Curtin-Hammett and Effective Molarity

This week’s Thursday Morning Problem Session involved a presentation by my graduate student, Soo, on some work published by Boons that concerned a stereoselective synthesis of 1,2-cis glycosides. The value of this chemistry is the bond forming event shown (in 2), guided by the attached chiral auxiliary:

Stereoselective Glycoside Formation by Geert-Jan Boons

The stereocentre in the appended group dictates the nature of the second ring, since the phenyl on the stereocentre goes equatorial. This therefore dictates the equatorial nature of the C-S bond. Because of all this, Soo suggested, the incoming nucleophile forms a bond that is axial stereoselectively. The chemistry behind the approach was originally described here.

At the end I questioned whether that’s a realistic explanation. Assume that the closure of that second ring by the intramolecular directing group is reversible, even to a very small degree (i.e. equilibrium between 1 and 2). Compound 1 will be much more reactive than compound 2 towards intermolecular attack by the hydroxyl (derived from the next sugar in the sequence). If we assume that the activation energy barriers of the attack of the intermolecular nucleophile are significantly larger than the barriers of the intramolecular reaction, then we can apply Curtin-Hammett and say that the form of the sugar that preferentially reacts with the incoming nucleophile is 1 (the more reactive of the two) no matter that it is the less prevalent form of the substrate. Naturally if that’s the case, then the oxacarbenium ion carbon that’s attacked is planar, ruining the argument about stereoselection. A possible way out is to say that immediately after formation of the oxacarbenium ion 1 from 2 (i.e. assuming it’s reversible), the local environment around the (soon-to-be) anomeric carbon is still influenced by the pendant chain (i.e. the molecule looks like 3), and that reaction is still therefore stereoselective. This argument is somewhat similar to the “ion pair” arguments that are invoked to explain nonracemic outcomes from apparently SN1 reactions of enantiopure substrates.

Professor Crossley objected that the effective molarity (sometimes called “intramolecularity”) of the thioether was very high. What he’s referring to here is the extraordinarily high effective concentration of nucleophiles that are performing intramolecular cyclisations. Anslyn and Dougherty (Section 9.2) have some nice data on this, derived from Tony Kirby’s paper from 1980:

Effective Molarity

It’s intuitively obvious that constraining a nucleophile within the same molecule for a cyclisation can have dramatic effects on the rates of that cyclisation. So it is certainly true that the cyclised form 2 (rather than 1) is likely to be kinetically greatly favoured because of this. But if there is any oxacarbenium ion present, I think it will react rapidly and in preference to 2. Question is, is that ring closure reversible?