In multiple protein sequence alignment of vertebrate carotenoid oxygenase family member sequences (Fig. 1), we found that all members conserved aromatics in positions comparable to F51, Y235, and W454. With respect to Y326 in BCMO1, Y was conserved in other BCMO1 orthologs, W was the paralogous residue in RPE65, but aromatic resides were not found in BCMO2 sequences at this position. It should be noted that BCMO1 and RPE65 both act at the 15 carbon position, whereas BCMO2 is thought to cleave carotenoids asymmetrically. The Swiss-PDB model of BCMO1 built on the ACO template predicts that F51, Y235, Y326 and W454 are aromatics located in the carotenoid binding cleft. The possibility that some or all of these residues participate in catalytically important functions was thus an important consideration. Tyrosines are known to play a key role in coupled proton-shuttling in a large number of metalloenzymes [39, 40], and, additionally, can directly participate in electron transfer by forming tyrosyl radicals [41, 42]. They may also play a structural role by participating in the architecture of the enzyme active site, or play a role in the correct positioning of the substrate and cofactor [43, 44] and are known to take part in interaction with substrate and/or cofactors and determine the substrate specificity in a variety of enzymes [44–47]. Finally, tyrosine and other aromatic amino acids are implicated in cation-π interactions stabilizing carbocation intermediates generated during enzymatic reactions [27, 48, 49].
Our results indicate that mutations of Y235 and Y326 to leucine significantly impair catalytic activity of BCMO1. We found that the aromaticity of amino acids Y235 and Y326 are important for enzymatic activity, whereas mutation of F51 or W454 to leucine had little effect. Mutation of either of these two tyrosines to other aromatic residues (phenylalanine or tryptophan) does not affect enzyme activity by more than 10-20% relative to wild type, but replacement to non-aromatic residues (leucine or glutamine) leads to significant loss of enzymatic activity. The studies presented here support a mechanism implicating cation-π stabilization of a putative carbocation intermediate. We can exclude the possibility of radical formation by these tyrosines in BCMO1 catalysis, as the tyrosine to leucine mutants, Y235L and Y326L, although reduced in activity (sometimes significantly), exhibited some level of residual activity. Additionally, EPR analyses of BCMO1 failed to detect tyrosyl radical formation (Poliakov and Krishna, unpublished results). However, this finding does not exclude the possibility of other functional roles for these residues. The Swiss-PDB model of BCMO1 built on the ACO template predicts close localization of Y235-E140 and Y326-E405 residues with possible hydrogen bonding for the latter pair (~2.73 Å) (Fig. 2B). However, our mutagenesis data suggest that these hydrogen bonds cannot be crucial for the catalytic mechanism of BCMO1; moreover, simultaneous replacement of Y235F/E140A does not eliminate activity.
From our data it is reasonable to propose that aromaticity of the two tyrosines Y235 and Y326 plays an important role in the catalytic mechanism of BCMO1. To explore this possibility, we mounted an ab initio model of β-carotene onto the BCMO1 structural model by thermodynamic optimization. It was observed that tyrosines Y235 and Y326 fix the position of the substrate on two sides of the 15-15' double bond of the substrate (Fig. 2C). The predicted hydrophobic interactions with substrate are in accordance with our catalytic activity data for Y235F, Y235W and Y235L, and Y326W, Y326F and Y326L mutants. Consequently, we can speculate that in the case of a mechanism involving one-electron oxidation by dioxygen bound to the ferrous ion and the subsequent radical-cation formation , or two one-electron oxidation steps and dication formation , the aromatic residues (Y235, Y326) and glutamates (E140, E405) stabilize a cationic intermediate with charge delocalization along the polyene chain. On the other hand, we could not rule out the possibility of protonation of a double bond and formation of a carbocation as in the case of isopentyl diphosphate:dimethylallyl diphosphate isomerase . Distances between these tyrosines and β-carotene chain in our model are less that 5 Å which is in good agreement with cation-π stabilization (Fig. 2C) . The occurrence of such stabilization was previously observed in squalene-hopene cyclase and oxidosqualene-lanosterol cyclase [26, 27]. In the polycyclization cascade of squalene and oxidosqualene, aspartates initiate the cyclization process by protonation of the terminal double bond and stabilize the first carbocation intermediate . The π-electrons of several aromatic residues stabilize cationic intermediates in other enzymes [49, 51, 52]. The cation intermediate would explain the observation of transient substrate isomerization in the binding cleft of ACO . Transient carbocations are formed in many steps of sterol and carotenoid synthesis, and various positively charged amino derivatives could noncompetitively inhibit these enzymes. Of the three inhibitors tested here, diphenylamine, an oxidation inhibitor, was the only one to have a substantial inhibitory effect on BCMO1 activity. Diphenylamine is reported to inhibit electron transfer in photosynthetic membranes  and carotenogenesis by inhibiting carotenoid ketolase and phytoene desaturase. However, this is in line with the common mechanism of inhibition by diphenylamine of iron oxygenases and is not necessarily conclusive for carbocation formation.
As expected, replacement of Y326 by tryptophan, which has the highest π-binding energy of the aromatic series , did not negatively effect activity in vivo. The altered biochemical parameters (especially the higher Km) for the Y326W mutant could be explained by steric hindrance. Also, it is important to note that the in vivo assay shows the cumulative effect of mutation and thus cannot estimate the catalytic efficiency of the enzyme. Replacement of tyrosine by glutamine (a partially positively charged residue) would be expected to destabilize a cation, and therefore, effectively inhibit catalytic activity of BCMO1. The results presented here correlate with these hypotheses. Productive replacement of Y326 by tryptophan, such as found at the paralogous residue W331 in RPE65  suggests that BCMO1 and RPE65 may share a similar catalytic mechanism. In fact, the formation of a carbocation intermediate had been proposed in the catalytic mechanism of retinoid isomerization in the RPE [57, 58] prior to the identification of RPE65 as the isomerase [6–8].