Chemistry Department

Enzymes in the methionine salvage pathway: structure and function

Both enzymes are monomeric and relatively small in size. E1 is 24 kDa (229 residues) and ARD is 18 kDa (176 residues). The genes for both have been cloned into pET expression vectors, and high level expression of both enzymes has been achieved in E. coli. All of these factors make these enzymes, for which no crystal structures exist and which are without homologues in the PDB database, attractive targets for NMR structure/function studies.

A particularly interesting feature of ARD is that its functionality changes depending on which metal ion is bound to it. ARD to which Ni(II) (NiARD) is bound catalyzes the decomposition of acireductone substrate to the (n-2) carboxylic acid, carbon monoxide and formate. From the same substrate, ARD to which Fe(II) is bound generates the (n-1) a-keto acid and formate. The function (if any) of CO is unknown [1, 2, 3]. However, CO has been implicated as a potential neurotransmitter in vertebrates [4].

We are currently refining the solution structure of Ni-ARD, and have begun an investigation into the roles played by the metal ions in the course of the chemistry catalyzed by ARD [5]. We have determined the 1H, 13C and 15N NMR resonance assignments of the diamagnetic regions of NiARD and have determined the global fold by NMR methods [6]. Preliminary NMR studies of the ARD‚ (Fe) protein confirm that subtle but clear structural differences exist between the two proteins.

We are using a wide variety of tools in this project besides structural NMR methods. In collaboration with Prof. Barry Snider, we have developed a simple synthesis for substrate analogs of I, the precursor of acireductone II. We are now in a position to examine the mechanism of this enzyme using enzymological techniques. Molecular biological methods are also being employed by our lab for this work. Sequence alignment of ARD with genes and ORFs from other organisms suggest that conserved His residues H96, H98 and H140 are potential candidates for ligation of the Ni, along with conserved Glu residues E95 and E102. These are consistent with our observations of paramagnetic broadening of these regions of the protein in NMR spectra. We are currently mutating residues in this region to observe the effect that these mutations have on expression and activity.

We are also beginning the structural and functional characterization of the enolase-phosphatase E1, that catalyzes the conversion of the acyclic diketone intermediate I (Scheme I) to the enol phosphate, followed by dephosphorylation to yield aci-reductone II. We have expressed E1 in E. coli and have performed preliminary enzymatic activity assays and multidimensional NMR characterization of 15N,13C-labeled E1.

References:

1) J. W. Wray and R. H. Abeles (1993) J. Biol. Chem. 268, 21466-21469. Return to text

2) J. W. Wray and R. H. Abeles (1995) J. Biol. Chem. 270, 3147-3150. Return to text

3) Y. Dai, P. C. Wensink and R. H. Abeles (1999) J. Biol. Chem. 274, 1193-1195. Return to text

4) A. Verma, D. J. Hirsch, C. E. Glatt, G. V. Ronnett and S. H. Snyder (1993) Science 259, 381-384. Return to text

5) Y. Dai, T. C. Pochapsky and R. H. Abeles (2001) "Mechanistic studies of two dioxygenases in the methionine salvage pathway of Klebsiella pneumoniae" Biochemistry 40, 6379-6387. Return to text

6) H. Mo, Y. Dai, S. S. Pochapsky and T. C. Pochapsky (1999) "1H, 13C and 15N assignments for a carbon monoxide generating metalloenzyme from Klebsiella pneumoniae" J. Biomol. NMR 14, 287-288. Return to text

Monooxygenase enzyme systems: structure and dynamics

We are actively investigating structure and function of a class of ferredoxins that transfer electrons to monooxygenases in bacterial and mammalian cells. Using multidimensional NMR methods, we determined a model for the solution structure of putidaredoxin (Pdx), the physiological reductant of cytochrome P-450cam. Pdx was the first protein of its class for which a structure was determined, and the first metalloprotein of any sort for which a structure was determined exclusively by NMR without reference to a homologous crystal structure [1, 2, 3]. The interaction between Pdx and cytochrome P-450cam is modulated by the redox state of Pdx, and we have characterized structural and dynamic differences between the oxidized and reduced forms of Pdx using 1H-15N correlated spectroscopy [4,5].

Currently, the ferredoxin project is taking several directions. We are using NMR and site-directed mutagenesis to determine the origins of functional differences between Pdx and related bacterial and mammalian ferredoxins, including terpredoxin (Tdx), which is the redox partner of cytochrome P-450terp, and the human ferredoxin adrenodoxin (Adx), which transfers electrons to cytochromes P450 in the steroidogenic pathways in humans. These three ferredoxins all have similar tertiary structures [6], but show remarkable specificity for interacting only with their own cognate P450s. We have identified a number of sequence differences between the three proteins that may be important in determining the low cross-reactivity of the ferredoxins, and we are attempting to engineer cross-reactive hybrid ferredoxins using site-directed mutagenesis, thereby identifying structural elements important for molecular recognition between ferredoxins and cytochrome P450s. We have constructed a gene for the human Adx, and we have recently submitted a paper describing an in-depth structural comparison of Pdx, Adx and Tdx that uses NMR and site-directed mutations to characterize structural and dynamic differences between these three ferredoxins [7].

We have also recently used site-directed mutagenesis to examine the nature of redox-dependent dynamics in Pdx. We know that binding between Pdx and cytochrome P450cam is oxidation state -dependent, that is, reduced Pdx (Pdxr) binds more tightly than the oxidized Pdx (Pdxo) to P450cam. We previously suggested that this behavior is dynamic in origin, and that Pdxo occupies more conformational substates than Pdxr. If the subset of conformations occupied by Pdxr is the same that are occupied when Pdx binds to P450cam, then we have found the mechanism by which oxidation state of Pdx modulates binding to P450cam. The problem was being able to actually observe these conformational substates individually, since they are interconverting fast on the NMR chemical shift time scale. By mutating a residue in the metal binding loop, Gly 40, to an asparagine, we were able to slow down substate interconversion sufficiently that we can actually observe these conformational shifts by NMR [8].

Backbone representation of the solution structure of oxidized Cys4Fe2S2 ferredoxin Pdx [3]. The iron-sulfur cluster is shown as four spheres. Backbone positions marked in red are dynamically affected by the presence of the G40N mutation [8] .

Fe2S2 binding loop and C-terminal cluster of Pdx. Residues His 49, Tyr 51, Ala 76, and Ser 82 are involved in the hydrogen bonding network proposed to transmit redox-dependent conformational changes from the metal cluster binding loop (backbone shown from Cys 39 to His 49) to the C-terminal cluster. Side chains of cysteinyl ligands to the metal cluster are shown as red lines.

One of the most frustrating problems facing researchers that examine paramagnetic proteins is the inability of standard NMR methods to probe the metal sites of these proteins due to paramagnetic broadening. We have developed some alternative methods for looking at the active sites of Pdx, Adx and Tdx that should be applicable to many paramagnetic proteins. The extent to which a resonance is broadened by paramagnetic relaxation is proportional to the gyromagnetic ratio of the nucleus being observed. For this reason, 15N and 13C NMR spectroscopy are useful structural probes in the vicinity of the metal center of Pdx. Both 15N and 13C have lower gyromagnetic ratios and so are less affected by the paramagnetism of the metal center than 1H. Using a combination of residue-selective isotope labeling and selective decoupling NMR experiments, we have sequence-specifically identified almost all of the amide 15N and carbonyl 13C resonances in the metal binding site of Pdx for both oxidation states. Interestingly, we see some evidence of significant conformational changes as a function of oxidation state in Pdx in these NMR studies [7, 9, 10].

Another way of dealing with the paramagnetism problem is simply to replace the paramagnetic metal with a diamagnetic metal. We have reconstituted Pdx with a diamagnetic metal ion, Ga+3, and have generated a stable, folded, diamagnetic protein (GaPdx) which retains the overall global fold of Pdx [11]. This protein has been prepared with uniform 13C,15N-labeling, and we have refined the structure of GaPdx using data from multidimensional NMR experiments [12, 13]. However, GaPdx is considerably more dynamic than the native protein and may represent a trapped late intermediate in the folding pathway. Also, since only a single gallium atom is taken up by the protein, the metal center replacement is not isosteric, and although we can obtain insight into general features of the metal binding site from GaPdx, we cannot assume that the binding site is unperturbed relative to that of the native protein. For this reason, we are continuing efforts to determine the structure of the native Pdx crystallographically. We have succeeded in obtaining a low-resolution crystal structure of reduced Pdx, and are currently refining this structure.

We are actively investigating structure/activity relationships in the Pdx-cytochrome P450 couple. We have constructed a model for the electron transfer complex between the two proteins based their observed behavior and their structures [14]. We are mutating some of the residues which are implicated by the model in protein-protein recognition, and will test these mutants in enzyme activity assays. We are also using mutagenesis to explore the structural requirements of metal binding in Pdx. One of the most exciting approaches that we have begun is to use new methods for NMR investigations of larger proteins (fractional and complete deuteration, 13C and 15N labeling, TROSY-based NMR methods) to begin sequential assignments for amide resonances in cytochrome P450cam. This will allow us to employ the methods of high-resolution solution state NMR to examine in detail structure-function relationships in cytochrome P450cam.

References:

1) X. M. Ye, T. C. Pochapsky and S. S. Pochapsky, "1H NMR Sequential Assignments and Identification of Secondary Structural Elements in Oxidized Putidaredoxin, an Electron Transfer Protein from Pseudomonas" Biochemistry 31, 1961-1968 (1992). Return to text

2) T.C. Pochapsky, X.M. Ye, G. Ratnaswamy and T.A. Lyons, "An NMR-Derived Model for the Solution Structure of Oxidized Putidaredoxin" Biochemistry 33, 6424-6432 (1994). Return to text

3) T. C. Pochapsky, M. Kuti, T. A. Lyons and J. Heymont, "A refined model for the solution structure of oxidized putidaredoxin" Biochemistry 38, 4681-4690 (1999). PDF version. Return to text

4) T.C. Pochapsky, G. Ratnaswamy and A. Patera, "Redox-Dependent 1H NMR Spectral Features and Tertiary Structural Constraints on the C-Terminal Region of Putidaredoxin" Biochemistry 33, 6433-6441 (1994). Return to text

5) T. A. Lyons, G. Ratnaswamy and T.C. Pochapsky, "Redox-Dependent Dynamics of Putidaredoxin Characterized by Amide Proton Exchange" Protein Science 5, 627-639 (1996). Return to text

6) H. Mo, S. S. Pochapsky and T. C. Pochapsky, "A model for the solution structure of oxidized terpredoxin, a Fe2S2 ferredoxin from Pseudomonas" Biochemistry 38, 5666-5675 (1999). PDF version. Return to text

7) M. Kostic, S. S. Pochapsky, J. Obenauer, H. Mo, G. Pagani, R. Pejchal and T. C. Pochapsky, "Comparison of functional domains in vertebrate-type ferredoxins", submitted. Return to text

8) T. C. Pochapsky, M. Kostic, N. Jain and R. Pejchal, "Redox-dependent conformational selection in a Cys4Fe2S2 ferredoxin" Biochemistry 40, 5602-5614 (2001). Return to text

9) N. Jain and T. C. Pochapsky, "Redox dependence of hyperfine-shifted 13C and 15N resonances in putidaredoxin" J. Am. Chem. Soc. 120, 12984-12985 (1998). Return to text

10) N. Jain and T. C. Pochapsky, "A new assignment strategy for the hyperfine shifted 13C and 15N resonances in Fe2S2 ferredoxins" Biochem. Biophys. Res. Commun. 258, pp. 54-59 (1999). Return to text

11) S. Kazanis, T.C. Pochapsky, T. M. Barnhart, J. E. Penner-Hahn, U. A. Mirza and B. T. Chait, "Conversion of a Fe2S2 Ferredoxin into Ga+3 Rubredoxin" J. Am. Chem. Soc. 117, 66256 (1995). Return to text

12) S. Kazanis and T. C. Pochapsky, "Structural features of the metal binding site and dynamics of gallium putidaredoxin, a diamagnetic derivative of a Fe2S2 ferredoxin" J. Biomol. NMR 9, 337-346 (1997). Return to text

13) M. Kuti, S. Kazanis and T. C. Pochapsky, "The solution structure of a gallium-substituted putidaredoxin mutant: GaPdx C85S" J. Biomol. NMR 12, 407-415 (1998). Return to text

14) T.C. Pochapsky, T.A. Lyons, S. Kazanis, T. Arakaki and G. Ratnaswamy, "A structure-based model for cytochrome P450cam-putidaredoxin interactions" Biochimie 78, 723-733 (1996). Return to text

Closed-shell ion pair structure and dynamics

Ion pairing is a phenomenon of considerable importance in chemical and biological systems, and has been studied by physical and spectroscopic methods for many years. NMR spectroscopy permit us to examine the time-average structures and dynamics of ion pairs. Of particular interest to us are the closed-shell ion pairs formed in non-polar solvents by large lipophilic cations such as quaternary ammonium ions and small hard anions such as chloride and tetrahydridoborate (BH4-). Such ion pairs are important in organic phase-transfer chemistry, and insight into their behavior will be used to design improved catalysts for phase-transfer reactions. Our observations of interionic NOEs in the tetrabutylammonium tetrahydridoborate (TBABH4) ion pair in CDCl3 showed that structural studies of ion pairs in non-polar media by NMR are feasible [1]. Temperature effects on the interionic NOE in TBABH4 and related ion pairs indicated that aggregation was probably taking place in these moderately concentrated solutions [2], and self-diffusion coefficients of tetrabutylammonium chloride relative to those measured for tetrabutylsilane in CDCl3 by pulsed field gradient NMR methods yielded an estimate of aggregate size as a function of temperature and concentration [3]. More recently, we have shown that the cation and anion diffuse at the same rate using borohydride anion diffusion as a probe [4]. The dynamics of the anion reorientation in the TBABH4 ion pair under a variety of conditions were characterized by 11B{1H}, 10B{1H} NOE and 11B, 10B relaxation measurements [5]. We are currently analyzing the results of 13C{1H} and 15N{1H} NOE and 13C,15N relaxation experiments in order to characterize the dynamics of the cation and overall aggregate motion in the TBABH4 ion pair [6]. Using selectively deuterated tetrabutylammonium salts, we are examining interionic NOEs in aggregates of tetrabutylammonium chloride in chloroform. Finally, we are continuing our investigation of complex ion pairs involved in asymmetric reductions of ketones in order to understand the mechanism of asymmetric induction. We will then apply this information to the design of more selective phase transfer catalysts (PTCs). The extension of NOE methodology to the study of more complex ion pairs with rotating-frame NOEs was accomplished using the ion pair derived from quinine by quaternarization with benzyl chloride ("benzylquininium chloride" or BQCl), followed by anion exchange with NaBH4 to yield "benzylquininium tetrahydridoborate" or BQBH4 [7]. This reagent has been shown to yield modest asymmetry in the reduction of prochiral ketones, and we have used the structural model to improve the asymmetric induction by rational catalyst redesign [8].

We have found further practical application of our model by using it to modify the reaction conditions under which asymmetric PTC reductions are performed. We have achieved an enantiomeric excess (ee) of 86% in the PTC reduction of a prochiral ketone that under standard PTC conditions yields only a 30% ee (Pochapsky and Hofstetter, patent pending). Finally, we have developed methods for examining ion pair structures using 19F{1H} NOEs [9].

References:

1) T. C. Pochapsky and P. M. Stone, "A Study of Ion Pair Solution Structure Using Nuclear Overhauser Effects: Interionic 1H{1H} and 11B{1H} NOEs in the (n-Butyl)4N+,BH4- Ion Pair" J. Am. Chem. Soc. 112, 6714-6715 (1990). Return to text

2) P. M. Stone, T. C. Pochapsky and E. Callegari, "The Sign of the Nuclear Overhauser Effect as a Function of Temperature in Contact Ion Pairs" J. Chem. Soc., Chem. Comm. 178-179 (1992). Return to text

3) S. Pochapsky, H. Mo and T.C. Pochapsky, "Closed-Shell Ion Pair Aggregation in Non-Polar Solvents Characterized By NMR Diffusion Measurements" J. Chem. Soc., Chem. Comm. 2513 (1995). Return to text

4) H. Mo and T. C. Pochapsky, "Self-diffusion coefficients of paired ions" J. Phys. Chem. B 101, 4485-4486 (1997). PDF version,Return to text

5) T. C. Pochapsky, A.-P. Wang and P.M. Stone "Closed-Shell Ion Pair Structure and Dynamics: Steady State 1H{1H}, 10B{1H} and 11B{1H} Nuclear Overhauser Effects and 10B, 11B Nuclear Relaxation of Tetraalkylammonium Tetrahydridoborates in Ion-Pairing and Dissociative Solvents" J. Am. Chem. Soc. 115, 11084-11091 (1993). Return to text

6) H. Mo, A. Wang, P. Stone Wilkinson and T. C. Pochapsky, "Closed-shell ion pairs: Cation and aggregate dynamics of tetraalkylammonium salt in an ion-pairing solvent" J. Am. Chem. Soc. 118, 11666-11673 (1997). Return to text

7) T.C. Pochapsky, P.M. Stone and S.S. Pochapsky, "Interionic Contacts in Complex Ion Pairs Detected by Rotating Frame Nuclear Overhauser Effects" J. Am. Chem. Soc. 113, 1460-1462 (1991). Return to text

8) C. Hofstetter, P.M. Wilkinson and T. C. Pochapsky, "NMR structure determination of ion pairs derived from quinine: A model for templating in asymmetric phase transfer reductions by BH4- with implications for rational design of phase transfer catalysts" J. Org. Chem. 64, 8794-8800 (1999). PDF version. Return to text

9) C. Hofstetter and T. C. Pochapsky, "BF4- as a probe for ion pair solution structure using interionic one- and two-dimensional 19F{1H} NOEs" Mag. Reson. Chem. 38, 90-94 (2000). Return to text

Amino acid interaction free energies for protein folding simulations: experimental measurements and theoretical considerations

The protein folding problem remains one of the major unsolved challenges of structural biology. Although much has been learned experimentally concerning the later stages of folding, the initial collapse of the polypeptide chain to a compact state is still poorly understood. This is a critical step, since much of the search space is excluded at the point of collapse, and any algorithm which successfully folds a protein based on sequence information must be able to select those compact states which are most likely to form. The most important component of such a search is information concerning the interaction free energies of amino acid side chains, and for this, a set of pairwise interaction potentials is required. Many such potential tables have been devised, often from statistics concerning interaction frequencies from the protein structure data base [1]. However, it has been shown recently that such potentials are likely to contain systematic errors in estimates of side chain interaction free energies, and as such may not be suitable for realistic folding simulations [2].

Our approach to this problem is two-fold. Experimentally, we are attempting to measure relative interaction free energies using affinity chromatography. We covalently bind functionality identical to a particular amino acid side chain to silica gel, and use this bonded phase as a stationary phase for liquid chromatography. Separation factors for amino acid derivatives in aqueous media on this support are then related to the relative free energies of interaction between the analytes and the bound side chain. To date, we have prepared phenylalanine and leucine side chain phases [3]. We have established that aromatic-aromatic and aromatic-aliphatic interactions are qualitatively different than aliphatic-aliphatic interactions, both in magnitude and in the nature of their temperature dependence. Our work suggests that aromatic amino acid side chains play a critical role in the initiation of the collapse of the polypeptide chain, and are likely to also be important in mediating protein-protein and protein-substrate interactions [4]. Besides experimental measurements, we also use lattice model Monte Carlo simulations to determine the minimal requirements of a potential for use in folding simulations. Our recent publications discuss the accuracy required for such a potential and test our theoretical predictions against a well-characterized lattice model for folding [5, 6].

References:

1) S. Miyazawa and R. L. Jernigan, Macromolecules 18, 534 (1985). Return to text

2) P. D. Thomas and K.A. Dill, J. Mol. Biol. 257, 457 (1996). Return to text

3) T. C. Pochapsky and Q. Gopen, "A Chromatographic Approach to the Determination of Relative Free Energies of Interaction Between Hydrophobic and Amphiphilic Amino Acid Side Chains" Protein Science 1, 786-795 (1992). Return to text

4) A. F. Pereira de Araujo, B. Joughin and T. Pochapsky, "Thermodynamics of interactions between amino acid side chains: Experimental differentiation of aromatic-aromatic, aromatic-aliphatic and aliphatic-aliphatic side chain interactions in water" Biophys. J. 76, 2319-2328 (1999). Return to text

5) A. F. Pereira de Araujo and T. C. Pochapsky, "Monte Carlo Simulations of Protein Folding Using Inexact Potentials: How Accurate Must Parameters Be In Order to Preserve Essential Features of the Energy Landscape?" Folding and Design 1, 299-314 (1996). postscript version, pdf version. Return to text

6) A. F. Pereira de Araujo and T. C. Pochapsky, "Estimates for the Potential Accuracy Required in Realistic Folding Simulations and Structure Recognition Experiments" Folding and Design 2, 135-139 (1997). postscript version, dvi version, figure 1 in postscript Return to text