Fine specificity of TCR complementarity-determining region residues and lipid antigen hydrophilic moieties in the recognition of a CD1-lipid complex

alphabeta TCR can recognize peptides presented by MHC molecules or lipids and glycolipids presented by CD1 proteins. Whereas the structural basis for peptide/MHC recognition is now clearly understood, it is not known how the TCR can interact with such disparate molecules as lipids. Recently, we demonstrated that the alphabeta TCR confers specificity for both the lipid Ag and CD1 isoform restriction, indicating that the TCR is likely to recognize a lipid/CD1 complex. We hypothesized that lipids may bind to CD1 via their hydrophobic alkyl and acyl chains, exposing the hydrophilic sugar, phosphate, and other polar functions for interaction with the TCR complementarity-determining regions (CDRs). To test this model, we mutated the residues in the CDR3 region of the DN1 TCR beta-chain that were predicted to project between the CD1b alpha helices in a model of the TCR/CD1 complex. In addition, we tested the requirement for the negatively charged and polar functions of mycolic acid for Ag recognition. Our findings indicate that the CDR loops of the TCR form the Ag recognition domain of CD1-restricted TCRs and suggest that the hydrophilic domains of a lipid Ag can form a combinatorial epitope recognized by the TCR.     

Analysis of a nucleic-acid-binding antibody fragment: construction and characterization of heavy-chain complementarity-determining region switch variants

The display of antibody (Ab) fragments (Fab) on the surface of filamentous bacteriophage (phage) and selection of phage that interact with a particular antigen (Ag) has enabled the isolation of Fab that bind nucleic acids. Nucleic acid (NA) binding Ab occur in vivo in connective tissue disease patients and certain inbred strains of mice and are thought to be pathogenic. Although there is ample data concerning the amino acid (aa) sequence of murine monoclonal Ab (mAb) reactive with DNA, significantly less is known about how autoAb interact with NA. The complementarity-determining regions (CDR) contained in the Fab contribute the most to Ag binding, especially through heavy (H)-chain CDR 3. We have examined the role of individual H-chain CDR of a previously isolated recombinant single-stranded DNA-binding Fab (DNA-1) in nucleic acid interaction using a combination of H-chain CDR switching and solution-binding experiments. The three H-chain CDR of DNA-1 Fab were independently switched with the H-chain CDR of a Fab (D5) with very similar sequence and framework (FR) that binds DNA poorly in order to create all possible H-chain CDR combinations. The chimeric Fab genes were bacterially expressed, and their products were purified and analyzed. Results indicated that the H-chain CDR 3 of DNA-1 Fab, in the context of the remainder of the H-chain of D5 Fab, restored binding to oligo(dT)₁₅ to 60% of DNA-1 levels, whereas H-chain CDR 1 and 3 of DNA-1 with CDR 2 of D5 Fab restored binding to 100%. A combination of H-chain CDR 2 and 3 of DNA-1 Fab with H-chain CDR 1 of D5, unexpectedly resulted in the ability of the chimeric Fab to bind RNA preferentially over DNA. These studies demonstrate the importance of both H-chain CDR 1 and 3 in DNA recognition and further suggest that the specificity of the type of NA recognized by a particular Fab can be drastically altered by exchanging CDR.     

Engineering of conserved residues near antibody heavy chain complementary determining region 3 (HCDR3) improves both affinity and stability

Affinity and stability are crucial parameters in antibody development and engineering approaches. Although improvement in both metrics is desirable, trade-offs are almost unavoidable. Heavy chain complementarity determining region 3 (HCDR3) is the best-known region for antibody affinity but its impact on stability is often neglected. Here, we present a mutagenesis study of conserved residues near HCDR3 to elicit the role of this region in the affinity-stability trade-off. These key residues are positioned around the conserved salt bridge between V_H-K94 and V_H-D101 which is crucial for HCDR3 integrity. We show that the additional salt bridge at the stem of HCDR3 (V_H-K94:V_H-D101:V_H-D102) has an extensive impact on this loop's conformation, therefore simultaneous improvement in both affinity and stability. We find that the disruption of π-π stacking near HCDR3 (V_H-Y100E:V_L-Y49) at the V_H-V_L interface cause an irrecoverable loss in stability even if it improves the affinity. Molecular simulations of putative rescue mutants exhibit complex and often non-additive effects. We confirm that our experimental measurements agree with the molecular dynamic simulations providing detailed insights for the spatial orientation of HCDR3. V_H-V102 right next to HCDR3 salt bridge might be an ideal candidate to overcome affinity-stability trade-off.     

Molecular analysis of TCR and peptide/MHC interaction using P18-I10-derived peptides with a single D-amino acid substitution

For the structural analysis of T-cell receptor (TCR) and peptide/MHC interaction, a series of peptides with a single amino acid substitution by a corresponding D-amino acid, having the same weight, size, and charge, within P18-I10 (aa318-327: RGPGRAFVTI), an immunodominant epitope of HIV-1 IIIB envelope glycoprotein, restricted by the H-2Dd class I MHC molecule, has been synthesized. Using those peptides, we have observed that the replacement at positions 324F, 325V, 326T, and 327I with each corresponding D-amino acid induced marked reduction of the potency to sensitize targets for P18-I10-specific murine CD8+ cytotoxic T lymphocytes (CTLs), LINE-IIIB, recognition. To analyze further the role of amino acid at position 325, the most critical site for determining epitope specificity, we have developed a CTL line [LINE-IIIB(325D)] and its offspring clones specific for the epitope I-10(325v) having a D-valine (v) at position 325. Taking advantage of two distinct sets of CD8+ CTLs restricted by the same Dd, three-dimensional structural analysis on TCR and peptide/MHC complexes by molecular modeling was performed, which indicates that the critical amino acids within the TCRs for interacting with 325V or 325v appear to belong to the complementarity-determining region 1 but not to the complementarity-determining region 3 of Vbeta chain.     

Preparation of an alloantiserum to murine C3 and demonstration of multiple alleles.

An antiserum was produced in C3H/He mice that can recognize the antigenic variations of murine C3. The alloantiserum is directed to the gene product of one of the alleles (C3-1a allele) of the C3-1 locus that controls genetic variations of murine C3 and is linked to H-2. The antiserum formed a single precipitin line in the Ouchterlony test with sera obtained from mice that carry the C3-1a/a or C3-1a/b genotype, but it did not react with sera from mice that do not have C3 coded for by the C3-1a allele. By the use of the alloantiserum, we detected a new C3 allotype (C3-1 CC) in SWR/J mice. The C3-1 CC is distinct from the other two allotypes on the basis of antigenicity and isoelectric point (pI). Therefore, at least three alleles have been identified for the C3-1 locus.     

Central actions of angiotensin in cardiovascular control: multiple roles for a single peptide.

Angiotensin II (ANG II) acts peripherally as a hormone, with actions on the vasculature, adrenals, and kidney. In addition, certain actions of ANG II in the central nervous system are directed toward cardiovascular control and fluid volume homeostasis. Dense binding sites for ANG II are found at circumventricular organs, which apparently have the ability to relay information to cardiovascular centers via neural circuitry. Microinjection of ANG II into the subfornical organ (SFO) or area postrema (AP) produces site-specific increases in blood pressure. In addition, electrophysiological studies demonstrate profound effects of ANG II, acting at the SFO, on activity of neurohypophysial neurons and release of oxytocin and vasopressin, which can be antagonized by ANG II blockers or attenuated by SFO lesions. Evidence from microinjection, electrophysiological, and lesion studies indicate a complex interaction between central sites involved in mechanisms of cardiovascular control: the SFO, AP, organum vasculosum of the lamina terminalis, and paraventricular and supraoptic nuclei of the hypothalamus. Not only is ANG II a humoral messenger in this central scenario, but evidence suggests it acts as a neurotransmitter or neuroendocrine substance within specific CNS pathways, suggesting multiple roles for this peptide in central cardiovascular control.     

Multiple, autoreactive TCR V beta genes utilized in response to a small pathogenic peptide of an autoantigen in EAU.

The restricted usage of particular T cell receptor beta chain genes in autoimmune disease was studied in LEW rats using T cell hybridomas specific for an immunodominant sequence of bovine retinal S-Ag, which induces experimental autoimmune uveoretinitis. T cell hybridomas from a pathogenic T cell line, R858, specific for residues 273-289 of bovine retinal S-Ag were analyzed in order to determine the contribution of their TCR V beta to self specificity as determined by recognition of the pathogenic epitope represented in the autologous rat S-Ag sequence. Six different, functional TCR rearrangements were expressed by the panel of hybridomas, including two distinct V beta 8.2 rearrangements and functional V beta 10, V beta 14, V beta 19 rearrangements, and an unidentified V beta gene. All hybridomas were Ag specific and reacted both to nonself-peptide derivatives as well as to self-peptide homologues. No unique pattern of peptide reactivity distinguished V beta 8.2+ hybridomas from V beta 8.2- hybridomas; all of the hybridomas were most reactive to the nonself sequences and reacted to self peptide with one to three orders of magnitude less sensitivity. However, all V beta 8.2+ hybridomas were much better responders overall and were activated by lower concentrations of all peptides than were V beta 8.2- hybridomas. Although V beta 8.2 gene usage is strongly associated with autoimmune pathology, these data show that in LEW rats several different TCR V beta genes are utilized in response to a short pathogenic sequence of this autoantigen and show that V beta 8.2 receptors are not uniquely self-reactive. However, the enhanced reactivity to Ag of V beta 8.2+ hybridomas relative to V beta 8.2- hybridomas specific for the same peptide may help explain the close association of V beta 8.2 TCR gene usage with pathogenicity found in autoimmune disease models.     

Olfactory response of a predatory mirid to herbivore induced plant volatiles: multiple herbivory vs. single herbivory

Plants infested with a single herbivore species can attract natural enemies through the emission of herbivore‐induced plant volatiles (HIPVs). However, under natural conditions plants are often attacked by more than one herbivore species. We investigated the olfactory response of a generalist predators Macrolophus caliginosus to pepper infested with two‐spotted spider mites, Tetranychus urticae, or green peach aphid, Myzus persicae, vs. plants infested with both herbivore species in a Y‐tube olfactometer set up. In addition, the constituents of volatile blends from plants exposed to multiple or single herbivory were identified by gas chromatography‐mass spectrometry (GC‐MS). The mirid bugs showed a stronger response to volatiles emitted from plants simultaneously infested with spider mites and aphids than to those emitted from plants infested by just one herbivore, irrespective of the species. Combined with results from previous studies under similar conditions we infer that this was a reaction to herbivore induced plant volatiles. The GC‐MS analysis showed that single herbivory induced the release of 22 additional compounds as compared with the volatiles emitted from clean plants. Quantitative analyses revealed that the amount of volatile blends emitted from pepper infested by both herbivores was significantly higher than that from pepper infested by a single herbivore. Moreover, two unique substances were tentatively identified (with a probability of 94% and 91%, respectively) in volatiles emitted by multiple herbivory damaged plants: α‐zingiberene and dodecyl acetate.     

Entrapping unusual folding characteristics of the beta-Ala residues in a model peptide: Boc-beta-Ala-Aib-beta-Ala-NHCH3.

Crystal structure analysis of a model peptide: Boc-beta-Ala-Aib-beta-Ala-NHCH3 (beta-Ala: 3-amino propionic acid; Aib: alpha-aminoisobutyric acid) revealed distinct conformational preferences for folded [phi approximately 136 degrees, mu approximately -62 degrees, psi approximately 100 degrees] and semifolded [phi approximately 83 degrees, mu approximately -177 degrees, psi approximately -117 degrees] structures of the N-and C-terminus beta-Ala residues, respectively. The overall folded conformation is stabilized by unusual Ni...H-Ni+1 and nonconventional C-H...O intramolecular hydrogen bonding interactions.     

Composition effect on peptide interaction with lipids and bacteria: variants of C3a peptide CNY21

The effect of peptide hydrophobicity and charge on peptide interaction with model lipid bilayers was investigated for the C3a-derived peptide CNY21 by fluorescence spectroscopy, circular dichroism, ellipsometry, z-potential, and photon correlation spectroscopy measurements. For both zwitterionic and anionic liposomes, the membrane-disruptive potency for CNY21 variants increased with increasing net positive charge and mean hydrophobicity and was completely lost on elimination of all peptide positive charges. Analogous effects of elimination of the peptide positive net charge in particular were found regarding bacteria killing for both Pseudomonas aeruginosa and Bacillus subtilis. The peptides, characterized by moderate helix content both in buffer and when attached to the liposomes, displayed high adsorption for the net positively charged peptide variants, whereas adsorption was non-measurable for the uncharged peptide. That electrostatically driven adsorption represents the main driving force for membrane disruption in lipid systems was also demonstrated by a drastic reduction in both liposome leakage and peptide adsorption with increasing ionic strength, and this salt inactivation can be partly avoided by increasing the peptide hydrophobicity. This increased electrolyte resistance translates also to a higher antibacterial effect for the hydrophobically modified variant at high salt concentration. Overall, our findings demonstrate the importance of the peptide adsorption and resulting peptide interfacial density for membrane-disruptive effects of these peptides.     

Ligation-free gene synthesis by PCR: synthesis and mutagenesis at multiple loci of a chimeric gene encoding OmpA signal peptide and hirudin.

A unique kination and ligation-free method that allows de novo synthesis of a gene through a novel application of polymerase chain reaction (PCR) involving stepwise elongation of sequence (SES) is described. SES-PCR is simple and efficient. Optimal utilization of nucleotides, ability to use only partially purified oligodeoxyribonucleotides, and elimination of kination and ligation of intermediates make SES-PCR-mediated gene synthesis more economical in terms of time, labour and money. Site-directed mutagenesis and/or gene fusion by SES-PCR is not limited by the prior availability of the gene(s) in question. The potentials of this novel method in gene synthesis, mutagenesis at multiple loci of DNA and gene fusion have been demonstrated using a chimeric gene encoding fusion between OmpA signal peptide and hirudin, as an example. The SES-PCR product was cloned and sequencing of positive clones demonstrated the presence of genes with expected sequence and bearing only the desired mutations. A nearly 100% efficiency of mutation was easily achieved by the design of the method.     

Cross-positive selection of thymocytes expressing a single TCR by multiple major histocompatibility complex molecules of both classes: implications for CD4+ versus CD8+ lineage commitment

This study has investigated the cross-reactivity upon thymic selection of thymocytes expressing transgenic TCR derived from a murine CD8+ CTL clone. The Idhigh+ cells in this transgenic mouse had been previously shown to mature through positive selection by class I MHC, Dq or Lq molecule. By investigating on various strains, we found that the transgenic TCR cross-reacts with three different MHCs, resulting in positive or negative selection. Interestingly, in the TCR-transgenic mice of H-2q background, mature Idhigh+ T cells appeared among both CD4+ and CD8+ subsets in periphery, even in the absence of RAG-2 gene. When examined on beta2-microglobulin-/- background, CD4+, but not CD8+, Idhigh+ T cells developed, suggesting that maturation of CD8+ and CD4+ Idhigh+ cells was MHC class I (Dq/Lq) and class II (I-Aq) dependent, respectively. These results indicated that this TCR-transgenic mouse of H-2q background contains both classes of selecting MHC ligands for the transgenic TCR simultaneously. Further genetic analyses altering the gene dosage and combinations of selecting MHCs suggested novel asymmetric effects of class I and class II MHC on the positive selection of thymocytes. Implications of these observations in CD4+/CD8+ lineage commitment are discussed.     

Structure and site-directed mutagenesis of a flavoprotein from Escherichia coli that reduces nitrocompounds: alteration of pyridine nucleotide binding by a single amino acid substitution

The crystal structure of a major oxygen-insensitive nitroreductase (NfsA) from Escherichia coli has been solved by the molecular replacement method at 1.7-A resolution. This enzyme is a homodimeric flavoprotein with one FMN cofactor per monomer and catalyzes reduction of nitrocompounds using NADPH. The structure exhibits an alpha + beta-fold, and is comprised of a central domain and an excursion domain. The overall structure of NfsA is similar to the NADPH-dependent flavin reductase of Vibrio harveyi, despite definite difference in the spatial arrangement of residues around the putative substrate-binding site. On the basis of the crystal structure of NfsA and its alignment with the V. harveyi flavin reductase and the NADPH-dependent nitro/flavin reductase of Bacillus subtilis, residues Arg(203) and Arg(208) of the loop region between helices I and J in the vicinity of the catalytic center FMN is predicted as a determinant for NADPH binding. The R203A mutant results in a 33-fold increase in the K(m) value for NADPH indicating that the side chain of Arg(203) plays a key role in binding NADPH possibly to interact with the 2'-phosphate group.     

Essential motions and energetic contributions of individual residues in a peptide bound to an SH3 domain

We have studied protein-ligand interactions by molecular dynamics simulations using software designed to exploit parallel computing architectures. The trajectories were analyzed to extract the essential motions and to estimate the individual contributions of fragments of the ligand to overall binding enthalpy. Two forms of the bound ligand are compared, one with the termini blocked by covalent derivatization, and one in the underivatized, zwitterionic form. The ends of the peptide tend to bind more loosely in the capped form. We can observe significant motions in the bound ligand and distinguish between motions of the peptide backbone and of the side chains. This could be useful in designing ligands, which fit optimally to the binding protein. We show that it is possible to determine the different contributions of each residue in a peptide to the enthalpy of binding. Proline is a major net contributor to binding enthalpy, in keeping with the known propensity for this family of proteins to bind proline-rich peptides.     

Membrane binding of peptides containing both basic and aromatic residues. Experimental studies with peptides corresponding to the scaffolding region of caveolin and the effector region of MARCKS

We have studied the binding of peptides containing both basic and aromatic residues to phospholipid vesicles. The peptides caveolin(92-101) and MARCKS(151-175) both contain five aromatic residues, but have 3 and 13 positive charges, respectively. Our results show the aromatic residues insert into the bilayer and anchor the peptides weakly to vesicles formed from the zwitterionic lipid phosphatidylcholine (PC). Incorporation of a monovalent acidic lipid (e.g., phosphatidylserine, PS) into the vesicles enhances the binding of both peptides via nonspecific electrostatic interactions. As predicted from application of the Poisson-Boltzmann equation to atomic models of the peptide and membranes, the enhancement is larger (e.g., 10(4)- vs 10-fold for 17% PS) for the more basic MARCKS(151-175). Replacing the five Phe with five Ala residues in MARCKS(151-175) decreases the binding to 10:1 PC/PS vesicles only slightly (6-fold). This result is also consistent with the predictions of our theoretical model: the loss of the attractive hydrophobic energy is partially compensated by a decrease in the repulsive Born/desolvation energy as the peptide moves away from the membrane surface. Incorporating multivalent phosphatidylinositol 4, 5-bisphosphate (PIP(2)) into PC vesicles produces dramatically different effects on the membrane binding of the two peptides: 1% PIP(2) enhances caveolin(92-101) binding only 3-fold, but increases MARCKS(151-175) binding 10(4)-fold. The strong interaction between the effector region of MARCKS and PIP(2) has interesting implications for the cellular function of MARCKS.     

Interaction between bovine trypsin and a synthetic peptide containing 28 residues of the bait region of human alpha 2-macroglobulin.

The time course of the interaction between trypsin and a synthetic peptide corresponding to a segment (residues 676-703) of the bait region (residues 666-706) of human alpha 2-macroglobulin (alpha 2M) was studied by measuring the generation of cleavage products as a function of time by HPLC. Three primary cleavage sites for trypsin were present in the synthetic peptide. The fastest cleavage occurred at the bond corresponding to Arg696-Leu in alpha 2M with an estimated kcat/Km = 1-2 x 10(6) M-1.s-1. This value is of the same magnitude as that characterizing the interaction of alpha 2M and trypsin when taking into account the fact that alpha 2M is a tetramer, kcat/Km = 5 x 10(6) M-1.s-1 [Christensen, U. & Sottrup-Jensen, L. (1984) Biochemistry 23, 6619-6626]. The values of kcat/Km for cleavage at bonds corresponding to Arg681-Val and Arg692-Gly in alpha 2M were 1.5 x 10(5) M-1.s-1 and 1.3 x 10(5) M-1.s-1, respectively. Cleavage of intermediate product peptides was slower, with kcat/Km in the range 13-1.3 x 10(6) M-1.s-1. The value of Km determined for fast cleavage in the synthetic peptide was 8-10 microM. 1H-NMR spectroscopy indicated no ordered structure of the peptide. Hence, the very fast cleavage of the peptide is compatible with a loose structure that readily adopts a conformation favorable for recognition and cleavage by trypsin.     

Identification of essential residues in 2',3'-cyclic nucleotide 3'-phosphodiesterase. Chemical modification and site-directed mutagenesis to investigate the role of cysteine and histidine residues in enzymatic activity

2',3'-Cyclic nucleotide 3'-phosphodiesterase (CNP; EC ) catalyzes in vitro hydrolysis of 3'-phosphodiester bonds in 2',3'-cyclic nucleotides to produce 2'-nucleotides exclusively. N-terminal deletion mapping of the C-terminal two-thirds of recombinant rat CNP1 identified a region that possesses the catalytic domain, with further truncations abolishing activity. Proteolysis and kinetic analysis indicated that this domain forms a compact globular structure and contains all of the catalytically essential features. Subsequently, this catalytic fragment of CNP1 (CNP-CF) was used for chemical modification studies to identify amino acid residues essential for activity. 5,5'-Dithiobis-(2-nitrobenzoic acid) modification studies and kinetic analysis of cysteine CNP-CF mutants revealed the nonessential role of cysteines for enzymatic activity. On the other hand, modification studies with diethyl pyrocarbonate indicated that two histidines are essential for CNPase activity. Consequently, the only two conserved histidines, His-230 and His-309, were mutated to phenylalanine and leucine. All four histidine mutants had k(cat) values 1000-fold lower than wild-type CNP-CF, but K(m) values were similar. Circular dichroism studies demonstrated that the low catalytic activities of the histidine mutants were not due to gross changes in secondary structure. Taken together, these results demonstrate that both histidines assume critical roles for catalysis.     

Distinct mode of G protein activation due to single residue substitution of active IGF-II receptor peptide Arg2410-Lys2423: evidence for stimulation acceptor region other than C-terminus of Gi alpha.

Arg2410-Lys2423 (RVGLVRGEKARKGK, peptide 14) of the human insulin-like growth factor II receptor directly activates Gi and deletion of C-terminal 4 residues from peptide 14 nullifies this activity. A study was thus made of the effects of peptides modified in the C-terminal structure. RVGLVRGEKAAKGK and RVGLVRGEKARKGA scarcely activated Gi, whereas RVGLVRGEKARAGK (peptide A5) activated Gi as much as peptide 14 did. However, peptide A5 action did not depend on Mg2+ concentration and was little affected by pertussis toxin modification of Gi alpha. Peptide A5 may thus recognize the region on Gi alpha that is distinct from the extreme C-terminus. It is consequently considered that (i) the first and the last basic residues in the C-terminal motif of peptide 14 determine the capacity for recognition of Gi and (ii) there is a region different from the C-terminus of Gi alpha, through which the C-terminal second basic residue-altered peptide 14 activates Gi in a Mg(2+)-independent manner.     

iLoc-Hum: using the accumulation-label scale to predict subcellular locations of human proteins with both single and multiple sites

Although numerous efforts have been made for predicting the subcellular locations of proteins based on their sequence information, it still remains as a challenging problem, particularly when query proteins may have the multiplex character, i.e., they simultaneously exist, or move between, two or more different subcellular location sites. Most of the existing methods were established on the assumption: a protein has one, and only one, subcellular location. Actually, recent evidence has indicated an increasing number of human proteins having multiple subcellular locations. This kind of multiplex proteins should not be ignored because they may bear some special biological functions worthy of our attention. Based on the accumulation-label scale, a new predictor, called iLoc-Hum, was developed for identifying the subcellular localization of human proteins with both single and multiple location sites. As a demonstration, the jackknife cross-validation was performed with iLoc-Hum on a benchmark dataset of human proteins that covers the following 14 location sites: centrosome, cytoplasm, cytoskeleton, endoplasmic reticulum, endosome, extracellular, Golgi apparatus, lysosome, microsome, mitochondrion, nucleus, peroxisome, plasma membrane, and synapse, where some proteins belong to two, three or four locations but none has 25% or higher pairwise sequence identity to any other in the same subset. For such a complicated and stringent system, the overall success rate achieved by iLoc-Hum was 76%, which is remarkably higher than that by any of the existing predictors that also have the capacity to deal with this kind of system. Further comparisons were also made via two independent datasets; all indicated that the success rates by iLoc-Hum were even more significantly higher than its counterparts. As a user-friendly web-server, iLoc-Hum is freely accessible to the public at or . For the convenience of most experimental scientists, a step-by-step guide is provided on how to use the web-server to get the desired results by choosing either a straightforward submission or a batch submission, without the need to follow the complicated mathematical equations involved.     

iLoc-Plant: a multi-label classifier for predicting the subcellular localization of plant proteins with both single and multiple sites

Predicting protein subcellular localization is a challenging problem, particularly when query proteins may simultaneously exist at, or move between, two or more different subcellular location sites. Most of the existing methods can only be used to deal with the single-location proteins. Actually, multiple-location proteins should not be ignored because they usually bear some special functions worthy of our notice. By introducing the "multi-labeled learning" approach, a new predictor, called iLoc-Plant, has been developed that can be used to deal with the systems containing both single- and multiple-location plant proteins. As a demonstration, the jackknife cross-validation was performed with iLoc-Plant on a benchmark dataset of plant proteins classified into the following 12 location sites: (1) cell membrane, (2) cell wall, (3) chloroplast, (4) cytoplasm, (5) endoplasmic reticulum, (6) extracellular, (7) Golgi apparatus, (8) mitochondrion, (9) nucleus, (10) peroxisome, (11) plastid, and (12) vacuole, where some proteins belong to two or three locations but none has ≥ 25% pairwise sequence identity to any other in a same subset. The overall success rate thus obtained by iLoc-Plant was 71%, which is remarkably higher than those achieved by any existing predictors that also have the capacity to deal with such a stringent and complicated plant protein system. As a user-friendly web-server, iLoc-Plant is freely accessible to the public at the web-site or . Moreover, for the convenience of the vast majority of experimental scientists, a step-by-step guide is provided on how to use the web-server to get the desired results without the need to follow the complicated mathematic equations presented in this paper for its integrity. It is anticipated that iLoc-Plant may become a useful bioinformatics tool for Molecular Cell Biology, Proteomics, Systems Biology, and Drug Development.