High performance liquid chromatography

The first High performance liquid chromatography (HPLC) separation of Neu5Ac and Neu5Gc was carried out by Roland Schauer’s group in 1982 using an anion-exchange resin (Shukla et al., 1982). Since then, HPLC has not only been used to quantify Sia, but also to reliably separate and identify a wide range of different Sia residues. A wide variety of column types, elution methods, derivatisation schemes and detection modes have been employed. A smaller number of HPLC methods which do not require Sia derivatisation have also been reported. For example, Ogawa and co- workers determined the Neu5Ac and Neu5Gc content of bovine vitronectin using N- acetylglycine as an internal standard. HPLC was performed without pre-column or post- column labelling and with ultraviolet (UV) detection at 205nm (Ogawa et al., 1993). Siskos and Spyridaki (1999) used a reversed-phase ion-pair HPLC method with a C18 column, tri-isopropanolamine ion-pair reagent and UV detection, for the separation of Neu5Ac and Neu5Ac2en in serum, urine and saliva (Siskos and Spyridaki, 1999).

Although derivatisation constitutes an additional step, it is far more widely employed, as it increases the specificity and intensity of the instrument signal, while reducing interferences. Hence, it provides far more accurate Sia quantitation. The group of Nikos K. Karamanos used p-toluene-sulphonyl-chloride to generate tosyl-derivatives and benzoic anhydride/pyridine to generate per-O-benzoylated derivatives that are highly UV-absorbent (Makatsori et al., 1998). Typically, a sample clean-up step prior to derivatisation with ion-exchange chromatography is required to remove neutral monosaccharides and amino acids. In addition, after derivatisation a solid phase extraction with a Sep-Pack C18 is used to remove excess reagents and non-tosylated or

un-benzoylated derivatives. Both Neu5Ac and Neu5Gc Tosyl and per-O-benzoylated derivatives can be separated on a Supelcosil LC18 column by isocratic elution using a

mobile phase of water-acetonitrile and UV detection. The per-O-benzoylated derivative of Neu5Gc elutes later than that of Neu5Ac due to the extra benzoyl group on the N- glycolyl moiety. These HPLC methods have a limit of detection in the picomolar range and have been used to detect Sia in animal tissues, blood serum and human tissues (Lamari et al., 2002).

Fu and O’Neill (1995) developed a method for the quantitative analysis of Sia using the derivatising agent 1-phenyl-3-methyl-5-pyrazolone (PMP). In this procedure, Sia is first released by neuraminidases or by mild acid treatment with TFA. Sia are then converted to their N-acyl mannosamine derivatives with neuraminic acid aldolase before being labelled with PMP after which the mannosamine derivatives are separated by reverse-phase HPLC and detected by UV absorption. Fu and O’Neill (1995) used this method to determine the Neu5Ac and Neu5Gc content of bovine fetuin, human serotransferrrin and porcine thyroglobulin (Fu and O’Neill, 1995). Yasuno et al. (1999) also reported a similar assay, whereby Neu5Ac and Neu5Gc in bovine fetuin and 3´- sialyllactose were quantitatively converted to acid-stable mannosamine derivatives [N- acetly-D-mannosamine (ManNAc) and N-glycolyl-D-mannosamine (ManNGc) respectively] by N-acetylneuraminic acid aldolase. In contrast to Fu et al. (1995), the fluorophore p-aminobenzoic acid ethyl ester (ABEE) was used to derivatise the Sia. The ABEE-converted Sia were analysed by reverse-phase HPLC on a C18 honepak™

One of the most commonly used fluorescent tags for the determination of Sia, is the compound 1,2-diamino-4,5-methylenedioxybenzene (DMB). Hara et al. (1986) reported the first reversed-phase HPLC fluorimetric separation of DMB-derivatised Neu5Ac and Neu5Gc isolated from human serum, urine and rat serum. This method had a 40 femtomole limit of detection for both Neu5Ac and Neu5Gc (Hara et al., 1986). Since then Sia from many biological sources have been successfully identified and quantified using this method. In this derivatisation reaction using mild acidic conditions, the α-keto carboxylic acid group of the Sia reacts with DMB to form quinoxaline derivatives which are stabilised by reduction with a high concentration of sodium dithionite. It is only carbon C-1 and C-2 of the Sia that are involved in the DMB derivatisation reaction as the other Sia carbons remain intact (Lamari et al., 2003). The addition of the DMB methylenedioxy group to the Sia results in an increase in mass and contributes to the fluorescence properties of the molecule (Anumula, 2006). The quinoxaline derivatives show strong fluorescence at 448nm (emission wavelength) on irradiation at 373nm (excitation wavelength). KDN elutes before Neu5Gc which in turn elutes before Neu5Ac when these DMB-derivatised samples are separated on a C18

column using an isocratic methanol-acetonitrile-10mM acetate buffer (pH 5.0) (Ito et

al., 2002). Most known naturally occurring substituted Sia can be derivatised and

separated by DMB-HPLC. The exceptions are 2,3-dehydro Sia and glycosides which do not have a free α-keto group available for reaction (Powell et al., 2001).

A number of researchers have used an internal standard in DMB-HPLC to improve accuracy. Stanton et al. (1995) used Neu5Gc as an internal standard for the determination of Sia content in human pituitary gonado-tropins (Stanton et al., 1995). To avoid potential interference from endogenous Sia, Hikita and co-workers (2000) generated an N-propionyl analogue of the ganglioside GM3, which was used as an internal standard for the quantification of Neu5Ac and Neu5Gc in the gangliosides isolated from the kidneys of bovine, dog and cat (Hikita et al., 2000). Ito and co- workers used a similar approach; however, they chemically synthesised N- propionylneuraminic acid (NPNA) and used it as an internal standard for the quantification of Neu5Ac from human apolipoprotein E. They showed that the coefficient of variation values for the estimation of Neu5Ac with NPNA as an internal standard were significantly less than those calculated without NPNA (Ito et al., 2002).

A major challenge in the quantitative analysis of O-acetylated Sia is their low abundance and instability. However, highly sensitive DMB-HPLC combined with mild acid conditions allows the complete cleavage of Sia, with no loss of the O-acetyl group (Butor et al., 1993). DMB-HPLC has also been used to identify and to determine the degree of polymerization of polySia structures in bacteria, fish, mammalian brain and milk (Yabe et al., 2003).

The fluorophore o-phenylenediamine (OPD) has also been used in Sia determination but is not as widely used as DMB. However, OPD has a number of advantages over DMB in Sia HPLC analysis. The OPD-HPLC method has a greater linear range (2-450 picomoles) in comparison to the DMB-HPLC method (3.5-28 picomoles). In addition, OPD is an inexpensive reagent and it does not require either 2- mercaptoethanol in the reaction mixture or an internal standard for quantitation. In addition, OPD in comparison to DMB retains its fluorescent stability for a significantly longer period of time (Anumula, 1995; Anumula, 2006).