hxxp://pubs.acs.org/doi/abs/10.1021/ja01171a022

Preparation of benzoic anhydride:

hxxp://www.orgsyn.org/orgsyn/orgsyn/prepContent.asp?prep=cv1p0091

Benzoic anhydride and alanine, refluxed in pyridine, should produce cathinone by the Dakin-West reaction upon hydrolysis of the generated amide. Addition of a catalytic quantity of DMAP allows this name reaction to occur at room temperature.

Has this been tried?

Thanks for the paper.

a-Benzamidopropiophenone. A mixture of 3 g. (0.034
mole) of DL-alanine, 12 g. (0.15 mole) of pyridine and 34 g.
(0.15 mole) of benzoic anhydride was heated with stirring
at 130-135C for two and one-half hours, and the reaction
product worked up as previously described. The solid
obtained was recrystallized once from ligroin and twice
from aqueous ethanol to give 3.6 g. (42%) of a-benzamidopropiophenone,
colorless prisms, m. p. 104-105" (cor.).
a-Benzamidopropiophenone has been reported to have a
m. p. of 103C and 104-105°C

Guess it works! Just have to hydrolyze that amide and you've got cathinone.

Chuck this product into a yeast reduction and you'd probably get some racemic ephedrine. Hit it with Zn/Hg amalgam + HCl (Clemmenson Reduction) or Pt/H2 and you should get amphetamine.

Perhaps, using tryptophan as the amino acid, refluxing with acetic anhydride and pyridine... and hydrolyzing the produced amide, and reducing the carbonyl with the Clemmenson Reduction... you could get alpha-ethyltryptamine.

It may be possible to produce these anhydrides with pyrosulfate (made by dehydrating NaHSO4 with heat - it's sold as pool pH down)  and the carboxylic acid salt (see jon's method)... then simply extracting a solution of the anhydride using toluene or DCM.

I would be very interested if anyone could find the yeast- or other organism- mediated reduction of an amide/amine/etc. (or more probably an imino/imido - ie. double-bond between the second carbon and the nitrogen) propiophenone derivative to the relevant chiral ephedrine.*

* There are of course two chiral centers in ephedrine and the yeast reduction to the chiral phenyl-1-propanol (ie. the position of the two available bonds on the first carbon in the chain, OH and H (the others being C and C), is also apparently capable of being used to directly influence the stereochemistry of the second carbon on the chain IF there is a double-bond available - so if there is a keto/azido/etc group - on that second carbon. IF it were possible to throw an imino-ketone into the mixture and reduce that (losses from hydrolysis would be a bitch), possibly in a non-fermenting environment (ie. not aqueous), THEN it might be possible and grab a shitload of interest real fucking quick.