Molecular gatekeeper discovery to link exposome with health outcomes through metabolomics

class: center, middle, inverse, title-slide

# Molecular gatekeeper discovery to link exposome with health outcomes through metabolomics
## Miao Yu
### 69th ASMS Conference, Pennsylvania
### 2021-11-04

---

## Background

- ### One compound, one outcome -> regular epidemiology study

- ### One compound, multiple outcomes -> regular toxicology study

- ### Multiple compounds, one outcome -> regular metabolomics/non-targeted analysis

- ### Multiple compounds, multiple outcomes -> exposomics

---

## Exposures-Metabolites-Disease model

- ### What's the relationship between exposures and metabolites?

- ### What's the relationship between small molecule and health outcome?

- ### Which metabolites are the most important?

---

## Gatekeeper theory for EMD model: active metabolites

- #### Active metabolites will connect to other metabolites

---

## Gatekeeper theory for EMD model: gatekeeper

- #### The small molecule linked both exposures and other small molecule should be the frontier of biological process

- #### Such molecules may regulate the adverse influences of exposures

---

## Adolescent girls blood samples

- #### Breast Cancer and the Environment Research Program (BCERP)

- #### 152 girls with data on exposures, health outcomes, and potential confounding variables

- #### **Exposures** measured by CDC
  - PFCs (n-PFOS, Sm-PFOS, n-PFOA, PFHxS, PFNA)
  - trace elements (BCD, Cadmium; BMN, Manganese; BPBLead, Lead; BSE, Selenium; THG, Mercury)

- #### **Metabolomics** by LC-Q-TOF-MS
  - 18750 (ZHP) and 25937 (RPN) features were found in plasma samples

- #### **Health outcomes**
  - BMI/BMI group
  - age at menarche

---

## Gatekeeper molecular discovery

- ### Step 1: Correlation clusters among metabolites

- ### Step 2: Association between metabolites and exposures

- ### Step 3: Screen the network structure to discover gatekeeper

- ### Step 4: Association between gatekeeper and health outcome

---

## Step 0: From metabolomics features to independent peaks

- #### Remove redundant peaks by GlobalStd algorithm

- #### 964 (ZHP) and 1784 (RPN) independent peaks as potential metabolites

.half[
Yu, M., Olkowicz, M., & Pawliszyn, J. (2019). Structure/reaction directed analysis for LC-MS based untargeted analysis. Analytica Chimica Acta, 1050, 16–24. doi:10.1016/j.aca.2018.10.062
]

---

## Step 1: Correlation clusters among metabolites

- #### Cutoff for correlation: Spearman’s rho > 0.9
- #### 62 (ZHP) and 98 (RPN) metabolites correlation network clusters found
- #### 786 out of 964 metabolites have no correlation with other metabolites (ZHP) and 1416 out of 1784 metabolites have no correlation with other metabolites (RPN)

---

## Step 2: Association between metabolites and exposures

---

## Step 3: Screen the network structure to discover gatekeeper

- #### 28 (ZHP) and 43 (RPN) peaks connected with other peaks and exposures
- #### 87 (ZHP) and 137 (RPN) peaks connected with exposures only

---

## Step 3: Screen the network structure to discover gatekeeper

- #### 5 (ZHP) and 5 (RPN) annotated gatekeepers are associated with exposure
- #### 2 annotated gatekeepers were associated with multiple environmental exposures

---

## Step 4: Association between exposure and health outcome

|            Exposure            | BMI (β±SE) | BMI group (β±SE) | Age at menarche in years (β±SE) | Associated exposure(s) |
|:------------------------------:|:----------:|:----------------:|:-------------------------------:|:----------------------:|
| n-PFOA                         | -4.5±3.1   | -0.3±0.2         | 1.2±1.5                         | -                      |
| n-PFOS                         | **-4.2±2.1**   | -0.3±0.2         | -0.2±1                          | -                      |
|           **Gatekeeper**           |            |                  |                                 |                        |
| SM(d18:2/14:0)                 | **30.5 ±5.2**  | **2.2 ±0.5**         | **0.8 ±0.2**                        | n-PFOS                 |
| Dehydroepiandrosterone sulfate | **6.0 ±2.5**   | **0.5 ±0.2**         | -0.1 ±0.1                       | n-PFOA                 |
| Androsterone sulfate           | **10.8 ±2.4**  | **1.0 ±0.2**         | -0.1 ±0.1                       | n-PFOA                 |
| taurodeoxycholate              | **-3.9 ±1.6**  | **-0.3 ±0.1**        | 0.1 ±0.1                        | n-PFOA, n-PFOS         |
| GPC(P-18:0/20:4)               | **-13.4 ±5.1** | **-1.2 ±0.4**        | -0.1 ±0.2                       | n-PFOS                 |

---

## Conclusion

- ### Gatekeeper discovery could be used to screen metabolites associated with exposures and other metabolites

- ### Some gatekeepers are sensitive to multiple environmental exposures

- ### Gatekeepers can reveal mechanistic aspects of relationships between environmental exposures and metabolome to generate hypothesis for future health related studies

- ### As a general data analysis framework, the mediation molecular can also be gene, protein, RNA, and/or metagenomics

- ### enet package on GitHub: https://github.com/yufree/enet

---

## Acknowledgement

.pull-left[

- #### Dr. Petrick's research group

- #### Lam-ha Dang

- #### Prof. Mary Wolff

- #### Prof. Susan Teitelbaum

- #### National Center for Environmental Health Laboratories at CDC

### Fundings

- NIEHS P30ES23515
- CHEAR pilot program
- U2C ES026561
- 1U2CES030859

]

.pull-right[

]

---

class: inverse, center, middle

# Thanks

## Q&A

## miao.yu@mssm.edu