On 2017 May 18, Jen Herman commented:

I would like to offer a possible alternative interpretation to explain the gain of interaction variants we identified for both SirA and DnaA that we did not conceive of at the time of publication.

In the gain of interaction screen (bacterial two-hybrid - B2H) we obtained the surprising result that none of the variants we identified (either in SirA or DnaA) occurred near the known SirA-DnaA interaction interface. The DnaA gain of interaction substitutions occurred primarily in the region of DnaA important for DnaA oligomerization (Domain III). If these variants are defective for DnaA self interaction, then they might also be more available to interact with SirA in the B2H.

If SirA, like DnaA, is also capable of forming higher order oligomers (at least at the higher copy numbers likely present in the B2H), then it is also conceivable that the gain of interaction variants we identified within SirA are also defective in this form of self-interaction. One piece of data to suggest this hypothesis might be correct is that truncating several amino acids from SirA's C-terminus (including the critical P141T residue) increases SirA solubility following overexpression. Previously, we and others were unable to identify conditions to solubilize any overexpressed wild-type SirA. Of course, this could simply be due to a propensity of SirA to form aggregates/inclusion bodies; however, another possibility is that SirA has an intrinsic tendency to oligomerize/polymerize at high concentrations, and that SirA's C-terminal region facilitates this particular form of self-interaction.

If any of this is true, one should be able to design B2H gain of interaction screens to identify residues that likely disrupt the suspected oligomerization of any candidate protein suspected to mutltimerize (as we may have inadvertently done). This could be potentially useful for identifying monomer forms that are more amenable to, for example, protein overexpression or crystallization.

In the bigger picture, one wonders how many proteins that are "insoluble" are actually forming ordered homomers of some sort due to their chiral nature. Relatedly, would this tendency be of any biological significance or simply a consequence of not being selected against in vivo (especially for proteins present at low copy number in the cell)? (see PMID 10940245 for a very nice review related to this subject).

On 2017 May 18, Jen Herman commented:

I would like to offer a possible alternative interpretation to explain the gain of interaction variants we identified for both SirA and DnaA that we did not conceive of at the time of publication.

In the gain of interaction screen (bacterial two-hybrid - B2H) we obtained the surprising result that none of the variants we identified (either in SirA or DnaA) occurred near the known SirA-DnaA interaction interface. The DnaA gain of interaction substitutions occurred primarily in the region of DnaA important for DnaA oligomerization (Domain III). If these variants are defective for DnaA self interaction, then they might also be more available to interact with SirA in the B2H.

If SirA, like DnaA, is also capable of forming higher order oligomers (at least at the higher copy numbers likely present in the B2H), then it is also conceivable that the gain of interaction variants we identified within SirA are also defective in this form of self-interaction. One piece of data to suggest this hypothesis might be correct is that truncating several amino acids from SirA's C-terminus (including the critical P141T residue) increases SirA solubility following overexpression. Previously, we and others were unable to identify conditions to solubilize any overexpressed wild-type SirA. Of course, this could simply be due to a propensity of SirA to form aggregates/inclusion bodies; however, another possibility is that SirA has an intrinsic tendency to oligomerize/polymerize at high concentrations, and that SirA's C-terminal region facilitates this particular form of self-interaction.

If any of this is true, one should be able to design B2H gain of interaction screens to identify residues that likely disrupt the suspected oligomerization of any candidate protein suspected to mutltimerize (as we may have inadvertently done). This could be potentially useful for identifying monomer forms that are more amenable to, for example, protein overexpression or crystallization.

In the bigger picture, one wonders how many proteins that are "insoluble" are actually forming ordered homomers of some sort due to their chiral nature. Relatedly, would this tendency be of any biological significance or simply a consequence of not being selected against in vivo (especially for proteins present at low copy number in the cell)? (see PMID 10940245 for a very nice review related to this subject).