Crystallographic studies on the structure of vitamin B12 and related compounds

• Main Author: Pickworth, Jenny
• Published: University of Oxford 1955
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.729049

Vitamin B₁₂ has been shown, by hydrolysis experiments, to contain 5:6-dimethylbenziminazole, ribose, phosphate, propanolamine, five or six amide groups, a cyanide group, and a cobalt containing fragment of unknown chemical composition. Todd and his coworkers at Cambridge obtained a cobalt-containing fragrant, a hexacarboxylic acid with a great tendency to form gums, in a crystalline form and in this thesis an X-ray analysis of these crystals is described. Very little of the material was available, and chemical analyses were performed on only a polymorphic modification which was not suitable for X-ray studies. Three dimensional intensity data were collected on Weissenberg photographs, some of which were found to have spots to the limit of sin ⊖ using copper radiation. The intensities, estimated visually, are corrected and scaled. The position of the heavy atom cobalt was found by two dimensional Patterson synthesis. It is in a more general position in the unit cell than it is in vitamin B₁₂ and its selenocyanide. A three dimensional electron density calculation using observed structure amplitudes and phase angles determined by the positions of the cobalt atom was calculated. The resulting sections show many peaks, some of which must be spurious, and others which were considered to correspond to actual atomic peaks. There are high diffraction maxima and minima around the cobalt atom but peaks corresponding to a porphyrin-like group (I), with one bridge atom missing, were clearly distinguished. This agreed with an atomic arrangement. suggested tentatively by crystallographic studies on the vitamin. There was a high peak at about 2.3 Å from the cobalt atom and it was suggested that this was the chlorine atom. From this stage onwards work continued in collaboration with Dr. K.N. Trueblood and Dr. R.J. Prosen in California. They calculated structure factors and three dimensional electron density series on the electronic computor SWAC. The co-ordinates of the 26 atoms around the cobalt atom (leaving out one uncertain atom on the edge of a diffraction maximum) and including chlorine and a cyanide group were used in the next phase angle calculation. The general appearance of the resulting three-dimensional electron density series was very different from that of the previous ones. Carbon and nitrogen atoms put into the calculation appeared with peak heights of 7.6 - 12.7 e/A³ , while smaller round peaks up to 4.5 e/A³ appeared in positions which corresponded stereochemically to a reasonable formula. While almost the entire arrangement of atoms was found at this stage these were not all put into the next calculation. The highest peaks, over 2.5 e/A³ in height, were included in the next phase angle calculation involving 54 atoms. The method used in the structure analysis was to compute phase angles for gradually increasing groups of succeptable atomic positions and impose those on observed structure factors. The resulting terms were used to calculate the three dimensional electron density series. Several such series were calculated and only the positions for atoms in one side chain have not been found satisfactorily. There has recently been chemical evidence which establishes the nature of the last side chain and supports the formula (II) tentatively suggested for the It was found that, in a non-centrosymmetric structure of this kind, an atom appeared in the position in which it was put in the phase angle calculation, regardless of whether that position was correct or not. If the position was totally incorrect the peak height was often rather lower, but certainly more than half that expected. If the position was only slightly incorrect a bulge was shown towards the correct position. The heights of peaks were determined to a certain extent by the rigidity of the structure in which they take part. There are very few hydrogen bonds in ths crystal structure. In the calculation involving 72 atoms there were no peaks over 1.5 e/A³ which did not correspond to atomic positions. The general arrangement of atoms has been found but the chemical nature of each atom is not always clear. The molecule is porphyrin-like but one bridge atom is missing. The central ring system round the cobalt is planar except at the points where the bridge atom is missing. The four five-membered rings have their β and β' positions fully saturated. Acetic acid (-CH₂-COOH) and propionic acid (-CH₂-CH₂-COOH) groups were attached in the sequence found for most naturally occurring porphyrins. Peaks, probably due to methyl groups, are found attached to two of the bridge atoms, at the direct link, and in a regular sequencer on the positions of the five-membered rings. There is a γ-lactone or γ-lactam ring attached to one of the five-membered rings. The atomic arrangement suggested has been confirmed by other workers for vitamin B₁₂, except that, during the hydrolysis, a hydroxyl group has been introduced into the molecule to glve a lactone (or loctam) ring, and an inversion has occurred so that the nucleotide group in the vitamin is replaced by a cyanide group, and the cyanide group by chloride.