Network-based approaches to quantify multicellular development

Abstract

Multicellularity and cellular cooperation confer novel functions on organs following a structure-function relationship. How regulated cell migration, division and differentiation events generate cellular arrangements has been investigated, providing insight into the regulation of genetically encoded patterning processes. Much less is known about the higher-order properties of cellular organization within organs, and how their functional coordination through global spatial relations shape and constrain organ function. Key questions to be addressed include: why are cells organized in the way they are? What is the significance of the patterns of cellular organization selected for by evolution? What other configurations are possible? These may be addressed through a combination of global cellular interaction mapping and network science to uncover the relationship between organ structure and function. Using this approach, global cellular organization can be discretized and analysed, providing a quantitative framework to explore developmental processes. Each of the local and global properties of integrated multicellular systems can be analysed and compared across different tissues and models in discrete terms. Advances in high-resolution microscopy and image analysis continue to make cellular interaction mapping possible in an increasing variety of biological systems and tissues, broadening the further potential application of this approach. Understanding the higher-order properties of complex cellular assemblies provides the opportunity to explore the evolution and constraints of cell organization, establishing structure-function relationships that can guide future organ design.

Keywords: complexity; development; multicellularity; network science; self-organization; structure–function.

Figure 1.

Discretization and abstraction of cellular organization into networks. (a) Cellular interaction mapping leads to the generation of networks where the nodes represent cells and edges their physical interactions. (b) A diagram of a cell interaction network typical of epithelial tissues in plants and animals. (c) A diagram of a part of a directed network of neuronal interactions. Information flows from the axon of a neuron to the dendrites of connected neurons giving the edges directions.

Figure 2.

Structural and functional networks of the Birmingham, UK, rail system. (a) The structural network of the Birmingham rail system, showing possible routes (edges) that can be taken between stations (nodes). (b) Functional annotation of the rail network, where edges are false coloured by the frequency of trains running between stations between 08.00 and 10.00 on weekdays. (c) Functional annotation of the rail network, where edges are false coloured by the average speed of trains between 08.00 and 10.00 during a weekday.

Figure 3.

Topological features of virtually generated planar cellular connectivity network. (a) Degree false coloured on the virtual tissue and the corresponding networks according to the scale provided. (b) Same as (a) with betweenness centrality. (c) Same as (a) with random walk centrality.

Figure 4.

Using a network framework to understand multicellular systems. (a) Experimental evolution of multicellularity by Ratcliff et al. [49]. As the number of sequential transfers increases, a unicellular (UC) yeast population evolves to become multicellular (MC) with higher sedimentation speeds. (b) Newly evolved phenotype called ‘snowflake’ (inset), with ensembles of cells remaining physically attached due to defective fission. Cells undergoing apoptosis are stained red. These events will split up the aggregate, leading to further growth. Below, the actual network underlying this particular ‘snowflake’, with node betweenness centrality depicted in shades of green and an arrow pointing at the sole apoptotic cell in this ‘snowflake’. (c) Violin plot of betweenness centrality, separating the cells into apoptotic and non-apoptotic. Dots and lines represent the means of each distribution, three aggregates and 118 cells.