• We tested the effects of varying the membrane/nucleoid diffusion constants: in the wild-type model, D₃ was set to zero. However, setting it equal to 0.006 m m²·s^-1 (the same as the other membrane/nucleoid diffusion constants) did not appreciably alter our results (with the exception of widening the Soj membrane distribution, see Fig. 5). Setting D₄ = D₅ = 0 m m²·s^–1 also did not significantly alter the model dynamics. However, setting D₂ = 0 m m²·s^–1 did destroy the Soj relocations. The key point here is that the membrane/nucleoid diffusion constants must be small compared with the cytoplasmic diffusion constants; with the exception of removing Soj nucleoid diffusion altogether, their precise values appear to be unimportant. Varying the cytoplasmic diffusion of Soj also had only a weak effect on the Soj dynamics: any diffusion constant larger than about D₁ = 1 m m²·s^–1 yielded significant Soj relocations.

• We also varied the reaction rates: Soj relocations persisted even when k₁, k₂, k₃, k₅, s ₁, and s ₂ were each separately increased by a factor of 2. However, increasing k₄ and k₆ by a factor of 2 did abolish the relocations. Separately reducing k₄, k₆, and s ₃ by a factor of 2 again led to Soj relocations. Hence, the Soj relocations are fairly robust to changes in the kinetic constants. The variations in k₁, k₂, k₃, k₄, s ₁, and s ₂ were performed on the Spo0J19 mutant model I, and the variations in k₅, k₆, and s ₃ on the wild-type model.

• We altered the exponents governing the cooperativity of the Soj binding/unbinding dynamics. Setting either s ₁ or s ₃ to zero, with all other parameters having their wild-type values, suppressed the Soj relocations. Setting s ₂ to zero abolished relocations in simulations of the Spo0J19 mutant (both models). Cooperativity, therefore, appears to be an important element in generating realistic Soj dynamics. However, provided the s terms are nonzero, there is still flexibility in the choice of the exponents in the cooperativity terms (previously set equal to 2 for the s ₁ and s ₃ terms, and unity for the s ₂ term). For example, using otherwise wild-type parameters but setting the exponent to 1.5 for the s ₁ and s ₂ terms and with s ₁ and s ₂ increased by factors of 10 and 40 also yielded Soj relocations, albeit of an increased frequency and reduced amplitude. Using an exponent of 2.5 for the s ₁ and s ₂ terms and with s ₁ and s ₂ reduced by factors of 20 and 10, relocations were observed to be similar to those in the wild-type model.

• We also tested several variations on the nature of the cooperative Soj unbinding from the nucleoid. For example, we altered the model so that the Spo0J condensation process was cooperative, meaning that the Spo0J condensed preferentially where condensed Spo0J was already present. A high local density of condensed Spo0J could then be responsible for the cooperativity in the Soj expulsion process. However, despite some effort, these modifications suppressed the Soj dynamics. Hence, we currently favor our model where the Soj catalyzes its own disassociation in the presence of condensed Spo0J. However, given the complexity of the Spo0J/chromosome condensation, we cannot rule out the possibility that this failure is a consequence of an oversimplification in our modeling. To test this point further, more data are needed so that more realistic models can be constructed of the Spo0J/chromosome reorganization dynamics.

• Adding protein production and degradation with a Soj half-life of 30 min did not significantly alter the dynamics in simulated wild-type cells.

• We also tested a model where the requirement that Soj first bind to the polar membrane before rebinding to the nucleoid was relaxed. In this implementation, there was only a single species of cytoplasmic Soj. This form was assumed able to bind to either the nucleoid or MinD/polar membrane (by using the same k₁ and k₅ terms as in the wild-type model equations): these regions simply competed for Soj binding. However, using this modified model, we were unable to reproduce the results found in experiments. The difficulty here is that to obtain relocations in simulated wild-type cells, the Soj must be able to bind relatively easily to any given nucleoid. However, in filamentous cells, this ability has the consequence that it is easy for the Soj to diffuse away and populate nucleoids well away from polar regions. In essence, the presence of extra Soj binding sites close to the poles is insufficient to "pin" the Soj close to the pole and prevent the Soj from diffusing away. Finally, we mention that this model is also made less likely by experimental results in minD mutants, where membrane Soj binding is prevented and the Soj dynamics virtually abolished (at least in exponential phase). This behavior is not what one would expect if the MinD/polar membrane were simply competing for Soj binding.