Evidence that NADP+ is the physiological cofactor of ADP-L-glycero-D-mannoheptose 6-epimerase.

ADP-L-glycero-D-mannoheptose 6-epimerase is required for lipopolysaccharide inner core biosynthesis in several genera of Gram-negative bacteria. The enzyme contains both fingerprint sequences Gly-X-Gly-X-X-Gly and Gly-X-X-Gly-X-X-Gly near its N terminus, which is indicative of an ADP binding fold. Previous studies of this ADP-l-glycero-D-mannoheptose 6-epimerase (ADP-hep 6-epimerase) were consistent with an NAD(+) cofactor. However, the crystal structure of this ADP-hep 6-epimerase showed bound NADP (Deacon, A. M., Ni, Y. S., Coleman, W. G., Jr., and Ealick, S. E. (2000) Structure 5, 453-462). In present studies, apo-ADP-hep 6-epimerase was reconstituted with NAD(+), NADP(+), and FAD. In this report we provide data that shows NAD(+) and NADP(+) both restored enzymatic activity, but FAD could not. Furthermore, ADP-hep 6-epimerase exhibited a preference for binding of NADP(+) over NAD(+). The K(d) value for NADP(+) was 26 microm whereas that for NAD(+) was 45 microm. Ultraviolet circular dichroism spectra showed that apo-ADP-hep 6-epimerase reconstituted with NADP(+) had more secondary structure than apo-ADP-hep 6-epimerase reconstituted with NAD(+). Perchloric acid extracts of the purified enzyme were assayed with NAD(+)-specific alcohol dehydrogenase and NADP(+)-specific isocitric dehydrogenase. A sample of the same perchloric acid extract was analyzed in chromatographic studies, which demonstrated that ADP-hep 6-epimerase binds NADP(+) in vivo. A structural comparison of ADP-hep 6-epimerase with UDP-galactose 4-epimerase, which utilizes an NAD(+) cofactor, has identified the regions of ADP-hep 6-epimerase, which defines its specificity for NADP(+).

contained the fingerprint sequence Gly-X-Gly-X-X-Gly or Gly-X-X-Gly-X-X-Gly, which is characteristic of the ADP binding ␤␣␤␣␤ fold (or Rossmann fold) associated with NAD(P) binding and also FAD-binding proteins (4,5). In addition, it was observed that one nicotinamide coenzyme was tightly bound to each ADP-hep 6-epimerase subunit; enzymatic analyses suggested that NAD ϩ was the cofactor used by ADP-hep 6-epimerase (6). ADP-L-glycero-D-mannoheptose 6-epimerase showed 24% sequence identity with UDP-galactose 4-epimerase (UGE) based on calculation with BLAST (7). The crystal structure of UGE showed two NAD molecules bound to the dimeric enzyme (8,9). However, the crystal structure of ADP-hep 6-epimerase (10) indicated that NADP ϩ is the more likely natural cofactor for this enzyme, because only NADP was found on each of the five subunits of ADP-hep 6-epimerase. The conformation of NADP bound to ADP-hep 6-epimerase is less extended than that of NAD bound to UGE as judged by the distance between adenine C6 and C2 of the nicotinamide ring.
Here we report that the enzyme ADP-L-glycero-D-mannoheptose 6-epimerase is an NADP ϩ -dependent enzyme. NAD ϩ can substitute for NADP ϩ , but enzymatic activity is reduced. Several important structural differences between ADP-hep 6-epimerase and UGE give rise to the different cofactor specificities.
Expression and Purification of ADP-hep 6-Epimerase-Plasmid pJP6 containing the ADP-hep 6-epimerase gene was transformed into the E. coli strain BL21 (DE3) plysS. The transformed bacteria were cultured in LB medium. Cells were harvested 3 h after induction with 0.4 mM isopropyl-␤-D thiogalactopyranoside. Buffers and crude extracts from bacterial cell pellets were prepared as described previously by Pegues et al. (4). ADP-hep 6-epimerase was purified to homogeneity employing the following three-step purification protocol: a hydrophobic interaction and an affinity chromatographic step as described (6), followed by an additional ion exchange step (11).
ADP-hep 6-Epimerase Activity Assay-The activity of ADP-hep 6-epimerase was assayed as described previously (2,6). The reaction mixture containing 0.1 M Tris acetate, pH 8.5, ADP-D-glycero-D-mannoheptose (5 nmol), and enzyme in a final volume of 50 l was incubated at 37°C for 30 min. The reaction was terminated by heating for 5 min at 100°C. Enzyme activity was determined by monitoring the formation of ADP-L-glycero-D-mannoheptose by high performance liquid chromatography (2, 6). 1 unit of enzyme activity is defined as the ADP-hep 6-epimerase activity capable of producing 1 nmol of ADP-L-glycero-Dmannoheptose in 30 min at 37°C in 0.05 ml of reaction mixture.
Preparation and Reconstitution of Apoenzyme-ADP-hep 6-epimerase was treated with saturated acidic ammonium sulfate (adjusted to pH 2.7 with concentrated sulfuric acid) containing 5 mM dithiothreitol on ice, as described by Gomi et al. (12). For reconstitution of ADP-hep 6-epimerase, the acidic ammonium sulfate-treated apo-ADP-hep 6-epi-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ To whom correspondence should be addressed. 1 The abbreviations used are: ADP-hep 6-epimerase and AGME, merase was dialyzed against 50 mM sodium phosphate buffer, pH 7.0, at 4°C for 3 h or longer. Subsequently, the dialyzed apoenzyme was incubated with 100 or 200 M NADP ϩ , NAD ϩ , or FAD at room temperature for 1 h. Unbound NAD ϩ , NADP ϩ , or FAD was removed from the reconstituted enzyme by dialysis against 50 mM sodium phosphate buffer, pH 7.0, at 4°C for 3 h or longer. K d Determinations-The values of K d for NADP ϩ and NAD ϩ were determined using freshly prepared apoenzyme. Apoenzyme was incubated for 1 h at room temperature with concentrations of NAD or NADP of 40, 80, 120, 160, and 200 M. The reconstituted enzyme samples, following dialysis for 5 h at 4°C (as described above), were then added to the standard assay system, which contained 2.7 nmol of ADP-Dglycero-D-mannoheptose. Apparent K d values for NAD ϩ and NADP ϩ and V max were evaluated by reciprocal plots of induced activity against nucleotide concentrations.
Enzymatic Identification of the ADP-hep 6-Epimerase Cofactor-NADP ϩ -specific isocitric dehydrogenase and NAD ϩ -specific alcohol dehydrogenase were used to define which nucleotide was the ADP-hep 6-epimerase cofactor. The cofactor was extracted by adding 10 l of 35% perchloric acid to 500 g of holo-ADP-hep 6-epimerase in 90 l of 10 mM Tris-HCl buffer, pH 7.0. After standing on ice for 20 min, the extracts were adjusted to pH 7.0 with saturated NaHCO 3 . The precipitate in the perchloric extracts was removed by centrifugation at 6000 ϫ g for 5 min. The supernatant solutions were used to assay the nature and amount of released coenzyme. To assay for NADP the reaction mixture consisted of 60 l of supernatant solution, 6 l of NADP ϩ -specific isocitrate dehydrogenase (1 mg/ml, 9.5 units/mg), 5 l of 0.04 M isocitrate, and 50 mM sodium phosphate buffer, pH 7.0. The NAD assay mixture contained 60 l of supernatant solution, 6 l of NAD ϩ -specific alcohol dehydrogenase (1 mg/ml, 440 units/mg), 5 l of 100% ethanol, and 50 mM sodium phosphate buffer, pH 8.8. The final volume for both assays was 120 l. The formation of reduced NAD or NADP was measured at 340 nm at room temperature.
Chromatographic Identification of ADP-hep 6-Epimerase-bound Cofactor-ADP-hep 6-epimerase-bound cofactor, NADP(H), was identified and detected by HPLC analysis of the perchloric acid extract of the enzyme by a modification of the method of Daussmann et al. (13). Enzyme solution (50 l) containing 150 g of purified ADP-hep 6-epimerase was extracted with 50 l of 5% perchloric acid at 0°C for 20 min. Saturated NaHCO 3 was then added to the extract to bring the pH value to 7.0. After neutralization, the mixture was centrifuged at 14,000 ϫ g for 10 min, and the supernatant was used for analysis. Quantitative determination of NADP ϩ and NADPH was carried out by reversed phase HPLC (Amersham Pharmacia Biotech Smart System). The column (Hypersil ODC-5 M; 125 ϫ 4-mm diameter; Hypersil Inc.) was equilibrated in 20 mM sodium phosphate, pH 5.0. A gradient of 20 mM sodium phosphate, pH 5.0, and acetonitrile (13) resulted in unambiguous resolution of NADP ϩ and NADPH at 12.2 and 15.7 min, respectively. The standards used in the HPLC analysis were subjected to the same treatments as the perchloric acid-extracted materials. Ultra-

FIG. 2.
A, cofactor analyses. The presence of NADP ϩ in perchloric acid extracts of ADP-hep 6-epimerase was monitored by the NADP ϩspecific isocitric dehydrogenase. The reaction mixture contained 5 l of 0.04 M isocitrate and 6 l of isocitric dehydrogenase (1 mg/ml, 9.5 units/mg). Perchloric acid extract (300 g of enzyme protein equivalent in 60 l) was added at 0 min. NAD ϩ (50 M) and NADP ϩ (50 M) were added at 40 and 50 min, respectively, as controls. The activity of isocitrate dehydrogenase was monitored spectrophotometrically at 340 nm. B, the presence of NAD ϩ was monitored with NAD ϩ -specific alcohol dehydrogenase. The reaction mixture contained 5 l of ethanol, and 6 l alcohol dehydrogenase (1 mg/ml, 440 unit/mg). 60 l of perchloric acid extract (300 g of enzyme protein equivalent in 60 l) was added at time 0. NADP ϩ (50 M) and NAD ϩ (50 M) were added at 15 and 30 min, respectively, as controls.  Table II). violet circular dichroism spectra of holo-ADP-hep 6-epimerase, apo-ADP-hep 6-epimerase, apo-ADP-hep 6-epimerase ϩ NADP ϩ , and apo-ADP-hep 6-epimerase ϩ NAD ϩ were measured in a Jasco J715 spectropolarimeter, using a 1-mm path length quartz cuvette at 25°C.
Smoothed spectra from 4 scans were analyzed in terms of secondary structure, using the CONTIN program (14).
Structural Comparison with UGE-The structure of ADP-hep 6-epimerase was determined recently by x-ray crystallography (10). A single molecule of ADP-hep 6-epimerase (PDB code 1EQ2) was superimposed on a molecule of UGE (PDB code 1XEL) using the graphics program O (15). Only core residues from the cofactor-binding domain were used in the superposition, resulting in a root mean square deviation of 1.62 Å for 124 ␣-carbon atoms.

RESULTS
Reconstitution of apo-ADP-hep 6-Epimerase with Adeninebased Dinucleotides-To identify the cofactor preference of ADP-hep 6-epimerase, we used NAD ϩ , NADP ϩ , and FAD to reconstitute apo-ADP-hep 6-epimerase. Enzyme activities of the reconstituted enzyme were detected by the standard ADPhep 6-epimerase assay as described under "Experimental Procedures." As shown in Table I, the apo-ADP-hep 6-epimerase is inactive. However, when 100 M NADP ϩ was used to reconstitute the apoenzyme, 88% of the native ADP-hep 6-epimerase activity was restored. Restoration of 40% of ADP-hep 6-epimerase activity was observed when 100 or 200 M NAD ϩ was used to reconstitute the apoenzyme. Thus, the reactivation of apo-ADP-hep 6-epimerase by NAD ϩ is about 50% of that achieved by NADP ϩ . In contrast, no restoration of ADP-hep 6-epimerase activity was achieved in the presence of FAD. These results demonstrate a clear preference among the selected adenine dinucleotides for reconstituting ADP-hep 6-epimerase, which probably reflects the cofactor preference of ADPhep 6-epimerase.
Enzymatic Identification of the in Vivo ADP-hep 6-Epimerase-bound Cofactor-NADP ϩ -specific isocitric dehydrogenase and NAD ϩ -specific alcohol dehydrogenase were employed to assay the perchloric acid extracts from purified ADP-hep 6-epimerase. When the perchloric acid extract was incubated with isocitric acid and the NADP ϩ -specific isocitric dehydrogenase, there was a marked increase in absorbance at 340 nm ( Fig. 2A), which reached a plateau after 30 min. As a control, either 50 M NAD ϩ or 50 M NADP ϩ was added after the initial reaction reached the plateau. When NAD ϩ was added to the reaction mixture at 40 min there was no increase in absorbance at 340  nm. However, when 50 M NADP ϩ was added to the reaction cuvette, a dramatic increase in absorbance at 340 nm was observed. The opposites were observed when the released cofactor was analyzed with the NAD-specific ethanol/alcohol dehydrogenase system (Fig. 2B). There was no increase in absorbance at 340 nm when the perchloric acid extract of ADP-hep 6-epimerase was added upon addition of NADP ϩ to the reaction cuvette. However, when NAD ϩ was added to the reaction vessel, a significant increase in absorbance at 340 nm was observed. These results provide evidence that NADP ϩ is the major dinucleotide cofactor present in the perchloric acid extracts. HPLC Analysis of the Perchloric Acid Extracts of ADP-hep 6-Epimerase-To confirm the presence of NADP, the neutralized perchloric acid extracts of purified ADP-hep 6-epimerase were analyzed by HPLC as described under "Experimental Procedures" (Fig. 3). Two clearly resolved peaks (Fig. 3) were observed when the perchloric acid extracts were analyzed. The chromatographic analysis showed that the molar ratio of NADP ϩ /NADPH in the extracts was 0.4:0.6. The occurrence of both oxidized and reduced forms of the cofactor bound to ADPhep 6-epimerase was consistent with earlier UV and fluorescence studies (4). Our study also suggested that NADP(H) is stable during perchloric acid extraction and that the coenzyme is tightly, but non-covalently, bound to the enzyme.
Secondary Structure of Native and Reconstituted ADP-hep 6-Epimerase-Circular dichroism spectroscopy, which is sensitive to secondary structural changes, was used to evaluate the secondary structure of native ADP-hep 6-epimerase and apo-ADP-hep 6-epimerase reconstituted with either NAD or NADP. The estimated secondary structures of various protein forms, apo, holo, NAD, or NADP, bound are shown in Fig. 4. These data are clearly in agreement with the enzymatic activity assays (see below). Both cofactors can partially restore the structure of the protein. However, NADP is more effective than NAD. The circular dichroism experiments showed that no more refolding occurred after the standard reactivation procedure. Consequently, the data derived from the enzymatic assays are characteristic of the reconstituted proteins. They do not reflect differences in the rate of reconstitution by the two cofactors.
K d Determinations-The determination of the K d values for NADP ϩ and NAD ϩ provided further convincing evidence for the coenzyme preference. As shown in Table II, the apparent K d values for NADP ϩ and NAD ϩ were 26 and 45 M, respectively, showing a higher binding affinity for NADP ϩ . These are apparent K d values, because they are determined by measurement of enzymatic activity and not by direct measurement of binding.

FIG. 5. Structural comparison of NADP-dependent ADP-hep 6-epimerase and NAD-dependent UGE.
A, ribbon diagram of ADP-hep 6-epimerase with bound NADP and substrate analog inhibitor ADP-glucose indicated in ball and stick representation. Lighter-colored regions (yellow) correspond to structural differences with UGE that define NAD/NADP specificity. B, ribbon diagram of UGE with bound NAD and substrate indicated in ball and stick representation. Lighter colored regions (cyan) correspond to structural differences with ADP-hep 6-epimerase that define NAD/NADP specificity. C, stereo diagram of the NADP-binding site (color scheme as in A and B). A high degree of structural and sequence conservation is observed for many of the cofactor binding residues (blue and red); however, large differences are observed in the region around the 2Ј-phosphate (yellow and cyan).

Structural Basis for ADP-hep 6-Epimerase NADP Binding
Preference-The recently determined crystal structure of ADPhep 6-epimerase (10) (Fig. 5A) closely resembles the structure of UGE (Fig. 5b) (8,9). The two enzymes, both from E. coli, share the same fold and 24% sequence identity. Despite the obvious structural similarity, UGE utilizes only NAD ϩ as its cofactor, whereas ADP-hep 6-epimerase prefers NADP ϩ in vitro. There are several important structural features that define this modified cofactor specificity. It can be seen that the largest structural difference in the N-terminal domain (i.e. the dinucleotide-binding domain) occurs in the region immediately surrounding the 2Ј-phosphate in the ADP-hep 6-epimerase structure ( Fig. 5A; see arrow). There are five additional amino acid residues present in the UGE structure in the polypeptide chain from residues 30 to 74, which correspond to residues 30 to 69 (Fig. 5B) in ADP-hep 6-epimerase. These extra residues in UGE extend the conformation of the interconnecting loop, between the secondary structure elements, away from the dinucleotide and thus prevent any direct interaction with a 2Јphosphate. They also allow significant changes in the orientation of one of the short helices. On closer inspection this region contributes all the specific residues that participate in coordinating the 2Ј-phosphate in ADP-hep 6-epimerase (Fig.  5C). Although there is a high degree of structural and sequence homology for much of the nicotinamide dinucleotide-binding site, there is little similarity around the 2Ј-phosphate in ADPhep 6-epimerase (Table III). Thus, as shown in Fig. 5C, the NAD/NADP binding fingerprint motifs of the two proteins match very closely (Gly 9 and Ile 11 in ADP-hep 6-epimerase are shown), as well as the catalytically important residues (Ser 116 , Tyr 140 , and Lys 144 in ADP-hep 6-epimerase). However, major differences occur where there are positively charged residues (Lys 38 and particularly Lys 53 ) in ADP-hep 6-epimerase that play an important role in conferring the additional NADPϩ specificity. Lys 53 extends along and forms a large hydrophobic contact with the adenine base, and it also then coordinates the 2Ј-phosphate through its side chain-terminal side-chain nitrogen atom. The orientation of Lys 53 precludes any direct H-bond interactions with the adenosine N1 and N3, which are present in UGE and also serves to allow Asn 32 to interact directly with the 2Ј-phosphate (Table III). DISCUSSION Many dinucleotide-binding proteins specifically require either NAD ϩ or NADP ϩ as cofactor, although some of them show a dual specificity (5, 16 -20). Our study provides strong evidence that NADP ϩ is the natural coenzyme of ADP-hep 6-epimerase, although NAD ϩ can substitute for NADP ϩ and still allow the epimerization reaction to proceed at a slower rate. The apparent dissociation constant determined for NADP ϩ versus that of NAD ϩ is consistent with preferential binding of NADP to ADP-hep 6-epimerase.
Since the 1970s, there has been substantial interest in elucidating the fundamental basis for NAD ϩ /NADP ϩ specificity of many enzymes. The fingerprint sequence Gly-X-Gly-X-X-Gly (20) has been recognized as the stereotypic hallmark of dinucleotide binding, and therefore it is not surprising that the study of NAD ϩ /NADP ϩ specificity has focused predominately on this sequence.
To investigate the coenzyme preference of ADP-hep 6-epimerase, we have linked the biochemical and molecular modeling studies with information derived from our recent threedimensional structural analysis. This combined approach has allowed us to (1) determine kinetic parameters that indicated a preference for NADP by ADP-hep 6-epimerase and (2), the detailed analysis of the amino acids and structural attributes of enzyme that confer preference for nicotinamide adenine dinucleotide with or without an additional 2Ј-phosphate. The structural studies of ADP-L-glycero-D-mannoheptose 6-ADP-hep 6-epimerase indicate that positively charged Lys 38 and Lys 53 are the major contributors to the electrostatic compensation for the 2Ј-phosphate group of NADP. The occurrence of these positively charged residues conferred the preference of the enzyme for NADP ϩ . This is consistent with the results of Scrutton et al. (19), which demonstrated that the introduction of positively residues (Arg 198 and Arg 204 ) in the NAD ϩ -binding site of E. coli glutathione reductase conferred NADP ϩ binding preference to the redesigned enzyme. Rizzi et al. (21) reported in a recent structural study of GDP-4-keto-6-deoxy D-mannose epimerase/ reductase that in its NADP ϩ -binding site two positively charged residues (Arg 12 and Arg 36 ) play important roles in providing electrostatic compensation for the NADP ϩ ribose 2Ј-phosphate group.
Further, residues (Gly 9 and Ile 11 ) present in the Gly-X-X-Gly-X-X-Gly motif (21) are positioned near the diphosphate bridge of the ADP-hep 6-epimerase-bound NADP. Residues Ser 116 , Tyr 140 , and Lys 144 , thought to be involved in the catalytic mechanism of ADP-hep 6-epimerase, are located near the nicotinamide ring and its attached ribose. Thus, there exists a TABLE III A comparison of the interactions between NADP with ADP-L-glycero-D-mannoheptose 6-epimerase and NAD with UDP-galactose 4-epimerase NAD/NADP atom name Interaction Protein residue/atom in AGME Protein residue/atom in UGE significant degree of structural and sequence similarity for the NAD(P) ϩ -binding sites of ADP-hep 6-epimerase and other NAD(P)-binding proteins. Further mutagenesis studies will be used to provide additional insight into the cofactor specificity of ADP-hep 6-epimerase and to determine in more detail the significance of the residues involved.