Discovery, identification and comparative analysis of non-specific lipid transfer protein (nsLtp) family in Solanaceae.

Plant non-specific lipid transfer proteins (nsLtps) have been reported to be involved in plant defense activity against bacterial and fungal pathogens. In this study, we identified 135 (122 putative and 13 previously identified) Solanaceae nsLtps, which are clustered into 8 different groups. By comparing with Boutrot's nsLtp classification, we classified these eight groups into five types (I, II, IV, IX and X). We compared Solanaceae nsLtps with Arabi-dopsis and Gramineae nsLtps and found that (1) Types I, II and IV are shared by Solanaceae, Gramineae and Arabidopsis; (2) Types III, V, VI and VIII are shared by Gramineae and Arabidopsis but not detected in Solanaceae so far; (3) Type VII is only found in Gramineae whereas type IX is present only in Arabidopsis and Solanaceae; (4) Type X is a new type that accounts for 52.59% Solanaceae nsLtps in our data, and has not been reported in any other plant so far. We further built and compared the three-dimensional structures of the eight groups, and found that the major functional diversification within the nsLtp family could be predated to the monocot/dicot divergence, and many gene duplications and sequence variations had happened in the nsLtp family after the monocot/dicot divergence, especially in Solanaceae.

Introduction

Plant non-specific lipid transfer protein (nsLtp) was first purified and characterized from spinach leaves and named for its capability of transferring phospholipids from liposome to mitochondria or chloroplasts 1, 2. Two nsLtps have been discovered from germinated castor bean seeds (, and Tsuboi et al. demonstrated organ-specific occurrence and expression of the nsLtp isoforms in castor bean seedlings (. NsLtps are encoded by a multigene family in many plant species and most nsLtps have been identified on the basis of sequence homology from cDNA clones or their sequences. For instance, 52, 49 and 122 putative nsLtps have been identified in rice (Oryza sativa), Arabidopsis thaliana and wheat (Triticum aestivum), respectively (.

Plant nsLtps are involved in many useful physiological functions, such as mediating phospholipid transfer, involving in plant defense activity against bacterial and fungal pathogens, participating in the assembly of hydrophobic protective layers of surface polymers like cutin, and associating with abiotic stresses 1, 6, 7, 8, 9, 10, 11, 12, 13. In addition, several members of the nsLtp family have been identified as relevant allergens in plant foods and pollens 11, 14, 15, 16. NsLtps display a complex tissue- and development-specific expression pattern 17, 18. NsLtp expression is also mediated by Ca2+ and calmodulin, which interact specifically with nsLtp 19, 20.

All known plant nsLtp precursors are synthesized with an N-terminal signal peptide. Plant nsLTPs are small (usually 6.5 to 10.5 kDa except for type VII nsLTPs, which are about 15 kDa) and their isoelectric point (pI) is usually in the range of 8.5 to 12 (some are below 5). NsLtps have eight cysteine motifs (8CM) in a backbone sequence: C-Xn-C-Xn-CC-Xn-CXC-Xn-C-Xn-C (. According to the sequence similarity, Boutrot et al. divided nsLtps into nine types (I, II, III, IV, V, VI, VII, VIII and IX) (.

nsLtps are predicted to be located in the extracellular matrix and possess a broad lipid-binding specificity, which is defined mostly based on their three-dimensional (3D) structures. The typical nsLtp fold, which is characterized by an α-helical compact domain composed of four α-helices (H1-H4), is connected by short loops (L1-L3) and a non-structured C-terminal tail. The compact domain is firmly held by a network of four conserved disulphide bridges [Pattern 1: C1-C6, C2-C3, C4-C7 and C5-C8 for type I nsLtps (; Pattern 2: C1-C5, C2-C3, C4-C7 and C6-C8 for type II ( and type IV (], such as LTP1 gene, which belongs to pattern 1 and type X (Figure 1) (. Additionally, a large number of intra-molecular H-bonds contribute to the stabilization of their 3D structures. This particular folding structure presents a large internal tunnel-like cavity that accommodates different types of lipids and also exhibits unusual stability against thermal processing and digestion 25, 26.

Figure 1

The 1T12 nsLtp structure of tobacco (Nicotiana tabacum). A. The sequence length, start and end position of 1T12 nsLtp gene in NCBI genomic DNA sequence. B. The exon and intron positions in 1T12 gene. The start of first exon and the end of second exon are not clear, so we use protein translation start and end site instead. C. The protein sequence of 1T12 and its position in the gene sequence. Signal peptide is yellow and mature peptide is green. D. The secondary structure of 1T12 and its position in protein sequence. E. The 3D structures of 1T12 in three styles shown from left to right. The first style shows the eight Cys residues in yellow color. The second style shows the four disulfide bonds in yellow color. The third style shows the eight Cys residues in yellow color and labels the Cys amino acids.

Solanaceae plants are major economic crops, such as potato (Solanum tuberosum), tomato (Solanum lycopersicum), tobacco (Nicotiana tabacum), and pepper (Capsicum annuum). Moreover, some Solanaceae plants have pharmacologic and horticultural value, such as Chinese wolfberry (Lycium barbarum) and Garden Petunia (Petunia hybrida Vilm). Therefore, it is not surprising to see 11 genome projects on Solanaceae plants at NCBI (http://www.ncbi.nlm.nih.gov/genomeprj). However, the Solanaceae nsLtp family has not been thoroughly annotated. In our study, we classified Solanaceae (the six Solanaceae species analyzed: S. tuberosum, S. lycopersicum, N. tabacum, Nicotiana benthamiana, C. annuum and P. hybrida) nsLtps by data mining, phylogenetic analysis, and 3D structure building. We also compared Solanaceae nsLtps with those of Arabidopsis and Gramineae (rice and wheat). Our analysis provides a detailed picture and novel insights on Solanaceae nsLtp gene family to support further functional and evolutionary research on nsLtp gene family among plants.

Results

Discovery and identification of nsLtp genes based on EST data mining in six Solanaceae species

In order to identity the non-redundant Solanaceae nsLtp sequences, we downloaded the identified sequences of nsLtp mRNAs, genes and proteins of Solanaceae, O. sativa and A. thaliana from NCBI. We found 22, 6 and 23 mRNAs; 5, 1 and 6 genes; and 13, 8 and 11 proteins in Solanaceae, O. sativa and A. thaliana, respectively (Table S1). Since none of the Solanaceae plant genome sequences have been finished until September, 2008, for our analysis we used 61,372, 46,849, 33,522, 16,127, 14,249 and 8,729 tentative consensus (TC) sequences from six Solanaceae species (including S. tuberosum, S. lycopersicum, N. tabacum, N. benthamiana, C. annuum and P. hybrida) from the TGI database of DFCI, respectively.

In order to identify all putative Solanaceae nsLtp genes, we first blasted the identified mRNA, gene and protein sequences of Solanaceae, O. sativa and A. thaliana nsLtp against the Solanaceae TC sequence databases. We obtained 246 blast hits at E-values less than 10−10 and performed alignment using the Estwisedb program to validate the blast hits, yielding 235 putative nsLtp genes; we then used the CLC Sequence Viewer to translate all putative nsLtp genes into protein sequences and examined the 8CM motif manually and aligned all the nsLtp sequences using ClustalX program 27, 28. In order to reduce the sequence redundancy because of SNP and EST sequence error, we just considered the difference at protein level. After removing those redundant nsLtp sequences according to the alignment result, we identified 122 new Solanaceae putative nsLtp sequences (22 members have been identified in S. tuberosum, 24 in S. lycopersicum, 30 in N. tabacum, 17 in N. benthamiana, 19 in C. annuum and 10 in P. hybrida) based on EST data mining in the TGI database of DFCI (Table 1). Importantly, we first reported the putative nsLtp in N. benthamiana, C. annuum and P. hybrid. To distinguish those different species of Solanaceae, we named S. tuberosum, S. lycopersicum, N. tabacum, N. benthamiana, C. annuum and P. hybrida nsLtp as StLtp, SlLtp, NtLtp, NbLtp, CaLtp and PhLtp, respectively. These 122 newly identified and 13 previously identified protein sequences were employed for subsequent protein analysis and phylogenetic study.

Table 1

Categorization of Solanaceae nsLtp

Species	Total	Previously identified	Newly identified	Type Ia	Type Ib	Type Ic	Type Id	Type II	Type IV	Type IX	Type X
ST	28	6	22	2	0	1	0	2	3	1	19
SL	28	4	24	6	2	0	0	1	2	1	16
NT	33	3	30	4	10	2	2	3	3	1	8
NB	17	0	17	4	0	1	0	1	1	0	10
CA	19	0	19	3	2	0	1	0	1	1	11
PH	10	0	10	3	1	0	0	0	0	0	6
Total	135	13	122	22	15	4	3	7	10	4	70

Note: ST, Solanum tuberosum; SL, Solanum lycopersicum; NT, Nicotiana tabacum; NB, Nicotiana benthamiana; CA, Capsicum annuum; and PH, Petunia hybrida. The bold numbers show the species in which nsLtps were never identified before.

Characteristics of Solanaceae nsLtps

We summarized the characteristics of 135 (122 new identified and 13 previously identified) Solanaceae nsLtps in Table S2. All protein sequences were predicted for the presence of the signal peptides using SignalP 3.0 program 29, 30. 97% of the Solanaceae nsLtp precursors are synthesized with a signal peptide of 15 to 33 amino acids, except for 4 proteins (CaLtpIb.1, CaLtpId.1, NbLtpX.7 and StLtpX.14), in which we did not find signal peptides. We believe that these sequences may not be complete or they may just simply not have signal peptides. NtLtpX.1 and NtLtpX.2 have different signal peptides, but their mature proteins are identical. This phenomenon was also reported in rice and wheat. We further predicted the subcellular targeting of 131 Solanaceae nsLtps using TargetP 1.1 Server program to find their secretory pathway (. CaLtpIX.1, StLtpIc.1, SlLtpIX. 1 and NbLtpIV.1 have been predicted to contain a chloroplast targeting peptide and others are secreted into the extracellular space. Solanaceae nsLtps are small and their molecular masses usually range from 7,253 Da to 10,835 Da. The average molecular mass and the theoretical pI are 9,049 Da and 10.06 pI, respectively. Some Solanaceae nsLtp gene sequences, which have genomic DNA sequence (such as AF525363, AF525362, U66465, U66466 and X62395), have one intron and two exons (Figure 1).

The main characteristic of plant nsLtps is the presence of eight cysteine (Cys) residues in a highly conserved position C-Xn-C-Xn-CC-Xn-CXC-Xn-CXn-C. The consensus sequence from mature protein sequence alignments of Solanaceae nsLtp showed the 8CM features except for type Ic, which has nine Cys residues (Table 2). So we predicted that most of Solanaceae nsLtps have the typical nsLtp structure. The mature protein sequence alignments of Solanaceae nsLtp and some sequences of O. sativa and A. thaliana are shown in Figure S1.

Table 2

The consensus motif of each nsLtp type

NsLtp type	8CM and number of flanking amino acid residues
Cysteine order		1		2		3 4		5 6		7		8
Type Ia (Group 1)	X3,4,6,14,15	C	X9	C	X13-16	CC	X19	CXC	X22,23	C	X13,14,10	C	X4,7,1
Type Ib (Group 2)	X2,3,10,12	C	X9	C	X13	CC	X19	CXC	X22	C	X13	C	X4,5
Type Ic (Group 3)	X3	C	X9	C	X14	CC	X2CX16	CXC	X21	C	X13	C	X4
Type Id (Group 4)	X4	C	X9	C	X13,16,17	CC	X19	CXC	X22,23	C	X14	C	X6,15
Type II (Group 5)	X2	C	X7	C	X13	CC	X8	CXC	X23	C	X6	C	X0
Type IV (Group 6)	X3	C	X9	C	X15	CC	X9	CXC	X24	C	X7	C	X0
Type IX (Group 7)	X2	C	X13	C	X15	CC	X9	CXC	X22	C	X6	C	X4
Type X (Group 8)	X2,3,12	C	X9	C	X12,14	CC	X6,19,20	CXC	X22,25	C	X13,14	C	X4,6,7,8,14,17

Note: The specific elements of consensus motif that could be used to identify nsLtp types are shown in bold. The consensus motif of each nsLtp type is deduced from the analysis of the mature sequences of Solanaceae nsLtps or putative nsLtps presented in Table 1.

Phylogenetic analysis of the Solanaceae nsLtps

In order to analyze the phylogenetic relationship of the Solanaceae nsLtp families, we performed multiple sequence alignments with Solanaceae (135), O. sativa (49), A. thaliana (45) and T. aestivum (122) nsLtp mature sequences using the ClustalX program (Figure S2). We used AtI.1 and OsI.1 instead of AtLTPI.1 and OsLTPI.1. The rest had been done in the same manner. The unrooted phylogenetic trees were constructed using the neighbor-joining method in Phylip package and drawn by MEGA4 software (Figure S3) (.

Based on the identity matrix of the alignments and the protein sequence 8CM property, the 135 Solanaceae nsLtps were clustered into 8 different groups (Figure S4). We compared the result with Freddy Boutrot’s nsLtp classification and classified these eight groups into five types (group 1 in type Ia, group 2 in type Ib, group 3 in type Ic, group 4 in type Id, group 5 in type II, group 6 in type IV, group 7 in type IX and group 8 in type X). The type X is the largest one, which has 70 Solanaceae nsLtps and is unique to Solanaceae. Types I, II, IV and IX are consisted of 44, 7, 10 and 4 members, respectively. Type I is further classified into four subtypes (type Ia, type Ib, type Ic and type Id) according to their sequence similarity and phylogeny, and we named them as LtpIa, LtpIb, LtpIc and LtpId, respectively (Table S2). The Solanaceae nsLtps are not homogeneously distributed in each type, and such a distribution may be related to their functional differentiation and complementarities.

Molecular modeling and comparison

We chose eight typical members (NtLtpIa.4, type Ia; NtLtpIb.4, type Ib; NtLtpIc.2, type Ic; NtLtpId.2, type Id; NtLtpII.3, type II; NtLtpIV.3, type IV; SlLtpIX.1, type IX and NtLtpX.1, type X) from each type or subtype, and performed molecular modeling using the Modeller program 33, 34, 35, 36, 37, 38. The atomic coordinates of nsLtps (1T12, 1FK5, 1T12, 2ALG, 1L6H, 2RKN, 2RKN and 1T12) were used to build the 3D models according to sequence identity and resolution of 3D structure. The electric charge and water environment are considered by using Amber 9 program, and the final models were validated by PROCHECK (. We viewed the primary and the energy minimized structure of these models (Figure 2) using VMD and Discovery Studio Visualizer software and the final structures are listed in Table S3 (.

Figure 2

Comparison of the structural models before and after energy minimization.

We compared the 8 nsLtp models with 1T12 structure using MultiProt program, and the alignment size is only 22 because of the non-conserved residues of every type (. The alignment size is 81 for NtLtpIa.4 (type Ia), NtLtpIb.4 (type Ib), NtLtpIc.2 (type Ic), NtLtpId.2 (type Id) and NtLtpX.1 (type X), and 71 for NtLtpIV.3 (type IV) and SlLtpIX.1 (type IX). We did not show NtLtpII.3 (type II) alignment result since it is too short. We further compared the eight structures with that of 1T12 based on the combinatorial extension (CE) method (Figure 3) (. The sequence identities according to the structures are shown in Table 3. The result is in agreement with the phylogenetic analysis.

Figure 3

Comparison of eight protein models with 1T12 based on combinatorial extension calculation of the two chains.

Table 3

The RMSD values and the sequence identities of the eight typical nsLtps

	NtLtpIa.4 (type Ia)	NtLtpIb.4 (type Ib)	NtLtpIc.2 (type Ic)	NtLtpId.2 (type Id)	NtLtpII.3 (type II)	NtLtpIV.3 (type IV)	SlLtpIX.1 (type IX)	NtLtpX.1 (type X)
RMSD value	0.6856	0.6960	0.8163	0.9116	0.9403	0.5881	0.8373	0.6486
sequence identity	54.9%	31.1%	42.2%	26.7%	8.2%	14.3%	11.3%	64.4%

Discussion

In this study, we carried out an analysis on nsLtp gene family in six Solanaceae species, which enabled us to identify 122 novel putative nsLtp genes. We have also clustered them into five types according to their phylogenies, where type X is a new type identified in this work. Most of the different types are monophyletic and supported by convincing bootstrap values except for type Ia (the bootstrap values are 53 and 43 for the two type Ia groups), which is largely due to their sequence diversity.

Comparing Solanaceae nsLtps in six Solanaceae species with those of Arabidopsis and Gramineae (rice and wheat), we found that: (1) Types I, II and IV are shared by Solanaceae, Gramineae and A. thaliana; (2) Types III, V, VI and VIII are not identified in six Solanaceae species so far, but exist in Arabidopsis and Gramineae; (3) Type VII was only found in Gramineae, whereas type IX was found in Solanaceae and A. thaliana, suggesting that type VII and IX may be unique to monocots or dicots but not both (. The existence of linage-specific types suggests that they may evolve before the monocot-dicot split. Type X is a new group, as it has not been reported in any other plants so far. It accounts for 52.59% of total nsLtp among the six Solanaceae species. Based on the identity matrix of the multiple sequence alignments, type X may be a subtype of type I. However, according to their positions in the phylogenetic trees and the sequence 8CM property, these sequences probably do not share the same common ancestor with other type I members. The major conclusion we could draw from these observations is that the ancestral nsLtp gene family already evolved to three types before the separation of monocots and dicots, and subsequently, the family expanded specifically in each species by gene duplication.

Our phylogenetic result turned out to be more informative about the evolutionary relationships and amended the previous results (. We classified types not only according to the percentage identity, but also the property of 8CM sequence and the comparison of 3D structures. By comparing the 3D structures of the eight groups in Solanaceae, we found that type X is derived from type I based on protein structural features (for example, the alignment size between type X and type I). The root mean square deviation (RMSD) values between the eight structures and 1T12 structure are also accordant with their sequence homology (Table 3).

Conclusion

In this study we identified four known types and a new type (type X) for nsLtp, yielding a total of ten types in plants. Phylogenetic analysis and 3D structural modeling further confirmed that type X is a novel type. The diversity and complexity of type X nsLtp genes are the evolutionary consequence of adaptation to Solanaceae-specific functions or various environmental changes. Our work provides essential information on the classification of the Solanaceae nsLtp gene family and gives novel insights into their functional and evolutionary studies.

Materials and Methods

Sequence retrieval

Previously identified nsLtp sequences of Solanaceae, O. sativa and A. thaliana were retrieved from NCBI according to their annotations (Table 4). In order to identify all the other putative Solanaceae nsLtp genes, TC sequences of six Solanaceae plants (S. tuberosum, S. lycopersicum, N. tabacum, N. benthamiana, C. annuum and P. hybrida) were downloaded from the TGI databases in DFCI (http://compbio.dfci.harvard.edu/tgi/), which come from the Gene Index Project, and installed on our local server.

Table 4

The numbers of nsLtp mRNA, gene and protein sequence

	S. tuberosum	S. lycopersicum	N. tabacum	C. annuum	O. sativa	A. thaliana	Total
No. of nsLtp mRNA sequences	6	4	5	7	6	23	51
No. of nsLtp gene sequences	2	2	1	0	1	6	12
No. of nsLtp protein sequences	6	4	3	0	8	11	32

In silico discovery and identification of Solanaceae nsLtp genes by EST data mining

We performed BLAST (version 2.2.17) search using all known nsLtp molecular sequences of Solanaceae, O. sativa and A. thaliana against each Solanaceae TC sequences database. Sequences that satisfied E-value less than 10−10 and score-value more than 100 were considered as new candidate nsLtp genes. For validating the results of blast, we further compared the 32 known nsLtp sequences with all blast results using Estwisedb (version 2.2.0) (. Sequences with more than 20 score-value were identified as putative nsLtp gene homologues. We translated all the TC sequences of putative nsLtp genes into proteins using CLC Sequence Viewer 5.0.1, and multiple alignment was then performed using ClustalX (version 2.0.6) (. A protein was considered as a new putative nsLtp if at least one mutation was observed.

All identified nsLtps were analyzed for presence of potential signal peptide cleavage sites using SignalP 3.0 29, 30. After removing the signal peptide of all nsLtps, we got putative mature nsLtp sequences. Each of the putative mature nsLtp sequences was manually validated through the analysis of the Cys residue pattern: the 8CM in a strongly conserved position (Cys1-Xn-Cys2-Xn-Cys3Cys4-Xn-Cys5XCys6-Xn-Cys7-Xn-Cys8). After removing those sequences that did not have the 8CM (except for the sequences that lose Cys residue because of losing some sequences or having some sequence errors), we identified 122 new Solanaceae putative nsLtp sequences in total. These protein sequences were employed for subsequent protein analysis and evolutionary study.

Sequence alignment and phylogenetic tree reconstruction

Solanaceae, O. sativa and A. thaliana nsLtp sequences were aligned with ClustalX (version 2.0.6). After refining the alignment results manually, we built the phylogenetic tree using neighbor-joining method in Phylip package and drew it by MEGA4 software package (. We used AtI.1 and OsI.1 instead of AtLTPI.1 and OsLTPI.1. The rest had been done in the same manner. The confidence level of each node was estimated by bootstrap procedure using 1,000 re-sampling repetitions of the data.

We constructed the models of the eight typical Solanaceae nsLtps with Modeller program using known nsLtp 3D structures [PDB ID: 1T12 (, 1FK5 (, 1T12, 2ALG (, 1L6H (, 2RKN (, 2RKN and 1T12, respectively] according to sequence identity and resolution of 3D structure. The lowest sequence identity between the target protein and the template protein is 33%. A total of 50 models were built for each nsLtp molecule, and the model quality was evaluated by DOPE assessment score, GA341 assessment score and PROCHECK program (. The model selection was based on the model quality result. At last, we performed the model optimization using the Amber 9 program by adding Na+ or Cl− and water molecular to the atomic coordinate of models. The structure alignment between eight final Solanaceae nsLtp models and 1T12 structure was performed by MultiProt program, and CE program calculates the two chains 41, 42.

Authors’ contributions

WL and SS conceived and carried out the study design, performed the bioinformatics analysis, interpreted the analysis results and drafted the manuscript. DH, KL, SH, JY and GG participated in the study design and discussion, and helped in revising and editing the manuscript. All authors read and approved the final manuscript.

Competing interests

The authors have declared that no competing interests exist.