13C-labeled N-acetyl-neuraminic acid in aqueous solution: Detection and quantification of acyclic keto, keto hydrate, and enol forms by 13C NMR spectroscopy

Article abstract of DOI:10.1021/ja077565g

Aqueous solutions of N-acetyl-neuraminic acid (Neu5Ac, 1) labeled with ¹³C at C1, C2, and/or C3 were analyzed by ¹³C NMR spectroscopy to detect and quantitfy the acyclic forms (keto, keto hydrate, enol) present at varying pHs. In addition to pyranoses, solutions contained the keto form, based on the detection of C2 signals at ～198 ppm (～0.7% at pH 2). Spectra of [2-¹³C] and [3-¹³C] isotopomers contained signals arising from labeled carbons at ～143 and ～120 ppm, respectively, which were attributed to enol forms. Solution studies of [1,2,3- ¹³C₃]I substantiated the presence of enol (～0.5% at pH 2). Enol was not detected at pH > 6.0. A C2 signal observed at ～94 ppm was identified as C2 of the keto hydrate (～1.9% at pH 2), based partly on its abundance as a function of solution pH. Density functional theory (DFT) calculations were used to study the effect of enol and hydrate structure on J_CH and J_CC values involving C2 and C3 of these forms. Solvated DFT calculations showed that ²J_C2,H3 in cis and trans enols have similar magnitudes but opposite signs, making this J-coupling potentially useful to distinguish enol configurations. Solvent deuterium exchange studies of 1 showed rapid incorporation of ²H from ²H₂O at H3_axial in the pyranoses at p ²H 8.0, followed by slower exchange at H3_equatorial. The acyclic keto form, which presumably participates in this reaction, must assume a pseudo-cyclic conformation in solution in order to account for the exchange selectivity. Weak ¹³C signals arising from labeled species were also observed consistently and reproducibly in aqueous solutions of ¹³C-labeled 1, possibly arising from products of lactonization or intermolecular esterification.

Full text of DOI:10.1021/ja077565g

Published on Web 08/19/2008
13
C-Labeled N-Acetyl-neuraminic Acid in Aqueous Solution:
Detection and Quantification of Acyclic Keto, Keto Hydrate,
13
and Enol Forms by C NMR Spectroscopy
†
‡
,†
Thomas Klepach, Ian Carmichael, and Anthony S. Serianni*
Department of Chemistry and Biochemistry and the Radiation Laboratory,
UniVersity of Notre Dame, Notre Dame, Indiana 46556-4670
Received October 5, 2007; E-mail: aseriann@nd.edu
1
3
Abstract: Aqueous solutions of N-acetyl-neuraminic acid (Neu5Ac, 1) labeled with C at C1, C2, and/or
13
C3 were analyzed by C NMR spectroscopy to detect and quantitfy the acyclic forms (keto, keto hydrate,
enol) present at varying pHs. In addition to pyranoses, solutions contained the keto form, based on the
13
13
detection of C2 signals at ∼198 ppm (∼0.7% at pH 2). Spectra of [2- C] and [3- C] isotopomers contained
signals arising from labeled carbons at ∼143 and ∼120 ppm, respectively, which were attributed to enol
1
3
forms. Solution studies of [1,2,3-
C
3
]1 substantiated the presence of enol (∼0.5% at pH 2). Enol was not
detected at pH > 6.0. A C2 signal observed at ∼94 ppm was identified as C2 of the keto hydrate (∼1.9%
at pH 2), based partly on its abundance as a function of solution pH. Density functional theory (DFT)
calculations were used to study the effect of enol and hydrate structure on JCH and JCC values involving C2
2
and C3 of these forms. Solvated DFT calculations showed that JC2,H3 in cis and trans enols have similar
magnitudes but opposite signs, making this J-coupling potentially useful to distinguish enol configurations.
2
2
Solvent deuterium exchange studies of 1 showed rapid incorporation of H from
2
H O at H3axial in the
2
pyranoses at p H 8.0, followed by slower exchange at H3equatorial. The acyclic keto form, which presumably
participates in this reaction, must assume a pseudo-cyclic conformation in solution in order to account for
the exchange selectivity. Weak C signals arising from labeled species were also observed consistently
13
13
and reproducibly in aqueous solutions of C-labeled 1, possibly arising from products of lactonization or
intermolecular esterification.
Introduction
Scheme 1. Anomerization of Neu5Ac (1) and Abundances of
Forms in Aqueous Solution at pH 2.0
N-Acetyl-neuraminic acid (Neu5Ac, 1, Scheme 1) is a C9
R-ketoacid often encountered as the terminal residue of complex-
type N-glycans covalently bound to human glycoproteins such
1
,2
as immunoglobulin G (IgG). Neu5Ac is commonly installed
on glycoproteins and glycolipids in 2f3 or 2f6 linkage with
terminal ꢀ-Gal residues via the involvement of the biologically
activated Neu5Ac donor, CMP-N-acetyl-neuraminic acid (CMP-
Neu5Ac), and CMP-Neu5Ac sialyltransferase (EC 2.4.99.1,
3
2
.4.99.4). The presence of Neu5Ac on glycoproteins has been
shown to influence their biological functions. For instance, the
anti- and pro-inflammatory activities of human IgG have been
shown to correlate with the presence and absence of Neu5Ac,
4
respectively, on the Fc fragment. Neu5Ac is derived from C6
and C3 metabolites in ViVo, although the specific forms of these
building blocks depend on the organism. In Escherichia coli,
Neu5Ac is produced from N-acetyl-D-mannosamine (ManNAc)
and pyruvate in an aldol condensation reaction catalyzed by
5
N-acetyl-neuraminate lyase (EC 4.1.3.3). In other organisms,
†
ManNAc or ManNAc-6P and phosphoenolpyruvate (PEP) are
substrates for sialic acid synthases, yielding Neu5Ac or Neu5Ac-
9P, respectively.
Studies in this laboratory of the conformational and dynamic
properties of biologically important oligosaccharides containing
Department of Chemistry and Biochemistry.
Radiation Laboratory.
‡
6
(
(
(
(
1) Dwek, R. A.; Lellouch, A. C.; Wormald, M. R. J. Anat. 1995, 187,
2
79–292.
2) Wormald, M. R.; Rudd, P. M.; Harvey, D. J.; Chang, S.-C.; Scragg,
I. G.; Dwek, R. A. Biochemistry 1997, 36, 1370–1380.
3) Breton, C.; Snajdrova, L.; Jeanneau, C.; Koca, J.; Imberty, A.
Glycobiology 2006, 16, 29R–37R.
(5) Maru, I.; Ohnishi, J.; Ohta, Y.; Tsukada, Y. J. Biosci. Bioeng. 2002,
93, 258–265.
4) Kaneko, Y.; Nimmerjahn, F.; Ravetch, J. V. Science 2006, 313, 670–
6
73.
(6) Tanner, M. E. Bioorg. Chem. 2005, 33, 216–228.
1
1892 9 J. AM. CHEM. SOC. 2008, 130, 11892–11900
10.1021/ja077565g CCC: $40.75  2008 American Chemical Society

13
C-Labeled Neu5Ac in Aqueous Solution
A R T I C L E S
Neu5Ac glycosidic linkages required 1 containing specific sites
of C-enrichment to allow measurements of NMR parameters
cooled, and filtered to remove the resin, and the filtrate was
1
3
1
concentrated in Vacuo to give ∼85 mg of crude syrup. 1D H and
1
3
C NMR analysis showed that methyl 2-O-methyl-ꢀ-D-N-acetyl-
such as trans-glycoside JCH and JCC spin-spin coupling
1
3
1
3
neuraminate methyl ester was the primary product. C NMR: δ
171.81 (C1); 100.60 (C2); 40.61 (C3); 67.82, 53.11, 71.93, 69.41,
constants. The availability of Neu5Ac singly labeled with
C
13
at C1, C2, and C3 stimulated tangential C NMR investigations
7
5
1.24, 64.77 (C4-C9); 176.21 (acetyl CO); 23.48 (acetyl CH
2.40 (glycoside CH ); 54.94 (ester CH ).
The methyl ester was saponified for 12 h in 1 M NaOH, after
which the solution pH was lowered to ∼1.4 with the addition of
3
);
of the solution composition of 1. Prior work has shown that
1
3
3
3
selective C-labeling at the anomeric carbons of aldoses and
ketoses allows the detection and quantification not only of major
tautomeric forms in solution, which are often cyclic, but also
of acyclic forms (aldehydo and keto forms and their hydrates),
+
1
13
Dowex (H ) ion-exchange resin. 1D H and C NMR analyses
showed conversion to the free acid methyl glycoside 2. C NMR:
1
3
7
–9
which are normally present in very low abundance. Knowl-
edge of the solution behavior of 1 contributes to a better
understanding of its biological properties, especially with respect
to identifying potential monomeric forms bound by Neu5Ac-
δ 171.46 (C1); 95.40 (C2); 38.74 (C3); 66.72, 52.13, 70.40, 68.26,
70.15, 63.22 (C4-C9); 174.86 (acetyl CO); 22.15 (acetyl CH
3.56 (ester CH ).
D. NMR Spectroscopy. Solutions (∼300 µL, ∼0.1 M) of C-
3
);
5
3
1
3
1
2
1
3
2 2
labeled compounds in 95:5 v/v H O: H O solvent were prepared
13 1
recognizing enzymes and receptors. We describe herein
C
13
and transferred to 3-mm NMR tubes. Quantitative 1D C{ H}
NMR studies of aqueous solutions of Neu5Ac C-isotopomers
and show that these solutions contain detectable amounts of
acyclic keto, keto hydrate, and enol forms.
NMR spectra were obtained at 25 °C on a Varian UnityPlus 600-
13
MHz FT-NMR spectrometer operating at 150.856 MHz for
C
13
1
and equipped with a 3-mm C/ H microprobe (Nalorac).
13
1
For non-quantitative measurements, C{ H} NMR spectra were
typically obtained with a 36 496 Hz spectral window and 27.75 s
Experimental Section
1
3
1
3
13
13
13
recycle time ( C T
experimental solution conditions, as determined from τnull values
in an inversion-recovery T experiment). Free induction decays
FIDs) were zero-filled once or twice to give final digital resolutions
1
’s were not longer than ∼5 s under the
A. C-Labeled Neu5Ac. [1- C]-, [2- C]-, [3- C]-, and
1
3
[
3
1,2,3- C ]Neu5Ac isotopomers were obtained from Omicron
1
Biochemicals, Inc. (South Bend, IN) and used without further
purification.
B. Preparation of Unlabeled Neu5Ac. Unlabeled 1 was
(
of <0.05 Hz/pt, and FIDs were processed with resolution enhance-
ment (Gaussian or sine-bell functions) to improve resolution and
facilitate the measurement of small J-couplings. The degree of
enhancement was chosen empirically on the basis of the observed
spectral S/N and quality.
prepared by a modification of the procedure described by Czarniecki
10
and Thornton. Two batches of edible bird’s nest (batch A, 27.7 g;
batch B, 28.1 g), obtained from Hsu’s Ginseng Enterprises, Inc.
(
http://english.hsuginseng.com), were each homogenized in 50 mL
of distilled H O using a Waring blender. The homogenate was
diluted to a final volume of 3 L containing H SO at a concentration
of ∼0.025 M. The solution was stirred at 60-70 °C for 2.5 h. A
saturated aqueous solution of Ba(OH) was then added slowly to
13
For quantitative measurements, C NMR spectra were collected
2
1
with H-decoupling during FID acquisition and without nuclear
2
4
Overhauser enhancement (NOE) buildup during the interpulse
delays (35 s) to allow for more reliable determinations of the
percentage of the different forms in solution. A minimal line-
broadening window function was applied prior to signal integration.
2
the mixture until the pH was between 5 and 6. The mixture was
incubated at 4 °C overnight and then vacuum filtered through a
glass microfiber filter. The clear filtrate was concentrated in Vacuo
to ∼250 mL and passed through a 0.3 µm filter, which was washed
with water after the filtration. The combined filtrate and washings
were split into two equal volumes (∼130 mL each) and individually
loaded onto a Dowex 1 × 8 (200-400 mesh) ion-exchange column
Calculations
A. Selection and Geometric Optimization of Model Com-
pounds. Structures 3 and 4-7 were chosen as mimics of 1h and
1e, respectively. Solvated density functional theory (DFT) calcula-
1
2
(
3.0 × 30.0 cm) in the formate form. After loading, the column
tions were conducted within Gaussian03 using the B3LYP
13 14
was washed with 1 L of distilled water and then eluted with a 1.5
L linear formic acid gradient (0-2 M). Fractions (23 mL) were
collected and assayed by spotting fraction aliquots on silica gel
thin layer chromatography (TLC) plates, spraying with molybdate
functional and 6-31G* basis set for geometric optimization, as
15,16
described previously.
Initial torsion angle constraints in 3 were
asfollows:C2-C3-C4-C5torsionfixedat-165°;C3-C2-O2-H,
C3-C2-O2′-H, C3-C4-O4-H, and C2-C1-O1′-H torsions
fixed at 180°; O1-C1-C2-C3 torsion set initially at 90° and
allowedtooptimize.For4and5,theC3-C4-O4-H,C2-C1-O1-H,
and C3-C2-O2-H torsions were fixed at 180°. For 6 and 7, the
C3-C4-O4-H, C3-C2-O2-H, and C2-C1-O1′-H torsions
were fixed at 180°. The C1-C2-C3-C4 torsion angle was fixed
at 0° in 4 and 6 (cis enols) and at 180° in 5 and 7 (trans enols).
Two orientations of the COOH group in 4 and 5 were examined in
which the double bonds were conjugated (O1′-C1-C2-C3 torsion
fixed at 180°) and unconjugated (O1′-C1-C2-C3 torsion was
fixed at 90°). In 6 and 7, the C2-C3-C4-C5 torsion angle was
rotated in 30° increments from 90° to 270° in two O1′-C1-C2-C3
rotamers (90° and 180°). All remaining geometric parameters were
optimized except those identified above.
reagent (1% (w/v) CeSO
H SO ), and charring. Fractions testing positive for Neu5Ac were
2 4
4
-2.5% (w/v) (NH
4 6 7
) Mo O24-10% aq
11
combined, and the resulting solution was lyophilized. Total yield
of Neu5Ac: ∼1.7 g each from batches A and B. The white powder
had a melting point of 184-186 °C and was found to be >98%
1
13
pure by 1D H and C NMR analyses.
C. Preparation of 2-O-Methyl-ꢀ-D-N-acetyl-neuraminic Acid
1
0
(
2). Unlabeled 1 (100 mg, 0.32 mmol) was dissolved in 80 mL
of dry methanol, 3 g of Dowex 50 × 8 (200-400 mesh)
+
ion-exchange resin in the H form was added, and the suspension
was refluxed at 76 °C. The reaction was monitored by silica gel
TLC (butanol:acetic acid:H
visualization (see above). The reaction was stopped after 48 h,
2
O, 2:1:1) using molybdate reagent for
1
1
(
(
7) Maple, S. R.; Allerhand, A. J. Am. Chem. Soc. 1987, 109, 3168–3169.
8) Drew, K. N.; Zajicek, J.; Bondo, G.; Bose, B.; Serianni, A. S.
Carbohydr. Res. 1998, 307, 199–209.
(12) Frisch, M. J.; et al. Gaussian03, Revision A.1; Gaussian, Inc.,
Pittsburgh, PA, 2003.
(13) Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
(14) Hehre, W. J.; Ditchfield, R.; Pople, J. A. J. Chem. Phys. 1972, 56,
2257–2261.
(
9) Zhu, Y.; Zajicek, J.; Serianni, A. S. J. Org. Chem. 2001, 66, 6244–
6
251.
(
10) Czarniecki, M. F.; Thornton, E. R. J. Am. Chem. Soc. 1977, 99, 8273–
(15) Cloran, F.; Zhu, Y.; Osborn, J.; Carmichael, I.; Serianni, A. S. J. Am.
Chem. Soc. 2000, 122, 6435–6448.
8
279.
(
11) Tropper, F. D.; Andersson, F. O.; Grand-Maitre, C.; Roy, R.
(16) Cloran, F.; Carmichael, I.; Serianni, A. S. J. Am. Chem. Soc. 2001,
123, 4781–4791.
Carbohydr. Res. 1992, 229, 149–154.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 36, 2008 11893

A R T I C L E S
Klepach et al.
13
1
13
13
B. DFT Calculations of C- H and C- C Spin-Coupling
Constants. JCH and JCC values were calculated in 3-7 with
1
2
13
Gaussian03 using DFT (B3LYP). Couplings were calculated
1
7
using the previously reported [5s2p1d|3s1p] basis set, and raw
(
unscaled) calculated couplings are reported.
For all calculated geometries and J-couplings, the effect of
solvent water was included by using the Self-Consistent Reaction
1
8
Field (SCRF) and the Integral Equation Formalism (polarizable
1
9
12
continuum) model (IEFPCM) as implemented in Gaussian03.
Results and Discussion
A. Cyclic Pyranose and Acyclic Keto Forms of 1. Neu5Ac
13
labeled with C at the anomeric (C2) carbon was used to detect
and quantify monomeric forms present in aqueous solution
1
3
1
13
(
2
Scheme 1). The C{ H} NMR spectrum of [2- C]1 at pH
.0 contained intense signals arising from labeled carbons in
the R-pyranose (1Rp, 96.53 ppm) and ꢀ-pyranose (1ꢀp, 95.97
2
0,21
ppm)
forms (Figure 1A). The ꢀ-pyranose was significantly
more abundant (91.2%) than the R-pyranose (5.8%) at pH 2.0;
these percentages were 92.1% and 7.5%, respectively, at pH
8
.0. The preference for the ꢀ-pyranose presumably derives from
22–24
the anomeric effect,
which favors an axial C2-O2 bond,
and from the greater steric demand of the COOH group relative
to OH, which favors an equatorial C1-C2 bond. Both effects
are reinforcing. The Rp/ꢀp ratio at pH 2.0 (0.064) was slightly
smaller than the same ratio at pH 8.0 (0.081).
1
3
1
13
The C{ H} NMR spectrum of [2- C]1 at pH 2.0 also
contained a relatively intense signal at 198.20 ppm (Figure 1B),
which was attributed to C2 of the acyclic keto (carbonyl) form
1
3
1
13
Figure 1. Partial C{ H} NMR spectrum of [2- C]1 (150 MHz) at pH
.0 and 25 °C showing signal assignments. (A) Anomeric region. X1 and
2
X2 are putative C2 signals of the acyclic hydrate form (1h). (B) Region
containing the labeled C2 keto (1k) and enol (1e) carbons, and the natural
abundance amide CO (COam) and C1 (COOH) carbons of 1ꢀp and 1ꢀp.
The C1 signals appear as doublets due to the presence of JC1,C2. Several
weak signals in this region were not identified.
(
1k). This signal appeared farther upfield than observed previ-
13
ously for the acyclic keto forms of D-[2- C]ribulose-5P (213.7
ppm), D-[2- C]ribulose-1,5-bisphosphate (211.7 ppm), and
25
13
25
1
13
26
the D-[2- C]pentuloses (∼214 ppm) and probably reflects the
effect of C1 structure on δC2 (COOH vs CH2OH/CH2OP).
Integration of the spectrum acquired at pH 2.0 and 25 °C
indicated that 1k comprises ∼0.7% of the total forms in solution;
this percentage decreased to ∼0.4% at pH 8.0. At pH 2.0, an
Rp:ꢀp:keto ratio of 8:130:1 was observed, whereas at pH 8.0,
the ratio was 19:230:1.
2
5
expected for the acyclic keto hydrate form, 1h. The X2 signal
was assigned tentatively to C2 of 1h on the basis of its relative
abundance (1.9%) and the observed keto hydrate/keto ratio of
∼
2.7, which is comparable to the corresponding ratio for pyruvic
2
7
acid (2.31 ( 0.02). By comparison, 2-ketose phosphates give
keto hydrate/keto ratios of <1 (e.g., 0.14 for D-ribulose-1,5-
bisphosphate; δC2 97.6 ppm), while no keto hydrate form was
detected in C NMR spectra of aqueous solutions of the
D-[2- C]2-pentuloses.
B. Acyclic Hydrate Form of 1. The anomeric region of the
C{ H} NMR spectrum of [2- C]1 at pH 2.0 contained two
1
3
1
13
1
3
relatively strong signals arising from labeled carbons at 95.90
and 93.60 ppm (Figure 1A, signals denoted X1 and X2,
respectively). These chemical shifts are consistent with those
1
3
25,26
Since C21h is unprotonated, experimental support for its
assignment could not be obtained by examining the magnitude
1
of JCH, which has been used previously to identify hydrate C1
(
(
(
(
17) Stenutz, R.; Carmichael, I.; Widmalm, G.; Serianni, A. S. J. Org.
Chem. 2002, 67, 949–958.
1
3
8,9
carbons in aqueous solutions of [1- C]aldoses. Therefore,
two alternative assignment strategies were investigated.
Previous studies have shown that pyruvate anion is much less
extensively hydrated than pyruvic acid in aqueous solution (keto
18) Canc e´ s, M. T.; Mennucci, B.; Tomasi, J. J. Chem. Phys. 1997, 107,
3
032.
19) Cammi, R.; Mennucci, B.; Tomasi, J. J. Phys. Chem. A 2000, 104,
5
631.
20) The chemical shift of C21ꢀp (free acid) was assigned as described
previously (Gervay, J.; Batta, G., Tetrahedron Lett. 1994, 35, 3009–
27–30
hydrate/keto ratios in the anion range from 0.054 to 0.087).
13
1
3
This behavior was confirmed by inspection of C NMR spectra
of pyruvic acid and sodium pyruvate (Figure S1). By analogy
to pyruvate, the intensity of the authentic C21h signal should
decrease significantly upon raising the solution pH from 2.0 to
3
012) using internal dioxane (δ 67.4 ppm) as the reference. All
C
NMR spectra of 1 at pH 2 were referenced in this fashion.
(
21) Assignments of the anomeric distribution of the pyranose forms of l
and related structures have been reported previously from analyses of
3
3
JC1,H3ax and JC1,H3eq couplings: Czarniecki, M.; Thornton, E. R. J. Am.
1
3
Chem. Soc. 1977, 99, 8273–8279; Hori, H.; Nakajima, T.; Nishida,
Y.; Ohrui, H.; Meguro, H. Tetrahedron Lett. 1988, 29, 6317–6320;
M a´ jer, G.; Borb a´ s, A.; Illy e´ s, T. Z.; Szil a´ gyi, L.; B e´ nyei, A. C.; Lipt a´ k,
A. Carbohydr. Res. 2007, 342, 1393–1404.
7.0. Titration of a solution of [2- C]1 at pH 2.0 with NaOH
(26) Vuorinen, T.; Serianni, A. S. Carbohydr. Res. 1990, 209, 13–31.
(27) Chiang, Y.; Kresge, A. J.; Pruszynski, P. J. Am. Chem. Soc. 1992,
114, 3103–3107.
(
(
22) Edward, J. T. Chem. Ind. 1955, 1102.
23) Lemieux, R. U. In Molecular Rearrangements; de Mayo, P., Ed.;
Interscience Publishers: New York, 1964; p 709.
(28) Greenzaid, P.; Luz, Z.; Samuel, D. J. Am. Chem. Soc. 1967, 89, 749–
756.
(24) Juaristi, E.; Cuevas, G. The Anomeric Effect; CRC Press: Boca Raton,
FL, 1995.
(29) Pocker, Y.; Meany, J. E.; Nist, B. J.; Zadorojny, C. J. Phys. Chem.
1969, 73, 2872–2882.
(
25) Serianni, A. S.; Pierce, J.; Barker, R. Biochemistry 1979, 18, 1192–
1
199.
(30) Cooper, A. J. L.; Redfield, A. G. J. Biol. Chem. 1975, 250, 527–532.
1
1894 J. AM. CHEM. SOC. 9 VOL. 130, NO. 36, 2008

13
C-Labeled Neu5Ac in Aqueous Solution
A R T I C L E S
∼
2.9 Hz, and the other is small or zero. Application of the
3
2
projection rule to the three staggered C2-C3 rotamers of 1h
2
predicted small or 0 Hz values for both JCH’s in one rotamer
(that having C1 anti to C4), whereas the remaining two rotamers
yielded a small or 0 Hz coupling to one H3 and a negative
coupling to the other (Scheme S2). Thus, the observation of a
2
single non-zero JCH is consistent with 1h having C2 anti to
C5, and C1 gauche to C4 (either latter gauche conformation
appears consistent with the experimental couplings). The second
2
putative C21h signal, X1, showed small or zero JCH values,
consistent with 1h having C2 anti to C5, and C1 anti to C4.
Solvated DFT calculations on Neu5Ac hydrate mimic 3 were
conducted to evaluate the effect of C2-C3 bond rotation on
2
calculated JC2,H3R/S values and total energies. In these calcula-
tions, the C2-C3-C4-C5 torsion angle was fixed at -165°,
3
which gave calculated JC2,H4 values of ∼0.6 Hz that are
consistent with the NMR results. This treatment assumes that
3
the small experimental JC2,H4 correlates with a relatively rigid
C2-C3-C4-H4 torsion angle of ∼75° in solution. Conforma-
tions of 3 containing C1-C2-C3-C4 torsion angles of 60°
and -60° were found to be lower in energy than that containing
a torsion angle of 180° (Figure S2A), favoring assignment of
the hydrate C2 signal to X2. However, calculated JC2,H3R/S
couplings show that, for torsions near 90°, one coupling is
1
13
13
Figure 2. Partial H-coupled C NMR spectrum (150 MHz) of [2- C]1
at pH 2.0 and 25 °C. (A) C2 signals of 1Rp, 1ꢀp, X1, and X2. (B) The C2
keto signal. (C) The C2 enol signals.
2
resulted in the disappearance of signals X1 and X2 at pH ∼6-7,
but only X2 reappeared at 93.60 ppm upon lowering the pH of
the same solution back to 2.0 with HCl. These results suggest
that X2 is the correct C21h carbon signal.
A second approach to assign the C21h signal involved an
interpretation of the X1 and X2 signal multiplicities observed
∼
-1.5 Hz and the other ∼ -7.5 Hz. These results are in modest
agreement with experiment (∼0 and ∼ (2.9 Hz) for X2. Other
conformational variables not investigated here (e.g., conforma-
tion about the C1-C2 bond) may contribute to the discrepancy.
1
1
1
13
13
JC1,C2 and JC2,C3 values were also computed in 3 as a
in the H-coupled C NMR spectrum of [2- C]1. The
multiplicities of the C2 signals of 1Rp and 1ꢀp (Figure 2A)
served as internal controls. The C21Rp signal appeared as a
doublet of doublets with JCH values of ∼3.9 and ∼7.5 Hz,
whereas the C21ꢀp signal appeared as a pseudo-triplet containing
JCH values of ∼3.5 Hz. These couplings were consistent with
function of the C2-C3-C4-C5 torsion angle (Figure S2B).
Both couplings are relatively insensitive to the C2-C3 bond
torsion, with the former considerably larger than the latter. The
calculated couplings are in reasonable agreement with those
1
1
observed experimentally for X2 ( JC1,C2 ) 69.5 Hz; JC2,C3 )
2
2
44.0 Hz).
the more accurate JC2,H3ax and JC2,H3eq values measured directly
1
3
1
1
13
31 3
3
C. Acyclic Enol Forms of 1. The C{ H} NMR spectrum
from the H NMR spectrum of [2- C]1.
JC2,H4 and JC2,H6
1
3
of [2- C]1 contained relatively strong signals associated with
labeled carbons at 142.75 and 142.85 ppm (Figure 1B). Signals
in this spectral region commonly arise from unsaturated carbons.
Possible candidates are enol forms 1e (Scheme 1) in cis and
trans configurations (Scheme 2). An additional small signal was
also observed at 144.39 ppm (Figure 1B). Integration of these
three signals at pH 2.0 gave a total abundance of ∼0.5%.
Experimental support for the presence of 1e was obtained
values in 1Rp and 1ꢀp are associated with C2-C3-C4-H4
and C2-O6-C6-H6 torsion angles of ∼ -60° and ∼60°,
respectively, and thus their small magnitudes appear as line-
13
broadening in the C multiplets. Signal X1 appeared broadened,
whereas X2 appeared as a doublet with a splitting of ∼2.9 Hz
(
Figure 2A). By comparison, the C21k signal appeared as a
broadened doublet with a splitting of ∼2.1 Hz (Figure 2B).
An inspection of molecular models of 1h showed that, in
geometries having C2 anti to H4, destabilizing 1,3-interactions
exist between the substituents at C2 and C4 that are likely to
destabilize this conformation. Conformations having C2 and H4
1
3
1
13
from the C{ H} NMR spectrum of [3- C]1. This spectrum
1
2
(
31) H NMR data showed significantly different (absolute) JC2,H3ax
2
and JC2,H3eq values in 1Rp (7.7 and 4.3 Hz, respectively), whereas
3
1
gauche would yield a small JC2,H4, undetectable in the H-
nearly identical couplings were observed in 1ꢀp (∼3.6 Hz). As
1
3
3
3
coupled C NMR spectrum (the conformation having C2 anti
to C5 appears more stable). Thus, non-zero JCH values involving
C21h are predicted to involve only the geminal C3 protons. If
reported previously for JC1,H3ax and JC1,H3eq (Hori, H.; Nakajima,
T.; Nishida, Y.; Ohrui, H.; Meguro, H. Tetrahedron Lett. 1988,
2
2
2
9, 6317–6320), JC2,H3ax and JC2,H3eq can be used to distinguish
between 1Rp and 1ꢀp (see ref 34).
(32) Bock, K.; Pedersen, C. Acta Chem. Scand. 1977, B31, 354–358.
2
the signal at 93.60 ppm (X2) is C21h, then one putative JCH is
J. AM. CHEM. SOC. 9 VOL. 130, NO. 36, 2008 11895

A R T I C L E S
Klepach et al.
Scheme 2. Enol Forms of 1
contained intense signals at 40.70 and 39.50 ppm attributed to
C31Rp and C31ꢀp, respectively (data not shown). While additional
weaker signals appeared in the same region, they could not be
assigned confidently to C3 of 1k and 1h. However, a moderately
strong signal arising from a labeled carbon was observed at
1
19.85 ppm (Figure 3A) that was tentatively assigned to C3 of
1
e.
1
13
13
13
H-Coupled C NMR spectra of [2- C]1 and [3- C]1 were
obtained to measure JCH values involving the putative C2 and
C3 carbons of 1e. The putative C2 signals appeared as doublets
with JCH ) (3.2 (major signal) and (3.4 Hz (minor signal)
2
(
Figure 2C), consistent with C2 being coupled to H3 ( JC2,H3)
3
or H4 ( JC2,H4). The putative enol C3 signal was split into a
doublet of doublets with JC3,H3 ) 177.5 Hz and a longer-range
1
JCH ) (4.7 Hz (Figure 3B), results consistent with the structure
of 1e. These data indicate that one of the two possible JCH values
involving C2 ( JC2,H3, JC2,H4) and C3 ( JC3,H4, JC3,H5) must be
1
3
1
13
Figure 3. Partial C{ H} NMR spectrum of [3- C]1 at pH 2.0 and 25
C. (A) Region showing the natural abundance C1 and amide CO signals
°
2
3
2
3
of 1Rp and 1ꢀp, and the C3 signal (labeled) of 1e. (B) C3 signal of 1e in
1 13 13 1
small or zero.
the H-coupled C spectrum of [3- C]1, showing the JCH and longer-
range JCH values.
Given the low intensities of the putative C2 and C3 enol signals
and the possibility of attributing their presence to low-level
The effects of C1-C2 and C3-C4 bond rotation on
calculated JCH values were investigated in simplified enol
mimics 6 and 7 (Scheme 4). These structures were geometrically
optimized as a function of the C2-C3-C4-C5 torsion angle
13
contaminants, 1 was prepared with contiguous C-labeling at
1
C1-C3 to confirm backbone carbon connectivities via JCC and
2
13
JCCC values. The C NMR spectrum of the trilabeled isotopomer
contained C2 and C3 signals at the expected chemical shifts with
comparable intensities and with the expected signal multiplicities
(
90°-270°) in two O1-C1-C2-C3 rotamers (90° and 180°).
The latter torsions were chosen to evaluate the effect of double-
bond conjugation on the calculated couplings. In most cases,
trans enol 7 was lower in energy than cis enol 6, although in a
few cases the opposite was observed, presumably due to
(
Figure 4). Backbone connectivity between C2 and C3 was
1
established from the mutual JCC observed in each carbon signal
1
1
(
JC2,C3 ) 80.6 Hz). JC1,C2 ) 76.6 Hz, as measured at C2, and
2
33
JC1,C3 ) (12.4 Hz, as measured at C3 (sign unknown). The
2
intramolecular H-bonding in 6 that was not present in 7. JC2,H3
values showed a significant dependence on the C2-C3-C4-C5
torsion angle (Figure S5B,D), and calculated JC2,H3 values were
similar to those found in 4 and 5 (negative in 6 and positive in
). JC2,H3 values shifted to more negative (less positive) values
1
large JC2,C3 is consistent with one-bond couplings in CdC
1
fragments, and JC1,C2 is ∼10 Hz larger than the corresponding
2
34
coupling in 1Rp and 1ꢀp at pH 2.0. The distinctive values of
these couplings also allowed an assignment of the weak C1 carbon
signal of the enol form at 172.41 ppm (Figure S3). Overall, these
results support the conclusion that 1e is present in aqueous solutions
of 1 at low pH.
2
7
as the C1-O1 bond (carbonyl) was brought into conjugation
1
with the C2-C3 bond. Calculated JC3,H3 values shifted to
smaller values in the conjugated form of 6, whereas conjugation
JCH values were calculated in enol structural mimics 4 and 5
in their protonated states and with O1′ anti to C3 to optimize
1
1
increased JC3,H3 in 7 (Figure S5A,C). JC3,H3 was uniformly
larger in 7 than in 6 in the conjugated structures, whereas JC3,H3
values were similar in the non-conjugated structures.
1
2
conjugation of the double bonds (Scheme 3). Calculated JC2,H3
values in 4 and 5 were similar in magnitude, but the coupling
in cis enol 4 was negative, while that in trans enol 5 was
3
Calculated JC2,H4 values in 6 and 7 were smallest (∼2 Hz)
3
in 7. The experimental JC2,H4 was predicted to be small or zero
2
positive. This result suggests that the sign of JC2,H3 could be
(
see above). However, it should be noted that the electronegative
N-acetyl substituent at C5 in 1 (not present in 6 and 7) may
affect this coupling, especially if it orients anti to H4. The latter
used to distinguish between cis and trans enol configurations.
1
JC3,H3 is ∼10 Hz larger in the trans enol (Scheme 3).
Calculations of total energies of both structures showed that
3
conformation would reduce JC2,H4 by ∼1 Hz, thus rendering it
the trans configuration 5 is more stable than 4 by ∼3 kcal/mol.
more consistent with the experimental data.
1
1
2
JC1,C2, JC2,C3, and JC1,C3 values were calculated in 6 and
7 as a function of the C2-C3-C4-C5 torsion angle (Figures
S6 and S7). Calculated JC2,C3 values were larger than JC1,C2
values, in agreement with the experimental observations,
although the difference was larger in the calculations.
1
(
33) These values differ from those measured in 1k ( JC1,C2 = 65.1 Hz,
1
1
1
JC2,C3 = 42.2 Hz) and for 1h (X2) ( JC1,C2 ) 69.5 Hz, JC2,C3 ) 44.0
1
1
Hz).
(
34) Klepach, T.; Zhang, W.; Carmichael, I.; Serianni, A. S. J. Org. Chem.
2
008, 73, 4376–4387.
1
1896 J. AM. CHEM. SOC. 9 VOL. 130, NO. 36, 2008

13
C-Labeled Neu5Ac in Aqueous Solution
A R T I C L E S
1
3
1
13
Figure 4. (A) Partial C{ H} NMR spectrum of [1,2,3- C3]1, showing the putative C2 and C3 enol signals. Signals labeled “u” were unassigned and
1
2
possibly arise from other enols. (B,C) Expansions of the C2 and C3 signals in (A), respectively, showing their multiplicities and constituent JCC and JCC
values.
Scheme 3
These results are consistent with those reported for the
enol-keto equilibrium of pyruvate, in which the [enol]/[keto]
ratio is approximately 100-fold greater for pyruvic acid than
for the pyruvate anion. A rationale has been offered for this
behavior based on the proposed greater energy gap between
keto pyruvate and its enol compared to that for the protonated
2
7
species.
1
3
C NMR chemical shifts for the enol forms of R-keto acids
like 1 have not been well documented in the literature. However,
the C shifts of propen-2-ol 8 (acetone enol) show CR at 156.8
ppm and Cꢀ at 95.3 ppm in 2-propanol solvent at 44 °C. These
shifts provide a crude calibration for the putative C2 and C3
shifts in 1 and suggest that C21e should be downfield of C31e,
as observed experimentally.
1
3
36
Scheme 4
D. pK
Rp, 1ꢀp, 1k, and 1e exhibited sufficient pH dependencies to
allow estimations of pKa values for COOH ionization. Titration
a
Values of Tautomers of 1. The C2 chemical shifts of
1
1
3
data (Figure 5) obtained on [2- C]1 and fit to a modified form
3
7
of the Henderson-Hasselbach equation gave the following
results: 1Rp, 2.5 ( 0.1; 1ꢀp, 2.6 ( 0.1; 1k, 2.1 ( 0.1 and 1.8
(
0.1; 1e, 5.5 ( 2.6. The large error in the pKa of 1e results
from the lack of data points at the higher pHs of the titration
curve (detection of the C2 enol signal is lost at the higher pHs).
The true pKa is expected to be close to 5.5, based on visual
inspection of the curve (Figure 5D).
The nearly identical pKa values for 1Rp and 1ꢀp were
intermediate in magnitude between those for 1k and 1e, with
2
Calculated JC1,C3 values were positive in sign and averaged
+7 Hz, compared to the experimental value of 12.4 Hz. It
∼
should be appreciated, however, that the effect of the C2-O2
bond torsion on these couplings could be appreciable,
mediated by O2 lone-pair orbital effects on the C1-C3
molecular fragment. In the present calculations, only one
C3-C2-O2-H torsion angle was examined (180°), and
presumably appropriate averaging of this torsion that properly
mimics behavior in solution would improve the agreement
1
k being most acidic. These results are comparable to those
found with pyruvic acid, where pKa values of the keto and enol
2
7
forms have been reported to be 1.97 and 3.79, respectively.
3
5
(
35) DFT calculations on 6, in which the C1-C2-O2-H and C3-C2-
O2-H torsion angles were fixed at 60° and-60°, respectively, gave
between experiment and theory.
Detection of the putative C2 and C3 enol signals in
1
3
C
2
calculated JC1,C3 values of +8.8 and +9.0 Hz, respectively (C2-
1
3
13
NMR spectra of [2- C]1 and [3- C]1, respectively, de-
pended on solution pH. Signals were observed in spectra
obtained on solutions at pH 2.0 but not on the same solutions
adjusted to pH 7.0. Enol signal detection could be recovered
by adjusting solutions at pH 7.0 back to pH 2.0 with HCl.
C3-C4-C5 torsion fixed at 180°). These results confirm that the effect
2
of the C2-O2 bond torsion on JCCC values in unsaturated systems
mimics that observed in saturated systems ( Klepach, T.; Carmichael,
I.; Serianni, A. S., unpublished results).
(
36) Sojka, S. A.; Poranski, C. F., Jr.; Moniz, W. B. J. Am. Chem. Soc.
1975, 97, 5953–5955.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 36, 2008 11897

A R T I C L E S
Klepach et al.
Figure 5. Dependencies of the C2 chemical shifts of 1Rp (A), 1ꢀp (B), 1k (C), and 1e (D) on solution pH. Filled blue circles and blue lines are fits to the
experimental data shown in either red or green. Curve fitting gave the pKa values reported in the text.
-
1
It was not possible to estimate the pKa of the putative hydrate
form of 1 since the intensity of signal X2 decreased rapidly with
increasing pH, preventing sufficient digitization of a titration
curve.
furanose and furanose-pyranose rate constants of 0.02-2.2 s
were determined, but rate constants for pyranose-pyranose
exchange were estimated at less than 0.01 s-
1
.
F. Solvent Deuterium Exchange in 1. Given the relatively
1
3
1
E. Exchange Rates between Tautomers of 1. C Saturation-
high abundance of enol forms of 1 in aqueous solution, H NMR
3
7
2
2
transfer (ST) experiments were conducted on an aqueous
studies were conducted on unlabeled 1 in H2O solvent at p H
.0 to investigate solvent deuterium exchange at C3. The H3ax
1
3
solution of [2- C]1 at pH 2.0 and 30 °C to identify species
in chemical exchange with the acyclic keto form. Selective
irradiation of the C2 signal of 1k did not affect the C2 signal
intensities of 1Rp, 1ꢀp, 1e, or 1h, suggesting that the
unidirectional rate constants for ring-opening, dehydration,
8
1
and H3eq signals in 1Rp and 1ꢀp in H2O solvent at pH 8.0
were assigned (Figure 6A) on the basis of the relative
3
3
magnitudes of JH3ax,H4 and JH3eq,H4.
-
1
2
and enolization were very small (<0.05 s ) under these
Solvent deuterium exchange was very rapid at p H 8.0 and
25 °C (Figure 6B), with virtually full and selective exchange
observed at H3ax after 47 min. H NMR data obtained after
incubation of the same reaction mixture for 6 days were
identical to those obtained after 47 min (spectrum not shown).
These results are consistent with solvent deuterium exchange
properties for 1 reported previously by Schmidt and Friebo-
1
3
solution conditions. Control C ST experiments on aqueous
1
3
1
solutions of D-[1- C]threose at pH 2.0 showed strong
reduction of the C1 signals from furanose and hydrate forms
upon selective irradiation of C1 of the acyclic aldehydo form,
3
8
as expected. ST experiments on Neu5Ac at higher tem-
peratures were not conducted to avoid thermal degradation
of the sample.
4
0
lin.
The small unidirectional rate constants of anomerization
observed for 1 are consistent with overall rates of anomerization
of 3-deoxy-D-manno-octulosonic acid (KDO) at pH 7.90 and
The facile and selective exchange of H3ax in 1Rp and 1ꢀp
presumably occurs via the acyclic keto form 1k. If so, 1k
probably assumes a relatively stable pseudo-cyclic conformation
in solution, containing relatively distinct and persistent C3-H3ax
and C3-H3eq bonds to account for the observed exchange
selectivity. The possibility of H/ H exchange via cyclic
structures was eliminated through studies of 2-O-methyl-ꢀ-D-
N-acetyl-neuraminic acid (2). In contrast to 1, solvent deuterium
1
3
39
2
5 °C measured by 2D C exchange spectrosoopy. Under
solution conditions more basic than those used for 1, which
should enhance exchange rates between anomers, furanose-
1
2
(
37) The following equation was used to fit the NMR chemical shift data:
a a
n(pK -pH) n(pK -pH)
δ(pH) ) [δbase + δacid × 10
]/[1 + 10
], where δacid
and δbase are the chemical shifts at the acid and base plateaus and n is
the Hill coefficient. Parameters δbase, δacid, pKa, and n were optimized
simultaneously during the data fit using the Solver Routine in Microsoft
Excel.
(39) Fesik, S. W.; Kohlbrenner, W. E.; Gampe, Jr., R. T.; Olejniczak, E. T.
Carbohydr. Res. 1986, 153, 136–140.
(
38) Serianni, A. S.; Pierce, J.; Huang, S.-G.; Barker, R. J. Am. Chem.
Soc. 1982, 104, 4037.
(40) Schmidt, H.; Friebolin, H. J. Carbohydr. Chem. 1983, 2, 405–413.
1
1898 J. AM. CHEM. SOC. 9 VOL. 130, NO. 36, 2008

13
C-Labeled Neu5Ac in Aqueous Solution
A R T I C L E S
Scheme 5. Potential Lactones of 1 (Keto, Keto Hydrate, and Enol
Forms)
confirmed. The latter problem is caused by the relatively limited
1
3
literature on C chemical shifts of these forms and the
1
3
underlying concern that weak signals observed in C NMR
spectra may be due to low-level contaminants present in the
sample.
13
The observation of weak C2 signals in the C NMR spectrum
1
3
of [2- C]1, consistent with the presence of enol forms, was
1
3
13
followed up by studies of the [3- C] and [1,2,3- C3] isoto-
pomers. The data, when taken collectively, strongly support the
presence of detectable levels of Neu5Ac enol in solution. While
the predominant configuration could not be determined experi-
mentally, DFT calculations suggest that the trans form is
probably preferred. DFT calculations also predict that the cis
and trans enol configurations can be distinguished on the basis
2
of the different signs of JC2,H3 in their structures. This may
prove useful in future structural characterization of these and
related species.
Of the three acyclic monomeric forms of Neu5Ac, confirma-
tion of the presence of the keto hydrate proved most challenging,
and the present data cannot be considered unequivocal. How-
ever, when the present data are viewed collectively, a likely
C2 hydrate signal emerges (X2).
1
Figure 6. Partial H NMR spectrum (600 MHz) of 1 immediately after
2
dissolution in H2O (A) and after incubation at 25 °C for 150 h (B). Signals
from H3ax in 1Rp and 1ꢀp are nearly absent, whereas those from H3eq
doublets) remain observable.
Exploitation of the pH dependencies of C2 chemical shifts
yielded pKa values for the cyclic, keto, and enol forms of
Neu5Ac. The enol form is least acidic, and it is assumed that
the ionization being measured is that of the COOH group, and
not the enol C2 OH. Presumably, raising the pH above 6-7,
where the abundance of the enol form decays, results in the
ionization of the C2 hydroxyl, thus destabilizing the enol due
to the presence of two negative charges in close proximity.
(
exchange was not observed in 2 at either H3ax or H3eq after 21
days at p H 12.0 and 25 °C.
2
13
13
Particularly striking about the C NMR spectra of C-labeled
Neu5Ac was the presence of weak but reproducible signals that
could not be attributed to the expected cyclic and acyclic
monomeric forms of the molecule. This observation contrasts
13
8
9
Conclusions
with prior studies of C-labeled aldopentoses and aldohexoses,
where essentially all signals arising from labeled species could
be assigned to monomeric forms. An identification of these
unknown Neu5Ac species was not attempted, but it is noted
that Neu5Ac lactonization can potentially occur in aqueous
solution, especially at low pH, giving the structures shown in
Scheme 5. These species may account for the additional keto,
enol, and hydrate signals observed in C NMR spectra of C-
labeled Neu5Ac. In addition to monomeric lactones, intermo-
lecular dimerization and oligomerization (esterification) involv-
ing the COOH and OH functionalities in different Neu5Ac
Aqueous solutions of N-acetyl-neuraminic acid (1) contain
mainly the ꢀ-pyranose form, with the R-pyranose second in
abundance. In addition to these cyclic forms, three acyclic forms
were detected at low levels (0.5-1.9% at pH 2). These acyclic
1
3
forms are difficult to observe by C NMR spectroscopy at
natural abundance, but detection and quantification are greatly
1
3
13
1
3
improved when measurements are made with C-labeled
samples, especially those singly labeled at C2. Prior studies have
13
utilized C-labeling of 1 and its homopolymers (e.g., polysialic
acid, PSA) to investigate the anomeric specificity of CMP-N-
4
1
acetyl-neuraminic acid synthetase and PSA structure in Vitro
4
2
and in ViVo (E. coli).
(41) Ambrose, M. G.; Freese, S. J.; Reinhold, M. S.; Warner, T. G.; Vann,
W. F. Biochemistry 1992, 31, 775–780.
The characteristic C2 chemical shift of the keto form of 1
makes its detection and quantification relatively straightforward,
whereas the presence of hydrate and enol forms is not easily
(
42) Azurmendi, H. F.; Vionnet, J.; Wrightson, L.; Trinh, L. B.; Shiloach,
J.; Freedberg, D. I. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 11557–
11561.
J. AM. CHEM. SOC. 9 VOL. 130, NO. 36, 2008 11899

A R T I C L E S
Klepach et al.
molecules may also contribute to these weak signals, with their
relative intensities dependent on Neu5Ac concentration.
Solvated DFT calculations conducted on structural mimics
of 1h and 1e have provided additional information on which to
base signal assignments. These calculations, however, involve
inherent structural assumptions that were not fully explored,
especially those regarding the preferred conformations of these
acyclic species in solution. Related to this limitation is the
fact that Neu5Ac undergoes facile solvent deuterium exchange
in aqueous solution, and this exchange occurs more rapidly at
the axial H3 in both 1Rp and 1ꢀp. This exchange correlates
with the relatively high abundance of the keto form in solution.
The observed exchange selectivity implies that the acyclic keto
form probably retains some cyclic character in solution, possibly
assuming a pseudo-cyclic conformation in which H3ax and H3eq
retain orientations similar to those found in the cyclic pyranoses.
The fact that solutions of Neu5Ac contain acyclic keto,
hydrate, and enol forms may have implications for its biological
functions. At low pH uncharacteristic of most biological
environments, the enol and hydrate forms are relatively abundant
in aqueous solution. However, as the pH approaches the
physiological value of ∼7.0, these forms decrease substantially,
thus reducing (but not necessarily eliminating) the likelihood
of their participation as substrates in enzyme-catalyzed reactions.
However, in subcellular regions of low pH (e.g., lysosomes) or
in the low-pH environment of the stomach, these forms are more
prevalent. The extent to which these solution behaviors influence
Neu5Ac biological functions remains to be explored.
4
3
Acknowledgment. This work was supported by a grant from
the National Institutes of Health (GM059239, A.S.). The Notre
Dame Radiation Laboratory is supported by the Office of Basic
Energy Sciences of the United States Department of Energy. This
is Document No. NDRL-4742 from the Notre Dame Radiation
Laboratory. The authors thank Omicron Biochemicals, Inc. for the
1
3
generous gift of the C-labeled Neu5Ac isotopomers.
Supporting Information Available: Scheme S1, projections
(
43) Larger disparities between calculated and experimental J-couplings
are not considered to arise from inherent limitations of the DFT
methodology but rather from simplifications and assumptions in the
structural models used to calculate the couplings. In saccharides, the
latter often involve the conformational properties of exocyclic OH
groups, which can exert a major effect on both JCH and JCC values.
Density functional theory with the B3LYP functional is known to
produce accurate structural and energetic parameters in most organic
molecules. This particular functional is specifically parametrized to
reproduce accurate thermochemistry and matches closely results from
high-level composite methods such as the Gaussian03 approach. Work
in this laboratory and by other investigators has shown high agreement
between B3LYP predictions of indirect nuclear spin-spin coupling
constants and calculated couplings obtained by accurate high-level
coupled-cluster-based calculations. When structural models that closely
2
for JC2,H3R/S in 1h conformers and predicted couplings; Figure
13
1
S1, C{ H} NMR spectra of pyruvic acid and sodium pyruvate;
Figure S2, effect of the C2-C3-C4-C5 torsion angle in 3 on
2
1
calculated total energies and JC2,H3R/S and JCC values; Figure
1
3
1
S3, C1 signals of 1Rp, 1e, and X1 in the partial C{ H} NMR
spectrum of [1,2,3- C3]1; Figure S4, C2 signals of 1Rp, 1ꢀp,
and X1 in the partial C{ H} NMR spectrum of [1,2,3- C3]1;
Figure S5, effect of the C2-C3-C4-C5 torsion angle on
calculatedJCH in6and7;FigureS6,effectoftheC2-C3-C4-C5
1
3
13
1
13
1
torsion angle on calculated JCC in 6 and 7; Figure S7, effect of
the C2-C3-C4-C5 torsion angle on calculated JC1,C3 in 6
2
3
2
mimic behavior in solution are available, J and J values (C-H and
C-C) calculated by DFT are usually within 0.2-0.4 Hz of the
experimental values. In general, absolute errors in calculated couplings
are 5-6% (underestimated) for JCH and 5-10% (overestimated) for
JCC (Cloran, F.; Carmichael, I.; Serianni, A. S. J. Phys. Chem. A 1999,
and 7; and complete ref 12. This material is available free of
charge via the Internet at http://pubs.acs.org.
1
03, 3783–3795).
JA077565G
1
1900 J. AM. CHEM. SOC. 9 VOL. 130, NO. 36, 2008