Cosubstrate Binding Site of Pseudomonas sp, AK1 y-Butyrobetaine

‘PHE J O U R N A L OF BIOLOOICAL CHEMISTRY (L: 1991 by The American Society for Biochemistry and Molecular Btology, Inc.

Val. 266, No. 3. Issue of danuary 25, pp. 1526-1533.1991 Printed iri U. S. A.

Cosubstrate Binding Site of Pseudomonas sp, AK1 y-Butyrobetaine Hydroxylase INTERACTIONS WITH STRUCTURAL ANALOGS OF a-KETOGLUTARATE*

(Received for publication, February 15, 1990)

Sze-Fong NgS, Hartmut M. Hanauske-Abel9, and Sasha Englardll From the Department of Biochemistry, Albert Einstein College of Medicine, Bronx, New York 10461

Forty-one aromatic and aliphatic analogs of a-ketoglutarate were studied kinetically for their interaction with the a-ketoglutarate binding site of y-butyrobetaine hydroxylase obtained from Pseudomonas sp. AK1. Together, the compounds represent structural permutations probing the contribution of: 1) the C 5 carboxyl group of a-ketoglutarate (domain I); 2) the C1-C2 keto acid moiety of a-ketoglutarate (domain 11); 3) the distance between domains I and 11; and 4) the spatial relationship of the two domains required for optimal interaction with the cosubstrate binding site. All compounds were competitive inhibitors for a-ketoglutarate ( K , 0,018 mM). Functionally, two subsites of the cosubstrate binding site were evident: subsite I for polar interaction with the C5 carboxyl group, and subsite 11, comprising of two distinct cis-oriented coordination sites of the catalytic ferrous ion which interact with the Cl-C2 keto acid moiety. The most efficient inhibitors were pyridine 2,4-dicarboxylate (Ki 0.0002 mM) and 3,4-dihydroxybenzoate (Ki 0.0006 mM). Both compounds contain a carboxyl group and a chelating moiety corresponding to domains I and I1 of a-ketoglutarate, respectively. The fixed orientation of these groups in both analogs was used to assess intersubsite distance and spatial relationship required for optimal interaction with the cosubstrate binding site. Binding at subsite I and chelation at subsite I1 were indispensible for effective competitive inhibition. The distance between these two domains also helped determine whether attachment at the cosubstrate binding site would be catalytically productive. This was em- phasized by the failure of either oxaloacetate or a- ketoadipinate to promote hydroxylation. Optimal interdomain distance, however, was not sufficient for cosubstrate utilization, as pyridine 2,4-dicarboxylate, with an interdomain distance identical to a-ketoglutarate in its staggered conformation, did not sustain hydroxylation. In the overall, these studies suggest that a-ketoglutarate utilization occurs in a ligand reaction at the active site ferrous ion of y-butyrobetaine hydroxylase. This is of particular interest since the delin-

* This work was supported by United States Public Health Service Grant 5ROI AM-21194 and Grant BC-642 from the American Cancer Society. The costs of publication of this article were defrayed in part hy the payment of page charges. This article must therefore he hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

$ Present address: Dept. of Biological Chemistry and Molecular Pharmacology and Dana Farber Cancer Institute, Harvard Medical School, Boston, MA 02115.

$ Present address: Dept. of Pediatrics, Rush Medical Center, Chi- cago, IL 60612.

ll To whom correspondence should be addressed. Tel.: 212-430- :3031; Fax: 212-892-0703.

eated stereochemical mode of oxidative decarboxylation could generate the reactive oxo-iron species that was shown experimentally to promote 7-butyrobetaine hydroxylation by an abstraction-recombination mechanism (Blanchard, J. S., and Englard, S. (1983) Bio- chemistry 22, 5922-5928; Englard, s., Blanchard, J. S., andMidelfort, C. F. (1985) Biochernistry24,lllO- 1116).

~~

Carnitine functions as the carrier molecule in the transport of medium and long chain fatty acids across the inner mito- chondrial membrane, and therefore is important in thermo- genesis and ketogenesis that depend on the process of @- oxidation (1-3). Carnitine may also participate in shuttling acyl residues out of the peroxisomes, the site of shortening of very long chain fatty acids (3, 4).

The biosynthetic pathway of carnitine is remarkable in that it includes two molecular oxygen-utilizing hydroxylations at aIiphatic carbon centers, each mediated by a ferrous ion- dependent enzyme that requires a-ketoglutarate as a cosubstrate and ascorbate as a reductant (3). These enzymes are similar in their cosubstrate and cofactor requirements to the three collagen hydroxylases (1,3). The enzymes are classified as a-ketoglutarate-dependent, ascorbate-linked, non-heme ferrous ion dioxygenases. Of these, prolyl 4-hydroxylase (EC 1.14.11.2) and y-butyrobetaine hydroxylase (EC 1.14.11.1) are the most extensively studied (1).

y-Rutyrobetaine hydroxylase catalyzes the terminal step in carnitine biosynthesis, the hydroxylation of 4-N-trimethylaminobutyrate (2, 3); it stoichiometrically consumes molecular oxygen and a-ketoglutarate. As with prolyl4-hydroxylase (5) , a-ketoglutarate decarboxylation can be uncoupled from substrate hydroxylation and may exceed it significantly under certain conditions (6, 7). While the hydroxylation phase of the catalytic cycle of prolyl 4-hydroxylase is yet to be probed more deeply (8), the events during the hydroxylation phase of y-butyrobetaine hydroxylase have been extensively studied (9, 10). The results indicate that an active site iron-oxo complex stereospecifically abstracts, by homolytic C-H bond scission, the pro-R hydrogen atom at C3 of the substrate. This leads to intermediate formation of a planar, sp’-hybridized carbon radical and an iron-hydroxyl radical complex (9, 10). It has been suggested that these two topographically fixed odd electron species rapidly recombine with retention of configuration at the carbon atom, giving the stereospecifically hydroxylated product of the catalytic cycle, R-carnitine, and regenerating the enzyme-bound ferrous ion (9, 10).

The hydroxylation phase during the catalytic cycle of prolyl 4-hydroxylase also has been considered to occur via a radical abstraction-recombination mechanism (11, 12). In this sug-

1526

a-Ketoglutarate Binding Site of y-Butyrobetaine Hydroxylase 1527

gested mechanism, the formation of the hydroxylating species was also addressed. The crucial cosubstrate decarboxylation step was postulated to occur in a ligand reaction at the enzyme-bound ferrous ion that would involve an SN2t-like interaction between the end-on coordinated dioxygen and the chelating C1-C2 keto acid moiety of a-ketoglutarate, charac- terized by sp2-sp3-sp2 transition of its C2 atom. In that way, carbon dioxide, succinate, and the hydroxylating oxygen atom species should be generated inside the coordination sphere of the enzyme-bound metal ion, without any stoichiometric reductant requirement or any immediate catalytic contribution of protein side chains (11, 12). This hypothesis was crucial to the successful design of several mechanism-derived inhibitors of prolyl4-hydroxylase (11-17).

The decarboxylation phase of the catalytic cycle of y- butyrobetaine hydroxylase has not been addressed experimentally. The purpose of the present work was to study the effects of a series of aliphatic and aromatic antagonists of utilization of a-ketoglutarate by y-butyrobetaine hydroxylase of Pseu- domonas sp. AK1. These compounds were used as reporter molecules to help map the enzyme's cosubstrate binding site and to delineate the early events of the catalytic cycle that produce the required oxo-iron complex. The findings reveal significant differences between the organization of the a- ketoglutarate binding site of the bacterial y-butyrobetaine hydroxylase and that of the chicken prolyl 4-hydroxylase. Those differences occur although both enzymes show similar K,,, values for a-ketoglutarate and both require metal chelation and polar interaction for optimal cosubstrate binding.

EXPERIMENTAL PROCEDURES

Materials-Bovine liver catalase (65,000 units/mg) was obtained from Boehringer Mannheim. a-Keto-[1-"CJglutarate (sodium salt, 59.4 mCi/mmol) was purchased from Du Pont-New England Nuclear. The 0.1 M a-ketoglutarate solution (in 0.1 M phosphate buffer, pH 7.5), used to dilute the high specific activity cosubstrate, was from Sigma. y-Butyrobetaine (4-N-trimethylaminobutyrate), listed in the supplier's catalog as (3-~arboxypropyI)-trimethyl ammonium chloride, was obtained from Aldrich. Hydrofluor liquid scintillation count- ing solution was obtained from National Diagnostics. The compounds tested as inhibitors of y-butyrobetaine hydroxylase and listed in Fig. 1 and Table I were of the highest purity commercially available and obtained as follows: compounds 1, 2, 5-11, 13, and 34 from Sigma; compounds 3, 12, 14-17, 19-24, 28-33, 35, and 38-41 from Aldrich compounds 4,18,25-27,36, and 37 from Fluka.

Enzyme Purification-y-Butyrobetaine hydroxylase was isolated and purified from Pseudomonas sp. AKI by a newly devised rapid and convenient procedure (I@.' Briefly, Pseudomonas sp. AK1 cells, as a 20% suspension (wet weight/volume) in buffer A (10 mM potassium phosphate buffer, pH 6.5, containing 0.05 mM dithiothreitol, 0.5 mM EDTA, 10% glycerol) were disrupted in a French press. The broken cell suspension was centrifuged and the supernatant adjusted to 2% (w/v) with streptomycin sulfate. The clear supernate obtained by centrifugation was fractionated with ammonium sulfate to 30% saturation and centrifuged. The supernatant obtained was adjusted to 20% saturation by addition of buffer A and applied to a 10 X 120- mm column of octyl-Sepharose CL-4B (Pharmacia LKB Biotechnol- ogy Inc.) preequilibrated with buffer B (10 mM potassium phosphate, pH 6.5, containing 0.05 mM dithiothreitol, 0.5 mM EDTA, 10% glycerol, and ammonium sulfate to 20% saturation). After extensive washing of the column with buffer B, the chromatogram was devel- oped with a linear gradient made with 100 ml each of buffer B and buffer A. Pooled fractions containing the enzyme activity were equilibrated with buffer C (50 mM Tris-HC1, pH 7.8, 0.05 mM dithiothreitol, 0.5 mM EDTA, 10% glycerol) and concentrated to 10 ml by ultrafiltration through a PM-10 membrane. The concentratedprotein fraction was then applied to a 5 X 55-mm Mono Q column preequilibrated with buffer D (10 mM triethanolamine HC1, pH 7.8). After washing, the column was eluted with a linear gradient of potassium chloride (0-1 M in buffer D). The fractions containing

' S.-F. Ng and S. Englard, manuscript in preparation.

enzyme activity were pooled and the sample was then applied, in aliquots of 500 pl , to a TSK-3000 SW gel filtration HPLC' column (7.5 x 600 mm) pre-equilibrated with buffer A. The collected fractions containing enzyme activity were pooled and used for the inhibitor studies. The preparation was homogeneous as judged from sodium dodecyl sulfate-polyacrylamide gel electrophoresis analysis. As determined by a-ket~-[l-'~C]gIutarate decarboxylation, the preparation showed a specific activity of 25 pnol of carbon dioxide produced min" mg". That value compares favorably with a previously reported specific activity of 22 pmol of carbon dioxide produced min-I mg" reported for Pseudomonas sp. AK1 y-butyrobetaine hydroxylase purified by a different procedure (19).

Enzyme Assays-Stock solutions of y-butyrobetaine hydrochloride and of the various inhibitors were carefully neutralized to pH 7.0. Solutions of ascorbate, ferrous ammonium sulfate, and inhibitors were prepared freshly and stored on ice until use.

All assays for y-butyrobetaine hydroxylase activity were carried out in 16 X 100-mm test tubes sealed with rubber stoppers holding filter paper strips (1.5 X 3 cm) that were impregnated with 200 el of 0.1 N NaOH. The standard 0.7-1111 incubation mixtures contained: 15 mM potassium phosphate buffer, pH 7.4, 14 mM sodium ascorbate, 0.2 mM a-ket~-[l-'~C]glutarate (0.23 pcilpmol), 1 mg of catalase, 29 mM y-butyrobetaine, 0.6 mM ferrous ammonium sulfate, and 0.05 pg of the purified enzyme. Following incubation with moderate shaking at 37 "C for 10 min, the reactions were terminated by injecting 1 ml of 10% (w/v) trichloroacetic acid into the stoppered test tubes, and these were incubated at 37 "C for 90 min to allow maximal trapping of released carbon dioxide. The paper strips were then transferred to vials containing 5 ml of Hydrofluor and counted in an LKB Rackbeta model 1219 liquid scintillation counter.

In the assay for y-butyrobetaine hydroxylase activity based on the release of ["CC]CO, from a-keto-[ l-'4C]glutarate, essentially identical K , values for a-ketoglutarate were calculated from data obtained in separate experiments in which the cosubstrate concentration was varied at either a constant or decreasing specific radioactivity, the latter being achieved by varying over a 5-fold range the ratio between unlabeled and l-'4C-labeled a-ketoglutarate. For all of the structural analogs examined, inhibitory activities were determined with variable concentrations of a-ketoglutarate at three different concentrations of the compounds. Linearity of the time course of the reaction up to 20 min under the defined conditions of assay was determined and established for compounds 3, 6, 8, 12-16, 18-26, and 31-33 at the highest concentration used in the determination of their respective inhibition constants. These were representative of the 41 aliphatic and aromatic compounds listed in Table I, with K, values ranging from 0.0002 mM determined for pyridine 2,4-dicarboxylate to 16.9 mM determined for benzene 1,2-dicarboxylate. A t least three different enzyme preparations were used in the course of the present study, and with each nearly identical kinetic constants were obtained as determined for a-ketoglutarate and compounds 12, 14, 25, and 26. For data analysis, reciprocal velocities were plotted against the reciprocal of the variable a-ketoglutarate concentration, and the data were fitted to the rate equations for competitive, noncompetitive and uncompetitive inhibition by the least squares method using the For- tran programs of Cleland (20). Evaluation of the results on the basis of the enumerated criteria for selection of the most appropriate fit of the data (20) revealed that all of the inhibitors listed in Table I were linearly competitive with respect to a-ketoglutarate.

To study the effect on the overall enzyme reaction of a-ketoglutarate replacement by either a-ketoadipinate, oxaloacetate, or pyridine 2,4-dicarboxylate, formation of [ methyl-'4C]carnitine under standard conditions was determined using [ methyl-'4C]y-butyrobetaine (0.09 pCi/pmol) as substrate. !metl~yl-'~C]Carnitine was separated from [methyl-'4C]r-butyrobetaine by HPLC, using a Varian model 5500 system equipped with a Varian MicroPak amino acid hydrolysate column.

RESULTS

Inhibition of y-Butyrobetaine Hydroxylase by Aliphatic StructuralAnulogs of a-Ketoglutarate-The importance of the three functional groups of cy-ketoglutarate, and of the length of the aliphatic carbon bridge joining the C5 carboxyl and the Cl-CZ keto acid moieties, was investigated with a series of 11

' The abbreviation used is: HPLC, high performance liquid chro- matography.

1528 a-Ketoglutarate Binding Site of y-Butyrobetaine Hydroxylase

analogs in which these structural regions were selectively altered (Fig. 1, compounds 1-11). The extent and type of inhibition were determined in a competitive assay system with a-ketoglutarate as the variable substrate, by measurement of the hydroxylation-coupled release of I4CO, from a- keto-[l-"C]glutarate. The K, for a-ketoglutarate under these conditions of assay for y-butyrobetaine hydroxylase was determined to be 0.018 mM.

All of the compounds tested were found to inhibit the enzyme competitively with respect to a-ketoglutarate; this is shown for pyruvate in Fig. 2. The Kt values ranged from 0.07 to 13.1 mM (Table I). Succinate and oxaloacetate were the most effective a-ketoglutarate antagonists, with K; values of 0.07 and 0.12 mM, respectively.

The C5 carboxyl group of a-ketoglutarate, domain I (circled in the first structure depicted in Fig. l), appeared to be crucial for attachment to the cosubstrate binding site of the enzyme, probably through polar interaction. This conclusion is based on the observation that elimination of this negatively charged group, as in a-ketobutyrate, or its replacement with a methyl moiety, as in a-ketovalerate, resulted in an increase of the K; value to 1.3 mM. Despite the differences in chain length, pyruvate, a-ketobutyrate, and a-ketovalerate inhibited the enzyme with very similar K, values, suggesting that the interaction with the aliphatic region does not contribute significantly to cosubstrate binding site attachment.

The a-keto acid moiety, domain I1 (surrounded by a rectan- gle in the first structure depicted in Fig. l), was required for efficient interaction with the cosubstrate binding site. The importance of this structure can be deduced from the K; values determined for compounds with appropriate modification. Thus, substitution of the C2 carbonyl by a methylene group, as in glutarate, or shift of the keto moiety to the C3 position, as in 3-ketoglutarate, resulted in K; values of 0.32 and 0.29 mM, respectively. Replacement of the C1 carboxyl with a methyl group, as examplified by levulinate, had an even stronger effect, giving a K, of 13.1 mM.

The distance between these two functional domains, defined in a-ketoglutarate by the C2 and the C5 carbon centers (or in its analogs, by the corresponding atoms), was identified as another crucial parameter for attachment to the enzyme's cosubstrate binding site. An aliphatic bridge of two methylene groups, as in a-ketoglutarate, appeared to be optimal. Short- ening to one methylene group, as in oxaloacetate, or extension to three methylene groups, as in a-ketoadipinate, resulted in increased K, values of 0.12 and 0.17 mM, respectively. The cosubstrate binding site generated product of the enzyme's catalytic cycle, succinate, was found to have a Ki of 0.07 mM, the lowest value of all aliphatic compounds tested. This result reflects the requirement for an a-ketoglutarate-like interdomain distance of two methylene groups for optimal attachment, although the domain I1 equivalent structure of succinate can only coordinate, not chelate to the active site ferrous ion. This optimal interdomain distance of two methylene groups does not, however, define a specific steric arrangement and therefore does not allow an accurate numeric measurement.

The interdomain distance also determined whether the attachment to the cosubstrate binding site occurred in a manner that would allow utilization of an a-keto acid moiety (domain 11) for productive contribution to the overall enzyme reaction. Thus, neither oxaloacetate nor a-ketoadipinate (up to 10 times their Kt values) promoted the hydroxylation of [methyl-14C]y-butyrobetaine to [rnethyl-'4C]carnitine under conditions in which as little as 0.1% conversion of substrate t o product would have been detected (data not shown).

Inhibition of -y-Butyrobetaine Hydroxylase by Pyridine Di- carboxylates and Related Structural Analogs-All pyridine di- carboxylates and related compounds tested were found to inhibit the enzyme competitively with respect to a-ketoglutarate, this is shown for pyridine 2,4-dicarboxylate in Fig. 3. The determined Kt value of 0.0002 mM for this particular pyridine dicarboxylate was 350-fold lower than the Kt value obtained for the most efficient aliphatic inhibitor succinate, and a 100-fold lower than the K, value of 0.018 mM determined for a-ketoglutarate.

Systematic structural variations of the pyridine dicarboxy- lates again demonstrated the importance of both polar interaction and metal ion chelation for attachment to the a - ketoglutarate binding site. The pyridine 2-carboxylate structure is known to interact with ferrous ions by acting as a bidentate, i.e. chelating ligand (21-23, 32, 33). Compounds in which this specific structural feature was altered, showed increases in K; values. Disruption of the chelating moiety is achieved by replacement of the crucial C2 carboxyl group with a hydrogen atom, by a shift of this carboxyl group into position 3, or by substitution of the aromatic ring nitrogen with a carbon atom. For pyridine 2,4-dicarboxylate (Ki 0.0002 mM), those modifications result in structures that correspond to pyridine 4-carboxylate, pyridine 3,4-dicarboxylate, and benzene 1,3-dicarboxylate, respectively. In that order, these compounds yielded K, values of 0.08, 0.74, and 3.8 mM. For pyridine 2,5-dicarboxylate (Ki 0.014 mM), analogous structural modifications result in structures that correspond to pyridine 3-carboxylate, pyridine 3,5-dicarboxylate, and benzene 1,4-dicarboxylate; these exhibited K; values of 0.41, 1.03, and 7.6 mM, respectively.

While preserving the pyridine 2-carboxylate structure, in- troduction of a second carboxyl group at the various available ring positions significantly affected inhibitory potency. These changes also revealed the optimal spatial orientation relative to the metal binding moiety required for ionic interaction at the cosubstrate binding site. Comparison of the K; values of pyridine 2-carboxylate (K; of 0.039 mM) with those of the 2, 3-, 2,4-, 2,5-, and the 2,6-isomers ( K ; values of 0.15, 0.0002, 0.014, and 0.29 mM, respectively) indicated that the second carboxyl group has to be in either ring position 4 or 5 to produce a K; value that is lower than that of the monocarbox- ylate. The most effective inhibitory interaction with the enzyme occurred when the second carboxyl group was located at ring position 4. The preferential binding of compounds with a carboxyl group in that spatial orientation occurred whether or not the mode of interaction with the metal ion was mono- or bidentate. This is evident by comparison of the K, values of pyridine 4-carboxylate, pyridine 3,4-dicarboxylate, and benzene 1,3-dicarboxylate with those of pyridine 3- carboxylate, pyridine 3,5-dicarboxylate, and benzene 1,4-dicarboxylate, respectively.

These experiments demonstrated that pyridine 2,4-dicarboxylate was the most effective competitive inhibitor of '%02 release from a-keto-[1-"CC]glutarate. The interdomain distance of this rigid structure is about 375 pm, and thus is virtually identical to that of a-ketoglutarate in its staggered conformation.

The inhibition of a-keto-[ 1-"CJglutarate decarboxylation by pyridine 2,4-dicarboxylate in the competitive assay, however, does not rule out its utilization (instead of a-ketoglutarate) by the enzyme, and does not provide unequivocal evidence for inhibition of substrate hydroxylation. Catalytic utilization of this a-ketoglutarate analog as a highly effective alternative cosubstrate would yield results identical to those reported here. To determine whether pyridine 2,4-dicarboxylate indeed


I i o

3 4 5

6

0 0

1 6 0

1 s

o+?"o - 0

2 0 0 0

1 7 1 8 0

1 9

2 3 2 4 2 1 22

YO

2 9 3 0 2 8

UO I

10 no NO

3 4 3 s 3 6 3 7 3 1 3 2 3 3

3 8 "0 YO

3 9 4 0 4 1

FIG. 1. Structural formulae for the aliphatic and aromatic compounds used in this study. The names and K , values of all of these substances are given in Table I. The carboxyl groups are drawn symmetrically to indicate their anionic form under the pH conditions ofthe assay procedure. Arrows indicate iron chelating capacity and are shown for wketoglutarate and compounds 12 and 25 as representatives of the aliphatic, pyridinecarboxylate, and dihydroxyhenzoate series, respectively. Compounds are grouped to emphasize structural homology, such as compounds 1-6 or 12-16, t.hat allowed probing of specific aspects of the cosubstrate binding site; those aspects include the requirement for and the relative orientation ofthe anionic and the chelating moieties of cu-ketoglutarate, i t? . domains I and 11, respectively. For wketoglutarate, domain I is indicated hy the circle, and domain I1 by the rc'ctangle.


30 -

20 -

10 -

Y n -

25 50 75 1 1 a-ketogtut.rate 1 1 m n .’)

FIG. 2. Double reciprocal plots of the inhibition of y-butyrobetaine hydroxylase by pyruvate, with &-ketoglutarate as the variable substrate. Concentration of inhibitor: 0 (X), 3.5 (A), 7.0 (O), and 10.5 (0) mM. The points are the experimental values and the lines represent fitted values of the experimental data to the equation describing linear competitive inhibition.

can substitute for a-ketoglutarate in the overall reaction, we measured the formation of [n~ethyl-’~C]carnitine from [ n~ethyl-’~C]y-butyrobetaine in the presence of pyridine 2,4- dicarboxylate alone. Although its cosubstrate binding site affinity is several orders of magnitude higher than that of a- ketoglutarate (Table I), pyridine 2,4-dicarboxylate, at concentrations up to 10 times its Ki value, did not promote detectable hydroxylation of y-butyrobetaine (data not shown)). Thus, the suppression of I4CO2 release from a-ket~-[l-’~C]glutarate did not reflect the favored utilization of pyridine 2,4-dicarboxylate instead of a-ket~-[l-’~C]glutarate, but resulted from the inertness of the inhibitor’s domain I1 toward enzymatic decarboxylation. Remarkably, that same substructure is very sensitive to nonenzymatic decarboxylation, and pyridine 2,4- dicarboxylate preferentially decarboxylates to form pyridine 4-carboxylate (24).

Inhibition of y-Butyrobetaine Hydroxylase by Dihydroxy- benzene Derivatives and Related Structural Analogs-All the hydroxyphenyl derivatives were found to be inhibitors of the enzyme with respect to a-ketoglutarate. This is shown for 3,4- dihydroxybenzoate in Fig. 4. The effect of this class of inhibitors on ascorbate binding will be reported separately.

The importance of having a carboxyl group in proper orientation to the two vicinal, chelating hydroxyl groups (23,25, 32, 33) was again evident. 3,4-Dihydroxybenzoate, displaying the structural prerequisites for both ionic interaction and for metal chelation, and an interdomain distance of about 375 pm, was the most potent inhibitor in this class of compounds. Its Ki value of 0.0006 mM was 6 one-thousandths the Ki value of the 2,3-isomer and 3 one-hundredths the K,,, value of a- ketoglutarate. Omission of the carboxyl function (domain I), as exemplified by 1,2-dihydroxybenzene, resulted in a KC value increased by a factor of 200. Disruption of the vicinal hydroxyl arrangement (domain 11), as in 3- and 4-hydroxybenzoate, raised the Ki values by a factor of about 30 and 850, respectively.

The distance between the carboxyl group and the chelating dihydroxyphenyl moiety of 3,4-dihydroxybenzoate affected inhibitor potency, re-emphasizing the significance of the interdomain distance as established with the aliphatic analogs and the series of pyridine carboxylates. A stepwise increase of this distance by one and by two aliphatic carbon atoms, in 3&dihydroxyphenylacetate and in 3,4-dihydroxyphenylpro-

TABLE I Inhibition constants for the aliphatic and aromatic compounds tested

in the reaction of y-butyrobetaine hydroxylase In all cases, inhibition was competitive with respect to a-ketoglu-

tarate ( K , for a-ketoglutarate = 0.018 4 0.001). The K, values (2 S.E.) were determined as described under “Experimental Procedures.”

No. in Fig. 1 Compound“ K,

1. 2. 3. 4. 5. 6. 7. 8. 9.

10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41.

a-Ketoadipinate Oxaloacetate a-Ketovalerate a-Ketobutyrate Pyruvate Glutarate Adipinate Succinate 3-Ketoglutarate Levulinate Malonate Pyridine 2-carboxylate Pyridine 2,3-dicarboxylate Pyridine 2,4-dicarboxylate Pyridine 2,5-dicarboxylate Pyridine 2,6-dicarboxylate Benzoate Benzene 1,2-dicarboxylate Benzene 1,3-dicarboxylate Benzene l,4-dicarboxylate Pyridine 4-carboxylate Pyridine 3-carboxylate Pyridine 3,4-dicarboxylate Pyridine 3,h-dicarboxylate 3,4-Dihydroxybenzoate 2,3-Dihydroxybenzoate 2,4-Dihydroxybenzoate 2-Hydroxybenzoate 3-Hydroxybenzoate 4-Hydroxybenzoate 2,5-Dihydroxybenzoate 3,5-Dihydroxybenzoate 2,6-Dihydroxybenzoate 3,4,5-Trihydroxybenzoate 1,2-Dihydroxybenzene 1,3-Dihydroxybenzene 1,4-Dihydroxybenzene 3,4-Dihydroxyphenylacetate 3,4-Dihydroxymandelate 3,4-Dihydroxyphenylpropior 3,4-Dihydroxvcinnamate

mM 0.17 f 0.02 0.12 f 0.02

1.3 f 0.3 1.3 f 0.2 1.4 f 0.3

0.32 f 0.06 0.15 f 0.01 0.07 -t 0.01 0.29 -t 0.05 13.1 ? 1.9 7.0 ? 1.7

0.039 ? 0.009 0.15 f 0.03

0.0002 f 0.0000 0.014 f 0.002 0.29 ? 0.07 2.6 f 0.1

16.9 f 2.2 3.8 f 0.4 7.6 ? 0.9

0.08 k 0.02 0.41 0.08 0.74 f 0.08 1.03 f 0.18

0.0006 ? 0.0001 0.10 2 0.02 1.08 ? 0.14 1.32 ? 0.27 0.02 k 0.00 0.51 ? 0.11 1.23 ? 0.14 0.05 f 0.01 0.26 f 0.04 0.19 0.02 0.12 f 0.02 13.2 f 2.2 11.7 ? 2.0

0.013 ? 0.001 0.028 f 0.006

late 0.020 f 0.004 0.006 0.001

For structural formulae, see Fig. 1.

pionate, caused approximately 20- and 30-fold increases in Ki values, respectively.

DISCUSSION

The oxygen atom transfer step of y-butyrobetaine hydroxylase purified from Pseudomonas sp. AK1 and calf liver was found to proceed stereospecifically with retention of configuration (9, 10). An iron-oxo complex was suggested as the hydroxylating agent, with its oxygen atom located in immediate proximity to the pro-R hydrogen atom at C3 of y- butyrobetaine. The radical mechanism proposed on the basis of the kinetic isotope effects, the stereospecific course of the reaction and the retention of configuration at the substrate methylene group imply topological constraints during the hydroxylation of y-butyrobetaine which effectively shield the C3 pro-S hydrogen atom from the oxidant. In view of these studies, the oxygen atom transfer step during the catalytic cycle of y-butyrobetaine hydroxylase can be better delineated than that of other a-ketoglutarate-linked, ascorbate-dependent ferrous iron dioxygenases.


25 50 75

1 I I o-keloalularate 1 ( mM .’) FIG. 3. Double reciprocal plots of the inhibition of y-buty-

robetaine hydroxylase by pyridine 2,4-dicarboxylate, with a- ketoglutarate as the variable substrate. Concentration of inhibitor: 0 (X ) , 0.35 (A), 0.70 (O), and 1.4 (0) p ~ . The points are the experimental values and the lines represent fitted values of the experimental data to the equation describing linear competitive inhibition.

25 50 75

1 I [ a.kaIoglulsrale ] ( mM ” )

FIG. 4. Double reciprocal plots of the inhibition of y-butyrobetaine hydroxylase by 3,4-dihydroxybenzoate, with a-ketoglutarate as the variable substrate. Concentration of inhihibor: 0 ( X ) , 0.35 (A), 0.70 (O), and 1.4 (0) pM. The points are the experimental values and the lines represent fitted values of the experimental data to t.he equation describing linear competitive inhibition.

The decarboxylation step of the catalytic cycle of y-butyrobetaine hydroxylase purified from Pseudomonas sp. AK1 and the structure of its cosubstrate binding site were investigated in the present paper. By determining the K , values of systematically modified aliphatic and aromatic compounds in an assay system that measured their competition with a- ketoglutarate, we obtained strong evidence that the cosubstrate binding site of the enzyme engages in two distinct modes of interaction with compounds having structural fea- tures related to a-ketoglutarate: polar interaction at subsite I and metal chelation at subsite I1 (Fig. 5a). This functional differentiation of the cosubstrate binding site can account for the attachment of such different compounds as a-ketoglutarate (Fig. 56), pyridine 2,4-dicarboxylate (Fig. 5c), and 3,4- dihydroxybenzoate (Fig. 5d). Accordingly, the cosubstrate structures allowing these interactions are the polar domain I and the metal ion binding domain 11.”

:’ As defined in Ref. 11, the term “domain” as used here applies to the cosubstrate or its analogs, while “subsite” refers to the cosubstrate binding site of the enzyme; e.g. in case of a-ketoglutarate, domain I (its C 5 carboxyl group) interacts with subsite I, while domain I1 (the C1-C2 keto acid moiety) interacts with subsite 11.

The requirement for an anionic domain I is evident from the significant changes in inhibitory potency of various ah- phatic and aromatic compounds caused by its omission or by its replacement with a methyl group. The most pronounced effects of such structural modifications were observed with the aromatic series of compounds. Thus, the requirement for domain I is evident from the observation that the Kt values of both pyridine 2,4-dicarboxylate and 3,4-dihydroxybenzoate are about 5 one-thousandths the K, values of pyridine 2- carboxylate and 1,2-dihydroxybenzene, respectively.

The requirement for the metal ion binding domain 11 is suggested by the finding that the most potent inhibitors were those compounds that contained a moiety able to act as a bidentate ligand. Although substances like pyridine 2,4-dicarboxylate and 3,4-dihydroxybenzoate form complexes with ferrous ions (21-23, 25, 32, 33), they behaved as competitive inhibitors of y-butyrobetaine hydroxylase with respect to CY-

ketoglutarate. These compounds also competitively inhibit prolyl 4-hydroxylase with respect to a-ketoglutarate and are noncompetitive inhibitors of that enzyme with respect to ferrous ion (15, 16). If the chelating moiety was modified, e.g. by elimination of the C2 carboxyl function in the pyridinecarboxylate series, or of one of the vicinal hydroxyl groups in the dihydroxybenzene series, the inhibitory potential was substantially reduced pyridine 4-carboxylate was 400 times less effective than pyridine 2,4-dicarboxylate, while 3-hydroxybenzoate and 4-hydroxybenzoate were less effective than 3,4- dihydroxybenzoate by a factor of 33 and 850, respectively. The latter result also suggested that the two hydroxyl groups that constitute the chelating domain I1 and mediate interaction to subsite 11, are not functionally equivalent. In general, the binding of meta-hydroxybenzoates was preferred over that of para-hydroxybenzoates. Thus, the K, value of 3-hydroxybenzoate was about 4 one-hundredths of the Ki value of 4- hydroxybenzoate. In the pyridinecarboxylate series, the K; value of pyridine 4-carboxylate was 0.20 the Ki value of pyridine 3-carboxylate. Therefore, withs respect to metal ion interaction, the N1 position of the pyridine carboxylates corresponds to the oxygen atom attached a t C3 of the hydroxybenzoates, if domain I-subsite I interaction is allowed (com- pare Fig. 5, c and d) . These results indicate that binding to subsite I1 occurs preferentially at the IIa position, the ferrous ion coordination site being oriented toward subsite I (Fig. 5a). In summary, the findings strongly suggest that two nonequivalent cis-positioned ferrous ion coordination sites constitute an essential part of the a-ketoglutarate binding site at the catalytic center of this enzyme.

It should be noted that the capacity to chelate metal ions in solution is not, however, the only or the main determinant for the inhibitory action of a given compound for enzymes containing the metal. Thus, in the case of y-butyrobetaine hydroxylase, metal chelation is necessary but not sufficient for optimal interaction with the a-ketoglutarate binding site. Indeed, the structural analogs of a-ketoglutarate used in this study could be subject to specific steric requirements imposed by the enzyme protein. In accord with that contingency, disparities were noted between enzyme inhibitory potencies as reflected by the determined K, values of particular compounds and their respective iron-chelatingcapacities as shown by the first iron complex formation constant K,[Fe]. Intro- duction of a carboxyl group to 1,2-dihydroxybenzene to give 3,4-dihydroxybenzoate, decreases the metal chelating potency by a factor of 10, K1[Fe] values of 1.02 X 10’‘) ( 2 6 ) and 9.97 X 10l8 (27), respectively, but increases the inhibitory efficacy by 200-fold (Table I). Sterically, that carboxyl group is in a position to interact with subsite I (Fig. 5d) . The position of


FIG. 5. Suggested subsite organization of the cosubstrate binding site of y-butyrobetaine hydroxylase and mode of attachment of a-ketoglutarate and its analogs. The a-ketoglutarate binding site ( a ) of the enzyme is subdivided into subsite I and subsite I1 showing an intersubsite distance X of 1000 k 50 pm. Subsite I, suggested to be a positively charged amino acid side chain, is capable of ionic interaction with the C5 carboxyl group of a-ketoglutarate or the corresponding moiety of its analogs (61, pyridine 2,4-dicarboxylate (c) and 3,4- dihydroxybenzoate ( d ) . In b, c, and d , these polar interactions are arbitrarily shown as salt bridges. Subsite I1 consists of the two cis-oriented, nonequivalent coordination sites of the enzyme-hound octahedral ferrous ion which are oriented along its n and y axes ( a ) . Subsites IIa and IIb allow chemically different molecules to chelate to the active site metal center, as shown in b, c, and d. Binding of molecular oxygen occurs along the z axis of the metal ion, perpendicular to the plane of the paper in this schematic presentation. a-Ketoglutarate is drawn in its staggered conformation ( b ) , with all the carbon atoms in the plane defined by the iron nucleus and the atoms of the chelating keto acid moiety. The shaded area to the right of the C2-C3 bond indicates the position of the aromatic ring in pyridine 2,4-dicarboxylate (c), the shaded area to the left of the C2-C3 bond indicates the position of the aromatic ring in 3,4-dihydroxybenzoate ( d ) . The a-ketoglutarate analogs pyridine 2,4-dicarboxylate (c) and 3,4-dihydroxybenzoate ( d ) are rigid, completely planar molecules, with all the carbon atoms in one plane. Note that the position of the N1 nitrogen in pyridine 2,4-dicarboxylate ( c ) corresponds to the position of the oxygen atom attached to C3 in 3,4-dihydroxyhenzoate ( d ) . The interdomain distance Y of about 375 pm, measured between the carbon nuclei that carry the heteroatoms for interaction with subsite I and subsite IIa, is virtually identical in @-ketoglutarate ( b ) , pyridine 2,4-dicarboxylate (c), and 3,4-dihydroxybenzoate ( d ) .

the carboxyl group in 2,3-dihydroxybenzoate as compared to the position in 3,4-dihydroxybenzoate, caused an increase in the Kl[Fe] value to 3.16 X 10'' (26). Despite this enhanced metal chelation capacity by a factor of about 30, the inhibitory potency of the 2,3-isomer decreases by more than 160-fold (Table I). Thus, the observed reduced inhibitory potency of 2,3-dihydroxybenzoate relative to the 3,4-isomer, cannot be attributed to the changing ratios between the fixed amount of free ferrous ions in the assay solution and the variable inhibitor concentrations tested. Rather, the disparity between K,[Fe] and K; reflects the steric restrictions under which chelation to the active site metal ion occurs. Apparently, steric conditions favor the selection of the 3,4-isomer for chelation to the enzyme-bound ferrous atom. That in turn suggests significant involvement of subsite I in the binding of cosubstrate to the enzyme. A similar disparity in Kl[Fe]-K; also occurred in the pyridine carboxylate series of compounds that was examined. The Kl[Fe] values for pyridine 2,3- and pyridine 2,5-dicarboxylate are virtually identical, 1.59 x lo3 and 2.51 X lo3 (28), respectively; nonetheless, the K, value of the 2,3-isomer is 10 times that of the 2,5-isomer (Table I). As with the inhibitors of the dihydroxybenzene series of compounds, the relative orientations of the anionic and the chelating moieties appear to determine the inhibitory effective- ness of a given pyridine carboxylate. For optimal activity to occur, those orientations must approximate the relative orientations of domains I and I1 in a-ketoglutarate.

In a-ketoglutarate, rotation around the C2-C3, C3-C4, and C4-C5 bonds is not sterically hindered. Accordingly, the conformation optimal for attachment to the cosubstrate binding site of y-butyrobetaine hydroxylase has been difficult to define. However, the a-ketoglutarate antagonists pyridine 2,4- dicarboxylate and 3,4-dihydroxybenzoate, are rigid planar structures in which the orientation of domain I with respect to domain I1 is fixed; that allows measurement of the distance between subsite I and subsite 11, i.e. the active site metal center and the positively charged locus participating in ionic interaction (Fig. 5a). This intersubsite distance X is estimated to be 1000 5 50 pm, using Cochrane's skeletal models. To

bridge this distance, a-ketoglutarate would have to bind in its staggered extended conformation as shown in Fig. 5b, with C3, C4, and C5 in the plane defined by C1, C2, the chelating oxygen atoms, and the iron nucleus. The distance between domain I and domain I1 of a-ketoglutarate, measured between the carbon nuclei that carry the heteroatoms for interaction with subsite I and subsite IIa, respectively, is measured to be about 375 pm for this conformation. In any given a-ketoglutarate analog, this interdomain distance has to conform closely to the value of 375 pm for optimal interaction with the cosubstrate binding site. Thus, 3,4-dihydroxyphenylacetate and 3,4-dihydroxyphenylpropionate, in which the interdomain distance significantly exceeds that value irrespective of the particular conformation, the Ki value was increased by a factor of about 20-30-fold relative to that of 3,4-dihydroxybenzoate (interdomain distance about 375 pm). More significantly, this crucial interdomain distance appears to be the parameter that determines whether or not attachment to the cosubstrate binding site is catalytically effective. a-Ketoadi- pinate, which differs from a-ketoglutarate only by an additional methylene group, did not promote the hydroxylation of y-butyrobetaine. However, a-ketoadipinate has been shown to promote the hydroxylation of proline residues in protocol- lagen by purified chicken prolyl 4-hydroxylase (15). The cosubstrate binding site of that enzyme was also less sensitive to an increase in interdomain distance than that of y-butyrobetaine hydroxylase, the Ki values for 3,4-dihydroxyphenylacetate and for 3,4-dihydroxyphenylpropionate, compared to 3,4-dihydroxybenzoate, were increased by only %fold (16).

There are other significant differences between the a-ketoglutarate binding sites of these two hydroxylases. Although both enzymes have an almost identical K,,, value for a-ketoglutarate, 3,4-dihydroxybenzoate was about 10 times more effective as an a-ketoglutarate antagonist of y-butyrobetaine hydroxylase than of prolyl 4-hydroxylase, with K, values of 0.0006 and 0.005 mM (16), respectively. Pyridine 2,4-dicarboxylate also was about 10 times more potent as an inhibitor of y-butyrobetaine hydroxylase than of propyl 4-hydroxylase, exhibiting K, values of 0.0002 and 0.002 mM (15), respectively.


The molecular architecture of the a-ketoglutarate binding site also appeared to be unique for each of the two hydroxylases, as evidenced by the spatial relationships between subsites I and 11. Thus, for the pyridinedicarboxylate series, the 2,4- isomer was the most efficient inhibitor of a-ketoglutarate decarboxylation by y-butyrobetaine hydroxylase, whereas the 2,5-isomer was most effective for prolyl 4-hydroxylase (15). Furthermore, for the dihydroxybenzoate group of analogs, the meta-isomers were more effective than the para-isomers with y-butyrobetaine hydroxylase, while the reverse was true for prolyl4-hydroxylase (16).

The demonstration that metal chelation, i.e. subsite I1 interaction, is crucial for attachment to the cosubstrate binding site of y-butyrobetaine hydroxylase has direct implica- tions for the catalytic events preceding the formation of the oxo-iron complex active in substrate hydroxylation (9, lo), and is consistent with the prior suggestion that a-ketoglutarate decarboxylation and ferry1 formation occur in a ligand reaction at the active site ferrous ion (11, 12). According to this view, the metal center would bind molecular oxygen and domain I1 of a-ketoglutarate. Such an interaction should allow dative and retrodative ?r bonding between these ligands and the central ion; that would result in enhanced susceptibility of the C2 atom of domain I1 to nucleophilic attack by the noncoordinated atom of the dioxygen unit. That addition reaction, producing a tetrahedral C2 intermediate, would then be followed by an SN2t-like elimination of carbon dioxide and the concomittant formation of both succinate and an iron- oxo species, all in a 1:1:1 stoichiometry. In that way, the active site iron atom serves as a template to hold and orient the reactants participating in both the decarboxylation and the hydroxylation phases of the catalytic cycle (11, 12). Signifi- cantly, pyridine 2,4-dicarboxylate did not promote the formation of carnitine, although it shows an interdomain distance identical to that of a-ketoglutarate and is known to preferentially decarboxylate at the C2 atom that results in the disruption of its chelating domain I1 (24). The inability of pyridine 2,4-dicarboxylate to function as a cosubstrate despite its high affinity for the cosubstrate binding site (see Table I) and despite its inherent disposition for decarboxylation at C2 (24), may well be explained by the mode of interaction between its domain I1 moiety and the active site ferrous ion. Thus, due to the aromatic nature of the C2 of the pyridine ring, this carbon atom cannot, in contrast to the aliphatic C2 of a-ketoglutarate, participate in the nucleophilic addition-elimination ligand reaction that produces the hydroxylating iron-oxo complex. In the prolyl 4-hydroxylase assay, pyridine 2,4-dicarboxylate also was not able to substitute for a-ketoglutarate (15).

It is not clear whether the observed in vitro inhibition of y- butyrobetaine hydroxylase by key intermediates of metabolism such as pyruvate, oxaloacetate, and succinate, is of any physiological significance in the control of carnitine formation in vivo. Also, 3,4-dihydroxyphenylacetate and 3,4-dihydroxymandelate, both of which are among the most effective in vitro inhibitors of the y-butyrobetaine hydroxylase (see Table I ) , are formed in the catabolic pathways of dopamine and of norepinephrine/epinephrine, respectively (29). In humans, these compounds are generated in milligram amounts each day. Finally, both 3,4-dihydroxyphenylacetate- and pyridine 2,4-dicarboxylate-like structures occur in pyrroloquinoline

quinone, and homology of the charge distribution on these domains of pyrroloquinoline quinone and the corresponding individual molecules has been demonstrated (30). Accord- ingly, it was predicted that pyrroloquinoline quinone would be an inhibitor of prolyl4-hydroxylase activity (30), a predic- tion that was subsequently verified (31).

Remarkably, among the 41 compounds examined as potential inhibitors of the bacterial y-butyrobetaine hydroxylase, five naturally occurring and chemically very different structures were identified as some of the most effective inhibitors of the enzyme: pyridine 2,4-dicarboxylate (Ki 0.0002 mM); 3,4-dihydroxyphenylacetate (Kt 0.013 mM); 3,4-dihydroxymandelate (Ki 0.028 mM); succinate (Kt 0.07 mM); oxaloacetate (Ki 0.12 mM). The inhibitory effect of these compounds, however, for mammalian y-butyrobetaine hydroxylase re- mains to be established, although this enzyme mechanistically is very similar to that of Pseudomonas sp. AK1 (9, 10).

Acknowledgments-H. M. H.-A. is indebted to Prof. K. I. Kivirikko, Oulu/Finland, for important suggestions and to Profs. J . F. Crigler, Jr., and P. M. Gallop, Boston for continuous support. He is grateful to the Evariste Galois Fund and Hoechst AG for additional assistance.

REFERENCES 1. Englard, S., and Seifter, S. (1986) Annu. Reo. Nutr. 6, 365-406 2. Rehouche, C. J., and Paulson, D. J. (1986) Annu. Reu. Nutr. 6, 41-66 3. Bremer, J. (1983) Physiol. Reu. 63, 1420-1480 4. Vamecq, J., and Draye, J.-P. (1989) Essays Biochem. 24,115-225 5. Myllyla, R., Majamaa, K:, Gunzler, V., Hanauske-Ahel, H. M., and Kivi-

6. Holme, E., Lindstedt, S., and Nordin, I. (1982) Biochem. Biophys. Res.

7. Holme, E., Lindstedt, S., and Nordin, I. (1984) Biosci. Rep. 4 , 433-440 8. Ehert, P. S., and Prockop, D. J. (1969) Biochem. Biophys. Res. Commun.

9. Blanchard, J. S., and Englard, S. (1983) Biochemistry 22,5922-5929

rikko, K. I. (1984) J . B~ol. Chem. 259,5403-5405

Commun. 107,518-524

8, 305-309

10. Englard, S., Blanchard, J. S., and Midelfort, C. F. (1985) Biochemistry 24,

12. Hanauske-Ahel, H. M. (1983) Uber einen stereochernischen Vorschlag fu r 11. Hanauske-Ahel, H. M., and Gunzler, V. (1982) J . Theor. Bid . 94,421-455

den katalytischen Mechanismus der Prolylhydroxylase, seine Anwendung zur Klassifizierung und Formulierung uon Hemmstoffen sowie die En- tutcklung und Erprobung eines “massgeschneiderten” neuartigen lnhibi- tors. M.D.-Ph.D. thesis, Phillips Universitat Marhurg, FRG

13. Gunzler, V., Hanauske-Ahel, H. M., Myllyla, R. Mohr, J., and Kivirikko, K. I. (1985) Biochem. J . 242, 163-169

14. Majamaa, K., Gunzler, V., Hanauske-Ahel, H. M., Myllyla, R., and Kivi- rikko, K. 1. (1986) J . Biol. Chem. 261, 7819-7823

15. Majamaa, K., Hanauske-Ahel, H. M., GUnzler, V., and Kivirikko, K. I . (1984) Eur. J . Biochem, 13.8,239-245

16. Malamma, K., Turpeennleml-Hujanen, T. M., Latipaa, P., Gunzler, V., Hanauske-Ahel, H. M., Hassinen, I. E., and Kivirikko, K. I. (1985) Biochem. J . 2 2 9 , 127-133

17. Kivirikko, K. I., Myllyla, R., and Pihlajaniemi, T. (1989) FASEB J . 3,

18. Ng, S F . (1988) Metabolic and Enzymological Studies of Biosynthesis of 1609-1617

19. Lindstedt, G., Lindstedt, S., and Nordin, I. (1977) Biochemistry 16, 2181- Carnitine. Ph.D. thesis, Albert Einstein College of Medicine, New York

20. Cleland, W. W. (1979) Methods Enzymol. 63, 103-138 2188

21. Acheson, R. M., and Taylor, G. A. (1959) J. Chem. Soc. 4140-4141

23. Martell, A. E., and Smith, R. M. (eds) (1973) Critical Stability Constants, 22. Cheng, K. L., and Riddick, J. A. (1954) Anal. Chem. 26,536-538

24. Bylicki, A. (1959) Bull. Acad. Pol. Sci. Ser. Sci. Chim. VII, 117-121 Vol. 1, pp. 367-370, Plenum Publishing Corp., New York

25. Martell, A. E. (1973) in Metal Ions in Biological Systems (Sigel, H., ed) Vol.

26. Avdeef, A., Sofen, S. R., Bregante, T. L., and Raymond, K. N. (1978) J .

27. Migal, P. K., and Ivanov, V. A. (1973) Russ. J . lnorg. Chem. 1 8 , 536-540 28. Morimoto, I . , and Sato, T. (1963) Bull. Chem. Soc. Jpn. 36, 605-610 29. Levine, R. J., and Landsherg, L. (1974) in Duncan’s Diseases of Metabolism:

Endocrinology (Bondy, P. K., and Rosenherg, L. E., eds) p. 1197. W. B. Saunders Co., Philadelphia, PA

30. Hanauske-Ahel, H. M., Tschank, G., Gunzler, V., Baader, E., and Gallop, P. M. (1987) FEBS Lett. 214, 236-243

31. Hanauske-Ahel, H. M., GUnzler, V., Duine, J . A,, and Kivirikko, K. 1. (1988) Collagen Relat. Res. 8, 530-531

32. Martell, A. E., and Smith, R. M. (eds) (1977) Critical Stabilit.), Constants. Vol. 3, pp. 200-202, Plenum Publishing Corp., New York

33. Smith, R. M., and Martell, A. E. (eds) (1989) Critical Stability Constants, Val. 6, pp. 378-379, Plenum Publishing Corp., New York

1110-1116

2, p. 207, Marcel Dekker, New York

Am. Chem. SOC. 100,5362-5370

Documents

Cosubstrate Binding Site of Pseudomonas sp, AK1 y-Butyrobetaine