Two Distinct Pathways can Control Expression of the Gene ...€¦ · Two Distinct Pathways can Control Expression of the Gene Encoding the Drosophila Antimicrobial Peptide Metchnikowin

J. Mol. Biol. (1998) 278, 515±527

Two Distinct Pathways can Control Expression of theGene Encoding the Drosophila AntimicrobialPeptide Metchnikowin

Elena A. Levashina, Serge Ohresser, Bruno Lemaitre andJean-Luc Imler*

ReÂponse Immunitaire etDeÂveloppement chez lesInsectes, UPR 9022 du CNRSInstitut de Biologie MoleÂculaireet Cellulaire, 15 rue ReneÂDescartes, 67000 StrasbourgFrance

Abbreviations used: ORF, open rgreen ¯uorescent protein from AeqDORSAL-related immunity factor;regulatory factor 1.

0022±2836/98/180515±13 $25.00/0/mb

Metchnikowin is a recently discovered proline-rich peptide fromDrosophila with antibacterial and antifungal properties. Like most otherantimicrobial peptides from insects, its expression is immune-inducible.Here we present evidence that induction of metchnikowin geneexpression can be mediated either by the TOLL pathway or by the imdgene product. We show that the gene remains inducible in Toll-de®cientmutants, in which the antifungal response is blocked, as well as in imdmutants, which fail to mount an antibacterial response. However, in Toll-de®cient;imd double mutants, metchnikowin gene expression can nolonger be detected after immune challenge. Our results suggest thatexpression of this peptide with dual activity can be triggered by signalsgenerated by either bacterial or fungal infection. Cloning of the metchni-kowin gene revealed the presence in the 50 ¯anking region of severalputative cis-regulatory motifs characterized in the promoters of insectimmune genes: namely, Rel sites, GATA motifs, interferon consensusresponse elements and NF-IL6 response elements. Establishment oftransgenic ¯y lines in which the GFP reporter gene was placed under thecontrol of 1.5 kb of metchnikowin gene upstream sequences indicatesthat this fragment is able to confer full immune inducibility and tissuespeci®city of expression on the transgene.

# 1998 Academic Press Limited

Keywords: Drosophila immunity; antimicrobial peptides; metchnikowin;Rel; TOLL pathway
*Corresponding author
Introduction

Insects have long been known to be particularlyresistant to microbial infections. This resistance isdue to an ef®cient host defense system that com-prises the proteolytic cascades leading to coagu-lation and melanization of the hemolymph, thephagocytosis and encapsulation of invading micro-organisms by blood cells, and the secretion into theblood of a battery of potent antimicrobial peptides(for a recent review see Hoffmann & Reichhart,1997). Infection leads to the rapid and transientsynthesis of these peptides, mainly by the fat body,a functional equivalent of the mammalian liver.Recent studies indicate that in Drosophila at least

eading frame; GFP,uonea victoria; DIF,IRF-1, interferon

981705

seven distinct peptides, and their isoforms, partici-pate in the humoral immune response. Five ofthese peptides, cecropins (Kylsten et al., 1990), dip-tericin (Wicker et al., 1990), drosocin (Bulet et al.,1993), attacin (Asling et al., 1995) and insect defen-sin (Dimarcq et al., 1994), are selectively activeagainst bacteria, whereas drosomycin, is solelyactive against fungi (Fehlbaum et al., 1994). Theseventh and most recently discovered molecule,metchnikowin, is unique among the antimicrobialpeptides of Drosophila in that it is active againstboth bacteria and fungi (Levashina et al., 1995).

The study of the regulatory mechanisms control-ling the rapid synthesis of the antimicrobial pep-tides after immunological challenge has become a®eld of intense research. Analysis of the humoralimmune response in different mutant strains ofDrosophila has led to the de®nition of two distinctpathways regulating antimicrobial peptide geneexpression. Lemaitre et al. (1995) have recently

# 1998 Academic Press Limited

516 Regulation of Metchnikowin Gene Expression

described a recessive mutation, immune de®ciency(imd), which impairs the inducibility of the genesencoding antibacterial peptides, while only mar-ginally affecting the inducibility of drosomycin.This gene has not yet been cloned. On the otherhand, these authors have shown that thewell characterized TOLL pathway (Morisato &Anderson, 1995) controls drosomycin geneexpression. During embryonic dorsoventral pat-terning in Drosophila, the TOLL receptor is acti-vated by a processed form of the product of thegene spaetzle (spz). A cascade of proteolytic clea-vages involving the products of the genes gastrula-tion defective (gd), snake (sn) and easter (ea), whichall code for secreted serine proteases, results in therelease of active SPAETZLE in the perivitelline¯uid on the ventral side of the embryo (Morisato &Anderson, 1995). Activation of TOLL results in therelease of the Rel protein DORSAL from the inhibi-tor CACTUS. The cytoplasmic proteins TUBE andPELLE intercalate between TOLL and CACTUS inthis pathway. Interestingly, this signaling cascadebears striking structural and functional similarityto the activation cascade of the Rel protein NF-kBin mammals: CACTUS is a structural homologueof IkB, which retains NF-kB in the cytoplasm ofmammalian cells (Siebenlist et al., 1994). The intra-cytoplasmic domain of TOLL shows sequencehomology with the corresponding domain in thetype I interleukin-1 (IL-1) receptor (Gay & Keith,1991); and the kinase PELLE is related to thekinase IRAK, which interacts with the IL-1 receptor(Cao et al., 1996). In mutants de®cient for spz, Toll(Tl), tube (tub) or pelle (pll), drosomycin inducibilityis severely reduced, while the induction of dipteri-cin and drosocin is not affected. Furthermore, in Tlgain-of-function mutants (TlD), as well as inmutants de®cient for the inhibitor CACTUS, droso-mycin is expressed in the absence of immune chal-lenge (Lemaitre et al., 1996). The TOLL pathwaytherefore appears necessary and suf®cient for thecontrol of drosomycin expression. It also appearsnecessary for inducibility of cecropin, defensin andattacin, since expression of these genes is reducedafter immune challenge in loss-of-function mutantsof spz, Tl, pll or tub. However, none of these threepeptides is expressed in the absence of immunestimulation in TlD or cactus (cact)-de®cient mutants,indicating that the TOLL pathway is not suf®cientfor the control of their expression (Lemaitre et al.,1996).

The importance of the TOLL and IMD pathwaysin the Drosophila host defense system has beenillustrated by experiments in which mutant ¯ieswere challenged with fungi or bacteria. Tl-de®cient¯ies rapidly succumb to fungal, but not bacterial,infection, whereas imd mutants resist fungi but aremarkedly susceptible to bacteria (Lemaitre et al.,1995, 1996). These results are consistent with ourpresent knowledge of the functions of the antimi-crobial peptides in Drosophila, since Tl controlsexpression of the antifungal peptide gene droso-mycin, while imd appears to control, or to partici-

pate in the control of, the antibacterial peptides.Here, we have investigated the regulation ofexpression of the newly discovered peptide metch-nikowin, which has both antibacterial and antifun-gal activity (Levashina et al., 1995). We report thatthe regulation of this peptide is unique comparedto the other peptides, in that it appears to be regu-lated independently by the TOLL and the IMDpathways. We also report cloning of the metchni-kowin gene and sequencing of 1.5 kb of upstreamnon-transcribed sequences. We show by functionalstudies using transgenic ¯y lines that this fragmentcontains the metchnikowin promoter.

Results

Transcription of the metchnikowin gene

We have previously shown that metchnikowinexpression is induced upon immune challenge inDrosophila adults, and reaches a maximum level 14hours after septic injury (Levashina et al., 1995). Asstated in the Introduction, two pathways havebeen shown to participate in the regulation of thehumoral immune response in Drosophila. The ®rstpathway is de®ned to date only by the imd geneproduct, which controls antibacterial peptide geneexpression (Lemaitre et al., 1995), whereas theTOLL pathway regulates the expression of the anti-fungal peptide drosomycin (Lemaitre et al., 1996).In order to determine if these pathways regulatemetchnikowin expression, we performed Northernblots with RNA prepared from wild-type ormutant ¯ies. Two types of mutants affecting theTOLL pathway were analyzed: ®rst, strains carry-ing strong loss-of-function mutations of gd, snk, ea,spz, Tl, tub, pll and dorsal (dl), which are known toblock the dorsoventral signaling pathway resultingin dorsalized embryos; and second, gain-of-func-tion mutations in Tl (TlD) and loss-of-functionmutations in cact that are strongly ventralizing inembryos. Mutants of the ®rst type, as well as theimd mutant ¯ies in which induction of the genesencoding antibacterial peptides is impaired, werechallenged by an injection of bacteria before RNAextraction six hours later, whereas ventralizingmutants were analyzed in the absence of challenge.

As shown in Figure 1, metchnikowin remainedsigni®cantly inducible by septic injury in Drosophilaadults carrying a loss-of-function allele of the Tlgene. The variability from experiment to exper-iment, which can be observed in all mutant back-grounds, was also observed for all other peptides,and most probably re¯ects the complexity of thisin vivo system (Lemaitre et al., 1996). Nonetheless,only in one out of seven experiments performedwas metchnikowin inducibility reduced to 50% ofthe wild-type inducibility (Figure 1, compare lanes2 and 9). In all other experiments, the inducibilityof metchnikowin gene expression in Tl-de®cientmutants was similar or superior to the level ofinducibility in wild-type ¯ies. These experimentswere extended to other mutations affecting the

Figure 1. Quanti®cation ofmetchnikowin expression in wild-type and mutant adults. Total RNAprepared from control (unchal-lenged: uc) or bacteria-challenged(six hours) wild-type (wt) ormutant adult ¯ies were analyzedby Northern blot using a metchni-kowin cDNA probe. The signals onseveral Northern blots were quanti-®ed with a Bio-Imager system. Ineach experiment the blot was alsoprobed with the ribosomal proteinrp49 cDNA and the signals formetchnikowin were normalizedwith the corresponding value ofthe rp49 signal. The levels ofexpression in wild-type adults sixhours after bacterial challenge weretaken arbitrarily as 100 and theresults are presented as relativeactivities. Each bar corresponds toan independent experiment. Lanesare as follows: 1 and 2, OrR; 3, dl1/dl1; 4, dlT/Df(dl); 5, tub238/tub118; 6,tub3/tub118; 7, pll078/pll21; 8, pll078/pllrm8; 9, Tlr632/Tl1-RXA (29�C); 10,Tlr444/Tl9QRE; 11, spz197/spz197; 12,spzrm7/spzrm7; 13, ea1/ea1; 14, ea818/ea818; 15, snk073/snk073; 16, gd7/gd7;17, gd8/gd8; 18, Tl1-RXA/ � ; 19,imd/imd; 20, cactA2/cactA2; 21,Tl10b/ � .

Regulation of Metchnikowin Gene Expression 517

TOLL pathway. Mutants de®cient in the gd, snkand ea genes, which code for serine proteases act-ing upstream in the pathway during embryonicdevelopment, and which are not essential for thecontrol of drosomycin expression (Lemaitre et al.,1996), showed a wild-type response to bacterialchallenge (Figure 1, compare lanes 13 to 17 to lane2). In addition, loss-of-function mutations in tub, pllor spz, which affect drosomycin expression(Lemaitre et al., 1996), clearly did not result indecreased inducibility of metchnikowin (Figure 1,compare lanes 5 to 12 to lane 2). On the contrary,metchnikowin RNA expression level in theseinduced mutants ¯ies was often higher than inwild-type. Similar results were obtained with dl-de®cient mutant ¯ies (Figure 1, compare lanes 3and 4 to lane 2). Although we do not have anyexplanation to date for these variations, we notethat similar upregulation of diptericin mRNA inthese mutants has been observed (Lemaitre et al.,1996). Altogether, these data indicate that metchni-kowin gene expression can be turned on by bac-terial challenge in the absence of a functionalTOLL pathway.

We next examined metchnikowin geneexpression in imd mutants, in which the antibacter-ial response is affected. We observed that metchni-kowin gene expression was slightly reduced in imdmutants. In one out of four experiments, the indu-cibility of metchnikowin in imd mutants was

reduced by ®vefold, in another experiment it wasequivalent to wild-type inducibility, whereas in theremaining two experiments, induction level wasroughly half of wild-type (Figure 1, compare lanes19 and 2). Therefore, we conclude that metchniko-win expression remains signi®cantly inducible bybacterial challenge in the absence of a functionalimd gene product.

These data suggest either that in immune-chal-lenged ¯ies metchnikowin expression is not regu-lated by the IMD or TOLL pathway, or that bothpathways can control metchnikowin geneexpression. To discriminate between these two pos-sibilities, we examined the inducibility of metchni-kowin gene expression six hours after septic injuryin double mutants for Toll and imd, and observedthat in this context the induction of metchnikowinis completely abolished (Figure 2). These resultssustain the hypothesis that both pathways partici-pate in the regulation of metchnikowin expression.

In order to study further the role of the TOLLpathway in metchnikowin gene regulation, wethen analyzed metchnikowin gene expression inthe absence of immune challenge in TlD gain-of-function and cact-de®cient adults. Previous studieshad shown that drosomycin is constitutivelyexpressed in these mutants (Lemaitre et al., 1996),thus suggesting that the TOLL pathway is suf®-cient for regulation of drosomycin expression.Strikingly, a signi®cant expression of metchniko-

Figure 2. Transcriptional pro®le of metchnikowin inwild-type or imd;Tl-de®cient double mutant Drosophilaadults. A representative Northern blot of total RNAextracted from control (0 hour; lanes 1 and 5) or bac-teria-challenged (1, 6 or 24 hours after challenge) wild-type (wt) or mutant adult Drosophila is shown. The blotwas ®rst probed with the metchnikowin cDNA, andafter dehybridization, with the ribosomal protein rp49cDNA. Lanes 1 to 4, wild-type Drosophila (OregonRstrain); lanes 5 to 8, imd;Tl-de®cient double mutant.


win RNA was observed in the absence ofimmune challenge in cact-de®cient mutants(Figure 1, compare lanes 20 to lane 1) and TlD

mutants (Figure 1, compare lanes 21 to 22 to lane1). In all four experiments performed with cact-de®cient ¯ies, the metchnikowin expression levelwas between 25% and 40% of that in immune-challenged wild-type ¯ies. Similar results wereobtained in TlD mutant ¯ies, with the exceptionof two out of seven experiments in which metch-nikowin RNA expression level only reached 15%of that in immune-challenged wild-type ¯ies. Weconclude from these data that activation of theTOLL pathway is suf®cient to induce a signi®-cant level of metchnikowin expression.

These results indicate that metchnikowin regu-lation differs from that of the other antimicrobialpeptides of Drosophila characterized to date, inthat it remains inducible in Toll-de®cient and inimd mutant ¯ies. We have therefore comparedexpression of metchnikowin in different mutantbackgrounds with that of the antibacterial pep-tides diptericin and cecropin A1, and with that ofthe antifungal peptide drosomycin. For that pur-pose, Northern blots representing the populationsof mRNAs from wild-type, Tl-de®cient, imd, cact-de®cient and TlD mutant ¯ies were repeatedlydehybridized and successively probed with metch-nikowin, drosomycin, diptericin and cecropin A1probes. Figure 3 shows that regulation of metchni-kowin gene expression in immune challenged Tl-de®cient mutants is similar to that of diptericin,whereas in imd mutants it is similar to that of dro-somycin. On the contrary, the regulation of cecro-pin A1 gene expression upon immune challenge issimilar to that of drosomycin in Tl-de®cientmutants, and to that of diptericin in imd mutants.In addition, metchnikowin is expressed at a higherlevel than diptericin and cecropin A in unchal-

lenged cact-de®cient and TlD mutants, albeit at alower level than drosomycin (Figure 3).

To summarize these results, we propose that themetchnikowin gene is unique in that it can beregulated by the IMD pathway in the absence of afunctional TOLL pathway, and by the TOLL path-way in imd mutants. Thus, the particularity ofmetchnikowin, which has both antibacterial andantifungal activity, extends to regulation of itsexpression. The rest of the paper describes the iso-lation and organization of the promoter of thegene encoding this peptide.

Cloning of the metchnikowin gene

A genomic library of Drosophila melanogaster wasscreened using metchnikowin cDNA as a probe.Ten hybridization-positive clones were isolated.A 9 kb SalI fragment, which hybridized to themetchnikowin cDNA, was subcloned into the vec-tor pTZ18R for restriction mapping. Subsequently,a 2.5 kb SalI-XhoI fragment was subcloned in M13for sequencing (Figure 4(a)). This region consists of1.5 kb of 50 ¯anking sequences, followed by a 52-codon open reading frame (ORF) corresponding tothe metchnikowin precursor peptide. This ORFcomprises a putative signal peptide and the 26amino acid residue long mature metchnikowin(Figure 4(b)). No additional sequences were foundin the transcribed region of the metchnikowingenomic clone in comparison to the cDNA clone(Levashina et al., 1995), indicating that the geneencoding metchnikowin is intron-less. The locationof the cap site was determined by primer extension(data not shown). It is located 29 nucleotidesupstream of the translation initiation codon(Figure 4(b)). The TATA box (TATAAAAG) islocated between positions ÿ31 and ÿ24 withregard to the transcription start site (�1), and puta-tive polyadenylation consensus sequences arefound 45 and 58 nucleotides downstream of thetranslation stop codon (Figure 4(b)).

As stated in the Introduction, several studiesaimed at functional characterization of the promo-ters of the genes coding for Drosophila antimicro-bial peptides have highlighted the importance ofnucleotide sequence motifs related to NF-kB DNAbinding sites (referred to as Rel sites hereafter)(reviewed by Hoffmann & Reichhart, 1997). Com-puter analysis revealed three such nucleotidesequence motifs in metchnikowin gene upstreamsequences. They are located on either strand ofDNA, at positions ÿ638 (upper strand), ÿ176(lower strand), and ÿ63 (lower strand)(Figure 4(b)). The most distal Rel site (at positionÿ638) resembles two Rel sites found in the promo-ter of the drosomycin gene (50-GGGTTTTTCA-30and 50-GGGTTTTTAC-30; L. Michaut et al., unpub-lished results). Interestingly, these two drosomycinRel sites are located more than 500 base-pairsupstream from the start site, in contrast to the Relsites in the genes encoding antibacterial peptides,which are usually found close to the coding region,

Figure 3. Comparison of drosomycin, metchnikowin, diptericin and cecropin A1 expression patterns in wild-type andmutant Drosophila adults. Northern blot with total RNA extracted from unchallenged (uc) or bacteria-challenged (sixhours after challenge) wild-type (wt) or mutant ¯ies was successively hybridized with probes corresponding to thefollowing cDNAs: drosomycin, metchnikowin, diptericin, cecropin A1 and rp49. The signals were quanti®ed with aBio-Imager system. In each experiment the values for the immune peptides were normalized with the correspondingvalue of the rp49 signal. The level of expression of each antimicrobial peptide gene in wild-type adults after bacterialchallenge was taken arbitrarily as 100 and the results are presented as relative activities. The data represent themeans � standard deviation of four to nine independent Northern blotting experiments. Other indications: wt, Ore-gon R; Tlÿ, Tl632/Tl1-RXA; imd, imd/imd; cactÿ, cactA2/cactA2; TlD, Tl10B/ � .


between positions ÿ40 and ÿ200 (Charlet et al.,1996; Dimarcq et al., 1994; Engstrom et al., 1993).Our analysis also revealed other consensus nucleo-tide sequence motifs in the metchnikowin promo-ter, corresponding to mammalian NF-IL6 responseelements (Hocke et al., 1992), GATA sites (Orkin,1992), and interferon consensus response elements(Georgel et al., 1995) (Figure 4(c)).

In summary, sequencing of 1.5 kb of DNAlocated upstream of the start site allowed theidenti®cation of three Rel sites. Additional studiesare needed to establish the functionality of thesesequence motifs and their role in the control ofmetchnikowin gene expression. In order to showin vivo that this 1.5 kb DNA fragment contains themetchnikowin promoter, we have fused it to areporter gene, and have analyzed its function intransgenic Drosophila.

Analysis of expression of a metchnikowin-GFPfusion gene

In our ®rst attempt to functionally characterizethe metchnikowin promoter, we have constructeda reporter gene containing 1.5 kb of metchnikowinupstream sequences fused to the jelly®sh Aequoreavictoria green ¯uorescent protein (GFP) cDNA;

790 bp of 30 ¯anking region of the drosomycingene were used to signal transcription terminationand polyadenylation. The advantage of this repor-ter gene is that it allows direct imaging of the ¯u-orescent gene product in living cells without theneed for prolonged and lethal histochemical stain-ing procedures. Unlike the product of the metchni-kowin gene, which is targeted by a signal sequencefor secretion in the hemolymph, the GFP expressedfrom these constructs remains within the cellswhere it is synthesized. The transposon containingthis reporter construct was injected into embryos,and 19 independent insertions were obtained andcharacterized. GFP expression was monitoredqualitatively using a ¯uorescence stereomicro-scope, or quantitatively using a ¯uorimeter.

Third instar larvae were pricked with a bacteria-soaked needle, and examined 12 hours later underthe ¯uorescence microscope. A strong ¯uorescencecould be seen in the fat body of larvae 16 hoursafter immune challenge (compare (a) and (b) inFigure 5). Similar results were obtained with all 19independent lines. When adults were challengedby septic injury, a strong ¯uorescence developed inthe fat body 24 hours later (data not shown). Thesedata are consistent with in situ hybridization exper-iments, which detected abundant metchnikowin

Figure 4 (legend opposite)


Figure 5. Inducible GFP ex-pression in transgenic larvae con-taining the GFP cDNA underthe control of 1.5 kb of metchni-kowin gene upstream sequences.(a) Non-immunized transgeniclarva. (b) Transgenic larva 16 hoursafter immune challenge with bac-teria. (c) Constitutive ¯uorescencein the fat body of an unchallengedtransgenic larva carrying one copyof the Tl10b gain-of-function Tollallele. (d) Fluorescent circulatinghemocytes (arrows) in a transgeniclarva 16 hours after bacterial chal-lenge. Larvae were illuminatedwith a wavelength of 495 nm, andvisualized using a ¯uorescencestereomicroscope.


transcripts in the fat body of immune challengedadults (Levashina et al., 1995). In addition, whenthird instar larvae where pricked with a bacteria-soaked needle, ¯uorescent hemocytes could beseen through the cuticle 12 hours later (Figure 5(d)),but only in some larvae (�10%). This discrepancybetween larvae most probably re¯ects the intensityof the immune response, with only the strongestresponding larvae reaching the GFP protein con-centration threshold required to detect ¯uorescencebefore puparation. Inducible expression of thegenes encoding antimicrobial peptides in larvalhemocytes has been documented for diptericin(Meister et al., 1994) and cecropin (Samakovlis et al.,1990).

In order to compare regulation of GFPexpression with that of the resident metchnikowingene, we performed quantitative measurements.First we compared the kinetics of RNA synthesisinduction on the resident metchnikowin gene andthe GFP reporter gene by Northern blot analysis.RNA was extracted from transgenic ¯ies before orat various time points after pricking with a bac-teria-soaked needle, and analyzed by Northernblotting using metchnikowin and GFP probes. Asshown in Figure 6(a), GFP was induced uponimmune challenge with kinetics similar to metchni-

Figure 4. Organization of the metchnikowin gene. (a) SchXhoI genomic fragment comprising the metchnikowin cDNthe arrow indicates the 50 to 30 orientation of the transcribed(bp) is shown below. (b) Sequence of the metchnikowin genis shown (cDNA) together with the amino acid sequence (obold type. Asterisks indicate the putative sites of cleavage opolyadenylation sites. The start site (arrow) and the putativtive Rel sites (underlined) located on either strand. (c) Schopen reading frame is represented by a hatched box, the aof sequence motifs identical to Interferon Consensus Reselements (NF-IL6 RE), or GATA sites (GATA) are shown.

kowin: RNA transcripts could be detected threehours after stimulation, and reached a plateau after14 hours. Expression levels started to slowlydecrease after 24 hours. We also examined GFPexpression at the protein level by Western blotting,and found that the protein started to be detectablesix hours after stimulation, and accumulated until72 hours after stimulation, before levelling off(Figure 6(b)). This accumulation re¯ects the highstability of the GFP (Cubitt et al., 1995). We nextmeasured the intensity of ¯uorescence in immune-challenged ¯ies. Fluorescence could be detected asearly as six hours after induction, and reached amaximum at 72 hours, in good agreement with theWestern blot data (Figure 7(a)). These experimentswere performed on a representative transgenicstrain. Quanti®cation of ¯uorescence intensity forfour other independent lines is presented inFigure 7(a), and shows only moderate variabilityfrom line to line, corresponding to positionaleffects. These data indicate that 1.5 kb of metchni-kowin gene 50 ¯anking sequences contain all theelements necessary to confer full inducibility on areporter gene upon bacterial challenge in trans-genic ¯ies.

As discussed above, metchnikowin is expressedin the absence of immune stimulus in Toll gain-of-

ematic representation and restriction map of a 2.5 kb SalI-A. The coding region is marked by a hatched box (ORF),

sequence. The size of the genomic fragment in base-pairse. The extent of the longest cDNA (Levashina et al., 1995)

ne letter code). The mature peptide sequence is marked inf a signal peptide. Broken lines correspond to the putativee TATA box (boxed) are marked as well as the three puta-ematic organization of the metchnikowin promoter. The

rrow indicates the transcriptional start point. The locationponse Elements (ICRE), Rel sites (Rel), NF-IL6 response

Figure 6. Inducibility of GFP expression in transgenicDrosophila upon immune challenge. (a) Northern blotanalysis: total RNA was extracted from transgenic Dro-sophila either untreated (0 hour) or at various timepoints post-immunization with a mixture of Gram-posi-tive and Gram-negative bacteria, and analyzed byNorthern blot. The blot was ®rst probed with the metch-nikowin cDNA, and after dehybridization successivelywith the GFP and rp49 cDNAs. (b) Western blot anal-ysis: protein extracts were prepared from transgenicDrosophila (line 1) untreated (0 hour) or at various timespost-immune challenge and analyzed by Western blot-ting using an anti-GFP antibody.


function mutants. In order to check whether the1.5 kb of upstream sequences from the metchniko-win gene fused to the GFP cDNA are suf®cient tomimick regulation of the endogenous metchniko-win gene by the TOLL pathway, we crossed twoindependent transgenic lines containing the metch-nikowin-GFP reporter construct with a strain car-rying a gain-of-function Tl allele. As shown inFigure 5(c), the resulting larvae exhibited ¯uor-escence in the fat body in the absence of immunechallenge. The fact that our reporter construct con-tains 30 ¯anking sequences from the drosomycingene raises the possibility that TOLL might induceGFP expression through these sequences. To rulethis out, similar experiments were performed withtransgenic ¯ies in which the diptericin or drosocinpromoter was placed upstream of the GFP reportercassette containing the same drosomycin termin-ator sequences. When these reporter genes wereput in TlD context, no ¯uorescence could bedetected in the larval fat body (S. Ohresser et al.,unpublished results), thus implying that the targetsequences for TOLL signaling are in the metchni-kowin upstream regulatory sequences and not inthe drosomycin terminator.

Recent studies in our laboratory have shownthat natural infection of adult Drosophila with theentomopathogenic fungus Beauveria bassiana resultsin the selective induction of the genes encoding thepeptides with antifungal activity: drosomycin andmetchnikowin (Lemaitre et al., 1997). Transgenic

¯ies containing the metchnikowin promoter-GFPfusion gene, as well as transgenic ¯ies containing adrosomycin promoter-GFP (Ferrandon et al., 1998)or a drosocin-GFP (S. Ohresser et al., unpublishedresults) fusion gene, were covered with B. bassianaspores, and placed at 29�C. Fluorescence intensitywas measured 72 hours later. A strong ¯uor-escence could be detected in challenged transgenic¯ies containing the drosomycin or metchnikowinpromoter fused to the GFP coding sequences. Incontrast, no induction of ¯uorescence followingfungal infection could be observed in transgenic¯ies expressing GFP under the control of the dro-socin promoter (Figure 7(b)).

Altogether, these data strongly suggest that1.5 kb of upstream sequences from the metchniko-win gene are suf®cient to confer proper inducibilityon a reporter gene in transgenic Drosophila uponimmune challenge, or in TlD context. Further anal-ysis of these sequences in vivo and in vitro shouldhelp us to understand how the immune responseis triggered by the TOLL pathway and the imdgene product.

Discussion

Metchnikowin gene expression can beregulated by either the imd gene product orthe TOLL pathway

Here we describe the regulation of metchniko-win gene expression in wild-type and mutant Dro-sophila, and provide evidence that this gene can beregulated independently by the TOLL or the IMDpathway. Analysis of the expression pattern ofantimicrobial peptide genes in mutants defectiveeither for the TOLL pathway or the imd geneenables us to divide them into four groups. The®rst category of genes is exempli®ed by the genefor drosomycin, which appears to be regulatedmainly by the TOLL pathway. The second categoryof genes, to which the genes encoding diptericinand drosocin belong, appears to be regulatedmainly by the imd gene product. The third categoryof genes encompasses the cecropin A, attacin anddefensin genes, which require both the TOLL path-way and IMD for full inducibility upon immunechallenge (Lemaitre et al., 1996). The fourth cat-egory, which we describe here, is represented todate only by the gene encoding metchnikowin.Inducibility of this gene by immune challenge isnot affected in spz-, Tl-, tub- or pll-de®cientmutants, and is only slightly reduced in imdmutants. However, this gene is no longer immune-inducible in strains in which both pathways areinactivated. Thus, the TOLL pathway is to someextent suf®cient for the induction of metchnikowinexpression in imd mutants, while the IMD pathwayappears suf®cient to achieve this goal in Tl-de®cient mutants. In keeping with this interpret-ation, we observed that the mutations causing con-stitutive activation of the TOLL pathway (TlD andcactÿ) reproducibly resulted in low but signi®cant

Figure 7. Inducibility of GFP ¯uor-escence in transgenic Drosophilaupon immune challenge. (a) Inten-sity of GFP ¯uorescence in proteinextracts prepared from transgenicDrosophila untreated (0 hour) or atvarious time points after challengewith bacteria. The level of ¯uor-escence 72 hours after immunechallenge is compared for ®ve inde-pendent transgenic lines. (b) Induc-tion of GFP expression by naturalinfection with B. bassiana. Trans-genic ¯ies expressing GFP underthe control of the drosocin, metch-nikowin or drosomycin promoters(four groups of ten ¯ies for eachreporter construct) were leftuntreated (0 hour) or covered withspores of B. bassiana. Proteinextracts were prepared 72 hourslater, and ¯uorescence emissionat 507 nm was measured. Datarepresent means � standard devia-tions.


expression of metchnikowin in the absence of anyimmune challenge (see Figures 1 and 3). As forthe genes of the third category, both the TOLLand the IMD pathways participate in metchniko-win regulation, but in this case the two pathwaysregulate gene expression independently of oneanother.

The existence of two distinct regulatory path-ways controlling the expression of antimicrobialgenes points to the existence of distinct recognitionmechanisms. Recent data indicate that the Droso-phila immune system can discriminate among var-ious classes of microorganisms (Lemaitre et al.,1997). When different categories of microorganismswere injected into ¯ies, Gram-negative bacteriawere found to preferentially induce antibacterialgenes, while drosomycin was strongly induced byfungi and to a lesser extent by Gram-positivebacteria. Most interestingly, metchnikowin geneexpression was found to be strongly inducible byboth bacteria and fungi (Lemaitre et al., 1997). Inaddition, when ¯ies are naturally infected with theentomopathogenic fungus B. bassiana, in theabsence of pricking and the non-speci®c effects ofinjury, only drosomycin and, to a lesser extent,metchnikowin are induced, while the genes encod-ing antibacterial peptides remain silent (Lemaitreet al., 1997; and Figure 7(b)). It is therefore nowestablished that Drosophila can discriminatebetween microorganisms of fungal or bacterial ori-gin, and, at least in the case of fungal pathogens,mount an appropriate response by selectively turn-ing on the genes encoding peptides with antifungal

activity (e.g. drosomycin, metchnikowin). Thesedata suggest that the two pathways that participatein the regulation of metchnikowin expression canbe turned on independently, by the recognition ofspeci®c microorganisms, and agree well with whatwe know from the function of metchnikowin, sincethis peptide is active against both bacteria andfungi.

Altogether, these experiments indicate that thecomplex interplay between two pathways allowsfor the selective induction of subsets of genesduring the immune response in Drosophila, andsuggest that this speci®city results from differencesin the organization of the promoters of the genesencoding the antimicrobial peptides.

Organization of the metchnikowin promoter

We have cloned and sequenced 1.5 kb ofupstream sequences from the metchnikowin gene.In order to determine if these sequences are necess-ary and suf®cient for metchnikowin geneexpression, transgenic strains were constructed, inwhich the reporter gene encoding GFP was placedunder the control of this DNA fragment. Analysisof the resulting ¯ies and larvae indicates that wehave cloned the metchnikowin promoter: ®rst, the1.5 kb fragment confers inducibility to the GFPgene upon immune challenge; second, the kineticsof inducibility of GFP expression parallel those ofmetchnikowin mRNA expression with a peak ofexpression at 14 hours; third, in larvae carrying thegain-of-function mutation Tl10b, which constitu-


tively activates the TOLL pathway, the GFP isexpressed in the fat body in the absence of immunechallenge; fourth, natural infection by B. bassianainduces GFP expression in the fat body of trans-genic ¯ies. We therefore conclude that this DNAfragment contains the regulatory elements necess-ary for the control of metchnikowin expression inresponse to immune challenge or stimulation bythe TOLL pathway.

Sequence analysis of the metchnikowin promoterrevealed several features in common with the pro-moters of insect genes coding for other antimicro-bial peptides, and of mammalian genes expressedduring the acute phase response (Hoffmann &Reichhart, 1997). The most striking feature is thepresence of sequence motifs recognized by Relfamily members (Rel sites). These sites are presentin most mammalian genes involved in in¯amma-tory responses, and bind the inducible transactiva-tor NF-kB (Siebenlist et al., 1994). These sites havealso been described in all the promoters of anti-microbial peptide genes characterized to date inDrosophila. Three Rel sites are present in themetchnikowin promoter at positions ÿ63, ÿ176and ÿ638. Interestingly, one of these sites islocated relatively far upstream from the cap site,whereas for most antimicrobial peptide genes Relsites are located in the near upstream region. It isworth noting here that, in the case of the droso-mycin promoter, the Rel sites that are likely to befunctionally important are located between ÿ500and ÿ800 (L. Michaut et al., unpublished results).To date, the strongest evidence that Rel proteinsare involved in the control of antimicrobial geneexpression in Drosophila comes from the resultsobtained with cact-de®cient mutants. Indeed,CACTUS/IkB proteins retain Rel family membersin the cytoplasm (Siebenlist et al., 1994); hence theexpression of drosomycin (Lemaitre et al., 1996)and metchnikowin (this paper) in the absence ofimmune challenge in cactus-de®cient mutantssuggests that a Rel protein (possibly the DOR-SAL-related factor DIF (Ip et al., 1993), or DIF inassociation with DORSAL, both of which areexpressed in the fat body) translocates to thenucleus to activate the genes encoding these twopeptides. The fact that the antibacterial peptides(e.g. diptericin, drosocin, attacin, cecropin A1 ordefensin) are not expressed in such cact-de®cientmutants (Lemaitre et al., 1996) suggests a role fora Rel protein that does not interact with CACTUS,such as RELISH (Dushay et al., 1996). Alterna-tively, DORSAL or DIF might need an additionalchallenge-dependent modi®cation to becomeactive on these promoters.

In addition to Rel sites, the metchnikowin pro-moter contains sequence motifs homologous to cis-acting regulatory elements binding the mammaliantranscription factors GATA (Orkin, 1992), NF-IL6(Hocke et al., 1992), as well as several GGAAANNsequence motifs present in the interferon consensusresponse elements of mammalian interferon-induced genes (Georgel et al., 1995). In mammals,

cooperative interaction of NF-kB with multipletranscription factors is well documented (Siebenlistet al., 1994). The few studies that have dissectedthe organization of immune-inducible promoters inDrosophila reveal similar features. Footprinting andgel-shift experiments performed on the diptericinpromoter have shown that the proximal upstreamregion of the gene contains multiple functionalregulatory sequence elements (Georgel et al., 1993;Kappler et al., 1993). The functional importance ofthese elements was demonstrated by analysis oftransgenic strains of Drosophila bearing mutationsin these sequence motifs (Meister et al., 1994). Inter-estingly, the most proximal of the regulatoryregion was shown to contain two motifs overlap-ping with the Rel sites, and similar in sequence toNF-IL6 response elements and a core motif recog-nized by interferon regulatory factor 1 (IRF-1). Thececropin A1 gene promoter, which has also beenstudied in detail, similarly contains conservedsequence motifs adjacent to the Rel site (Engstromet al., 1993; Tryselius et al., 1992). Gel shift exper-iments and analysis of mutations in transienttransfection assays have recently shown that oneof these motifs, AGATAA, is functionally import-ant; this motif is identical to the consensussequence WGATAR, recognized in mammalianpromoters by transcriptional activators of theGATA family (Kadalayil et al., 1997). In themetchnikowin promoter, the most distal Rel siteoverlaps with a GGAAANN motif, and is adja-cent to an NF-IL6 response element, a situationreminiscent of the organization of the proximalelement in the diptericin promoter (Georgel et al.,1993; Meister et al., 1994). Another Rel site, atposition ÿ176, is located 11 bp downstream of aGATA site, a situation reminiscent of the cecropinpromoter (Kadalayil et al., 1997). A GATA site isalso present 8 bp upstream of a Rel site in thedrosocin gene promoter (Charlet et al., 1996).Finally, the third, most proximal, Rel site is notin the vicinity of any of the response elementsdiscussed above. Functional characterization ofthese sequence motifs, and puri®cation of theproteins that recognize them, should enable us tounderstand how the antimicrobial genes promo-ters are regulated.

The picture that is emerging from the study ofthe organization of the promoters of immune-inducible peptides in Drosophila therefore suggestsa situation in which Rel proteins combine withother DNA binding factors, the nature of whichdetermines the selection of speci®c Rel sites. Acti-vation of different subsets of Rel proteins anddifferent accessory factors by speci®c microorgan-isms might explain the selective induction ofsome genes during the immune response in Dro-sophila. It will be of interest to determine whetherdifferent Rel sites are responsible for induction ofmetchnikowin expression by the TOLL and IMDpathways, or if both pathways converge on asingle Rel site, and activate different accessoryfactors.


Materials and Methods

Drosophila stocks

The lines used in this study have been described(Lemaitre et al., 1995, 1996). In order to demonstrate theinducibility of the metchnikowin promoter by the TOLLpathway, transgenic females containing the GFP geneunder the control of the metchnikowin promoter werecrossed with males heterozygous for the Tl10b allele.Crosses were performed at 25�C, and third instar larvaewere collected. wÿ ¯ies were used as wild-type ¯ies, andwere also used as recipients for transformation.

Immunization procedures

Bacterial challenge was performed by pricking adults(3 to 5-day-old) with a thin needle previously dippedinto a concentrated bacterial culture of Escherichia coliand Micrococcus luteus. Flies were then kept at 25�C, andwere snap frozen in liquid nitrogen after different timeintervals. For fungal infection, ¯ies were anesthetizedand covered with spores of B. bassiana (strain 80.2). Theywere then placed at 29�C, and collected for ¯uorescencequanti®cation 48 hours later.

RNA analysis

Total RNA extraction and Northern blotting exper-iments were performed as described (Lemaitre et al.,1995). The following probes were used: cecropin A1cDNA (Kylsten et al., 1990); diptericin cDNA (Wickeret al., 1990); drosomycin cDNA (Fehlbaum et al., 1994);metchnikowin cDNA (Levashina et al., 1995); and rp49cDNA (O'Connell & Rosbach, 1984). Hybridization sig-nals were quanti®ed using a Bio-Imager system, and thesignals obtained for immune genes RNA were normal-ized with the corresponding value of the ubiquitouslyexpressed ribosomal protein rp49 RNA signal.

Isolation and sequencing of the metchnikowin gene

A total of 100,000 plaque-forming units of an ampli-®ed Drosophila Oregon R genomic library constructed inl DASH II vector (gift from Dr J.-A. Lepesant) werescreened with a 32P-labeled 280 bp cDNA probe (insertA1; Levashina et al., 1995). Hybridization of nitrocellu-lose ®lters and washing were performed as described(Dimarcq et al., 1994). Nine hybridization-positive pla-ques were obtained and shown by restriction mappingto correspond to four overlapping genomic fragments.A 9 kb SalI fragment was cloned in the pTZ18R phage-mid by standard methods (Sambrook et al., 1989). Afterrestriction mapping a 2.5 kb SalI-XhoI fragment wassubcloned into a M13mp18 vector. Sequencing wasperformed by the dideoxy method of Sanger et al. (1977)using a Pharmacia Sequencing kit. The products wereseparated on a polyacrylamide gel (6% acrylamide, 8 Murea). Routine computer analysis of the sequencing datawas facilitated by DNASTARTM software for a Macin-tosh microcomputer. The putative cis-regulatory motifswere determined using the MatInspector V2.1 program(Copyright''1997Transfac-Team).

Construction of the metchnikowin reporter gene andisolation of transformed fly lines

The F64L mutation (replacement of the phenylalanineresidue in position 65 by a leucine) in GFP was intro-duced by PCR-directed mutagenesis (Higuchi et al.,1988) using the high ®delity Vent polymerase (New Eng-land Biolabs). The plasmid pJM705 (Ferrandon et al.,1998), which contains the S65 T version of GFP, servedas a template. The PCR fragment was sequenced to con-trol the ®delity of the reaction. The modi®ed GFP cDNAwas fused to a 790 bp fragment of drosomycin gene 30¯anking region (Ferrandon et al., 1998). A fragment con-taining metchnikowin gene sequences from positionÿ1497 (SalI restriction site) to position �32 was sub-sequently fused upstream of the GFP sequences. Thecomplete metchnikowin 50 sequences±GFP±drosomycin30 ¯anking region cassette was then subcloned into thetransformation vector pCasper, before injection intoembryos. The helper plasmid ppq2-3 was co-injectedwith the pCasper vector containing the fusion gene intowÿ embryos, and transformants were recovered as G1w� ¯ies. Crosses with ¯ies carrying appropriate bal-ancers were performed to establish stable heterozygousor homozygous lines, as well as to determine thechromosome carrying the insertion. Nine insertions werefound on the second chromosome, and ten on the thirdchromosome. The established lines were maintained at25�C on a standard cornmeal medium.

Quantification of GFP expression

For Western blot analysis, 20 mg of protein extract pre-pared from transgenic female ¯ies at various time pointsafter immunization were resuspended in Laemmli buffer,and analyzed by SDS-PAGE on a 10% acrylamide gel(Laemmli, 1970). After migration, the proteins were elec-troblotted onto a nitrocellulose ®lter, which was probed(overnight, 4�C) with a polyclonal rabbit anti-GFP anti-body (Clontech), at a 1:2000 dilution in Tris-bufferedsaline (TBS) supplemented with 0.3% low fat dry milkand 0.05% Tween 20. A horseradish peroxidase±anti-rabbit conjugate (Amersham Life Science) was used as asecond antibody at a dilution of 1:10,000, and wasincubated with the ®lter for one hour at 37�C. The GFPwas detected using the enhanced chemiluminescencetechnique (Amersham Life Science).

Fluorescent GFP in live ¯ies was visualized using aphotoautomat (wild MPS48/52, Leica) equiped for ¯uor-escence microscopy. For quantitative measurements ofGFP activity, ten female adults were homogenized inChal®e buffer (Chal®e et al., 1994), and extracts werecentrifuged for ten minutes at 10,000 g to remove debris.Fluorescence in the supernatant was measured with aspectro¯uorimeter (Jobin Yvon ¯uoromax-2). The emis-sion spectra were measured from 495 mm to 560 nm,with a ®xed excitation wavelength of 495 nm. The peakof emission was observed at 507 nm.

Acknowledgments

E. A. Levashina and S. Ohresser made equal contri-butions to this work. We thank Dominique Ferrandonfor injection into embryos as well as for his manyadvice; Jean-Marc Reichhart for stimulating discussions;Jules A. Hoffmann for continued support and interest,


as well as critical reading of the manuscript. The techni-cal expertise of Reine Klock and Marie Eve Moritz isgratefully acknowledged. This work was supported inpart by a grant from the Human Frontiers ScienceProgram.

The nucleotide sequence for the metchnikowin genehas been submitted to GenBank and will be accessibleunder the accession number AF030959.

References

Asling, B., Dushay, M. & Hultmark, D. (1995).Identi®cation of early genes in the Drosophilaimmune response by PCR-based differencialdisplay: the Attacin A gene and the evolution ofattacin-like proteins. Insect Biochem. Mol. Biol. 25,511±518.

Bulet, P., Dimarcq, J., Hetru, C., Lagueux, M., Charlet,M., Hegy, G., Van Dorsselaer, A. & Hoffmann, J.(1993). A novel inducible antibacterial peptide ofDrosophila carries an O-glycosylated substitution.J. Biol. Chem. 268, 14893±14897.

Cao, Z., Henzel, W. & Gao, X. (1996). IRAK: a kinaseassociated with the interleukin-1 receptor. Science,271, 1128±1131.

Chal®e, M., Tu, Y., Euskirchen, G., Ward, W. & Prasher,D. (1994). Green ¯uorescent protein as a marker forgene expression. Science, 263, 802±805.

Charlet, M., Lagueux, M., Reichhart, J., Hoffmann, D.,Braun, A. & Meister, M. (1996). Cloning of the geneencoding the antibacterial peptide drosocininvolved in Drosophila immunity. Expression stu-dies during the immune response. Eur. J. Biochem.241, 699±706.

Cubitt, A., Heim, R., Adams, S., Boyd, A., Gross, L. &Tsien, R. (1995). Understanding, improving andusing green ¯uorescent proteins. Trends Biochem.Sci. 20, 448±455.

Dimarcq, J., Hoffmann, D., Meister, M., Bulet, P., Lanot,R., Reichhart, J. & Hoffmann, J. (1994).Characterization and transcriptional pro®les of aDrosophila gene encoding an insect defensin.A study in insect immunity. Eur. J. Biochem. 221,201±209.

Dushay, M., Asling, B. & Hultmark, D. (1996). Originsof immunity: Relish, a compound Rel-like gene inthe antibacterial defense of Drosophila. Proc. NatlAcad. Sci. USA, 93, 10343±10347.

Engstrom, Y., Kadayalil, L., Sun, S., Samakovlis, C.,Hultmark, D. & Faye, I. (1993). kB-like motifs regu-late the induction of immune genes in Drosophila.J. Mol. Biol. 232, 327±333.

Fehlbaum, P., Bulet, P., Michaut, L., Lagueux, M.,Broeckaert, W., Hetru, C. & Hoffmann, J. (1994).Insect immunity: septic injury of Drosophilainduces the synthesis of a potent antifungal peptidewith sequence homology to plant antifungalpeptides. J. Biol. Chem. 269, 33159±33163.

Ferrandon, D., Jung, A., Criqui, M., Lemaitre, B.,Uttenweiler-Joseph, S., Michaut, L., Reichhart, J. &Hoffmann, J. (1998). A drosomycin-GFP reportertransgene reveals a local immune response in Droso-phila that is not dependent on the Toll pathway.EMBO J, 17, 1217±1227.

Gay, N. & Keith, F. (1991). Drosophila Toll and IL-1receptor. Nature, 351, 355±356.

Georgel, P., Meister, M., Kappler, C., Lemaitre, B.,Reichhart, J. & Hoffmann, J. (1993). Insect immu-

nity: the diptericin promoter contains multiple func-tional regulatory sequences homologous tomammalian acute-phase response elements. Bio-chem. Biophys. Res. Commun. 197, 508±517.

Georgel, P., Kappler, C., Langley, E., Gross, I., Nicolas,E., Reichhart, J. & Hoffmann, J. (1995). Drosophilaimmunity. A sequence homologous to mammalianinterferon consensus response element enhances theactivity of the diptericin promoter. Nucl. Acids Res.23, 1140±1145.

Higuchi, R., Krummel, B. & Saiki, R. K. (1988). A gen-eral method of in vitro preparation and speci®cmutagenesis of DNA fragments: study of proteinand DNA interactions. Nucl. Acids Res. 16, 7351±7364.

Hocke, G., Barry, D. & Fey, G. (1992). Synergistic actionof interleukin-6 and glucocorticoids is mediated bythe interleukin-6 response element of the rat alpha2 macroglobulin gene. Mol. Cell Biol. 12, 2282±2294.

Hoffmann, J. & Reichhart, J. (1997). Drosophilaimmunity. Trends Cell Biol. 7, 309±316.

Ip, Y., Reach, M., Enstrom, Y., Kadalayil, L., Cai, H.,Gonzalez-Crespo, S., Tatei, K. & Levine, M. (1993).Dif, a dorsal-related gene that mediates an immuneresponse in Drosophila. Cell, 75, 753±763.

Kadalayil, L., Petersen, U. & Engstrom, Y. (1997).Adjacent GATA and kB-like motifs regulate theexpression of a Drosophila immune gene. Nucl.Acids Res. 25, 1233±1239.

Kappler, C., Meister, M., Lagueux, M., Gateff, E.,Hoffmann, J. & Reichhart, J. (1993). Insect immu-nity. Two 17 bp repeats nesting a kB-relatedsequence confer inducibility to the diptericin geneand bind a polypeptide in bacteria-challengedDrosophila. EMBO J. 12, 1561±1568.

Kylsten, P., Samakovlis, C. & Hultmark, D. (1990). Thececropin locus in Drosophila: a compact gene clus-ter involved in the response to infection. EMBO J. 9,217±224.

Laemmli, U. K. (1970). Cleavage of structural proteinsduring the assembly of the head of bacteriophageT4. Nature, 227, 680±685.

Lemaitre, B., Kromer-Metzger, E., Michaut, L., Nicolas,E., Meister, M., Georgel, P., Reichhart, J. &Hoffmann, J. (1995). A recessive mutation, immunede®ciency (imd), de®nes two distinct control path-ways in the Drosophila host defense. Proc. NatlAcad. Sci. USA, 92, 9365±9469.

Lemaitre, B., Nicolas, E., Michaut, L., Reichhart, J. &Hoffmann, J. (1996). The dorsoventral regulatorygene cassette spaÈ tzle/Toll/cactus controls thepotent antifungal response in Drosophila adults.Cell, 86, 973±983.

Lemaitre, B., Reichhart, J. & Hoffmann, J. (1997).Drosophila host defense: differential induction ofantimicrobial peptide genes after infection by var-ious classes of microorganisms. Proc. Natl Acad. Sci.USA, 94, 14614±14619.

Levashina, E., Ohresser, S., Bulet, P., Reichhart, J.,Hetru, C. & Hoffmann, J. (1995). Metchnikowin, anovel immune-inducible proline-rich peptide fromDrosophila with antibacterial and antifungalproperties. Eur. J. Biochem. 233, 694±700.

Meister, M., Braun, A., Kappler, C., Reichhart, J. M. &Hoffmann, J. A. (1994). Insect immunity. A trans-genic analysis in Drosophila de®nes several func-tional domains in the diptericin promoter. EMBO J.13, 5958±5966.


Morisato, D. & Anderson, K. (1995). Signaling pathwaysthat establish the dorsal-ventral pattern of thedrosophila embryo. Annu. Rev. Genet. 29, 371±399.

O'Connell, P. & Rosbach, M. (1984). Sequence, structureand codon preference of the Drosophila ribosomalprotein 49 gene. Nucl. Acids Res. 12, 5495±5513.

Orkin, S. (1992). GATA-binding transcription factors inhematopoietic cells. Blood, 80, 575±581.

Samakovlis, C., Kimbrell, D., Kylsten, P., Engstrom, A. &Hultmark, D. (1990). The immune response inDrosophila: pattern of cecropin expression andbiological activity. EMBO J. 9, 2969±2976.

Sambrook, J., Fritsch, E. & Maniatis, T. (1989). MolecularCloning: A Laboratory Manual, 2nd edit., Cold SpringHarbor Laboratory Press, Plain View, NY.

Sanger, F., Nicklen, S. & Coulson, A. R. (1977). DNAsequencing with chain-terminating inhibitors. Proc.Natl Acad. Sci. USA, 74, 5463±5467.

Siebenlist, U., Franzoso, G. & Brown, K. (1994).Structure, regulation and function of NF-kB. Annu.Rev. Cell Biol. 10, 405±455.

Tryselius, Y., Samakovlis, C., Kimbrell, D. & Hultmark,D. (1992). CecC, a cecropin gene expressed duringmetamorphosis in Drosophila pupae. Eur. J. Bio-chem. 204, 395±399.

Wicker, C., Reichhart, J., Hoffmann, D., Hultmark, D.,Samakovlis, C. & Hoffmann, J. (1990). Insect immu-nity: characterization of a Drosophila cDNA encod-ing a new member of the diptericin promoter. J. Biol.Chem. 265, 22493±22498.

Edited by M. Yaniv

(Received 28 October 1997; received in revised form 5 February 1998; accepted 6 February 1998)

Documents

Two Distinct Pathways can Control Expression of the Gene ...€¦ · Two Distinct Pathways can Control Expression of the Gene Encoding the Drosophila Antimicrobial Peptide Metchnikowin