Cross-platform transcriptomic profiling of the response to recombinant human erythropoietin

Guan Wang, Traci Kitaoka, Ali Crawford, Qian Mao, Andrew Hesketh, Fergus Guppy, Garrett I. Ash, Jason Liu, Mark Gerstein, Yannis Pitsiladis

This study, published on Research Square, provides a rigorous head-to-head comparison of modern transcriptomic technologies to see which can best "catch" the biological signature of recombinant human erythropoietin (rHuEPO).

TL;DR

Researchers compared four different gene-expression platforms to track how the human body responds to rHuEPO (a common blood-doping agent and anemia treatment). They found that total RNA sequencing using the MGI DNBSEQ™ platform was significantly more sensitive than traditional microarrays or standard Illumina mRNA-seq, uncovering a wealth of new "gene signatures" that could lead to more robust anti-doping tests and better therapeutic monitoring.

What Problem Does This Paper Address?

Recombinant human erythropoietin (rHuEPO) is a hormone that stimulates red blood cell production. While it’s a vital medicine for patients with kidney disease or anemia, it is also infamously used by athletes to gain an unfair aerobic advantage.

The challenge for anti-doping authorities is that the hormone itself leaves the system quickly, but its biological effects—the "transcriptomic fingerprint"—linger much longer. However, the field lacks a consensus on which technology is best for detecting this fingerprint. Previous studies often relied on older microarray tech or specific RNA-seq protocols that might miss subtle but important changes. This paper seeks to identify the most sensitive and reliable platform for defining these gene signatures.

Methods

The researchers conducted a longitudinal study involving 18 endurance-trained males who received rHuEPO injections over four weeks. They collected whole blood samples at multiple time points: before, during, and after administration.

To compare technologies, they ran the same samples across four major platforms:

  1. MGI DNBSEQ-G400RS: Total RNA-seq (using rRNA depletion and globin reduction).
  2. Illumina NextSeq 500: mRNA-seq (using poly-A enrichment).
  3. Affymetrix GeneChip™ HTA 2.0: A high-density microarray.
  4. Illumina HumanHT-12 v4 BeadChip: A standard benchmarking microarray.

The team used a "data triangulation" approach, cross-referencing results from different platforms and using Gene Set Enrichment Analysis (GSEA) and Reactome pathway mapping to ensure the findings were biologically meaningful rather than just statistical noise.

Datasets

  • Primary Dataset: 143 whole blood samples from 18 subjects (with a subset of 50 samples from 10 subjects used for the direct four-platform comparison).
  • Reference Databases: The study utilized the MSigDB (Molecular Signatures Database) for hallmark pathways and Gene Ontology (GO) terms, and the Reactome database for functional pathway visualization.

Key Insights and Ideas

  • The Superiority of Total RNA-seq: The MGI DNBSEQ platform (total RNA-seq) was the clear winner in sensitivity. It identified 1,552 differentially expressed features at the peak of treatment, compared to only 582 for Illumina mRNA-seq and 252 for the GeneChip microarray.
  • Why MGI Performed Better: The authors suggest that by using rRNA depletion rather than poly-A enrichment (which only captures mature mRNA), the MGI protocol captured a broader slice of the transcriptome, including intronic sequences and low-abundance transcripts that other methods missed.
  • Beyond Red Blood Cells: While "heme metabolism" was the most enriched pathway across all platforms (as expected), the high sensitivity of the MGI platform revealed that rHuEPO affects much more than just erythropoiesis. It unveiled responses related to mitophagy (cleaning up damaged mitochondria), oxidative stress response, and even neuronal synaptic plasticity.
  • The Power of Triangulation: By looking for genes that showed up across multiple platforms (like ALAS2, BPGM, and SLC4A1), the researchers identified a "core" signature that is highly resistant to platform-specific biases.

Key Results

  • Sensitivity Boost: MGI RNA-seq showed a 2.28-fold increase in significantly expressed genes compared to the combined results of the other two primary platforms.
  • Unique Signatures: At the four-week mark of EPO administration (EPO4), 66.8% of the genes identified by MGI were unique to that platform, meaning other technologies were effectively blind to them.
  • Validated Candidates: The study narrowed down the results to a high-confidence list of 80 candidate genes (from MGI) and 41 genes (from BeadChip) that are biologically relevant to rHuEPO response. These serve as a "knowledge base" for future diagnostic tests.

Limitations and Future Work

The authors acknowledge that while 18 subjects is a solid number for this type of clinical trial, it is still a relatively small cohort for defining a universal "gold standard" signature.

Future Work: 1. Confounding Factors: Researchers need to test these gene signatures against "confounders"—things like high-altitude training, extreme heat, or intense exercise—to ensure a "false positive" isn't triggered by natural physiological stress. 2. Practical Testing: The next step is to translate these 80+ candidate genes into a streamlined, cost-effective test that can be used in World Anti-Doping Agency (WADA) labs globally. 3. Therapeutic Application: Beyond sports, these insights could help doctors better monitor patients receiving EPO therapy to ensure they are responding correctly at a molecular level.

Summary updated

Abstract

Abstract RNA-seq has matured and become an important tool for studying RNA biology. Here we compared two RNA-seq (Illumina sequencing by synthesis and MGI DNBSEQ™) and two microarray platforms (Illumina Expression BeadChip and GeneChip™ Human Transcriptome Array 2.0) in healthy individuals administered recombinant human erythropoietin for transcriptome-wide quantification of differential gene expression. The results show that total RNA sequencing combined with DNB-seq produced a multitude of genes of biological relevance and significance in response to recombinant human erythropoietin, in contrast to other platforms. Through data triangulation linking genes to functions, genes representing the processes of erythropoiesis as well as non-erythropoietic functions of erythropoietin were unveiled. This study provides a knowledge base of genes characterising the responses to recombinant human erythropoietin through cross-platform comparison and validation.
BibTeX
@inproceedings{wang2021cross,
  author = {Guan Wang and Traci Kitaoka and Ali Crawford and Qian Mao and Andrew Hesketh and Fergus Guppy and Garrett I. Ash and Jason Liu and Mark Gerstein and Yannis Pitsiladis},
  title = {{Cross-platform transcriptomic profiling of the response to recombinant human erythropoietin}},
  booktitle = {Scientific},
  year = {2021},
  doi = {10.1038/s41598-021-00608-9},
}
Figure