Purpose
The purpose of this work was to create an in silico Monte Carlo model that predicts the expected association of intravitreally injected adipose-derived stem cells (ASCs) with retinal vasculature, assuming a random distribution of cells. These predictions were then compared to experimental measurements of ASC associations with in vivo retinal microvessels. This comparison allowed us to validate this Monte Carlo method as a means for determining whether the observed in vivo data is due entirely to random chance or might suggest preferential adhesion or migration of injected ASCs towards the retinal microvasculature.
Methods
Microvessel network architecture in the model is derived from confocal images of excised murine retinae labeled with lectin to visualize the vascular endothelium, as described in more detail below. The simulation records the percent of ASCs that overlap with vascular networks for each trial, then repeats the process and displays the average percent of the ASC population that co-localizes with the vasculature across all of the trials. A histogram is generated to show the distance each ASC is away from the nearest vasculature landmark.
Results
To determine the efficacy of the simulation, we used fluorescent confocal images (20X magnification) of retinal wholemounts. Eyes were previously injected intravitreally with DiI labeled murine ASCs in eight week old Akimba or Akita mice. Eyes were harvested 1 month later, retinal wholemounts prepared, and retinal vasculature labeled with lectin. ASC association with the vasculature was determined through blind counting the percentage of murine ASCs contacting the vasculature. Using a Bland-Altman analysis, we showed that there was no difference between the Monte Carlo simulation and the blind counting method to measure the percent of intravitreally murine ASCs contacting retinal vasculature (Figure 1).
Conclusions
Our results suggest the Monte Carlo simulation is a valuable stochastic simulation to determine the number of intravitreally injected ASCs that make contact with retinal vasculature purely by chance. The simulation can be extended to other cellular applications in the eye to determine whether cell location and co-localization is caused by random chance or endogenous directional signals.
Keywords: 473 computational modeling •
721 stem cells •
688 retina