Landscape dynamic network biomarker analysis reveals the tipping point of transcriptome reprogramming to prevent skin photodamage

Chengming Zhang, Hong Zhang, Jing Ge, Tingyan Mi, Xiao Cui, Fengjuan Tu, Xuelan Gu, Tao Zeng, Luonan ChenJournal of Molecular Cell Biology 2021 · DOI

TL;DR

This study introduces a novel "landscape dynamic network biomarker" (l-DNB) approach to analyze time-series transcriptome data from 3D skin models exposed to UV radiation. It identifies a critical "tipping point" in the skin's response, characterized by a specific set of 13 genes that can predict skin lightening and serve as early warning signals for photodamage.

What Problem Does This Paper Address?

The skin, our outermost barrier, is constantly exposed to environmental stressors like UV radiation, which can lead to various skin disorders, including premature aging and cancer. While we know UV exposure triggers a response, the underlying molecular mechanisms, especially the critical transition points where the skin's state changes irreversibly, are not well understood. Traditional studies often look at single time points or individual genes, missing the dynamic, non-linear nature of the skin's response. This paper aims to uncover these critical transition points and identify molecular signatures that can predict and potentially prevent skin photodamage.

Methods

The researchers employed a "landscape dynamic network biomarker" (l-DNB) analysis, an advanced method built upon previous dynamic network biomarker (DNB) theories. The core of their approach involves:

  • 3D Skin Model (LSE): A pigmented living skin equivalent (LSE) model was used to mimic human skin's response to repetitive UVB irradiation. This model allows for controlled experimental conditions and time-series data collection.
  • Time-Series Transcriptome Data: RNA sequencing was performed at multiple time points (from hours to days) after UVB exposure to capture the dynamic changes in gene expression.
  • Single-Sample Network (SSN) Construction: For each sample, a gene co-expression network was built to represent the molecular interactions at that specific time point.
  • Dynamic Network Biomarker (DNB) Scoring: A scoring system was developed to quantify the "network biomarker" activity of genes. Genes with high DNB scores are considered critical players in state transitions.
  • Tipping Point Identification: By analyzing the DNB scores across time, the researchers identified a "tipping point"—a critical moment before a significant, potentially irreversible change in the skin's state.
  • Core DNB Gene Identification: A set of 13 "core DNB genes" were identified as crucial for detecting this tipping point. These genes were prioritized based on their DNB scores and their presence in the network at the tipping point.
  • Validation: The findings were validated using:
  • In silico validation: Comparing DNB scores derived from core DNB genes versus all genes.
  • In vivo validation: Using an independent human skin dataset to assess the predictive power of core DNB genes for skin lightening.
  • In vitro validation: Testing the effect of a skin lightening agent (USBT2627) applied before and after the identified tipping point.

Datasets

  • Primary Dataset: Time-series transcriptome data from a 3D pigmented living skin equivalent (LSE) model subjected to repetitive UVB irradiation.
  • Independent Validation Dataset: Human skin data from a clinical trial, including gene expression and phenotype (L* value for skin lightness) measurements from two groups (placebo and active treatment).

Key Insights and Ideas

  • The Skin's Response is Biphasic and Nonlinear: The study reveals that skin's reaction to UVB isn't a simple, continuous increase in damage. Instead, it exhibits an acute response followed by an adaptation phase, with a critical tipping point occurring during this adaptation.
  • A Tipping Point Exists Before Significant Phenotypic Change: The most crucial insight is the identification of a "tipping point" in gene expression dynamics that precedes observable changes in skin lightness. This suggests that molecular changes are occurring that can be detected before visible signs of photodamage manifest.
  • A Network-Based Biomarker for Skin Lightness: A specific set of 13 core DNB genes were identified as robust predictors of skin lightening. These genes are not necessarily the most obvious pigmentation genes but are involved in broader cellular processes like proliferation and cell cycle regulation, highlighting the complex interplay of factors.
  • Early Warning System Potential: The identified core DNB genes can serve as early warning signals, allowing for proactive interventions to prevent or mitigate skin photodamage.

Key Results

  • Tipping Point Identified: The study pinpointed a tipping point around the first hour of Day 6 (Day6-1h) in the LSE model's response to UVB.
  • 13 Core DNB Genes Identified: These genes, including COL7A1 and CTNNB1, were found to be critical network biomarkers for predicting skin lightening.
  • Predictive Power Demonstrated:
    • In silico validation showed that DNB scores calculated using core DNB genes mirrored the overall DNB score trend.
  • In vivo validation with human skin data confirmed that core DNB genes could consistently and significantly predict skin lightening outcomes.
  • In vitro experiments showed that applying a skin lightening agent before the tipping point was significantly more effective than applying it afterward, reinforcing the importance of timing and the tipping point concept.

Limitations and Future Work

  • LSE Model vs. Human Skin: While the LSE model is a good mimic, it's still a simplified representation of complex human skin. Future work could involve validating these findings in more complex in vivo models or directly in human studies.
  • Mechanism of Core DNB Genes: While the study identified core DNB genes and their involvement in pathways, further research could delve deeper into the precise molecular mechanisms by which these genes drive the tipping point and influence skin lightening.
  • Broader Applications: The l-DNB approach, demonstrated here for skin photodamage, could potentially be applied to identify tipping points and network biomarkers for other complex diseases and biological processes.
  • Personalized Risk Prediction: The authors suggest that these core DNB genes could be developed into personalized network biomarkers for assessing an individual's risk of skin photodamage.

Summary updated

Abstract

Skin, as the outmost layer of human body, is frequently exposed to environmental stressors including pollutants and ultraviolet (UV), which could lead to skin disorders. Generally, skin response process to ultraviolet B (UVB) irradiation is a nonlinear dynamic process, with unknown underlying molecular mechanism of critical transition. Here, the landscape dynamic network biomarker (l-DNB) analysis of time series transcriptome data on 3D skin model was conducted to reveal the complicated process of skin response to UV irradiation at both molecular and network levels. The advanced l-DNB analysis approach showed that: (i) there was a tipping point before critical transition state during pigmentation process, validated by 3D skin model; (ii) 13 core DNB genes were identified to detect the tipping point as a network biomarker, supported by computational assessment; (iii) core DNB genes such as COL7A1 and CTNNB1 can effectively predict skin lightening, validated by independent human skin data. Overall, this study provides new insights for skin response to repetitive UVB irradiation, including dynamic pathway pattern, biphasic response, and DNBs for skin lightening change, and enables us to further understand the skin resilience process after external stress.
BibTeX
@article{zhang2021landscape,
  author = {Chengming Zhang and Hong Zhang and Jing Ge and Tingyan Mi and Xiao Cui and Fengjuan Tu and Xuelan Gu and Tao Zeng and Luonan Chen},
  title = {{Landscape dynamic network biomarker analysis reveals the tipping point of transcriptome reprogramming to prevent skin photodamage}},
  journal = {Journal of Molecular Cell Biology},
  year = {2021},
  doi = {10.1093/jmcb/mjab060},
}
Figure