Two principal modern methods for enantioselective aldol reactions — the Evans aldol and the Masamune aldol — use a carboxylic-acid derived enolate rather than a ketone enolate as e.g. in the Paterson aldol reaction. For ‘traditional’ ketone enolates, a rule of thumb dictates that (E) enolates give anti-products and (Z) enolates give syn-products, typically rationalised by the Zimmermann-Traxler transition state model. Assigning (E) or (Z) to the enolate of an ethyl ketone poses no problems, as oxygen always has a higher Cahn-Ingold-Prelog priority than carbon, so it is all about the orientation of the enolate oxygen and the methyl group (see figure 1).
An Evans aldol reaction, the enolate of which is also shown in figure 1, can be analysed similarly. Oxygen has a higher priority than nitrogen, so the same basic rules apply.
A Masamune aldol reaction, however, has two oxygens attached to the enolate carbon, formally making it an O,O-ketene acetal. The CIP rules would dictate to follow the connectivities atom by atom. The Masamune auxilliary, like any ester, has a carbon attached to the oxygen. The enolate anion may be strongly attached to a certain atom (e.g. when using a boron- or titanium(IV)-mediated aldol reaction) but also may not be, or the attachment may be ambiguous (e.g. a potassium salt used for enolating in the presence of $\ce{LiCl}$).
Figure 1: The enolates in a Paterson-, Evans- and Masamune-aldol reaction and their double bond configurations where known.
To establish whether the double bond geometry is (E) or (Z), must I know if and which atom is attached to said oxygen? What if two different possibly attaching atoms (e.g. potassium and lithium) would give two different geometries? Is it possible to ignore said attachment altogether and just define the enolate oxygen as a free monoanion?
And do the CIP rules differ from common usage?