Use of Metarhizium anisopliae Chitinase Genes for Genotyping and Virulence Characterization

Virulence is the primary factor used for selection of entomopathogenic fungi (EPF) for development as biopesticides. To understand the genetic mechanisms underlying differences in virulence of fungal isolates on various arthropod pests, we compared the chitinase genes, chi2 and chi4, of 8 isolates of Metarhizium anisopliae. The clustering of the isolates showed various groups depending on their virulence. However, the analysis of their chitinase DNA sequences chi2 and chi4 did not reveal major divergences. Although their protein translates have been implicated in fungal virulence, the predicted protein structure of chi2 was identical for all isolates. Despite the critical role of chitin digestion in fungal infection, we conclude that chi2 and chi4 genes cannot serve as molecular markers to characterize observed variations in virulence among M. anisopliae isolates as previously suggested. Nevertheless, processes controlling the efficient upregulation of chitinase expression might be responsible for different virulence characteristics. Further studies using comparative “in vitro” chitin digestion techniques would be more appropriate to compare the quality and the quantity of chitinase production between fungal isolates.


Introduction
Entomopathogenic fungi (EPF) based products are being developed for the control of insect pests in agricultural systems [1][2][3]. Entomopathogenic fungi infect their hosts through the cuticle and do not need to be ingested like bacteria, viruses, and protozoa [4]. During the process of infection, EPF secrete chitinase to digest insect cuticle [5][6][7][8].
The entomopathogenic fungus M. anisopliae produces at least six types of chitinases [9,15,19]. However, the respective role of these proteins in the process of pathogenicity as well as their contribution to virulence on arthropod pests has not been clearly elucidated [20,21]. Nonetheless, chitinase chi2 gene isolated from M. anisopliae var. anisopliae strain E6 has been reported to be responsible for virulence in the genus M. anisopliae [22]. Overexpression of chi2 constructs showed higher efficiency in host killing, while the absence of the same chitinase reduced fungal infection efficiency [20]. Recent studies on differential expression of chitinase genes in vitro and in vivo established the role of substrate differences in the process of pathogenesis [23]. To understand the role of chitinase genes underlying differences in virulence between fungal isolates, we compared the virulence against various arthropod pests and characterized the chitinase genes of 8 isolates of M. anisopliae from the International Centre of Insect Physiology and Ecology (icipe)'s Arthropod Germplasm Centre.

Statistical Analysis.
Records on the performance of each of the M. anisopliae isolates were obtained from icipe archives. Virulence data (percentage mortality and lethal time to mortality (LT)) of each isolate was used in the cluster analysis. For each pest, a virulence factor for each isolate was determined by using the average mortality value of the total percentage mortality of all isolates. The same procedure was used for LT values. Data were then subjected to a -mean clustering model to determine the difference in their virulence. The centroid, which is the mean vector of each cluster, was used to define cluster membership of each isolates. The within-groups inertia was used as a criterion to define cluster compactness.
The number of clusters was fixed at 4 ( = 4) according to the major taxonomic groups that were considered in this study. Missing values were estimated. A factor analysis based on Spearman correlation (Quartimax rotation) was used to determine the relation between the isolates. The number of iterations performed was 11 and the overall iterations were 200. All statistical analyses were performed using XLSTAT-Pro (Version 7.2, 2003, Addinsoft, Inc., Brooklyn, NY, USA); the significance level was set at = 0.05.

Sequence Diversity and Phylogeny.
Chitinase nucleotide sequences were edited and aligned to remove ambiguous base calls before they were translated into proteins using Geneious [25]. A search to identify protein sequences similar to chi2 and chi4 was performed using tBLASTx algorithm of NCBI GenBank. Geneious Software was used to estimate phylogeny with the neighbour-joining, minimum evolution, or maximum parsimony method. A dendrogram was constructed using Molecular Evolutionary Genetics Analysis (MEGA) software version 4.0 with 10,000 bootstrap replicates [26]. All methods returned trees with similar topology and approximate bootstrap values; therefore only the neighborjoining tree is presented. Percentage homology among similar chitinases to chi2 and chi4 were computed using MEGA software.
The 3D structure was predicted using Swiss PdB Viewer, v 4.0.1 (http://www.expasy.org/spdbv/). The conserved residues of the Carbohydrate Insertion Domain (CID) [27] were identified through multiple sequence alignment with the characterized chitinase genes.

Clustering of Insects Based on the Virulence Data of the M.
anisopliae Isolates. The grouping of the arthropods into clusters based on virulence data showed that cluster1 (inertia = 0.0) includes fruit-fly species C. rosa and C. capitata; clus-ter2 (inertia = 8.2) comprises ornamental pests such as F. occidentalis, M. sjostedti, L. huidobrensis, and T. urticae; cluster 3 (inertia = 9.3) includes five hosts belonging to various taxonomic groups: C. cosyra, P. duboscqi, T. evansi, M. michaelseni, and C. puncticollis. Cluster 4 with the highest inertia of 73.1% corresponded to the highest diversity of arthropods pests (Table 3).

Relation between the M. anisopliae Isolates.
Factor analysis using correlation matrix showed various levels of similarity between the isolates based on their performances on the 11 insect pests. ICIPE7 has similarities with ICIPE20, ICIPE69, and ICIPE78 whereas ICIPE20 is closely related to ICIPE41  and ICIPE62; ICIPE 30 is only related to ICIPE7 and ICIPE78, although the correlations were not strong; ICIPE41 was strongly related to ICIPE62, ICIPE63, and ICIPE69; ICIPE62 and ICIPE63 have closed virulence patterns as IMI330189. ICIPE20 and ICIPE41 also are related to IMI330189. There were also similarities in virulence patterns between ICIPE78, ICIPE20, and ICIPE41 (Table 4).

Analysis of Chitinase2 Gene Sequence.
Comparison of the chi2 nucleotide sequences from all selected M. anisopliae isolates originating from three different parts of Africa showed no differences in the open reading frames composed of 229 amino acid residues. However, when compared with the similar chitinase sequences retrieved from NCBI database, there were differences in amino acid composition (Figure 2).

Homology
Modeling of Chitinase2. The Swiss-Pdb Viewer (http://www.expasy.org/spdbv/) server was used to predict the 3D structure of chi2. The conserved residues of the Carbohydrate Insertion Domain (CID, Y × R and V × I) were present in all selected M. anisopliae isolates that exhibited no differences in their coding regions. In M. anisopliae var. acridum the "Y × R" motif is replaced by "Y × K" (Figure 4).

Analysis of Chitinase4 Gene Sequence.
All M. anisopliae var. anisopliae isolates had identical chi4 nucleotide sequences. After the editing process to remove the ambiguous base calls a BLAST analysis using chi4 sequence on NCBI GenBank database revealed highest amino acid identities to M. anisopliae var. anisopliae M34412, ARSEF7524, and M. anisopliae var. acridum IMI330189 ( Figure 5).

Discussion
The clustering analyses based on virulence data on various taxonomic groups revealed differences between the icipe's isolates. Cluster1 comprises fruit flies C. rosa and C. capitata, against which ICIPE 20 is most virulent, although other isolates have been reported to be pathogenic [28]. ICIPE20       also fits in Cluster2, which comprises L. huidobrensis, F. occidentalis, T. urticae, and M. sjostedti against which it has been reported to be pathogenic [29][30][31][32]. Cluster2 also accommodates ICIPE 69 which has been reported to be virulent against thrips [2,29,33] and is currently commercialised for the control of insect pests of horticulture in Africa [32]. ICIPE7 which has been reported to be most virulent isolate against T. urticae [30] can also be considered in that cluster. Cluster3, on the other hand, includes flies, termites, and mites and therefore involves a larger number of isolates. Previous records on their virulence indicate that ICIPE7, ICIPE20, ICIPE30, ICIPE78, and ICIPE62 could be included in that cluster because of their virulence on T. urticae, M. michaelseni, and C. puncticollis [30,34,35]. ICIPE69 has been reported to be the least virulent isolate against M. michaelseni [36] and thus cannot be considered in that cluster. This may explain the absence of thrips species in cluster3. Cluster4 comprises 11 arthropod pests, suggesting that each of the isolates is virulent to some extent to each of these pests and their related species. For instance, ICIPE30 has been used for the control of tsetse fly Glossina spp. [30,37]. ICIPE7, which is virulent against mites T. urticae and T. evansi [30,38], is also indicated for the control of the tick species Rhipicephalus appendiculatus and R. pulchellus [39,40], both belonging to Acari group. ICIPE78, known to be the most virulent isolate for the control of T. evansi [37,41], is closely related to ICIPE7.
All the M. anisopliae isolates used in this study showed the same chi2 and the same chi4 protein structure despite the fact that they originated from different localities in Africa. Only IMI330189 (M. anisopliae var. acridum) which originated from Niger had a nonsynonymous substitution in the chi4 sequence. The analysis of the common predicted structure of the chitinase showed folding patterns and conserved amino acids of the Carbohydrate Insertion Domain (CID) described in many fungal species [9,27,44] including NCBI outgroup sequences.
Chitinase gene chi2 was reported to be mainly responsible for M. anisopliae virulence [20,23]. The present molecular results suggest either that chitinase genes are differentially regulated (i.e., different expression levels) in different isolates or that there are other parameters that affect the process of infection. Regarding the first hypothesis, chi2 gene has been reported to be upregulated by chitin (which serves as a carbon source to the fungus) in conditions of fungus autolysis, and is downregulated by glucose [25]. Chitin composition of insect cuticle can affect chitinase production level [23,45], which would justify the difference in virulence. Since insect pests have special cuticle compositions, the virulence of EPF may vary accordingly, even between life stages [23]. In that regard, Moritz [46] reported that adult thrips and larvae have different cuticle structures, which could explain, in part, the difference in susceptibility to EPF between arthropod pests [32,[47][48][49]. Posttranscriptional regulation of chitinase genes [50] may also account for the observed virulence difference in our isolates. This needs to be further investigated by comparing chitinase gene expression of isolates with different virulence patterns. Additionally, other relevant factors, such as conidiation and toxin production genes, that affect fungal virulence need to be considered as well. Niassy et al. [32] observed that ICIPE 69 produced more conidia than ICIPE 20 and ICIPE 7 and was virulent to larvae of F. occidentalis. Fang et al. [24] demonstrated that gene disruption of a conidiation-associated gene (cag8) in M. anisopliae resulted in the lack of conidiation on agar plates and on infected insects reduced mycelial growth and decreased virulence, suggesting the involvement of cag8 in the modulation of conidiation, virulence, and hydrophobin synthesis in M. anisopliae. All these gene-regulatory processes need to be considered when developing molecular techniques for genotyping EPF.