Brain Tumor Growth Model

For several years, members of the Weitz lab have been working in collaboration with several other groups (Tom Deisboeck at Massachusetts General Hospital, Mike Berens at Tgen, Len Sander at the University of Michigan, and Antonio Chiocca at Ohio State University) to understand details about the motion of a particular type of brain tumor cell, Glioblastoma Multiforme (GBM). GBM is a brain cancer that kills its victims quickly because it is highly invasive and because surgery to remove such tumors inevitably leaves much of the tenuous network of invasive cancer cells behind. There are many different aspects of this project, all aimed at understanding the motion of GBM and ways of curtailing it. The Weitz group is involved in the following aspects of this work:

 

1. Mechanical properties of the ECM

Previous work in the lab, done by V. Gordon, has shown that the tumor is exerting significant force outward on its surroundings as it volumetrically expands, but that it also exerts tractional, inward forces as the tips invade. The data previously acquired also suggests that the tumor is remodelling the matrix of the surrounding gel.

Both figures to the right contain particle tracks over several hours with early times in blue and late times in red. The closer image is from the day on which the tumor was implanted, at which point it had not yet developed invasive tips. The second image is taken the next day, further from the spheroid, where one invasive tip had developed, as can be seen faintly in the bright field image.

We are currently attempting to model the surrounding matrix using a finite element method (FEM) so as to quantify the existing strains and stresses.
 

 

2. Cell-ECM interactions

Previous work in the lab, done by L. Kaufman, K. Kasza, C. Brangwynne and E. Filippidi has focused on determining the dependence of tumor growth and invasion as a function of ECM density. To accomplish this, tumor spheroids were placed in collagen gels of various concentrations and their growth was monitored. We accomplish simultaneous imaging of the collagen fibers using confocal reflectance microscopy and of the cells using CARS (coherent anti-Stokes Raman scattering) microscopy (top right). Imaging of the collagen fibers exclusively has allowed some characterization of the correlation between mesh size and gel stiffness (bottom right).

In the future, we wish to label the collagen ECM to visualize the activity of MMPs (matrix metallo-proteinases) secreted by the invasive cells.

 

 
Cells sparsely seeded in our collagen type I matrices tend to contract those gels over the course of several hours to several days. The amount of contraction exerted by the cell population will depend on collagen density, as well as cell number density.

With values of collagen density ranging from 0.5mg/mL (top) to 1.8mg/mL (bottom) and cell number from 103 to 6x104, U87 glioblastoma cells contract the gel to various sizes in LabTek chamber slides.

 
When sparsely seeded cells in collagen contract the gel, they pull on the fibers (until those break?) and remodel their environment, almost acting as "force dipoles", which could be characterized by i) a direction, ii) a force amplitude and iii) a decay length. We wish to quantitatively study these three characteristics of cells as a function of collagen and cell number density.

One U87 glioblastoma cell (GFP-labeled nucleus, green) seen remodeling the collagen fibers (confocal reflectance imaging, red) surrounding it.

  

 

 

3. Cell motion and cell-cell interactions

Cells transfected with the h2B-GFP gene can have their nuclei imaged and tracked under fluorescence microscopy, both 2D wide field and 3D confocal. Tumor spheroids generated by the hanging droplet method are inserted into a 3D collagen matrix (0.5-2.0 mg/mL) and subsequently imaged. A timelapse acquisition of the position of the cells allows us to track and analyze their motion.

Tracking in 2D and 3D is an important tool, which will help us determine the spread of the tumor at a given time. Additionally, plating cells at different densities will allow us to quantify how closely cells interact.

Top image is a set of tracks generated during overnight imaging of a tumor spheroid. Blue cells are less mobile than red cells. Middle image is one cell track, where color coding corresponds to displacement amplitude, red and orange being greater than yellow and green. Bottom image plots vertical position z of cells versus their y position, and shows a peculiar motion of cells outward and downward from the MTS, which we are attempting to explain.


 

4. Spheroid density

By imaging h2B-GFP transfected glioblastoma cells with a confocal microscope, we get quantitative data on the location and density of cells within the tumor spheroid. Characterization of the evolution of density over time can give us some information on the development and span of the tumor.

Plot to the right is an example of density as a function of radius, which follows closely an exponential decay.

5. Facilitated tracking through a graphical user interface

 

The Weitz lab has developed many tracking algorithms and programs over the past years, both in the IDL and Matlab programming languages. Matlab already has the benefit of a reasonably "friendly" user interface, but IDL is a text-based programming language, which may seem quite unfriendly to start with.

To facilitate the use of tracking for researchers who don't necessarily want too much programming, we are developing a graphical user interface (GUI) for particle tracking, where parameters can be set through sliders, and modules can be chosen through drop-down menus.

Image to the right displays the main tracking GUI, with parameter sliders to the left, and the output of the bpass.pro function (original image, top; bandpassed image, bottom).

 

Last updated: February 2005.
If you have any questions, feel free to contact David Vader, Cliff Brangwynne or Karen Kasza from the Weitz lab.