Insect Immunity ISOLATION FROM A COLEOPTERAN INSECT OF A NOVEL INDUCIBLE ANTIBACTERIAL PEPTIDE AND OF NEW MEMBERS OF THE INSECT DEFENSIN FAMILY*

Injection of heat-killed bacteria into larvae of the large tenebrionid beetle Z o p h o b utrutus (Insecta, En-dopterygota, Coleoptera) results in the appearance in the hemolymph of a potent antibacterial activity as evidenced by a plate growth inhibition assay. We have isolated three peptides (A-C) from this immune hemo- lymph which probably account for most of this activity. Their primary structures were established by a com-bination of peptide sequencing and molecular mass determination by mass spectrometry. Peptide A, which is bactericidal against Gram-negative cells, is a 74-residue glycine-rich molecule with no sequence homology to known peptides. We propose the name coleop- tericin for this novel inducible antibacterial peptide. Peptides B and C are isoforms of a 43-residue peptide which contains 6 cysteines and shows significant sequence homology to insect defensins, initially reported from dipteran insects. This peptide is active against Gram-positive bacteria. The results are discussed in connection with recent studies on inducible antibacter- ial peptides present in the three other major orders of the endopterygote clade of insects: the Lepidoptera, Diptera, for the other enzymes. Each enzyme was assayed at a concentration of 5, 10, and 20 IU/mg of protein (cell-free hemolymph). After enzymatic treatment, the samples were di- rectly assayed against M. luteus or E. coli strains D22 and D31 in the plate growth inhibition assay. In parallel, the inocuity of the various enzymes on the bacterial growth was tested in strictly identical conditions (pH, concentration). Control experiments were run in the absence of enzymes.

Insect immunity is at present a rapidly expanding field of research. A series of pioneering studies had established by 1930 that larvae of Lepidopteran insects are able to build up a potent humoral antibacterial response when challenged with low doses of bacteria (1)(2)(3)(4). It however took nearly 50 years until the first molecules responsible for this inducible humoral antibacterial activity of Lepidopterans were isolated and their structures characterized. Studies in Hyalophora cecropia, Galleria mellonella, and Manduca sexta have since established that these insects synthesize, in response to a bacterial challenge or a septic injury, three major groups of antibacterial peptides: (i) cecropins, which are 4-kDa basic peptides (5-8); (ii) attacins, which are 20-22-kDa basic or acidic proteins (9,IO), and (iii) lysozyme (11)(12)(13). These isolation studies of antibacterial peptides were rapidly extended to Diptera, which were shown to contain in their immune blood, in addition to cecropins (Sarcophnga (14), Drosophila (15)) and to an attacin-related 27-kDa peptide (Sarcophaga (16)), two novel an-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Interestingly, within the vast class of the insects, inducible antibacterial peptides have so far been isolated only from species belonging to the endopterygote clade. This huge group of insects, which contains more species than the rest of the living world together, is characterized by a non-feeding and immobile last immature instar (the pupa), from which the adult organism emerges after far-reaching metamorphic events. The overwhelming species richness in the Endopterygota results from their diversification in only four of the constituant orders: the Coleoptera, the Hymenoptera, the Diptera, and the Lepidoptera.
Surprisingly, no inducible immune peptide had been characterized so far from the coleopteran branch of the endopterygote clade, although an antibacterial response can be elicited in this order (23,24). In the present study we have used larvae of a large tenebrionid beetle, Zophobas atratus, for the isolation and structural characterization of antibacterial peptides which appear in the blood in response to injection of heatkilled bacteria. We report the identification of a novel 74residue antibacterial peptide, which we propose to name coleoptericin, and two isoforms of a new member of the insect defensin family of antibacterial peptides. We were unable to detect in the immune larvae molecules such as cecropins, attacins, diptericins, or apidaecins which participate in the antibacterial response in other orders of the Endopterygota.

Bacterial Strains
The bacterial strains were gifts from the following colleagues: Escherichia coli D31 (streptomycin resistant) and Enterobacter cloacae Dl2 (nalidixic acid resistant) from H. Boman
Immunization 3rd instar larvae (6-8-month-old) of 2. atratus (average weight, 1 g/larva) received a 10-pl injection of overnight heat-killed cultures of M. luteus and E. coli D31 containing approximately 1 million cells of each germ. After various time intervals, the insects were chilled for 1 min in ice-cold water, and several drops (-30 plllarva) of hemolymph were recovered by sectioning a metathoracic leg and gently squeezing the abdomen. The hemolymph was pooled in a precooled plastic tube in the presence of aprotinin (Sigma; final concentration, 10 pg/ml of hemolymph). After centrifugation at 70,000 X g for 50 min at 4 "C, the cell-free hemolymph was clarified through a Millex 0.8-pm filter.
Antibacterial Assays (a) Plate Growth Inhibition Assay-Assay conditions were essentially those described by Lambert et al. (18): sterile Petri dishes (9cm diameter) received 7.5 ml of melted agar in buffered nutrient medium, pH 7.2, containing -2 X 1 0 ' logarithmic-phase cells of a given bacterial strain. Wells (2-mm diameter) were cut into the freshly poured plates after the solidification of the agar. Each well received a 2 4 aliquot of the fraction suspected to contain antibacterial molecules. The plates were incubated overnight at 37 "C, and the diameters of the clear zones were recorded, after subtraction of the well diameter.
(b) Bactericidal Assay-250 pmol of purified peptide A in 10 p1 of distilled water were incubated in microtiter plates in the presence of 100 p1 of a logarithmic-phase culture of E. coli D22 at a starting ODW of 0.15. Aliquots were removed at different time intervals and plated on nutrient agar to determine, by an overnight culture at 37 "C, the number of colony forming units.

Purification of Antibacterial Peptides
Step I: Sep-Pak Prepurification-The cell-free filtered hemolymph was acidified (0.05% trifluoroacetic acid) and applied on a Sep-Pak CIS cartridge (Waters Associates). Stepwise elution was performed with increasing proportions of acetonitrile (up to 60%) in water acidified with 0.05% trifluoroacetic acid. Antibacterial activity was monitored on aliquots of the fractions that had been vacuum-dried to remove acetonitrile.
Step II: Reuersed-phase HPLC'-The active fractions from the Sep-Pak prepurification were vacuum-dried, and the residue was dissolved in 250 pl of 10% acetonitrile in acidified water (0.05% trifluoroacetic acid) and subjected to reversed-phase HPLC on an Aquapore RP 300 Ca (250 X 4.6-mm) column (Brownlee Associates). Elution was performed with a linear gradient of 10-60% of acetonitrile in acidified water over 90 min at a flow rate of 1 ml/min. Ultraviolet absorption was monitored at 225 nm. The eluted fractions were vacuum-dried and dissolved each in 50 pl of distilled water; the antibacterial activity was monitored on 2-pl aliquots.
Step IIZ: Final Purification-Different methods were used for the final purification of the active compounds present in the three peaks referred to as A, B, and C under "Results" (Fig. 2): A, reversed-phase HPLC on an Aquapore RP 300 C, (250 X 4.6-mm) column with a relatively mild gradient elution of 18-38% acetonitrile in acidified water over 120 min; B, isocratic size exclusion chromatography on a Protein-Pak 125 column (Waters Associates) with 30% acetonitrile in acidified water; C, reversed-phase HPLC on an Aquapore RP 300 C, (250 X 4.6-mm) column with an elution gradient of 20-40% acetonitrile in acidified water over 120 min. The antibacterial activity was monitored on 2 -4 aliquots of the concentrated fractions as in step 11.
The abbreviation used is: HPLC, high performance liquid chromatography.

Amino Acid Sequence Analysis
Automated Edman degradation of peptides and detection of phenylthiohydantoin derivatives were performed on a pulse liquid automatic sequentator (Applied Biosystems, model 473).

Mass Spectrometry
The mass spectra were recorded on a Bio-Ion 10 K plasma desorption mass spectrometer (Bio-Ion AB). Peptides were analyzed using nitrocellulose targets.

Enzymatic Digestion
Endoproteinase Glu-C (Staphylococcus aureus VS protease) was purchased from Pierce Chemical Co. Peptide A (10 pg) was dissolved in 100 p1 of 25 mM HC03NH4, pH 4, and 750 ng of enzyme were added. Enzymatic reaction lasted for 18 h at 25 "C. The digestion peptides were recovered from the mixture by reversed-phase HPLC (conditions as above for (step 11)).
Protease Assay 100-pl samples of immune cell-free hemolymph (see above) were subjected to protease treatment during 18 h at 37 "C. The following enzymes were used pepsin (EC 3.4.23.1, Sigma), protease from Streptomyces griseus (Sigma), trypsin (EC 3.4.21.4, Worthington) and achymotrypsin (EC 2.4.21.1, Worthington). The enzymes were diluted in appropriate buffers: 50 mM KC1-HCl, pH 2.0, for pepsin, and 50 mM NH,Cl-NaOH, pH 8.5, for the other enzymes. Each enzyme was assayed at a concentration of 5, 10, and 20 IU/mg of protein (cellfree hemolymph). After enzymatic treatment, the samples were directly assayed against M. luteus or E. coli strains D22 and D31 in the plate growth inhibition assay. In parallel, the inocuity of the various enzymes on the bacterial growth was tested in strictly identical conditions (pH, concentration). Control experiments were run in the absence of enzymes.

RESULTS
Appearance of Antibacterial Activity in the Hemolymph of Larvae of Z. atratus after Injection of Heat-killed Bacteria (Immunization)-In a pilot experiment, nine groups of seven 3rd instar larvae received an injection of heat-killed bacteria, and their hemolymph was collected after various time intervals up to 4 weeks. The presence of antibacterial activity was monitored for each larva on an aliquot of cell-free hemolymph in the plate growth inhibition assay against the Gram-positive M. luteus. The results are presented in Fig. 1 and show that a strong antibacterial activity appeared between 6 and 12 h following the inoculation of bacteria. A high activity was monitored at 24 h and was maintained for at least 12 h. Significant activity was still detectable after 4 weeks. In contrast, untreated larvae were devoid of antibacterial sub- stances in their cell-free hemolymph.
Essentially similar results were obtained when the antibacterial activity was tested on the Gram-negative E. coli strains D22 and D31 (data not shown).
When the cell-free hemolymph of challenged larvae was subjected to protease treatment (see "Materials and Methods''), no antibacterial activity could be evidenced, indicating that the molecules responsible for this activity are peptides.
Isolation of Three Antibacterial Peptides (Peptides A-C) from Immune Hemolymph of 2. atratus-150 3rd instar larvae of 2. atratus were immunized as above, and their hemolymph was collected in the cold after 48 h, yielding a total volume of 5 ml. The hemocytes were removed by centrifugation, and the supernatant was filtered through a Sep-Pak C1, cartridge. The antibacterial activity was recovered by elution with 60% acetonitrile in acidified water. The eluate was applied to a reversed-phase HPLC column and eluted with a linear gradient of acetonitrile in acidified water as shown in Fig. 2. Aliquots of the eluted fractions were tested in the plate growth inhibition assay on M. luteus and E. coli D31 and D22. Anti-E. coli activity was observed in absorption peak A (at 27.5% of acetonitrile) while two peaks, B and C (at 33 and 35% acetonitrile), contained anti". luteus activity. No other fraction was observed to contain antibacterial activity in our conditions. The active compounds were further purified by reversed-phase HPLC (A and C) and size exclusion chromatography (B). Apparently pure substances were recovered, as judged by UV monitoring at 225 nm (Fig. 3). The estimated purification yields were as follows: from 5 ml of hemolymph containing 150 mg/ml of total proteins, we recovered in pure form 40 pg of peptide A, 15 pg of peptide B, and 15 pg of peptide C. This extraction procedure was repeated several times to obtain sufficient pure material for the subsequent studies.

Primary Structure Determination of Peptides A, B, and C-
The three peptides were sequenced by automated Edman degradation. The sequences are presented in Fig. 4 and show two distinct types of molecules: A is a 74-residue peptide devoid of cysteines, and B and C are isoforms of a 43-residue cysteine-rich peptide with sequence similarities to the insect defensins identified in the immune blood of the dipteran Phormia terranovae (18).

FIG. 4. Amino acid sequences (in one letter code) of the three antibacterial peptides (A, B, C) from 2. atratus.
The amino acid sequences of peptides B and C are compared with the sequence of defensin A from P.
terranovae (18). The sequences are aligned for homology; identical residues are in bold characters. 24523 fragments were separated by reversed-phase HPLC and sequenced by Edman degradation. One fragment of 38 residues corresponded to the 38 NHz-terminal residues identified above, while the other fragment of 36 residues gave the COOH-terminal sequence of peptide A. Mass spectrometric measurement of the intact peptide yielded a molecular mass of m/z 8114.7 which is in good agreement with the calculated average isotopic mass (MH') of 8110.8, Peptide A is strikingly rich in Gly residues (18%) which are evenly distributed in the molecule. The NHp-terminal third (up to residue 23) is devoid of charges. The central part of the molecule is highly charged, both positively and negatively, while the COOH terminus of the peptide has a marked basic character. The overall PI of the molecule is 10.9. The peptide contains no cysteines.
Peptides B and C-The sequence of 43 residues of peptide C was obtained by subjecting 400 pmol of pure material to Edman degradation. The repetitive yield of the sequencator was about 95%. No phenylthiohydantoin was present at positions 3, 20, 24, 34, 40, and 42; the intensities of the signals preceding and following these six blanks clearly indicated that they correspond to cysteines, as is the case in defensins from Phormia, allowing for a gap of 4 residues in the dipteran peptides (see Fig. 4 and "Discussion"). This assumption was corroborated by determination of the molecular mass which was found to be at m/z 4395. Indeed, the calculated monoisotopic molecular mass (MH+) is 4396 when the six blanks are considered to correspond to cysteines engaged in three intrachain disulfide bridges. For peptide B, 400 pmol of pure material were subjected to Edman degradation, and a sequence was obtained which was fully superposable on that of peptide C with a single replacement of Thr-30 in C by Arg-30 in B.
Both peptides are basic with a PI of 7.9 for C and 8.2 for B (the presence of an Arg in B in place of a Thr in C accounts for the difference in basicity of the two peptides).
Activity Spectrum of Peptides A and C and Mode of Action of Peptide A from Immune Hemolymph of Z. atratus-Purified peptide A and peptide C were tested in the plate growth inhibition assay against various bacterial strains as illustrated in Table I. Peptide A was found to be highly active against E. coli D31 and D22. Lower but significant activity was also observed against two other Gram-negative strains, A. baumanii and P. maltophilia. M. luteus was the only Grampositive cell affected by this peptide in our conditions. Peptide C was strongly active against M. luteus, S. pyogenes, and 8.
subtilis QB 935 and affected moderately the growth of Corynebacterium D2. No Gram-negative cell was found to be sensitive to peptide C, except for the D22 strain of E. coli.
Pure peptide A was tested in the liquid growth inhibition assay at 2.5 j~h4 against the highly sensitive strain D22 of E.
coli. As shown in Fig. 5, a 0.5-h contact with the peptide was sufficient to kill growing cells of this strain. Pure peptides B and C yielded similar results when tested against M. luteus at the same concentration (data not shown). This result is in agreement with our previous observations on the bactericidal activity of insect defensins from P. terranovae (18).

TABLE I Activity spectrum of peptides A fcoleoptericin) and C (insect defensin)
from immune hemolymph of Z. atratus 500 ng of pure peptide were tested in the plate growth inhibition assay (two independent assays/bacterial strain), and the diameter of growth inhibition was recorded. Expression is noted as follows: +, 2 to 4 mm; ++, 4 to 8 mm; +++, 8

DISCUSSION
Our study establishes that larvae of a coleopteran species respond to a bacterial challenge by the rapid appearance in their hemolymph of antibacterial peptides. This result corrob-orates earlier investigations in Eleodes, another coleopteran insect (23; see also Ref. 24). The time course of the appearance is roughly equivalent to that observed in representative species of Lepidoptera and Diptera. The activity persists for a longer time period, which may reflect the remarkably long duration (several months) of the 3rd larval instar of 2. atratus.
AS shown in this report, the inducible antibacterial activity monitored by the plate growth inhibition assay in the cellfree hemolymph of challenged larvae of 2. atratus is essentially attributable to two types of molecules. Cecropins, attacins, diptericins, apidaecins, and lysozymes, which are active against one or all of the three routinely used test organisms ( M . luteus, E. coli D22 and D31) were not observed although our working conditions were favorable for their detection.
Peptide A has no sequence similarity to other known peptides, and we propose therefore the name coleoptericin for this novel inducible antibacterial peptide. Coleoptericin is basic in character (PI 10.9) and is strikingly rich in glycine residues (18%). This is not uncommon among insect antibacterial peptides. The 8-kDa diptericins of Phormia and Drosophila contain 18 and 22%, respectively, of glycine residues and within the larger attacin (20)(21)(22)(9)) and the attacinrelated sarcotoxin I1 (27 kDa, (16)) 60-residue glycine-rich domains are observed which contain up to 20% glycine residues (see also "Discussion" in (25)). Coleoptericin, like diptericin, is bactericidal against Gram-negative bacteria.
The presence of insect defensins, peptides B and C, in Coleoptera is interesting in many respects. Insect defensins were initially discovered in Diptera (immune blood of P. terranouae; (18); cf. also sapecin, a homologous molecule secreted by an embryonic cell line of Sarcophuga peregrina, (26)). Their presence has not been reported so far in Lepidoptera, where they are probably absent, given the efforts which have been devoted to the isolation of antibacterial peptides in this order. However, a defensin-related 51-residue peptide is present in the royal jelly of the honey bee (referred to as royalisin, (27)). The presence in insects of antibacterial peptides with sequence similarities to the mammalian defensins has attracted attention and prompted the suggestion that defensins were ancestral antibacterial peptides (18,20). Mammalian defensins form a relatively large family of variably cationic peptides comprised of 29-34 amino acid residues. They all contain a characteristic motif of 6 cysteines engaged in three intramolecular disulfide bridges. However, the spacing between the cysteine residues differs markedly between insect and mammalian defensins and the connectivity between the three disulfide bridges has been shown recently to be different (28-30), which casts some doubt on the proposed homology.
The sequence similarities between the defensins of 2. atratus and the three dipteran defensins (defensins A and B of P. terranouae and sapecin of S. peregrim) is especially high in the COOH-terminal part of the molecule. A detailed NMR analysis of recombinant Phormia defensin A produced in yeast (31) has recently shown that the COOH-terminal part of defensin forms a short a-helix followed by two antiparallel @-sheets.' This arrangement is probably conserved between Phormia and 2. atratus given the 80% sequence similarity for this region. In contrast, the NH2-terminus of Phormia defensin forms a large loop and we observe that the amino acid sequence is only poorly conserved (40%) in this region between the two species with the exception of a stretch of 5 residues (TCDV/LL) comprising the first NHz-terminal cysteine (which in Phormia defensin anchors the loop to the psheet). In other words, if we accept the idea that the coleop-* M. Ptak, personal communication.
teran defensin has a three-dimensional organization basically similar to that of dipteran defensin, the major change in sequence affects the NHz-terminal loop. Whether and how this change reflects on the mode of action or the spectrum of activity remains to be established. So far in our biological tests, both defensins were bactericidal against the same Grampositive bacteria at similar concentrations (comprised between 0.5 and 2.5 pM).
Taken in conjunction with the results we obtained in Lepidoptera, Diptera, and Hymenoptera, our data show that insects of the four major orders of the endopterygote clade all respond to a bacterial challenge by the production of several groups of antibacterial peptides. The full characterization of these peptides has only been performed in a few species and frequently only at one stage of development, as is the case in this paper. It is therefore premature to draw firm conclusions as to the distribution of the various antibacterial peptide families within this insect clade. However, it is now clear that the four orders under investigation do not necessarily produce the same array of antibacterial peptides in contrast to earlier assumptions. Defensins probably play a paramount role in the anti-Gram-positive response of Coleoptera, Diptera, and possibly Hymenoptera while Lepidoptera rely on cecropins and lysozyme. Gram-negative cells are also countered by cecropins in Lepidoptera and Diptera, and in addition, by relatively large-sized glycine-rich peptides in Coleoptera (coleoptericin) and Diptera (diptericins) and by small prolinerich peptides in Hymenoptera (abaecins, apidaecins).
It is at present unclear whether insects belonging to other clades also synthesize antibacterial peptides in response to a bacterial challenge. In fact with the exception of lysozyme, no antibacterial peptide has been chemically characterized so far from any insect species outside the Endopterygota. It will be interesting to extend these studies to other insect groups; it is obviously also a necessity to complete our information on the four orders of the endopterygote clade.
In conclusion, we have shown that within a few hours after an injection of heat-killed bacteria, three basic antibacterial peptides of relatively low molecular weights appear in the hemolymph of 3rd instar larvae of 2. atratus. These molecules certainly participate in the antibacterial response of the injured insects as they exhibit potent activity against several of the bacterial strains which were tested. It is probable nevertheless that the immune response of these insects is not limited to the synthesis of coleoptericin and defensins. In particular, larger proteins with antibacterial activity, such as some lectins (32), would not have been detected in the plate growth inhibition assay. Also the role of phagocytosis and capsule formation by hemocytes, as evidenced in other insect orders (33-34), remains to be investigated in Coleoptera. Clearly, we are only at the beginning of our understanding of the immune response in this group which outnumbers in terms of species the sum of all the other animal groups.