B.R.S.M. Pain is temporary, publications are eternal


The Splenda Effect and Chlorosulfolipids

Acknowledgement: I took the term 'Splenda Effect' from Carreira's talk at the Bristol Synthesis Meeting on Tuesday. It's a useful term to describe the low reactivity of alkyl halides near electron withdrawing groups. This post is mostly sourced from that, wikipedia and the Nature paper cited below.

Anyone eat any primary alkyl chlorides today? My dad had a few spoons in his morning coffee but that's not unusual. I only found out at the start of the week that sucralose, the main ingredient in Splenda and other sweetners, actually contains not one BUT TWO primary alkyl chlorides.[1] That makes it rather indigestable and also about 600 times sweeter than the starting material. In fact, as it contains fewer than 5 calories per serving the FDA allow it to be sold as 'zero calorie'. Because anything below five is basically zero.[2]

But why bring that up? Well, in considering synthetic approaches to the chlorosulfolipids Carreira asked the obvious question, 'can't we just start from a polyol and do a bunch of halogenations?'

"The result of initial synthetic efforts involving model systems led us to conclude that construction of such systems would have to take into account the unique behaviour and properties of a polychlorinated backbone with electron-withdrawing groups. As a relevant benchmark, sucralose, the key ingredient of Splenda, incorporates two 1° and one 2° chlorides on a disaccharide core and is sufficiently stable and safe to be widely used as an artificial sweetener. In a similar fashion, in preliminary investigations we observed that displacement reactions of activated alcohol derivatives to furnish the corresponding alkyl chlorides proved unworkable when the carbinol bears two methine substituents with electron-withdrawing groups, such as chlorides. Additionally, we noted that α- and β-chlorinated aldehydes are fleeting intermediates, which undergo enolization, hydration or elimination too rapidly. Thus, we sought to implement strategic approaches that would circumvent these limitations in crafting a synthesis route to chlorosulpholipid 5." - Carreira in Nature, 2009, 457, 573

The route the group actually used was interesting, and worth a minute's consideration. It's from three years ago, but I didn't have a blog back then so I'll quickly cover it now.

The synthesis began with the chlorination and reduction of ethyl sorbate, followed by TBS protection of the new alcohol. Next the group needed to perform an epoxidation of the protected allyl alcohol, which was done using a two step procedure. First, dihydroxylation with osmium tetroxide gave a seperable mixture of diastereomers, fortunately 5.6:1 in favour of the one required. Then, treatment of the diol with triflic anhydride in the presence of excess DABCO lead to selective activation of the less hindered alcohol, followed by epoxide formation. Removal of the TBS protecting group, oxidation of the primary alcohol and Wittig reaction with phosphonium salt A gave a vinyl epoxide ready for the introduction of three more chlorine atoms. Treatment with TMSCl introduced the first, giving a major product along with traces of another two as well as a large amount of recovered starting material. As it wasn't obvious which diastereomer had been formed, and the synthesis was nearly finished, the group pressed on, planning to assign the configuration of the newly formed stereogenic centre by comparison with the natural product.

Unfortunately, it turned out that the data did not match and the group initially suspected that the structure had been misassigned. However, upon closer inspection it seem that the epoxide opening step appeared to have somehow occurred with retention of stereochemistry!

Carreira rationalised the outcome by invoking anchiomeric assistance from a nearby chlorine atom, an effect with some literature precedent, having been previously observed in the solvolysis of alkyl tosylates. Of the two possible chloronium transition state I think the five membered one looks less silly, but a four membered one can also be drawn. In the end, simply changing the epoxide stereochemistry allowed the group to obtain the correct chloride to complete the target.

Great work! The group has since produced a library of different diastereomeric polychloroalkanes to allow assignment of  the stereochemistry of different motifs by comparison of NMR data and to prevent confusion over stereochemistry as seen here. The 'Splenda effect' also has important consequences for the mode of action of these marine toxins, as it means that they might not just be simple alkylating agents as was once believed.


If you want to learn more, the group has just published a review on the subject in EJOC (Eur. J. Org. Chem. 2012, 1685).


Tangental Information

1. This, perhaps, is why European PhDs are held in such low regard by American academics.

2.  "But how? How do you swap just one secondary alcohol?". Like this:The are many patented routes, and I'm not sure which is actually used. Plus I hate reading patents. The key step appears to the migration of just one acetate group when the pentaacetate is warmed in the presence of pyridine.

The story of the development as a sweetner is quite amusing. According to wikipedia:

"Sucralose was discovered in 1976 by scientists from Tate & Lyle, working with researchers Leslie Hough and Shashikant Phadnis at Queen Elizabeth College (now part of King's College London). While researching ways to use sucrose as a chemical intermediate in non-traditional areas, Phadnis was told to test a chlorinated sugar compound. Phadnis thought that Hough asked him to taste it, so he did. He found the compound to be exceptionally sweet."

According to Carreira, Hough was Scottish, which might go someway to explain the confusion.

Comments (5) Trackbacks (0)
  1. Wow, first the snarky “Footnote #1,” and then the crack at Scottish chemists…
    Banner day for UK science on B.R.S.M. ; )

  2. There seems to be an extra methylene in phosphonium salt A.
    Does the configuration at the epoxide always invert during the Swern oxidation (and why wouldn’t the inversion open the epoxide)?
    Wouldn’t the chlorination via cyclic chloronium occur by a double inversion? It’s drawn as a double retention (both times).
    Wouldn’t it have made more sense to call sucralose suchlorose?
    Has anyone studied the fluorine analog of sucralose?

    • There’s no inversion during the Swern – that was a copy/paste error on my part. I also made a mistake with *both* cyclic chloronium structures as intramolecular epoxide opening, being SN2, of course occurs with inversion. Retention (as seen overall here) is only ever (as far as I know) the result of double inversion. Regarding suchlorose… maybe. Disadvantages I can see are 1. ambiguous pronounciation – non-chemists will probably call it such-lorose 2. chlorine is scary.

      Thinking about it, point 1. reminds me of something I read somewhere about how a person pronounces the word unionised being a good test for whether or not they’ve had any chemical training. Think it might be something Asimov once said.

    • Also, a quick reaxys search reveals sufluorose to be unknown in the literature

  3. Great post. Much more comprehensive than the one I wrote in Russian 🙂

    However, “the low reactivity of alkyl halides near electron withdrawing groups” is something I still cannot accept. Alpha-haloketones are unbelievably reactive (I have an evidence when secondary chloride next to a ketone reacts faster than primary one! Under typical SN2 conditions (you know, DMSO or DMF etc).

Leave a comment

No trackbacks yet.