Although these fractions appeared as single peaks in their chromatograms, the 1H NMR spectra indicated that they were not pure compounds. The two fractions were subjected to an LC-MSTOF experiment to determine the molecular weight of the apparent mixture. Both fractions appeared as single peaks in the HPLC chromatogram and gave m/z values (negative ESI) of 1185.3568 (tR = 4.40 min) and 1185.3470 (tR = 4.70 min). These molecular weights indicate identical molecular formulas (M − 1) of C52H65O33
On the basis of the observed 1H NMR spectra, mass spectrum, and HPLC chromatogram it was concluded that the two separable peaks represented regioisomeric compounds. Further, the presence of four peaks in the 1H NMR spectra of the two compounds between 1.96 and 2.05 ppm suggested that the separable peaks were a set of
inseparable regioisomers that differed in the position of the acetate groups. Further attempts to purify these peaks using other forms of bonded-phase HPLC were fruitless. Attempting to solve the structure of the two compounds as mixtures of acetates was not viable because the mixture gave a spectrum that could not be interpreted. To circumvent this problem, we subjected the compounds to mild base hydrolysis. When the isomers associated with the peak at 4.40 min were treated with a methanolic solution of 0.5 M NaOH for 5 min
followed by neutralization with HOAc and separated using HPLC (NH2), a single peak (tR = 9.1 min) was observed. An accurate mass (negative ESI) gave a molar mass of 1039.3229 corresponding to a molecular formula of C43H59O29 (calculated 1039.3142). When the peak at 4.70 min was treated under the same conditions, a single peak (tR = 8.00) was obtained. Accurate mass of that peak gave an m/z of 1039.3218, which corresponds to C43H59O29. The loss of 146 amu from the native compounds upon hydrolysis was inconsistent with our initial
hypothesis that these native compounds were mixtures of regioisomeric monoacetates (i.e., anticipating a loss of 42 amu).