1. Brett Macey, Matthew Jenny, Lindy Thibodeaux,
Physiological Responses of the Eastern Oyster Crassostrea virginica Exposed to Mixtures of Copper, Cadmium and Zinc
Brett Macey, Matthew Jenny, Lindy Thibodeaux,
Heidi Williams, Jennifer Ikerd, Marion Beal, Jonas Almeida, Charles Cunningham, AnnaLaura Mancia, Gregory Warr, Erin Burge, Fred Holland, Paul Gross, Sonomi Hikima, Karen Burnett, Louis Burnett, and Robert Chapman

2. Environmental changes
Biological Response Networks
Environmental changes
Physiological
responses
Immune
responses
Genomic and proteomic
responses

4. Environmental changes
Can we generate a predictive model
that links physiological responses to environmental change?
Physiological
responses

5. Environmental change: exposure to multiple metals
216 C. virginica
27 combinations:
Cu (0 – 200 ppb)
Cd (0 – 50 ppb)
Zn (0 – 200 ppb)
0 – 27 days exposure

6. Physiological Responses
Physical
weight, width, length
accumulated metals
Respiratory/acid-base/ redox status
hemolymph Po2, pH,
& total CO2
gill & hepatopancreas glutathione (GSH)
gill & hepatopancreas lipid peroxidation (LPx)
Immune response
culturable bacteria
culturable Vibrio spp.
hemocyte count

7. Glutathione (GSH)
Oxidative Damage
(e.g. Lipid peroxidation)

8. What We Learned metal accumulation in tissues
physiological responses to mixed metal exposure
linear analysis
modelling interactions of metals to predict physiological effects
Non-linear analysis
(Artificial Neural Networks)

9. Cu++ content of tissues did not change with exposure to Cu++
Patterns of metal accumulation are complex
and interdependent
Metal exposure [uM*days]

10. Zn++ content of tissues did not change with exposure to Zn++
●…Gill
□…Hepatopancreas
Metal exposure [uM*days]

11. Cd++ content of tissues increased with exposure to Cd++
●…Gill
□…Hepatopancreas

12. Physiological Responses Correlated with Metal Exposure
NONE

13. Physiological Responses Correlated with Metal Contents of Gill
Correlation Coefficient
LPx

14. Physiological Responses Correlated with Metal Contents of Hepatopancreas
Correlation Coefficient
LPx

15. Conclusions of Linear Analyses
Lipid Peroxidation (Oxidative Damage) was the most reliable marker for metal tissue content across tissue and treatments.
General Linear Models showed significant interaction between measured Cu and Zn in predicting oxidative damage.

16. Environmental changes
Systems Modeling
Environmental changes
Cu, Zn, Cd
LPx
Can we find a model that better predicts
the relationship between
oxidative damage and metal content?

17. Artificial Neural Networks
non-linear statistical data modeling tools
used to model complex relationships
- between inputs and outputs
- find patterns in data

18. Artificial Neural Networks
Tissue metals Cu Zn Cd
LPx or
GSH
Hemolymph pH
PO2
CO2

19. Artificial Neural Networks (cont’d)
Generated 30 ANNs for each tissue and each output (LPx or GSH).
Looked for models with
high R2
cross-validation with high R2
low variance among models

20. Artificial Neural Networks Results
Poor prediction of GSH
Gill
Average #nodes =
Average R2 =
Hepatopancreas
Average #nodes =
Average R2 =
Stronger prediction of LPx
Gill
Average #nodes =
Average R2 =
Hepatopancreas
Average #nodes =
Average R2 =

21. Sensitivity Analysis for Gill - LPX: best-fit model
observed variance in LPx
% Contribution to
# nodes = 7
R2 =

22. Sensitivity Analysis for Gill - LPx: best-fit models
Hepatopancreas LPx

23. Sensitivity Analysis for Hepatopancreas - LPx: best-fit model
# nodes = 8
R2 =
observed variance in LPx
% Contribution to

24. Sensitivity Analysis for Hepatopancreas - LPx: best-fit models
Gill LPx

25. Importance of these findings
Oxidative damage, measured by LPx, is a broad-based biomarker for metal-induced toxicity in oysters.
ANNs incorporating markers of oxidative damage (e.g. LPx) along with markers of redox status (hemolymph pH, Po2, Pco2) provide powerful predictive models for the complex relationships between mixed metal exposure and oxidative damage in whole oysters.

26. Thanks