Pulsed-alkylation mass spectrometry for the study of protein folding and dynamics: development and application to the study of a folding/unfolding intermediate of bacterial luciferase.

A new method employing the classical techniques of chemical modification of proteins and the new technology of mass spectrometry, known as pulsed-alkylation mass spectrometry (PA/MS), has been developed to probe the dynamic structure of folding intermediates and folded complexes of proteins under a variety of conditions. This method is fast and simple, and the results are easily interpreted. PA/MS may provide an alternative to H/D exchange monitored either by NMR or by electrospray ionization mass spectrometry for some experiments; for others, it may provide access to questions not readily answered by available methods. The objective of PA/MS is to determine simultaneously the location and the extent of labeling of functional groups in a protein by measuring the reactivity of cysteines with N-ethylmaleimide, within the context of the conformation of the protein under specific conditions. The method can also be applied to chemical modification of other amino acid residues employing any of a vast array of reagents, depending upon the specifics of the protein under investigation. The enormous range of reactivity of the thiol groups of the cysteinyl residues in proteins and the change in reactivity upon denaturation or conformational rearrangement afford a large signal change that can be correlated with changes in accessibility of the thiol group. The information obtained from the correlation of observed thiol reactivity with the local environment of each cysteinyl residue in the structure of the folded protein can be supplemented by results obtained from fluorescence, circular dichroism, or other methods, to develop an understanding of the structure and dynamics of altered conformational states. With bacterial luciferase as a model system, we have applied PA/MS to investigate the structural differences between the native heterodimeric enzyme and a folding intermediate that is well-populated in 2 M urea. The thiol residues at positions 307, 324, and 325 of the alpha subunit were much more reactive with N-ethylmaleimide in the presence of 2 M urea than in the native enzyme, suggesting that the C-terminal region of the alpha subunit was less tightly packed in the folding intermediate. The apparent unfolding of the C-terminal region of the alpha subunit of the alphabeta structure in 2 M urea appears to mimic the unfolding of the C-terminal domain of the free alpha subunit, also in 2 M urea, described by Noland, B. W., Dangott, L. J., and Baldwin, T. O. (1999) Biochemistry 38, 16136-16145. The approach described here should be applicable to a wide array of problems that have in common the need to determine the locations of conformational changes in proteins. Application of PA/MS to the investigation of the relative thermodynamic stability of the coordination complexes of zinc within each of the six zinc-finger domains of MRE-binding transcription factor-1 (Zn(6) MTF-zf) in its free and DNA-bound forms is presented in the companion paper in this issue [Apuy, J. L., Chen, X., Russell, D. H., Baldwin, T. O., and Giedroc, D. P. (2001) Biochemistry 40, 15164-15175].

Ratiometric pulsed alkylation/mass spectrometry of the cysteine pairs in individual zinc fingers of MRE-binding transcription factor-1 (MTF-1) as a probe of zinc chelate stability.

Metal-response element (MRE)-binding transcription factor-1 (MTF-1) is a zinc-regulated transcriptional activator of metallothionein (MT) genes in mammalian cells. The MRE-binding domain of MTF-1 (MTF-zf) has six canonical Cys(2)-His(2) zinc finger domains that are distinguished on the basis of their apparent affinities for zinc and their specific roles in MRE-binding. In this paper, pulsed alkylation of the zinc-liganding cysteine thiolate pairs with the sulfhydryl-specific alkylating reagent d(5)-N-ethylmaleimide (d(5)-NEM) is used as a residue-specific probe of the relative stabilities of the individual zinc finger coordination complexes in Zn(6) MTF-zf. A chase with excess H(5)-N-ethylmaleimide (H(5)-NEM) to fully derivatize MTF-zf concomitant with complete proteolysis, followed by MALDI-TOF mass spectrometry allows quantitation of the mole fraction of d(5),d(5)-, d(5),H(5)-, and H(5),H(5)-NEM derivatized peptides corresponding to each individual zinc finger domain as a function of d(5)-NEM pulse time. This experiment establishes the hierarchy of cysteine thiolate reactivity in MTF-zf as F5 > F6 >> F1 > F2 approximately F3 approximately F4. The apparent second-order rate of reaction of F1 thiolates is comparable to that determined for the DNA binding domain of Sp1, Zn(3) Sp1-zf, under identical solution conditions. The reactivities of all Cys residues in MTF-zf are significantly reduced when bound to an MREd-containing oligonucleotide. An identical experiment carried out with Zn(5) MTF-zf26, an MTF-zf domain lacking the N-terminal F1 zinc finger, reveals that MTF-zf26 binds to the MREd very weakly, and is characterized by strongly increased reactivity of nonadjacent F4 thiolates. These findings are discussed in the context of existing models for metalloregulation by MTF-1.

Functional implications of the unstructured loop in the (beta/alpha)(8) barrel structure of the bacterial luciferase alpha subunit.

Bacterial luciferase catalyzes the conversion of FMNH(2), a long-chain aliphatic aldehyde, and molecular oxygen to FMN, the corresponding carboxylic acid, and H(2)O with the emission of light. The light-emitting species is an enzyme-bound excited state flavin. The enzyme is a heterodimer (alphabeta) of homologous subunits each with an (beta/alpha)(8) barrel structure. A portion of the loop in the alpha subunit that connects beta strand 7 to alpha helix 7 is disordered in the crystal structure. To test the hypothesis that this loop closes over the active site during catalysis and protects the active site from bulk solvent, a mutant was constructed in which the 29 residues that are disordered in the 2.4 A crystal structure were deleted. Deletion of this loop results in a heterodimer with a subunit equilibrium dissociation constant of 1.32 +/- 1.25 microM, whereas the wild-type heterodimer shows no measurable subunit dissociation. This mutant retains its ability to bind substrate flavin and aldehyde with wild-type affinity and can carry out the chemistry of the bioluminescence reaction with nearly wild-type efficiency. However, the bioluminescent quantum yield of the reaction is reduced nearly 2 orders of magnitude from that of the wild-type enzyme.

Process of biosynthetic protein folding determines the rapid formation of native structure.

Biosynthetic folding, beginning with the growing nascent chain and leading to the biologically active structure within its proper cellular context, is one function shared by all proteins. We show that the bacterial luciferase beta subunit reaches its final native form in the alphabeta heterodimer much more rapidly during biosynthetic folding than during refolding from urea. The rate of formation of active enzyme is determined by a short-lived folding intermediate, which is able to associate with the alpha subunit very rapidly following release from the ribosome. This intermediate appears to involve a transient interaction of the C-terminal region of the beta subunit, a region distant from the subunit interface, but intimately involved in heterodimerization. Refolding of the beta subunit under similar conditions proceeds much more slowly. We have characterized both pathways and show that the basic difference between biosynthetic folding and refolding from urea is that the newly synthesized beta subunit enters the folding pathway at a point beyond the slow, rate-determining step that limits the rate of the renaturation process and constitutes a kinetic trap. This mechanism embodies a major strategy, the avoidance of slow-folding intermediates and kinetic traps, that may be employed by many proteins to achieve fast and efficient biosynthetic folding.

Folding, stability, and physical properties of the alpha subunit of bacterial luciferase.

Bacterial luciferase is a heterodimeric (alphabeta) enzyme composed of homologous subunits. When the Vibrio harveyi luxA gene is expressed in Escherichia coli, the alpha subunit accumulates to high levels. The alpha subunit has a well-defined near-UV circular dichroism spectrum and a higher intrinsic fluorescence than the heterodimer, demonstrating fluorescence quenching in the enzyme which is reduced in the free subunit [Sinclair, J. F., Waddle, J. J., Waddill, W. F., and Baldwin, T. O. (1993) Biochemistry 32, 5036-5044]. Analytical ultracentrifugation of the alpha subunit has revealed a reversible monomer to dimer equilibrium with a dissociation constant of 14.9 +/- 4.0 microM at 18 degrees C in 50 mM phosphate and 100 mM NaCl, pH 7.0. The alpha subunit unfolded and refolded reversibly in urea-containing buffers by a three-state mechanism. The first transition occurred over the range of 0-2 M urea with an associated free-energy change of 2.24 +/- 0.25 kcal/mol at 18 degrees C in 50 mM phosphate buffer, pH 7.0. The second, occurring between 2.5 and 3.5 M urea, comprised a cooperative transition with a free-energy change of 6.50 +/- 0.75 kcal/mol. The intermediate species, populated maximally at ca. 2 M urea, has defined near-UV circular dichroism spectral properties distinct from either the native or the denatured states. The intrinsic fluorescence of the intermediate suggested that, although the quantum yield had decreased, the tryptophanyl residues remained largely buried. The far-UV circular dichroism spectrum of the intermediate indicated that it had lost ca. 40% of its native secondary structure. N-Terminal sequencing of the products of limited proteolysis of the intermediate showed that the C-terminal region of the alpha subunit became protease labile over the urea concentration range at which the intermediate was maximally populated. These observations have led us to propose an unfolding model in which the first transition is the unfolding of a C-terminal subdomain and the second transition represents the unfolding of a more stable N-terminal subdomain. Comparison of the structural properties of the unfolding intermediate using spectroscopic probes and limited proteolysis of the alpha subunit with those of the alphabeta heterodimer suggested that the unfolding pathway of the alpha subunit is the same, whether it is in the form of the free subunit or in the heterodimer.

Deuterium kinetic isotope effects and the mechanism of the bacterial luciferase reaction.

A combined experimental and theoretical investigation of the deuterium isotope effects on the bacterial luciferase reaction is described. The experimental studies focus on determining if the unusual aldehydic deuterium isotope effect of approximately 1.5 observed in these reactions is an intrinsic isotope effect resulting from a single rate-limiting step or is a composite of multiple rate-limiting steps. The isotope effect observed is not significantly affected by variation in the aldehyde chain length, changes in the pH over a range of 6-9, use of alphaC106A and alphaC106S site-directed mutants, or chloride substitution at the 8-position of the reduced flavin, though the isotope effect is decreased when the 8-methoxy-substituted flavin is used as a substrate. From these observations it is concluded that the aldehydic isotope effect arises from the change in rate of a single kinetic step. A stopped-flow kinetic analysis of the microscopic rate constants for the reactions of 1-[1H]decanal and 1-[2H]decanal in the bacterial luciferase reaction was carried out, and aldehyde hydration isotope effects were determined. From the results it is estimated that the aldehydic deuterium isotope effect is approximately 1.9 after formation of an intermediate flavin C4a-hydroperoxy hemiacetal. Ab initio calculations were used to examine the transformation of the aldehyde into a carboxylic acid and to predict isotope effects for possible mechanisms. These calculations indicate that the mechanism involving rate-limiting electron transfer from the flavin C4a-hydroxide to an intermediate dioxirane is consistent with the enigmatic aldehydic isotope effect and that the intermediacy of a dioxirane is energetically plausible.

GroE modulates kinetic partitioning of folding intermediates between alternative states to maximize the yield of biologically active protein.

The central issue of chaperone function is the mechanism whereby partitioning of folding polypeptides along the productive pathway may be maximized, while non-productive folding pathways are minimized. We have found that the GroE chaperone is capable of accelerating the rate of the productive pathway of bacterial luciferase alphabeta heterodimer formation. At intermediate temperatures at which the productive pathway and non-productive pathways leading to dimerization-incompetent monomeric forms of the subunits coexist, GroE enhances the yield of native enzyme while minimizing the yield of misfolded protein. These results suggest that GroE releases the subunits in forms capable of achieving the native structure faster than the forms initially bound by the chaperone. At higher temperatures, at which the native enzyme is stable but the dimerization reaction is diminished, GroE is unable to force the productive folding reaction to occur. However, the chaperone decreases the rate of formation of the heterodimerization-incompetent species, thereby enhancing the final yield of active enzyme when the temperature is reduced to the permissive range. Our results suggest a mechanism by which the chaperone functions to maximize the yield of the biologically active form of the protein while maintaining or even accelerating the essential rapid kinetics of folding reactions.

Kinetic mechanism of luciferase subunit folding and assembly.

The kinetic mechanism in vitro of the folding and assembly of the heterodimeric flavin monooxygenase bacterial luciferase has been defined by a unique set of rate constants which describe both the productive refolding pathway and competing off-pathway reactions in 50 mM phosphate, pH 7.0 at 18 degrees C. The individual alpha and beta subunits fold independently to form heterodimerization-competent species, alpha i and beta i. The alpha i beta i species can interact to form an inactive heterodimeric intermediate, [alpha beta ]i, which isomerizes to form the active alpha beta structure; the structure of the enzyme has been determined to 1.5 A resolution [Fisher, A. J., Thompson, T. B., Thoden, J. B., Baldwin, T. O., & Rayment, I. (1996) J. Biol. Chem. 271, 21956-21968]. In the absence of alpha i, beta i can form a kinetically trapped homodimer, beta 2, with a second-order rate constant of about 180 M-1 s-1 [Sinclair, J. F., Ziegler, M. M., & Baldwin, T. O. (1994) Nat. Struct. Biol. 1, 320-326]; the structure of beta 2 has recently been reported [Thoden. J. B., Holden, H. M., Fisher, A. J., Sinclair. J. F., Wesenberg, G., Baldwin, T.O., & Rayment, I. (1997) Protein Sci. 6, 13-23]. The beta i species, or some other form that precedes beta i on the refolding pathway, can also undergo a first-order conversion into a form (designated beta x) that cannot associate with alpha i to form the native enzyme. The rate constant for this process, assigned here, accounts well for the previously observed dependence of final yield on concentration of refolding species [Ziegler, M.M., Goldberg, M.E., Chaffotte, A. F., & Baldwin, T. O. (1993) J. Biol. Chem. 268, 10760-10765]. In simulations of the refolding reaction, all processes associated with the refolding of the individual subunits were combined into single first-order rate constants for each subunit which were consistent with the rate constants determined from stopped-flow circular dichroism studies. The first-order rate constant for the folding of the alpha subunit, estimated from the concentration-independent lag preceding the appearance of active enzyme, and the second-order rate constant for assembly of alpha i and beta i into the heterodimer, estimated from the concentration-dependent rate of appearance of active enzyme, were consistent with the rates of first- and second-order processes monitored by changes in fluorescence of an extrinsic probe [the product of modification with N-(4-anilino-1-naphthyl)maleimide] on the alpha subunit during refolding. The rate constant for the isomerization of [alpha beta]i to form the active heterodimer was estimated from the kinetic data of a secondary dilution experiment and from fluorescence measurements of protein diluted 20-fold from 2.1 M urea-containing buffer. The rate constants reported here for the kinetic mechanism of refolding permitted simulation of the time courses and yields for activity recovery during the refolding of luciferase from about 1 to 25 micrograms/mL which are in excellent agreement with our previously reported data.

Structure of the beta 2 homodimer of bacterial luciferase from Vibrio harveyi: X-ray analysis of a kinetic protein folding trap.

Luciferase, as isolated from Vibrio harveyi, is an alpha beta heterodimer. When allowed to fold in the absence of the alpha subunit, either in vitro or in vivo, the beta subunit of enzyme will form a kinetically stable homodimer that does not unfold even after prolonged incubation in 5 M urea at pH 7.0 and 18 degrees C. This form of the beta subunit, arising via kinetic partitioning on the folding pathway, appears to constitute a kinetically trapped alternative to the heterodimeric enzyme (Sinclair JF, Ziegler MM, Baldwin TO. 1994. Kinetic partitioning during protein folding yields multiple native states. Nature Struct Biol 1: 320-326). Here we describe the X-ray crystal structure of the beta 2 homodimer of luciferase from V. harveyi determined and refined at 1.95 A resolution. Crystals employed in the investigational belonged to the orthorhombic space group P2(1)2(1)2(1) with unit cell dimensions of a = 58.8 A, b = 62.0 A, and c = 218.2 A and contained one dimer per asymmetric unit. Like that observed in the functional luciferase alpha beta heterodimer, the major tertiary structural motif of each beta subunit consists of an (alpha/beta)8 barrel (Fisher AJ, Raushel FM, Baldwin TO, Rayment I. 1995. Three-dimensional structure of bacterial luciferase from Vibrio harveyi at 2.4 A resolution. Biochemistry 34: 6581-6586). The root-mean-square deviation of the alpha-carbon coordinates between the beta subunits of the hetero- and homodimers is 0.7 A. This high resolution X-ray analysis demonstrated that "domain" or "loop" swapping has not occurred upon formation of the beta 2 homodimer and thus the stability of the beta 2 species to denaturation cannot be explained in such simple terms. In fact, the subunit:subunit interfaces observed in both the beta 2 homodimer and alpha beta heterodimer are remarkably similar in hydrogen-bonding patterns and buried surface areas.

Autoinducer-independent mutants of the LuxR transcriptional activator exhibit differential effects on the two lux promoters of Vibrio fischeri.

The LuxR protein is a transcriptional activator which, together with a diffusible small molecule termed the autoinducer [N-(3-oxohexanoyl)-L-homo-serine lactone], represents the primary level of regulation of the bioluminescence genes in Vibrio fischeri. LuxR, in the presence of autoinducer, activates transcription of the luxICDABEG gene cluster and both positively and negatively autoregulates transcription of the divergently oriented luxR gene, activating transcription at low levels of autoinducer, and repressing synthesis at high autoinducer concentration. Seven LuxR point mutants which activate V. fischeri lux transcription in the absence of autoinducer (LuxR*) have been characterized. The LuxR* proteins activated transcription of the bioluminescence genes to levels 1.5-40 times that achieved by wild-type LuxR without autoinducer. All of the LuxR* mutants retained responsiveness to autoinducer. However, in each case the degree of stimulation in response to autoinducer was lower than that observed for wild-type LuxR. The LuxR* proteins retained the requirement for autoinducer for autoregulation of the luxR gene. We propose that the LuxR protein exists in two conformations, an inactive form, and an active form which predominates in the presence of autoinducer. The LuxR* mutations appear to shift the equilibrium distribution of these two forms so as to increase the amount of the active form in the absence of autoinducer, while autoinducer can still convert inactive to active species. The differential effects of the LuxR* proteins at the two lux promoters suggest that LuxR stimulates PluxR transcription by a different mechanism to that used at the PluxI promoter, implying that binding of LuxR to its binding site, known to be necessary for transcriptional activation, may not be sufficient.

The 1.5-A resolution crystal structure of bacterial luciferase in low salt conditions.

Bacterial luciferase is a flavin monooxygenase that catalyzes the oxidation of a long-chain aldehyde and releases energy in the form of visible light. A new crystal form of luciferase cloned from Vibrio harveyi has been grown under low-salt concentrations, which diffract x-rays beyond 1.5-A resolution. The x-ray structure of bacterial luciferase has been refined to a conventional R-factor of 18.2% for all recorded synchrotron data between 30.0 and 1.50-A resolution. Bacterial luciferase is an alpha-beta heterodimer, and the individual subunits fold into a single domain (beta/alpha)8 barrel. The high resolution structure reveals a non-prolyl cis peptide bond that forms between Ala74 and Ala75 in the alpha subunit near the putative active site. This cis peptide bond may have functional significance for creating a cavity at the active site. Bacterial luciferase employs reduced flavin as a substrate rather than a cofactor. The structure presented was determined in the absence of substrates. A comparison of the structural similarities between luciferase and a nonfluorescent flavoprotein, which is expressed in the lux operon of one genus of bioluminescent bacteria, suggests that the two proteins originated from a common ancestor. However, the flavin binding sites of the nonfluorescent protein are likely not representative of the flavin binding site on luciferase. The structure presented here will furnish a detailed molecular model for all bacterial luciferases.

Algorithms for finding the axis of a helix: fast rotational and parametric least-squares methods.

Several methods for finding the axis of a helix are presented and compared. The most accurate determines the helix axis as the axis of rotation necessary to map point i to point i + 1 of the helix. The fastest method calculates the helix axis as the best-fit line through the coordinates by a three-dimensional parametric linear least-squares algorithm, taking advantage of the sequential nature of the data.

Implications of N and C-terminal proximity for protein folding.

We have investigated the proximity of the N and C termini in protein structures, and developed a model to test the theoretical possibility that proteins fold with their termini closely associated. On average, the distance between the termini is not significantly different from what would be expected based on chance. However, the theoretical model indicated that it is possible to greatly decrease the N-to-C terminal distance by allowing small (approximately six amino acid residues) solvent-accessible terminal fragments to move. Subsequent to this distance minimization method, more than 90% of the proteins studied had smaller-than-expected N-to-C distances, but only minor structural modification.

Firefly luciferase: the structure is known, but the mystery remains.

The structure of firefly luciferase reveals a new protein fold which may be representative of a growing family of homologous enzymes.