Differential affinities between coats and SNARE proteins can generate distinct compartments.
The new model only worked when it incorporated differential affinity of one coat protein for one set of SNARE proteins versus the other. If both coat proteins bound all SNAREs with equal affinity, then bidirectional vesicle transport would result in two uniform organelles. If the model assumed that one coat protein bound one pair of SNAREs with at least a 10-fold preference over the other SNARE pair, then the model accurately maintained two unique compartments.
One nonintuitive aspect of the model is that it predicts that both members of a SNARE pair, the v- and t-SNAREs, accumulate in the target compartment, a prediction borne out by experimental observations from other groups.
COPII is known to have high affinity for SNAREs that target the Golgi, and the model predicts that COPI should preferentially bind to ER SNAREs. If, however, the coat protein's binding affinity for a SNARE pair is too strong, then all of those SNARE proteins would accumulate in the target compartment, leaving vesicles to bud from the other compartment without any SNAREs that would allow it to fuse to the target membrane. The model also predicts that if one SNARE pair is overexpressed, then the size of its target compartment would increase.