Formulae are suggested for brevianamide-A and brevianamide-E.PENICILLIUM BREVI-COMPACTUM gives a low yield of a mixture of related neutral compounds, mostly pigments, brevianamides A-E. Brevianamide-A is a yellow pigment, C₂₁H₂₁N₃O₃, with a β-indoxyl chromophore, the presence of which is confirmed by borohydride reduction and acid dehydration to deoxybrevianamide-A. Chemical and spectral evidence was obtained for the presence of a 2,3-disubstituted indole ring in the latter. No double bonds other than those in the β-indoxyl system could be detected in brevianamide-A, and ozonolysis and methylation gave C₁₄H₁₅N₃O₃, as expected for fission of the pyrrole ring, loss of the benzene ring and indolic N.Amide bands at 1670 and 1690 cm⁻¹, lack of the amide-II band, and reduction by diborane of deoxybrevianamide-A to the basic C₂₁H₂₇N₃ suggested the presence of a diketopiperazine ring. Hydrolysis gave no simple amino-acids, but base gave an amino-acid which rapidly regenerated brevianamide-A on acidification. No evidence for the NCHCO unit of a diketopiperazine could be obtained (τ 5.5-6.1 absent); there was no exchange with MeOD and base, and no racemisation. With six rings to be expected, these positions were assumed to be involved in ring-closures.Chemical and spectral evidence of possible biogenetic precursors was sought. The presence of CMe₂ in brevianamide-A (τ 8.70, 9.10), with differences in environment, obliterated in deoxybrevianamide-A (τ 8.48), indicates proximity of the Me₂ to the CO and NH of the β-indoxyl system. Both compounds show spectral evidence for an isolated CH₂ group, which in the latter compound is adjacent to the aromatic ring. Lack of other CMe resonances and the presence of a resonance at τ 6.7 (2H) similar to that due to NCH₂ in proline dimer suggested a possible proline unit. The mass spectrum showed a major loss of C₅H₉ and suggested a terpene unit.The evidence above, biogenetic considerations, and relations to known fungal metabolites suggested a number of possible formulae including (I). Feeding experiments were conducted with [3-¹⁴C]tryptophan, [2-¹⁴C]mevalonic lactone, [2-¹⁴C]acetate, [Me-¹⁴C]methionine and L-[U-¹⁴C]proline (incorporations 0.6, 0.003, 0.025, 0.000, and 0.09% respectively). In view of the low yields (2-10 mg./l.) the positive incorporations were considered to be significant. Examinations in detail of mass spectra and ¹H n.m.r. spectra, checked by irradiation, strongly support (I) for brevianamide-A and (II) for deoxybrevianamide-A. Notably dibromination of the aromatic ring, and methylation of the CONH and exchange with MeOD, permitted assignments of many mass spectral peaks.The colourless brevianamide-E, C₂₁H₂₃N₃O₃, has an indoline chromophore, unaffected by acid or base. It contains OH (3600 cm⁻¹), NH (3370 cm⁻¹), and amide (1680, 1690 cm⁻¹) but lacks the amide-II band at 1550 cm⁻¹, indicating the likelihood of the presence of a diketopiperazine ring. In this case the NCHCO hydrogens are detectable in the ¹H n.m.r. spectrum, one coupled to the nonequivalent protons of an otherwise isolated CH₂ (ABX system JAB 13, JAX 11, JBx 8Hz.). Hydrolysis gave one molar equivalent of proline. The terpene unit is present as CMe₂CH:CH₂; τ 8.75 s (2Me) and three vinylic H which could be assigned to CH:CH₂, confirmed by hydrogenation to Et. The mass spectrum shows a major loss of 69 units, increased to 71 in the dihydro-derivative. Reduction with zinc and acetic acid gave deoxybrevianamide-E with removal of one oxygen and production of an indole chromophore. We consider brevianamide-E to be (III) and this is supported by detailed spectra.