Elicitation of an oxidase activity in bacterial luciferase by site-directed mutation of a noncatalytic residue.

Flavin-dependent external monooxygenases and oxidases could catalyze the same flavin oxidation reaction involving distinct mechanisms. To gain insights into enzyme structure-function relationship, site-directed mutagenesis was carried out for Vibrio harveyi luciferase, a monooxygenase. The substitution of the alpha subunit cysteine 106 by alanine shows unambiguously that the alphaCys106 is not essential to catalysis. The corresponding substitution by valine resulted in a substantial reduction of the bioluminescence activity correlatable with the induction of a new flavin oxidation activity typical for oxidases. These findings indicate that mutation of a single noncatalytic residue at the active center of a flavoenzyme could transform one enzyme type to another, thus highlighting the subtlety of enzyme active site structure in relation to catalysis and the versatility of enzyme evolution.

Flavoenzymes catalyze a great variety of biological redox reactions. The diversity is marked by not only the types of substrates utilized but also the underlying reaction mechanisms (I, 2). Therefore, flavoenzymes provide one of the most challenging systems for the elucidation of enzyme structurefunction relationships.
For flavin-dependent oxidases and external monooxygenases, the initial oxidation of a substrate is coupled with the reduction of the enzyme-bound flavin (E-F). The subsequent oxidation of the bound reduced flavin ( (8,9).
Vibrio harveyi luciferase has eight cysteinyl residues on LY (M, 40,100) (10) and an additional six on p (M, 36,400) (11). The cysteinyl residue at position 106 on the (Y subunit (12,13), which is particularly reactive, can be selectively modified by a number of chemical reagents leading to luciferase inactivation (14)(15)(16)(17). Recently, we have selectively methylated the V. harveyi luciferase CUC~S'~' by methyl p-nitrobenzene sulfonate and have found that the modified enzyme retains the ability to bind aldehyde and FMNH2, but the formation of the peroxyflavin intermediate II is impaired (13). Therefore, the intriguing question as to whether or not the aCyPmodified luciferase is active in catalyzing the oxidation of FMNHz or aldehyde in a dark reaction should be addressed. In this work, we have conducted site-directed mutagenesis to replace the CUC~S'~~ with either an alanine (cuC106A) or a valine (cuC106V) to gain further insights into the luciferase structure-function relationship.  (27).

AND DISCUSSION
Both the &lOGA and the &106V enzymes were found to be active in catalyzing the bioluminescence emission (Table   I). On the basis of either the peak intensity (ZJ or the total quantum output (Q), the aC106A enzyme is substantially more active than the &106V luciferase. The substitution of alanine for cysteine did not affect much the K,,, for aldehyde but resulted in a 6.5-fold increase in the Km for FMNH2. On the other hand, a 2.7-fold increase in the Km for FMNH* as well as 3.7-and 5.1-fold increases in the K,,, for decanal and dodecanal, respectively, resulted from the substitution of valine for the cysteine. When compared with the wild-type  further that the activities observed for the mutant enzymes are not due to any contamination of wild-type luciferase.
Previously, the V. harveyi luciferase aCy~'~' has been converted to a serine residue with the resulting luciferase variant (&lOGS) found active in bioluminescence (28,29). Since serine and cysteine side chains are similar in size and are homologous nucleophiles, we have argued (13) that the retention of bioluminescence activity by the cuC106S enzyme does not necessarily rule out a catalytic function for the cyCys'OG residue. Consistent with this argument, activity retention has been observed for subtilisin (30,31) and trypsin (32) with their active site essential serine converted to a cysteine. In the present case of cuClO6A luciferase, the chemical reactivity of alanine side chain is drastically different from that of cysteine, but this mutant enzyme is highly active in bioluminescence. Such  Although the oxidation of the aClOGV-bound FMNH, is only twice as fast as the nonenzymatic oxidation, one should note that autoxidation is itself remarkably fast and also that luciferase is an unusually slow enzyme. For the wild-type luciferase under similar conditions, the rate constant for the formation of FMN via the dark decay of II was 0.05 s-i (Table  I) and that via the light pathway was 0.29 s-l (with decanal) and 0.04 s-l (with dodecanal) (20). In comparison, 90-630folds of rate enhancements are exhibited by the aC106V luciferase for the oxidation of FMNH,. both luciferase variants and the properties of decanal binding for the aC106A enzyme reported by these authors (33) are similar to but quantitatively somewhat different from our findings.
We propose that the CUC~S'~' has an important structural role such that its modification could lead to a complete or partial inactivation of luciferase due to a critical, and perhaps subtle (13), structural perturbation of the active site. Moreover, a small size residue at position 106 of the 01 subunit appears to be important to keep luciferase in a more active state; substitution of CYC~S'~~ by serine (28,29) or alanine retains the full or much of the activity, whereas luciferase is inactivated by methylation of CYC~S'~' (13) and substantially deactivated by substitution of valine for the smaller cysteine. A very interesting finding revealed by the present study is that the substitution of valine for the catalytically nonessential ~tCys'~~ of luciferase results in a change of its catalytic reaction mechanism. The kinetics of the nonenzymatic oxidation of FMNH, are complex (34) but, for the present work, can be approximated as a pseudo-first order process with a rate constant of 13.1 s-i (Fig. 1, truce A). The oxidation of wild-type luciferase-bound FMNHz showed two distinct phases (Fig. 1, truce B). The initial phase was associated with a small signal change and, on the basis of spectral and kinetic characterizations reported previously (9,20,35), was attributed to the formation of II. The second phase exhibited a much larger signal rise following apparent first order kinetics (lz = 0.02 s-l) and was assigned to the subsequent slow decay of the 4a-hydroperoxy FMN intermediate II to form FMN. A biphasic time course was also observed for the oxidation of &lOGV-bound FMNH, (Fig. 1, truce C). However, the fast phase (Iz = 25.2 s-l) was associated with most of the signal increase whereas the second phase involved only about 3% of the total U&O. We believe that these phases represent two independent, rather than consecutive, reactions; the major It has been shown that the luciferase emission is coupled to the hydroxylase pathway (Scheme 1, A) in which the aldehyde is converted to acid (36)  The recovery of tetradecanal was determined by bioluminescence assays using excess Photobacterium phosphoreum luciferase and aliquots of the sample solution as the aldehyde supply. When tetradecanal was similarly reacted with FMNH* and oxygen in the presence of aC106V (4.8 PM), 89 + 2% of the expected aldehyde recovery was obtained. Therefore, the &106V enzyme has little or no activity to oxidize aldehyde. This finding is consistent with the observation that the major pathway for the oxidation of aClOGV-bound FMNHz does not involve 4ahydroperoxy FMN as an intermediate. It is important to note that the major route of FMNHz oxidation by &lOGV, depicted as pathway C in Scheme 1, mimics the activity of a typical flavin-dependent oxidase (Equation 1) in contrast to that for the pseudooxidase activities of flavomonooxygenases (Equation 2b; Scheme 1, pathway B). Furthermore, this new oxidase activity for aC106V luciferase was elicited by a single mutation of a residue not directly involved in chemical catalysis. We propose that the positioning of oxygen and the enzyme-bound flavin has a crucial role in dictating the chemical mechanism of FMNH, oxidation.
For flavomonooxygenases, the microenvironment of the active site is such that it favors the attack of the flavin 4a position by the molecular oxygen to form the key 4ahydroperoxyflavin intermediate. However, for flavin-dependent oxidases the reduced flavin cofactor and oxygen are positioned in ways that disallow the accessibility of the flavin 4a site to OS. In the case of luciferase, a point mutation of the catalytically nonessential ~&ys'~' is apparently a sufficient structural perturbation leading to the induction of a new oxidase-like pathway of flavin oxidation. The subtlety of enzyme active center microenvironment in relation to the expression of catalytic mechanism and the versatility of enzyme evolution are thus indicated.