Molecular Characterization of Two Novel Antibacterial Peptides Inducible upon Bacterial Challenge in an Annelid, the Leech Theromyzon tessulatum *

Two novel antimicrobial peptides named theromacin and theromyzin were isolated and characterized from the coelomic liquid of the leech Theromyzon tessulatum . Theromacin is a 75-amino acid cationic peptide containing 10 cysteine residues arranged in a disulfide array showing no similarities with other known antimicrobial peptides. Theromyzin is an 86-amino acid linear peptide and constitutes the first anionic antimicrobial peptide observed in invertebrates. Both peptides exhibit activity directed against Gram-positive bacteria. Theromacin and theromyzin cDNAs code precursor molecules containing a putative signal sequence directly followed by the mature peptide. The enhancement of theromacin and theromyzin mRNA levels has been observed after blood meal ingestion and upon bacterial challenge. In situ hybridization revealed that both genes are expressed in large fat cells in contact with coelomic cavities. Gene products were immunodetected in large fat cells, in intestinal epithelia, and

Two novel antimicrobial peptides named theromacin and theromyzin were isolated and characterized from the coelomic liquid of the leech Theromyzon tessulatum. Theromacin is a 75-amino acid cationic peptide containing 10 cysteine residues arranged in a disulfide array showing no similarities with other known antimicrobial peptides. Theromyzin is an 86-amino acid linear peptide and constitutes the first anionic antimicrobial peptide observed in invertebrates. Both peptides exhibit activity directed against Gram-positive bacteria. Theromacin and theromyzin cDNAs code precursor molecules containing a putative signal sequence directly followed by the mature peptide. The enhancement of theromacin and theromyzin mRNA levels has been observed after blood meal ingestion and upon bacterial challenge. In situ hybridization revealed that both genes are expressed in large fat cells in contact with coelomic cavities. Gene products were immunodetected in large fat cells, in intestinal epithelia, and at the epidermis level. In addition, a rapid release of the peptides into the coelomic liquid was observed after bacterial challenge. The presence of antimicrobial peptide genes in leeches and their expression in a specific tissue functionally resembling the insect fat body provide evidence for the first time of an antibacterial response in a lophotrochozoan comparable to that of holometabola insects.
All of the multicellular organisms including plants, invertebrates, and vertebrates have developed an immediate response to defend themselves against infectious microorganisms (1,2). Recent studies on the components of the innate immune system have demonstrated the contribution of antimicrobial peptides to the host defense. Antibiotic peptides are small molecules.
Based on their structural features, three classes were defined: 1) linear ␣-helical peptides without cysteines of which the prototype of this family is the cecropin; 2) open-ended cyclic cysteine-rich peptides among which defensins are the most widespread; and 3) linear peptides containing a high proportion of one or two amino acids like indolicidin, for example (3). Despite great diversity in the primary structure, the majority of antimicrobial peptides documented are characterized by a preponderance of cationic and hydrophobic amino acids. This amphipathic structure allows them to interact with bacterial membrane. Brogden et al. (4) have determined in ovine that a novel class of peptides with anionic properties also has antimicrobial properties (4). In marked contrast to cationic antimicrobial peptides, no anionic antibacterial molecules have been described hitherto in invertebrates.
Antimicrobial peptides appear to be essential anti-infectious factors that have been conserved during evolution. Meanwhile, their implications in immune processes are different according to species, cells, and tissues (5). The involvement of antimicrobial peptides in natural resistance to infection is sustained by their strategic location in phagocytes, in body fluids, and at the epithelial level, i.e. at interfaces between the organisms and its environment. This action is strengthened by the rapid induction of such antibacterial peptide genes in bacteria-challenged plants or animals. Drosophila melanogaster, which is subjected to septic injury, massively induces and synthesizes a battery of potent antibiotic molecules from fat body, a functional equivalent of the mammalian liver (6). The active peptides then are released into the hemolymph where they exert their antifungal and/or antibacterial activities. Adding to their systemic response, antibiotic molecules of the fruitfly are produced by cells of most of the barrier epithelia such as tracheal epithelium, gut lining, or salivary glands where they provide a local first defense against microorganisms (7,8).
In worms, even if antimicrobial activities have been detected in the body fluids of several species, antimicrobial peptides have been characterized fully in nematodes and annelids only. A 71-amino acid peptide named ASABF 1 (for Ascaris suum antibacterial factor) possessing structural and functional similarities to the insect/arthropod defensins has been characterized in the parasitic nematode as A. suum (9). Recently, Pillai et al. (10) show an enhancement of the ASABF transcript in the body wall upon bacterial challenge. ASABF presents a significant sequence identity with peptides deduced from a cDNA sequence (yk150c7) and from a putative gene (T22H6.5) of the free-living nematode Caenorhabditis elegans. Although no variation of the C. elegans ASABF gene expression was detected using high density cDNA microarray, Mallo et al. (11) demonstrate that Gram-negative bacteria infection of this worm provokes a marked up-regulation of genes encoding lysozyme and lectins that are known to be involved in the immune responses of other species. These authors (11) also establish that certain infection-inducible genes are under the control of the DBLtransforming growth factor-␤ pathway. The existence of an inducible immune system in both nematodes and insects is consistent with the new molecule-based phylogeny, which considers that protostomes can be divided into two large groups: 1) ecdysozoans comprising nematodes, arthropods, and other phyla that have emerged and 2) lophotrochozoans regrouping annelids, flat worms, and mollusks (12). In lophotrochozoans, no inducible antimicrobial peptides have been described so far in the animals studied. The gene encoding lumbricin I, a proline-rich antibacterial peptide from the earthworm Lumbricus rubellus, is expressed constitutively (13). The constitutive expression also has been reported in the mussel Mytilus galloprovincialis in which antibiotic peptides stored in hemocyte granules are released after bacterial challenge (14).
In this study, we report the primary and secondary structures of two new antibacterial peptides, theromacin and theromyzin, isolated from the body fluid of an annelid, the leech Theromyzon tessulatum. Interestingly, both molecules do not present significant similarities with other known molecules. Theromacin is a novel cysteine-rich antimicrobial peptide, and theromyzin constitutes a new anionic antibiotic peptide. In an experimental model of infection, we obtained evidence for an enhancement of transcription levels of both genes in specific tissues that could be assimilated to the fat body of D. melanogaster. We have found that peptides are released massively into hemolymph where they might exert their antimicrobial activity by a systemic action. These data, reminiscent of the Drosophila antimicrobial defense, are the first reports of antimicrobial peptide induction by a specific tissue in a lophotrochozoan.

Animals
T. tessulatum rhynchobdellid leech is an ectoparasite of aquatic birds. Its life cycle is subdivided in stages defined by taking as indicators three blood meals (stage 0, from hatching until the first blood meal; stages 1, 2, and 3, respectively, after the first, second, and third blood meals). The stage 3 corresponds to the gametogenesis phase.

Theromacin and Theromyzin Purification Procedure
Stage 3 animals were pricked individually with saline solution or with 10 l of a solution containing 10 9 killed Escherichia coli (D31). Coelomic fluid and mucus covering the animals were collected with a tuberculin syringe at t ϭ 0, 3, and 24 h after challenge. Liquids were centrifuged immediately at 800 ϫ g for 15 min at 4°C, and supernatants were diluted (1:1 v/v) in pure water (Beckman) containing 0.1% of trifluoroacetic acid. The pH value was brought to 3.9 with 1 M HCl. Centrifugation (10,000 ϫ g, 20 min, at 4°C) then was used to clarify the supernatants, which were loaded onto Sep-Pak C18 Vac cartridges (Waters) according to Bulet et al. (15). Elution steps were performed with 5 and 40% acetonitrile (ACN) in acidified water. The prepurified fractions then were concentrated, reconstituted in pure water, and tested for antimicrobial activity as described below. Only the 40% ACN eluted fractions were active and submitted to purification by reversed-phase high pressure liquid chromatography (RP-HPLC). All of the following HPLC steps were carried out on a Beckman TM Gold HPLC system.
Step 1-Aliquots of the 40% Sep-Pak fractions were subjected to RP-HPLC on a Sephasyl C18 column (250 ϫ 4.1 mm, model 218TP54, Vydac). Elution was performed with a linear gradient of 2-52% ACN in acidified water over 90 min at a flow rate of 1 ml/min. The fractions corresponding to absorbance peaks were collected in polypropylene tubes, dried, reconstituted in water, and tested for antimicrobial activity.
Step 2-The active fractions were loaded further onto a C18 column (250 ϫ 2.1 mm, model 218TP52, Vydac) with a gradient consisting of 2-25% ACN in acidified water for 10 min and 25-35% ACN for 40 min at a flow rate of 0.2 ml/min. The fractions were collected and treated as above.
Step 3-One additional step was performed on a Narrowbore C18 reversed-phase column (150 ϫ 2 mm, Waters) at a flow rate of 0.2 ml/min using the ACN gradient described in Step 2.
The purity assessment and the molecular weight determination of theromacin and theromyzin were carried out by mass spectrometry in an electrospray-positive ion mode using Quattro II triple quadrupole (Micromass) and by the matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) instrument (DE STR PRO, Applied Biosystems). Amino acids analysis of the purified peptides was performed by automated Edman degradation on a pulse liquid automatic peptide sequenator (Procise, Applied Biosystems).

Secondary Structure Determination of Theromacin
MALDI-TOF MS analyses were used to determine the mass of the reduced-alkylated theromacin and the mass of the fragments derived from protease digestions.
Reduction and Alkylation-50 pmol of purified peptide were dissolved in 100 l of 0.2 M amino bicarbonate buffer, pH 8, containing 0.05 M dithiothreitol and 6 M guanidine hydrochloride. The sample was flushed with argon and incubated at 70°C for 30 min. 50 l of 6 M acrylamide was added, and the sample was flushed with argon again and incubated at 4°C for 1 h.
C-Lysyl Endopeptidase Digestion-10 pmol of purified theromacin were dissolved in 10 l of 25 mM Tris-HCl buffer, pH 8.5, containing 1 mM EDTA. 0.05 g of lysyl endopeptidase (Roche Applied Science) was added. The digestion was carried out overnight at room temperature.
C-Glutamyl Endopeptidase Digestion-10 pmol of purified peptide were dissolved in 10 l of 25 mM ammonium carbonate buffer, pH 8. 0.05 g of glutamyl endopeptidase (Roche Applied Science) was added. This enzyme cleaves peptide bonds C-terminally at glutamic acid and with a lower rate at the aspartic residue.
MALDI-TOF MS analyses were used to measure the mass shifts of theromacin, resulting from the reduction and alkylation steps as well as the mass of the peptide fragments derived from protease digestions. The samples were dissolved in 0.5% formic acid containing methanol/water (1:1) at a concentration of 1 pmol/l. 1 l of the sample solution was mixed directly onto the target with 1 l of the 2,5-dihydroxybenzoic acid matrix solution (12 mg/ml prepared in methanol/water, 7:3).

Bioassays
Microorganisms-The bacteria E. coli D31, Micrococcus luteus IFO12708, and the filamentous fungus Fusarium oxysporum were used for the antimicrobial assays.
Antimicrobial Tests-After each purification step, antibacterial activity was monitored by a liquid growth inhibition assay as described by Bulet et al. (15). The minimal inhibitory concentration (MIC) and the minimal bactericidal concentration were determined according to the method of R. E. Hancock (www.interchg.ubc.ca/bobh/methods.html). Antifungal activity was monitored as described by Fehlbaum et al. (17) by a liquid growth inhibition assay. The bactericidal test was performed as described previously (18).
cDNA Cloning cDNA for theromacin and cDNA for theromyzin were cloned using two-step PCR amplification.
Step 2: Rapid Amplification of 5Ј-cDNA End-Reverse transcription was performed using antisense oligonucleotides 5Ј-TTGCAGTTCTGTAG-CGTTTATGC-3Ј and 5Ј-ATCAACTGGAGTCTTATCGGCGTA-3Ј deduced, respectively, from the theromacin and theromyzin cDNA sequences obtained previously. After first strand cDNA synthesis and the addition of a poly(dA) tail at its 3Ј end using a terminal transferase (Invitrogen, the protocol of the manufacturer), PCR was performed with an oligo(dT) anchor primer and the internal antisense primers deduced from the cDNA obtained in Step 1 (5Ј-CATTCTCCCCTATCAGCTTTG-3Ј for theromacin and 5Ј-CAGCGTTAATTCTTCGTGTTC-3Ј for theromyzin). PCR parameters were identical to those described in step 1. All of the PCR products were subcloned into pGEM-T Easy vector (Promega), and several different cDNA clones were sequenced.

Gene Expression Site
In Situ Hybridization (ISH)-Leeches were fixed overnight in a solution containing 13% formalin, 39% ethanol, and 0.65% ammonium hydroxide, pH 6.4. After dehydration, animals were embedded in paraplast and 7-m sections were cut, mounted on poly-L-lysine-coated slides, and stored at 4°C until use.
The plasmids containing the coding region of theromacin and theromyzin probes were used as templates for the synthesis of the probes. Digoxigenin-UTP-labeled and [ 35 S]UTP-labeled antisense and sense riboprobes were generated from linearized cDNA plasmids by in vitro transcription using an RNA-labeling kit (Roche Applied Science). Digoxigenin-labeled riboprobes (40 -100 ng/slide) and 35 S-labeled riboprobes (100 ng or 1 ϫ 10 6 cpm/slide) were hybridized as described previously (19). The slides were observed under a Zeiss TM Axioskop microscope.
The control consisted in replacing antisense riboprobe with the sense riboprobe. RNase control sections were obtained by adding a preincubation step with 10 g/ml RNase A prior to hybridization.

Gene Expression Analysis
Northern Blot-Leeches were homogenized in TRIzol using a Polytron. Total RNA was extracted according to the manufacturer's protocol (Invitrogen). 10 g of total RNAs and size markers were separated on a 1% agarose gel electrophoresis containing 0.02 M sodium phosphate buffer, pH 7.2, and 8.5% formaldehyde and transferred to Hybond N membrane (Amersham Biosciences). Theromacin and theromyzin probes were synthesized from 332-and 405-pb fragments, respectively, amplified in the first PCR (see above). An 18 S subunit probe was also performed as control. A sense oligonucleotide, 5Ј-GCGGGAACGAGCGCGCGTTTAT-TAGAT-3Ј, and an antisense oligonucleotide, 5Ј-TGGCACCAGACTTGC-CCTCCAATT-3Ј, were designed from the ribosomal RNA 18 S subunit of T. tessulatum (GenBank TM accession number AFO99942) and used in PCR experiments before being subcloned into pGEM-T Easy vector. The plasmids containing theromacin cDNA or 18 S cDNA were digested with EcoRI, and inserts were purified from 1% agarose gel using the Wizard PCR-prep DNA purification system (Promega). 25 ng of each fragments then were 32 P-labeled by random priming using the Ready-to-Go DNAlabeling kit (Amersham Biosciences). Hybridization and detection were performed as described previously.
In Situ Hybridization-Quantification of the radiolabeling at the cellular level was performed using an Axiophot Zeiss microscope and a Biocom quantification system as established previously (20).

Immunohistochemical Procedures
Antibodies-The chemically synthesized regions of theromacin (Gln 49 -Arg 63 ) and theromyzin (Lys 26 -Val 41 ) were used for the immunization procedure of rabbits and mice, respectively, according to the protocol of Agrobio.
For light microscopy on 1-m thick semi-thin sections, a post-embedding immunogold procedure was used. Tissues were fixed for 2 h at 4°C in a mixture containing 4% paraformaldehyde, 0.1% glutaraldehyde, and 0.2% picric acid in 0.1 M phosphate buffer, pH 7.4. Tissues were post-fixed in 1% OsO 4 for 3-5 min and dehydrated in graded alcohol before embedding in LR-White (TAAB Laboratories). Theromacin and theromyzin immunostainings were performed using a gold-tagged secondary antibody and silver amplification as described above. Controls were incubations of immune serum preadsorbed by synthetic peptides.

Variation of Plasmatic Amount
Aliquots of 40% ACN Sep-Pak and ␤-lactoglobulin (Applied Biosystems) at different concentrations (250, 375, and 500 pmol) were subjected to RP-HPLC on a C18 column (250 ϫ 4.1 mm, model 218TP54, Vydac). Elution was performed as described above. The area of peak corresponding to theromacin or theromyzin was calculated, and the amount of peptide was evaluated by comparing the values with the ␤-lactoglobulin standard. The amount obtained divided by the volume of hemolymph collected initially allowed us to estimate the concentration of theromacin in the coelomic fluid of the leech.

RESULTS
Biochemical and Molecular Characterization of Theromacin and Theromyzin-Theromacin and theromyzin were isolated from leech body fluid in acidic conditions (Fig. 1). A 30-residue N-terminal sequence of the purified molecules then was obtained by automatic Edman degradation. In parallel, electro- After prepurification by solid phase extraction, the 40% ACN-eluted material was loaded onto a C18 column (250 ϫ 4 mm). Each individually collected peak was tested for its antimicrobial activity by liquid growth inhibition assays. Two fractions containing theromacin and theromyzin that were eluted at 33% ACN and at 38% ACN, respectively, were found to be active against M. luteus. These fractions were purified further by two additional RP-HPLC purification steps. The exact mass of purified peptides was determined by electrospray ionization MS and by MALDI-TOF MS, respectively. A, for theromacin mass calculation, quadrupole was scanned over 300 -1700 Da. Two unequivocal pseudomolecular ions ([Mϩ6H] 6ϩ /6 ϭ 1,419.66 and [Mϩ7H] 7ϩ /7 ϭ 1,216.85) were observed that correspond to the protonated species whose molecular mass was 8,517.98 Da. B, an analysis of theromyzin by MALDI-TOF MS shows a m/z value of 9,973.1 MHϩ. spray and MALDI-TOF mass spectrometry analysis revealed a deduced molecular mass of 8,517.98 Da for theromacin and 9,973.1 Da for theromyzin, illustrating that the sequences obtained by Edman degradation were incomplete (Fig. 1).
To fully determine the amino acid sequences of the peptides and to obtain information on the precursors, cDNA cloning was carried out using a two-step PCR-based approach (see "Experimental Procedures" and Fig. 2). Amino acid sequences deduced from the cDNAs allowed us to conclude that theromacin is a cationic 75-amino acid peptide (calculated pI of 8.6) and that theromyzin is an anionic 86-amino acid peptide (calculated pI of 6.03). Data bank analysis (BLAST program in Swiss-Prot) revealed no obvious sequence similarities of these peptides with other known peptides or proteins. Moreover, it appeared that the mass value corresponding to the theromacin sequence deduced from the cDNA was 10 Da in excess of the biochemically calculated mass of the native peptide (8,527.81 versus 8,517.98 Ϯ 0.72). The difference in mass was attributable to the arrangement of the 10 cysteine residues contained in the sequence in five disulfide intramolecular bridges.
In contrast with theromacin, the calculated mass of theromyzin was in perfect agreement with the measured mass, demonstrating that theromyzin is a peptide devoid of posttranslational modifications.
Precursor sequences also were deduced from the cDNA sequences (Fig. 2). The Signal PVI software analysis revealed that 1) theromacin and theromyzin are flanked at the N terminus by a putative hydrophobic signal peptide, Met 1 -Ala 22 , and Met 1 -Ala 21 , respectively, and that 2) the cleavage site for signal peptidase is most probably located after the Ala 22 preceding the Gly 23 for theromacin and after the Ala 21 preceding the Asp 22 for theromyzin. This leads to the notion that mature peptides may be generated through conventional processing mechanisms and can be secreted to the extracellular medium.
Determination of the Disulfide Array of Theromacin-To decipher the pattern of the cysteine residues engaged in disulfide bridges, the molecular mass of theromacin was subjected first to alkylation with acrylamide and then measured by MALDI-TOF mass spectrometry. The mass spectrum revealed that reducing the protein prior to the alkylation step lead to a mass shift of 710.37 Da, consistent with the incorporation of 10 acrylamide molecules, whereas alkylation alone did not lead to a mass shift (data not shown). This result demonstrates that the 10 cysteine residues of theromacin are engaged in five disulfide bridges. From the analysis of the primary structure of the antibacterial peptide showing that theromacin did not harbor a cysteine motif described in other known antimicrobial peptides, the determination of the disulfide array was visualized by protease digestion of the native peptide ( Fig 3A).
Biological Activity of Theromacin and Theromyzin-In liquid growth inhibition assay, the purified theromacin was active against M. luteus (MIC 16.5-33 nM; minimal bactericidal concentration 0.26 -0.52 M). No activity was found toward E. coli or F. oxysporum at the same concentration. When theromacin was incubated with M. luteus at 165 nM, a concentration 10 times higher than the MIC value, all of the bacteria were killed in Ͻ5 h (Fig. 4). Consequently, theromacin exerts a bactericidal activity. The loss of antibacterial property was noticed when theromacin was reduced, demonstrating that disulfide bridges play an important role in the biological activity of the molecule. Theromyzin was shown to be active against M. luteus (MIC 250 -500 nM). The growth of bacteria on the plate was observed when theromyzin was tested for the minimal bactericidal concentration determination showing that the molecule possesses bacteriostatic activity against M. luteus.

Expression and Induction of Theromacin and Theromyzin
Transcripts-The gene expression pattern of theromacin and theromyzin during immune response was investigated by Northern blot. An analysis of the RNA level in stage 3 leeches was assessed first at 0, 6, 12, 24, 48, and 72 h after saline or killed bacteria (E. coli or M. luteus) injections. As illustrated in Fig. 5A, the theromacin and theromyzin transcripts level is enhanced rapidly (6 h) by Gram-positive bacteria, Gram-negative bacteria, or saline challenges. The expression levels were still high 3 days after infection and are stronger with a bacterial injection than with a saline injection. In contrast, no differences were observed between Gram-positive-and Gramnegative-injected leeches. Taken together, these Northern blot data indicate that treatment with bacteria induces a rapid, strong, and persistent increase of theromacin and theromyzin mRNA levels and that this enhancement does not depend on the type of septic injury. This variation of theromacin and theromyzin transcripts upon bacterial challenge also has been proven by a DNA microarray approach (21).
To corroborate these results and to better understand the physiological role of theromacin and theromyzin in stage 3 leeches, we investigated the localization of the gene expression site using ISH (Figs. 5B and 6A). Fig. 5 gives results obtained with the digoxygenin dUTP-labeled theromacin antisense probe (Fig. 5A) on unchallenged stage 3 leech sections. Both genes are expressed strongly and exclusively in large cells evenly distributed in whole animals. Because they contained numerous lipidic granules, we named them large fat cells (LFC). Interestingly, LFC are often in close contact with cavities containing coelomic fluid from which theromacin and theromyzin peptides were isolated. As all of these experiments were carried out with sections of non-challenged stage 3 leeches, we concluded that the RNA base-line level of stage 3 animals observed by Northern blot could be attributed to the transcripts localized in the LFC.
We then further attempted to analyze whether the inducibility evidenced by Northern blot experiments was relevant in LFC or in other tissues. The induction was monitored by injecting stage 3 leeches with bacteria or saline solution. The sections were hybridized with radiolabeled antisense riboprobes. The signal appeared under light microscopy as an accumulation of spots (Figs. 5C and 6A). The counting of the spots was performed to quantify the transcript level in the LFC (see "Experimental Procedures"). An enhancement of the theromacin (Fig. 5C) transcript levels was observed only in the LFC of challenged stage 3 leeches. Markedly, the level of induction in LFC was stronger with bacteria than with saline injection for both genes (41.87 versus 27.27 arbitrary units).
These gene expression studies were extended to stage 2 animals pricked or not (control) with saline buffer or with a mixture of killed Gram-positive and Gram-negative bacteria. ISH on sections of leeches killed 24 h post-injury was performed using radiolabeled theromacin and theromyzin riboprobes. The results are presented in Fig. 6A, demonstrating the detection of theromacin and theromyzin RNA in the LFC of challenged animals. The expression level was higher after bacteria injection than after saline injection (73.65 versus 27.27 arbitrary units for theromacin and 52.2 versus 18.3 arbitrary units for theromyzin). No signal was noticed in other cells or tissues.
By contrast to stage 3 animals (Fig. 5A), no RNA base-line level (theromacin) or very few (theromyzin) were observed in stage 2 leeches (Fig. 6A). The highest sensitivity of the ISH compared with Northern blot could explain why theromyzin transcripts were detected in the control by this technique only.
Thus, we hypothesized that several physiologic events that occurred during the transition stage 2/stage 3 may be inducers of theromacin and theromyzin gene expression. In support of this hypothesis, stage 2 leeches were collected 24, 48, and 72 h after the blood meal, which corresponds to the tripping factor of the transition. Total RNAs of stage 3 leeches were extracted and probed with the theromacin and theromyzin cDNA successively. The results presented in Fig. 6B underlined a low but progressive increase of both theromacin and theromyzin transcripts after the blood meal, confirming our hypothesis.
These results define the cells implicated in theromacin and theromyzin mRNA synthesis. We concluded that in all of the cases of induction, i.e. bacteria, saline injections, or transition stage 2/stage 3, the enhancement of theromacin and theromyzin transcripts was observed only in LFC. This finding suggests that the leech LFC may constitute an organ implicated in the synthesis of antimicrobial molecules comparable to the insect fat body.
Immune Localization of Theromyzin and Theromacin Peptides and Dynamic of Their Release in the Body Fluid-To specify the participation of theromacin and theromyzin in antiinfectious processes, immunocytochemical and plasmatic titration procedures were carried out in leeches subjected to bacterial challenge. Theromacin-and theromyzin-storing sites were investigated by immunocytochemistry on the sections of unchallenged leeches (Fig. 7). Theromacin immunoreactivity was found in LFC (Fig. 7A), in the intestinal epithelium (Fig. 7B), FIG. 5. Analysis of theromacin and theromyzin gene expression in stage 3 leeches. A, Northern blot analysis of RNA from stage 3 leeches extracted at different times (0, 6, 12, 24, 48, and 72h) after challenge with saline buffer, killed E. coli, and killed M. luteus. RNAs were probed successively with theromacin, theromyzin, and the 18 S ribosomal subunit cDNA. B, ISH with digoxygenin dUTP-labeled theromacin antisense riboprobe. The signal was detected in LFC, which are in contact with coelomic cavities (CC). No hybridization was noticed with the sense riboprobe, confirming the specificity of the antisense probe. C, ISH with radiolabeled dUTP theromacin antisense riboprobe. The signal, which appears as dark silver deposits, is restricted also to LFC. Individual LFC titrations of theromacin transcripts were performed using a Biocom system. The histogram shows a strong increase of the level of expression following bacterial challenge. and at the epidermis level. Theromyzin immunolabeling was observed also in LFC (Fig. 7F). Semi-thin sections confirmed the labeling of LFC and showed a strong immunoreactivity in coelomic cavities (CC) in contact with the LFC when using theromacin (Fig. 7E) and theromyzin (Fig. 7, F and G) antisera. No immunolabeling was detected in the controls (Fig. 7, D and  H), confirming the specificity of the reaction. Interestingly, these coelomic cavities contain the body fluid from which theromacin and theromyzin were isolated. These results argue in favor of an exocytosis of the antibacterial peptides from the LFC to the body fluid of the leech.
To study the effect of injury on exocytosis mechanism, a variation of theromacin and theromyzin plasmatic amounts was carried out. Leech liquid coelomic fluid was extracted at We can conclude that, similar to antimicrobial peptides of Drosophila, theromacin and theromyzin peptides are induced principally in specialized tissue (in this case, LFC) and are re-leased immediately into the coelomic fluid after septic injury (22).
In addition to their participation in the systemic response, theromacin and theromyzin may have a role in mucosal defense. Immunocytochemistry (Fig. 7C) underlined the theromacin-positive structures in the epithelia in contact with the external environment, the intestine epithelium, and epidermis. The localization at the epidermis level drove us to search for the peptides in the mucus covering the animals. This investigation also was based on the physiological observations made during the injection. Indeed, we noted a strong enhancement of mucus production by leech after bacterial challenge. After acidic extraction and RP-HPLC purification of the mucus, two peaks eluted at the same ACN percentage as theromacin and theromyzin. Sequencing by Edman degradation and MALDI-TOF MS analysis confirmed that the two molecules isolated from the mucus corresponded to theromacin and theromyzin (data not shown). DISCUSSION Our study described both the isolation and characterization of two novel antibacterial peptides from the body fluid of the leech T. tessulatum. These are theromacin, a cationic peptide exhibiting bactericidal activities, and theromyzin, an anionic peptide with bacteriostatic properties. The peptide sequences deduced from the theromacin and theromyzin genes contain putative signal peptides, indicating that mature peptides correspond to secreted molecules. Importantly, in the context of this study, the genes encoding these peptides are up-regulated by immune challenge in a diffused tissue constituted by the LFC.
Over the last decade, several types of antimicrobial peptides containing cysteine residues have been described in multiple  (23). In addition to having ten cysteine residues instead of six, theromacin does not harbor this consensus sequence. Furthermore, the data bank analysis (Blast program in Swiss-Prot) showed that the peptide has no significant similarity with other known peptides. Thus, theromacin can be considered as a new cysteine-rich antimicrobial peptide. Few cysteine-rich antimicrobial peptides different from the insect/arthropod defensin have been reported in other invertebrates such as the nematode A. suum (9). This finding suggests that cysteine-rich antibacterial peptides could be very ancient in origin, which leads to the questions already advanced by Kato et al. (24). Are antibiotic cysteine-rich molecules found in plants, vertebrates, or invertebrates evolutionarily related? Do they possess a common ancestral molecule?
The majority of antimicrobial peptides including members of the cysteine-rich family possess a global positive charge, allowing their interaction with the negatively charged bacterial membrane (25). Our study shows that theromyzin, in contrast to theromacin, is an anionic molecule. This is the first report of an anionic antibacterial peptide in invertebrates. In vertebrates, antimicrobial peptides with anionic properties were evidenced in the human and in sheep lung (4). Anionic properties are anionic because of the homopolymeric regions of aspartate and require zinc as a cofactor for bactericidal activity. Histatins, a family of histidine-rich antimicrobial peptides found in human saliva, also need the presence of zinc ions for bactericidal activities (26). Circular dichroism studies showed that the antimicrobial activities of histatin 5 require a conformational change that results from the interaction of the peptide with both zinc ions and negatively charged membranes (27). The abundance of histidine residues at the N-terminal part of theromyzin could argue in favor of some common structures between the leech antibacterial peptide and histatins. Based on these data, we can hypothesize that the active part of theromyzin might be reduced to the N-terminal part enriched in histidine and aspartate residues and that theromyzin could require a cofactor for bactericidal activity.
Theromacin and theromyzin genes are expressed exclusively in LFC evenly distributed in the leech. Their transcriptional level is enhanced after bacterial challenge. These results underlined a regulation of the theromacin gene similar to that of the insect antimicrobial peptides genes. Indeed, in the fruitfly, genes encoding antibiotic peptides are induced rapidly following a septic injury and their expression is continuous over 1-3 days following a septic injury (22). Moreover, no difference in gene expression was observed after Gram-positive or Gramnegative injection, suggesting that the antibacterial response of Theromyzon is aspecific. This nonspecificity also has been assumed in Drosophila until the work of Lemaitre et al. (22) demonstrated that the humoral antimicrobial response of the fruitfly discriminates among various classes of microorganisms and mounts a response that is adapted to the infection. This suggests that, in a more natural mode of infection, the leech also could adapt its antimicrobial response.
In addition, it appeared that some physiological events occurring during the transition stage 2/stage 3 could also be inducers of the theromacin and theromyzin gene expression. FIG. 7. Immune localization of theromacin and theromyzin peptides in non-injected stage 3 animals. Theromacin immunoreactivity on paraffin sections was observed in LFC (A), in intestinal epithelia (IE) (B), and at the epidermis level (Ep) (C). On semi-thin sections using an immunogold procedure, positive signals with both theromacin (D and E) and theromyzin antisera were detected at the periphery of the LFC (F) and into coelomic cavities (CC) (G). Controls (theromacin (D) and theromyzin (H)) were incubations of immune serum preadsorbed by synthetic peptides.
Since the stage 3 leeches entered their gametogenesis phase, we hypothesized that such hormonal factors implicated in this sexual maturation may participate in the induction of theromacin and theromyzin genes. This kind of developmental induction had been reported already for the Drosophila diptericin gene. Meister and Richards (28) present evidence that an important increase in the response of the gene occurred during the third larval stage, leading to a maximum in late larvae and early prepupae after a septic injury.
The massive production of theromacin and theromyzin is followed immediately by a rapid release of the peptides into the coelomic fluid after septic injury. Thus, theromacin and theromyzin may play their antimicrobial activities through a systemic action. Moreover, the presence of these peptides in the intestinal epithelial cells and at the epidermis level suggests that leech antimicrobial peptides could also play a role in epithelial defense. The localization of antibiotic molecules in the gastrointestinal tract also has been reported in insects and in vertebrates where they provide a rapid local immune response against exogenous pathogens brought in during feeding (29 -31). The detection of theromacin and theromyzin in leech mucous evokes the local defensive response reported in frogs in which antibacterial peptides secreted in the mucous prevent bacteria colonization and/or subsequent infection (30). In addition, because a mucous membrane covers the eggs after laying, we hypothesized a protective role of leech theromacin and theromyzin against bacteria during egg development.
The sum of these data allows us to present a scheme for the role of antimicrobial peptides in the immune system of the leech. Septic injury provokes an important production of mucous that would trap bacteria present in the external environment of the leech. The trapped bacteria then would be killed by theromacin and theromyzin present in the mucous. At the same time, injury would induce theromacin and theromyzin gene expression and the secretion of the gene products from the LFC into the body fluid of the animal. Thus, theromacin and theromyzin would exert their antimicrobial property through a systemic action. This would lead, in conjunction with the phagocytic action of coelomocytes, to the microbial killing in the body fluid.
The results presented here are in contrast with those reported for other lophotrochozoans such as the mollusk M. galloprovincialis and the annelid L. rubellus (13,19). In these animals, antibiotic peptides are expressed constitutively and stored in hemocytes; no specific organ like the fat body is implicated. The production and secretion of active theromacin and theromyzin by a specific tissue and the rapid transcriptional up-regulation on infection evoke for the first time in a lophotrochozoan an antimicrobial response developed by holometabola insects (32,33). Indeed, the recent data made a distinction between insects undergoing complete (holometabolas) and incomplete (heterometabolas) metamorphosis. In many studied holometabolas, i.e. in dipterans or in lepidopterans, antibiotic peptide genes are induced by bacterial challenge (16). In contrast, no inducible system was found in heterometabolas like the termite (isopteran). These results suggest that, in arthropods, such inducible systems of antimicrobial peptide gene could have appeared at the time of the divergence of holometabolas from heterometabola insects. The similarity between the antibacterial response of the leech and that of holometabolas also is supported by the functional resemblance between the leech LFC and the insect fat body, which possess the common capacity to produce egg yolk proteins.