Cysteine racemase: an impossible enzyme?

Cysteine racemase is an enzyme (EC 5.1.1.10) that was characterised in lysates long ago, but have never been found since. Is it a real enzyme? Can the reaction happen well? The problem is that racemising via a carbanion intermediate something with a leaving group is not an easy feat.




The reaction is PLP-driven. The Schiff base is transaldiminated to cysteine from a lysine residue, one residue deprotonated the α-carbon and the PLP stabilises the carbanion long enough for it to be reprotonated at the opposite face. There is a catch: thiol is a good leaving group and the PLP may be outcompeted for the electron pair by the thiol. A lot of PLP-driven β-elemination reactions are used in biochemistry: cysteine synthase, cystathionine β-lyase and cysteine lyase*. The alanine racemase inhibitor β-chloroalanine works thanks to the fact that the enzyme tries to racemise it, but instead quickly eliminates it and the PLP π-stacking tyrosine (eventually) gets alkylated (=electrophilic aromatic substitution). Thiols are weaker leaving group than halogens, but are still really good.
One way around the thiol leaving is to have it as a thiolate, which is doable in light of the pKa of 8 of cysteine. The active site has the phosphate of the PLP close by to there and carboxyl group of the amino acid needs also a positive charge, but aspartate transaminase manages.
So the question is whether a cysteine racemase with a >99% product yield (product over substrate) is possible.

The enzyme might exist and might have a role. D-cysteine has inhibitory properties, but D-cysteine desulfhydrase (=lyase) eliminate it. It is not an overly useful compound, but it may be used to regenerate luciferin: luciferin is luminously decarboxylated/oxidised by luciferase into oxyluciferin and the latter is regenerated back to luciferin by luciferin regenerating enzyme (LRE), which may use D-cysteine (ref). Apart from this, it might be produced solely as a toxin, so not overly useful and could account why it is not reported.

A similar enzyme is serine racemase (EC 2.6.1.3). The hydroxyl on serine is a weaker leaving group than a thiol —but it cannot be deprotonated as easily. There are two types of serine racemase, the eukaryotic and archaeal one, srr encoded, and a bacterial one, vanT encoded. The former evolved from a serine lyase and still performs an elimination in addition to a racemisation: in the human one the fate of a serine is half-half elimination and racemisation, in Pyrobaculum islandicum elimination is a lot higher and may possess both activities physiologically. In the bacterial one, vanT, the elimination has not been reported: it has been assayed in several papers, but none explicitly mention serine elimination either because there was none or because they did not run a control without one of the coupled enzymes (DAAO).

Broad spectrum racemases, such as the one encoded by bsrV in Vibrio, also can racemise cysteine and this is the strongest proof that the enzyme activity is actually possible. Irksomely, cysteine is an inhibitor of peroxidase for the coupled assays so on a paper that looks at BsrV only used MS: curiously the peaks are really low, which could mean that an overwhelming amount of cysteine elimination occurred. Nevertheless, this indicates that the enzyme is possible, but not what the product yield is.

A possibly harder reaction is cysteine transaminase. The dehydrocysteine–quinoid PLP carbanion would have to not eliminate for long enough for the secondary ketimine to be substituted by water (hydrolysis). This reaction is known. Although if it eliminates cysteine or not as a side reaction is not clear (cell lysates).

This cysteine racemisation is mentioned here because enzymes with product yield less than 100% are mighty uncommon as discussed in a past post. Some secondary metabolite enzymes, such as geranylgeranyl pyrophosphate cyclases produce a broad spectrum of products, because they are in a gray zone between promiscuous and physiological —they are under selection to be plastic, but the products are marginally under selection. But really few are fully physiological and not high yielding. Cysteine synthase makes cysteine, but can also break it, but there are order of difference between the two. Many other synthases have a negligible destructive activity: it stands to reason it is hard to eradicate. One of the few enzymes which makes lesser products is probably ThiC. This is an enzyme with a complex reaction and is explicitly mentioned in a paper to produce 50% of the time a dephosphorylated product (and maybe others). Most likely other enzymes with complex enzymes have less than perfect yields, but they have not been characterised. Cysteine racemase is different though: it is a simple reaction that has to fight another.

In biotin biosynthesis there is a rather telling step. KAPA synthase, encoded by bioF, catalyses a PLP-driven decarboxylative aldol condensation (or whatever an acyl-CoA attack to a ketimino is). The mechanism is the same as aminolevulinic acid synthase. The odd thing is that the methyl group of the alanine substrate is inert and has to be activated in a later step in a rather complex manner (2x rSAM with sacrificial FeS cluster). If cysteine had been used as the substrate it would have been a walk in a park. If elimination happened, in this reaction a 50% yield would be not okay as, ironically, a transaminase would be needed followed by the pathway nature uses currently...


*) Not to be confused with cysteine desulfurylase, which is an enzyme that transfers the thiol to a cysteine residue oddly without playing the PLP card.