Quantification of Nucleotide-Activated Sialic Acids by a Combination of

Apr 29, 2010 - nucleotide-activated form is one pacemaker for the highly variable sialylation of glycoconjugates. Hence, the deter- mination of CMP-si...
0 downloads 0 Views 3MB Size
Anal. Chem. 2010, 82, 4591–4598

Quantification of Nucleotide-Activated Sialic Acids by a Combination of Reduction and Fluorescent Labeling Sebastian P. Galuska,*,† Hildegard Geyer,† Birgit Weinhold,‡ Maria Kontou,§ Rene´ C. Ro¨hrich,† Ulrike Bernard,‡ Rita Gerardy-Schahn,‡ Werner Reutter,§ Anja Mu¨nster-Ku¨hnel,‡ and Rudolf Geyer† Institute of Biochemistry, Faculty of Medicine, University of Giessen, Friedrichstrasse 24, D-35392 Giessen, Germany, Institute of Cellular Chemistry, Hannover Medical School, Carl-Neuberg-Strasse 1, D-30625 Hannover, Germany, and Institute of Biochemistry and Molecular Biology, Charite´-University Berlin, Arnimallee 22, D-14195, Berlin-Dahlem, Germany Sialic acids usually represent the terminal monosaccharide of glycoconjugates and are directly involved in many biological processes. The cellular concentration of their nucleotide-activated form is one pacemaker for the highly variable sialylation of glycoconjugates. Hence, the determination of CMP-sialic acid levels is an important factor to understand the complex glycosylation machinery of cells and to standardize the production of glycotherapeutics. We have established a highly sensitive strategy to quantify the concentration of nucleotide-activated sialic acid by a combination of reduction and fluorescent labeling using the fluorophore 1,2-diamino-4,5-methylenedioxybenzene (DMB). The labeling with DMB requires free keto as well as carboxyl groups of the sialic acid molecule. Reduction of the keto group prior to the labeling process precludes the labeling of nonactivated sialic acids. Since the keto group is protected against reduction by the CMP-substitution, labeling of nucleotide-activated sialic acids is still feasible after reduction. Subsequent combination of the DMB-high-performance liquid chromatography (HPLC) application with mass spectrometric approaches, such as matrix-assisted laser desorption/ ionization time-of-flight-mass spectrometry (MALDI-TOFMS) and electrospray-ionization (ESI)-MS, allows the unambiguous identification of both natural and modified CMP-sialic acids and localization of potential substituents. Thus, the described strategy offers a sensitive detection, identification, and quantification of nucleotide-activated sialic acid derivatives in the femtomole range without the need for nucleotide-activated standards. Sialic acids are the most abundant terminal monosaccharides of eukaryotic glycoconjugates. So far, more than 50 different sialic acids have been described.1 All members are derivatives of neuraminic acid (5-amino-3,5-dideoxy-D-glycero-D-galacto-non-2ulopyranosonic acid), an R-keto acid with a nine-carbon backbone. * To whom correspondence should be addressed. Phone: +49 641 9947415. Fax: +49 641 9947419. E-mail: [email protected]. † University of Giessen. ‡ Hannover Medical School. § Charite´-University Berlin. (1) Angata, T.; Varki, A. Chem. Rev. 2002, 102, 439–469. 10.1021/ac100627e  2010 American Chemical Society Published on Web 04/29/2010

The most frequently expressed sialic acids are N-acetylneuraminic acid (Neu5Ac), N-glycolylneuraminic acid (Neu5Gc), and 2-keto3-deoxy-D-glycero-D-galacto-nonoic acid (KDN). The high diversity arises from the various substitutions of the amino and/or hydroxyl groups by sulfation, methylation, lactylation, lactonization, and acetylation.2-4 The terminal positions of sialic acid monomers and/or polymers in glycoconjugates make them suitable for cellular communication elements, involved in the formation of contacts with components of the same (cis-contacts) or neighboring (transcontacts) cell, with extracellular matrix and pathogens that invade the system.5-7 Moreover, sialylation influences the quality of effector-molecules (e.g., erythropoietin) and the biological stability of glycoconjugates. These phenomena explain the essential role of sialic acids in many biological processes and the lethal phenotype of sialic acid-negative mice.8 Furthermore, sialylated glycoconjugates are involved in many pathological events, such as, virus infection as well as metastasis formation and progression of a variety of tumors.9-12 Such pathological events can be influenced by the administration of unnatural sialic acid precursors.13,14 For example, the polysialylation of glycoproteins on tumor cells can be reversibly inhibited by N-propanoylmannosamine (ManNProp).15,16 ManNProp is converted to an unnatural sialic acid instead of the natural sialic (2) (3) (4) (5) (6) (7) (8)

(9) (10) (11) (12) (13) (14) (15) (16)

Schauer, R. Glycoconjugate J. 2000, 17, 485–499. Schauer, R. Zoology (Jena, Ger.) 2004, 107, 49–64. Varki, A. Glycobiology 1992, 2, 25–40. Varki, A. Cell 2006, 126, 841–845. Varki, A. Nature 2007, 446, 1023–1029. Traving, C.; Schauer, R. Cell. Mol. Life Sci. 1998, 54, 1330–1349. Schwarzkopf, M.; Knobeloch, K. P.; Rohde, E.; Hinderlich, S.; Wiechens, N.; Lucka, L.; Horak, I.; Reutter, W.; Horstkorte, R. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 5267–5270. Varki, N. M.; Varki, A. Lab. Invest. 2007, 87, 851–857. Crocker, P. R.; Paulson, J. C.; Varki, A. Nat. Rev. Immunol. 2007, 7, 255– 266. Comstock, L. E.; Kasper, D. L. Cell 2006, 126, 847–850. Ohtsubo, K.; Marth, J. D. Cell 2006, 126, 855–867. Keppler, O. T.; Horstkorte, R.; Pawlita, M.; Schmidt, C.; Reutter, W. Glycobiology 2001, 11, 11R–18R. Dube, D. H.; Bertozzi, C. R. Nat. Rev. Drug Discovery 2005, 4, 477–488. Mahal, L. K.; Charter, N. W.; Angata, K.; Fukuda, M.; Koshland, D. E., Jr.; Bertozzi, C. R. Science 2001, 294, 380–381. Bork, K.; Gagiannis, D.; Orthmann, A.; Weidemann, W.; Kontou, M.; Reutter, W.; Horstkorte, R. J. Neurochem. 2007, 103, 65–71 (Suppl. 1).

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4591

acid precursor N-acetylmannosamine (ManNAc).17 Although the resulting N-propanoylneuraminic acid (Neu5Prop) differs from Neu5Ac by only a single methylene group, polysialylation, for instance, is inhibited.18 In addition, an increased resistance of Neu5Prop against circulating sialidases has been shown to improve the stability and medical effectiveness of therapeutic glycoproteins.19 Incorporation of natural as well as artificially modified sialic acids into glycoconjugates needs metabolic activation of the molecules.20,21 In contrast to other monosaccharides, however, sialic acids are activated by CTP and not by UTP or GTP. Moreover, the localization of activation in the nucleus is unique, whereas all other monosaccharides are activated in the cytosol. The reason for the nuclear localization of CMP-sialic acid synthetase is still unknown.22 After activation, CMP-sialic acid is transported into the Golgi lumen with the support of a specific carrier, thus allowing the transfer of sialic acid to nascent glycoproteins and glycolipids by a panel of different sialyltransferases. The sialylation of glycoconjugates is variable and depends inter alia on the concentration of CMP-sialic acids.23 Dysfunction in the regulation of the sialic acid metabolism can lead to several diseases.24-26 In the case of sialuria, for example, the feed-back regulation of the key enzyme of sialic acid biosynthesis by CMPsialic acid is impaired. This defect leads to an increased urinary excretion of free sialic acid accompanied by mental retardation of the affected patients.27,28 To analyze the complex role of CMP-sialic acids in the physiology of vertebrates in detail and to control the activation of different sialic acids to produce defined glycoconjugates for therapeutic application, sensitive methods for the quantitative analysis of CMP-sialic acids are necessary. So far, two highperformance liquid chromatography (HPLC)-based methods are available which require less than 1 µg of material for an exact quantification of CMP-Neu5Ac.29,30 In these cases, CMP-Neu5Ac is visualized during HPLC by ultraviolet light (UV) or pulsed (17) Kayser, H.; Zeitler, R.; Kannicht, C.; Grunow, D.; Nuck, R.; Reutter, W. J. Biol. Chem. 1992, 267, 16934–16938. (18) Horstkorte, R.; Mu ¨ hlenhoff, M.; Reutter, W.; Nohring, S.; ZimmermannKordmann, M.; Gerardy-Schahn, R. Exp. Cell Res. 2004, 298, 268–274. (19) Bork, K.; Horstkorte, R.; Weidemann, W. J. Pharm. Sci. 2009, 98, 3499– 3508. (20) Kean, E. L. Glycobiology 1991, 1, 441–447. (21) Kean, E. L.; Münster-Kühnel, A. K.; Gerardy-Schahn, R. Biochim. Biophys. Acta 2004, 1673, 56–65. (22) Münster-Kühnel, A. K.; Tiralongo, J.; Krapp, S.; Weinhold, B.; Ritz-Sedlacek, V.; Jacob, U.; Gerardy-Schahn, R. Glycobiology 2004, 14, 43R–51R. (23) Bork, K.; Reutter, W.; Gerardy-Schahn, R.; Horstkorte, R. FEBS Lett. 2005, 579, 5079–5083. (24) Reinke, S. O.; Lehmer, G.; Hinderlich, S.; Reutter, W. Biol. Chem. 2009, 390, 591–599. (25) Galeano, B.; Klootwijk, R.; Manoli, I.; Sun, M.; Ciccone, C.; Darvish, D.; Starost, M. F.; Zerfas, P. M.; Hoffmann, V. J.; Hoogstraten-Miller, S.; Krasnewich, D. M.; Gahl, W. A.; Huizing, M. J. Clin. Invest. 2007, 117, 1585–1594. (26) Malicdan, M. C.; Noguchi, S.; Hayashi, Y. K.; Nonaka, I.; Nishino, I. Nat. Med. 2009, 15, 690–695. (27) Mochel, F.; Yang, B.; Barritault, J.; Thompson, J. N.; Engelke, U. F.; McNeill, N. H.; Benko, W. S.; Kaneski, C. R.; Adams, D. R.; Tsokos, M.; Abu-Asab, M.; Huizing, M.; Seguin, F.; Wevers, R. A.; Ding, J.; Verheijen, F. W.; Schiffmann, R. Ann. Neurol. 2009, 65, 753–757. (28) Weiss, P.; Tietze, F.; Gahl, W. A.; Seppala, R.; Ashwell, G. J. Biol. Chem. 1989, 264, 17635–17636. (29) Fritsch, M.; Geilen, C. C.; Reutter, W. J. Chromatogr., A 1996, 727, 223– 230. (30) Rabina, J.; Maki, M.; Savilahti, E. M.; Jarvinen, N.; Penttila, L.; Renkonen, R. Glycoconjugate J. 2001, 18, 799–805.

4592

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

amperiometric detection (PAD). However, especially, the UV detection requires complex extraction steps before a sensitive analysis can be performed. Furthermore, only the quantification of CMP-Neu5Ac was described for these two HPLC systems. Nothing is known about the possibility of separating further nucleotide-activated sialic acids. Moreover, a standard of the analyzed activated sialic acid would still be needed. Of the total panel of more than 50 different sialic acids, however, only CMPNeu5Ac is commercially available. In the present study, we developed an ultrasensitive strategy for the analysis of different kinds of CMP-sialic acids. The described method is more sensitive than previously described techniques and does not require defined sialic acid standards because resulting 1,2-diamino-4,5-methylenedioxybenzene (DMB)derivatives can be easily identified by mass spectrometry (MS). EXPERIMENTAL SECTION Materials. The monosaccharide standards CMP-Neu5Ac, Neu5Ac, Neu5Gc, and KDN were purchased from Sigma-Aldrich (Taufkirchen, Germany). All reagents used were of analytical grade. Reduction of Free Nonactivated Monosaccharides. To reduce free nonactivated monosaccharides dried samples were carefully dissolved in 64 µL of ice-cold 0.2 M sodium borate buffer, pH 8.0, containing 0.2 M sodium borohydride.31 The mixtures were incubated overnight at 0 °C and dried in a SpeedVac concentrator. DMB-Labeling of Sialic Acids. For fluorescent labeling of sialic acids, samples were hydrolyzed in 100 µL of 0.2 N trifluoroacetic acid (TFA) for 4 h at 80 °C, dried, redissolved twice in methanol, and dried again. Under these conditions no cleavage of substituents such as O-acetyl groups was observed.32 Alternatively, hydrolysis can be performed in 2 M acidic acid for 90 min at 80 °C, which has been successfully employed in the case of more than 40 different sialic acid derivatives.33 To label sialic acids with DMB (Dojindo, Kumamoto, Japan), hydrolysates were dissolved in 80 µL of DMB-reaction buffer (500 mM 2-mercaptoethanol, 9 mM sodium hydrosulfite, 20 mM TFA) and incubated for 2 h at 56 °C as described previously.32,34 Reactions were stopped by adding 10 µL of 0.2 M NaOH. HPLC Separation. DMB-labeled samples were analyzed on a Superspher 100 C-18 column (250 mm × 40 mm, Merck-Hitachi, Darmstadt, Germany) at 40 °C using a Merck-Hitachi HPLC system. Mobile phases methanol/acetonitrile/water/TFA (4:4:92: 0.1) (M1) and methanol/acetonitrile/water/TFA (45:45:10:0.1) (M2) were used for separation of fluorescently labeled sialic acids.35 A linear gradient was applied from 0% to 20% M2 in 35 min at a flow rate of 0.3 mL/min. A fluorescence detector was set at 372 nm for excitation and 456 nm for emission. Preparation of Mouse Organs for Sialic Acid and CMPSialic Acid Analyses. For separation of cytosolic and nuclear fractions, cells or tissues were homogenized in 10 mM Hepes (31) Sato, C.; Inoue, S.; Matsuda, T.; Kitajima, K. Anal. Biochem. 1998, 261, 191–197. (32) Galuska, S. P.; Geyer, R.; Mu ¨ hlenhoff, M.; Geyer, H. Anal. Chem. 2007, 79, 7161–7169. (33) Zanetta, J. P.; Pons, A.; Iwersen, M.; Mariller, C.; Leroy, Y.; Timmerman, P.; Schauer, R. Glycobiology 2001, 11, 663–676. (34) Inoue, S.; Inoue, Y. Methods Enzymol. 2003, 362, 543–560. (35) Galuska, S. P.; Geyer, R.; Gerardy-Schahn, R.; Mu ¨ hlenhoff, M.; Geyer, H. J. Biol. Chem. 2008, 283, 17–28.

buffer (10 mM Hepes, pH 7.9, 10 mM KCl, 1.5 mM MgCl2, 0.2 mM EDTA, 20% glycerol, 1 mM DTT, 0.5 µg/mL aprotinin, and 0.5 mM PMSF). After centrifugation (5 min, 4 °C, 9 300g), the supernatant (cytosolic extract) was removed and the nuclei were resuspended in 20 mM Hepes buffer containing 20 mM KCl and lysed in 20 mM Hepes buffer containing 1.2 M KCl and 20 µg of DNase I (Roche, Mannheim, Germany). After centrifugation (15 min, 4 °C, 9 300g), the nuclear extracts were obtained as supernatant. Protein concentrations of the cytosolic fraction as well as nuclear extracts were determined by the BCA protein assay (Pierce, Rockford, Illinois). The purity of the extracts was controlled by Western blot detection of poly ADP-ribose polymerase and CMP-sialic acid synthetase or GAPDH which were used as nuclear or cytosolic marker proteins, respectively. To analyze the content of nonactivated and CMP-activated sialic acid molecules, purified cell compartments (equivalent to 25 µg protein) were dried. Reduction, hydrolysis, DMB-labeling, and HPLC separation were performed as described above. To control for the absence of glycoconjugates, extracts were applied to a gel-filtration column using Biogel P2 material with an exclusion limit of 1 800 Da (BioRad, Mu ¨ nchen, Germany). The sialic acid contents of the void and the elution volumes were determined in the same way as the extracts after reduction. Preparation of ManNAc- and ManNProp-Treated PC 12 Cells for Sialic Acid and CMP-Sialic Acid Analyses. The unnatural precursor of sialic acid, ManNProp, was synthesized as described in Kontou et al.36 Rat PC12 cells were cultivated in Falcon plastic flasks in RPMI 1640 medium supplemented with 10% horse serum, 2 mM L-glutamine, and 0.1 mg/mL penicillin/ streptomycin. PC12 cells (1 × 107) were incubated with 5 and 10 mM ManNAc or 5 and 10 mM ManNProp as well as with PBS. After 48 h, the cells were counted and harvested. Membrane fractions were separated after homogenization by ultrasonification of 105 cells in ice-cold Tris/HCl buffer, pH 8.0, containing 5 mM EDTA, 150 mM NaCl, 200 U/mL aprotinin, 1 mM PMSF, and 20 µg/mL leupeptin by centrifugation for 1 h at 4 °C at 30 000g. The pellets represented the membrane fraction. In order to separate free sialic acids from soluble proteins, the supernatant was subjected to a chloroformmethanol treatment. The aqueous-phase was further filtered through a centrifugal filter unit (Millipore, Billerica, MA) with an exclusion limit of 3 000 Da to separate the eventually remaining glycoconjugates from the unbound sialic acids. The dried filtrate and cell membrane fraction were used for sialic acid analysis as mentioned above. For determination of glycoconjugate-bound sialic acids, pellets were dried and applied to DMB-derivatization. Mass Spectrometric Analysis. Peaks of interest were collected during HPLC and analyzed by matrix-assisted laser desorption/ionization time-of-flight-MS (MALDI-TOF)-MS and nanoLc-electrospray-ionization (ESI)-MS(/MS). For MALDI-TOF-MS analysis, 1 µL sample was loaded onto a stainless steel target and mixed with 1 µL of 2,5-dihydroxybenzoic acid (DHB) matrix (5 mg/mL in 50% methanol containing 0.5% phosphoric acid). MALDI-TOF-MS analysis was performed on an Ultraflex time-of(36) Kontou, M.; Bauer, C.; Reutter, W.; Horstkorte, R. Glycoconjugate J. 2008, 25, 237–244.

flight mass spectrometer (Bruker-Daltonik, Bremen, Germany) equipped with a nitrogen laser and a LIFT-MS/MS facility. The instrument was operated in the positive-ion reflector mode. For nano-Lc-ESI-MS(/MS) analysis of DMB-labeled sialic acid residues, 1 µL of the collected sample was separated on a reversedphase (RP) column (PepMap, 3 µm spheres 75 µm × 100 mm, LC Packings) using an Ultimate nano-LC system, which was directly coupled with an Esquire 3000 ESI-ion trap (IT)-MS (Bruker Daltonik). Mobile phases acetonitrile/water/formic acid (8:92:0.1) (M1) and acetonitrile/water/formic acid (90:10:0.1) (M2) were used for nano-LC. A linear gradient was applied from 0% to 20% M2 in 30 min at a flow rate of 0.3 µL/min. Typical ESI source conditions were spray voltage 1.4 kV, capillary temperature of 250 °C, end plate offset of -500 V, and capillary exit of 140 V. RESULTS AND DISCUSSION Strategy for the Analysis of CMP-Sialic Acids. Twenty years ago, Okhura and co-workers established a HPLC application for the quantification of sialic acids using DMB as fluorophore, which is specific for R-keto acids.37,38 After hydrolysis and DMB-labeling under acidic conditions, the derivatized sialic acid residues were separated by RP-HPLC, making identification and quantification of sialic acids possible. This procedure results in the labeling of both nonactivated and activated sialic acids since CMP is cleaved during hydrolysis (Scheme 1A). As mentioned above, for the fluorescent tagging of sialic acids with DMB both functional groups are necessary, the keto and the carboxyl groups. A possible reduction of the keto group to a hydroxyl group would prevent a DMB-labeling of nonactivated sialic acids (Scheme 1B). Studies in the 1950’s had shown that only nonactivated sialic acids were reducible by sodium borohydride. This divergent reaction behavior can, therefore, be used to distinguish activated from nonactivated forms.39 Hence, DMB-labeling of activated sialic acids is still possible after a reduction step followed by the cleavage of the CMP-substitution under acidic conditions. A combination of both illustrated workflows, thus, allows the quantification of the total amount of sialic acids present (“classical” DMB-labeling; Scheme 1A) and the amount of nucleotide-activated sialic acids (reduction before DMB-labeling; Scheme 1B). However, sialic acid linked to a glycoconjugate will also be resistant to reduction. For this reason, glycoconjugates have to be removed prior to DMBanalysis. In order to prove the feasibility of this strategy, we tagged CMP-Neu5Ac with DMB after hydrolysis with or without preceding reduction with sodium borohydride and separated the samples by RP-HPLC. Both approaches led to a fluorescent signal at the predicted retention time for DMB-Neu5Ac (Figure 1B,C). The detected peak areas indicated that obviously no side-reactions occurred during reduction. As expected, the detection of nonactivated sialic acids was only possible without prior reduction (Figure 1D,E). Whereas KDN was discernible after “classical” DMB-labeling, no signal was detectable after DMB-labeling of reduced KDN. (37) Hara, S.; Takemori, Y.; Yamaguchi, M.; Nakamura, M.; Ohkura, Y. Anal. Biochem. 1987, 164, 138–145. (38) Hara, S.; Yamaguchi, M.; Takemori, Y.; Furuhata, K.; Ogura, H.; Nakamura, M. Anal. Biochem. 1989, 179, 162–166. (39) Warren, L.; Blacklow, R. S. J. Biol. Chem. 1962, 237, 3527–3534.

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4593

Scheme 1. Schematic Illustrations of (A) the “Classical” DMB-Labeling of Nonactivated As Well As Activated Sialic Acids and (B) the Exclusive DMB-Labeling of CMP-Sialic Acids As a Result of the Prior Reduction of Nonactivated Sialic Acids

To estimate the linearity and limit of detection of the described method, we analyzed different concentrations of CMP-Neu5Ac. The range of linearity for this sugar nucleotide is presented in Figure 1F,G. We observed a similar linear relationship between the peak area and the amount of nucleotide sugar in a range from 15 fmol to 10 pmol. The coefficient of determination approached

a value of 0.9989 demonstrating that the complete calibration range can be used for quantification. Because of this fact, a direct quantification of samples from different biological systems is feasible without a preceding rough estimation of the activated as well as nonactivated sialic acid concentrations. The limit of quantification was about 15 fmol. In comparison to other methods

Figure 1. Chromatograms of sialic acid derivates obtained (A, B, D) after the “classical” DMB-labeling or (C and E) DMB-labeling after reduction. (A) DMB-labeled KDN, Neu5Gc, and Neu5Ac were used for determination of the retention time. (B and C) CMP-Neu5Ac was detectable in an equal amount with or without reduction prior to hydrolysis and DMB-labeling. (D and E) DMB-KDN was only detectable without preceding reduction. Each sialic acid (1 ng) was fluorescently labeled by DMB after hydrolysis and separated by RP-HPLC. Additionally, calibration lines for CMP-Neu5Ac were generated in the ranges of (F) 500 fmol to 10 pmol and (G) 15 fmol to 500 fmol. The area of each peak was plotted against the applied amount of CMP-Neu5Ac, and the coefficient of determination (R2) is displayed in both graphs. 4594

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

Figure 2. Analysis of the intracellular distribution of activated and nonactivated sialic acids in cells of newborn mouse organs: (A) RP-HPLCprofiles of DMB-labeled standards (upper chromatogram) and DMB-Neu5Gc as well as DMB-Neu5Ac after reduction, hydrolysis, and DMBlabeling of cytosolic extracts of kidney cells (lower chromatogram). KDN (1 ng) was added to each sample to control the completeness of reduction of nonactivated sialic acids. The remaining diagrams illustrate the amounts of CMP-activated sialic acid and sialic acid in the nuclei and cytosol of (B) kidney, (C) liver, (D) lung, and (E) brain tissues. The diagrams represent mean values of three independent experiments. The amounts were determined by comparison with DMB-Neu5Ac standard calibration (see Figure 1F,G).

used for the quantification of CMP-Neu5Ac, our new strategy is approximately 1000-fold more sensitive than other available methods.29,30,40 Quantification of Nucleotide Activated Sialic Acids in the Cytosol and Nuclei from Cells of Different Mouse Organs. In order to evaluate whether the described strategy is applicable to biological samples, we analyzed the distribution of activated sialic acids and nonactivated sialic acids in the nuclei as well as the cytosol from cells of organs isolated from newborn mice. To this end, newborn mice were sacrificed, and organs were explanted and separated into cellular compartments. Nuclear as well as cytoplasmic fractions were further analyzed using the two workflows illustrated in Scheme 1. For the quantification of the total amount of sialic acids, samples were hydrolyzed prior to DMB-labeling. The amount of CMP-activated sialic acids was analyzed after reduction of nonactivated sialic acids, hydrolysis, and DMB-labeling (Figure 2A). The difference between the total

amount of sialic acids (nonactivated sialic acids + CMP-activated sialic acids) and CMP-linked sialic acids represents the amount of nonactivated species. As a prerequisite, the absence of glycoconjugates was controlled by gel filtration at an exclusion volume of approximately 1.8 kDa. Before and after the separation by gel filtration, identical amounts of CMP-sialic acids were observed (data not shown). Completeness of reduction was verified by use of KDN as an internal standard. A potential presence of KDN in the analyzed organ homogenates of newborn mice had been ruled out in preceding control experiments in agreement with recent studies of Rinninger and co-workers demonstrating the absence of KDN in murine brain and liver homogenates.41 As exemplified in Figure 2A for the cytosol of kidney cells, reduction was complete and no signal corresponding to KDN was detectable in any of the reduced samples. Hence, all visualized signals resulted from the activated forms, since nonactivated sialic acids were

(40) Nakajima, K.; Kitazume, S.; Angata, T.; Fujinawa, R.; Ohtsubo, K.; Miyoshi, E.; Taniguchi, N. Glycobiology DOI: 10.1093/glycob/cwq44.

(41) Rinninger, A.; Richet, C.; Pons, A.; Kohla, G.; Schauer, R.; Bauer, H. C.; Zanetta, J. P.; Vlasak, R. Glycoconjugate J. 2006, 23, 73–84.

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4595

Figure 3. Analysis of CMP-sialic acids of ManNAc- and ManNProp-treated PC 12 cells. (A) Following reduction, DMB-labeled sialic acid residues of PC 12 cells were separated by RP-HPLC. The chromatographic profiles represent solely nucleotide-activated sialic acid molecules. Disappearance of the internal standard KDN verified complete reduction. The observed peaks were collected and analyzed by (B) MALDITOF-MS and (C) ESI-IT-MS to identify the type of sialic acid. (D) The assignments were confirmed by ESI-IT-MS/MS(MS2) analysis. (E) Fragmentation pathway of DMB-labeled sialic acid as proposed by Manzi and co-workers.45

similarly reduced and sialic acid carrying glycoconjugates were absent in the nuclear and cytosolic extracts. The result of kidney analysis showed that the distribution of CMP-Neu5Ac was nearly the same in both the nucleus and the cytosol (Figure 2B and Table S-1 in the Supporting Information). In both cellular compartments, the amount of CMP-Neu5Ac was between 2 and 3 times higher than the Neu5Ac-amount. In addition to CMP-Neu5Ac, we observed the presence of CMPNeu5Gc. As for Neu5Ac, the amount of CMP-Neu5Gc was higher than that of Neu5Gc. However, the difference between the CMPNeu5Gc and Neu5Gc amounts was more pronounced. Nearly 4 to 5 times more Neu5Gc was present in the activated form. Differences in the biosynthetic pathways of Neu5Ac and Neu5Gc could be the reason for the discrepancy in activation rates of these two sialic acids. While CMP-Neu5Ac is always made from free Neu5Ac, the main route to CMP-Neu5Gc starts from CMP-Neu5Ac but also recycled Neu5Gc can serve as a substrate.3 On the other 4596

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

hand, the ratio of activation could also be a cell type and/or stagespecific process which is influenced by the current glycosylation requirements of different glycoconjugates. To investigate whether the described discrepancy in the ratio of activation between Neu5Ac and Neu5Gc is an organ specific effect or a general difference between Neu5Ac and Neu5Gc, we extended our analysis to liver, lung, and brain samples. The data shown in Figure 2C-E demonstrate significant variations between organs (in addition, Table S-1 in the Supporting Information). The highest amount of Neu5Ac and, in addition, the largest difference between CMP-Neu5Ac and Neu5Ac was found in the brain, where more than 90% of the sialic acid pool was nucleotide-activated. The smallest divergence between CMP-Neu5Ac and Neu5Ac was measured in the lung. As shown for Neu5Ac, also in the case of Neu5Gc, the contents of CMP-Neu5Gc and Neu5Gc vary between the organs demonstrating that the activation of sialic acids is a cell-specific process. Further analyses together with bioinformatics

calculations will be necessary to obtain a more complete picture of the sialic acid metabolism in vivo. Nevertheless, our data demonstrated that the newly developed strategy is a very useful tool for the determination of different CMP-sialic acid levels in biological samples. Determination of Nucleotide-Activated Unnatural Sialic Acid Derivatives. The use of artificial sialic acids for biochemical approaches to study and/or to influence cellular processes as well as to increase the half-life of therapeutic glycoproteins gets more and more in the focus of biomedicine.13,19,42-44 During the processing of artificial sialic acids several key steps have to be considered: (1) the uptake of the unnatural sugar, (2) the activation rate, (3) the transport into the Golgi apparatus, and (4) the substrate suitability for the sialyltransferases. We have employed the described strategy to analyze the synthesis of such unnatural sialic acids as well as their activation rates and transfer to membrane-bound glycoconjugates. To this end, we pretreated rat PC 12 cells with the same amounts of either the natural sialic acid precursor ManNAc or the unnatural precursor ManNProp which is transformed to the unnatural sialic acid Neu5Prop. In addition, untreated cells were included as reference in the study. The obtained chromatographic profiles of the nucleotideactivated sialic acids displayed one peak in the case of ManNAc pretreatment and two peaks if ManNProp was used (Figure 3A). As shown for ManNAc-treated cells, the obtained chromatograms of control cells displayed only the first peak. On the basis of its retention time, this peak can be assumed to be Neu5Ac. Since no standard for Neu5Prop was available, the identity of the second peak could only be surmised. The fact that ManNProp was utilized as a sialic acid precursor presupposed Neu5Prop as the second detected sialic acid. To verify this presumption, we collected both peaks during HPLC-separation. The collected samples were dried and further analyzed by MALDI-TOF-MS and ESI-IT-MS (Figure 3B,C). DMB-labeling of sialic acids leads to an increase in the molecular mass by a mass increment of 116.2 Da. As expected, we observed a signal at m/z 426.1 in the MALDI-TOF-MS spectrum of the first peak which fits the calculated mass of DMBlabeled Neu5Ac (Figure 3B). In the case of the second peak, only one signal was detectable at m/z 440.1 which could be assigned to DMB-Neu5Prop due to the mass difference of 14 Da corresponding to an additional methylene group. Both molecules could also be visualized by nano-Lc-ESI-IT-MS together with sodium adducts and products formed by loss of H2O (Figure 3C). To assign the position of the additional methylene group, further fragmentation cycles were performed by ESI-IT-MS2 (Figure 3D). The obtained MS2 spectrum of DMB-Neu5Ac displayed fragment ions at m/z 283 and 313, which resulted from elimination of formaldehyde and/or acetamide and three water molecules as described by Manzi and co-workers (Figure 3E).45 The fragment ion spectrum obtained from the supposed DMB-labeled Neu5Prop residue included the same fragment ions, thus indicating that the additional methylene group is located in the substitu(42) Prescher, J. A.; Bertozzi, C. R. Cell 2006, 126, 851–854. (43) Laughlin, S. T.; Bertozzi, C. R. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 12–17. (44) Prescher, J. A.; Dube, D. H.; Bertozzi, C. R. Nature 2004, 430, 873–877. (45) Klein, A.; Diaz, S.; Ferreira, I.; Lamblin, G.; Roussel, P.; Manzi, A. E. Glycobiology 1997, 7, 421–432.

Figure 4. Analysis of sialic acid derivates in PC 12 cells after cultivation in the presence/absence of ManNAc or ManNProp. The amounts of (A) glycoconjugate-bound, (B) nucleotide-activated, as well as (C) nonactivated sialic acid residues were calculated as described in Figure 2. Data represent means (SD of three independent experiments.

ent at position C-5 and verifying DMB-Neu5Prop as the second DMB labeled component in the case of ManNProp-treated cells. Next, the amounts of glycoconjugate-bound, nucleotideactivated as well as nonactivated sialic acid residues were determined (Figure 4). The results revealed that addition of ManNAc had almost no influence on the sialylation status of membrane-bound glycoconjugates (Figure 4A). Both concentrations of ManNAc led to the same amount of glycoconjugate-bound Neu5Ac as shown for untreated cells. Intriguingly, the determination of the intracellular levels of activated and nonactivated sialic acids demonstrated, however, that treatment with ManNAc resulted in increased amounts of CMP-Neu5Ac and particularly of Neu5Ac (Figure 4B,C). The question whether the transport into the Golgi apparatus or the sialyltransferase activity is ratelimiting could, however, not be answered by the performed analysis. Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

4597

In contrast to ManNAc pretreatment, addition of ManNProp to the cell culture clearly influenced the degree of sialylation of membrane-bound glycoconjugates. The treatment with the artificial sialic acid precursor resulted in a partial replacement of bound Neu5Ac by Neu5Prop. However, the total amount of bound Neu5Ac and Neu5Prop agreed again with the amount of Neu5Ac determined for ManNAc-treated and untreated cells. This finding is in line with previous studies in which MDCK II cells were treated with ManNAc, ManNProp, and other ManN-analogues.46 A further disparity to ManNAc-treated cells was that less Neu5Prop was synthesized even though the same concentrations of precursors were applied. Two reasons are possible: (1) ManNAc is more easily transported into the cells or (2) ManNAc is converted to Neu5Ac with higher efficacy than ManNProp to Neu5Prop. Interestingly, the ratios of Neu5Ac to Neu5Prop in both CMPsialic acids and bound sialic acids are approximately the same, suggesting that once CMP-sialic acids are formed, the two sialic acid species are equally well transported into the Golgi and accepted by the sialyltransferases. Taken together, our results demonstrated that the combination of reduction and DMB-labeling is a highly sensitive strategy allowing the identification and quantification of different nucleotide-activated sialic acids from less than 105 cells without the need for standards due to combination with MS approaches. CONCLUSIONS The analysis of the activation status of sialic acids in biological samples requires highly sensitive methods. So far, however, only low-sensitivity methods were available requiring quantities in the range of low picomolar up to nanomolar amounts depending on contaminating peaks.29,30,40 In addition, only the quantification of (46) Keppler, O. T.; Herrmann, M.; von der Lieth, C. W.; Stehling, P.; Reutter, W.; Pawlita, M. Biochem. Biophys. Res. Commun. 1998, 253, 437–442.

4598

Analytical Chemistry, Vol. 82, No. 11, June 1, 2010

CMP-Neu5Ac or all CMP-sialic acids together had been achieved by these techniques in biological samples. In the present study, we have, therefore, developed a highly sensitive strategy to distinguish between activated as well as nonactivated sialic acids and applied this protocol to several tissue and cell culture samples. Subsequent analysis by MALDI-TOF-MS and/or ESI-IT-MS(/MS) allowed the identification of fluorescently labeled sialic acids without the need of standards. Thus, the described approach enabled studies which were impossible until now since only CMPNeu5Ac is commercially available. Taken together, we have shown that the new strategy is a powerful tool allowing the separation, identification, and quantification of nucleotide-activated natural as well as artificial sialic acids in the femtomole range. ACKNOWLEDGMENT We thank Werner Mink, Siegfried Ku¨hnhardt, Daniela Wittenberg, and Ulrike Bernard for expert technical assistance and Roger Dennis, Melanie Oschlies, Wiebke Schaper, as well as Christina Bleckmann for many helpful discussions during the preparation of the manuscript. This work received financial support from the Excellent Cluster Cardiovascular System (DFG), German Research Foundation (DFG) to A.K. Mu¨nster-Ku¨hnel (Grant MU1849/1), and R. Gerardy-Schahn (DFG Research Unit 548 “PolySia”). In addition, financial support of the SonnenfeldStiftung, Berlin, is greatly acknowledged. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review March 9, 2010. Accepted April 14, 2010. AC100627E