The Nag1 N-acetylglucosaminidase of Trichoderma atroviride is essential for chitinase induction by chitin and of major relevance to biocontrol

RESEARCH ARTICLE

The Nag1 N -acetylglucosaminidase of Trichoderma atroviride is essentialfor chitinase induction by chitin and of major relevance to biocontrol

Received: 29 January 2003 / Revised: 2 April 2003 / Accepted: 8 April 2003 / Published online: 14 May 2003� Springer-Verlag 2003

Abstract The nag1 gene of the mycoparasitic fungusTrichoderma atroviride encodes a 73-kDa N-acetyl-b-D-glucosaminidase, which is secreted into the medium andpartially bound to the cell wall. To elucidate the role ofthis enzyme in chitinase induction and biocontrol, anag1-disruption mutant was prepared. It displayed only4% of the original N-acetyl-b-D-glucosaminidase activ-ity, indicating that the nag1 gene product accounts forthe majority of this activity in T. atroviride. The nag1-disruption strain was indistinguishable from the parentstrain in growth and morphology, but exhibited delayedautolysis. Northern analysis showed that colloidal chitindisruption does not induce ech42 gene transcription inthe nag1-disruption strain. Enzyme activities capable ofhydrolysing p-nitrophenyl-N,N¢-diacetylchitobiosideand p-nitrophenyl-N,N¢-diacetylchitotriose were alsoabsent from the nag1-disruption strain under the sameconditions. Retransformation of the T. atroviride nag1-disruption strain with the nag1 gene essentially led to theparent-type behaviour in all these experiments. How-ever, addition of N-acetyl-b-D-glucosaminidase to themedium of the nag1-disruption strain did not rescue the

mutant phenotype. The disruption-nag1 strain showed30% reduced ability to protect beans against infectionby Rhizoctonia solani and Sclerotinia sclerotiorum. Thedata indicate that nag1 is essential for triggering chitin-ase gene expression in T. atroviride and that its func-tional impairment reduces biocontrol by T. atroviride bya significant extent.

Keywords Trichoderma Æ Biocontrol Æ InductionChitinase regulation

Introduction

Several species of the imperfect soil fungus Trichoderma(e.g. T. harzianum, T. atroviride) are potent mycopara-sites against several economically important plant path-ogenic fungi and are therefore used as biocontrol agents.Biocontrol of plant pathogens is an attractive alternativeto the strong dependence of modern agriculture on fun-gicides, which may cause environmental pollution anddevelopment of resistant strains. Unfortunately, theapplication of biocontrol strains is still not easy to pre-dict and thus not yet able to compete with chemicalfungicides (cf. Harman and Bjorkman 1998; Hjeljord andTronsmo 1998). Consequently, knowledge of the bio-chemical events which determine mycoparasitism andtheir regulation would offer strategies to improve thereliability of Trichoderma as a biocontrol agent.

Several genes, which are specifically induced undermycoparasitic conditions, have been cloned from myco-parasitic Trichoderma spp (Chet et al. 1998). Chitinase-encoding genes are prominent among these. So far, threeendochitinases (ech42, chit33, chit36) and two exoch-itinases [nag1(=exc1), exc2] have been cloned andcharacterized (Lorito 1998; Kubicek et al. 2001; Viterboet al. 2001; Kim et al. 2002). However, a critical inves-tigation of the role of these chitinases in biocontrol has sofar only been performed with ech42 and the results arecontroversial (Woo et al. 1998; Baek et al. 1999; Carsolio

Curr Genet (2003) 43: 289–295DOI 10.1007/s00294-003-0399-y

Kurt Brunner Æ Clemens K. Peterbauer Æ Robert L. Mach

Matteo Lorito Æ Susanne Zeilinger Æ Christian P. Kubicek

Communicated by U. KuckK. Brunner and C.K. Peterbauer contributed equally to the presentwork and are listed in alphabetical order

K. Brunner Æ R.L. Mach Æ S. Zeilinger Æ C.P. Kubicek (&)Abteilung fur Angewandte Biochemie und Gentechnik,Institut fur Verfahrenstechnik,Umwelttechnik und Technische Biowissenschaften,TU Wien, Getreidemarkt 9/1665, 1060 Vienna, AustriaE-mail: [email protected]

M. LoritoDipartimento di Arboricoltura,Botanica e Patologia Vegetale,Sezione di Patologia Vegetale,Universita degli Studi di Napoli ¢¢Federico II¢¢ and Centro diStudio CNR per le Tecniche di Lotta Biologica (CETELOBI),Via Universita 100, 80050 Portici (NA), Italy

C.K. PeterbauerInstitut fur Lebensmitteltechnologie,Universitat fur Bodenkultur, Muthgasse 11,1190 Vienna, Austria

et al. 1999). In addition, the mechanism by which thesechitinases are induced under mycoparasitic conditions isunknown. Present evidence suggests that a low-molecu-lar-weight, diffusible factor is released from the host andtriggers chitinase gene expression in T. atroviride (Corteset al. 1998; Zeilinger et al. 1999). Kullnig et al. (2001)provided evidence that this factor is produced by an asyet unidentified chitinase of Trichoderma. Inbar and Chet(1995), using a biomimetic system, detected a N-acetyl-b-D-glucosaminidase as the first chitinase formed whenTrichoderma comes into contact with its host.

Although two N-acetyl-b-D-glucosaminidases arepresent inT. harzianum andT. virens (Draborg et al. 1995;Kim et al. 2002), only a single gene has been detected inT. atroviride P1 (Peterbauer et al. 1996), thus offering aconvenient source for studying the hypothesis that theN-acetyl-b-D-glucosaminidase activity of Trichoderma maybe triggering chitinase gene expression. Using an nag1-disruption strain, we show here that Nag1 accounts forthe majority of secreted N-acetyl-b-D-glucosaminidase ofT. atroviride and that its deletion strongly affects thegeneral induction of the chitinolytic enzyme system buthas only a moderate effect on the biocontrol of Rhizoc-tonia solani and Sclerotinia sclerotiorum.

Materials and methods

Strains

T. atroviride strain P1 (T. harzianum ATCC 74058; Kullnig et al.2001) was used throughout this study and was maintained on po-tato dextrose agar (PDA; Merck, Darmstadt, Germany). Botrytiscinerea strain 26 and S. sclerotiorum strain 1450 were used asmodel pathogens and were obtained from the collection of theInstitute of Plant Pathology, Universita degli Studi di NapoliFederico II (Naples, Italy). Escherichia coli JM109 (Yanisch-Perronet al. 1985) was the host for plasmid amplification.

Cultivation conditions

T. atroviride and recombinant strains prepared from it were grownin liquid synthetic medium (SM, pH 5.4) containing (per litre): 2 gKH2PO4, 1.4 g (NH4)2SO4, 0.3 g CaCl2Æ2H2O, 0.3 g MgSO4Æ7H2O,0.6 g urea, 0.01 g FeSO4Æ7H2O, 0.0028 g ZnSO4Æ2H2O, 0.0032 gCoCl2Æ6H2O. Glucose or glycerol (10 g/l, except when statedotherwise) were used as carbon sources. For induction experiments,T. atroviride was pre-cultivated for 36 h in SM containing glycerolas carbon source, harvested by filtration through sterile Miracloth(Calbiochem, La Jolla, Calif.), washed with sterile tap water andtransferred to 100-ml Erlenmeyer flasks containing 25 ml of SMmedium and 1.5% (w/v) of colloidal chitin. Mycelia were harvestedafter different times of incubation, as indicated at the respectiveresults.

B. cinerea and S. sclerotiorum were cultivated on malt extract/peptone broth/agar and potato dextrose broth/PDA, respectively.

DNA and RNA manipulation

Details of methods for DNA and RNA extraction were reported byPeterbauer et al. (2002a). Briefly, DNA was isolated by the CTABmethod (Ausubel et al. 1990) and plasmid DNA by means of amidiprep kit (Qiagen, Chatsworth, Calif.). Northern blots of total

RNA were done using BiodyneB Nylon membranes (Pall Corp.,Ann Arbor, Mich.). All restriction enzymes and DNA-modifyingenzymes were obtained from Promega Corp. (Madison, Wis.).DNA probes for hybridizations were radioactively labelled byrandom priming (Ausubel et al. 1990). Other molecular techniqueswere performed according to standard protocols (Sambrook et al.1989; Ausubel et al. 1990).

Plasmids and plasmid constructions

A 2,696-bp fragment of the T. atroviride nag1 locus (Peterbauer etal. 1996), spanning the genomic copy from –289 (MluI site, filled-inwith Klenow fragment) to +2,407 (XbaI site) was inserted betweenthe Acc65I site (blunt-ended, but reconstituted by ligation with thefilled-in MluI site) and the XbaI site of pBluescript SK+ (Strata-gene Corp. ) to yield pCN1. To construct a disruption cassette fornag1, a 3.9-kb SalI/XbaI fragment of the Aspergillus nidulans amdS(acetamidase-encoding) gene was released from plasmid p3SR2(Kelly and Hynes 1985). The sticky ends were filled-in with Klenowfragment and the resulting DNA-fragment was then ligated into theEcoRV site of pCN1. The resulting cassette consisted of 1.1 kb and1.7 kb of nag1, respectively, flanking the amdS selectable marker.To obtain a linear DNA fragment for gene disruption, this cassettewas cut out with Acc65I and XbaI.

Fungal transformation

The 6.7-kb, linearized nag1 disruption cassette was introduced intoT. atroviride by protoplast transformation, as described by Peter-bauer et al. (2002b). Mitotically stable transformants were obtainedby at least three sequential transfers of conidia from non-selectiveto selective media. Transformation of the nag1-disruption strainT. atroviride P1ND1 with pCN1 was performed by co-transfor-mation with pHATa (carrying the E. coli hph gene, encodinghygromycin B-phosphotransferase), as described by Peterbaueret al. (2002a).

Preparation of N-acetyl-b-D-glucosaminidase from T. atroviride

N-acetyl-b-D-glucosaminidase from T. atroviride was prepared byinducing mycelia, pre-grown on glycerol, with N-acetyl-b-D-gluco-samine as described by Mach et al. (1999). SDS-PAGE showed thatthe 73-kDa N-acetyl-b-D-glucosaminidase, identified by the use ofan antibody (Peterbauer et al. 1996), accounted for more than 95%of total protein in culture filtrates harvested after 4 h of incubation.They were then dialysed, concentrated 20-fold, lyophilized and kepton ice until use.

Enzyme assays

Extracellular culture filtrates were filtered through a 0.22-lmmembrane (Millipore, Bradford, Mass.), dialysed against distilledwater for 24 h at 4 �C, concentrated about 20-fold with polyeth-ylene glycol (8,000 MW; Fluka Biochemika, Buchs, Switzerland),and stored at –20 �C with 20% glycerol until use. Enzymatic assayswere performed as described by Harman et al. (1993) and Lorito etal. (1994), using p-nitrophenyl N-acetyl-b-D-glucosamine, p-nitro-phenyl b-D-N,N¢¢-diacetylchitobiose and p-nitrophenyl b-D-N¢¢,N¢¢¢¢-triacetylchitotriose (all from Sigma, St. Louis, Mo.) assubstrates.

Biocontrol assays

Tests of inhibition of B. cinerea spore germination were performedin ELISA plates containing 20 ll of culture filtrate of T. atrovirideP1 and the recombinant strains thereof and 2,000 Botrytis conidia

290

(prepared by growing B. cinerea for 8 days on agar plates con-taining 3%malt extract, 1% peptone) in 100 ll of 5 mM potassiumphosphate buffer (pH 6.7) per well. The plates were incubated at25 �C for 5 h and the number of germinated spores was countedunder an inverted microscope (Lorito et al. 1993).

Plate confrontation assays were performed as described byLorito et al. (1996).

For in vivo biocontrol tests in a greenhouse, bean seeds werecoated with a 10% suspension of Pelgel (Liphatech, Milwaukee,Wis.) in 20 mM potassium phosphate buffer augmented with20 mM glucose, using 1 ml of a Trichoderma conidial suspension(108 conidia/ml) per 10 g of seeds. Pathogen-infested soil wasprepared by adding 1.5–3.0 g fungal biomass (wet weight ofR. solani, S. sclerotiorum) to 1 l of sterile soil. After 2 days, theinfested soil was diluted 1:4 with sterile soil and used for biocontrolassays, as described above. Trichoderma-coated seeds were planted4 cm deep in the pathogen-infested soil. The number and height ofsurviving plants was evaluated for a period of up to 4 weeks.Control experiments included: (a) coated seeds in soil withoutpathogen and Trichoderma, (b) coated seeds in soil with pathogenbut without Trichoderma, (c) coated seeds in soil with T. atrovirideP1 only and (d) coated seeds in soil with T. atroviride P1ND1 only.For each of these controls, 14 seeds were planted.

Chemical assays

Total soluble carbohydrate was determined by the phenol–sulfuricacid method (Dubois et al. 1956).

Results

Construction of a nag1-disruption strainof T. atroviride

To prepare a nag1-disruption strain of T. atroviride, weconstructed a plasmid vector in which the A. nidulansamdS gene was inserted into the coding region of nag1(Fig. 1A). The resulting disruption cassette was trans-formed as a linear fragment into T. atroviride andmitotically stable progeny were tested for integrationinto the nag1 locus by Southern hybridization, usinggenomic DNA digested with SalI and a fragment of thenag1-coding region as probe. Insertion of the 3.9 kbamdS marker gene increased the size of the SalI nag1fragment from approximately 4 kb to 8 kb (Fig. 1B);and this was proven to be the case in some transfor-mants, as given in Fig. 1B for strain P1ND1. Furtherand in order to confirm that disruption of the genomiclocus did eliminate the formation of any nag1 transcript,we performed Northern analyses on mycelia in whichnag1 gene expression had been induced by N-acetylglu-cosamine (Mach et al. 1999). While the nag1 transcriptwas abundant in the parent strain P1, it was shown to beabsent in strain P1ND1 (Fig. 1C). Thus, the nag1-dis-ruption strain P1ND1 is defective in Nag1 formation.

In order to confirm this finding is not due to anyeffect other than the lack of nag1, the intact nag1 genewas retransformed into strain P1ND1. Three strainswere obtained (P1ND1r1, P1ND1r2, P1ND1r3), whichcontained one, two and one copy of nag1, respectively,integrated into the T. atroviride P1ND1 genome atunknown, heterologous loci. All three strains exhibited

Nag1-activity in the same range as or even higher than(P1ND1r2) the wild-type strain P1. We therefore con-clude that the loss of Nag1 activity in the nag1-disrup-tion mutant is due to a loss-of-function of the nag1 gene.

nag1 deletion affects autolysis of T. atroviride

Because chitinases are also involved in fungal cell-wallturn-over (Reyes et al. 1989a, 1989b; Sahai and Mano-cha 1993; White et al. 2002), we first tested whether thenag1-disruption strain would have any general physio-logical defects, such as altered growth rates, sporulationor morphological abnormalities, which could affectsubsequent experiments. However, strain P1ND1behaved essentially as the parent strain in all theseproperties (data not shown), therefore showing nag1 isnot involved in growth, morphology and sporulation ofT. atroviride.

In contrast, a major difference was observed inthe autolysing capability of T. atroviride P1 and the

Fig. 1A–C Construction and identification of a nag1-disruptionmutant of Trichoderma atroviride. A Schematic representation ofthe strategy for disruption of the T. atroviride nag1 genomic locus.A 3.9-kb blunt-ended SalI/XbaI fragment from p3SR2 containingthe Aspergillus nidulans amdS gene was inserted into the EcoRV siteof a 2.7-kb MluI/XbaI fragment of nag1 in plasmid pCN1(Peterbauer et al. 1996), resulting in plasmid pND1. B Identifica-tion of T. atroviride P1ND1 as nag1-disruption mutant by Southernhybridization, using SalI-digested genomic DNA and a 1.9-kb SalI/XbaI fragment of nag1 as a probe (the location of the probe is givenby the grey horizontal bar over the nag1 gene in A. A double cross-over event results in replacement of the intact genomic copy withthe disrupted construct, thereby increasing the size of the indicatedSalI fragment to 8 kb. C Evidence for lack of nag1 genetranscription in T. atroviride P1ND1 by Northern hybridizationof total RNA. Mycelia, pre-grown on 1% glycerol, were placedonto 0.5% N-acetyl-glucosamine and harvested after 2 h. The same1.9-kb nag1 fragment as above was used as a probe. The lower panelshows the ethidium bromide-stained rRNA bands

291

nag1-disruption strain. When both strains were pre-cultivated on glycerol and then placed onto colloidalchitin, strain P1 started to partially autolyse after 16 h,as indicated by the continuous release of soluble car-bohydrates into the culture fluid (4.7 lg/ml after 16 h,20 lg/ml after 26 h). In contrast, a significantly loweramount (0.1 lg/ml after 16 h, 0.14 lg/ml after 26 h) ofcarbohydrate was released by the nag1-disruption strain.The retransformed strain P1ND1r3 essentially behavedlike the wild-type strain in these experiments, thus vali-dating that the effect was due to nag1 loss-of-function.

nag1 encodes the major, extracellularN-acetylglucosaminidase activity of T. atroviride

Having a nag1-disruption strain in hand, we then testedwhether the nag1 gene product accounts for all theN-acetyl-b-D-glucosaminidase activity of T. atroviride.To this end, we induced the fungus with N-acetylglu-cosamine (Mach et al. 1999) and measured N-acetyl-b-D-glucosaminidase activity in both the culture supernatantand a mycelial suspension of the fungus. Table 1 showsthat the activity in the culture supernatant was reducedto <1% of the control. However, a residual activity ofabout 10% of that of the wild-type strain could bemeasured in the cell walls of the fungus, indicating thatT. atroviride has a second N-acetyl-b-D-glucosaminidaseactivity, which accounts for 4% of the total extracellularactivity in the nag1-disruption-strain and which is al-most exclusively located within the cell wall.

nag1 deletion impairs ech42 gene expression andformation of other chitinase activities by T. atroviride

In order to investigate the hypothesis of this work, i.e.whether the absence of nag1 would affect the formation

of other chitinases in T. atroviride, we incubated theparent strain P1 and the nag1-disruption strain P1ND1on colloidal chitin. Neither strain grows on chitin underthese conditions, but chitinase activities {measured bythe ability to hydrolyse the low-molecular-weightchromogenic substrates p-nitrophenyl-N,N¢-diacetylchi-tobioside [PNP-(GlcNAc)2] and p-nitrophenyl-N,N¢-di-acetylchitotriose [PNP-(GlcNAc)3]; Haran et al. 1995}form after 3 days (Mach et al. 1999). As shown inTable 2, no chitinase activities could be detected in thenag1-disruption strain, whereas chitinase activity wasclearly present in the parent strain P1 under theseconditions.

To test whether this lack of chitinase activity was dueto a lack of chitinase gene transcription, we performed aNorthern analysis of the endochitinase ech42 genetranscript. Figure 2 shows that, in contrast to the parentstrain P1, the ech42 transcript is virtually absent in thenag1-disruption strain, thus indicating that ech42 tran-scription cannot be triggered by these conditions in astrain lacking nag1. In order to confirm this finding isnot due to any effect other than the lack of the nag1 geneproduct, these experiments were repeated with the threerandomly chosen P1ND1 strains into which nag1 hadbeen retransformed. The result with only one of these(P1ND1r3), which contained one copy of nag1integrated in the T. atroviride P1ND1 genome at anunknown, heterologous locus, is shown in Fig. 2, butthose of the two others (containing one and two copies,respectively) were essentially consistent and demon-strated that retransformation led to a resumption ofN-acetyl-b-D-glucosaminidase activity, inducibility ofech42 gene transcription and resumption of otherchitinase activities. The effect observed in the nag1-dis-ruption mutant P1ND1 is thus specific for nag1.

Addition of exogenous Nag1 cannot rescuethe effect of the nag1 deletion on inductionof ech42 gene expression in T. atroviride

To further test our hypothesis, i.e. that the lack of Nag1enzyme activity is responsible for the effect of nag1deletion on the induction of chitinases in T. atroviride,we attempted to rescue the inducibility of chitinases inT. atroviride P1ND1. Thus, a >95% pure preparationof Nag1 (judged by SDS-PAGE; data not shown) wasprepared and added to cultures of T. atroviride P1ND1,pre-grown on glycerol and transferred to colloidal chitinin an amount to yield the same volumetric activity in theculture filtrate as that originally present in it when the

Table 1 Effect of nag1 gene disruption in Trichoderma atroviride onits extracellular N-acetyl-b-D-glucosaminidase activity. Cell-wallactivity was determined from the difference in the measurement ofactivity for whole mycelial broth and mycelium-free culturesupernatant. Data presented are means ±SD from three separateexperiments

Source Enzyme activity (units/ml)

T. atroviride P1 T. atroviride P1ND1

Total extracellular activity 500±52 22.0±3.9Culture filtrate 293±18 2.0±0.5Cell walls 207±15 20.0±4.0

Table 2 Effect of nag1 genedisruption in T. atroviride onextracellular chitinolyticenzyme activity. Data presentedare means ±SD from threeseparate experiments

Source Enzyme activity (units/ml)

T. atroviride P1 T. atroviride P1ND1

p-Nitrophenyl-N,N¢-diacetylchitobioside 65±4 1.0±0.3p-Nitrophenyl-N,N¢-diacetylchitotriose 40±2 0.5±0.2

292

glycerol cultures were harvested (0.04 units/ml). How-ever, this addition failed to rescue the phenotype of thenag1-disruption-mutant, as neither the formation ofchitinase enzyme activities nor the induction of expres-sion of ech42 was observed (Fig. 2). This was not due toa denaturation of Nag1 enzyme activity, because theculture filtrates clearly showed N-acetyl-b-D-glucosa-minidase activities. Also, this lack of effect was not dueto an inhibition of ech42 gene expression by someunknown components present in the Nag1 preparation,because addition of Nag1 to T. atroviride P1 incubatedwith colloidal chitin resulted in the expected ech42 genetranscription. Therefore, the effect of nag1 deletion onchitinase induction is not due to the loss of the extra-cellular Nag1 activity.

Effect of nag1 deletion on the biocontrol abilityof T. atroviride

In order to learn whether the nag1 deletion would affectthe biocontrol activity of T. atroviride, strain P1 and thenag1-disruption strain P1ND1 were first used in plate-confrontation assays against R. solani and S. sclerotio-rum. However, no differences in the antagonistic abilitiesof the two strains were observed (data not shown).

As results from in vitro biocontrol analyses weresometimes not concordant with those from in vivoexperiments (cf. Woo et al. 1998), the ability of theparent strain P1 and the nag1-disruption strain to pro-tect beans infected by R. solani and S. sclerotiorum wascompared (Table 3). In these experiments, the nag1-disruption strain was significantly less effective thanstrain P1, enabling plant survival at only 61% (R. solani)and 65% (S. sclerotiorum), respectively.

Discussion

Our results show that the protein encoded by the nag1gene has a major impact on the induction of T. atroviridechitinases by chitin, its deletion virtually blocking the

formation of other chitinases, including ech42. Thisfinding is of interest, as it offers new insights into themechanism of chitinase induction in Trichoderma. Pre-vious studies showed that chitinase gene expression issubject to an array of external signals, including coldshock, osmotic stress and carbon or nitrogen starvation(Mach et al. 1999; de laMercedes Dana et al. 2001) and issubject to induction by fungal cell walls and amorphouschitin (Carsolio et al. 1994; Garcia et al. 1994; Zeilingeret al. 1999; de la Mercedes Dana et al. 2001). However,while both ech42 and chit33 were induced by chitinoligomers in T. harzianum (de la Mercedes Dana et al.2001), no such induction was observed in T. atroviride(Mach et al. 1999). We previously interpreted thesefindings as lack of inducibility of T. atroviride ech42 bychitin-degradation products. However, the present datasuggest this may not be fully correct. Unless the Nag1protein itself is also a regulator of chitinase biosynthesis(for which at the moment no evidence is available), wehave to assume that its effect on chitinase gene inductionis due to its enzymatic activity, thereby being involved inthe formation of the physiological inducer. The fact thatchito-oligomers (N-acetyl-glucosamine or glucosamine),which would be the logical products of the action ofNag1 on chitin, do not induce ech42 expression (Mach etal. 1999) does not contradict this possibility, because therole of Nag1 in induction of chitinases may rather be tomodify them by its transglycosylating activity (Koga etal. 1991). This interpretation, however, is in conflict withearlier findings by us and others (Margolles-Clark et al.1996; Mach et al. 1999) that T. atroviride cannot grow oncolloidal chitin and that the expression of ech42 uponincubation with colloidal chitin under our conditions isdue to carbon starvation. In view of the present findings,this hypothesis cannot be maintained. Rather, the factthat the nag1-disruption strain has a reduced ability toautolyse suggests that ech42 gene expression may be dueto autolytic cell-wall oligomers of T. atroviride. Thefinding of an involvement of Nag1 in autolysis is also inagreement with the data by Reyes et al. (1989a), whodemonstrated that a N-acetyl-b-D-glucosaminidase ofA. nidulans is involved in autolytic cell wall degradation.Thus, if Nag1 forms a low-molecular-weight inducer ofech42, it is derived from its own cell wall rather than fromcolloidal chitin. The structure of oligochitosides thereby

Fig. 2 Effect of nag1 gene disruption in T. atroviride on inductionof ech42 gene expression, shown by Northern analysis of ech42gene expression upon incubation with glucose (Gluc) or colloidalchitin (Chit) for 72 h. Plusses (+) indicate the addition of0.04 units/ml (final concentration) of N-acetyl-b-D-glucosaminidaseto the medium at the start of incubation. P1ND1r3 is aretransformant of T. atroviride P1ND1 containing two copies ofthe nag1 locus. A 1.0-kb PstI fragment of ech42 (Mach et al. 1999)and a 1.9-kb Acc65I fragment of act1 were used as probes

Table 3 Antagonistic activity of T. atroviride P1 and the nag1-disruption mutant P1ND1 against Rhizoctonia solani and Scle-rotinia sclerotiorum infection of beans. Plant growth (height, cm)was used as a measure of vitality. Data are presented for a total offour independent experiments

Days Vitality/growth (cm)

R. solani S. sclerotiorum

P1 P1ND1 P1 P1ND1

19 6.9±1.3 4.3±0.6 14.2±7.8 6.2±4.331 15.7±2.3 10.8±0.5 20.0±1.3 12.8±7.341 23.7±1.5 17.8±3.8 23.4±4.2 15.5±8.2

293

released may be significantly different from thosereleased from colloidal chitin because chitin in the fungalcell wall is covalently linked via a peptide bridge tob-glucan (Wrathall and Tatum 1973). In support of thisassumption, we recently obtained evidence for a stronginduction of chitinases by chito-oligosaccharides whichwere enzymatically released from fungal cell walls(Zeilinger et al. 1999; Lorito et al., manuscript in prep-aration). In this context, we would also like to note thatan induction of chitinases by autolysis may also explaintheir induction by cold, osmotic and pH stresses, as allthese conditions also lead to immediate stops in thedelivery of precursors and energy for cell wall synthesisand thus may cause partial or transient autolysis. Thus, amore thorough investigation of autolysis in T. atroviridewould be worthwhile.

The fact that extracellular addition of purified Nag1did not restore the inducibility of chitinases in thenag1-disruption strain appears to contradict theseconclusions at first glance. However, one must bear inmind that autolysis occurs at the inner part of the cellwall (White et al. 2002) and the polymers of ascomy-cete cell walls are assembled in a layered structure(Griffiths 1994). The cell-wall-located Nag1 maytherefore release oligosaccharides different from thosepresent in the extracellular medium. While novelexperimental approaches need to be designed to testthis hypothesis, the genetic evidence clearly shows thatNag1 is strongly involved in chitinase induction bychitin in T. atroviride.

Only a single N-acetyl-b-D-glucosaminidase-encod-ing gene has been described so far from T. atroviride(Peterbauer et al. 1996). Here, however, the nag1-dis-ruption strain was shown to have a residual, mainlycell-wall-located N-acetyl-b-D-glucosaminidase activity,thus indicating the presence of a second enzyme, al-though this accounted for only 4% of the total extra-cellular activity. These data suggest that an orthologueof the gene encoding the second N-acetyl-b-D-glucosa-minidase, which has been cloned from T. harzianumand T. virens (Draborg et al. 1995; Kim et al. 2002), isalso present in T. atroviride. One might argue that the4% is an underestimate, because the deletion of nag1may also eliminate the inducibility of this gene. How-ever, we should like to note that the 4% refers tocultures induced by N-acetylglucosamine and at themoment we have no evidence and do not consider itlikely that nag1 deletion also affects induction by sol-uble compounds.

Another intriguing finding from this study wasthat—despite the almost complete absence of chitinaseactivity in the nag1-disruption strain—there was appar-ently no complete reduction in the biocontrol activity ofthis strain. Although a reduction in biocontrol activity by30% is considerable (cf. Woo et al. 1998; Baek et al.1999), it suggests that even a complete loss of chitinasesonly partially impairs biocontrol. One possible explana-tion for this would be that the nag1-disruption strainoverexpresses other biocontrol-relevant enzymes or

proteins and thus can largely compensate for the loss ofchitinase activities. This could be due either to thewithdrawal of trans-acting factors by strongly expressedpromoters, such as that of nag1, or to the saturation ofsecretory vesicles. In fact, a similar finding was made inT. reesei, where a cbh1 knock-out mutant overexpressedother cellulase genes (Suominen et al. 1993; Seiboth et al.1997). In support of this hypothesis, we observedelevated b-glucanase activities in the nag1-disruptionstrain (K. Brunner, unpublished data). In addition, itshould be stressed that we do not know whether the effectof nag1 deletion on chitinase gene expression, as found inthe laboratory experiments, may also occur in vivo dur-ing biocontrol as e.g. the loss of Nag1 may be compen-sated by the plant itself. In fact, Dal Soglio et al. (1998)reported that soybean seedlings, incubated together withR. solani and T. harzianum, showed high expression of aplant N-acetyl-b-D-glucosaminidase together with theT. harzianum endochitinase 42. Thus, although neces-sary for the induction of chitinase gene expression inT. atroviride, this effect of the nag1 deletion may notbecome relevant during biocontrol, as it could becompensated by the plant; and thus no conclusion can bedrawn as to the role of total chitinase activity forbiocontrol. In situ demonstration of expression of thechitinase genes from the nag1-disruption strain will benecessary to clarify this point.

Acknowledgements This work was supported by grants from theAustrian Science Foundation to C.K.P. (P-13170 MOB) and S.Z.(P-15483). M.L. and C.P.K. also recognize the support of the EU-funded TRICHOEST project (QLRT-2001-02032). M.L. recognizesthe grants FIRB 2002 [Genomica funzionale dell¢interazionetrapiante e micro-organismi (patogeni, antagonisti e simbiotici):determinanti coinvolti nella produzione agricola e protezionedell¢ambiente] and MIPAF (Risorse genetiche di organismi utili peril miglioramento di specie di interesse agrario e per un¢agricolturasostenibile). K.B.¢s stay in Naples was supported by a grant from theOAD (Scientific-technological collaboration between Austria andItaly) and by the Ausseninstitut of Technische Universitat Wien.S.Z. is the recipient of an APART fellowship (number 10764).

References

Ausubel FM, Brent R, Kingston RE, Moore DD, Seidman JG,Smith JA, Struhl K (1990) Current protocols in molecularbiology. Wiley-Interscience, New York

Baek JM, Howell CR, Kenerley CM (1999) The role of an extra-cellular chitinase from Trichoderma virens Gv29-8 in the bio-control of Rhizoctonia solani. Curr Genet 35:41–50

Carsolio C, Gutierrez A, Jimenez B, Van Montagu M, Herrera-Estrella A (1994) Characterization of ech42, a Trichodermaharzianum endochitinase gene expressed during mycoparasit-ism. Proc Natl Acad Sci USA 91:10903–10907

Carsolio C, Benhamou N, Haran S, Cortes C, Gutierrez A, Chet I,Herrera-Estrella A (1999) Role of the Trichoderma harzianumendochitinase gene, ech42, in mycoparasitism. Appl EnvironMicrobiol 65:929–935

Chet I, Benhamou N, Haran S (1998) Mycoparasitism and lyticenzymes. In: Harman GE, Kubicek CP (eds) Enzymes, bio-logical control and commercial application. (Trichoderma andGliocladium, vol 2) Taylor and Francis, London, pp 153–171

Cortes C, Gutierrez A, Olmedo V, Inbar J, Chet I, Herrera-EstrellaA (1998) The expression of genes involved in parasitism by

294

Trichoderma harzianum is triggered by a diffusible factor. MolGen Genet 260:218–225

Dal Soglio FK, Bertagnolli BL, Sinclair JB, Yu G-Y, EastburnDM (1998) Production of chitinolytic enzymes and endoglu-canase in the soybean rhizosphere in the presence of Tricho-derma harzianum and Rhizoctonia solani. Biol Control 12:111–117

Draborg H, Kaupinnen S, Dalboge H, Christgau S (1995) Molec-ular cloning and expression in Saccharomyces cerevisiae of twoexochitinases from Trichoderma harzianum. Biochem Mol BiolInt 36:781–791

Dubois M, Gilles KA, Hamilton JK, Rebers PA, Smith F (1956)Colorimetric method for determination of sugars and relatedsubstances. Anal Chem 28:350–356

Garcia I, Lora JM, Cruz J de la, Benitez T, Llobell A, Pintor-ToroA (1994) Cloning and characterization of a chitinase (chit42)cDNA from the mycoparasitic fungus Trichoderma harzianum.Curr Genet 27:83–89

Griffiths D (1994) Fungal physiology. Wiley, New YorkHaran S, Schickler H, Oppenheim A, Chet I (1995) New compo-

nents of the chitinolytic system of Trichoderma harzianum.Mycol Res 99:441–446

Harman GE, Bjorkman T (1998) Potential and existing uses ofTrichoderma and Gliocladium for plant disease control andplant growth enhancement. In: Harman GE, Kubicek CP (eds)Enzymes, biological control and commercial application.(Trichoderma and Gliocladium, vol 2) Taylor and Francis,London, pp 229–265

Harman GE, Hayes CK, Lorito M, Broadway RM, Di Pietro A,Tronsmo A (1993) Chitinolytic enzymes of Trichoderma har-zianum: purification of chitobiosidase and endochitinase. Phy-topathology 83:313–318

Hjeljord L, Tronsmo A (1998) Trichoderma and Gliocladium inbiological control: an overview. In: Harman GE, Kubicek CP(eds) Enzymes, biological control and commercial application.(Trichoderma and Gliocladium, vol 2) Taylor and Francis,London, pp 129–151

Inbar J, Chet I (1995) The role of recognition in the induction ofspecific chitinases during mycoparasitism by Trichoderma har-zianum. Microbiology 141:2823–2829

Kelly JM, Hynes MJ (1985) Transformation of Aspergillus niger bythe amdS gene of Aspergillus nidulans. EMBO J 4:475–479

Kim DJ, Baek JM, Uribe P, Kenerley CM, Cook DR (2002)Cloning and characterization of multiple glycosyl hydrolasegenes from Trichoderma virens. Curr Genet 40:374–384

Koga K, Iwamoto Y, Sakamoto H, Hatano K, Sano M, Kato I(1991) Purification and characterization of b-N-acetylhexosa-minidase from Trichoderma harzianum. Agric Biol Chem55:2817–2833

Kubicek CP, Mach RL, Peterbauer CK, Lorito M (2001) Tricho-derma chitinases: from genes to biocontrol. J Plant Pathol83:11–23

Kullnig CM, Krupica T, Woo SL, Mach RL, Rey M, Benitez T,Lorito M, Kubicek CP (2001) Confusion abounds over identityof Trichoderma biocontrol isolates. Mycol Res 105:769–772

Lorito M (1998) Chitinolytic enzymes ands their genes. In: HarmanGE, Kubicek CP (eds) Enzymes, biological control and com-mercial application. (Trichoderma and Gliocladium, vol 2)Taylor and Francis, London, pp 73–99

Lorito M, Harman GE, Hayes CK, Broadway RM, Tronsmo A,Woo SL, Di Pietro A (1993) Chitinolytic enzymes produced byTrichoderma harzianum: antifungal activity of purified en-dochitinase and chitobiosidase. Phytopathology 83:302–307

Lorito M, Hayes CK, Di Pietro A, Woo SL, Harman GE (1994)Purification, characterization and synergistic activity of a glu-can 1,3-b-glucosidase and N-acetyl-b-glucosaminidase fromTrichoderma harzianum. Phytopathology 84:398–405

Lorito M, Woo SL, D¢Ambrosio M, Harman GE, Hayes CK,Kubicek CP, Scala F (1996) Synergistic interaction between cellwall degrading enzymes and membrane affecting compounds.Mol Plant Microbe Interact 9:206–213

Mach RL, Peterbauer CK, Payer K, Jaksits S, Woo SL, ZeilingerS, Kullnig CM, Lorito M, Kubicek CP (1999) Expression oftwo major chitinase genes of Trichoderma atroviride (T. har-zianum P1) is triggered by different regulatory signals. ApplEnviron Microbiol 65:1858–1863

Margolles-Clark E, Harman GE, Penttila M (1996) Enhancedexpression of endochitinase in Trichoderma harzianum with thecbh1 promoter of Trichoderma reesei. Appl Environ Microbiol62:2152–2155

Mercedes Dana M de la, Limon CM, Meijas R, Mach RL, Pintor-Toro JA, Kubicek CP (2001) Regulation of chitinase 33 (chit33)gene expression in Trichoderma harzianum. Curr Genet 38:335–342

Peterbauer C, Lorito M, Hayes CK, Harman GE, Kubicek CP(1996) Molecular cloning and expression of nag1 (N-acetyl-b-D-glucosaminidase-encoding) gene from Trichoderma harzianumP1. Curr Genet 29:812–820

Peterbauer CK, Brunner K, Mach RL, Kubicek CP (2002a)Identification of the N-acetyl-D-glucosamine-inducible elementin the promoter of the Trichoderma atroviride nag1 geneencoding N-acetyl-glucosaminidase. Mol Genet Genomics267:162–170

Peterbauer CK, Litscher D, Kubicek CP (2002b) The Trichodermaatroviride seb1-(stress response element binding) gene encodesan AGGGG-binding protein which is involved in osmotic stressresponse. Mol Genet Genomics 268:223–231

Reyes F, Calatayud J, Vazquez C, Martinez MJ (1989a) Beta-N-acetylglucosaminidase from Aspergillus nidulans which degradeschitin oligomers during autolysis. FEMS Microbiol Lett 53:83–87

Reyes F, Calatayud J, Martinez MJ (1989b) Endochitinases fromAspergillus nidulans Beta-N-acetylglucosaminidase fromAspergillus nidulans implicated in the autolysis of the cell-wall.FEMS Microbiol Lett 51:119–124

Sahai AS, Manocha MS (1993) Chitinases of fungi and plants: theirinvolvement in morphogenesis and host-parasite interaction.FEMS Microbiol Rev 11:317–338

Sambrook J, Fritsch EF, Maniatis T (1989) Molecular cloning: alaboratory manual, 2nd edn. Cold Spring Harbor LaboratoryPress, Plainview, N.Y.

Seiboth B, Hakola S, Mach RL, Suominen PL, Kubicek CP (1997)Role of four major cellulases in triggering of cellulase geneexpression by cellulose in Trichoderma reesei. J Bacteriol179:5318–5320

Suominen PL, Mantyla AL, Karhunen T, Hakola S, Nevalainen H(1993) High frequency one-step gene replacement in Tricho-derma reesei. II. Effects of deletions of individual cellulasegenes. Mol Gen Genet 241:523–530

Viterbo A, Haran S, Friesem D, Ramot O, Chet I (2001) Anti-fungal activity of a novel endochitinase gene (chit36) fromTrichoderma harzianum Rifai TM. FEMS Microbiol Lett200:169–174

White S, McIntyre M, Berry DR, McNeil B (2002) The autolysis ofindustrial filamentous fungi. Crit Rev Biotechnol 22:1–14

Woo SL, Donzelli B, Scala F, Mach RL, Harman GE, KubicekCP, Del Sorbo G, Lorito M (1998) Disruption of ech42 (en-dochitinase-encoding) gene affects biocontrol activity inTrichoderma harzianum strain P1. Mol Plant Microbe Interact12:419–429

Wrathall CR, Tatum EL (1973) The peptides of the hyphal wall ofNeurospora crassa. J Gen Microbiol 78:139–153

Yanisch-Perron C, Vieira J, Messing J (1985) Improved M13 phagecloning vectors and host strains: nucleotide sequences of theM13mp18 and pUC19 vectors. Gene 33:103–119

Zeilinger S, Galhaup C, Payer K, Woo SL, Mach RL, Fekete C,Lorito M, Kubicek CP (1999) Chitinase gene expression duringmycoparasitic interaction of Trichoderma harzianum with itshost. Fungal Genet Biol 26:131–140

295