S on adhesion heterogeneity. In PubMed ID:https://www.ncbi.nlm.nih.gov/pubmed/26240184 summary, even though it is known that spatio-temporal variability in adhesion receptor expression should not be neglected in tumour cell population models [35], the effects of those dynamics have not been studied systematically so far. We propose a two-dimensional multiscale cellular automaton (CA) model that couples cell-cell adhesionReher et al. Biology Direct (2017) 12:Page 3 ofwith intracellular adhesion receptor regulation. Cellular automata are widely used as models for different aspects of cancer dynamics [36?3]. A particular type of cellular automata, the lattice-gas cellular automaton (LGCA), is well-suited to model cell-cell interaction and cell migration [28?0]. Several LGCA models have been introduced to study tumour invasion. For instance, B tger et al. [44?6] and Hatzikirou et al. [47] developed LGCA models to study consequences of the `go-or-grow’ dichotomy on cancer growth and invasion. The `go-orgrow’ dichotomy is known to characterise glioma cancer cells [48]. Here, we analyse the effects of adhesion heterogeneity on tumour cell dissemination and couple an LGCA model for CBR-5884 web adhesive cell interaction to an adhesion receptor model adapted from Engwer et al. [49]. For this, we compare simulations of four model scenarios. In particular, we distinguish scenarios in which single cell adhesion receptor regulation is either independent from or controlled by the local cell density and further consider both homogeneous and heterogeneous cell populations, the latter ones with different degrees of adhesion heterogeneity. For our analysis, we count the number of disseminated cells that migrated beyond a threshold distance and measure mean adhesion receptor concentrations of nondisseminated and disseminated cells at a given time. This allows us to characterise potential adhesivity differences between these two subpopulations. We predict that the degree of adhesion heterogeneity determines the size of the disseminated cell subpopulation.MethodsDefinition of the multiscale modelthe cell surface. The adhesive states are regulated by an adhesion receptor regulation rule which accounts for extrinsic and intrinsic sources of adhesion heterogeneity in the model. Adhesive interactions between cells are realised by a migration rule, which depends on the adhesive states of the cells: cells perform biased random walks such that cells with high adhesive state values have a higher probability to be attracted to neighbouring cells than cells with lower adhesive state values. The LGCA model is described on a discrete ddimensional regular lattice L with periodic boundary conditions. Each lattice node r L is connected to its b nearest neighbours by unit vectors ci , i = 0, . . . , b, called velocity channels. The total number of channels per node is defined by > b, where – b is an arbitrary number of channels with zero velocity, called rest channels. Each channel can be occupied by at most one cell at a time, defined by the occupation state variable (r, k) 0, 1. We distinguish moving cells, which reside on the velocity channels, indexed by i = 1, . . . , b, and resting cells, which are located within the rest channels of the lattice, indexed by i = b + 1, . . . , . Adhesive cell states are defined by a variable ai (r) [ 0, ). The total number of cells at time k and node r is given by n(r, k). The parameter is a local cell number bound which is imposed, since the maximal cell number in a given volume is limited in.