I’ve been meaning to write this post for a while, but the right words never seemed close at hand. Thanks to Michelle for giving me the inspiration.
Around here model kits seem to be one of the most under-appreciated pieces of equipment, discarded after coursework is over. This may be because most kits are awkward to handle, seemingly designed more for style than functionality. Even the best kits take quite a bit of time to set up, and it’s unlikely that anyone could make a Viagra model in twenty seconds or less.
But models can provide insights that are very difficult to obtain by other means. Sketches on paper or ChemDraw are flat and lifeless, while 3D modeling suites show only a single conformation, hiding flexibility. With a good model you can actually feel the ring strain, and see the impact of high entropy and steric hindrance. Chemical reactions become a lot more intuitive when you can actually see Burgi-Dunitz angles and build endo/exo tet/trig/dig rings .
For the past few years I’ve been using Indigo’s Orbit kit, student edition . This kit is very similar to the older Chem-Tutor set I got from Aldrich, but about half the price and significantly more resilient to punishment. Sigma bonds are represented by rigid tubes that bend only slightly under pressure, which stiffens cyclic structures and mimics real systems. The tubes collapse under sufficient strain, but will usually recover.
Now let’s put this in the context of the question I asked yesterday. A few years back I was starting a project on the aminoglycoside tobramycin. Tobramycin has five hydroxyl groups and I wanted to protect all but the central site, which had a fair bit of literature precedence. Under some pretty standard conditions all but my site of interest are converted to acetyl groups.
Running the reaction myself, things were almost too good to be true. Provided I kept the temperature below 40 ℃ or so I had perfect selectivity, and the reaction could be run for days without peracetylation. The selectivity was puzzling, given that there were four secondary alcohols. Steric hindrance was unlikely to play a role, given the flexibility of tobramycin, and the electronics of each site were broadly similar.
Still, the reaction worked well. I searched the literature and built a model, but eventually had to move on without an explanation. The model sat by my computer for a few months collecting dust, until one day I started rotating bonds at random. One conformation caught my eye.
Barring the protected amines tobramycin has two strong hydrogen bond acceptors; the oxygens in the carbohydrate rings. The 5-OH is the sole hydroxyl that is able to form an effective hydrogen bond with these oxygens, both bond forming a seven atom pseudoring (the strongest type of intramolecular hydrogen bonds). The standard 2D structure doesn’t show this conformation well, though it’s visible if you twist things around a little. These bonds prevented the 5-OH from interacting with acetic anhydride, while raising the temperature above 40 ℃ opened the structure up and led to peracetylation.
 Just try and violate Bredt’s rule with a model kit. I dare you.
 Three copies of it. While you can also buy only the pieces you want, the price is about double the cost of their premade set.