(790e) Cell-to-cell variations define age-dependent cell migration through fractional redistributions among motility states | AIChE

(790e) Cell-to-cell variations define age-dependent cell migration through fractional redistributions among motility states

Authors 

Phillip, J. - Presenter, Johns Hopkins University
Zamponi, N., Weill Cornell Medicine
Daya, J., Weill Cornell Medicine
McGovern, S., Lummus Technology, a CB&I Company
Williams, W., Weill Cornell Medicine
Phillip, M. P., Weill Cornell Medicine
Jayatilaka, H., Johns Hopkins University
Wu, P. H., Johns Hopkins University
Walston, J., Weill Cornell Medicine
Wirtz, D., Johns Hopkins University
As humans age, cells within organs and tissues undergo profound biophysical and biomolecular changes, which significantly influence the rate of progressive functional decline. Recently we have demonstrated that aging information is encoded within biophysical properties of cells, and as such can be used as biomarkers of aging in health and disease. The coordinated movement of cells within organs and tissues are important in the contexts of both homeostasis and disease; such as in the surveillance of immune cells and during cancer metastasis. Although it is now recognized that there are ensemble decreases in the motile capacity of cells with increasing age, it is unclear how this motile regulation becomes impaired. To address this question we procured a panel of primary dermal fibroblasts from donors aged 2-92 years old, and tracked the movement of cells seeded on collagen-coated substrates for ten hours, taking images every 3 minutes. Quantification of bulk motility properties using the anisotropic persistent random walk model, revealed that there was indeed a decrease in the overall capacity of cells to move with increasing age (including displacement, directionality and velocity). Looking further into these findings, we plotted the trajectories of individual cells across all ages, and found that there was a high decree os cellular heterogeneity among cells, with some cells from elderly donors exhibiting highly-motile features, and vice versa. Prompted by the extent of cell-to-cell variations, we hypothesized that these age-dependent decreases in cell motility was not due to global decreases in the movement of all cells, but a re-distribution of the proportions of highly-motile and non-motile cells with increasing age. Taking advantage of the single-cell nature of the data, we compiled the trajectories of cells across all ages and computed eight motility parameters that described the spatiotemporal movement; e.g. velocity, displacement, anisotropy and diffusivity. Then from this data matrix we used unsupervised hierarchical clustering and identified eight clusters which defined ‘motility states’ exhibiting distinct spatial patterns, as indicated by the trajectories of individual cells within each cluster, and t-stochastic neighbor embedding analysis. Next, we asked whether certain clusters had increased propensities for cells of a particular age. Quantifying the abundance of cells classified per cluster for all ages, we found that there was an increased propensity of various cluster for young and elderly donors, however, for post-adolescent/adult donors there was a similar likelihood of either cluster. This finding was interesting as it points to increased heterogeneity during the ‘middle-age’ period. To further verify this result, we quantified the ‘activity’ of cells per age, as determined by the 1- dimensional displacements per cell based on the frequency and magnitude of displacement bursts. Utilizing a similar clustering approach as before, we identified five clusters that defined the activity of each cell across all ages. interestingly, while the temporal patterns of the bursts were different among groups, plotting the trajectories of individual cells we were unable to identify distinct spatial motility patterns. However, plotting the abundance of each ‘activity’ cluster per age, showed a less-defined but similar tendency in cellular heterogeneity. Combining the information regarding the spatial-temporal and activity per cell, we computed the abundance of cells within the 40 motility states (8 spatio-temporal clusters x 5 activity clusters) to define motility landscapes, and asked how this differed as a function of age group. Interestingly, we found that there was a clear difference among age groups regarding the propensity for certain states, with the significant gain and loss of states across ages. Quantifying the Shannon entropy of each landscape, we confirmed the results of a maximization of heterogeneity for donors classified as post-adolescent/adults.