With the continuous advancement of sequencing technologies and mass spectrometry detection methods, transcriptomics and metabolomics have become essential and widely used omics disciplines. By leveraging transcriptomics to explore gene expression differences at the transcriptional level and combining metabolomics to validate the regulation of downstream products by genes, a mechanistic study of upstream and downstream relationships has become a fundamental premise for integrated analysis of both omics. This article introduces two new approaches for associating gene expression and metabolite abundance changes to predict the potential regulatory mechanisms between genes and metabolites.

Integrated Transcriptomic and Metabolomic Analysis Methods

Four-quadrant plot strategy for comparing two sets of samples

Pairwise comparison is often the most commonly used analytical approach in differential studies. By conducting differential analysis, the differentially expressed genes or metabolites provide preliminary insights into potential factors that regulate phenotypic differences in organisms, such as the screening of candidate biomarkers and identification of key genes influencing desirable traits. Performing two-omics integration based on differential results is a common method to validate whether and how transcriptional-level differences regulate downstream metabolite products. Therefore, using a four-quadrant plot allows for an intuitive and quick presentation of the two issues mentioned above. Each point in the plot represents a gene-metabolite relationship pair. The quadrant assignment of each point is determined by the correlation between gene and metabolite abundance, where quadrants 1 and 4 indicate negative correlation, and quadrants 2 and 3 indicate positive correlation. The positions of points in the plot are determined by the log2(FC) values of both genes and metabolites.

The four-quadrant plot allows for the rapid identification of two key pieces of information:

• The trend of gene or metabolite changes between the two sample groups;
• Whether there is a positive or negative correlation between genes and metabolites.

By analyzing the trend of changes, the variations within the omics data can be interpreted, and potential regulatory relationships between the omics can be explored based on their correlation. This method facilitates the individual selection and downstream analysis of gene-metabolite relationship pairs with specific change trends and correlations.

WGCNA Strategy for Multi-group Sample Association

Multi-sample omics studies not only enhance the precision and depth of scientific research, enable biomarker screening, and promote clinical applications, but also provide more comprehensive data for studies on crop breeding population genetic diversity, economic traits in plants and animals, species origin, and precision breeding. However, multi-sample experimental designs often involve challenges such as a large number of samples, high data depth, and more complex gene or metabolite abundance variation patterns. It is difficult to effectively categorize genes and metabolites using simple pairwise sample comparisons.

WGCNA (Weighted Gene Co-expression Network Analysis) is capable of directly integrating genes or metabolites with similar expression patterns. It not only constructs gene-gene similarity networks but also combines trait data with gene interactions to identify key factors (genes, modules) influencing complex traits. This ability to integrate trait data makes WGCNA a foundational method for transcriptomics-metabolomics integration. Metabolites, being downstream of molecular mechanisms, can directly explain the factors underlying phenotypic differences. In association analysis, the abundance of target metabolites is treated as trait data, and WGCNA is applied to transcriptomic data to identify the gene modules most strongly correlated with the target metabolites, thus revealing the regulatory relationships between genes and metabolites. Further exploration of hub genes within the identified modules using Cytoscape can provide insights into the potential regulatory role of hub genes on target metabolites.

Case of Integrated Analysis of Transcriptomic and Metabolomics

Comprehensive metabolomics expands precision medicine for triple-negative breast cancer

Journal: Cell Res (Impact Factor = 28.2)
Published: May 2022
Sample Type: Triple-negative breast cancer (TNBC) samples, paired normal breast tissues
Research Method: Metabolomics, Lipidomics, Transcriptomics, Genomics

The metabolic reprogramming process in triple-negative breast cancer (TNBC) remains unclear. The following study utilized untargeted metabolomics and lipidomics to analyze 330 TNBC samples and 149 normal breast tissue samples, and integrated these findings with transcriptomic data from the same cohort. Based on the identification of TNBC subtype-specific metabolites, the study aimed to screen potential therapeutic targets for the disease.

Metabolomic landscape of TNBC

In this study, using Benjamini–Hochberg-corrected Mann–Whitney U tests, 452 metabolites (417 elevated and 35 reduced in tumor samples) were identified to exhibit significant differences in abundance between tumor and normal tissues. Among the dysregulated metabolites, certain polar metabolites, particularly those involved in oxidation reactions and glycosyl transfer (such as oxidized glutathione [GSSG] and uridine diphosphate glucose [UDPG]), were significantly enriched in tumors compared to normal tissues. Additionally, lipids such as phosphatidylinositols, fatty acids, and ceramides were also found to be enriched in triple-negative breast cancer (TNBC) tissues.

Figure 1 .volcano plots of the 594 annotated polar metabolites.

A comprehensive analysis linking polar metabolites and lipids to genomic features

Using the Recon 3D database, gene-metabolite substrate (substrate) and gene-metabolite product (product) relationships were identified, and correlations were calculated and visualized in a quadrant plot. The results showed a low correlation between the two omics datasets, indicating that the metabolic profile in cancer patients is more complex.

Figure 2 .Correlation of mRNA expression of metabolic genes with the abundances of paired metabolites as substrates or products.

Metabolomic subtyping refines the transcriptomic subtyping of BLIS tumors

To investigate metabolic heterogeneity, similarity network fusion (SNF) analysis was used to classify TNBC patients into three distinct subtypes based on metabolomic data: the C1 subtype (characterized by sphingolipids and fatty acids), the C2 subtype (characterized by upregulated metabolites related to carbohydrate metabolism and oxidation reactions), and the C3 subtype (characterized by metabolites with no significant differences compared to normal tissues). Correlation analysis between these metabolic subtypes and previously identified transcriptomic subtypes revealed that the transcriptomic luminal androgen receptor (LAR) subtype highly overlaps with the metabolomic C1 subtype. Additionally, transcriptomic subtypes, including basal-like immune-suppressed (BLIS), immune-regulatory (IM), and mesenchymal (MES), were classified into metabolomic C2 and C3 subtypes, with the BLIS-C2 subgroup showing poorer prognosis. These findings suggest that metabolomic data provides new insights into previously developed TNBC transcriptomic subtypes, highlighting the importance of integrating both omics data for subtype analysis and exploring metabolic targets for different subtypes. Experimental validation further identified sphingosine-1-phosphate (S1P), an intermediate in the ceramide pathway, as an effective therapeutic target for LAR tumors, and N-acetyl-aspartyl-glutamate as a potential therapeutic target for BLIS tumors.

Figure 3 .Metabolomic Subtype Associations with Transcriptomic Subtypes, Metabolic-Gene Subtypes, and TNBC Relapse Status.

Multiomics strategy reveals the accumulation and biosynthesis of bitter components in Zanthoxylum schinifolium Sieb. et Zucc

Journal: Food Res Int (Impact Factor = 7.0)
Published: December 2022
Sample Type: Zanthoxylum schinifolium
Research Method: Metabolomics, Transcriptomics

This study investigates the perceptible bitterness of Zanthoxylum schinifolium Sieb. et Zucc, a key factor that limits its consumption. The researchers employed metabolomics to identify potentially bitter metabolites during the plant's development, focusing on dynamic changes in 17 key bitter components. These components were primarily synthesized through the phenylpropanoid biosynthesis, flavonoid biosynthesis, and flavone/flavonol biosynthesis pathways. By combining this metabolomic data with transcriptomic analysis, the study revealed how the expression of specific genes regulates the accumulation of these bitter compounds. Additionally, quercetin-3-galactoside, isoquercitin, quercitrin, kaempferol, and isorhamnetin were identified as the main sources of bitterness in mature Z. schinifolium. The findings provide valuable insights into the biochemical mechanisms underlying the plant's bitterness and lay the groundwork for developing high-quality varieties of Z. schinifolium.

Weighted gene co-expression network analysis

WGCNA analysis was performed on genes with FPKM > 5, resulting in the identification of 23 modules. A correlation analysis between the abundance of 17 bitter metabolites, including quercetin, epicatechin, L-valine, proanthocyanidin B2, salicylic acid, and arbutin, obtained from metabolomics, and the module eigengenes revealed significant correlations between the turquoise and sienna modules and the bitter metabolites. Enrichment analysis of the genes within these two modules identified key pathways such as "phenylpropanoid biosynthesis," "flavonoid biosynthesis," and "phenylalanine metabolism." These findings are consistent with the results of differential transcriptomic analysis of Zanthoxylum schinifolium at different developmental stages, further indicating that these three pathways are critical in regulating the accumulation of bitter metabolites. Genes associated with these pathways play an important potential regulatory role in the accumulation of bitter metabolites.

Figure 4 .Correlation heatmap of gene expression modules and bitter metabolite abundance.

References

1. Xiao, Y., Ma, D., Yang, Y. S., et.al. (2022). Comprehensive metabolomics expands precision medicine for triple-negative breast cancer. Cell research, 32(5), 477–490. https://doi.org/10.1038/s41422-022-00614-0
2. Huang, S., Wang, L., Wang, Z., Yang, G., Xiang, X., An, Y., & Kan, J. (2022). Multiomics strategy reveals the accumulation and biosynthesis of bitter components in Zanthoxylum schinifolium Sieb. et Zucc. Food research international (Ottawa, Ont.), 162(Pt A), 111964. https://doi.org/10.1016/j.foodres.2022.111964