Latency escapes and model robustness. Figure showing the minimum instantaneous change in EBNA-1 protein number necessary to switch from latency I to III and vice versa, as a function of the number of Oct-2+Grg/TLR proteins. For small numbers only the latency III state exists, while for large numbers only latency I state exists, compare Figure 3 and Figure 4. Left figure: red lines are at the reference parameter values, in particular Oct-2 complex binding with affinity K
= 2.5 nM and an EBNA-1 dimerization dissociation constant of 1 nM. Blue lines show a five-fold weaker Oct-2 affinity (K
= 12.5 nM). Similar behaviour is then displayed at approximately five-fold higher Oct-2 level. Right figure: red solid and dotted lines at reference parameter values. Blue solid and dotted lines at a tenfold stronger EBNA-1 dimerization, and green solid and dotted lines at a tenfold greater volume. The latency III state is relatively robust towards either of these changes, while the latency I state changes more. Black solid and dotted lines show both a tenfold stronger EBNA-1 dimerization and a tenfold greater volume. This influences the latency III state more, essentially because EBNA-1 concentration in the latency III state is then comparable to the EBNA-1 dimerization dissociation constant.