We were able to detect antimicrobial activity in the skin mucus of Atlantic salmon, unlike Richards et al. (33
), who identified antimicrobial activity in extracts from the liver, intestine, and stomach but not in extracts from the skin. This may have been because their extraction method was not harsh enough. Our experience shows that acidic extraction, preferably with acetic acid, is necessary to release antimicrobial peptides from the mucus of fish. Richards et al. (33
) assigned the antimicrobial activity to the entire histone H1 protein; but the presence of specific fragments, like the one reported here, cannot be ruled out.
The purification of SAMP H1 was problematic due to large batch-to-batch differences in the chromatographic properties of the antimicrobial activity, especially in the step in which the Mono S column was used. This may have been due to unfavorable interactions with other compounds from the mucus, and the presence of urea eliminated the problem. The presence of urea in the subsequent Mono S runs was not necessary, however, probably because the interfering compounds were removed.
The sequence of the 30-mer peptide was quite unique in that it contained high amounts of alanine (43%), lysine (23%), and proline (20%); and a sequence homology search identified the peptide as a fragment of histone H1. Surprisingly, unlike other synthetic AMPs, the synthetic version of the peptide initially had no antimicrobial activity. A small fraction of it, however, became activated upon binding to the Mono S cation-exchange column. The active form apparently had an altered three-dimensional structure, since it eluted ahead of the inactive form upon chromatography on the Mono S column; it was (in contrast to the inactive form) not retained on the reversed-phase column, and it eluted with an apparent molecular weight of about 100 upon gel filtration, in contrast to the inactive form, which eluted at the expected molecular weight of about 2,800. Moreover, the CD spectra of the active form were entirely different from those of the inactive form, whose CD spectra were those of a nonstructured peptide. The activity cannot be due to a degraded product, since the active form had the expected molecular weight, as determined by MALDI-TOF MS. These observations demonstrate that the activated synthetic peptide has the same apparent physiochemical properties as the biological isolated SAMP H1. A peculiar finding was that the same amount of peptide was activated, irrespective of the amount applied to the column or the incubation time (Fig. ). Thus, the activation probably occurs at the solvent-matrix interface and is dependent only on the column cross section.
Comparison of the active and inactive forms of SAMP H1 by CD and chromatographic analyses revealed that the active form, both the natural and the synthetic forms, has a structure different from that of the inactive synthetic form. The active form is more structured and is probably not oligomerized but is more condensed, since it is retarded upon gel filtration and it passed through a filtration membrane with a nominal molecular weight cutoff of 3,000. The active form apparently exposes its hydrophilic residues and buries the hydrophobic ones to a greater extent than the inactive form, since the active form did not bind to the RPC column. The slightly earlier elution of the active peptide upon cation-exchange chromatography might be due to a more rigid structure that limits the number of positively charged groups that may interact with the column material at the same time. The inactive form, being nonstructured and presumably more flexible, may permit most of its positive groups to interact with the column.
It is known that the peptide bond on the N-terminal side of proline residues can exist in both the trans
and the cis
conformations, in contrast to the peptide bond adjacent to other residues, which is in the trans
). This prompted us to consider the possibility that the synthetic peptide did not contain the proper proline peptide bond conformation necessary for the peptide to be active. When the synthetic peptide was free in buffer, treatment with PPIase did not produce the active form of SAMP H1. The CD spectra suggested an unstructured random coil, in which the all-trans
form might be most stable. It has previously been shown that PPIase may be more effective in forming the correct structure of a protein when it is bound to a chromatographic support than when it is in solution and that even higher yields of correctly folded protein were achieved when the PPIase was supplemented with a minichaperone and a protein disulfide isomerase (1
). For SAMP H1 to obtain the necessary isomerization upon PPIase treatment, the peptide must probably be structurally stabilized, such that the cis
form becomes energetically favorable, and such stabilization apparently occurs upon binding of the peptide to the cation-exchange column.
An inducible mechanism for the proteolytic production of AMPs from histones has been found for the histone H2A-derived peptides buforin I and parasin I. Buforin I is produced from cytoplasmic histone H2A by the action of pepsin in the gastric gland cells of the Asian toad (17
). Parasin I is produced in a similar way in the skin mucus of catfish by the action of cathepsin D (4
), which itself is activated by matrix metalloproteinase 2 (5
). A similar mechanism may exist for the formation of SAMP H1 by a specific proteolytic degradation of the histone H1 during proliferation/degradation of the epidermal tissue of the fish. The antimicrobial active structure of the peptide is already formed by the folding during the biosynthesis of the histone and, consequently, is present in the antimicrobial entity. On the other hand, synthesized SAMP H1 is not active, since the conditions for chemical synthesis of the peptide do not support such a specific isomeric formation of the correct proline isomeric forms that are needed for an active peptide.
It is not possible at present to deduce which of the prolyl peptide bonds are responsible for the high degrees of stability of the two forms of SAMP H1. Even though extensive studies have been performed with model peptides (32
), the results cannot be directly transformed to natural peptides and proteins. Of the six proline residues in SAMP H1, four are preceded by alanine residues, and the remaining two are preceded by lysine and glycine, respectively. Neither of these peptide pairs showed a propensity to yield cis
amide bonds in the aforementioned study. It is likely that several possible isomeric structures of SAMP H1 possess antimicrobial activity.
Patrzykat et al. (29
) reported on the presence of a shorter form (26-
mer) of SAMP H1 in coho salmon in which the four C-terminal amino acid residues were missing but the synthetic form of the peptide was inactive. Judging from our results, this was probably because the correct proline peptide bond conformation was not present in their synthetic peptide. Their inactive peptide was, however, reported to potentiate the antimicrobial activity of pleurocidin and lysozyme, whereas our inactive SAMP H1 showed no synergy with pleurocidin (data not shown).
This work shows for the first time the importance of proline isomers in the activity of a peptide and that it can be enzymatically converted from an inactive form to an active form and also confirms that the structures of such small antimicrobial peptides can be essential for the antimicrobial activity.