The catalytic activities of the bifunctional Azotobacter vinelandii mannuronan C-5-epimerase and alginate lyase AlgE7 probably originate from the same active site in the enzyme.

The Azotobacter vinelandii genome encodes a family of seven secreted Ca(2+)-dependent epimerases (AlgE1--7) catalyzing the polymer level epimerization of beta-D-mannuronic acid (M) to alpha-L-guluronic acid (G) in the commercially important polysaccharide alginate. AlgE1--7 are composed of two types of protein modules, A and R, and the A-modules have previously been found to be sufficient for epimerization. AlgE7 is both an epimerase and an alginase, and here we show that the lyase activity is Ca(2+)-dependent and also responds similarly to the epimerases in the presence of other divalent cations. The AlgE7 lyase degraded M-rich alginates and a relatively G-rich alginate from the brown algae Macrocystis pyrifera most effectively, producing oligomers of 4 (mannuronan) to 7 units. The sequences cleaved were mainly G/MM and/or G/GM. Since G-moieties dominated at the reducing ends even when mannuronan was used as substrate, the AlgE7 epimerase probably stimulates the lyase pathway, indicating a complex interplay between the two activities. A truncated form of AlgE1 (AlgE1-1) was converted to a combined epimerase and lyase by replacing the 5'-798 base pairs in the algE1-1 gene with the corresponding A-module-encoding DNA sequence from algE7. Furthermore, substitution of an aspartic acid residue at position 152 with glycine in AlgE7A eliminated almost all of both the lyase and epimerase activities. Epimerization and lyase activity are believed to be mechanistically related, and the results reported here strongly support this hypothesis by suggesting that the same enzymatic site can catalyze both reactions.

alginate, a 1,4-linked linear polysaccharide consisting of ␤-Dmannuronic acid (M), 1 and its C-5 epimer, ␣-L-guluronic acid (G). The M-and G-moieties are distributed as heteropolymeric (MG) and/or homopolymeric (MM or GG) blocks in the polymers (10 -15). The alginates appear to be essential to the differentiation process, since non-mucoid strains fail to encyst (16 -18). A. vinelandii in addition synthesizes alginates during vegetative growth. This vegetative capsule seems to be responsible for bacterial adhesion to surfaces and may function as a diffusion barrier against O 2 or heavy metals (19). Bacterial alginates are O-acetylated in varying degrees at O-2 and/or O-3 on mannuronic acid moieties (20 -22).
In all studied alginate-producing organisms the epimerization of M to G is catalyzed by mannuronan C-5-epimerases at the polymer level (23)(24)(25)(26)(27)(28)(29). The A. vinelandii genome encodes a family of seven secreted Ca 2ϩ -dependent epimerases (AlgE1-7) (29) and a Ca 2ϩ -independent periplasmic epimerase (AlgG) (28) similar to the epimerase (AlgG) in Pseudomonas aeruginosa (25). The AlgE epimerases consist of varying numbers of two types of structural modules, A and R, of which the A-module alone is capable of catalyzing epimerization and determining the final distribution of G-moieties in the produced alginates (30). The R-modules contain four to seven Ca 2ϩ -binding motifs, and the direct binding of Ca 2ϩ by this module, as well as by the A-module, has been demonstrated previously (30). The R-modules stimulate reaction rates and are probably involved in export of the AlgE proteins (23,31,32). By producing a family of alginate epimerases, A. vinelandii is able to synthesize a variety of alginates with different physicochemical properties, the exact biological function of which is not yet clearly understood.
The genomes of alginate-synthesizing organisms encode alginases that degrade the polymers at specific uronic acid sequences, but most bacterial lyases display low activity against the native highly acetylated polymers (33)(34)(35). The alginases associated with A. vinelandii are reported to cleave M-rich alginates most effectively (36 -39), even though the periplasmic AlgL enzyme degrades M-G bonds with similar efficiency as M-M bonds (36). Interestingly, the AlgE7 protein displays both epimerase and lyase activity (29). In 1987 Gacesa (40) proposed that the reaction mechanism of these two enzyme activities were very similar (Fig. 1). Both mechanisms include removal of H-5 of the uronic acid, but instead of a replacement of this proton from the opposite side of the sugar ring as in epimeri-zation, the alginases catalyze a ␤-elimination of the 4-O-glycosidic bond, generating unsaturated 4-deoxy-L-erythro-hex-4enepyranosyluronate moieties at the non-reducing end. Based on stereochemical considerations, Feingold and Bentley (41) predicted that a similar reaction mechanism could apply to both M-specific and G-specific lyases. If these theoretical considerations are correct it seemed possible that the AlgE7 lyase and epimerase activities originate from the same active site in the enzyme. In this paper we examine the relationship between these two reactions mediated by AlgE7, and the results strongly support the hypothesis that a common site is responsible for both activities.

EXPERIMENTAL PROCEDURES
Growth of Bacteria-Escherichia coli cells were grown in L-broth with shaking (unless otherwise stated) or on L-agar at 37°C (42). When relevant the E. coli media were supplemented with 200 g/ml ampicillin unless otherwise stated. For enzyme purification, the E. coli cells were grown in a medium containing 30 g/liter tryptone, 15 g/liter yeast extract, and 5 g/liter NaCl (3ϫ concentrated L-broth).
Standard Laboratory Methods-Restriction endonuclease digestions, ligations, and agarose gel electrophoresis were performed as described by Sambrook et al. (42). Transformations were performed according to Chung et al. (43). Plasmids encoding AlgE6 (29), AlgE7 (29), AlgE7-E1, AlgE1-E7, and AlgE7-D152G were propagated in E. coli SURE (44), whereas pHH5 (45), encoding AlgE5, was propagated in JM109 (46). Plasmids were isolated by a midi kit from Qiagen (for sequencing) or a Wizard miniprep kit from Promega. DNA sequencing for verification of plasmid constructs was performed by using the ABI PRISM TM Dye Terminator Cycle Sequencing Ready Reaction kit (PerkinElmer Life Sciences) and an Applied Biosystems Model 373 automatic sequencer. Protein concentrations were measured using the Bio-Rad Coomassie Brilliant Blue-based protein assay using bovine serum albumin as standard.
Plasmid Constructions and Site-specific Mutagenesis-AlgE1 was previously shown to consist of two separate catalytic domains, of which the N-terminal part, AlgE1-1, introduces both G-and MG-blocks into the alginate substrates (encoded by pHE37) (47). AlgE1-1 has the same modular structure as AlgE7 (AR1-3), and its epimerization products have been extensively characterized (47). The expression vectors pHE37 and pBG27 (encodes AlgE7 (29)) were digested by NcoI and BglII, and the 5Ј-798 bp of algE1-1 were substituted with the same A-module encoding part of algE7 and vice versa. The resulting plasmids, pBG70 and pBG71, encode and express the hydrid enzymes AlgE7-E1 (N-terminal 266 amino acids from AlgE7) and AlgE1-E7 (N-terminal 266 amino acids from AlgE1), respectively.
Alginates Used as Substrates-Acetylated mannuronan was isolated from the epimerase negative P. aeruginosa strain FRD462 (48). A deacetylated (20) preparation of this alginate was used in all analyses involving mannuronan. For the NMR experiment referred to in Fig. 6, the mannuronan substrate was degraded to an average degree of polymerization (DP n ) ϳ 100 by mild acid hydrolysis (12) before incubation with AlgE7. Polyalternating alginate (MG-alginate) was produced in vitro from mannuronan by using recombinantly produced AlgE4. Since AlgE5 and AlgE6 do not display lyase activity, this MG-alginate was degraded to a DP n ϭ 30 -40 before incubation with these two enzymes (NMR analyses). The highly acetylated M-rich alginate was isolated from P. aeruginosa 8830 (49), and the acetylated G-rich alginate was isolated from vegetatively growing A. vinelandii E (26,50). Deacetylated versions (20) of these two alginates were also used. The Macrocystis pyrifera alginate is a commercially available high molecular weight alginate from Sigma. The G-block alginate was isolated from the outer cortex of Laminaria hyperborea stipes and degraded to DP n ϭ 30 -40 by acidic hydrolysis and fractionation (11) before incubation with AlgE7 and AlgE7-E1. 5-3 H-Labeled M-rich alginate (specific activity 144330 dpm/mg alginate) was prepared by growing P. aeruginosa 8830 (49) in a medium containing 5-3 H-labeled glucose (28). The composition of the alginate substrates used are summarized in Table I.
Preparation of Enzyme Extracts-E. coli cells containing the plasmid of interest were grown overnight in 3ϫ concentrated L-broth supplemented with 400 g/ml ampicillin. The cultures were diluted 1:100 in the same medium, and after 3 h of incubation the production of enzymes were induced by adding isopropyl-␤-D-thiogalactopyranoside at a concentration of 0.5 mM. The cells were harvested 3-4 h after induction and resuspended in 1 ⁄10 of the culture volume in MC buffer (MOPS (pH 6.9), 2.2 mM CaCl 2 ) and then disrupted by ultrasonication. After centrifugation at 27,000 ϫ g for 30 min, the crude cell extracts were filtered through a 0.2-m Millipore filter before purification by ion exchange chromatography.
The crude cell extracts were loaded onto a HiTrap Q Sepharose high performance column (Amersham Pharmacia Biotech) equilibrated with MC buffer. All tested proteins were eluted by a continuous NaCl gradient (0 to 1 M NaCl in MC buffer); AlgE7 was eluted between 36 and 39 mM NaCl, AlgE7-E1 was eluted between 33 and 38 mM NaCl, AlgE5 was eluted between 44 and 48 mM NaCl, and AlgE6 was eluted between 30 and 35 mM NaCl.
Measurements of Enzymatic Activities by Isotope Assays (Lyase and Epimerase) and Spectrophotometry (Lyase)-A radioisotope assay (29) was used for detection of epimerase and lyase activities in the crude extracts (AlgE7, AlgE7-E1, AlgE1-E7, AlgE7-D152G, AlgE5, and AlgE6) and for identification of active enzyme fractions after purification with ion exchange chromatography (AlgE7, AlgE7-E1, AlgE5, and AlgE6). The definition of the specific activity of AlgE7 and AlgE7-E1 (Fig. 2) is complex since two reactions (epimerization and degradation) run simultaneously, and both reactions are influenced by changes in the structure of the substrate (F G increases) as the reactions proceed. The two reactions cannot be easily quantified separately also because H-5 may be released twice from the same moieties after epimerization followed by cleavage at the same site. Measured radioactivity (dpm) in the supernatant (in a scintillation counter) after incubation and precipitation of alginate (29) includes, besides released 3 H from alginate to water upon epimerization/degradation, also 3 H in unprecipitated small oligomers produced by the lyase. The specific activities for AlgE7 and AlgE7-E1 were therefore defined as unprecipitated radioactivity (dpm/  (40)). AA1-3 refer to amino acid residues on the enzyme. The negative charge on the carboxylate anion is believed to be neutralized by a positively charged amino acid (AA1) in the active site of the enzyme. Note that the abstraction of the H-5 proton is believed to occur from below the sugar plane, whereas the replacement occurs from above. For simplicity the protons are omitted from the sugar rings. min)/mg of protein. The incubation time used (85 min) in the above calculations was based on reactions that were run to less than 10% completion. Generally we used samples without added enzyme as blanks (negative controls) in the experiments, but to get a relatively accurate estimate of the low activity of the AlgE7-D152G mutant protein it was found necessary to use the same cells without the algE7 gene inserted into the plasmid (pTrc99A) as negative controls. By also using the same amount of added extract for the mutant and the negative control we were able to design the experiments such that mutant activities were about 600 dpm, whereas those of the control extracts were around 50 dpm. Mutant activities also responded linearly to reductions in the amounts of added extract (values below 600 dpm), whereas the negative controls remained nearly constant.
Alginate lyases produce unsaturated 4-deoxy-L-erythro-hex-4enepyranosyluronate moieties (⌬) as a result of degradation of the alginate substrates, and these unsaturated uronic acids strongly absorb UV light at 230 nm (51). Therefore, the lyase reaction was measured as a continuous increase in absorbance at 230 nm at room temperature by adding enzyme extract to a mixture of alginate (0.91 mg/ml) in 50 mM Tris-HCl buffer (pH 7.1) and 2 mM CaCl 2 (unless otherwise stated). The total reaction volume was 1.1 ml. Lyase activity is given as the increase in absorbance units per minute.
Determination of the Dependence of the AlgE7 Lyase Activity on Divalent Cations, pH, and Ionic Strength-The lyase activity of AlgE7 was assayed in the pH range 6.5-8.8. The buffers used were MOPS (pH 6.5-7.0; pH adjusted with NaOH) and Tris-HCl (pH 7.0 -8.8; pH adjusted with HCl). The pH in each reaction was measured in the cuvette after the lyase reactions were stopped. Before measurements of the lyase activity as a function of divalent cations and NaCl concentrations, the AlgE7 preparation was dialyzed extensively against 20 mM MOPS (pH 6.9) and then concentrated against 20% polyethylene glycol (M r ϭ 20 000). For the substrate specificities and NMR analyses, the enzyme was not dialyzed, resulting in final NaCl concentrations of ϳ34 mM (substrate specificity analyses) and 10 mM (NMR spectroscopy).
Measurements of Epimerase and Lyase Activities by 1 H NMR Spectroscopy-Samples for characterization by NMR spectroscopy were incubated in a total volume of 7.2 ml containing the relevant enzyme, 8 mg of alginate in 50 mM Tris-HCl buffer (pH 7.1), and 2 mM CaCl 2 at 37°C. When using the G-block alginate (AlgE7 and AlgE7-E1), the acetylated M-rich P. aeruginosa alginate (AlgE7), and the degraded (in vitro) MG-alginate (AlgE5 and AlgE6), the reactions were stopped by adding Na 2 EDTA to a concentration of 10 mM (removes remaining Ca 2ϩ ). The samples were then dialyzed against deionized water (dH 2 O) before NMR spectroscopy. The acetylated alginate from P. aeruginosa was, in addition, deacetylated (20) after incubation with AlgE7 but before NMR. No other reaction mixtures involving AlgE7 and AlgE7-E1 were dialyzed before NMR spectroscopy due to the degradative activity of these enzymes and, hence, potential loss of small oligosaccharides through the dialysis membrane. The AlgE7-D152G mutant protein (crude extract) was incubated (24 h) with mannuronan and the M. pyrifera alginate. The reactions were then stopped and dialyzed against deionized water (dH 2 O). The alginates were then degraded by mild acid hydrolysis (DP n ϳ30 -40) (12) before NMR spectroscopy. All substrates and products were freeze-dried and dissolved in D 2 O (10 mg/ml at pD 6.8). The samples were then analyzed by NMR spectroscopy using a Bruker AM-300 (300 MHz) spectrometer at 90°C. The internal stand-ard used was 3-(trimethylsilyl)-propionic-2,2,3,3-d4-acid, Na ϩ salt (Aldrich).
For the specific NMR experiment referred to in Fig. 6 the AlgE7 reaction on mannuronan was performed in an NMR tube inside the spectrometer. 12.0 mg of mannuronan (DP n ϳ100) was dissolved in 400 l of D 2 O. 300 l of this solution was transferred into a 5-mm NMR sample tube, and CaCl 2 in D 2 O, Tris-HCl buffer (50 mM, pH 7.0 at 49°C), and 5 l of 1% 3-(trimethylsilyl)-propionic-2,2,3,3-d4-acid, Na ϩ salt (internal standard) were added to a total volume of 450 l. A 1 H NMR spectrum of this starting material was recorded. Then 50 l of the enzyme solution (2.2 mg of the enzyme preparation in 150 l of D 2 O) was added, and continuous NMR recording was started. The final concentrations in the NMR sample (500 l) were 18.0 mg/ml mannuronan, 1.32 mg/ml AlgE7 (as total protein), and 8.15 mM Ca 2ϩ .
Spectra were recorded on a Bruker DPX 400 spectrometer. To monitor the progress of the reaction, a series of 270 1 H NMR spectra was recorded successively (T ϭ 49°C). Spectra were obtained using a spectral width of 3612 Hz, a 32,000 data-block size, and 64 scans were accumulated. The resulting time interval between two successive spectra was 8.5 min. No presaturation or other technique to reduce the HDO signal was used to avoid distortion of neighboring signals. The peaks were assigned according to Grasdalen et al. (12,13) and Heyraud et al. (52).
Due to different reaction conditions, the degradation time for mannuronan found in the above-described NMR experiment cannot be directly compared with the incubation time needed for complete degradation given in the substrate specificity curve (see Fig. 4A). Additionally, the time courses of the epimerase and lyase activities were analyzed by stopping the reaction at different time intervals before completion (terminated by freezing, no dialysis), whereupon the reaction products were analyzed in ordinary NMR experiments (results not shown). The NMR data acquired from integration of these spectra were consistent with the corresponding results shown in Fig. 6B.
Interpretation of 1 H NMR Spectra-The molar fraction of the monomers G (F G ), M (F M ), the diads (F GG , F MM , F GM , F MG ), and the Gcentered triads (F GGG , F MGM , F GGM , F MGG ) were calculated as described by Grasdalen et al. (12,13). The average length of the G-blocks was Alginate lyases degrade alginates at specific sequences and produce unsaturated 4-deoxy-L-erythro-hex-4-enepyranosyluronate moieties (⌬) by a ␤-elimination reaction. The characteristic resonance signals from the reducing ends (G red and M red ) and from the unsaturated nonreducing ends, ⌬, may then be identified and their fractions calculated from the NMR spectrum of the degraded alginate sample (52). Moreover, since the latter resonance shifts depend on the nearest neighbor moiety, both ⌬M and ⌬G can be identified. Thus, when G2MM and/or G2GM bonds are cleaved, the result will be G red signals from the reducing end and ⌬M signals from the nonreducing end. If instead M2MG and/or M2GG bonds are attacked, the corresponding resonance signals will be M red and ⌬G. From the NMR spectrum, only the ␤-anomeric reducing end signals can be integrated due to overlap of the ␣-signals with the unsaturated nonreducing ⌬-1-G end signals. To calculate the total molar fraction of G-and M-moieties at the reducing ends, the intensities of the ␣-signals are found by the ratios of the anomeric protons M ␣ /M ␤ Х 2.2 and G ␣ /G ␤ Х 0.2 (52). The average degree of polymerization, DP n , was estimated from DP n ϭ [ , where I denotes the integrated intensities of the indicated signals in the spectra.

Purification and Biochemical Characterization of AlgE7-
The plasmid pBG27 (29) was used for expression of AlgE7 in E. coli SURE, and the enzyme was purified by ion-exchange chromatography in a one-step protocol, which was found to be sufficient for further biochemical characterization (Fig. 2). The band representing AlgE7 in the SDS-polyacrylamide electrophoresis gel corresponded to a higher molecular mass than expected based on the deduced amino acid sequence of the enzyme (105 kDa compared to 90.4 kDa, respectively). Such aberrant migration rates, possibly A-module-associated (30,53), have been observed also for other AlgE proteins (45,47,54).
To study the relation between the epimerase and lyase activities of AlgE7, the response of the lyase to the parameters, known to be important for epimerization, was analyzed. The experiments showed that Ca 2ϩ is an absolute requirement for the lyase activity (Fig. 3A). Maximal activity was reached at a concentration of 2.5 mM, and it remained constant up to the highest tested concentration, 18 mM. Similar to epimerization by AlgE1 (47), AlgE2 (55), and AlgE4 (54), only Sr 2ϩ could substitute for Ca 2ϩ , although the efficiency of the reaction was reduced by ϳ80% (Fig. 3B). At a suboptimal calcium concen-tration (0.5 mM), Mn 2ϩ was inhibitory, whereas Zn 2ϩ eliminated all activity. Mg 2ϩ on the other hand had a weak stimulatory effect. The detrimental effect of Zn 2ϩ was also observed for AlgE1 (47), AlgE4 (54), and for the periplasmic epimerase AlgG (28). In conclusion, the responses of the AlgE7 lyase to divalent cations are very similar to that of the epimerases, strengthening the hypothesis that the same catalytic site is involved in both enzymatic activities.
To ensure optimal lyase activity in further experiments we also determined the pH and ionic strength optima. The pH optimum was found to be between 6.9 and 7.3, with about 85 and 74% activity at pH 6.5 and pH 7.9, respectively (results not shown). The lyase activity of AlgE7 was inhibited by high NaCl concentrations, as also observed earlier for AlgE epimerases (47,54,55). At 300 mM only 11% of the activity was retained (Fig. 3C). The epimerase activity of AlgE7 was not followed in these experiments, since the isotope assay does not distinguish between epimerization and lyase activity (see "Experimental Procedures"). However, the conditions found to be optimal for the lyase are very close to those previously shown by NMR spectroscopy to support good epimerization activity of AlgE7 (29).
The Substrate Dependence of the AlgE7 Lyase Activity-The AlgE epimerases are known to display different activities on different types of alginates (45,47,56), for example by being sensitive to acetylation (57). Furthermore, since the epimerization reaction appears to act mostly or exclusively in the M to G direction (54), it follows that G-block alginates are very poor substrates for the epimerases. If the epimerase and lyase activities originate from the same site in AlgE7, one might therefore expect its lyase activity to act poorly on highly acetylated alginates and substrates with continuous stretches of G-moieties. To analyze this we used eight structurally different alginate substrates ( Table I). The experiments showed that AlgE7 displayed lyase activity on all tested alginates, except on the G-block alginate from L. hyperborea (Fig. 4A). Deacetylated M-rich alginates (mannuronan and alginate from P. aeruginosa 8830) and the M. pyrifera alginate were found to be the best substrates (equal maximum lyase activity), and the overall kinetics on the two M-rich alginates appeared nearly identical. Noteworthy, however, was that complete degradation of the M. pyrifera alginate required a longer incubation time with the enzyme than the M-rich substrates. One possible explanation for this may be the existence of more unfavorable G-block sequences in this substrate.
The native highly acetylated (d.a. ϳ 0.7) M-rich alginate from P. aeruginosa was a very poor substrate for the lyase reaction. The maximum activity on the acetylated alginate was only 15% that of the activity on the corresponding deactylated substrate, and after about 7 h the lyase reaction almost ceased. At this point more than 90% of the glycosidic linkages remained intact. Correspondingly, the maximum lyase activity on the less acetylated (d.a. ϭ 0.06) and deacetylated alginates from A. vinelandii was much more similar, although the final ϳ50% of the attacked bonds were cleaved faster in the deacetylated alginate. Note also that the rate of degradation of the deacetylated A. vinelandii alginate was significantly slower than that of the three most efficiently degraded alginates. These experiments therefore highlight the negative effects of acetylation and continuous stretches of G-moieties for the activity of the AlgE7 lyase, similar to what one would expect also for the AlgE7 epimerase activity (see above paragraph).
The MG-alginate was degraded much less efficiently than the M-rich alginates, and the maximum lyase activity was only 17% compared with the activity on mannuronan. Furthermore, AlgE7 degraded the amount of mannuronan used within 1.8 h, whereas the same quantity of MG-alginate was not completely degraded after 52 h.
Mannuronan and the M. pyrifera alginate were degraded with nearly similar efficiency (Fig. 4A), but a closer analysis of the kinetics revealed that the initial degradation profiles differed (Fig. 4B). Pure mannuronan was degraded significantly slower initially and was found to behave similarly to the deacetylated alginate from A. vinelandii. This suggested to us that G-moieties in M-rich areas somehow stimulate the lyase reaction, but that this effect requires a certain sequential distribution of the G-moieties (Table I). The epimerase activity of AlgE7 might introduce such favorable sequences in the alginates, explaining the overall efficient degradation of M-rich substrates.
1 H NMR Analyses of Alginates Incubated with AlgE7-To obtain more detailed information on the substrate specificity of the AlgE7 lyase and the interplay between the lyase and epimerase activities, the reaction products from the different substrates were analyzed by NMR spectroscopy. The spectrum of completely degraded mannuronan showed, as expected, that AlgE7 acted as a combined epimerase and lyase, and the G content in the degraded alginate after epimerization was 17% ( Fig. 5A and Table II). In the spectrum the signals from G red dominate over those from M red , whereas the unsaturated nonreducing end signals exclusively are ⌬M. This is in good agreement with previously reported results using an M-rich (F G ϳ 0.05) alginate from P. aeruginosa as substrate (29). Hence, potential cleavage sites for the AlgE7 lyase with mannuronan as substrate are G2MM and/or G2GM, and M2MM and/or M2GM sequences, with a clear preference for G at the reducing end of the cleavage site (i.e. the two first mentioned bonds). From the signal intensities of the end groups, the average degree of polymerization (DP n ) of the oligomers was calculated to represent a tetramer, of which 35% of the G-moieties were internal (Table II). The kinetics of the AlgE7 activities could also be followed by carrying out the reaction inside an NMR tube, allowing spectra to be obtained continuously as the reaction progressed (Fig.  6A). Visual inspection indicated that the internal G content (G-1 peak) reached its maximum relatively early and then slowly decreased, whereas the lyase activity, on the other hand, continued to degrade the substrate, as demonstrated by the increase in the end signals. By integration of the 17 spectra in Fig. 6A, a more quantitative relationship between moieties epimerized, bonds cleaved, and DP n could be obtained (Fig. 6B). The plot confirmed that the epimerase activity dominated initially and that the AlgE7 epimerase increased the G content to 26% (the fifth spectrum in Fig. 6A and Table II). In the same period of time the lyase activity degraded the substrate (DP n ϭ 100) to oligomers with an average size of 11 units (3-4 chain breaks per 26 units epimerized). From this point in the reaction more G-moieties were consumed in cleavage than were formed by epimerization, resulting in a slow decrease in F G (Fig. 6B). These data therefore indicate that the oligomer size of the alginate substrates also influences the balance between epimerization and degradation by AlgE7. In the final spectrum of the series, the G content had decreased from 26 to 18%, and on average every fourth bond in the alginate was cleaved by the lyase (DP n ϳ 4). This is in very good agreement with the NMR data obtained from completely degraded mannuronan (Fig. 5A and Table II).
The type and intensity of the end signals found in degraded M. pyrifera alginate were similar to those in the mannuronan spectra (the M. pyrifera spectrum is not shown). Hence, the degradation of this substrate involved the same glycosidic links as those attacked during the reaction on mannuronan, with the same apparent preference toward the G2MM and/or G2GM bonds. The M. pyrifera alginate was degraded to a DP n of about  (Table II).  (Table II), and most of the guluronic acids were found as internal G-moieties (83%). Taking into account the sequential distribution of G-moieties in this substrate, the end products of the lyase activity may be considered as a mixture of fairly long G-blocks and smaller oligosaccharides with a predominance of M-moieties. This is consistent with the finding that G-blocks are not cleaved by the lyase (Fig. 4A). Even though some newly epimerized moieties are likely to be converted to ⌬ units in the lyase reaction (with a concomitant decrease in total F G ), the NMR data indicate that the lyase activity dominated on this particular alginate (Table II). This observation is interesting in that it clearly shows that the composition of the substrate can determine whether the enzyme acts as an epimerase or a lyase. For a possible explanation of this, see "Discussion." The influence of acetylation on the lyase and epimerase activities of AlgE7 was also studied by NMR spectroscopy using the P. aeruginosa M-rich alginate (ϳ5% G) as substrate ( Fig.  5B and Table II). The analysis confirmed the inhibitory effect of the acetyl groups on the lyase activity (DP n ϭ 43), and the spectra in addition showed also that the epimerase activity was strongly inhibited by these groups (F G ϭ 0.11). In contrast, a deacetylated version of this alginate was readily degraded to a DP n of about 7, and the G content in the degraded alginate had increased to 30%. The chemical shift values and intensities of the lyase signals from the deacetylated M-rich alginate were similar to those obtained with mannuronan as substrate. The G content and especially the amount of internal G-moieties in the degraded alginates were nevertheless significantly higher with the M-rich alginate. Moreover, the degraded M-rich alginate contained more than twice as many GG sequences as depolymerized mannuronan. One possible explanation is to assume that the epimerase has a preference for introduction of new Gmoieties next to a pre-existing one in the M-rich alginate, leading to the formation of short G-blocks. Since mannuronan does not contain G-moieties initially, the formation of G-blocks may be much less efficient on this substrate. This hypothesis is interesting in view of the previously suggested processive or preferred attack modes of action of the AlgE4 epimerase (54).
AlgE7 Is Capable of Epimerizing Polyalternating Alginates-A strictly polyalternating alginate has only two possible cleavage sites for lyases, G2MG and M2GM, and according to results described above, only the latter of these bonds represents a potential site for the AlgE7 lyase. Furthermore, it was not expected that the AlgE7 epimerase could introduce FIG. 5. 1 H NMR spectra of alginates before and after incubation with AlgE7, AlgE7-E1, and AlgE5. A, mannuronan (AlgE7 and AlgE7-E1); B, acetylated and deacetylated M-rich alginates from P. aeruginosa (AlgE7 only); C, MG-alginate (AlgE7, AlgE7-E1, and AlgE5). In the different spectra G, M, G red , M red , and ⌬ denote internal G-and M-moieties, reducing end G-and M-moieties, and 4-deoxy-Lerythro-hex-4-enepyranosyluronate moieties, respectively. The numbers refer to the position of the proton in the pyranosyl ring that causes the signal, and the nonunderlined M and G refer to the neighboring moieties (see also "Experimental Procedures"). For all substrates except the MG-alginate, the lyase reactions were allowed to proceed to completion (no further increase in A 230 ). Since the specific activity of AlgE7 and AlgE7-E1 is very similar (Fig. 2), equimolar amounts of enzymes were used in reaction with the MG-alginate. The NMR data of the integrations of the peaks are summarized in Table II. AlgE7 (the seventeenth spectrum, Fig. 6A) AlgE7 (Fig. 5A) (Table I), signals for the reducing ends may be neglected (i.e. internal F G equals total F G ).
b No resonance signals caused by the H-5 proton could be recorded (see Fig. 6).
c NMR reactions were run to completion. d 24-h incubation. e The MG-alginate was degraded to a DP n ϭ 30 -40 before incubation with AlgE5. more favorable G2GM cleavage sites for the lyase, since it was previously reported that single M-moieties in a GMG sequence were inaccessible to epimerization (45,47,58). Thus, one would expect the lyase activity on this substrate to generate M-moieties at the reducing ends and ⌬M-moieties at the nonreducing ends and that no further epimerization of the substrate would take place. Surprisingly, however, the spectrum from the MGalginate contained both ⌬M and ⌬G signals in equal amounts (measured as intensities), whereas the reducing end signals were exclusively G red (Fig. 5C and Table II). Furthermore, the total G content increased from 46 to 62%, demonstrating that AlgE7 had epimerized M-moieties situated in a GMG sequence. Thus, cleavable AlgE7 bonds in the MG-alginate after epimerization were G2MG and G2GM, where the G at the reducing end in the latter was introduced by the AlgE7 epimerase. In addition, the absence of M red signals illustrated that M2GM sequences are inaccessible for AlgE7 on the polyalternating alginate. This leads to the conclusions that polyalternating alginates can be epimerized by AlgE7, that the new alginate structure thereby generated becomes a substrate for the lyase activity, and that ⌬G signals are generated at the nonreducing end. The latter was not observed with the other tested substrates. The results therefore suggest that there exists a very complex relation between substrate composition and the possible lyase cleavage sites.
Since the epimerization of the polyalternating alginate was unexpected, we also tested two other epimerases, AlgE5 and AlgE6, for this property. The results showed that the G content increased from 46 to 58 and 54%, respectively (Fig. 5C, Table   II; results only shown for AlgE5). This ability is therefore not unique to AlgE7.
The Properties of an Enzyme Hybrid (AlgE7-E1) and a Point Mutant Protein (AlgE7-D152G) Strongly Indicate That the Same Site Is Responsible for Both the Epimerase and the Lyase Activities-We have recently shown that the A-modules are sufficient for epimerization (30), and based on the hypothesis that the epimerase and lyase activities originate from the same catalytic site, we compared the amino acid sequences of the AlgE7 A-module to those of AlgE1-6 (24,29). The analysis showed that at least two distinct short motifs in the N-terminal part are particularly characteristic for AlgE7 (29). This might indicate that the N-terminal part of the A-module contains moieties that are needed to display lyase activity, and to test this hypothesis we substituted the 5Ј 798 base pairs in algE1-1 (47) with the corresponding sequence from algE7, generating algE7-E1, and vice versa (algE1-E7). The resulting two plasmids, pBG70 and pBG71, could then be used for expression of AlgE7-E1 and AlgE1-E7 in E. coli SURE. The crude extracts were analyzed with respect to epimerase and lyase activities (radioisotope assay), and the results showed that AlgE7-E1 displayed strong epimerase and/or lyase activity, whereas the activity of AlgE1-E7 was 1-2% that of AlgE7-E1. AlgE1-E7 was not studied further as it was difficult to obtain NMR spectra showing the activity. To elucidate if the active hybrid enzyme displayed both epimerase and lyase activity, AlgE7-E1 was purified (ion exchange chromatography) and incubated with different alginates before NMR spectroscopy.
Incubation with mannuronan followed by NMR spectroscopy FIG. 6. Epimerization and degradation of mannuronan by AlgE7. A, 1 H NMR spectroscopy demonstrating the interplay between epimerization and degradation of mannuronan (17 selected spectra) (see Fig. 5 for abbreviations). The AlgE7 reaction was allowed to proceed in an NMR tube inside the spectrometer. Since the enzymatic reaction took place in D 2 0, no resonance signals from H-5 could be recorded. The direction of the arrow indicates increasing reaction time (17 spectra); B, units epimerized (represented by the molar fraction of G moieties, total F G ), bonds cleaved (represented by the molar fraction of ⌬-1-M, F ⌬-1-M ), and DP n , plotted as a function of incubation time with the enzyme. The three parameters analyzed were calculated by integration of the corresponding spectra in A. To retain the clarity of the plot, the first four spectra in A were stacked wider than indicated by the time scale (compare with B).
of the reaction products showed that AlgE7-E1, similarly to AlgE7, displayed both epimerase and lyase activity (Fig. 5A). It could therefore be concluded that the N-terminal 266 amino acids of AlgE7 are sufficient to convert an epimerase to a combined lyase and epimerase. The type of the end signals were also identical to those generated by AlgE7 acting on the same substrate. However, the relative intensities of the reducing end signals were different, as AlgE7-E1 seemed to have a relatively higher preference for M-moieties at the reducing end (Fig. 5A). If one assumes that the epimerase activities of AlgE7 and AlgE7-E1 are important for producing good substrates for the lyase reaction, it was not unexpected that the total G content in the AlgE7-E1-degraded mannuronan was lower than for AlgE7 and that the internal F G after complete degradation was much smaller than the total F G for both enzymes (particularly significant for AlgE7-E1, Table II).
The difference in substrate specificity compared with AlgE7 was also reflected by the appearance of weak but clearly significant ⌬G signals from reaction products of the M. pyrifera alginate and the G-block alginate (spectra not shown). Since AlgE7-E1 appears to cleave the alginates more effectively than AlgE7, we expected that the MG-alginate might also serve as a better substrate for the hybrid enzyme. This assumption was confirmed by the observed strong intensity of the ⌬G signal relative to ⌬M in the corresponding NMR spectrum (Fig. 5C) and by the lower DP n value generated by the hybrid enzyme (Table II). Similar to AlgE7, the hybrid enzyme was also capable of epimerizing M-moieties between two G-moieties.
The experiments described above indicated that at least some of the N-terminal 266 amino acids in the AlgE7 A-module are of key importance for the lyase activity but also that the substrate specificity of the lyase is affected by amino acids located in the C-terminal part. Even though these experiments were consistent with the hypothesis that the two enzyme activities originate from the same site in the A-module, possibly in the N-terminal part, it was not obvious what specific amino acids would be essential for both activities. No inactive epimerase point mutants have so far been reported, and we therefore carried out a computer analysis to try to identify candidate amino acids that might affect one or both activities. A total of nine A-modules have been sequenced from the AlgE-type family, but all these sequences share too much similarity to point to a particularly critical amino acid. This problem can be approached in different ways, but here we decided to first focus on a comparison of the AlgG and AlgE epimerases (Fig. 7), which share a very low degree of overall sequence similarity. However, short stretches of similarity could be detected, and we focused on a region in which there is also some similarity to a pectate lyase (PelL) and an apparently unrelated virus protein (GP1). Based on this comparison, only one amino acid (Asp-152 in AlgE7) was found to be universally present (Fig. 7). Sitespecific mutagenesis was therefore used to mutate algE7 such that the aspartic acid at position 152 was changed to glycine. The mutant enzyme was expressed in the same vector system as AlgE7, and Western blot analyses showed that it was expressed equally well as the wild type enzyme (not shown). The crude extract was then subjected to activity analyses by the radioisotope assay (M-rich alginate, ϳ5% G), and interestingly, only a very weak activity could be detected (about 0.2% of the wild type enzyme). Since both the lyase and epimerase activities are measured by this assay, it seemed as both activities had been virtually eliminated by this single amino acid substitution. The result was also confirmed by NMR analyses (mannuronan substrate), as neither signals from epimerization nor lyase activity could be detected (NMR spectrum not shown).
Since there is a complex interplay between the epimerase and lyase activities, it was formally possible that the epimerase activity alone had been inactivated by the mutation and that lack of this activity resulted in an unfavorable substrate (mannuronan) for an intact lyase activity. To investigate this, we also repeated the NMR experiment by using the G-containing M. pyrifera alginate as substrate (NMR spectrum is not shown). No activity was detected, and it is therefore clear that both activities became inactivated by the point mutation. Based on this, many of the other experiments reported here and on previously reported data on the properties of the AlgE epimerases, we feel that it is very probable that the epimerase and lyase activities both originate from the same site in the A-module of AlgE7. DISCUSSION The results reported here strongly suggest that not only are lyase and epimerase reactions mechanistically related but that they can both be catalyzed by the same active site in an enzyme. However, despite the very large number of alginate lyases studied (59), only AlgE7 has been reported to display this capacity at comparable rates. This obviously raises the question as to what conditions have to be met for AlgE7 to act as an epimerase and/or a lyase. The activity patterns observed when different alginate substrates were used showed that acetylation and continuous stretches of G-moieties are inhibitory for both reactions ( Fig. 4A and Table II), and a simple explanation for this could be that the common alginate binding site is incompatible with such a substrate composition. It is, however, more difficult to envision what determines whether the enzyme acts as a lyase or an epimerase on substrates that can be converted by one or both activities (i.e. mannuronan and the M. pyrifera alginate). When using pure mannuronan as substrate, the reaction rate is much faster for the epimerase up to a point where the substrate has become highly degraded, where after the lyase activity strongly dominates on the small depolymerized oligomers (DP Ͻ 11) (Fig. 6). This indicates that small oligomers are readily bound by the enzyme but that for the epimerization pathway to be favored, longer alginates are needed. This assumption is in good agreement with a recent study that showed that epimerization by AlgE4 and AlgE2 cannot take place on mannuronan substrates smaller than 8 and 6 moieties, respectively, 2 whereas AlgE7 is clearly able to degrade to smaller fragments. The same way of thinking may also explain the strong dominance of the lyase activity on the M. pyrifera alginate. AlgE7 can obviously bind to this polymer, 2 7. Sequence alignment of protein segments found by BLAST software within the mannuronan C-5-epimerases and other proteins. The aspartic acid (D) residues conserved in all sequences are in bold and underlined. AlgE1-6 Av, mannuronan C-5epimerases AlgE1 (Q44494), AlgE2 (Q44495), AlgE3 (Q44496), AlgE4 (Q44493), AlgE5 (Q44492), and AlgE6 (Q9ZFH0) from A. vinelandii; AlgE7 Av, mannuronan C-5-epimerase AlgE7 (Q9ZFG9) from A. vinelandii; AlgY Av, A-module containing protein AlgY (Q9ZFG8) from A. vinelandii; AlgG Av. mannuronan C-5-epimerase AlgG (P70805) from A. vinelandii; AlgG Pa, mannuronan C-5-epimerase AlgG (Q51371) from P. aeruginosa; GP1 Esv, GP1 protein from Ectocarpus silliculosus virus (Q90190); PelL Ec, pectate lyase from Erwinia chrysanthemi (Q47473), which belongs to the family 9 of pectate lyases. a , an identical sequence is present at positions 989 -1002 and 987-1000 in AlgE1 Av and AlgE3 Av proteins, respectively. b , the PelL Ec protein displayed a very low BLAST score. but it may be that epimerization is somehow more sterically hindered by the nature of the substrate. These differences in substrate selectivity might be explained by assuming different stereochemical requirements for binding and positioning of the alginate substrate in the lyase compared with the epimerase reaction (Fig. 1).
Based on results from NMR spectroscopy, potential cleavage sites for the AlgE7 lyase were G2MM and/or G2GM, whereas M2MM and/or M2GM sequences are less favorable. G2MG sequences were also found to be cleavable (MG-alginates) but with strongly reduced efficiency. The fact that G-moieties dominated at the reducing end even when mannuronan was used as substrate highlights the complexity of the interplay between the lyase and epimerase activities. It probably means that the initially higher epimerase activity results in the formation of a new substrate that is more favorable for the lyase than the substrate provided from the start. Similarly, G2MG sequences appear to be poorly cleavable, but the activity on substrates containing many such triads (like the MG-alginate) is selfstimulated by introducing cleavable G2GM-sequences through the fill-in capacity of the epimerase activity.
Even if pure mannuronan and the deacetylated M-rich alginate (ϳ5% G) are very similar substrates, the data obtained from complete degradation differed in that the total G-content (and F GG ) and especially the amount of internal G-moieties was significantly higher in the M-rich alginate (Table II). As a consequence of this, the average size of the final oligomers was also higher. This may indicate that the AlgE7 epimerase preferentially attacks M-moieties next to a G in the alginate chains, as this mode of action (preferred attack) probably will result in formation of short G-blocks, which in turn inhibit the AlgE7 lyase from cleavage.
Although there are obviously certain amino acid residues that are essential for both cleavage and epimerization to take place (like Asp-152), the experiments reported here indicate that the specificity of these reactions appears to be affected by many other residues in the protein. This was clearly illustrated by the reaction products of the hybrid enzyme AlgE7-E1, which more frequently contained M-moieties at the reducing ends compared with the products of the native AlgE7 reaction. The hybrid enzyme also generated stronger ⌬G signals relative to those of ⌬M, when the MG-alginate was used as substrate (Fig.  5C), and the final DP n was lower (Table II). For epimerization, the previously reported variability in the epimerization patterns among the AlgE family enzymes support a similar conclusion (45,56), and here we also noted that the M. pyrifera substrate strongly favors the lyase reaction, whereas the same substrate could easily be epimerized from its original 41% G to 66% by AlgE6 (NMR data not shown). These observations all suggest that the specificity of both the lyase and epimerase activities are the result of presumably slight differences in the folding structures of the enzymes, and to understand the nature of these effects, it seems obvious that a three-dimensional structure of at least one of the A-modules would be very important.
Another interesting question not addressed here is what biological significance the dual properties of AlgE7 might have. It seems likely that this is somehow related to cyst formation, as it is known that such differentiated cells contain different types of alginates (15), and it is also known that the expression of the AlgE enzymes are affected by the conditions that lead to induction of cyst formation (53).