Potential for the Crosslinking and Aggregation of Proteins through Lysyl Adducts of Pyridoxal: Relevance to B6 Neurotoxicity

That diagram above (it also applies to pyridoxal and pyridoxamine) will help with the references to the sites of the reactions, as discussed in the posting.

I'm not finished with this yet, but the thought I'm trying to convey is that free pyridoxal, pyridoxal 5'-phosphate, or pyridoxamine could, potentially, mediate the crosslinking of proteins by their extremely reactive properties, and these crosslinks, initiated by the nucleophilic attack of nucleophilic amino acid residues [lysine is the most nucleophilic, in general, and cysteine and histidine residues can also act as nucleophiles in the formation of protein adducts of small-molecule drugs or other compounds, such as pyridoxal (the aldehyde form of vitamin B6), etc.] on the carbonyl carbon of pyridoxal or by the nucleophilic attack of the primary amine moiety of pyridoxamine on electrophilic carbonyl-containing compounds. Pyridoxal can basically form Schiff base adducts and Amadori adducts with any lysine residue (especially one that has a sufficiently-low pKa, thereby increasing the likelihood that the epsilon (primary) amine of lysine will be deprotonated within the physiological pH range) (http://scholar.google.com/scholar?hl=en&q=%22low+pKa%22+nucleophilic+lysine+residue&btnG=Search&as_sdt=100000000&as_ylo=&as_vis=1), and some of the general concepts (and some specifics) about the reactivity of pyridoxine-derived compounds are discussed in this article [Wondrak, 2008: (http://www.ncbi.nlm.nih.gov/pubmed/18473939)]. In my view, however, the pyridine ring of pyridoxamine is also so reactive as to potentially complicate the role of pyridoxamine as a "carbonyl scavenger." Although there's lots of research on the potential therapeutic effects of pyridoxamine in the prevention of advanced glycation end (AGE) product formation [I've shown the formation of an N-glutathionyl conjugate of the acyclic, aldehyde form of beta-D-glucose, followed by the formation of a protein AGE (via the nucleophilic attack by the thiol group of the cysteine moiety of the glutathionyl conjugate of glucose on one of the sulfur atoms of a protein disulfide linkage; I didn't show the rest of it, but the idea comes through--that particular conjugate has been shown to form, and I'll cite the article later], the capacity of pyridoxamine to form substituted pyrrole derivatives [I've shown a mechanism by which the pyridoxamine conjugate of levuglandin E2 (a 1,4-dicarbonyl derivative of PGE2 that is formed in vivo and discussed in the article by Wondrak (2008))] could potentially be problematic, in my view.

A major reason I'm thinking that the formations of these types of conjugates of B6-derived compounds could contribute to B6-induced neuropathy/neurotoxicity is that a major underlying mechanism of small-molecule-mediated neurotoxicity, such as can be induced by exposure to compounds such as acrylamide, 1,4-dimethylketones (i.e. 2,5-hexanedione), and carbon disulfide, is the formation of protein lysyl adducts and, subsequently, protein crosslinks that disturb axonal transport [Zhu et al., 1994: (http://www.ncbi.nlm.nih.gov/pubmed/7981420); LoPachin and Decaprio, 2005: (http://www.ncbi.nlm.nih.gov/pubmed/15901921)]. There's at least one article showing that pyridoxal phosphate induced the aggregation of proteins in the lenses (of the eyes) by forming lysyl adducts on the proteins, and glucose partially inhibited the pyridoxal-mediated increases in protein aggregation [Ganea and Harding, 1996: (http://www.ncbi.nlm.nih.gov/pubmed/8727969)]. The authors cautioned against the use of pyridoxal as an inhibitor of glycation, etc. It's important to note that pyridoxamine acts as a nucleophile, and pyridoxal acts as an electrophile. But I think some pyridoxamine can be formed from pyridoxal in vivo, even though most of the conversion is of pyridoxamine to pyridoxal. Anyway, another major point is that a big mechanism underlying the neurotoxicity of 1,4-dicarbonyl compounds (i.e. 2,5-hexanedione) is the formation of substituted pyrrole compounds, after the initial, nucleophilic attack by lysine, by a transimination reaction (I've shown this, in the diagrams, in the formation of a cyclic pyridoxal adduct of spermidine, formed via a transimination reaction, and in the formation of the pyrrole derivative of pyridoxamine), followed by the oxidation of pyrrole by triplet dioxygen and, along with the presence of a protein-bound pyrrole moiety of heme, for example, the formation of a bipyrrole crosslink.

Anyway, I stopped taking supplemental vitamin B6, but one should talk to one's doctor about these things. I wonder if the reason glutamate has been shown to ameliorate B6 neuropathy might be that B6 forms adducts with glutamate and other amino acids (glutamate in particular), such as glycine. If B6 caused glutamate depletion or something, maybe that could account for the effect of exogenous glutamate. And maybe the inverse relationship between liver glycine levels and liver PLP levels is due to the formation of glycine adducts of pyridoxal. I don't have time to cite those articles, but another possibility is that the inactivation of proteins (I've shown a possible mechanism by which the PLP-induced inactivation of succinate dehydrogenase, via the formation of lysyl adducts (this has been shown to occur, as cited below), could occur) could account for the strange effects of B6 on alkaline phosphatase expression, and I've talked about the way those effects don't really have any metabolic explanation. Anyway, it's interesting to think about. Also, the Schiff base formation and Amadori rearrangement mechanisms commonly occur when the carbonyl carbons of carboxylate groups on glucuronide conjugates of drugs or drug metabolites undergo nucleophilic attack by lysine (or other) residues to form drug-protein conjugates ("acyl glucuronide" mediated idiosyncratic adverse drug reactions, etc., with immunological sequelae).

I wonder if deferoxamine (DFO) might exert some of its therapeutic effects in diabetic neuropathy (or in the prevention of transglutaminase-mediated crosslinking of proteins) by serving as a carbonyl scavenger and not just by indirectly, via its iron chelation, limiting reactive-oxygen-species-mediated carbonyl stress (http://scholar.google.com/scholar?hl=en&q=carbonyl+cross-link+deferoxamine+OR+desferrioxamine&btnG=Search&as_sdt=100000000&as_ylo=&as_vis=0). Lysine is used to form the primary, terminal amine group of desferrioxamine B (a.k.a. deferoxamine), and lysine is the most nucleophilic of the amino acids. If DFO serves as a carbonyl scavenger, the N-(carbonyl-carbon-of-reactive-carbonyl-compounds)-DFO adducts might be substantially less likely than something like pyridoxamine to undergo "post-Amadori" oxidation and to form protein adducts. Also, the lysosomal localization of DFO might mean that the carbonyl scavenging by deferoxamine would be compartment-specific or whatever--"site-restricted" (such as to mitochondria, during lysosomal fusion events). Anyway, it's interesting to think about.

This shows the formation of a protein lysyl adduct of acrylamide:


This shows the formation of a glutathionyl adduct of glucose, following the hydrolysis of the cyclic lactone form of glucose that predominates in vivo, the Schiff-base-formation mechanism that applies to all or most of the reactions involving nucleophilic attack on the carbonyl carbon of pyridoxal (the catalytic mechanism of many or all PLP-dependent enzymes involves a transimination reaction, and protein-bound PLP exists as a Schiff base conjugate), the Amadori rearrangement that produces a stable, Amadori product (I guess this is an AGE also), and the nucleophilic attack by the thiol group of the glucose-GSH adduct on a disulfide linkage (the first step in the formation of a protein (a protein AGE) adduct of the glucose-GSH adduct):


Nucleophilic attack by primary amine group of glutathione (GSH) on the carbonyl carbon of the aldehyde group of acyclic D-glucose:

Formation of unstable Schiff base:

Amadori rearrangement of Schiff base:

Formation of Amadori product:

Attack by thiol group of glutathionyl-glucose adduct on disulfide linkage, to form protein adduct of glutathionyl-glucose (N-deoxyglucosyl-glutathione, I think, is the type of nomenclature used in the article):

This shows the formation of a substituted pyrrole-containing compound, via the nucleophilic attack of the primary amine group of pyridoxamine on levuglandin E2, I think it's called:


I think that the dehydration of this "pyrroline" (?) compound, to form the pyrrole ring, may involve carbocation intermediates, but I didn't show that:


This is the slow step by which triplet dioxygen forms a charge-transfer complex, in which an electronic transition from a HOMO of pyrrole to the unoccupied 4sigma(g) molecular orbital of dioxygen (the LUMO of dioxygen) occurs, followed by the transfer of that electron into one of the pi*x or pi*y antibonding molecular orbitals (these are each singly-occupied in triplet dioxygen, as shown in the diradical species I've drawn in the diagram) to form the superoxide radical anion, in a doublet state. There is actually an ion pair in the charge-transfer complex, with the anion being delocalized across dioxygen and the pyrrole nitrogen existing as a formal or actual cation, and the existence of this type of ion pair, specifically in the oxidation of pyrrole compounds, has been shown experimentally:





Formation of cyclic, pyridoxal adduct of spermidine via a transimination reaction of the Schiff-base "intermediate" (a cyclization reaction that precludes the formation of a more stable, Amadori product from the Schiff base):




This shows a nucleophilic attack by glutathione on the carbonyl carbon of pyridoxal:

This shows a possible mechanistic explanation for the apparent noncompetitive, mechanism-based inactivation of succinate dehydrogenase (SDH) by pyridoxal (actually, pyridoxal 5'-phosphate (PLP)), followed by the dissociation of succinate dehydrogenase from complex II, in the mitochondrial matrix [Choudhry et al., 1985: (http://www.ncbi.nlm.nih.gov/pubmed/3972121)]. This PLP-induced inhibition of SDH has been shown to occur and to, either definitely or almost definitely, depend on the formation of one or more protein lysyl adducts of PLP within or near to the ubiquinone binding site (Choudhry et al., 1985, p. 172). CoQ2 (this is a ubiquinone analog that is just, I think, ubiquinone with 2 isoprene units instead of 10) could block the PLP-induced inactivation, and the authors found that the formation of protein adducts of PLP did not occur at the active site of the enzyme (the dicarboxylate binding site). Apparently lysine residues are required for the association of SDH with the rest of the complex II multienzyme complex, and the PLP-induced formation of Schiff-base or Amadori lysyl adducts apparently prevented the normal electrostatic interactions that are required to maintain the binding of SDH to complex II (shown in my sketches as a single bond, for the sake of brevity/simplicity). So it sounds like PLP could form adducts at the ubiquinone binding site and also at the lysine-rich "domains" or portions of SDH that mediate its association with complex II (or the ubiquinone binding site is near to or a part of those complex II-binding residues). PLP and pyridoxal and pyridoxine (and pyridoxamine, presumably, such as after it forms adducts) are known to react with singlet oxygen at C6 of the pyridine ring [Ohta and Foote, 2002: (http://www.ncbi.nlm.nih.gov/pubmed/12371824)], and I've shown a slightly-abbreviated "mechanism" for the so-called "-ene" reaction of singlet oxygen with pyridoxal to form a C6 hydroperoxide derivative of pyridoxal. The mechanism of the -ene reaction is not the same as the singlet oxygen cycloaddition mechanism, and the -ene reaction may involve a perepoxide transition state or a so-called exciplex transition state. I think that the electron transfer is from a highest-occupied molecular orbital (HOMO) of pyridoxal to the lowest-unoccupied molecular orbital (LUMO) of singlet dioxygen (singlet delta dioxygen), and this LUMO is the 4sigma(g) orbital (the 3pz sigma antibonding molecular orbital) of singlet dioxygen. Then there's one or more transitions of electron(s) from the LUMO(s) of the transition-state molecular orbitals of the pyridoxal hydroperoxide adduct to a HOMO of the pyridoxal hydroperoxide adduct, etc., as far as I know. I meant to show a perepoxide transition state, but that particular chemistry drawing program didn't easily allow me to show what was going on by drawing a singly-bonded dioxygen species (that would potentially be confused with triplet dioxygen, which is often drawn that way) and by drawing the zwitterionic perepoxide in the transition state). So I show something closer to an "exciplex," I guess. The hydroperoxide then forms a cyclic endoperoxide that undergoes dehydration, yielding a quinone-like coenzyme Q10 analog. The mechanism shown by Ohta and Foote (2002), in contrast, showed the formation of the final quinone-like compound via the nucleophilic attack of methanol on the quaternary C2 of the pyridine ring of pyridoxal. Incidentally, I don't think coenzyme Q10 supplementation is likely to be a good idea, in general (and I say this emphatically), but one would obviously want to discuss this type of issue with one's doctor.












This shows 1) the formation of a monoalky dithiocarbamate protein lysyl adduct (Protein-Lys-NH-CS2, a lysyl dithiocarbamate adduct, as an example of an N-alkyl dithiocarbamate), following the nucleophilic attack by a lysine residue on carbon disulfide, 2) the oxidation of two lysyl dithiocarbamate residues to form a dialkyl dithiocarbamate disulfide linkage between two proteins or two lysyl residues of the same protein or polypeptide, and 3) a redox disulfide cleavage reaction in which the N-alkyl dithiocarbamate moieties of the oxidation of the dialkyl dithiocarbamate disulfide linkage are oxidized, indirectly, via the oxidation of the disulfide sulfur atoms by singlet delta dioxygen. So singlet delta dioxygen oxidizes the sulfur atoms of the disulfide moiety, and then the oxidized sulfur atoms serve as oxidants of the lysyl dithiocarbamate moieties to lysyl isothiocyanate residues and, in the process, undergo reduction in the "reductive cleavage" or "reduction" of the disulfide linkage, overall, I guess you could say. The cleavage of a disulfide linkage is usually discussed as being a "reduction" of the oxidized thiol (-SH) groups, and I'm trying to address that in terms of the reactions in question. Then the isothiocyanate residues can form more stable dialkyl di-isothiocyanate disulfide (I think that's the terminology) crosslinks, and I didn't get around to showing that part yet. The point is that these crosslinks begin with the nucleophilic attack by a lysine residue onto carbon disulfide and are mechanistically reminiscent of the attack by lysine on the carbonyl carbon of pyridoxal, thereby potentially forming pyridoxal-mediated crosslinks, either by transimination, prior to the formation of an Amadori product of the pyridoxal adduct of a lysine residue, or by the hydroxyl radical-mediated oxidation of the methylene carbon (C9) of the hydroxymethyl substituent of the pyridine ring of the protein-lysyl pyridoxyl adduct (pyridoxine has two hydroxymethyl substituents, but there's only one on an adduct of pyridoxal that's formed as a result of nucleophilic attack by lysine at C8) (followed by the formation of a protein crosslink at that allylic [quasi-benzylic) methylene carbon] or by the oxidation by singlet oxygen of the C6 (or C2, I think) carbon of the pyridine ring, followed by the formation of a quinone-like compound (shown above) or by the formation of a protein crosslink at the C9 site, etc. There are all sorts of other reactions that can occur with the lysyl dithiocarbamate, dilysyl dithiocarbamate, or lysyl isothiocyanate residues, such as the formation of hydrophobic thiophene derivatives, etc., that could become crosslinked to the pyrrole moieties of protein-bound heme species, etc.

Three Major Electronic States of Dioxygen (O2); Relevance to Differences in the Reactivities of Iron(IV)-Hemes

These diagrams show the three major electronic states of dioxygen (O2) [note: I can't change pictures, in the blogger software, after I upload them, without dismantling the whole posting, basically, but the final diagram should show either no single-electron-transfer arrow or a two-electron-transfer arrow on zwitterion II (showing transfer to zwitterion I), not a single-electron-transfer arrow]. I'll put up sketches of the molecular orbitals soon, here, but one of the main points of this article [see Yamaguchi et al., 2009: (http://linkinghub.elsevier.com/retrieve/pii/S0277538709000515)] and of many other articles is that triplet dioxygen exists in a robustly-diradical state, with regard to the 2pi*x and 2pi*y antibonding molecular orbitals that are singly-occupied by electrons of the same spin, and that, in all of the three major electronic states of dioxygen, the 2pz(sigma) (a.k.a. 2sigma(2pz) a.k.a. 3sigma(gerade)) bonding molecular orbital exhibits a less-robust, 2pz-2pz diradical character with "singlet coupling" between the two "partially unpaired" electrons [the 2pz atomic orbitals are oriented along the z-axis, which is the internuclear axis that intersects both oxygen nuclei, and form a "partial" 2sigma(2pz) (a.k.a. 3sigma(gerade)) sigma bond]. That's shown in one of the diagrams below. Also, the highest-energy singlet excited state (singlet sigma) exists as a pair of zwitterionic "resonance" forms (still within the same electronic state), and the first excited state (singlet delta) exists as four resonance forms (two zwitterions and two (2pi*x and 2pi*y) diradicals in which there is "sigma coupling" between the two orbitals that are singly-occupied by electrons of opposite spin). The other articles cited are these [Lyne et al., 1993: (http://pubs.acs.org/doi/abs/10.1021/ic00074a022); Bytautas and Ruedenberg, 2010: (http://link.aip.org/link/?JCPSA6/132/074109/1)]. The other point is that the FeO moieties of some of the electronic states of perferryl and ferryl heme (and other oxidation states of heme) exist as "singlet delta"-like diradicals or zwitterions, "singlet sigma"-like zwitterions alone, or triplet-O2-like diradicals [Yamaguchi et al., 2009; Filatov et al., 1999: (http://www.rsc.org/delivery/_ArticleLinking/DisplayArticleForFree.cfm?doi=a809385g&JournalCode=P2)(http://cat.inist.fr/?aModele=afficheN&cpsidt=10072829)].

Some of the Molecular Orbitals and Electronic States of Heme Species

I'm going to paste these diagrams for some of the major molecular orbitals of heme species and then for four "major" electronic states of perferryl heme species (in this case, the states are of compound I or another, very similar perferryl heme species that contains a proximal cysteinate ligand). In the doublet and quartet states that contain a sulfur ligand radical, there is still a pi-type bonding interaction (constructive overlap or "mixing") between a pi type lone-pair molecular orbital of cysteine. Iit's not clear to me if the 3pz or 3px nonbonding ("lone-pair") molecular orbital of the free, non-heme-bound cysteine residue forms a sigma-sigma type charge-transfer complex with the sigma*(z^2) antibonding molecular orbital of free heme, prior to its overlap with the a2u molecular orbital of heme, but the point is that some of the spin density of even the sulfur/thiolate ligand radical is in the porphyrin a2u molecular orbital (and not the a1u molecular orbital). It's not entirely clear to me what the nature of the "single" bond between the sulfur and the iron atoms is, but I've gotten a much more clear picture of the specific orbital occupations in the FeO (ferryl) moiety in different electronic states. I don't have time, at the moment, to put up diagrams that show the similarities between the different electronic states of the Fe=O moiety and the three major electronic states of dioxygen (O=O), but a lot of the articles have made the comparisons between various electronic states of ferryl and perferryl heme and the ground-state triplet, first excited singlet state ("delta singlet"), or second excited singlet state (sigma singlet) of dioxygen. In those different electronics states of dioxygen [the electronic state is more specific than the electronic configuration and specifies the relative energy levels of the molecular orbitals and the "spins" (the m(subscript)s quantum number) of the electrons in the various singly- and doubly-occupied molecular orbitals; within a given electronic configuration, any number of electronic states may exist (up to 25 or more, in some cases)] or of the Fe=O moiety, the key difference is that there is either a robust diradical (as in the states of heme species that are similar to ground-state triplet dioxygen, in which the spins of the electrons in the singly-occupied pi*(xz) and pi*(yz) molecular orbitals of O2 are both +1/2--this is the situation shown below, although other electronic states of heme species are more similar to triplet dioxygen, in terms of their orbital symmetries, than the electronic states shown below) or, I think, more of a partial diradical [in a manner analogous to the two major excited, singlet states of dioxygen, in which there is either a doubly-occupied pi*(xz) or pi*(yz) molecular orbital (singlet delta) or two singly-occupied pi*(xz) and pi*(yz) molecular orbitals containing electrons of opposite "spin" (sigma singlet)] character to the singly-occupied pi*(xz) and pi*(yz) antibonding molecular orbitals, such that the spin density is primarily localized to the Fe and the O, individually, and not shared to a significant extent. The triplet, singlet, and quartet terms in the diagrams are the spin multiplicities of the individual spin centers. The spin multiplicity is this: (the absolute value of 2S) + 1. S is the total spin for a given spin center and is this: [(# of alpha electrons ("positive spin")) x (+1/2)] + [(# of beta electrons ("negative spin")) x (-1/2)]. So S is (+1/2)(2) = 1 for the FeO moiety shown below, and 2(1)+1 = the spin multiplicity = 3 = triplet, for the FeO moiety. The FeO, porphyrin, and thiolate [as well as the iron atom itself, which can exist in more than one spin state (or, as it is sometimes referred to as being, configuration, such as d(superscript)5, d4, or d6 (or 5D or 5S or whatever the other terminology is--I'll put up diagrams on those aspects), within a given oxidation state, such as Fe(IV) or Fe(III), etc.] moieties are treated as being separate spin centers that may or may not participate in various types of "spin coupling" interactions with another spin center. Usually, the total spin for two "spin centers" can be "calculated" by just doing the calculation on the alpha and beta electrons, in the same manner as described above. In some cases, however, such as when the a1u and a2u orbitals are each singly-occupied by electrons of opposite spin, the singly-occupied porphyrin orbitals undergo "configuration interaction" and form a new orbital, basically or literally, that, effectively, contains two paired electrons that don't affect the total spin, when the new orbital(s) (I'm fairly certain that it can be a single orbital or pair of orbitals in A1g symmetry in the D4h point-group symmetry of the isolated, nonsubstituted, planar porphyrin moiety) are formed, of the would-be coupling between the porphyrin and FeO moieties. For example, in that situation, the overall singlet state [two singly-occupied molecular orbitals containing electrons of opposite spin (or referred to as alpha or beta orbitals, etc.)] of the a2u and a1u orbitals, taken together, does not figure in to the "calculation" of the total spin for the particular electronic state (it's still an overall triplet state, with the triplet designation being dictated entirely by the spins of the diradical electrons in the pi* xz and yz antibonding molecular orbitals of the FeO moiety (as shown below). I think it's either that it's like a new orbital an "A1g orbital" in a singlet state (a porphyrin molecular orbital of A1g symmetry in the D4h point group, which is different from "A1g state symmetry," referring to two doubly-occupied molecular orbitals, as shown in the diagrams below) or its symmetry is such that an interaction with the FeO moiety is symmetry-forbidden or spin-forbidden or something. I don't understand how to assign point-group symmetries to specific molecules, yet, but I can sort of work backwards and assign axes and planes of symmetry. The business of assigning molecules to point groups of symmetry and assigning orbital symmetry designations is very complex, and I need a lot more practice on that. Anyway, I don't have time to put up anything else right now.