Fagopyrins are phototoxic compounds that cause fagopyrism. They are chemical compounds related to hypericin and pseudohypericin. Some variations lack one or both methyl groups, which I think is to streamline the molecule so it better fits in membranes and so increase damage. The main difference from (pseudo)hypericin though is the presence of pyrrolidine or piperidine (or both) moieties, which may come from L-proline and L-piperidine-1-carboxylate (in turn from L-lysine) respectively. So, what is the function of those moieties in the fagopyrin molecules? Does it render fagopyrins more toxic than (pseudo)hypericin?


Radical generator

Looking at the molecules they are highly conjugated —must be dark red or similar! And the disease is photosensitivity. Aromatics + UV? That is a standard UV radical reaction. It may be enough as a catalyst, but you'll get more will a cyclic amine, such as pyrrolidine or piperidine. If you want to go technical confer: Hofmann–Löffler reaction.

The radical will be passed on to other molecules, such as oxygen (generating peroxide), which will damage DNA and will raise an alarm biologically ultimately attracting neutrophils, which are expecting some attack (not poisoning), thus causing local inflammation, which manifests as skin irritation.


With the caveat that secondary metabolism is highly promiscuous and that some modifications are more accidental that anything else, let's look at the methylation and membrane permeability.

Promiscuity and secondary metabolism evolvability

This figure I found has the methylation patterns. A first glance I have assumed it was the usual C-methyltransferase action adding methyl groups, but instead it is a removal twice in the case of fagopyrin B and once in C and D. Specifically, fagopyrin is clearly two condensed anthraquinones molecules (emodin?) —which is a rather interesting non-standard change, I have not looked but so hope that someone has found the enzyme responsible! As you point out an anthraquinone is made by polyketide synthesis: the last acetyl-CoA results in a methyl group on the aromatic structure, which is occassionally oxidised to a carboxyl group (e.g. rhein) and in this case decarboxylated —I do not think a C-methyltransferase can remove groups even with pushed stoichiometries nor do I believe I have heard of one. But anyway. The anthraquinone condensing enzyme may not be able to tell apart the standard and the demethylated metabolites, which is otherwise meant for another pathway, i.e. it could just be promiscuous.


Let's calculate the effect on solubility (python3 w/ RDKit).

Wikipedia has the SMILES string for the main fagopyrin and it's easy to spot the methyl groups.

from rdkit import Chem
from rdkit.Chem import AllChem
from rdkit.Chem.Draw import SimilarityMaps
# let's make the molecules
normal = Chem.MolFromSmiles('O=C1C2=C(C3=C(O)C(C4CCCCN4)=C2O)C(C5=C6C(C)=CC(O)=C15)=C7C(C3=C(O)C(C8NCCCC8)=C9O)=C9C(C%10=C7C6=C(C)C=C%10O)=O')
demeth = Chem.MolFromSmiles('O=C1C2=C(C3=C(O)C(C4CCCCN4)=C2O)C(C5=C6C=CC(O)=C15)=C7C(C3=C(O)C(C8NCCCC8)=C9O)=C9C(C%10=C7C6=CC=C%10O)=O')
# no need to orient them as the indices are the same. Whoohoo
img=Chem.Draw.MolsToGridImage([normal, demeth],
                              legends=['w/ methyls', 'w/o methyls'],


Now let's calculate logP (Partition coefficient). print('w/ methyls', Descriptors.MolLogP(normal)) print('w/o methyls', Descriptors.MolLogP(demeth))

  • w/ methyls 6.71
  • w/o methyls 6.0

They are really soluble in octanol and insoluble in water. The unmethylated version is less hydrophobic.

Let's look at what is causing the difference by looking at the Crippen contributions to logP. Using the BluePurple colormap (suggestions welcome for best colormap for this descriptor), so that purple is hydrophobic while light blue is hydrophilic. Note that this is not partial charge (Gasteiger).

for mol, name in ((normal, 'meth'),(demeth, 'demeth')):
    contribs = Chem.rdMolDescriptors._CalcCrippenContribs(mol)
    fig = SimilarityMaps.GetSimilarityMapFromWeights(mol, [x for x,y in contribs], colorMap='BuPu', contourLines=10)
    fig.savefig('/Users/matteoferla/Desktop/'+name+'_crippen.png', bbox_inches='tight')

normal demeth

The difference is a result of the methyl groups, being slightly more electron withdrawing than hydrogens making the carbon between it and the hydroxyl/ketone more hydrophobic. But overall does not change the polar sides, so is in effect unlikely to have an effect on its high membrane permeability.

| improve this answer | |
  • $\begingroup$ The methylated fagopyrins are in fact the first fagopyrins, from which the unmethylated fagopyrins derive. The methyl groups are in the aromatic structure and in fact derive all the way from the first polyketide synthetized in the pathway (which leads in St John's wort to hypericin and pseudohypericin, which also have methyl groups (two in hypericin and pseudohypericin having one, the other being hydroxymethyl)). My conjecture regarding membranes is from extrapolation from how cholesterol fits better in membranes than its precursor lanosterol, which may be faulty reasoning. $\endgroup$ – JMRD Dec 14 '19 at 21:20
  • $\begingroup$ I had mistakenly though they were extra added in addition. My bad. I have expanded that part of the answer —although it is a bit off topic now. The radical generator paragraph is satisfactory, though? $\endgroup$ – Matteo Ferla Dec 15 '19 at 12:32
  • $\begingroup$ Yes, it's sastifactory enough. $\endgroup$ – JMRD Dec 15 '19 at 14:23
  • $\begingroup$ Great. Don't forget to mark the question as solved then. $\endgroup$ – Matteo Ferla Dec 16 '19 at 9:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.