Bioinformatics-Based Identification of Chemosensory Proteins in African Malaria Mosquito, Anopheles gambiae

Abstract

Chemosensory proteins (CSPs) are identifiable by four spatially conserved Cysteine residues in their primary structure or by two disulfide bridges in their tertiary structure according to the previously identified olfactory specific-D related proteins. A genomics- and bioinformatics-based approach is taken in the present study to identify the putative CSPs in the malaria-carrying mosquito, Anopheles gambiae. The results show that five out of the nine annotated candidates are the most possible Anopheles CSPs of A. gambiae. This study lays the foundation for further functional identification of Anopheles CSPs, though all of these candidates need additional experimental verification.

Introduction

Anopheles gambiae (A. gambiae) is the principal vector of malaria that afflicts more than 500 million people and causes more than 1 million deaths each year. Analysis of the whole genome of A. gambiae revealed strong evidence for about 14,000 protein-coding transcripts, which need further annotation and experimental verification (1).

Olfaction plays a key role in host selection of agricultural pests and disease vectors. Recent advancement in understanding the molecular mechanism of olfaction is the result of multidisciplinary research efforts by using a variety of model organisms including insects. The reception of pheromones and general odorants is mediated by specific neurons located in specialized cuticular sensilla in insects (2). Chemosensory neurons extend their dendrites to a lymphatic cavity, where the soluble and low molecular weight proteins that are supposed to transfer the hydrophobic odorants across the fluid barrier to the receptive dendritic membrane are contained. Molecular cloning and biochemical surveys of insect antennae have identified two abundant but unrelated families of small soluble proteins with proposed odorant transport function, that is, odorant-binding proteins (OBPs) and olfactory specific-D (OS-D) related proteins 3., 4.. OS-D related proteins (average 13 kDa) were first identified by subtractive hybridisation experiments using antennae of Drosophila melanogaster 5., 6.. Many OS-D homologues were subsequently identified based on sequence similarity in different insect orders (Table 1).

Table 1.

Previously Identified Insect Chemosensory Proteins^*

Order	Species	Protein namea	Length(a.a)	Accession No.b	References
Hymenoptera	Apis mellifera	ASP3c	130	AF481963	7., 8.
Lepidoptera	Cactoblastis cactorum	CLP-1	130	U95046	9
	Manduca sexta	SAP1	105	AF117574	10
		SAP3	126	AF117585	10
		SAP2	127	AF117592	10
		SAP5	231	AF117594	10
		SAP4	127	AF117599	10
	Bombyx mori	BmorCSP2	120	AF509238	11
		BmorCSP1	127	AF509239	11
	Mamestra brassicae	CSP-MbraA1	112	AF211177	12
		CSP-MbraA2	112	AF211178	12
		CSP-MbraA3	112	AF211179	12
		CSP-MbraA4	112	AF211180	12
		CSP-MbraA5	112	AF211181	12
		CSP-MbraB1	108	AF211182	12
		CSP-MbraB2	108	AF211183	12
		CSP-MbraA6	128	AF255918	13
		CSP-MbraB3	108	AF255919	13
		CSP-MbraB4	108	AF255920	13
	Heliothis virescens	HvirCSP2	126	AY101511	14
		HvirCSP1	114	AY101512	14
		HvirCSP3	106	AY101513	14
	Mamestra brassicae	SAP	111	AY026760	unpublished
	Helicoverpa armigera	CSP-Harm	127	AF368375	unpublished
	Helicoverpa zea	CSP-Hzea	128	AF448448	unpublished
Diptera	D. melanogaster	A10	155	U05244	15
		RH70879p	124	BT001865	unpublished
		PEBmeIII	158	U08281	16
	Anopheles gambiae	SAP-1	127	AF437891	17
Orthoptera	Schistocerca gregaria	CSP-sg1	109	AF070961	18
		CSP-sg2	109	AF070962	18
		CSP-sg3	103	AF070963	18
		CSP-sg4	109	AF070964	18
		CSP-sg5	109	AF070965	18
	Locusta migratoria	OS-D1	103	AJ251075	19
		OS-D2	120	AJ251076	19
		OS-D3	125	AJ251077	19
		OS-D4	125	AJ251078	19
		OS-D5	125	AJ251079	19
Phasmatodea	Eurycantha calcarata	CSP-ec1	107	AF139196	20
		CSP-ec2	102	AF139197	20
		CSP-ec3	107	AF139198	20
Dictyoptera	Periplaneta americana	p10	130	AF030340	21., 22.

^*GenBank (04/2003);

^aRegistered names in GenBank;

^bGenBank accession numbers.

OS-D related proteins differ from OBPs in several aspects: they share no sequence similarity with OBPs and contain only four of the six spatially conserved Cysteine residues that are characterised by OBPs (4). Although two of the protein families are both represented by multiple genes within a given species, OS-D related proteins are more conserved than OBPs across evolution or between different phyla, with 40%-50% of identical residues even between most distant species. They have been identified in a variety of tissues, while most OBPs appear to be restricted to olfactory tissues. They are common in Orthopteroid (phasmid and grasshopper) and holometabolous (Lepidoptera and Diptera) insects (Table 1) and thus may be present throughout the Neoptera, while OBPs are only known within the holometabolous and hemipteroid lineages (4). There is no strong evidence for the physiological role of OS-Ds so far, and they may be involved in chemical communication and perception. To contrast with OBPs that are found in olfactory sensilla, the OS-D related proteins were designated as Chemosensory Proteins (CSPs; ref. 18., 23.).

Compared with the twenty-nine putative A. gambiae OBPs characterized for similarity to OBPs of Drosophila and other insects (24), no bioinformatics-based annotation has been carried out to identify the CSP candidates of A. gambiae. The NMR (Nuclear Magnetic Resonance) solution structure of chemosensory protein Csp2 (1K19_A) from moth Mamestra brassicae has been established (23), which is the best elucidated insect CSP and would be used as a model for homology modelling. We created an algorithm for identifying the conserved domains present in Anopheles putative CSPs through Perl programming.

Results

Conserved domain of insect CSP candidates

Exhaustive queries with all previously identified insect CSP sequences retrieved from GenBank (April 2003) and ClustalX multialignment resulted in an absolutely conserved structure for insect CSPs, that is, Cx(6,8)Cx(18)Cx(2)Cx(3) (Figure 1). We used a program developed by the authors through Perl programming to search the local database that contains the Fasta files of Anopheles gDNA (genomic DNA) and cDNA (complementary DNA) sequences downloaded from Ensembl Mosquito Genome Server for the identified pattern in their primary structure. Totally eight sequences were hit, including agCP10968, agCP11079, agCP11481, agCP11484, agCP11532, agCP11545, agCP6514, and agCP12965. The hits were in turn corroborated by BLAST searching in GenBank, and the prediction was made for their biochemical properties and secondary structure.

Fig. 1.

Multiple alignment showing the absolutely conserved domain in insect CSPs: Cx(6,8)Cx(18)Cx(2)Cx(3). All acronyms in the figure refer to the protein names listed in Table 1.

CSP candidates corroborated by BLAST

The hits obtained through pattern searching were corroborated by BLAST in GenBank (http://ncbi.nlm.nih.gov/). Six sequences (agCP10968, agCP 11079, agCP11481, agCP11484, agCP11532, and agCP11545) were found closely related to the previously identified CSPs, whereas two sequences (agCP6514 and agCP12965) matched no other proteins, so the CSP candidacy of agCP6514 and agCP12965 could not be excluded. The most interesting discovery was that a novel Anopheles CSP candidate (agCP11435) had been identified by BLAST searching, though the Expect (E) values are not very high (Table 2).

Table 2.

Anopheles CSP Candidates Found by CSPMOT and BLAST (E^*-values<0.0001)

Peptide ID	Previously identified insect CSPs
agCP10968	SAP-1	ASP3c	SAP2	CSP-Harm	HvirCSP2	OS-D3	CSP-sg1
	(6e-18)	(7e-18)	(6e-17)	(1e-16)	(2e-16)	(2e-16)	(2e-16)
	OS-D1	CSP-sg4	CSP-sg2	CSP-MbraA6	OS-D4	CSP-Hzea	OS-D5
	(2e-16)	(3e-16)	(3e-16)	(3e-16)	(4e-16)	(5e-16)	(3e-16)
	SAP4	PEBmeIII	CSP-sg5	OS-D2	A10	CSP-ec3	CSP-MbraA3
	(7e-16)	(9e-16)	(9e-16)	(2e-15)	(2e-15)	(2e-15)	(2e-15)
	CSP-MbraA1	CSP-MbraA2	CSP-sg3	p10	CSP-MbraA5	SAP3
	(2e-15)	(3e-15)	(3e-15)	(3e-15)	(7e-15)	(8e-15)
agCP11079	SAP-1	PEBmeIII	ASP3c	p10	HvirCSP2	SAP4	A10
	(1e-45)	(1e-34)	(2e-34)	(4e-28)	(3e-27)	(1e-26)	(2e-26)
	OS-D2	CSP-MbraA6	OS-D3	SAP3	BmorCSP1	SAP5	CSP-sg1
	(6e-26)	(2e-23)	(4e-23)	(7e-23)	(8e-23)	(9e-23)	(1e-22)
	CLP-1	CSP-MbraA3	CSP-sg4	HvirCSP1	CSP-sg2	OS-D4	OS-D5
	(2e-22)	(3e-22)	(4e-22)	(5e-22)	(6e-22)	(7e-22)	(7e-22)
	CSP-sg4	OS-D1	CSP-MbraA2	CSP-sg3	SAP2	CSP-MbraA3	agCP11435
	(1e-21)	(1e-21)	(2e-21)	(3e-21)	(4e-21)	(4e-21)	(5e-08)
agCP11481	ASP3c	PEBmeIII	A10	SAP4	HvirCSP2	SAP5	SAP3
	(4e-28)	(3e-27)	(3e-26)	(3e-25)	(4e-25)	(4e-24)	(2e-23)
	CSP-sg2	SAP-1	CSP-sg1	OS-D2	CSP-MbraA6	CLP-1	CSP-sg5
	(1e-22)	(1e-22)	(2e-22)	(3e-22)	(7e-22)	(1e-21)	(2e-21)
	CSP-sg4	CSP-MbraA3	CSP-sg3	CSP-MbraA2	OS-D3	HvirCSP1	CSP-MbraA1
	(2e-21)	(9e-21)	(1e-20)	(2e-20)	(3e-20)	(4e-20)	(6e-20)
	CSP-MbraA3	CSP-MbraB1	SAP2	CSP-MbraA4	CSP-MbraA5	BmorCSP1	agCP11435
	(7e-20)	(8e-20)	(1e-19)	(2e-19)	(2e-19)	(3e-19)	(1e-06)
agCP11484	SAP-1	ASP3c	PEBmeIII	HvirCSP2	p10	CSP-MbraA6	OS-D2
	(3e-55)	(1e-31)	(5e-32)	(6e-29)	(4e-28)	(3e-27)	(9e-27)
	SAP4	BmorCSP1	CLP-1	SAP2	A10	SAP3	SAP5
	(1e-26)	(2e-26)	(2e-26)	(1e-25)	(6e-25)	(4e-24)	(6e-24)
	CSP-Hzea	CSP-MbraA2	HvirCSP1	CSP-MbraA1	CSP-MbraA3	CSP-Harm	CSP-MbraA4
	(3e-23)	(5e-23)	(5e-23)	(7e-23)	(7e-23)	(9e-23)	(2e-22)
	CSP-MbraA5	CSP-sg1	CSP-sg4	CSP-sg2	OS-D3	OS-D1	CSP-ec1
	(2e-22)	(4e-22)	(8e-22)	(1e-21)	(2e-21)	(3e-21)	(3e-21)
	CSP-sg5	agCP11435
	(3e-21)	(2e-08)
agCP11532	RH70879	ASP3c	SAP5	CSP-sg4	CSP-MbraA6	CSP-sg5	CSP-sg2
	(5e-30)	(4e-10)	(6e-08)	(6e-08)	(1e-07)	(1e-07)	(2e-07)
	HvirCSP2	CSP-sg1	SAP2	CSP-ec3	BmorCSP2	CSP-sg3	OS-D3
	(2e-07)	(3e-07)	(4e-07)	(7e-07)	(7e-07)	(9e-07)	(1e-06)
	CLP-1	CSP-MbraA3	CSP-MbraA1	CSP-MbraA2	CSP-MbraA4	SAP4	CSP-MbraA5
	(1e-06)	(1e-06)	(1e-06)	(1e-06)	(1e-06)	(2e-06)	(2e-06)
	OS-D2	OS-D1	A10	CSP-Hzea	HvirCSP3	CSP-ec1
	(3e-06)	(5e-06)	(1e-05)	(1e-05)	(1e-05)	(3e-05)
agCP11545	SAP-1	PEBmeIII	ASP3c	p10	HvirCSP2	A10	OS-D2
	(3e-40)	(2e-33)	(2e-32)	(1e-30)	(6e-30)	(1e-28)	(1e-27)
	SAP4	CSP-MbraA6	CSP-sg1	BmorCSP1	SAP3	CSP-sg4	CSP-sg2
	(3e-26)	(1e-25)	(3e-24)	(3e-24)	(3e-24)	(6e-24)	(6e-24)
	OS-D3	SAP5	CSP-sg5	CSP-sg3	HvirCSP1	OS-D4	OS-D1
	(1e-23)	(2e-23)	(3e-23)	(3e-23)	(4e-23)	(5e-23)	(7e-23)
	CSP-MbraA3	CSP-MbraA5	CLP-1	CSP-ec1	CSP-MbraA2	CSP-Hzea	agCP11435
	(2e-22)	(2e-22)	(2e-22)	(5e-22)	(7e-22)	(1e-21)	(7e-09)
agCP6514	no matchesa
agCP12965	no matchesa
agCP11435b A10	OS-D1	OS-D3	OS-D5	OS-D4	SAP-1	SAP4
	(2e-12)	(5e-10)	(7e-10)	(1e-09)	(2e-09)	(2e-08)	(3e-08)
	PEBmeIII	OS-D2	HvirCSP2	ASP3c	CSP-sg5	CSP-ec2	CSP-sg4
	(3e-08)	(4e-08)	(8e-08)	(1e-07)	(3e-07)	(5e-07)	(5e-07)
	CSP-sg3	CSP-sg2	CSP-sg1	CSP-ec1	CSP-ec3	p10	CSP-MbraA6
	(6e-07)	(9e-07)	(9e-07)	(2e-06)	(2e-06)	(5e-05)	(5e-05)
	CSP-Hzea	CSP-Harm	CSP-MbraA3	CSP-MbraA1
	(1e-04)	(1e-04)	(1e-04)	(1e-04)

^aNo significant hits obtained by BLAST;

^bBLAST identified;

^*The E value is a parameter that describes the number of hits one can “expect” to see just by chance when searching a database of a particular size. All acronyms of insect CSPs refer to protein names listed in Table 1.

Biochemical properties, prediction of secondary structure and ORFs of Anopheles CSP candidates

The cDNA sequences of the Anopheles Genome Project stored in GenBank and in the Ensembl Mosquito Genome Browser (ftp://ftp.ensembl.org/pub/current_mosquito/) had been annotated jointly by the privately funded Celera and EBI (http://www.ensembl.org/Anopheles_gambiae/). All the entries are the results from preliminary prediction. Even the positions of start codon for most of the annotated transcripts were not determined in the database. We have predicted the complete coding sequences (CDSs) and open reading frames (ORFs) for the Anopheles CSP candidates identified through pattern searching and BLAST, based on the genomic DNA sequences of the candidates. The redefined ORFs were listed in Table 3, along with the corresponding Celera IDs, GenBank IDs (#EAA), chromosomal locations, and scaffold numbers (#AAAB). Positions of the signal peptides, isoelectric points (pI) and hydrophobicity were also presented. The information contained in the table shows that all of the CSP candidates except agCP12965 and agCP6154 are located on one scaffold (AAAB01008964) of the chromosome 3 R. This stimulating discovery may indicate that these olfactory genes were duplicated at some point of Anopheles evolution. Meanwhile, all of the CSP candidates have small molecular weights (<15 kDa in most cases) and are hydrophilic (<35% hydrophobic amino acids in most cases). Most of the candidates have a signal peptide, though no signal peptides have been found in two of the sequences, that is, agCP10968 and agCP12965. In fact, we have used several different tools to predict the ORFs of agCP10968, unfortunately no signal peptides have been found, which indicates that a sequencing error may have occurred in the Anopheles Genome Project.

Table 3.

Chromosomal Location, New ORFs, Signal Peptides, and Biochemical Properties of Anopheles CSP Candidates

Celera_ID	GB_ID	Chrom	Scaffold No.	Original length (a.a.)	New ORF length (a.a.)	Signal peptide	MW (kDa)	pI	Hydrophobic a.a. (%)
agCP10968	EAA12703	3R	AAAB01008964	127	109	none	12.3	9.5	25.7
agCP11079	EAA12353	3R	AAAB01008964	143	127	1–17	14.8	5.4	33.1
agCP11481	EAA12591	3R	AAAB01008964	137	123	1–19	14.3	9.4	29.3
agCP11484	EAA12322	3R	AAAB01008964	149	127	1–17	14.7	8.6	33.1
agCP11532	EAA12601	3R	AAAB01008964	150	117	1–33	12.9	9.8	41.0
agCP11545	EAA12338	3R	AAAB01008964	141	126	1–17	14.6	8.6	31.0
agCP11435	EAA12702	3R	AAAB01008964	102	137	1–16	15.7	5.0	35.0
agCP12965	EAA05664	3L	AAAB01008834	173	137	none	14.1	3.5	23.4
agCP6514	EAA10937	2L	AAAB01008960	132	117	1–19	13.0	8.6	25.6

Table 4 summarized the secondary structure prediction of Anopheles CSP candidates. The predictions showed that most of the candidates could be classified as all-alpha structure, with a high probability to form a globular domain. However, agCP10968 and agCP6514 did not appear to be globular.

Table 4.

Secondary Structure of Anopheles CSP Candidates^*

Peptide ID	Predicted secondory structure (%)				Globularity
	Helix	Sheet	Loop	Class
agCP10968	25.70	15.50	58.80	mixed	appears not to be globular
agCP11079	72.40	0.00	27.60	all-alpha	may be globular, but it is not as compact as a domain
agCP11481	71.50	0.00	28.50	all-alpha	may be globular, but it is not as compact as a domain
agCP11484	71.70	1.60	26.80	all-alpha	appears as compact, as a globular domain
agCP11532	72.70	0.00	27.40	all-alpha	appears as compact, as a globular domain
agCP11545	68.20	2.40	29.40	all-alpha	may be globular, but it is not as compact as a domain
agCP11435	75.90	0.00	24.10	all-alpha	may be globular, but it is not as compact as a domain
agCP12965	13.10	16.80	70.10	mixed	appears as compact, as a globular domain
agCP6514	10.30	0.00	89.70	mixed	appears not to be globular

^*Prediction server—http://www.sbg.bio.ic.ac.uk/3dpssm.

Homology modelling

The NMR solution structure of chemosensory protein Csp2 of Mamestra brassicae (1K19_A) was retrieved from Protein Data Bank (PDB ID: 1K19). The 3D structure of this model molecule is shown in Figure 2A, which is characterized by two disulfide bonds (CysI-CysII, CysIII-CysIV), instead of three disulfide bonds (CysI-CysIII, CysII-CysV and CysIV-CysVI) characterized by OBPs. The model has a typical hydrophobic core that is supposed to act as a pocket for ligand binding. The pocket is formed by hydrophobic amino acids, surrounded by hydrophilic amino acids.

Fig. 2.

Homology modelling of Anopheles CSP candidates with 1K19_A_MbraCSP (PDB ID: 1K19) as the model. A. 3D-structure of 1K19_A_MbraCSP showing the position of two disulfide bridges; B-J: predicted 3D-structure of Anopheles CSPs, including agCP11079, agCP10968, agCP11435, agCP11481, agCP11484, agCP11532, agCP 11545, agCP12695, and agCP6514. The figures are generated in Swiss-PdbViewer.

Homology modelling of the Anopheles CSP candidates was made in Swiss-PdbViewer (Figure 2). The figures showed that most candidates folded similarly as 1K19_A. However, a structurally weak linkage occurred between the two disulfide bridges of agCP11435 (Figure 2D), though a hydrophobic pocket was formed in the core of its structure. Further sequence analysis was carried out, which showed that the structural abnormality might be caused by a surplus sequence insertion (GRLACLALVL; Figure 3).

Fig. 3.

Partial alignment of candidate Anopheles CSPs with the previously identified CSPs shows the surplus sequence of agCP11435 that causes its 3D-structural abnormality. All acronyms of the gene names refer to the protein names listed in Table 1.

Discussion

Sequence similarity in CSP primary structure is significantly higher than that in OBPs. OBP sequences of the same species could be less similar than those of different species, whereas CSP sequences are more conserved at species level as well as between phylogenetically distant groups. In the present study, the sequence similarity between agCP11484 and other CSP candidates is: 77%, 76%, 47%, 39%, 21%, 15%, 13%, and 8%, respectively. Moreover, the positions of two disulfide bridges were highly conserved. The structural conservation should be extremely important in forming a strong hydrophobic core that function as an odorant-binding site.

Thirty-eight OBP candidates have been annotated in D. melanogaster, whereas only 29 OBP candidates have been conceptually identified in A. gambiae, based on sequencing and genome analysis 10., 24., 25., 26.. In the present study, we located nine CSP candidates in Anopheles genome, among which five candidates, i.e. agCP11079, agCP11481, agCP11484, agCP11532, and agCP11545, are the most possible Anopheles CSPs, considering the sequence similarity in their primary structure, the biochemical properties such as hydrophobicity and molecular weight (<15 kDa), in particular the secondary structure and 3-D structure. It would be very interesting to discuss the number of insect OBPs and CSPs (29 OBPs to 9 CSPs for Anopheles), because the numbers may be closely related to the functions. It has been shown that the previously identified OBPs are spatially distributed in insect olfactory sensilla, while CSPs are distributed in different tissues, including non-olfactory tissues. Pheromone-binding proteins (PBPs) are typical insect OBPs, which are highly specific. However, CSPs are less specific, which indicate that one CSP may be able to bind more than one odorant. We postulated that more OBPs might be needed to bind different odorants specifically, although any further hypothesis on the physiological function of CSPs can only be made when reliable experimental evidence has been presented, such as identification of the specific ligands that show CSPs function as chemosensory proteins.

It must be noted that agCP11435 may be a good model for discussion on its structural relations to its functions. We hope agCP11435 is a CSP candidate, but its structural peculiarity may provide some space for exploring its binding behaviour. It is expected that a minor insertion into the structural core may lead to destruction of the fundamental functions of a protein.

No strong evidence has been provided for the physiological role of CSPs so far, though researchers believed they might be involved in chemical communication and perception. We checked the spatial expression pattern of the putative Anopheles CSP candidates (data unpublished) and found they were distributed not only in mosquito antennae (olfactory tissues), but also in non-olfactory tissues such as heads stripped off antennae and maxillary palps, legs and bodies. The preliminary results indicated that CSPs might have other functions than olfaction. Further studies will be focused on functional research, such as ligand identification and crystallization of CSP recombinants.

Materials and Methods

Defining conserved domains in insect CSPs

The amino acid sequences of previously identified CSPs were downloaded from GenBank. All the sequences were then aligned in ClustalX 8.1 (27) using Multiple Alignment Mode with the default gappenalty parameters. The multiple alignment was manually checked, and the absolutely conserved Cysteine residues in the alignment were defined as CSP motifs.

Pattern search, selection of Anopheles CSP candidates in the whole genome sequences of A. gambiae

The Fasta files of gDNA, cDNA and the according translated peptide sequences of A. gambiae were downloaded from the Ensembl Mosquito Genome Browser (ftp://ftp.ensembl.org/pub/current_mosquito/). A pattern-searching program was created through standard PERL programming and named as CSPMOT by the authors. This program is run in Windows Commander, and can scan a local database for a pattern defined by the users. In this case, the CSP motif was used as a pattern to match every sequence stored in a local database that contains the Fasta files downloaded from the public databases. When a sequence matches the pattern, the program will display the sequence name, the position of the first residue where the pattern matches, and the pattern match alignment.

BLAST

BLAST was performed at NCBI (http://ncbi.nlm.nih.gov/BLAST/; April 2003). The Anopheles CSP candidates identified by the program described above were used as queries to Blast GenBank. The E values smaller than 0.0001 were accepted. The E value is a parameter that describes the number of hits one can “expect” to see just by chance when searching a database of a particular size.

Annotations

GENSCAN (http://genes.mit.edu/GENSCAN.html; ref. 28., 29.) and ORF Finder (http://www.ncbi.nlm.nih.gov/gorf/gorf.html) were used to predict the full-length genes, based on gDNA and cDNA sequences of Anopheles CSP candidates. The annotated CDSs were then used for calculation of hydrophobicity, pI and molecular weight of the CSP candidates, by using the comprehensive biosoftware Vector NTI (InforMax Inc., Bethesda, USA). The positions of signal peptides were determined online (http://www.cbs.dtu.dk/services/SignalP2.0/; ref. 30., 31.), and finally the secondary structure, in particular the globularity, was predicted (http://maple.bioc.columbia.edu/predictprotein/; ref. 32., 33., 34.).

Homology modelling

Csp2 of Mamestra brassicae (PDB ID: 1K19; ref. 35) is an elucidated insect chemosensory protein, which was used as a model to predict the 3-D structure of Anopheles CSP candidates in Swiss-Pdb Viewer (http://us.expasy.org/spdbv/; ref. 36). Swiss-Pdb Viewer is a program that allows analysis of several proteins at the same time. It could be used to deduce 3-D structure of proteins and compare the active sites of different molecules.