Tribolium castaneum and Drosophila melanogaster chitinases (Zhu et al. 2004; Zhang et al. 2011) were downloaded from GenBank (http://www.ncbi.nlm.nih.gov). and the sequences of Glyco_18 catalytic domain were used to search for similar genes in the silkworm genome sequence with tBLASTn search against database of B. mori genome draft sequences (http://www.ncbi.nlm.nih.gov/sutils/genom_ta ble.cgi?organism=insects, and http://silkworm.genomics.org.cn/). The obtained sequences were predicted using FGENESH+ to find the potential chitinase sequences. Determination of the chromosome location of B. mori chitinases was carried out with SilkMap program (http://silkworm.swu.edu.en/silksoft/silkmap.html).

Pfam (http://pfam.wustl.edu/) and SMART (http://smart.embl-heidelberg.de/) domain analysis programs were used to predict the domain architecture of potential chitinases. PEST sequence was predicted by ePESTfind program (http://emboss.bioinformatics.nl/cgibin/emboss/epestfind).

Multiple sequence alignment was performed using the ClustalW program (http://www.ebi.ac.uk/clustalw/). Previous studies have shown four highly conserved signature sequences in the amino acid sequences of all known insect chitinases (Zhang et al. 2011): FDGXDLDWEYP (the residue E is a putative proton donor in the mechanism, and is important for catalytic activity), KXXXXXGGW, MXYDXXG, and GXXXWXXDXD. The sequences that had none of these four conserved sequences were discarded.

Phylogenetic analyses to all the identified B. mori chitinase members were carried out in two steps. First, all obtained catalytic domain sequences of B. mori chitinases were used to build a neighbor-joining distance tree with the D. melanogaster and T. castaneum chitinases catalytic domain sequences using MAGA5.0 (Tamura et al. 2011). Then, each catalytic domain sequence of B. mori chitinases was used to conduct in-group phylogenetic analyses (Wang et al. 2007). The neighborjoining trees were bootstrapped with 5,000 replicates to provide information about their statistical reliability.

The catalytic domain sequences of chitinases from D. melanogaster and T. castaneum were used to carry out a tBLASTn search against the B. mori genome database. The four conserved regions of catalytic domain amino acid sequences were included in all the reference chitinases used for tBLASTn searching.

There were nine B. mori chitinase proteins identified by tBLASTn searches and sequence analysis. They had four conserved regions of catalytic domain (Table 1). Because the following phylogenetic analyses (Figure 1 and Figure 2) provided support for allocating the identified B. mori chitinases into their corresponding families, these chitinases were named as Bm followed by the family name plus a number if more than one member was found in that family. However, two B. mori chitinases cannot form monophyletic groups with T. castaneum, A. gambiae, or D. melanogaster chitinase. These two genes were named as BmCht25 and BmCht26 in accordance with the nomenclature used by Zhang et al (2011). BmCht5 (GenBank accession No.: AAB47538), BmCht26 (GenBank accession No.: BAC67246), and BmIDGF (Imaginai disc growth factor, GenBank accession No.: NP_001036847) were previously reported (Kim et al. 1998; Daimon et al. 2003; Wang et al. 2009). These three proteins were renamed based on the phylogenetic analysis of the catalytic domain. Three B. mori hypothetical chitinases protein were found to have incomplete catalytic domain, and these genes were named as B. mori hypothetical chitinase protein 1, 2, and 3, respectively.

Nakabachi et al. (2010) found 12 B. mori chitinases when they studied chitinase-like proteins encoded in the genome of Acyrthosiphon pisum. Ten of the sequences were the same as were found in the present study (BmCht5, BmCht6, BmCht7, BmCht10, BmCht11, BmCht20, BmIDGF, BmCht25, BmCht26, B. mori hypothetical chitinase protein 1), but the other two sequences they mentioned may not belong to glycoside hydrolase family 18, because they do not contain conserved sequences of Glyco_18 catalytic domain as shown by the analysis with SMART tool.

Identifying orthologous genes is accompanied with much uncertainty because there is no absolute criterion that can be used to decide if two genes are orthologous (Ledent and Vervoort 2001). In the present study, the criterion used was that a single B. mori chitinase must form a monophyletic group with another chitinase of a known family in phylogenetic trees constructed with different methods, and all the bootstrap values must exceed 50 (Zhu et al. 2004; Zhang et al. 2011).

The BmCht5, BmCht6, BmCht7, and BmCht11 formed monophyletic groups with T. castaneum, A. gambiae, and D. melanogaster catalytic domains, had bootstrap values greater than or equal to 50, and belong to Group I, Group VI, Group III, Group VIII, respectively (Figure 1). Bootstrap values of BmIDGF, BmCht10, BmCht20, BmCht25, and BmCht26 were not greater than or equal to 50. Further phylogenetic analyses showed that BmIDGF belongs to Group V (Figure 2A), BmCht10 belongs to Group II (Figure 2B), and BmCht20 belongs to VII (Figure 2C). BmCht25 and BmCht26 were not included in any of the chitinase groups (Groups I–VIII) (Zhang et al. 2011). They may belong to two new groups of the chitinase family (Figure 2D and 2E). BmCht25 had 78% sequence identity with Danaus plexippus chitinase (GenBank accession No. EHJ68452) and 44% identity with Culex quinquefasciatus chitinase (GenBank accession No. XP_001869617). BmCht26 had 92% sequence identity with Manduca sexta chitinase (GenBank accession No. ABB88891) and 91% identity with Scania Cynthia chitinase (GenBank accession No. BAE16586).

BmCht11 has only one exon and BmCht26 has two exons, whereas other B. mori chitinase genes have more than two exons (Figure 3). For example, has 9 exons and BmCht5 has 10 exons (Figure 3). B. mori contains 27 autosomes and one Z chromosome. In this study, all identified B. mori chitinases genes on the silkworm chromosomes were mapped successfully. BmCht5 and BmCht26 are located on chromosome 7. Other B. mori chitinases (BmCht6, BmCht7, BmCht10, BmCht11, BmCht20, BmIDGF) are located on separate chromosomes, respectively (Figure 4).

A multi-domain structural organization is often observed in polysaccharide-degrading enzymes, where one or more domains are responsible for hydrolysis and associating with the solid polysaccharide substrate. Catalysis in family 18 chitinases depends on catalytic domains. CBD domains in the chitinases are presumably to anchor the enzyme tightly onto the substrate. The linker region apparently helps to stabilize the enzyme and protects protease-susceptible bonds in the catalytic domain from hydrolysis in the gut. The PEST sequence in the linker region serves as proteolytic signals. With such domains, insect chitinases may evolve from efficient degradation of the insoluble polysaccharide to soluble oligosaccharides during the molting process (Kramer 1997; Arakane et al. 2003).

Domain analysis of the nine B. mori chitinases showed that they contain at least one Glyco_18 catalytic domain (Figure 5). The Glyco_18 catalytic domain with FDGXDLDWEYP sequence is predicted to be active catalytically, and the glutamate amino acid (E) substitutions in these sequences are predicted to be non-enzymatic. Most likely, these proteins are lectins that will bind to chitin or other carbohydrates containing Nacetylglucosamine. They may be involved in cell-to-cell communication or in insect immunity. BmCht10 has five catalytic domains. The first catalytic domain has a glutamate-valine (E-V), the second catalytic domain has a glutamate-asparagine (E-N), the fifth catalytic domain has a glutamateGlutamine (E-Q), and the other two catalytic domains retain glutamate residues. Both BmIDGF and BmCht25 contain one catalytic domain, which have glutamate-glutamine (EQ) and glutamate- leucine (E-L), respectively (Figure 6).

The function of the CBD domain in chitinases is presumably to anchor the enzyme tightly onto its substrate, thereby facilitating the hydrolytic process. In insect chitinases, the CBD domain has six conserved cysteines that probably form three disulphide bridges (Suetake et al. 2000; Behr and Hoch 2005). BmCht10 contains seven CBD domains, BmCht5 contains one CBD domain, and the other seven B. mori chitinases (BmCht6, BmCht7, BmCht11, BmCht20, BmCht25, BmCht26, and BmIDGF) do not contain a CBD domain (Figure 7).

The N-terminal sequence of BmCht5, BmCht11, BmCht25, BmCht26, and BmIDGF contains a highly hydrophobic amino acid that is likely to function as a signal peptide (Figure 3), which means that these chitinases are secreted. Some chitinases contain the sequences rich in proline (P), glutamic acid (E), serine (S), and threonine (T), which qualifies it as a PEST sequence (Rechsteiner and Rogers 1996). The linker region of Manduca sexta and T. castaneum chitinases also contain potential PEST sequences. It serves as proteolytic signals, necessary and sufficient for ubiquitination (Roth and Davis 2000). BmCht10 contains five catalytic domains that are separated by three potential PEST. It is speculated that BmCht10 may be expressed as zymogens that are subsequently cleaved by proteolysis to reveal multiple active enzymes. BmCht5 also contains two potential PEST sequences. This composition may promote the chitinase precursor to be converted into active form, or suggest that the insect chitinases might be rapidly degraded proteins.

BmIDGF shows high sequence similarity to chitinase and belongs to group V of the chitinase-like family (Zhu et al. 2008). IDGF is a soluble polypeptide growth factor that was first identified from the conditioned medium of D. melanogaster imaginai disc C1.8+ cells. It is also expressed in larval glands and fat body. It is secreted and transported to target tissues via the hemolymph (Kawamura et al. 1999). However, the D. melanogaster IDGF, D. melanogaster DS47, and T. castaneum IDGF were demonstrated to be devoid of chitinolytic activity, although they could act as carbohydrate-binding proteins and bind very tightly to an insoluble ligand-colloidal chitin. In Mamestra brassicae, recombinant M. brassicae IDGF could stimulate the growth of cells derived from the fat body and hemocytes without insulin (Zhang et al. 2006). Previous work (Wang et al. 2009; Pan et al. 2010) found that BmIDGF was expressed in all developmental stages of silkworm larvae and various larval tissues and was located in the extracellular space. It was predicted that BmIDGF is produced mainly by the fat body, secreted into the hemolymph, and then brought to other organs.

BmCht5 generates four mRNA products (GenBank accession No. NP_001166831, NP_001037480, NP_001166832, and NP_001166833) by alternative splicing. Furthermore, the four mRNA products showed chitinase activity when expressed in Escherichia coli, which demonstrates the role of the alternative splicing process in generating multiple isoforms of the silkworm's chitinases. The presence of this splicing mechanism in the B. mori chitinases also contributes to the interpretation of the variations reported in studies of the insect's cDNAs. All of the mRNA products are from a single gene and functionally active (Babiker et al. 2002). Mikitani et al. (2000) found a novel DNA type transposon in BmCht5 that shows similarity to the TC-like transposable element.

BmCht26 showed extensive homology with chitinase of bacteria and baculoviruses. BmCht26 showed 62% sequence identity with BmNPV ChiA and 72% with Serratia marcescens chiA. It was suggested that BmCht26 may be derived from a bacterial or baculovirus chitinase gene via horizontal gene transfer (Daimon et al. 2003, 2005, 2006). The insect chitinases have been found to be endochitinases. In a recent study, BmCht26 had exo-type substrate preference just as baculoviruses ChiA (Daimon and Katsuma 2007), which is different from other B. mori chitinases. BmCht26 localizes in the chitincontaining tissues during the molting stages, indicating that it plays a role in chitin degradation during molting.

Interestingly, B. mori hypothetical chitinase proteins 1, 2, and 3 were found to have an incomplete catalytic domain that had lost the FDGXDLDWEYP and KXXXXXGGW conserved regions, but all of them retained the CBD domain. They may be hypothetical chitinase proteins in the silkworm genome (Table 1). They cannot form monophyletic groups with T. castaneum, A. gambiae, or D. melanogaster chitinases (data not shown). Nevertheless, it was found that T. castaneum, Danaus plexippus, and Pediculus humanus corporis also have a number of hypothetical chitinase sequences (GenBank accession No. EFA05323, EHJ77971 and XP_002425481), all of which have high sequence identity with the three B. mori hypothetical chitinase proteins. Whether they are genuine chitinase proteins and have chitinase-like functions remains to be determined by further studies.

In D. melanogaster, sixteen chitinases were identified. In T. castaneum, twenty-two chitinases were identified. In A. gambiae, twenty chitinases were identified (Zhang et al. 2011). In this work, nine B. mori chitinases and three hypothetical chitinase proteins were identified in the current version of silkworm genome sequences. Therefore, identification of more B. mori chitinases is expected upon release of a new version of the silkworm genome. Protein expression and chitinase activity assays, as well as chitin-binding studies, are needed to verify that the sequences identified in this study are indeed B. mori chitinase proteins.