Extended Data Fig. 3: Tumor cell networks have scale-free and small-world properties, and their functional network hubs frequently display periodic calcium oscillations in vivo and in vitro.
From: Autonomous rhythmic activity in glioma networks drives brain tumour growth
a—c, Mean probability distribution of coactive cells per cell from in vitro and in vivo Ca2+ imaging plotted in a log-log scale with linear regression fit in red and the fitted random Poisson distribution in green; the grey box indicates highly connected network hubs; Pearson correlation. d, ‹k› (mean number of connections per cell) and σ (standard deviation of number of connections per cell) of all recorded networks from each condition. For a random network with Poisson degree distribution the standard deviation of the degrees follows σ = ‹k›1⁄2 shown as a red dashed line on the figure. For each network σ is larger than the value expected for a random network with the same ‹k›, which leads to the manifestation of scale-free properties in the networks7. e—h, Network parameters λ (mean shortest path length) and σ (clustering coefficient) calculated for recorded networks and their corresponding Erdős-Rényi random networks of equal number of nodes and edges; error bars show s.e.m.; paired two-sided t-test. a—h, n = 24 recordings from 3 mice (S24 in vivo), n = 22 recordings from 9 biologically independent experiments (S24 in vitro), n = 40 recordings from 4 mice (BG5 in vivo), and n = 10 recordings from 5 biologically independent experiments (BG5 in vitro). i—n, Network plot of cross-correlation coefficients larger than cut-off derived from the same Ca2+ recordings of tumor cells; BG5 line in vitro (i,l); S24 line in vivo (j,m); BG5 line in vivo (k,n); scale bars are 100 μm (i) and 50 μm (j,k). o—q, Ca2+ transients of indicated TM-connected tumor cells; dark red transients originated from a periodic cell. ns, not significant (P ≥ 0.05).
