CompuCell3D Workshop: Module 5.3: �Modeling Secondary Palate Fusion and Disruption

M. Shane Hutson

Department of Physics & Astronomy, Vanderbilt University

Thomas B. Knudsen

National Center for Computational Toxicology,

U.S. Environmental Protection Agency

References: Herbert Sauro, Systems Biology: Introduction to Pathway Modeling, for help with Tellurium visit https://tellurium.readthedocs.io/en/latest/index.html

Slides, cheat-sheets and demos available at: https://drive.google.com/drive/folders/1m1AHosct2F7xlVCZf9da57eCRfbf7elM?usp=sharing

Support: NIH NIBIB-U24EB028887, NIGMS-R01GM122424, NSF-2120200, NSF-2000281, NSF-1720625, NIGMS-R01GM076692, NIGMS-R01GM077138, EPA – 83573601, ORISE

Workshop is live-streamed, recorded and distributed

The views expressed in this presentation are those of the author[s] and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency.

Take-Home Messages:

• cell-based model of palate fusion
• includes biochemical regulation
• recapitulates genetic knockouts
• provides mechanistic insight for developmental toxicity of dioxin

J.O. Bush and R. Jiang (2012) Development 139:231-243.

OUTLINE

​

• Sketch of the computational model
• cellular biophysics
• biochemical regulation
• Simulations of WT and in silico knock-outs
• Simulating a toxicant’s effect (dioxin)
• Look at the model code
• Code-writing exercises

CompuCell3D Model of Palate Fusion

Each cell occupies multiple points on a hexagonal lattice.

CompuCell3D Model of Palate Fusion

Pseudo-energy minimization (Metropolis Monte Carlo) with terms for

• Target area of each cell
• Target perimeter of each cell
• Target length of some cells
• Contact energies between cells

​

​

​

​

​

​

​

​

​

​

CompuCell3D Model of Palate Fusion

Pseudo-energy minimization (Metropolis Monte Carlo) with terms for

• Target area of each cell
• Target perimeter of each cell
• Target length of some cells
• Contact energies between cells

​

​

​

​

​

​

CompuCell3D Model of Palate Fusion

Pseudo-energy minimization (Metropolis Monte Carlo) with terms for

• Target area of each cell
• Target perimeter of each cell
• Target length of some cells
• Contact energies between cells

​

​

​

​

​

​

CompuCell3D Model of Palate Fusion

Pseudo-energy minimization (Metropolis Monte Carlo) with terms for

• Target area of each cell
• Target perimeter of each cell
• Target length of some cells
• Contact energies between cells

​

​

​

​

​

​

Mesenchyme

Epithelium

NASAL SIDE

ORAL SIDE

MEE

Fgf7

Bmp2

Bmp4

Ptc1,Smo>Gli1

Bmpr1a>SMADs

Msx1

Dlx5

Fgfr2b>MAPK>c-Myc

Ptc1,Smo>Gli1

EphrinB?

EphrinB1-EphB2/3

Proliferation

Fgfr2b

MAPK

c-Myc

MAPK

PI3K

PTEN

MAPK, PI3K, PTEN

Noggin

Anterior Palate Fusion

BIOCHEMICAL CONTROL CIRCUITS

Mesenchyme

Epithelium

NASAL SIDE

ORAL SIDE

MEE

Fgf7

Bmp2

Bmp4

Ptc1,Smo>Gli1

Bmpr1a>SMADs

Msx1

Dlx5

Fgfr2b>MAPK>c-Myc

Ptc1,Smo>Gli1

EphrinB?

EphrinB1-EphB2/3

Proliferation

Fgfr2b

MAPK

c-Myc

MAPK

PI3K

PTEN

MAPK, PI3K, PTEN

Noggin

Anterior Palate Fusion

BIOCHEMICAL CONTROL CIRCUITS

Mesenchyme

Epithelium

NASAL SIDE

ORAL SIDE

MEE

Fgf7

Bmp4

Ptc1,Smo>Gli1

Bmpr1a>SMADs

Msx1

Dlx5

Fgfr2b>MAPK>c-Myc

Ptc1,Smo>Gli1

EphrinB?

EphrinB1-EphB2/3

Proliferation

Fgfr2b

MAPK

c-Myc

MAPK

PI3K

PTEN

MAPK, PI3K, PTEN

Noggin

Anterior Palate Fusion

BIOCHEMICAL CONTROL CIRCUITS

Mesenchyme

Epithelium

NASAL SIDE

ORAL SIDE

MEE

Fgf7

Bmp2

Bmp4

Ptc1,Smo>Gli1

Bmpr1a>SMADs

Msx1

Dlx5

Fgfr2b>MAPK>c-Myc

Ptc1,Smo>Gli1

EphrinB1-EphB2/3

Proliferation

Fgfr2b

MAPK

c-Myc

MAPK

PI3K

PTEN

Noggin

Anterior Palate Fusion

BIOCHEMICAL CONTROL CIRCUITS

Shh

Fgf10/Bmp2

Bmp4

Fgf7

Noggin

Tgfβ3

~BrdU

~TUNEL

WT (~E13.5)

~TUNEL

Shh

Fgf10/Bmp2

Bmp4

Fgf7

Noggin

Tgfβ3

~BrdU

WT (~E13.5)

Fgf10

Rice, R., B. Spencer-Dene, et al. (2004) J Clin Invest 113(12): 1692-1700.

Zhang, Z., Y. Song, et al. (2002) Development 129: 4135-4146.

~TUNEL

Shh

Fgf10/Bmp2

Bmp4

Fgf7

Noggin

Tgfβ3

~BrdU

WT (~E13.5)

Zhang, Z., Y. Song, et al. (2002) Development 129: 4135-4146.

~TUNEL

Shh

Fgf10/Bmp2

Bmp4

Fgf7

Noggin

Tgfβ3

~BrdU

WT (~E13.5)

Fgf7

Rice, R., B. Spencer-Dene, et al. (2004) J Clin Invest 113(12): 1692-1700.

~TUNEL

Shh

Fgf10/Bmp2

Bmp4

Fgf7

Noggin

Tgfβ3

~BrdU

WT (~E13.5)

Noggin

He, F., W. Xiong, et al. (2010) Developmental Biology 347: 109-121.

~TUNEL

Shh

Fgf10/Bmp2

Bmp4

Fgf7

Noggin

Tgfβ3

~BrdU

WT (~E13.5)

Jin, J.-Z., Q. Li, et al. (2008) Cell and Tissue Research 333: 29-38.

He, F., W. Xiong, et al. (2011) Developmental Biology 350: 511-519.

Shh

Fgf10/Bmp2

Bmp4

Fgf7

Noggin

Tgfβ3

~BrdU

~TUNEL

WT

Shh

Fgf10/Bmp2

Bmp4

Fgf7

Noggin

Tgfβ3

~BrdU

~TUNEL

WT

in silico knock-outs

in silico knock-outs

Linking the Model to Toxicology

Organ

Apposition

altered growth

Outcome

Structural malformation

Cleft Palate

altered reorientation

altered fusion

Cellular

Cell-level behaviors

Mesenchymal cell

↓ ECM production

↓proliferation

∆ migration

Molecular

Target pathways

Cleft Palate AOP Framework (N. Sipes)

TGFb Signaling

TGFb/BMP

Receptor TKs

EGF, FGF, EphB, PDGF

Epithelial

cell

∆ apoptosis

∆ proliferation

↓adhesion

∆ differentiation

Aryl Hydrocarbon

AhR

Tissue

Palatal shelf

↓ ECM remodeling/ matrix expansion

∆ medial edge degeneration

↓EMT

↓ Mesenchymal mass

Dioxin

(TCDD)

Mesenchyme

Epithelium

NASAL SIDE

ORAL SIDE

MEE

Proliferation

Apoptosis

Fgf7

Bmp2

Bmp4

Ptc1,Smo>Gli1

Fgfr2b>MAPK>c-Myc

Ptc1,Smo>Gli1

Bmpr1a>SMADs

Ptc1,Smo>Gli1

Shh

Msx1

Dlx5

EMT

Fgf10

Fgfr2b

MAPK

c-Myc

Tgfbr1

Tgfbr2

SMADs

MAPK

Fgfr2b>MAPK>c-Myc

Ptc1,Smo>Gli1

Proliferation,

Matrix Secretion

EphrinB?

Motility

EphrinB1-EphB2/3

Proliferation

Fgfr2b

MAPK

c-Myc

MAPK

PI3K

PTEN

MAPK, PI3K, PTEN

Noggin

Bmpr1a>SMADs

Anterior Palate Fusion

BIOCHEMICAL CONTROL CIRCUITS

Tgfβ3

Egf

Constant AhR Activation (~ TCDD)

Low Hysteresis Switch @ Threshold: 1.15x EGFR

Constant AhR Activation (~ TCDD)

High Hysteresis Switch @ Threshold: 1.2x EGFR

Transient AhR Activation (~ RA)

Low Hysteresis Switch: Up to 1.8x EGFR

Transient AhR Activation (~ RA)

High Hysteresis Switch: Up to 1.6x EGFR

Importance of Polarized Epithelia

​

• When we first developed the model with a single simple EPI cell type, the palate shelves fused WAY TOO EASILY. We could not prevent them from fusing. It was critical to make sure the epithelium was polarized.

Let’s look at the code for PalateModelUpdate.py and PalateModelSteppables.py to see how this was done.

​

We’ll then follow up with two exercises where you make a relatively simple polarized epithelium.

Exercise 5.3.1

​

Start with the NucleateCompartments.cc3d project. Comment out the random assignment of half the cells to type EPI. Instead, make differentiation from MESENCH to EPI a slower process that depends on whether a cell is on the outer rim of the blob, i.e., in contact with the medium.

​

To do so, change the steppable so that it checks to see which cells are in contact with the medium. If a cell of type MESENCH is in contact with the medium, then differentiate it to type EPI with a probability such that the outer rim of the blob - and only the outer rim - becomes type EPI in about 1000 MCS.

PSEUDOCODE (for each call to steppable)

- loop over all cells

- check whether each cell is in contact with the medium

- if a cell of type MESENCH is in contact with medium,

- convert to type EPI with an appropriate probability

CHALLENGE: Try to write your steppable so that the time required to differentiate the outer rim to EPI is the same regardless of whether you call the steppable every MCS or every 10th MCS or every 50th MCS?

Exercise 5.3.2

​

Build on solution to 5.3.1 so that after a cell has differentiated into an EPI cell, it then polarizes to have apical and basal compartments with different properties. The basal compartment can be the original EPI cell type, but we will need a new apical domain (say EPIa). Goal is to generate the polarization, maintain it, and keep the polarized epithelium well organized with all apical domains on the outside in contact with medium

and all basal domains on the inside in contact with MESENCH cells.

​

​

PSEUDOCODE (for each call to steppable)

- loop over all cells and do the following for each cell of type EPI

- check whether the cluster to which it belongs also has an EPIa compartment

- if not, then nucleate one on part of the EPI cell’s surface in contact with medium

- nucleation will require creating a new "cell" of type EPIa, assigning some lattice sites to it, and setting its volume and surface energy parameters

​

Feel free to copy over the getPixelsInContactWithMedium function from PalateModelUpdatedSteppables.py. It will be useful here.

​

CHALLENGE: Add a line that randomly kills off a few basal EPI compartments at each step. Then add code that will check whether any EPIa cells exist without a complementary EPI compartment and if so, nucleate an appropriate complement.

Module 3.5 Questionnaire

Support: NIH NIBIB-U24EB028887, NIGMS-R01GM122424, NSF-188553, NSF-186890, NSF-1720625, NIGMS-R01GM076692, NIGMS-R01GM077138, EPA – 83573601, ORISE

Please take a minute or two to let us know about your experience with this module by filling out the brief zoom survey.

​

Feel free to provide additional comments and suggestions in the slack or by email to us as well

haydenfennell@gmail.com

or

m.shane.hutson@vanderbilt.edu

​