Synthesis of a Highly Substituted N6-Linked Immobilized NAD+ Derivative Using a Rapid Solid-Phase Modular Approach: Suitability for Use with the Kinetic Locking-on Tactic for Bioaffinity Purification of NAD+-Dependent Dehydrogenases

Protein Expression and Purification 20, 421– 434 (2000) doi:10.1006/prep.2000.1314, available online at http://www.idealibrary.com on

Synthesis of a Highly Substituted N 6-Linked Immobilized NAD ⫹ Derivative Using a Rapid Solid-Phase Modular Approach: Suitability for Use with the Kinetic Locking-on Tactic for Bioaffinity Purification of NAD ⫹-Dependent Dehydrogenases Julie Tynan, Jessica Forde, Mary McMahon, and Patricia Mulcahy 1 Department of Applied Biology and Chemistry, Institute of Technology Carlow, Kilkenny Road, Carlow, Ireland

Received May 15, 2000, and in revised form July 11, 2000

This study is concerned with further development of the kinetic locking-on strategy for bioaffinity purification of NAD ⴙ-dependent dehydrogenases. Specifically, the synthesis of highly substituted N 6-linked immobilized NAD ⴙ derivatives is described using a rapid solid-phase modular approach. Other modifications of the N 6-linked immobilized NAD ⴙ derivative include substitution of the hydrophobic diaminohexane spacer arm with polar spacer arms (9 and 19.5 Å) in an attempt to minimize nonbiospecific interactions. Analysis of the N 6-linked NAD ⴙ derivatives confirm (i) retention of cofactor activity upon immobilization (up to 97%); (ii) high total substitution levels and high percentage accessibility levels when compared to S 6linked immobilized NAD ⴙ derivatives (also synthesized with polar spacer arms); (iii) short production times when compared to the preassembly approach to synthesis. Model locking-on bioaffinity chromatographic studies were carried out with bovine heart L-lactate dehydrogenase (L-LDH, EC 1.1.1.27), bakers yeast alcohol dehydrogenase (YADH, EC 1.1.1.1) and Sporosarcinia sp. L-phenylalanine dehydrogenase (LPheDH, EC 1.4.1.20), using oxalate, hydroxylamine, and D-phenylalanine, respectively, as locking-on ligands. Surprisingly, two of these test NAD ⴙ-dependent dehydrogenases (lactate and alcohol dehydrogenase) were found to have a greater affinity for the more lowly substituted S 6-linked immobilized cofactor derivatives than for the new N 6-linked derivatives. In contrast, the NAD ⴙ-dependent phenylalanine dehydrogenase showed no affinity for the S 6-linked immobilized NAD ⴙ derivative, but was locked-on strongly to 1 To whom correspondence and reprint requests should be addressed. Fax: 0503 70500. E-mail: [email protected].

1046-5928/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

the N 6-linked immobilized derivative. That this locking-on is biospecific is confirmed by the observation that the enzyme failed to lock-on to an analogous N 6linked immobilized NADP ⴙ derivative in the presence of D-phenylalanine. This differential locking-on of NAD ⴙ-dependent dehydrogenases to N 6-linked and S 6linked immobilized NAD ⴙ derivatives cannot be explained in terms of final accessible substitutions levels, but suggests fundamental differences in affinity of the three test enzymes for NAD ⴙ immobilized via N 6linkage as compared to thiol-linkage. © 2000 Academic Press Key Words: phenylalanine dehydrogenase; lactate dehydrogenase; alcohol dehydrogenase; bioaffinity chromatography; kinetic locking-on tactic.

NAD ⫹-dependent dehydrogenases continue to serve as a fertile field for enzymological research because of their physiological and biotechnological importance (1– 15). However, detailed structural studies and biotechnological application of this class of enzymes is often hampered by the requirement for large quantities of highly purified enzymes with appropriate kinetic and stability properties. One approach to overcoming restrictions inherent in the use of conventional protein purification procedures is the development of highly specific bioaffinity chromatographic systems capable of purifying the target enzyme in a single chromatographic step, with yields approaching 100%. The bioaffinity systems should also have a very high capacity for the target enzyme so that “large-scale” purifications can be achieved on the laboratory bench. We have developed such bioaffinity systems for several NAD ⫹dependent dehydrogenases (16 –23). The approach utilizes immobilized NAD ⫹ (general ligand) derivatives in 421

422

TYNAN ET AL.

conjunction with soluble-specific substrates (the locking-on ligand) or specific-substrate analogues, to promote biospecific binding. Application of this tactic to a range of dehydrogenases has proven the power of the approach (16 –23), while the introduction of auxiliary tactics (16, 17) has ensured that the locking-on tactic can produce a degree of biospecificity similar to that obtained for the most successful immobilized specific ligand systems (19). Furthermore, since immobilized general ligands are capable of interacting with a wide range of complementary proteins, a single adsorbent can be used for the one-step purification of a number of enzymes (the target enzyme being selected by controlling the soluble enzyme-specific substrate analogue in the chromatographic irrigant). The interrelationship between the position and nature of the NAD ⫹ immobilization linkage, the stability of the immobilized cofactor derivative, the spacer-arm composition of the affinity matrix, the accessible immobilized ligand concentration, the soluble locking-on ligand concentration, the dissociation constant of locking-on ligand, and identification and elimination of nonbiospecific adsorption effects, have been addressed in our earlier reports (16 –23). The present study is concerned with further development of the technologies involved through: 1. Synthesizing immobilized cofactor derivatives with improved chemical properties for protein purification applications. 2. Addressing difficulties traditionally associated with the scale-up of bioaffinity-based separations including the high cost of reagents used for synthesis of the affinity matrices and the time-consuming, often complex nature, of the synthetic methods currently employed. MATERIALS AND METHODS

All chemicals and biochemicals were obtained from Sigma Chemical Co. (Poole, Dorset, England), including the following commercial enzyme preparations: alcohol dehydrogenase from bakers yeast (EC 1.1.1.1), aldehyde dehydrogenase from yeast (EC 1.2.1.5), Llactate dehydrogenase from bovine heart (Type III) (EC 1.1.1.27), and L-phenylalanine dehydrogenase from Sporosarcinia sp. (EC 1.4.1.20). Enzyme Assays Alcohol dehydrogenase activity from Saccharomyces cerevisiae was monitored spectrophotometrically at 30°C by measuring the increase in absorbance at 340 nm. Assays were carried out in 30 mM sodium pyrophosphate buffer (pH 8.8) containing 0.8 mM NAD ⫹ and 0.12 M ethanol. The assay temperature was 30°C. L-Lactate dehydrogenase activity from bovine heart

(Type III) was monitored spectrophotometrically at 30°C by measuring the decrease in absorbance at 340 nm. Assays were carried out in 50 mM potassium phosphate buffer (pH 7.4) containing 0.5 mM pyruvate and 0.2 mM NADH. The assay temperature used was 30°C. L-Phenylalanine dehydrogenase activity from Sporosarcinia sp. was monitored spectrophotometrically at 25°C by measuring the increase in absorbance at 340 nm. Assays were carried out in 100 mM glycine buffer (pH 8.6) containing 0.1 M KCl, 0.5 mM NAD ⫹, and 10 mM L-phenylalanine. Synthesis of Immobilized NAD ⫹ Derivatives S 6-Linked immobilized NAD ⫹ derivatives were synthesized as described in Ref. (20) using 1,3 diaminopropanol as the spacer arm compound. N 6-Linked immobilized NAD ⫹ derivatives were synthesized using methods based on those described by Mosbach (24), but with important modifications. First, the immobilized derivatives were synthesized using a variety of spacer arm compounds other than the 1,6diaminohexane hydrophobic spacer arm favored by Mosbach and coworkers. Second, unlike the preassembly approach used in the earlier studies, the N 6-linked immobilized NAD ⫹ derivatives described in the present study were synthesized using a solid phase modular approach. In the latter case, N 6-carboxymethyl-NAD ⫹ was coupled directly to the Sepharose-4Bspacer arm complex, rather than synthesizing the spacer arm-ligand assembly in free solution prior to coupling of the entire “preassembled” complex to the inert matrix (24). The Sepharose-4B-spacer arm complexes used in the present study were synthesized as described in Ref. (20). N 6-Carboxymethyl-NAD ⫹ was synthesized as described in Ref. (24) but with modifications introduced by Buchanan (25). NAD ⫹ (1.0 g) was added to an aqueous solution of iodoacetic acid (3.0 g) that had been previously neutralized to pH 7.5 with 2 M and 0.2 M lithium hydroxide. The pH was adjusted to pH 6.5 and brought to a final volume of 30 ml. The reaction mixture was kept in the dark at 26°C for 10 days. The pH was checked daily and adjusted when necessary to pH 6.5 using 2 M lithium hydroxide. The reaction mixture was adjusted to pH 3.0 with 3 M HCl. Two volumes of chilled acetone (60 ml) were added, producing a milky suspension which was then added to 10 vol of vigorously stirred acetone at 0°C (300 ml). The solution produced a heavy flocculate precipitate. Crude N 1-carboxymethyl-NAD ⫹ was filtered, washed with acetone and ether, and dried under a vacuum to give a pink power (crude N 1-NAD ⫹). The crude N 1-carboxymethylNAD ⫹ was dissolved in 90 ml of 2% sodium hydrogen carbonate and the pH was adjusted to pH 8.5 with 1 M NaOH. The solution was deaerated by bubbling nitro-

BIOSPECIFIC AFFINITY PURIFICATION OF NAD ⫹-DEPENDENT DEHYDROGENASES

423

FIG. 1. Elution of N-carboxymethyl-NAD ⫹ nucleotides from a Dowex 1(⫻2) column (200 mesh, chloride form) with a linear gradient of CaCl 2 (peak I: N 1-carboxymethyl-NAD ⫹; peak II: N 6-carboxymethyl-NAD ⫹).

gen through the reaction mixture for 2 min. At this point, 1.5 g sodium dithionite was added before leaving the solution in the dark for 5 h. The absorbance of the sample was monitored at 340 nm prior to the addition of sodium dithionite and also at intervals after the addition of sodium dithionite. To terminate the reduction, the solution was oxygenated for 12 min with pure air followed by bubbling nitrogen through the solution for a 3-min period. The reduced solution was then adjusted to pH 11.5 with 1 M NaOH to bring about the rearrangement to N 6-carboxymethyl-NADH. Enzymatic reoxidation was carried out at room temperature: 6 ml of 2 M Tris (pH 7.5) and 1.5 ml of redistilled acetaldehyde was added to the solution. The pH was adjusted to pH 7.5 with 3 M HCl before the addition of yeast alcohol dehydrogenase (5150 units). The solution was left at room temperature for six h. During this period the absorbance was checked intermittently at 340 nm. One volume of acetone was added to the solution and then poured into 10 vol of vigorously stirred acetone at room temperature. This was left overnight to allow the precipitate to form (stored at ⫺20°C until required). The crude N 6 -carboxymethyl-NAD ⫹ was dissolved in 30 ml of distilled water and the pH was adjusted to pH 8.0 with 1 M lithium hydroxide. The solution was then loaded onto a Dowex 1(⫻2) (200 mesh) chloride form (dimensions 5 ⫻ 12.5 cm), which had been previously washed with 3 M HCl and equilibrated with 20 liters of water. The column was washed with distilled water (500 ml) and 5 mM CaCl 2 (1 liter, pH 2.0) until the effluent no longer

contained ultraviolet absorbing material. A linear CaCl 2 gradient was applied from 5 mM CaCl 2 (1 liter, pH 2.7) to 100 mM CaCl 2 (1 liter, pH 2.0). The fractions of the main peak (Fig. 1) were neutralized and concentrated on a Dowex 1(⫻2) column, chloride form (dimensions 1.25 ⫻ 7 cm). The solution was then lyophilized to give a white powder which was then desalted on a Sephadex G-10 column (2.5 ⫻ 19cm). The desalted solution was lyophilized once more to produce a white product that was stored at ⫺70°C. The purified N 6 -carboxymethyl NAD ⫹ was coupled to various Sepharose 4B-spacer arm assemblies (19.5 and 9 Å, see Figs. 2 and 3) at pH 7.4, using the following general procedure: N 6 -carboxymethylNAD ⫹ was dissolved in water to produce an absorbance reading of 90 at 266 nm. This solution (2 ml) was added to 2 g (dried by suction) of the Sepharose 4B-spacer arm assembly that had previously been washed with distilled water. The pH was adjusted to pH 7.5 using 8 M NaOH. The water-soluble carbodiimide, 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide (148 mg/g of gel), was dissolved in 0.3 ml of water. Aliquots (100 ␮l) of the water-soluble carbodiimide was added slowly and dropwise every 2 min, adjusting the pH to 7.5 with 0.5 M HCl after each addition and for the following h. The reaction was allowed to proceed (with gentle stirring) for 30 – 40 h at room temperature after which time the matrix was washed with 20 vol of 0.1 M NaCl and distilled water. The synthesis of N 6 -linked immobilized NAD ⫹ affinity matrices with varying substitution levels

424

TYNAN ET AL.

FIG. 2.

Structures of the affinity matrices used in the present study.

was achieved by varying the concentration of N 6 carboxymethyl-NAD ⫹ used in the final coupling reaction. The matrices showed no deterioration over a 6-month time period once they were stored in 100 mM potassium phosphate buffer pH 7.4 at 4°C. Analysis of Immobilized NAD ⫹ Derivatives The total amount of immobilized NAD ⫹ was determined using chemical digestion followed by phosphorus analysis (26). The amount of accessible immobilized NAD ⫹ was determined by monitoring the

enzymatic reduction of immobilized NAD ⫹ to NADH by alcohol and aldehyde dehydrogenase (20). Affinity Chromatography Chromatography was carried out at room temperature in miniature columns (1.7 cm internal diameter; 1 ml bed volume) with hydrostatic head pressure adjusted to give a flow rate of approximately 1-column vol/2.8 min. All columns were equilibrated with the appropriate irrigating buffer prior to application of the enzyme sample.

425

BIOSPECIFIC AFFINITY PURIFICATION OF NAD ⫹-DEPENDENT DEHYDROGENASES

FIG. 3. Absorption spectrum of N 6-linked immobilized NAD ⫹ derivative (9 Å hydrophilic spacer arm; total NAD ⫹, 1.86 ␮mol/g wet wt.; accessible NAD ⫹, 1.66 ␮mol/g wet wt.). The sample cell contained 250 mg of matrix and the reference cell contained 250 mg of the spacer arm-Sepharose 4B assembly unsubstituted with ligand (suspended in 3 ml of distilled water).

RESULTS AND DISCUSSION

Development of Biospecific Affinity Chromatographic Methods Based on the Locking-on Tactic The first step in the development of a biospecific affinity chromatographic method based on the locking-on tactic is identification of a suitable locking-on ligand (competitive inhibitor of the specific substrate) followed by “model” affinity chromatography studies with a variety of locking-on ligands and immobilized cofactor derivatives. The term “model” is used to describe preliminary studies carried out with pure/partially purified dehydrogenase preparations (when available). The rationale for this is that the use of partially purified enzyme activities in the initial stages of the research minimizes complications introduced by the use of crude cellular extracts. One good example of this is the interference of cellular racemases during the bioaffinity purification of L-phenylalanine dehydroge-

nase from microbial extracts using D-phenylalanine as locking-on ligand. Unless these racemases are inactivated prior to the chromatography, the locking-on ligand is converted to L-phenylalanine and the immobilized NAD ⫹-derivative is reduced to NADH (effectively destroying the affinity matrix). Once the potential for interference in crude extracts is identified, resolution of the problem is generally straightforward. In the example given above, a general irreversible inhibitor of racemases, which does not interfere with the target amino acid dehydrogenase activity, is added to all buffers used for extraction. Our earlier studies have confirmed that the dissociation constant (K i value) for the locking-on ligands dictates the effective concentration of soluble locking-on ligand for bioaffinity chromatography. For example, using various competitive inhibitors of bovine liver glutamate dehydrogenase, the percentage en-

TABLE 1 Properties of the N 6-Linked Immobilized NAD(P) ⫹ and S 6-Linked Immobilized NAD ⫹ Derivatives Synthesized in the Present Study Immobilization (nature of spacer arm) N 6 -linked NAD ⫹ (9 Å hydrophilic arm) N 6-linked NAD⫹ (19.5 Å hydrophilic arm) N 6-linked NADP ⫹ (9 Å hydrophilic arm) S 6-linked NAD ⫹ (9 Å hydrophilic arm) N 6-linked NAD ⫹ (9 Å hydrophilic arm)

Total [cofactor] (␮mol/g wet wt.)

Accessible [cofactor] (␮mol/g wet wt.)

Percentage reduction (enzymatic)

1.86 0.89 1.00 0.70 0.24

1.66 0.84 0.90 0.30 0.23

87.5 92.2 90.0 42.9 97.0

426

TYNAN ET AL.

TABLE 2 Affinity Chromatography of L-Lactate Dehydrogenase from Bovine Heart on N 6- and S 6-Linked Immobilized NAD ⫹ Derivatives Immobilization point (spacer arm) N 6-linked NAD ⫹ (9 Å hydrophilic arm)

N 6-linked NAD ⫹ (19.5 Å hydrophilic arm)

S 6-linked NAD ⫹ (9 Å hydrophilic arm)

[NAD ⫹] (␮mol/g)

Locking-on buffer

1.66

0.5 M KCl

Eluted in the protein breakthrough peak Locked-on in the presence of 100 mM oxalate and autoeluted upon discontinuation of oxalate Not locked-on in the presence of 20 mM oxalate

0.84

100 mM oxalate, and 0.5 M KCl 0.5 M KCl plus 20 mM oxalate 0.5 M KCl plus 5 mM oxalate 0.5 M KCl

Locked-on in the presence of 100 mM oxalate plus 10 mM 5⬘-AMP and autoeluted upon discontinuation of oxalate (Fig. 4B) Not locked-on in the presence of 20 mM oxalate (Fig. 4C) Not locked-on in the presence of 5 mM oxalate

0.7

100 mM oxalate, 0.5 M KCl and 10 mM 5⬘-AMP 0.5 M KCl plus 20 mM oxalate 0.5 M KCl plus 5 mM oxalate —

Eluted in the protein breakthrough peak (Fig. 4A)

0.5 M KCl 5 mM oxalate, 0.5 M KCl and 10 mM 5⬘-AMP

Locked-on in the presence of 5 mM oxalate plus 10 mM 5⬘-AMP and autoeluted upon discontinuation of oxalate (19)

0.2 mM oxalate and 10 mM 5⬘-AMP

0.2

Not locked-on in the presence of 5 mM oxalate

Adsorbed by the matrix in the absence of lockingon ligand and was competitively eluted using 10 mM 5⬘-AMP (18) Locked-on in the presence of 0.5 mM oxalate plus 10 mM 5⬘-AMP and autoeluted upon discontinuation of oxalate (18) Locked-on in the presence of 0.2 mM oxalate, partially stripped off upon addition of 10 mM 5⬘-AMP, autoeluted upon discontinuation of oxalate Eluted in the protein breakthrough peak (19)

0.5 mM oxalate and 10 mM 5⬘-AMP

S 6-linked NAD ⫹ (9 Å hydrophilic arm)

Chromatographic behavior

Note. The irrigant was 50 mM potassium phosphate buffer (pH 7.4) and indicated additions. (The concentration of immobilized cofactor indicated refers to accessible cofactor concentration.)

zyme activity locked-on to an S 6-linked immobilized NAD ⫹-derivative gradually decreased as the K i value of the competitive inhibitor increased (17). However, because K i values are influenced by pH, temperature, buffer type, ionic strength, and other physical parameters, there is scope to further optimize enzyme assay/ chromatographic conditions, which produce even lower K i values. It is of course essential that a compromise be reached between enzyme stability and those conditions producing optimum K i values for locking-on of the enzyme to its complementary immobilized cofactor derivative. The second step in the development of a biospecific affinity chromatographic method based on the locking-on tactic is identification of a suitable immobilized cofactor derivative. The range of parameters that must be considered here include position and nature of the immobilization linkage, the spacer-arm composition of

the affinity matrix, the accessible immobilized ligand concentration, identification, and elimination of nonbiospecific adsorption effects. The third step in the development of biospecific affinity chromatographic methods based on the locking-on tactic is application of the developed model systems to the purification of the target enzyme activity from its crude cellular extract. The primary consideration here is identification and elimination of interference from elements present in crude extracts. These include substances that interfere with biospecific adsorption (e.g., phosphatases or racemases), general problems involving nonbiospecific adsorption of “other” proteins in the extracts (nonbiospecific interactions), and gross effects such as high viscosity. The need for the introduction of auxiliary tactics is also established at this stage (e.g., the use of heterotrophic effectors for allosteric enzymes, the stripping ligand tactic using

BIOSPECIFIC AFFINITY PURIFICATION OF NAD ⫹-DEPENDENT DEHYDROGENASES

427

TABLE 3 Affinity Chromatography of Alcohol Dehydrogenase from Saccharomyces cerevisiae on N 6- and S 6-Linked Immobilized NAD ⫹ Derivatives Immobilization point (spacer arm) N 6-linked NAD ⫹ (9 Å hydrophilic arm) 6

⫹

N -linked NAD (19.5 Å hydrophilic arm)

[NAD ⫹] (␮mol/g)

Locking-on buffer

1.66

0.1 M KCl

0.84

100 mM hydroxylamine and 0.1 M KCl 0.1 M KCl 100 mM hydroxylamine 10 mM 5⬘-AMP and 0.1 M KCl

S 6-linked NAD ⫹ (9 Å hydrophilic arm)

0.7

20 or 5 mM hydroxylamine and 0.1 M KCl — 10 mM hydroxylamine 20 mM hydroxylamine

S 6-linked NAD ⫹ (9 Å hydrophilic arm)

0.3

— 100 mM hydroxylamine, 10 mM 5⬘-AMP

Chromatographic behavior

Eluted in the protein breakthrough peak (Fig. 5A) Not locked-on in the presence of 100 mM hydroxylamine (Fig. 5B) Eluted in the protein breakthrough peak (Fig. 6A) Locked-on in the presence of hydroxylamine plus 10 mM 5⬘-AMP; Partial autoelution upon discontinuation of hydroxylamine (Fig. 6B) Not locked-on in the presence of hydroxylamine (Fig. 6C) Eluted in protein breakthrough peak (20) Not locked-on in the presence of 10 mM hydroxylamine (20) Locked-on in the presence of hydroxylamine and autoeluted upon discontinuation of hydroxylamine (20) Eluted in protein breakthrough peak (16) Locked-on in the presence of 100 mM hydroxylamine plus 10 mM 5⬘-AMP and autoeluted upon discontinuation of hydroxylamine (16)

Note. The irrigant was 100 mM sodium pyrophosphate buffer (pH 8.8) and indicated additions. (The concentration of immobilized cofactor indicated refers to accessible immobilized cofactor concentration.)

5⬘-AMP for NAD ⫹-dependent dehydrogenases or the stripping ligand tactic using various ADP/NADP ⫹ analogues for NADP ⫹-dependent dehydrogenases). Characteristics of the N 6- and S 6-Linked Immobilized NAD ⫹ Derivatives Synthesized Using the Solid Phase Modular Approach Our investigations to-date on the development of kinetic strategies and tactics, which increase the specificity of immobilized NAD ⫹ affinity matrices for individual target NAD ⫹-dependent dehydrogenases, have focussed primarily on the use of 8⬘-azo-linked and S 6linked immobilized NAD ⫹ derivatives (16 –23). However, both of these affinity matrices have been found to be less than ideal in this mode of use. The 8⬘-azo-linked immobilized NAD ⫹ derivative is relatively unstable and the nature of the attachment introduces nonbiospecific interference when applied to certain dehydrogenases (20). In contrast, while the S 6-linked immobilized NAD ⫹ derivative is very stable, the synthetic methods currently employed are long and cumbersome, the total and accessible immobilized NAD ⫹ concentrations achieved are low, and heterogeneous ma-

trices containing significant quantities of immobilized AMP are produced (18). In a search for alternative and more effective immobilized cofactor derivatives, it was noted that in 1973 Mosbach reported synthesis of an N 6-linked immobilized NAD ⫹ derivative (24) that retained cofactor activity for selected NAD ⫹-dependent dehydrogenases. This study aimed to develop the matrix described by Mosbach in the following key areas: (a) Chemical nature of the spacer arm: The spacer arm compound used by Mosbach was 1,6-diaminohexane (24). Our earlier studies have confirmed that diaminohexane substantially increases the potential for nonbiospecific interactions, to the extent that nonbiospecific interference can completely eclipse the biospecific interaction (16 –23). Thus the present study focused on using the polar 1,3-diaminopropanol as the spacer arm compound. (b) Length of the spacer arm: It is well established that an affinity ligand must be placed sufficiently distant from the support matrix in order to minimize steric hindrance. The present study has thus employed spacer arms of varying length (9 and 19.5 Å) to establish the optimum spacer arm length necessary for un-

428

TYNAN ET AL.

TABLE 4 Affinity Chromatography of L-Phenylalanine Dehydrogenase on N 6- and S 6-Linked Immobilized NAD ⫹ Derivatives Immobilization point (spacer arm) N 6-linked NAD ⫹ (9 Å hydrophilic arm)

N 6-linked NAD ⫹ (9 Å hydrophilic arm)

N 6-linked NAD ⫹ (19.5 Å hydrophilic arm)

S 6-linked NAD ⫹ (9 Å hydrophilic arm)

[NAD ⫹] (␮mol/g)

Locking-on buffer

Chromatographic behavior

1.66

0.2 M KCl

Eluted in protein breakthrough peak (Fig. 7A) Locked-on in the presence of 0.1 mM phenylalanine plus 10 mM 5⬘-AMP and autoeluted upon discontinuation of phenylalanine (Fig. 7B)

0.23

0.1 mM Dphenylalanine, 10 mM 5⬘-AMP and 0.2 M KCl 0.2 M KCl

Locked-on in the presence of 1 mM phenylalanine plus 10 mM 5⬘-AMP and autoeluted upon discontinuation of phenylalanine (Fig. 8B) Locked-on in the presence of 0.1 mM phenylalanine but was stripped off the matrix upon addition of 10 mM 5⬘-AMP to the irrigant (Fig. 8C)

0.84

1 mM D-phenylalanine, 10 mM 5⬘-AMP and 0.2 M KCl 0.1 mM Dphenylalanine, 10 mM 5⬘-AMP and 0.2 M KCl 0.2 M KCl 5 mM D-phenylalanine, 10 mM 5⬘-AMP and 0.2 M KCl 1 mM D-phenylalanine, 10 mM 5⬘-AMP and 0.2 M KCl

Locked-on in the presence of 5 mM phenylalanine plus 10 mM 5⬘-AMP and autoeluted upon discontinuation of phenylalanine Locked-on in the presence of 1 mM D-phenylalanine but partially stripped off in the presence of 1 mM Dphenylalanine plus 10 mM 5⬘-AMP (the remainder autoeluting upon discontinuation of D-phenylalanine) Eluted in protein breakthrough peak

0.3

— 10 mM Dphenylalanine, and 0.2 M KCl

Eluted in protein breakthrough peak (Fig. 8A)

Eluted in protein breakthrough peak

Not locked-on in the presence of 10 mM Dphenylalanine

Note. The irrigant was 100 mM glycine buffer (pH 8.6) and indicated additions. (The concentration of immobilized cofactor indicated refers to accessible concentration.)

impeded locking-on to the N 6-linked immobilized cofactor derivative. (c) Substitution level and percentage accessibility: The substitution level of the immobilized cofactor derivative is an important factor that has been ascertained in the present study using acid hydrolysis followed by phosphate analysis. This gives an estimation of the total amount of NAD ⫹ immobilized. However, it is also important to determine the amount of immobilized NAD ⫹ that is available for interaction with the target enzyme (percentage accessibility, see Ref. 18). This has been calculated by enzymatically reducing the immobilized cofactor (the change in absorbance at 340 nm is monitored). With the bioaffinity approach under development in the present study, it is desirable to have the highest substitution levels possible, together with high percentage accessibilities. However, we have also synthesized a lowly substituted N 6-linked derivative for comparative studies with similarly substituted S 6-linked immobilized NAD ⫹. (d) “Preassembly” versus “solid-phase modular” approach to synthesis: These are essentially the two ap-

proaches used for the synthesis of immobilized NAD ⫹ derivatives. The preassembly approach involves synthesis of the ligand-spacer arm complex in free solution using conventional organic chemistry. This complex is then purified and the entire preassembled complex is coupled to the support matrix in one final step. In comparison, the solid-phase modular approach involves building up the assembly gradually starting with the inert matrix. With reference to the latter point, there are two disadvantages associated with synthesizing N 6 linked immobilized NAD ⫹ derivatives using the preassembly approach. The first is low substitution levels compared to those obtained with either 8⬘azolinked or S 6 -linked immobilized cofactors. The second disadvantage is that the preassembly approach is very cumbersome taking approximately 3 months to complete. The time involved, and the small quantities of matrices finally produced, makes future scale-up of this system unlikely. Although these disadvantages make the preassembly approach

BIOSPECIFIC AFFINITY PURIFICATION OF NAD ⫹-DEPENDENT DEHYDROGENASES

429

FIG. 4. Chromatography of L-lactate dehydrogenase from bovine heart on an N 6-linked immobilized NAD ⫹ derivative (19.5 Å hydrophilic spacer arm; total NAD ⫹, 0.89 ␮mol/g wet wt.; accessible NAD ⫹, 0.84 ␮mol/g wet wt.) and an N 6 linked immobilized NADP ⫹ derivative (9 Å hydrophilic spacer arm; total NADP ⫹, 1.0 ␮mol/g wet wt.; accessible NADP ⫹, 0.9 ␮mol/g wet wt.). The irrigant was 100 mM potassium phosphate buffer (pH 7.4) containing 0.5 M KCl and indicated additions. Chromatography was performed at room temperature.

the least favored method for synthesizing immobilized N 6 -linked NAD ⫹ derivatives, it is commonly cited as the best approach. The reason given for this is that production of ligand-less spacer arm assemblies is avoided and this minimizes nonbiospecific interference. However, we have shown that while this conclusion is correct when diaminohexane is used as the spacer arm compound, this is not the case when diaminopropanol is used. Properties of the N 6 -linked and S 6 -linked immobilized NAD ⫹ derivatives used in the present study are presented in Table 1. (The properties of an N 6 -linked immobilized NADP ⫹ derivative, used as a “control” matrix to confirm biospecificity, has also been included). These results confirm the following for the N 6 linked immobilized derivatives synthesized using

the solid-phase modular approach described in the experimental section: (i) Retention of cofactor activity (up to 97%); (ii) High total substitution levels and high percentage accessibility levels when compared to S 6 -linked immobilized NAD ⫹ derivatives (20); (iii) Short production times when compared to the preassembly approach; (iv) Synthetic procedures that, with further modifications, will facilitate scale-up for future large scale purifications. Model Bioaffinity Chromatographic Studies: Comparative Studies with S 6- and N 6-Linked Immobilized NAD ⫹ Derivatives Model bioaffinity studies were carried out using Llactate dehydrogenase from bovine heart, alcohol de-

430

TYNAN ET AL.

FIG. 5. Chromatography of yeast alcohol dehydrogenase from bakers yeast on an N 6-linked immobilized NAD ⫹ derivative (9 Å hydrophilic spacer arm; total NAD ⫹, 1.86 ␮mol/g wet wt.; accessible NAD ⫹, 1.66 ␮mol/g wet wt.) and an N 6-linked immobilized NADP ⫹ derivative (9 Å hydrophilic spacer arm; total NADP ⫹, 1.0 ␮mol/g wet wt.; accessible NADP ⫹, 0.9 ␮mol/g wet wt.). The irrigant was 30 mM sodium pyrophosphate buffer (pH 8.8) containing 0.1 M KCl and indicated additions. Chromatography was performed at room temperature.

hydrogenase from bakers yeast, and L-phenylalanine dehydrogenase from Sporosarcinia sp. (The locking-on ligands used were oxalate, hydroxylamine, and D-phenylalanine, respectively). A summary of the results obtained with the N 6-linked derivatives are presented in Tables 2– 4 and Figs. 4 – 8, along with some comparative studies using S 6-linked derivatives (Tables 2– 4). Bovine heart L-LDH showed no affinity for the N 6linked immobilized NAD ⫹ derivatives with accessible immobilized ligand concentrations of 1.66 and 0.84 ␮mol/g wet wt. matrix (9 and 19.5 Å spacer arms, respectively, e.g., see Fig. 4A). This is in contrast to results obtained in our earlier studies with the more lowly substituted S 6-linked immobilized NAD ⫹ derivatives: bovine heart L-LDH was strongly adsorbed by S 6-linked derivatives with accessible immobilized cofactor concentrations of 0.2– 0.7 ␮mol/g wet wt. in the absence of locking-on ligand. The adsorbed activity could be competitively eluted using either soluble NAD ⫹ or 5⬘-AMP (Table 2 and Ref. 18). The lower affinity of bovine heart L-LDH for the N 6-linked immobilized NAD ⫹ was also evident when attempts were made to use these derivatives in the locking-on mode.

For example, while bovine heart L-LDH was completely locked-on to the N 6-linked derivatives in the presence of 100 mM oxalate (e.g., Fig. 4B), 20 mM oxalate had a negligible effect (e.g., Fig. 4C). That the locking-on with 100 mM oxalate is biospecific was confirmed by the observation that the enzyme failed to lock-on to analogous N 6-linked immobilized NADP ⫹ derivatives in the presence of oxalate (e.g., Fig. 4D). In contrast, a progressive decrease of oxalate concentration from 100 mM produced no apparent weakening of the locking-on effect with the S 6-linked derivatives until submillimolar concentrations were reached (0.5 mM oxalate produced strong enhancement of the adsorption, 0.2 mM promoted weaker binding, and 0.05 mM oxalate had a negligible effect, see Table 2). The difference in percentage accessibilities between the S 6- and N 6-linked immobilized NAD ⫹ derivatives do not explain the difference in bovine L-LDH chromatographic behavior. Indeed, our earlier studies have confirmed that increasing the accessible immobilized cofactor concentration promotes locking-on at lower soluble concentrations of locking-on ligand (18). Similar results were obtained with yeast alcohol de-

BIOSPECIFIC AFFINITY PURIFICATION OF NAD ⫹-DEPENDENT DEHYDROGENASES

431

FIG. 6. Chromatography of alcohol dehydrogenase from bakers yeast on an N 6-linked immobilized NAD ⫹ derivative (19.5 Å hydrophilic spacer arm; total NAD ⫹, 0.89 ␮mol/g wet wt.; accessible NAD ⫹, 0.84 ␮mol/g wet wt.) and an N 6-linked immobilized NADP ⫹ derivative (9 Å hydrophilic spacer arm; total NADP ⫹, 1.0 ␮mol/g wet wt.; accessible NAD ⫹, 0.9 ␮mol/g wet wt.). The irrigant was 30 mM sodium pyrophosphate buffer (pH 8.8) containing 0.1 M KCl and indicated additions. Chromatography was performed at room temperature.

hydrogenase. This enzyme was not locked-on to an N 6-linked immobilized NAD ⫹ derivative [accessible (NAD ⫹), 1.66/g wet wt; spacer arm length, 9 Å] in the presence of 100 mM hydroxylamine (Fig. 5). However, when the 9 Å spacer arm was replaced with a 19.5 Å spacer arm, the ADH was locked-on completely to the N 6-linked derivatives in the presence of 100 mM hydroxylamine (although not at lower locking-on ligand concentrations, e.g., Fig. 6). In contrast, a decrease in hydroxylamine concentration from 100 to 20 mM produced no apparent weakening of the locking-on effect with S 6-linked derivatives (0.7 ␮mol/g wet wt; 20 mM hydroxylamine produced strong enhancement of the adsorption and 10 mM hydroxylamine had a negligible effect, see Table 3).

While lactate and alcohol dehydrogenase were found to have a greater affinity for the more lowly substituted S 6-linked immobilized cofactor derivatives than for the new N 6-linked derivatives, the NAD ⫹-dependent phenylalanine dehydrogenase showed no affinity for the S 6-linked immobilized NAD ⫹ derivative but was locked-on strongly to the N 6-linked immobilized derivatives over a wide range of accessible immobilized NAD ⫹ concentrations (Figs. 7 and 8 and Table 4). For example, using the N 6-linked immobilized NAD ⫹ derivative with a 9 Å spacer arm and 1.66 ␮mol accessible NAD ⫹/g wet wt (Fig. 7B), 0.1 mM D-phenylalanine (competitive inhibitor with respect to L-phenylalanine) proved to be an effective locking-on ligand. That the locking-on with 0.1 mM D-phenylalanine is biospecific

432

TYNAN ET AL.

FIG. 7. Chromatography of L-phenylalanine dehydrogenase from Sporosarcinia sp. on an N 6-linked immobilized NAD ⫹ derivative (9 Å hydrophilic spacer arm; total NAD ⫹, 1.86 ␮mol/g wet wt.; accessible NAD ⫹, 1.66 ␮mol/g wet wt.) and an N 6-linked immobilized NADP ⫹ derivative (9 Å hydrophilic spacer arm; total NADP ⫹, 1.0 ␮mol/g wet wt.; accessible NADP ⫹, 0.9 ␮mol/g wet wt). The irrigant was 100 mM glycine buffer (pH 8.6) containing 0.2 M KCl and indicated additions. Chromatography was performed at room temperature.

was confirmed by the observation that the enzyme failed to lock-on to an analogous N 6-linked immobilized NADP ⫹ derivative in the presence of D-phenylalanine (e.g., Fig. 7D). Reduction of the accessible N 6-linked immobilized NAD ⫹ concentration, from 1.66 to 0.23 ␮mol/g wet wt, required that the effective locking-on ligand concentration be increased to 1 mM D-phenylalanine (Fig. 8). However, the phenylalanine dehydrogenase was not locked-on to an S 6-linked immobilized NAD ⫹ derivative (accessible [NAD ⫹], 0.3/g wet wt; spacer arm length, 9 Å) in the presence of up to 100 mM D-phenylalanine (Table 4). Our earlier studies confirmed that locking-on to immobilized NAD ⫹ derivatives could be achieved at progressively lower concentrations of the soluble locking-on ligand, as the amount of accessible immobilized ligand is increased (18). This experimental observation was attributed to a decrease in K i value for the following ligand (soluble locking-on ligand) as the leading ligand concentration (immobilized cofactor) was increased (a general property of multisubstrate enzymes

following sequential ordered mechanisms of substrate binding). Thus, the differential locking-on of the three test NAD ⫹-dependent dehydrogenases to N 6-linked and S 6-linked immobilized NAD ⫹ derivatives used in this study cannot be explained in terms of the final accessible substitution levels. This suggests fundamental differences in affinity for NAD ⫹ immobilized via N 6-linkage as compared to thiol linkage. The results presented here further suggest that for at least one of the three test enzymes investigated in the present study (bakers yeast ADH), an extended spacer arm of 19.5 Å is required for the N 6-linked immobilized NAD ⫹ derivatives, if steric hindrance is to be avoided and locking-on to the immobilized cofactor is to be successful. Further studies are currently underway aimed at establishing the general applicability of the new N 6linked immobilized NAD ⫹ derivatives for use in the locking-on mode with other NAD ⫹-dependent dehydrogenases, with particular emphasis on the amino acid dehydrogenase class of enzymes.

BIOSPECIFIC AFFINITY PURIFICATION OF NAD ⫹-DEPENDENT DEHYDROGENASES

433

FIG. 8. Chromatography of L-phenylalanine dehydrogenase from Sporosarcinia sp. on an N 6-linked immobilized NAD ⫹ derivative (9 Å hydrophilic spacer arm; total NAD ⫹, 0.24 ␮mol/g wet wt.; accessible NAD ⫹, 0.23 ␮mol/g wet wt.) and N 6-linked immobilized NADP ⫹ derivative (9 Å hydrophilic spacer arm; total NADP ⫹, 1.0 ␮mol/g wet wt.; accessible NADP ⫹, 0.9 ␮mol/g wet wt.). The irrigant was 100 mM glycine buffer (pH 8.6) containing 0.2 M KCl and indicated additions. Chromatography was performed at room temperature.

ACKNOWLEDGMENTS This work was funded by the Graduate Training Programme under the Operational Programme for Industrial Development and the Higher Education Authority under their Programme for Research in Third Level Institutions. We are also grateful to Dr. Tadhg Griffin, Department of Biochemistry, National University of Galway Ireland, for helpful advice.

REFERENCES 1. Huang, T., Warsinke, A., Kuwana, T., and Scheller, F. W. (1998) Determination of L-phenylalanine based on an NADH-detecting biosensor. Anal. Chem. 70(5), 991–997. 2. Galkin, A., Kulakova, L., Yoshimura, T., Soda, K., and Esaki, N. (1997) Synthesis of optically active amino acids from alpha-keto

3.

4.

5.

6.

acids with Escherichia coli cells expressing heterologous genes. Appl. Environ. Microbiol. 63(12), 4651– 4656. Devaux-Basseguy, R., Bergel, A., and Comtat, M. (1997) Potential applications of NAD(P)-dependent oxidoreductases in synthesis: A survey. Enzymol. Microbiol. Technol. 20, 248 –258. Morishita, Y., Nakane, K., Fukatsu, T., Nakashima, N., Tsuji, K., Soya, Y., Yoneda, K., Asano, S., and Kawamura, Y. (1997) Kinetic assay of serum and urine for urea with use of urease and leucine dehydrogenase. Clin. Chem. 43(10), 1932–1936. Beckett, P. R., Hardin, D. S., Davis, T. A., Nguyen, H. V., WrayCahen, D., and Copeland, K. C. (1996) Spectrophotometric assay for measuring branched-chain amino acid concentrations: Application for measuring the sensitivity of protein metabolism to insulin. Anal. Biochem. 240, 48 –53. Kragl, U., Kruse, W., Hummel, W., and Wandry, C. (1996) En-

434

7.

8.

9.

10.

11.

12.

13.

14.

15. 16.

TYNAN ET AL.

zyme engineering aspects of biocatalysts: Cofactor regeneration as example. Biotechnol. Bioeng. 52, 309 –319. Nakamura, K., Fujii, T., Kato, Y., Asano, Y., and Cooper, A. J. (1996) Quantitation of L-amino acids by substrate recycling between an aminotransferase and a dehydrogenase: Application to the determination of L-phenylalanine in human blood. Anal. Biochem. 234(1), 19 –22. Randell, E. W., and Lehotay, D. C. (1996) An automated enzymatic method on the Roche COBAS MIRA S for monitoring phenylalanine in dried blood spots of patients with phenylketonuria. Clin. Biochem. 29(2), 133–138. Kiba, N., Oyama, Y., Kato, A., and Furusawa, M. (1996) Postcolumn co-immobilized leucine dehydrogenase-NADH oxidase reactor for the determination of branched-chain amino acids by highperformance liquid chromatography with chemiluminescence detection. J. Chromatogr. 724(1–2), 355–357. Keffler, S., Denmeade, R., and Green, A. (1994) Neonatal screening for phenylketonuria: Evaluation of an automated enzymatic method. Ann. Clin. Biochem. 31(2), 134 –139. Norbert, M., Brunhuber, W., and Blanchard, J. (1994) The biochemistry and enzymology of amino acid dehydrogenases. Crit. Rev. Biochem. Mol. Biol. 29, 415– 467. Britton, K. L., Baker, P. J., Engel, P. C., Rice, D. W., and Stillman, T. J. (1993) Evolution of substrate diversity in the superfamily of amino acid dehydrogenases. Prospects for rational chiral synthesis. J. Mol. Biol. 20(234, 4), 938 –945. Coolbear, T., Daniel, R. M., and Morgan, H. W. (1992) The enzymes from extreme thermophiles: Bacterial sources, thermostabilities and industrial relevance. Advs. Biochem. Eng. Biotechnol. 45, 57–99. Ohshima, T., and Soda, K. (1990) Biochemistry and biotechnology of amino acid dehydrogenases. Advs. Biochem. Eng. Biotechnol. 42, 187–209. Hummel, W., and Kula, M. (1989) Dehydrogenases for the synthesis of chiral compounds. Eur. J. Biochem. 184, 1–13. O’Flaherty, M., O’Carra, P., McMahon, M., and Mulcahy, P. (1999) Studies on the “stripping” ligand tactic for use with the kinetic locking-on strategy demonstrating its use in the resolu-

17.

18.

19.

20.

21.

22.

23.

24.

25. 26.

tion and purification of NAD ⫹-dependent dehydrogenases. Protein Exp. Purif. 16, 424 – 431. O’Flaherty, M., McMahon, M., Forde, J., and Mulcahy, P. (1999) The kinetic locking-on strategy for bioaffinity purification: Further studies with bovine liver glutamate dehydrogenase. Protein Exp. Purif. 16, 276 –297. Mulcahy, P., O’Flaherty, M., McMahon, M., and Oakey, L. (1999) Effect of accessible immobilized NAD ⫹ concentration on the bioaffinity chromatographic behavior of NAD ⫹-dependent dehydrogenases using the kinetic locking-on strategy. Protein Exp. Purif. 16, 261–275. Oakey, L., Martung, A., McMahon, M., O’Flaherty, M., and Mulcahy, P. (1999) A comparison of biospecific affinity chromatographic methodologies for the purification of NAD ⫹-dependent dehydrogenases; Studies with bovine L-lactate dehydrogenase. Process Biochem. 35, 349 –357. O’Flaherty, M., McMahon, M., and Mulcahy, P. (1999) A kinetic locking-on strategy for bioaffinity purification: Further studies on alcohol dehydrogenase. Protein Exp. Purif. 15, 127–145. Mulcahy, P., Cullina, A., and O’Carra, P. (1997) Both D- and L-lactate-specific dehydrogenases coexist in individual cephalopods. Comp. Biochem. Phys. [A], 116(2), 143–148. Mulcahy, P., Griffin, T., and O’Carra, P. (1997) Biospecific affinity purification of octopine dehydrogenase from molluscs. Protein Exp. Purif. 9(1), 109 –114. O’Carra, P., Griffin, T., O’Flaherty, M., Kelly, N., and Mulcahy, P. (1996) Further studies on the biospecific affinity chromatographic purification of NAD-dependent dehydrogenases using the locking-on tactic. Biochim. Biophys. Acta 1297, 235–243. Lindberg, M., Larsson, P-O., and Mosbach, K. (1973) A new immobilized NAD ⫹ analogue: Its application in affinity chromatography and as a functioning coenzyme. Eur. J. Biochem. 40, 187–193. Buchanan, M. (1988) M.Sc. Thesis, National University of Ireland, Dublin. Mosbach, K. (1974) AMP and NAD ⫹ as general ligands in “Methods in Enzymology” (Jakoby, W. B., and Wilcheck, M., Eds.), Vol. XXXIV, pp. 229 –242, Academic Press, New York.