Dynamic simulation of Bone Morphogenetic patterning in a 3D

model of the zebrafish embryo

Sang-Hun Lee; Ph.D.David M. Umulis; Ph.D.

Ag. & Biological Engineering;Weldon School of Biomedical Engineering

Purdue UniversityWest Lafayette, IN

Conflict disclosure: none

mikeText BoxPresented at the COMSOL Conference 2011 Boston

Canonical BMP pathway

PP

Co-Smad

R-Smad

Txn

P

P

P

R-Smad

type I and type II

P

cell membrane

nucleus

Tx

Hetero/homodimer ligands w/ different activity1

1Shimmi, Umulis, Othmer & O’Connor, Cell (2005) 2For review see Umulis, et al. Development (2009)

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Source Sink

Threshold A: Front

Threshold B: Rear

Diffusion

A.M. Turing

Lewis Wolpert

Problems with the classical view of Morphogenesis

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Problems?

* Figure from Umulis, O’Connor, & Othmer. CTDB(2008)

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Saturation?

Use of Comsol modeling in developmental biology

• Test alternative hypotheses• Tackle inverse problem• Measure biophysical properties• Understand phenomena and behavior• Estimate unknowns• Translate expectations to new contexts• Bridge genotype to phenotype

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

BMP regulation in zebrafish

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Process control: jet

United States Patent 6885917

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Using Comsol to reverse engineer zebrafish development

Aero/Jet Chemical product

Cellular

DataTheoryRegime

EnvironmentSensingControl

strategiesTradeoffs

Quantitative Quantitative Semi/QualMature Mature Early

Continuous Continuous +StochasticDynamic Steady-state Dynamic>103 /sec >10/sec up to daysClassical, model based,

relatively well characterizedChallenging- sparse data

Reduced by design, $$$ Inherent

• >10 genes, >3 states each gene = 59,049 combinations at minimum

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Discovery pipeline for modeling zebrafish embryonic patterning

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Geometry development

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Model development

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Model results

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

BMP-Alk8 (distributions of active signaling)

Chordin (distribution BMP inhibitor)

Model results

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Model performance (scale-invariance)

Chordin (Inhibitor secreted dorsally) Tld (Chordin protease)

Dimensionless position Dimensionless position

BMP-

ALK

8

BMP-

ALK

8

Other uses of Comsol in developmental biology: decipher stem cell regulation

!"#$%&&$

'()$

&'()$

*(+$

,-.$

*/(0$

1*23$

4+5/6$

)'78$9:)-870-.;.6?/=:=@(>$2/-+-0=.$);A=/=:B(B-:$

Harris et al. Dev. Cell (2011)

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Umulis et al. Dev. Cell (2010)Peluso et al. Dev. Cell (2011)

Other uses of Comsol in developmental biology: BMP pattern formation

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Concluding remarks

• Multidimensional data can be seamlessly integrated into mathematical models.

• Current zebrafish models are being used as a tool to drive discovery alongside experimental methods.

• BMP/Sizzled/Chordin/Tld network appears capable of providing automatic scale-invariance.

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

Future directions: Better integration

David M. Umulis Purdue UniversityDavid M. UmulisBackground Results Conclusions Future directions Ack.

!"

#"

$"

%"

&"

!"

#"

$"

%"

&"

Joe Zinski Mary Mullins, Ph.D.

Acknowledgements

David M. Umulis Purdue UniversityDavid M. UmulisExamples Ack.Overview Future directions

Purdue University, USADr. Sang-Hun Lee

Shahriar Karim

U Penn, USAMary Mullins

Joe Zinski

Thank you!

David M. Umulis Purdue UniversityDavid M. UmulisExamples Ack.Overview Future directions