response gene eSNPs in the meta-GWAS compared to resampled eSNPs. Source and code files for this plot are available in Figure 3—source data 3, and additional source data files available at https://doi.org/10.5061/dryad.zkh18938j, and Source code 10.
Figure 4.
Figure 4.. Diabetic retinopathy meta-GWAS for eSNPs of differential response genes to glucose.
(a) Manhattan plot of the results of the meta-GWAS for diabetic retinopathy showing association signals for the eSNPs from the differential response genes to glucose for individuals with and without retinopathy (RGPDR–nDR). Threshold lines represent Bonferroni correction (blue) and FDR < 0.05 (black). Association testing for diabetic retinopathy performed with 7253 eSNPs representing 103 differential response genes to glucose. Source and code files for this plot are available in Figure 4—source data 1 and Source code 11. (b) Bar plot comparing the true positive rate (π1), TPR, for association of diabetic retinopathy with all SNPs, all eSNPs, eSNPs from the 103 differential response genes to glucose (n = 7253), and eSNPs found in retina and >20 GTEx tissues for folliculin (FLCN) (n = 272). TPR is an estimate of the proportion of tests that are true under the alternative hypothesis. Plot reveals significant enrichment for glucose response gene eSNPs in general and for FLCN eSNPs (π1 = 0.9) in particular. Source and code files for this plot are available in Figure 4—source data 2, 3, and 4, and Source code 12.
Figure 4—figure supplement 1.
Figure 4—figure supplement 1.. Folliculin (FLCN) expression in the human retina.
(A) FLCN (green) is evident in the ganglion cell layer (GCL), neuronal cells of the inner nuclear layer (INL), and faintly in the outer nuclear layer (ONL). (B) Colocalization of FLCN with CD31 (red), a marker of endothelial cells, confirms FLCN expression in retinal blood vessels (white arrows).
Figure 4—figure supplement 2.
Figure 4—figure supplement 2.. Folliculin (FLCN) expression response to glucose by disease status (PDR vs. nDR).
Box and whisker plot of the change in FLCN expression in lymphoblastoid cell lines between standard glucose and high glucose conditions. Pink indicates the distribution of responses for individuals with proliferative diabetic retinopathy (PDR) (log2FC = 0.08), and blue indicates the same for individuals with diabetes but no retinopathy (nDR) (log2FC = −0.19). Y-axis measures the difference in expression fold change on the log2 scale. Each individual is represented by a dot. Difference in means between PDR and nDR is 0.27, p-value 0.003. Source and code files for this plot are available in Gene Expression Omnibus (GEO) at https://www.ncbi.nlm.nih.gov/geo/ under accession code GSE146615, and Source code 13.
Figure 4—figure supplement 3.
Figure 4—figure supplement 3.. QQ plot of diabetic retinopathy meta-GWAS p-values corresponding to 272 folliculin (FLCN) eSNPs.
Points represent observed and expected GWAS-meta-analysis p-values for each of 272 FLCN eSNPs identified in the retina and more than 20 GTEx tissues. The red line represents the null hypothesis of no difference between the observed and expected p-values. Source and code files for this plot are available in Figure 4—source data 5 and Source code 14.
Figure 4—figure supplement 4.
Figure 4—figure supplement 4.. QQ plot of UKBB diabetic retinopathy GWAS p-values corresponding to 272 folliculin (FLCN) eSNPs.
Points represent observed and expected UKBB GWAS p-values for each of 272 FLCN eSNPs identified in the retina and more than 20 GTEx tissues. The red line represents the null hypothesis of no difference between the observed and expected p-values. Source and code files for this plot are available in Figure 4—source data 6 and Source code 15.
Figure 5.
Figure 5.. Experimental design.
(a) Schematic representation of the experimental design for transcriptomic profiling. Lymphoblastoid cell lines (LCLs) from 22 individuals were cultured under both standard glucose (SG) and high glucose (HG) conditions. Gene expression was quantified using microarrays for three biological replicates of each LCL in each condition. The response to glucose was determined for all genes on a per-individual basis, by comparing expression in SG and HG conditions. The cell lines were derived from individuals with diabetes and no retinopathy (7), individuals with diabetes and proliferative diabetic retinopathy (8), and individuals without diabetes (7). (b) We identified 15 individuals based on retinopathy status from the Epidemiology of Diabetes Interventions and Complications (EDIC) cohort. We compared the differential response in gene expression to glucose for individuals with and without proliferative retinopathy (RGpdr–ndr). Expression quantitative trait loci (eQTL) for those genes that showed the greatest differential response between individuals with and without retinopathy were tested for their genetic association with diabetic retinopathy.
See this image and copyright information in PMC

References

    1. Al-Mashat HA, Kandru S, Liu R, Behl Y, Desta T, Graves DT. Diabetes enhances mRNA levels of proapoptotic genes and caspase activity, which contribute to impaired healing. Diabetes. 2006;55:487–495. doi: 10.2337/diabetes.55.02.06.db05-1201. - DOI - PubMed
    1. Arar NH, Freedman BI, Adler SG, Iyengar SK, Chew EY, Davis MD, Satko SG, Bowden DW, Duggirala R, Elston RC, Guo X, Hanson RL, Igo RP, Ipp E, Kimmel PL, Knowler WC, Molineros J, Nelson RG, Pahl MV, Quade SRE, Rasooly RS, Rotter JI, Saad MF, Scavini M, Schelling JR, Sedor JR, Shah VO, Zager PG, Abboud HE. Heritability of the severity of diabetic retinopathy: the FIND-Eye study. Investigative Opthalmology & Visual Science. 2008;49:3839–3845. doi: 10.1167/iovs.07-1633. - DOI - PMC - PubMed
    1. Becker RA. The New S Language. Wadsworth & Brooks/Cole; 1988.
    1. Ben-David U, Siranosian B, Ha G, Tang H, Oren Y, Hinohara K, Strathdee CA, Dempster J, Lyons NJ, Burns R, Nag A, Kugener G, Cimini B, Tsvetkov P, Maruvka YE, O'Rourke R, Garrity A, Tubelli AA, Bandopadhayay P, Tsherniak A, Vazquez F, Wong B, Birger C, Ghandi M, Thorner AR, Bittker JA, Meyerson M, Getz G, Beroukhim R, Golub TR. Genetic and transcriptional evolution alters Cancer cell line drug response. Nature. 2018;560:325–330. doi: 10.1038/s41586-018-0409-3. - DOI - PMC - PubMed
    1. Caramori ML, Kim Y, Moore JH, Rich SS, Mychaleckyj JC, Kikyo N, Mauer M. Gene expression differences in skin fibroblasts in identical twins discordant for type 1 diabetes. Diabetes. 2012;61:739–744. doi: 10.2337/db11-0617. - DOI - PMC - PubMed

Publication types

MeSH terms