Epimerization - American Chemical Society

Biomacromolecules 2002, 3, 1343-1352

1343

A High Field NMR Study of the Products Ensuing from Konjak Glucomannan C(6)-Oxidation followed by Enzymatic C(5)-Epimerization† Vittorio Crescenzi,‡ Gudmund Skjåk-Bræk,§ Mariella Dentini,*,‡ Giancarlo Masci,‡ Maria Scala Bernalda,‡ Daniela Risica,‡ Donatella Capitani,| Luisa Mannina,|,⊥ and Anna Laura Segre| Department of Chemistry, University of Rome “La Sapienza”, Rome, Italy, Department of Biotechnology, Norwegian University of Science and Technology, Trondheim, Norway, Institute of Chemical Metodologies, CNR, Research Area of Rome, Rome, Italy, and S.T.A.T. Department, University of Molise, Environmental Science Faculty, 86170 Isernia, Italy Received July 12, 2002; Revised Manuscript Received September 13, 2002

Konjak glucomannan (KGM) is a water-soluble linear copolymer of (1f4) linked β-D-mannopyranosyl and β-D-glucopyranosyl units. It has been selectively C6-oxidized by a 2,2,6,6-tetramethylpiperidin-1-oxy mediated reaction to obtain the corresponding uronan. Oxidized KGM has been treated with three different C-5 epimerases, AlgE4, AlgE6, and AlgE1, to obtain uronans with a various content of R-L-gulopyranuronate residues, namely, KGME4, KGME6, and KGME1. By use of 1D selective and 2D NMR techniques, a full assignment of the high field (600 MHz) NMR spectra of the purified native KGM and of the oxidized and epimerized derivatives has been obtained. Since in the anomeric region of the 1H NMR spectrum of native KGM, diads sensitivity is present, the glucose-glucose, glucose-mannose, mannose-mannose, and mannose-glucose distribution has been obtained. In the 13C spectrum of oxidized KGM, due to the presence of triad sensitivity on the C-4 resonance of glucuronic and mannuronic units, a better sequential investigation has been possible. As a result the average length of mannuronic blocks, NM is obtained. When AlgE4, AlgE6, and AlgE1 enzymes are used for the epimerization of oxidized KGM, the reaction products differ significantly both in the proportion and in the distribution of the mannuronic and guluronic residues. In epimerized KGM derivatives, a careful deconvolution of 1H spectra allows the measurement of the degree of epimerization. In the case of KGME1 and KGME6, the average blocks length, NG, of the guluronic blocks introduced in the polysaccharidic chain with the epimerization has also been calculated. Due to the shortness of mannuronic blocks in the oxidized KGM before the epimerization, NG in the epimerized compounds is also very low. Introduction Glucomannans are mannose-rich hemicellulose compounds found in softwoods and in roots, tubers, bulbs, and seeds of some plants, particularly in orchids.1 Konjak Glucomannan (KGM) is a water-soluble, nonionic polysaccharide found in tubers of Araceae. It is a linear copolymer of (1f4) linked β-D-mannopyranosyl (m) and β-D-glucopyranosyl units (g), with an m/g ratio equal to ≈1.6. Depending on its origin, native konjak contains a small percentage, 2-6%, of O-acetyl groups on C-6, which control the water solubility.2 A minimal amount of short-chain branches on the C-3 position of mannose has been previously reported.3 Tubers containing this polysaccharide have been cultivated in the Far East for many centuries and they are used in * Corresponding author: telephone, +390649913633; fax, +39064457112; e-mail, [email protected]. † Dedicated to the late Professor Salvatore Castellano. ‡ University of Rome “La Sapienza”. § Norwegian University of Science and Technology. | Institute of Chemical Metodologies, CNR. ⊥ University of Molise.

traditional Japanese cooking to make noodles and gels which are stable in boiling water. They have been recently introduced into the United State and Europe as food additives because when dissolved in water they form highly viscous solutions and have potential use as thickener for gravies, soups, and sauces. More recently the increased use of KGM as a food additive has led to increased research activity.4 Solution properties and gelling abilities of glucomannans strictly depend on their structural features; therefore it is worth obtaining detailed understanding of the structure of glucomannans in terms of distribution of m and g residues and a good quantitative analysis of the acetylated units. Fully deacetylated β-1f4-linked glucomannan is insoluble in water,2 while native KGM is rather soluble. Besides, KGM is able to give stable gels in water in the presence of mild alkali through the formation of a network structure supported by hydrogen bonding. High pH is necessary to cleave acetyl groups thus allowing the formation of junction zones between unsubstituted stretches of KGM backbones. Junction zones are essentially formed by acetate-free portions long enough to give energetically favorable chains pairing. The crystal

10.1021/bm025613d CCC: $22.00 © 2002 American Chemical Society Published on Web 10/17/2002

1344

Biomacromolecules, Vol. 3, No. 6, 2002

structure can be used as a model for junction zones in gels. In the solid state KGM packs isomorphously in the mannan II structure, showing that the replacement of mannose by glucose does not significantly affect the typical molecular association adopted by hydrated mannans.5 Recently we have been working with glucomannans and other polysaccharides containing β-1f4-linked mannose residues in order to prepare new charged derivatives through a selective C-6 oxidation mediated by the stable nitroxide free-radical TEMPO (2,2,6,6-tetramethylpiperidin-1-oxy).6,7 These β-1f4-linked polysaccharides containing mannuronic acid residues were then treated with different mannuronan C-5 epimerase in order to produce R-1f4-linked L-guluronic acid residues.6 These new polyuronans, which we can call “pseudo-alginates”, might be interesting since they may be cheaper then natural alginates and still have similar good gelling properties suitable for applications in the biomedical field as high added-value materials. By cloning and expressing the genome of the alginate-producing bacterium Azotobacter Vinelandii in Escherichia coli,8-10 seven different epimerizing enzymes (named AlgE1-AlgE7) were produced. These enzymes give products which differ significantly in the proportion and distribution of mannuronic and guluronic acid residues. AlgE4 predominantly forms alginates with mannuronic acid (M)-guluronic acid (G) blocks, AlgE6 preferentially introduces stretches of G blocks into the polymer, and AlgE1 makes stretches of MG and G blocks.8,9 Besides, during the epimerization, different epimerases can have a different mode of action in creating different uronic acid patterns. The mechanistic properties can be significantly influenced by the presence of nonalginate residues (glucuronic acid) and by side chain substituents (acetyl groups) when the enzymes are used on our artificial uronans prepared starting from glucomannans. Finally, it is worth noticing that a good knowledge of the KGM structure, in particular the distribution of mannose and glucose residues, can be very useful in understanding how mannuronan C-5 epimerases actually work on the oxidized substrates. In this work we report on the 1D and 2D NMR study of konjak glucomannan and its selectively oxidized and epimerized derivatives.Three different C-5-epimerases, AlgE1, AlgE4, and AlgE6, have been used to epimerize β-Dmannopyranuronate residues to R-L-gulopyranuronate residues. The degree of epimerization and the distribution of M and G residues will be calculated and related to the KGM sructure and to the different epimerises used. Experimental Section Nutricol konjak was supplied by FMC Europe NV and was further purified by following the procedures given in the literature.11 A 0.5% (w/w) solution of KGM was prepared in water and left for 48 h at 25 °C under stirring. The solution was centrifuged at 5000 rpm for 1 h to remove the insoluble material. The supernatant liquid was filtered on 3-0.45 µm pore size filters. It was then poured two times into an equal volume of methanol to precipitate KGM.

Crescenzi et al.

The precipitate was then dispersed in water, extensively dialyzed, and freeze-dried. This yielded a fluffy white material, approximately 70% of the original powder. Konjak Oxidation.7 The polysaccharide (1 g) was suspended in 800 mL of distilled water, and TEMPO (Sigma) (0.06 g) and NaBr (0.3 g) were added. The resulting solution was cooled overnight at 4 °C. Then, cold sodium hypochlorite, previously brought to pH 9, was added. Without further cooling, the reaction was monitored while 0.5 M NaOH was added to maintain a pH of 9-9.5. After 1-1.5 h, when the pH was decreasing slowly, EtOH was added to the reaction mixture to consume NaOCl still present, then NaBH4 (0.1 g) was added. The reaction was left overnight at room temperature and stopped by adding 4 M HCl, bringing the pH to ∼5. The resulting polysaccharide derivative was precipitated by addition of MeOH, solubilized in water and dialyzed. Yields of oxidized polysaccharide were about 8090%. The degree of C-6 oxidation of the samples determined by means of potentiometric titration of the acid form and NMR methods (see Results and Discussion) was higher than 90%. Konjak Epimerization. Three oxidized konjak samples (polymer concentration Cp ) 0.1% w/v) were epimerized using three different C-5 mannuronan epimerases: AlgE4 (Cp ) 0.005% w/v), AlgE6 (Cp ) 0.01% w/v), and AlgE1 (Cp ) 0.004%), to obtain the epimerized derivatives KGME4, KGME6, and KGME1, respectively. All the samples were incubated in 50 mM TRIS/HCl buffer pH 6.9, plus 50 mM NaClO4 and 2 mM CaCl2, at 40 °C under magnetic stirring. The reaction was stopped by chelation of Ca2+ ions (addition of 13 mM EDTA). The pH was adjusted to 7.0, and the samples were shaken gently for about 12 h. The protein was removed by cooling and filtration of the solution on 0.2 µm pore size filters, which are selective for the proteins but not for the konjak polymer chains. Samples were extensively dialyzed against distilled water and recovered by freeze-drying. NMR Experiments. Samples (≈2 mg) were dissolved both in D2O (700 µL) and in a phosphate buffered (pD ) 7) D2O solution, 0.01 M NaCl (700 µL). Due to the strong dependence of some proton resonances from the acidity of the solution,12 the control of pD is necessary. Epimerized konjak samples were also dissolved in a phosphate/citrate buffered D2O solution at pD ) 4. 1 H and 13C NMR experiments have been performed on a Bruker Avance600 spectrometer operating at 600.13 and 150.92 MHz, respectively. Since at room temperature proton resonances appear broad and poorly resolved, all spectra were performed at 343 K. At this temperature all spectra appear sufficiently resolved to be assigned. In all 1H spectra a soft presaturation of HOD residual signal was performed.13 Chemical shifts of 1H and 13C spectra are reported in ppm with respect to 2,2-dimethyl-2-silapentane-5-sulfonate sodium salt (DSS) used as internal standard. The 13C{1H} NMR spectra of KGM and oxidized KGM in D2O were obtained using the following acquisition parameters: time domain (td) 32K, spectral width (sw) 255 ppm, digital resolution 1.17 Hz per point, π/2 pulse width 7.5 µs, recycle delay (rd) 2 s; proton decoupling has been

Products from Konjak Glucomannan

Biomacromolecules, Vol. 3, No. 6, 2002 1345

Figure 1. 600.13 MHz 1H NMR spectrum of konjak in D2O at 343 K. In the insert, the experimental (b) and the simulated (a) spectral regions of anomeric protons of mannose (m) and glucose (g) units are shown. Diad assignments m-1g, m-1m, g-1g, and g-1m are reported.

performed with a GARP sequence14a,g using a 90° pulse length of 86 µs. 1 H and 13C assignments were obtained using 1H-1H COSY,14b,c 1H-1H TOCSY,14d,e and HMQC14f,g experiments. All these 2D experiments have been acquired using a time domain of 1024 data points in the F2 dimension and 512 data points in the F1 dimension, and the recycle delay was 2 s; the TOCSY experiment has been acquired with a spinlock duration of 80 ms. The HMQC experiment has been acquired using a coupling constant 1JC-H ) 150 Hz. The number of scans has been optimized for obtaining a good signal/noise ratio. All 2D experiments have been processed in the phasesensitive mode (TPPI) with 512 × 512 data points. Due to the strong signal overlapping, the assignment of the 1H NMR spectra of KGM has been obtained by means of 1D selective TOCSY using a different spin-lock duration,15 spanning from 5 to 100 ms. The 90° selective pulse was obtained by means of a selective spin-echo using a DANTE pulse train both for 90° and for 180° pulses. Typical values for DANTE were 60 repetitions of 1 µs. The deconvolution of the 1H and 13C NMR spectra was performed using the SHAPE2000 (ver.2.1) software package.16 In our notation, m ) mannose unit, g ) glucose unit, M ) mannuronic unit, G ) guluronic unit, and Gl ) glucuronic unit. Results Konjak and Oxidized Konjak. The 1H NMR spectra of KGM and oxidized KGM have been run both in D2O and in a phosphate buffered solution (pD ) 7) at 343 K. The spectra in D2O appear slightly more resolved than the corresponding ones in buffered solution. Details of the assignment will be discussed only for the case of KGM in D2O. Konjak. The 1H spectrum of KGM in D2O is shown in Figure 1; in the insert, the anomeric region in an expanded

Figure 2. 1D selective TOCSY experiments of konjak in D2O obtained by selective irradiation of the anomeric proton g-1 of the glucose units. At very short spin-lock duration, τ ) 5 ms (A), only the signal due to g-1 is observed. With increase of the spin-lock duration, τ ) 23 ms (B) at 3.361 ppm, the resonance due to g-2 is observed. At τ ) 40 ms (C) also the resonance at 3.687 ppm due to g-3 is well observable. At τ ) 58 ms (D) and τ ) 70 ms (E) the resonances at 3.669 and 3.606 ppm due to g-4 and g-5, respectively, are observed. In F, the 1H spectrum of KGM is shown for comparison.

scale is also shown. The spectral assignment was obtained by means of 1H-1H COSY and 1H-1H TOCSY experiments; however, due to the strong overlapping in the 3.3-4.1 ppm spectral region, 1D selective TOCSY experiments have been also performed at different mixing times. In this way, the whole spin system is assigned;15,17 see Figure 2. The full assignment of 1H spectrum of KGM in D2O is given in Table 1. In the same table, the assignment of the 13 C spectrum of KGM as obtained by means of an HMQC experiment, Figure 3a, is also reported. Note that, due to the diads sensitivity of the C6 resonance, two cross-peaks of methylene carbons at 62.59 and 62.85 for the mannose units and two cross-peaks at 62.48 and 62.72 ppm for the glucose units are observed. The 13C spectrum of KGM appears rather broad and unresolved, without any improvement of the sequential resolution with respect to the 1H spectrum. The diad sensitivity and the chemical shift of the anomeric proton of a saccharidic unit depend on the chemical nature of its neighbor adjacent units.18 In the insert b of Figure 1, the 4.40-4.80 ppm region of the 1H spectrum of KGM is shown. Resonances of the

1346


Table 1. 1H and

13C

Crescenzi et al.

NMR Assignments of KGM in D2O and in Buffered Solution (pD ) 7) at T ) 343 Ka konjak in D2O

mannose unit 1 2 3 4 5 6

mm mg

konjak in buffered solution glucose unit

1H

13C

4.734 4.748 4.102

102.30 102.30 72.29

3.788 3.822 3.544b 3.604 3.757b 3.899 3.732 3.894

73.86 78.69 77.34 77.44 62.85 62.59

gm gg

mannose unit

1H

13C

4.506 4.521 3.361

104.60 104.60 75.25

3.687 3.669 3.606b 3.637 3.994b 3.823 3.979 3.823

76.33 80.81 77.01 77.11 62.72

mm mg

glucose unit

1H

13C

4.734 4.748 4.102

102.12 102.30 72.30

3.788 3.822 3.55

73.72 78.64 77.30

3.732 3.900

62.61

gm gg

1H

13C

4.506 4.521 3.361b 3.370 3.687 3.669 3.610

104.60 104.60 75.22

3.990 3.892

62.61

76.26 80.78 77.06

62.48

a Chemical shifts are reported in ppm with respect DSS used as an internal standard. 13C NMR assignment has been obtained by means of 2D HMQC experiments. b Diad sensitivity.

Table 2. Experimental and Random Distributions of Diads in KGM diads

experimetal distribution

m-1m m-1g g-1g g-1m

0.33 ( 0.01 0.27 ( 0.03 0.098 ( 0.002 0.30 ( 0.01

random distributiona RDmm RDmg RDgg RDmg

0.37 0.24 0.16 0.24

a The areas A and A of resonances m-1 and g-1, respectively, have m g been obtained by applying the deconvolution procedure. These values allow calculation of the random distribution values: RDmm ) (Am)(Am); RDmg ) (Am)(Ag); RDgg ) (Ag)(Ag); RDgm ) (Ag)(Am).

Figure 3. HMQC experiments performed on konjak (a) and oxidized konjak (b), in D2O at T ) 343 K. The assignment is reported in Tables 1 and 3.

anomeric proton m-1 of mannose units are at 4.734 and 4.748 ppm; resonances of the anomeric proton g-1 of glucose units appear as two partly overlapped doublets respectively centered at 4.506 and 4.521 ppm. The deconvolution of this spectral region allows the area of the resonances to be obtained; the obtained mannose/glucose ratio is 1.6 in agreement with chemical analysis data. Moreover, using the area obtained from the deconvolution, it is possible to arrive at the sequential distribution in term of diads, i.e., mannose linked to mannose (m-1m), mannose linked to glucose, (m1g), glucose linked to mannose (g-1m), and glucose linked to glucose (g-1g). The experimental diad frequency distribution, see Table 2, is slightly different from a random one, favoring alternating sequences (mg, gm) with respect to blocks (mm, gg). The 1H spectrum of KGM allows the degree of acetylation of the sample to be obtained by using the resonances in the 2.0-2.3 ppm spectral region due to acetyl groups; see Figure 1. The percentage of acetylation is about 6%. Oxidized Konjak. The 1H spectrum of oxidized KGM shows rather broad resonances and poor resolution, both in

D2O and in the buffered solution. On the contrary, the 13C spectrum appears extremely well resolved and with triad sequences well observable (see Figure 4). The 1H and 13C assignments of oxidized KGM in D2O and in buffered solution were obtained by means of COSY, TOCSY, and HMQC experiments and are reported in Table 3. The degree of oxidation of the C6 oxidized konjak can be evaluated by means of 1D and 2D experiments. In fact, the comparison between the HMQC maps obtained for KGM (Figure 3a) and for C6-oxidized KGM (Figure 3b) gives direct information on the degree of oxidation of the sample: the cross-peaks due to methylene carbons in the KGM sample, fully disappear in the HMQC map of oxidized KGM. The high degree of oxidation is also confirmed by integrating the 1H spectrum of the oxidized sample. In the nonoxidized sample, seven nonexchangeable protons belong to each monomeric unit; the nonanomeric protons all resonate in the 3.3-4.2 ppm range. With the area of the resonances due to the anomeric protons equal to 1, the integral of all signals resonating in the 3.3-4.2 ppm range is equal to 6. In a fully oxidized sample the integral of all nonanomeric protons must be equal to 4. In the case of the sample discussed here, the value of this integral is found to be 4.2. As a consequence, the percentage of oxidation is about 90%. Moreover, the 13C spectrum of oxidized konjak shows the carbonyl resonances at 176.59 and 176.33, respectively (see the insert “a” in Figure 4), whereas no resonances of nonoxidized units are present. From all these observations, we can assume for the C6-oxidized KGM sample a nearly complete degree of oxidation.



where IM ) IMMM + IMMGl + IGlMM + IGlMGl. The average block lengths of mannuronic blocks in oxidized KGM is therefore rather short (3.0 ( 0.1), as expected. Epimerization of Oxidized KGM Using Mannuronan C-5-Epimerases. As previously shown, in the oxidized KGM short M blocks are present. Even if the average M-block length is rather short, indeed oxidized KGM appears a suitable substrate for the mannuronan C-5-epimerases. The mannuronan C-5 epimerases introduce some guluronic acid (G) residues in the polysaccharidic chain by C5-epimerization of some of the mannuronic acid residues (see sketch below).

Figure 4. 13C spectrum of oxidized konjak in D2O at T ) 343 K: M ) mannuronic unit, Gl ) glucuronic unit. In the insert “a” the carbonyl spectral region is shown. Inserts “b” and “c” show the spectral region of C4 resonances on the mannuronic and glucuronic units, respectively, with a triad assignment; the solid lines through the experimental spectra are obtained by applying the deconvolution procedure. Resonances evidenced with an asterisk do not belong to M-C4.

In the 13C spectrum of oxidized konjak, see Figure 4, diads are well observable on C1 and C3 resonances of mannuronic acid (M) and on the C3 resonance of glucuronic acid (Gl) (data not shown), while a triad resolution is observable on the C4 resonances of both glucuronic and mannuronic units (see insert “b” and “c” in Figure 4). A few simple remarks lead to the assignment of the observed multiplicities: n-ads distribution must be the same in konjak and oxidized konjak; due to spatial proximity, the chemical shift of C1 depends strongly on the next neighbor unit whereas the chemical shift of C4 depends strongly on the previous residue and only weakly on the next one. Using the results obtained from the deconvolution of the 13 C spectrum and choosing a spectral solution deviating as little as possible from a random distribution, we propose the assignment given in Table 4. Then, the deconvolution of the C4 spectral region allows the average block length, NM, of mannuronic blocks to be obtained; in fact, the deconvolution allows to calculate the areas IMMM, IMMGl, IGlMM, and IGlMGl of the resonances due to MMM, MMGl, GlMM, and GlMGl triads (see the insert “b” in Figure 4). According to a method previously reported for alginates,19 the average block length NM of mannuronic blocks can be calculated IM - IGlGl NM ) ) 3.0 ( 0.1 IMGl

Depending on the specific epimerase, the epimerized products may differ significantly both in the proportion and in the distribution of the guluronic acid residues. Three different epimerases, namely, AlgE1, AlgE4, and AlgE6, were tested and found active with the oxidized konjak as substrate. To characterize the epimerized compounds, a full 1D and 2D NMR study was performed at pD = 7 and at pD = 4. It must be noted that all resonances in the 1H spectrum of the KGME6 derivative are also present in the KGME1 derivative, whereas the spectrum of the KGME4 shows a minor signal multiplicity. Since the assignment obtained for KGME6 takes into account for all possible cases, the assignment of all resonances is discussed in detail only for this case. The 1H NMR spectrum of KGME6, in a phosphatebuffered solution, is shown in Figure 5. It must be noted that no resonances are observed in the 5.7-5.9 ppm range. Since these resonances have been previously assigned6 to the proton H-4 of the 4,5-unsaturated residues formed from M and G units, we conclude that no degradation by β elimination occurs during the epimerization reaction. The assignment of resonances due to G units is reported in Figure 5 and in Table 5. From the above assignment it is worth noting that, as expected, the epimerization on C-5 induces a ring inversion (see the sketch below); in fact, the axial proton on C-1 of some M units becomes equatorial and resonates at 5.043 ppm, the equatorial proton on C-2 becomes axial, and so on.

To understand the multiplicity on the resonance due to the proton G-5, it is worth to recollect the NMR literature on alginates.19-21

1348


Table 3. 1H and

13C

Crescenzi et al.

NMR Assignments of C6-Oxidized KGM in D2O and in Buffered Solution (pD ) 7) at T ) 343 Ka oxidized KGM in D2O

mannuronic acid

oxidized KGM in buffered solution

glucuronic acid

mannuronic acid

glucuronic acid

1H

13C

1H

13C

1H

13C

1H

13C

1 2 3 4

4.680 4.027 3.738 3.914

102.31 72.11 73.74 80.40

4.536 3.371 3.641 3.724

104.59 75.07 76.41 82.88

105.37 75.70 76.94 83.60

3.840b 3.784

77.89 77.89

3.838b 3.908

77.45 77.45

102.87 72.86 74.42 80.09 81.48 78.60 78.99

4.521 3.360 3.626 3.706

5

4.655 4.016 3.717 3.901b 3.867 3.790b 3.720

3.86b 3.79

78.30 78.60

a Chemical shifts are reported in ppm with respect DSS used as an internal standard. 13C NMR assignment has been obtained by means of 2D HMQC experiments. b Diad sensitivity.

Table 4.

13C

Assignment of C6-Oxidized KGM in D2O at 343 K

carbon atom

diads and triads

M-C1

M-M M-Gl M-M M-Gl M-M M-Gl M-M-M M-M-Gl Gl-M-M Gl-M-Gl M-M GlM + GlM Gl-Gl Gl-Gl Gl-M Gl-Gl Gl-M Gl-GL Gl-M Gl-Gl-Gl Gl-Gl-M M-Gl-Gl M-Gl-M

M-C2 M-C3 M-C4

M-C5 Gl-C5 Gl-C1 Gl-C2 Gl-C3 Gl-C4

13C

(ppm)

102.37 102.25 72.14 72.08 73.84 73.64 80.27 80.32 80.39 80.54 78.02 77.76 77.45 104.60 104.58 75.02 75.12 76.54 76.28 83.02 82.81 82.80 82.75

a Key: M ) mannose unit, Gl ) glucose unit; M-M ) mannosemannose diads, i.e., a mannose unit with a mannose unit as a first neighbor; M-M-M ) mannose-mannose-mannose triads, i.e., a mannose unit with mannose units as first neighbors and so on.

It is well-known that the chemical shift value of the proton G-5 of the guluronate residue depends on the chemical nature of the first neighbor sugar residues; i.e., it is triad sensitive. Due to spatial proximity, the chemical shift values of the proton G-5 depends strongly from the next neighbor unit and rather weakly from the adjacent previous unit. In the case of the epimerized KGM, the same effects are present (see the 1H spectrum in Figure 5). An explanation of these effects can be found in the alginate literature;12 in fact, the different orientation of the glycosidic bond in the GG, GM, MM MG diads causes a large variation of the chemical shift of G-5. In the GG and GM diads the proximity of the G-5 to the OH-3 of the adjacent unit causes a strong downfield effect:12 typically, depending on the pH, the G-5 resonance belonging to GG diads lies in the 4.404.50 ppm range, while the G-5 resonance belonging to GM diads lies in the 4.75-4.85 ppm range. In the case of KGME6 and KGME1, the triad sensitivity of the 1H spectrum, GG-5M, MG-5M, GG-5G, MG-5G,

Figure 5. 600.13 MHz 1H NMR spectrum of KGME6, at pD ) 7, and T ) 343 K. Resonances of guluronic acid units have been labeled with G.

allows a quantitative measurement of the enzymatic activity. In the case of KGME4 only the resonance due to the MG5M triad is observed. Therefore, AlgE4 epimerase introduces only MG blocks, whereas AlgE1 and AlgE6 epimerases also introduce G-blocks in the polysaccharidic chain. The 13C spectral assignment of KGME6 has been performed using an HMQC 2D map. An expanded region of the 2D map is shown in Figure 6. It is worth to note that the broad 1H resonance at 5.043 ppm gives two cross-peaks with carbon atoms at 101.79 and 102.84 ppm. Therefore, the carbon resonance C-1 of guluronic acid unit is sensitive to diads assigned to G-1M and G-1G, respectively. Carbon resonances of GG-5M and of MG-5M have been assigned at 69.41 and 69.53 ppm, respectively, whereas the carbon signals of GG-5G and MG-5G are overlapped at ≈69.44 ppm. The resonance due to C-1 belonging to M-1G diads and the resonance of the attached proton are observed at 102.21 and 4.655 ppm, respectively. Note that a shoulder in the 1H spectrum at ≈4.72 ppm


Products from Konjak Glucomannan Table 5. 1H and

13C

NMR Assignments of AlgE6 Epimerized KMG in Buffered Solution (pD ) 7) at T ) 343 Ka mannuronic acid

glucuronic acid

guluronic acid

1H

13C

1H

13C

1

4.655

102.21

4.522

104.60

2 3 4

4.016 3.717 3.905 3.876 3.721b 3.790

72.14 73.67 80.19 80.56 78.17 77.80

3.360 3.626 3.677 3.697 3.788b 3.860

74.95 76.13 83.04 82.60 77.80 77.43

5

MM GlM

GlGl MGl

1H

13C

G-M G-G

5.043 5.043 3.978 4.143 4.214

101.79 102.84 72.57 71.33 81.89

MGM GGM MGG GGG

4.737 4.758 4.441 4.454

69.53 69.41 69.44 69.44

a Chemical shifts are reported in ppm with respect DSS used as an internal standard. 13C NMR assignment has been obtained by means of 2D HMQC experiment. b Diad sensitivity.

Figure 6. Slice of the 4.4-5.1 ppm and 68-108 ppm regions of the 2D map resulting from the HMQC experiment performed on a KGME6 sample, pD ) 7 and T ) 343 K. In the horizontal projection, the 1H spectrum and its assignment are reported.

with the corresponding carbon atom at ≈70 ppm is still unassigned; see Figures 6 and 7. To perform the full triads assignment, all previously reported spectra have been repeated in a different buffer (citrate-phosphate) at pD ) 4. In fact, from previous studies on alginates it is known that partial protonation of the carboxyl group shifts downfield the H-5 resonance of the guluronic units. This diagnostic downfield shift provides the key for a direct identification of the G-5 resonances.12,19 A direct comparison of 1H spectra at pD ) 4 with the corresponding ones at pD ) 7 shows that only the resonances due to H-5 of the guluronic units experience a significant

variation of chemical shift. In parts a and b of Figure 7, the 4.36-5.13 ppm regions of the 1H spectra of KGME4 at pD ) 7 and at pD ) 4 are shown. Since no resonance due to GG blocks is present, the AlgE4 epimerase introduces only MG blocks in the polymeric chain. In agreement with literature data12 at pD ) 4, due to the partial protonation of the carboxyl group, the proton resonance of MG-5M shifts from 4.737 to 4.798 ppm. Note that the unassigned small resonance at 4.720 ppm shifts downfield to 4.777 ppm; thus, this resonance can be attributed to an H-5 of G units. A possible triad still unassigned which may be present in the polysaccharidic chain is the GlG-5M to which we tentatively attribute the resonance at 4.720 ppm. In Figure 8a, an expanded region of the 2D map obtained for a KGME4 sample at pD ) 4 is shown. Note the clear separation between the G-5 and M-1 resonances. In the same 2D map, only the cross-peak at 4.798 and 70 ppm due to MG-5M and the cross-peak at 4.777 and 70 ppm attributed to GlG5M are observable. A unique cross-peak due to the G-1M is observed at 5.049 and 102.264 ppm. In the case of KGME6, the 4.36-5.13 ppm proton region at pD ) 7 and pD ) 4 are shown in parts c and d of Figure 7, respectively. At pD ) 4, the resonances of GG-5M and MG-5M shift 0.082 ppm downfield, whereas the resonance identified with a “/” and attributed to GlG-5M shifts 0.075 ppm downfield. Note that at pD ) 4, due to the overlapping with the Gl-1 resonance, the resonance due to GG blocks is not directly observable; however, the corresponding crosspeak is clearly shown in the 2D HMQC map (see Figure 8b). In the same map, two cross-peaks due to G-1M and G-1G diads are observed, with a clear diad sensitivity of the C1 carbon in the guluronic unit. The full 1H and 13C assignments are reported in Table 5. Percentage of the Epimerization. Due to the strong overlapping of 1H resonances in the spectra of epimerized KGM to measure the signal areas, a full spectral deconvolution must be performed. A full deconvolution of the 4.4-5.1 ppm range of the 1H spectra of all epimerized KGM derivatives has been performed. From this deconvolution the area of all signals resonating in the selected ppm range has been obtained. The percentage of C5-epimerization with respect to the total amount of mannuronic and guluronic units can be calculated as the percentage of the area IG-1 of the signal due to the

1350


Crescenzi et al.

Figure 8. Slice of the 4.4-5.1 ppm and 63-108 ppm regions of the 2D HMQC. 2D map of (a) KGME4 at pD ) 4 and T ) 343 K and (b) KGME6 at pD ) 4 and T ) 343 K. In the horizontal projection, the 1H spectrum and the corresponding assignment are reported.

Note that the integral values obtained from the spectral deconvolution16 can be also used for a further check of the triads assignment. In fact, respecting the triad definition18 Figure 7. Expansion of the 4.4-5.1 ppm spectral region of the 600.13 MHz 1H NMR spectra at T ) 343 K: (a) KGME4 epimerized for 24 h and solubilized in a phosphate-buffered solution at pD ) 7; (b) KGME4 epimerized for 24 h and solubilized in a phosphate/citrate buffered solution at pD ) 4; (c) KGME6 epimerized for 24 h and solubilized in a phosphate-buffered solution at pD ) 7; (d) KGME6 epimerized for 24 h and solubilized in a phosphate/citrate buffered solution at pD ) 4. The resonance evidenced with an asterisk is possibly due to GlG-5M triads.

anomeric proton G-1 of the guluronic units with respect to the total area of anomeric protons of guluronic IG-1 and mannuronic IM-1 units % epimerization )

IG-1 × 100 IG-1 + IG-1

being IM-1 ) IM-1M + IM-1Gl + IM-1G. In Figure 9, the experimental and deconvoluted spectra at pD ) 7 of oxidized KGM treated with AlgE1, AlgE4, and AlgE6 C5-epimerases for 24 and 48 h are shown. The percentage of epimerization for these samples is reported in Table 6. The lengthening of the time of reaction from 24 to 48 h does not cause a marked increase of the percentage of epimerization.

IG-1 ) IMG-5M + IGlG-5M in the case KGME4 and IG-1 ) IMG-5M + IGlG-5M + IGG-5M + IGG-5G + IMG-5G in the case of KGME6. Average Block Lengths of G-Blocks in Epimerized Konjak Derivatives. The deconvolution of the 4.4-5.1 ppm region of the 1H spectra of KGME6 and KGME1 allows calculation12,19 of the average block length NG of guluronic blocks NG )

IG - IMG-5M - IGlG-5M IG-5M

where IG is the integral of all resonances due to G-5, IMG-5M is the integral of the resonance of the MG-5M triad, IGlG-5M is the integral of the resonance of the GlG-5M triad, and IGG-5M is the integral of the resonance of the GG-5M triad. The average block lengths calculated for KGME6 epimerized for 24 and 48 h are 1.44 ( 0.04 and 1.60 ( 0.04,



Figure 9. Expansion of the 4.4-5.2 ppm region in the 1H NMR spectra of oxidized KGM epimerized with different enzymes for different times t: KGME1, t ) 48 h, pD ) 7 (a, b); KGME4, t ) 24 h, pD ) 7 (c, d); KGME4 t ) 48 h, pD ) 7 (e, f); KGME6 t ) 24 h, pD ) 7 (g, h); KGME6, t ) 48 h, pD ) 7 (i,l). Experimental spectra on the left side, deconvoluted spectra on the right side. Table 6. Percentage of Epimerization in the AlgE1, AlgE4, and AlgE6 Epimerized KGMa enzyme AlgE1 AlgE6 AlgE6 AlgE4 AlgE4

epimerization time (h)

% epimerization

48 24 48 24 48

15 ( 1 20 (1 22 ( 1 16 ( 3 22 ( 3

NG 1.30 ( 0.01 1.44 ( 0.04 1.60 ( 0.04

a For AlgE1 and AlgE6 epimerized KGM, the average G-G block length (NG) is also reported.

respectively, whereas for KGME1 epimerized for 48 h the average blocks length is 1.30 ( 0.01 Since the average length of M blocks in oxidized KGM is 3.0 ( 0.1, it is reasonable to find a very short average G block lengths in the epimerized derivatives. It also worth noticing that in all cases IMG-5M > IGG-5M + IGlG-5M. Due to the shortness of M blocks in the starting oxidized KGM, even in KGME1 and KGME6, the probability of finding a G unit between two M units is higher than the probability of finding G blocks. Concluding Remarks. For all the three recombinant

mannuronan C-5 epimerases that we have tested, oxidized konjak can be used as a substrate. When these enzymes work on mannuronan or alginates with long consecutive stretches of M residues, they all act nonrandomly, introducing either long MGM sequences, in the case of AlgE4, or preferentially G-blocks for AlgE1 and AlgE6. The high-field NMR spectra of KGME1, KGME4, and KGME6 were fully assigned. NMR data show that in the case of AlgE4 only single G residues are introduced. In the case of AlgE6, triads MGM comprising isolated guluronic acid residues, that is, MGM, prevail with respect to triads comprising GG pairs. Therefore, despite its tendency to produce G blocks, AlgE6 works on the substrate even if the M blocks are so short that in the majority of cases it is unable to introduce more than one G residue. Although all the enzymes act on oxidized konjak, apparently the majority of the M residues are not accessible to the epimerization. Therefore, the overall effectiveness of the enzyme is significantly lower on artificial uronan than on native alginates.22 The shortness of the M blocks alone cannot rationalize this observation. In fact, in GMG sequences, AlgE1 and AlgE6 can epimerize even single M residues.23 The nonrandom behavior of the epimerases has previously been attributed to two possible mechanisms. One mechanism is a preferred attack mechanism, where the enzyme preferentially epimerizes residues adjacent to a G unit or a GM sequence. The other one is a processive mode of action where the enzyme, after an initial random attack, slides along the polymer chain. Note that both mechanisms imply that the enzyme recognizes more than one residue in the polymer substrate. Moreover, experiments with oligomanuronates of various sizes show that a DP > 6 is necessary to support epimerase activity.24 Even if the three-dimensional structure of the C-5 epimerases is currently lacking, a recent report on an alginate lyase suggests an active cleft interacting with six consecutive residues.25 This observation and the reported lyase23 activity in some of the AlgE epimerises suggest enzymes with an active cleft consisting of several subsites. According to this model, epimerization will occur only on substrates able to saturate all subsites. The partial epimerization of oxidized konjak supports such a model in which the presence of glucuronic acid in the polymer chain might impair the binding of the epimerase to the substrate or affect the sliding of the enzyme along the chain. The presence of the GlGM triad suggests however that glucuronic acid at least can bind in the subsite adjacent to the productive site of the enzyme. Acknowledgment. This work has been carried out with financial support of the European Union (Contract QLK3CT-1999-00034). References and Notes (1) Timell, T. E. AdV. Carbohydr. Chem. 1965, 20, 409-483. (2) Dea, I. C. M.; Morrison, A. AdV. Carbohydr. Chem. Biochem. 1975, 31, 241-312. (3) Maeda, M.; Shimahara, H.; Sugiyama, N. Agric. Biol. Chem. 1980, 44 (2), 245. (4) Okimasu, S.; Kondo, Y. Nippon Shokuhin kogyo Gakkaish 1967, 14, 345.

1352


(5) Yui, T.; Ogawa, K.; Sarko, A. Carbohydr. Res. 1992, 229, 41-55). (6) Crescenzi, V.; Hartmann, M.; de Nooy, A. E. J.; Rori, V.; Masci, G.; Skjåk-Bræk, G. Biomacromolecules 2000, 1 (3), 360. (7) Crescenzi, V.; Dentini, M.; Bernalda, M. S.; Masci, G.; Rori, V.; Skjåk-Bræk, G. Biomacromolecules 2001, 2 (3), 958-964. (8) Ertesvåg, H.; Doseth, B.; Larsen, B.; Skjåk-Bræk, G.; Valla, S. J. Bacteriol. 1994, 176, 2846. (9) Ertesvåg, H.; Høidal, H. K.; Skjåk-Bræk, G.; Valla, S. J. Biol. Chem. 1998, 273, 30927. (10) Svanem, B. I. G.; Skjåk-Bræk, G.; Ertesvåg, H.; Valla, S. J. Bacteriol. 1999, 181, 68. (11) Dave, V.; Sheth, M.; McCarthy, S.; Ratto, J. A.; Kaplan, D. L. Polymer 1998, 39, no. 5, 1139-1148, J. Org. Chem. 1997, 62, 2 (21). (12) Grasdalem, H. Carbohydr. Res. 1983, 118, 255. (13) Guere´ron, M.; Plateu, P.; Decorps, M. Prog. NMR Spectrosc. 1991, 23, 135. (14) (a) Braun, S.; Kalinowski, H.-O.; Berger, S. 150 and More Basic NMR Experiments: a Pratical Course, Wiley-VCH: Weinheim, 1998. (b) Aue, W. P.; Ernst, R. R. J. Chem. Phys. 1975, 64, 2229. (c) Bax, A.; Freeman, R. J. Magn. Reson. 1981, 44, 542. (d) Braunschweiler, L.; Ernst, R. R. J. Magn. Reson. 1983, 53, 521. (e) Bax, A.; Davis, D. G. J. Magn. Reson. 1985, 65, 355. (f) Bax, A.; Subramanian J. Magn. Reson. 1986, 67, 565. (g) Shaka, A. J.; Barker, P. B.; Freeman, R. J. Magn. Reson. 1985, 64, 547.

Crescenzi et al. (15) Bax, A.; Davis, D. G. J. Am. Chem. Soc. 1985, 65, 335. (16) Courtesy of Professor Michele Vacatello, University of Naples, Italy, [email protected]. (17) Bosco, M.; Miertus, S.; Dentini, M.; Segre, A. L. Biopolymers 2000, 54 (2), 115. (18) Tonelli,A. E. NMR spectroscopy and polymer microstructure: the conformational connection; VCH Publishers: New York, 1989. (19) Grasdalem, H.; Larsen, B.; Smidsrod, O. Carbohydr. Res. 1979, 23, 68. (20) Whittington, S. G. Biopolymers 1971, 10, 1481. (21) Whittington, S. G. Biopolymers 1971, 10, 1617. (22) Hartmann, M.; Duun, A. S.; Markussen, S.; Grasdalen, H.; Valla, S.; Skjåk-Bræk, G. Biochim. Biophys. Acta, in press. (23) Glærum, B. I.; Strand, W. I.; Ertesvåg, H.; Skjåk-Bræk, G.; Valla, S. J. Biol. Chem. 2001, 276, 31542. (24) Hartmann, M.; Holm, O. B.; Johansen, G. A.; Skjåk-Bræk, G.; Stokke, B. T. Biopolymers 2002, 63, 77. (25) Yoon, HJ.; Mikami, B.; Hashimoto, W.; Murata, K. J. Mol. Biol. 1999, 290, 505.

BM025613D

Epimerization - American Chemical Society

Recommend Documents