I did my PhD on total synthesis of a natural product, but via a particular route. We were interested in using carbohydrates as “chiral pool” starting materials, and for polyketides (which are decorated with lots of chiral hydroxy centers) that can be a good strategy. Getting your strings of chiral centers out of a vial can sound a lot more appealing than painstakingly setting them through synthetic chemistry, but in the end that’s a false dichotomy. For one thing, you’ll often have to flip or delete some of the hydroxy groups in your starting sugar molecules on their way to natural product synthons, because in the end, antibiotics and other polyketides aren’t really made out of strung-together sugar molecule pieces, either. And no matter how you get your chiral centers, you’re going to be doing a lot of protecting-group chemistry along the way.
For folks who haven’t had the total-synthesis baptism, a “protecting group” is a modification onto some reactive part of a molecule (a hydroxy group, an amine, a carboxylic acid, etc.) that can be stuck on to such a group to keep it from reacting as you go on with other synthetic steps. Sometimes you want to specifically work on a different hydroxy and need similar ones blocked off while that happens, and sometimes you need them all battened down because they’ll inactivate some reagent that you need to form carbon-carbon bonds or what have you. The keys to a good protecting group is for it to go on easily, put up with a wide range of other chemical conditions once it’s attached, and then leave gracefully under particular conditions when you want to cleave it off. That’s not always easy to realize! Oxygen-silicon ether bonds are classic protecting groups for hydroxys, because they can be pretty solid until they’re exposed to fluoride conditions. Benzyl carbamates (Cbz groups) can be handy for basic nitrogens, since they’re also pretty robust but fall off under catalytic hydrogenation. There are dozens, hundreds of these things in the chemical literature – I remember buying the first edition of this book back in 1984, and it has become a lot larger since then! Working on the synthesis of a really complex natural product has a way of turning into an annoying protecting group puzzle: figuring out which parts of the molecule need to have this sort of chemical masking tape on them and for how long, what the different ones need to be able to survive, and how you’ll remove tham and when.
Carbohydrate-based synthesis tosses you directing into the protecting group world, and it can be a tricky business. There is a rich literature on manipulating the more common sugars, with common protecting group combinations having their own shorthand names (“benzylidene glucose”, “diacetone glucose” and so on). Much like steroid chemistry, it’s a weird corner of organic synthesis with a lot of odd tricks and transformations. Every hydroxy group on those sugar-derivative starting materials has its own personality, mostly surly, and you have to get used to what you can get away with and what you can’t. There are some complex transformations that are startlingly easy to do on the right carbohydrate framework, and there other things that look perfectly trivial on paper but get grins and raised eyebrows from experienced carbohydrate chemists, because they know that some of them are practically impossible. Sugar rings are crammed with stereoelectronic chutes and ladders: all those oxygen molecules and their lone pairs of electrons sticking out in every direction from these often-rigid rings can assist you greatly or they can totally shut your ideas down.
But if you’re looking for an illustration of the weirdness of sugar chemistry, all you have to do is look at one of the most fundamental reactions, glycosylation. The “carbon 1” of a sugar ring is a masked carbonyl, part of a cyclic acetal or ketal, and that can leave you with a free OH group sticking out there, or an OR group if you come in with another hydroxy-bearing compound. With methanol you get a methyl glucoside, for example, and that O-methyl (like the original hydroxy) can stick out of the ring either “up” or “down”, in an equatorial or axial position. Glycosides like this are profoundly important in carbohydrate chemistry, not least in vivo, and in fact polysaccharide biomolecules (chains of sugars) are generally put together by such glycoside formation between the C1 of one sugar and a hydroxyl from another one, orver and over. But of course sugars have a whole selection of hydroxyls to choose from, and by linking these together into different sorts of glycoside chains you can get completely different sorts of polysaccharide molecules. Meanwhile, proteins are generally decorated with specific sugar molecules, glycosylated at various surface amino acid residues, and these can have big effects on a given protein’s activity, stability, and localization in the cell. Natural product molecules very often come with glycosides hanging onto their core structures as well, and are often inactive in vivo without them.
So forming glycosides under controlled conditions is of great importance, but after over a hundred years of effort it is still an unsolved problem. I last wrote about this here in 2018. For particular sugar derivatives there are usually good conditions known to form what you need, but if you change substituents on the sugar ring (by, for example, trying out a new protecting group!) all bets are off. And controlling glycosylation as you start stitching whole sugars together is the same beastly problem it’s always been. There are a lot of important cell-surface immune system molecules that depend on particular polysaccharide chains, but many of these things are still more or less impossible to make on demand. There are machines that will crank out oligonucleotide chains for you, and there are machines that will crank out proteins. There is no machine that will crank out polysaccharides to order. For that, you need to hire a lot of people, brew a lot of coffee, spend a lot of time and money, and probably stage a lot of sacrificial incantations under the light of the full moon. A lot of full moons – you might be at this for some months. The Koenigs-Knorr reaction is one place to start, but there are a lot of ways to run that one, and a lot of alternatives that have been devised under various levels of desperation.
So this recent paper certainly got my attention, as a old carbohydrate hand, with its title of “Site-selective, stereocontrolled glycosylation of minimally protected sugars“. Because frankly, for the most part, that’s just been impossible. This is achieved with some complex arylpyrrolidine molecules held together by thiourea spacers. These are taking advantage of a strategy that tries to bypass all that oxygen-ridden stereoelectonic complexing I was rambling on about above. Instead, this work (from the Jacobsen group at Harvard) takes a cue from the sorts of interactions that are often seen in carboydrate-binding protein sites, things like pi-electrons interacting with C-H bonds. That’s not the first thing you’d think of exploiting, but it really is orthogonal to wrestling with the oxygen atoms. The latter strategy has furnished some really ingenious chemistry, but all of those methods have intrinsic limitations – if you happen to have cis hydroxys and this particular sort of glycoside that you need, then you’re in luck, maybe, but otherwise not so much. Going after the sugar framework itself, the carbon-hydrogen stuff, has the promise of more generality.
The paper has a long list of examples, all of which use a fully protected (often methylated) glycosyl donor functionalized with a diphenylphosphate, chemistry that the Jacobsen groups has been working on for some years now. That earlier work has shown some very selective glycosylations with more simple alcohols, but this paper extends it to sugar-sugar couplings. The “accepting” carboyhydate partners here are indeed minimally protected, sometimes with just their (more reactive) primary alcohols blocked off by silyl ethers, and there are some really useful selectivities not only for the alpha/beta glycoside isomers that are formed but for where they reaction on the second carbohydrate rings themselves.
The chemistry in this paper is of course not the final answer to the glycosylation question, but it may point towards some of them. The whole idea of using such molecules with electron-donating aryl groups arranged on them to hold the various sugars in place is a big topic, but it’s clearly showing promise. What you’re looking at is a sort of artificial enzyme, mimicking the tyrosine residues on the natural binding sites to prearrange the reacting partners while at the same time activating the diphenylphosphate leaving group. That’s exactly the kind of work that enzymes do, and this paper (and its precursors) are worthy additions to the long project of trying to get human hands to mimic what a billion years of evolution have come up with.
