Validation of a Systems Model for Human Bronchial Epithelial Cells with and without Cystic Fibrosis

Cystic fibrosis (CF) is an autosomal recessive disorder caused by defects in the cystic fibrosis transmembrane conductance regulator (CFTR) gene. This defect leads to absent or malfunctioning chloride channels on the airway epithelium. Sodium absorption across the epithelium is also typically increased through the ENaC channel.  Together, these effects lead to increased airway surface liquid absorption, the accumulation of dehydrated mucus, and respiratory failure.

We seek to measure liquid and solute transport in a human bronchial epithelial (HBE) cell model in order to support the development of a systems model of airway epithelial physiology. Using a small radiolabeled molecule, diethylene triamine pentaacetic acid (DTPA), we are able to quantify airway surface liquid (ASL) volume changes in both CF and non-CF human bronchial epithelial cell cultures. Experiments used apical mannitol addition to modulate ASL absorption rate. Apical mannitol addition slows DTPA clearance by creating a transepithelial osmotic gradient determining direction of water absorption, and is used as a “zero convection case” which allows us to determine whether there are basic permeability differences between CF and non-CF epithelia. A hyposmotic case was also tested using apical addition of diluted (150 mOsm) Ringers solution, thereby creating an environment favoring liquid absorption. These results were compared to the findings for the same procedure using undiluted (300 mOsm) Ringers solution. An optical method is used to calculate these changes in ASL volume, which are compared to DTPA absorption findings. These studies are important tools to verify the accuracy of our systems model of cell volume regulation.

Validation of this model in response to osmotic challenges was performed using the data sets described previously. In response to hyposmotic Ringers addition, running the model under validation conditions produced the same trends that were shown experimentally: an increase in both DTPA and liquid absorption in both CF and non-CF HBE lines. In CF HBEs, the model predicts a subtle difference in absorption dynamics between the control and experimental groups, recognizing that CF cells hyperabsorb liquid even in an isosmotic environment. Model training in response to hyperosmotic mannitol showed deviations from the Ringers data. Validation of this parameter set against a separate set of filters obtained under identical conditions demonstrated a reduction in liquid and DTPA absorption in CF and non-CF HBEs – mirroring earlier outcomes of the model fitting data. So, while the model predictions and the data under mannitol challenge show similar trends, dynamic behavior, particularly at later time-points, is not matched between the model and the data, which may indicate the action of physiological mechanisms yet unincorporated in the model structure.

In vitro, apical mannitol addition is shown to reduce DTPA clearance and liquid absorption in CF and non-CF HBEs, while hyposmotic Ringers addition increases DTPA clearance and liquid absorption. Experimental data for these test cases serve as validation sets for cell behavior predictions via a systems model.  These experiments and the development of a systems model are important tools for determining more effective and personalized treatments for CF patients in the future.