The structure of a cyclic peptide toxin from Microcystis aeruginosa was established by chemical and enzymatic methods. The stereochemistry of two chiral centers in the unusual β-amino acid moiety was also deduced by chemical degradation. The toxins produced by the water bloom-forming cyanobacterium, Microcystis aeruginosa have been known to cause deaths among livestock consuming heavily contaminated water2-5. The highly hepatotoxic toxin, cyanogenosin-LA, isolated from a South African strain has been most extensively investigated, and its cyclic heptapeptide structure 1 containing D-amino acids and an unusual β-amino acid2 was proposed on the basis of mass spectral fragmentation and other data5,6. Several other minor toxins have been isolated from various strains, and it was assumed that they are structural variants with different amino acid compositions. In this communication we report the establishment of the structure of one of the toxins, cyanogenosin-RR by chemical and enzymatic degradation and sequencing. Cyanogenosin-RR, 2, [α]D -100°(c=0.52, MeOH), λmax 238 nm (ε=33,200), νmax (KBr) 3400,1680,1550,1400,980, was isolated from the cultured cells of a Japanese strain of M. aeruginosa strain No-15-1840 (Ichimura Collection) as the major toxin (LD50 mouse ip, 0.5 mg/Kg). High resolution FAB mass spectroscopy gave a molecular weight, m/z 1038.5733, which fits the molecular formula, C49H75N13O12. Amino acid analysis of 2 after hydrolysis with 6N HCl gave a composition: Ala (1 eq.), Arg (2 eq.), Glu (1 eq.) and β-methylAsp (1 eq.). The stereochemistry of Ala, Glu, and β-methylAsp was established as D-Ala, D-Glu, and erythro-methyl-D-Asp by the GC retention correlation of the trifluoroacetate-methyl ester (TFA-ME) derivatives on a chiral column (SP-300, Supelco)7. The L-configuration of both arginine molecules was determined by electrophoresis after treatment of D- and L-arginine decarboxylases8. NMR studies of 2 (500 MHz, ¹H decoupling, COSY and ¹³C) confirmed the presence of all the aforementioned amino acids, an N-methyl moiety, and β-amino acid group reported by Santikarn et al.9. The presence of an N-methyldehydroalanine residue was confirmed by the reduction of 2 with NaBH4 and acid hydrolysis, which afforded N-methylalanine. The toxin, 2 was treated with 70 % acetic acid at 100°C for 24 hrs and with 30 % H2O2 at 37°C for 30 min10. The purification of the reaction mixture by HPLC (C18, 70 % methanol-water) gave a ninhydrin-positive open peptide, 3. The N-terminal of 3 was determined as alanine by the dansylation method11. The linear peptide, 3 successfully underwent two cycles of Edman degradation, establishing alanine as N-terminal and arginine as the second amino acid. The third cycle of Edman degradation failed, but the third amino acid was identified as β-methylaspartate by the dansylation method. This failure of the third Edman degradation strongly suggested the iso-linkage of β-methylAsp in the molecule. Treatment of 3 with carboxypeptidase Y12 at pH 5.5 and 25°C released methylamine, N-methylglutamine, and glutamic acid (TLC and electrophoresis). This confirms that glutamate is linked in iso-form to N-methyldehydroalanine. In another experiment, 2 was subjected to partial hydrolysis with 2N HCl for 6 hr. at 110°C, which afforded three peptide fragments after TLC separation. The fragments were identified as Ala-Arg, Arg-MeAsp, and MeAsp-Arg by the dansylation method. Partial hydrolysis of 2 with 2N HCl at 100°C for 2 hr. afforded a tripeptide, Ala-(Arg, MeAsp) with Ala as N-terminal. These results confirm that 2 is indeed an analogue of cyanogenosin-LA 1, in which leucine and alanine are replaced by two arginines. They also provide the independent confirmation of the structure by conventional methods. In order to determine the stereochemistry of the β-amino acid moiety, 1 was ozonized, and oxidized with H2O2. The oxidized product was hydrolyzed with 6N HCl, and the hydrolysate was converted to the TFA-ME derivatives. In a parallel experiment, 2 was hydrolyzed and derivatized under the same condition. Comparison of the two hydrolysates by GC on a chiral column showed, in addition to the intrinsic erythro-β-methyl-D-aspartate, an extra peak for the ozonized product, which corresponds to erythro-β-methyl-L-aspartate, 4. This result suggests that the chiral centers at C-2 and C-3 of the β-amino acid are R,R. The erythro configuration seems to be in agreement with the fairly large vicinal coupling between H-2 and H-3 (10.2 Hz), if the molecule is assumed to be in the staggered form as normally is. The geometry of the diene was assigned as trans-trans.5 The remaining problem with the structure is the stereochemistry of C-8 and C-9 of the β-amino acid, which is currently under investigation.