| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
,1
* Cardiff School of Biosciences, Cardiff University, Museum Avenue, Cardiff, UK;
Department of Microbiology, Pasteur Institute, Paris, France;
Wuhan Institute of Virology, Chinese Academy of Sciences, Wuhan, China; and
Embrapa Recursos Geneticos e Biotecnologia, Parque Estação Biológica, Brasília, Brazil
2Correspondence: Cardiff School of Biosciences, Cardiff University, Museum Ave., Cardiff CF10 3US, UK. E-mail: berry{at}cf.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: insecticidal toxin • toxin evolution • toxin interactions
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, are well known for their production of various and specific Cry toxins, which constitute the main insecticidal factor in bioinsecticides and transgenic Bt-crops (2)
. In addition, strains of the bacterium Bacillus sphaericus also produce insecticidal toxins and are used to control mosquito vectors of human disease. These strains may produce mosquitocidal toxins named Mtx1, Mtx2, and Mtx3 during vegetative growth, but production levels are low and the proteins do not survive into the spores that are the active agent used in mosquito control (3
4
5)
. The major toxic moiety of commercial B. sphaericus strains is the spore-associated binary toxin (Bin), comprising two components; BinB that appears to have a major role in receptor binding (6
7
8)
and BinA that is able to form pores in target membranes (9
, 10)
. Although there is some evidence that BinA alone may be toxic in high doses (11)
, significant toxicity requires the presence of both proteins and is optimal at a 1:1 M ratio (12)
. Despite the excellent performance of B. sphaericus in the field, the presence of only the Bin toxin in spores has allowed insects to develop resistance (13
14
15)
that may limit its application or necessitate rotation with other insecticides. To identify B. sphaericus strains that may cope with resistance, screening of various isolates resulted in the detection of strains producing spores that are able to overcome laboratory resistance to the Bin toxin in Culex quinquefasciatus mosquito larvae (16
17
18)
. The presence of a protein of 
49 kDa (provisionally termed P49) in spore/crystal preparations of these strains, which is not observed in other B. sphaericus strains to which mosquito resistance has developed, led to the proposal that this might represent a new toxin from this species that could have an important role in the prevention of insect resistance (17)
. In this study, we characterize this protein and establish that its toxicity is dependent on a second protein. Interestingly, this second protein shares strong identity with known mosquitocidal toxins such as Cry4Aa from B. thuringiensis and this is the first report of the occurrence of a 3-domain family Cry-toxin in B. sphaericus. However, the in vivo activity studies indicate the necessity for a previously undescribed interaction between toxin components and suggest a novel mode of action to produce insect mortality. Thus, our findings could indicate that active genetic exchange between bacilli has occurred, probably within a mosquito host, and that this can give rise to new toxin gene combinations. This has wide significance, related to pathogen-host coevolution and the mobility/plasticity of virulence genes. This point is largely discussed in relation to the mode of action of this new two-component toxin. |
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
and pSTAB (also known as pPF-C5; ref. 20
) containing the STAB-SD region, which confers high mRNA stability (21)
were obtained from D. Lereclus (INRA, France) and B. Federici (University of California, Riverside, CA, USA), respectively. Bacilli were grown in the Embrapa medium [a nutrient broth/yeast extract based medium supplemented with a range of salts as described by Monnerat et al. (22)
] at 30°C, 250 rpm. E. coli strain DH5
was grown in LB medium at 37°C, 250 rpm.
SDS-PAGE and N-terminal sequencing
Cultures of bacilli were grown to >90% sporulation, as assessed by microscopic observation (typically 48–72 h), and the spore/crystal mixture was harvested by centrifugation. The pellet was resuspended in SDS-PAGE loading dye and boiled for 5 min, and the proteins were separated in 12% acrylamide resolving gels (23)
before visualization by staining with Coomassie blue. For N-terminal sequencing, gels were run under identical conditions, except that glycine in the running buffer was replaced with tricine. Proteins were transferred electrophoretically from the gel onto polyvinylidene difluoride membranes for rapid staining to identify proteins of interest. Selected protein bands were excised from the membrane with a clean scalpel for N-terminal sequencing (Alta Bioscience, Birmingham, UK).
Southern hybridization
Total DNA was isolated from B. sphaericus strain IAB59 (24)
before treatment with restriction endonucleases and separation of digest products by agarose gel electrophoresis. Resolved DNA fragments were transferred to nylon membrane (25)
and fixed by baking (80°C, 2 h). Hybridizations were performed at 10°C below the minimum Tm for each 3' digoxigenin-ddUTP degenerate oligonucleotide in hybridization buffer [5x SSC, 0.1%(w/v) N-lauroylsarcosine, 0.2%(w/v) SDS, 1%(w/v) dry skimmed milk)] containing 10 pmol/ml probe. Low stringency washes (2x SSC, 0.1%(w/v) SDS; 2x5 min, 25°C) and high stringency washes (0.5x SSC, 0.1%(w/v) SDS; 2x15 min, hybridization temperature) were followed by detection of probe using anti-DIG antibody and CSPD (Roche Diagnostics, Mannheim, Germany) according to the manufacturer’s recommendations. Colony hybridizations (26)
were performed using the same hybridization conditions as described for Southern blot.
PCR amplification and cloning of putative toxin genes
PCR amplification was achieved using the oligonucleotides described in Table 1
and EasyA polymerase (Stratagene, La Jolla, CA, USA) under the following conditions: 95°C, 5 min followed by 15 cycles of 95°C, 1 min; 60°C, 1 min; 72°C, 3.5 min. For cloning into the pHT304 shuttle vector, resulting amplicons were digested with BamHI and ligated into the BamHI site of the vector. For ligation into pSTAB, both amplicons and vector were first digested with SalI and SphI. After screening to identify clones containing the desired inserts and sequencing to ensure that no PCR mutations had occurred, plasmids were transferred to B. thuringiensis 4Q7 by electroporation and recombinant colonies were selected by growth on LB agar containing 15 µg/ml erythromycin.
|
Electrotransformation of B. thuringiensis
B. thuringiensis subsp. israelensis 4Q7 was grown to an attenuance at 600 nm of 0.2–0.3 in 10 ml LB medium (30°C, 250 rpm). Cells were harvested by centrifugation (2,000 g, 5 min) and washed twice in ice-cold 10% (w/v) sucrose before final resuspension in 250 µl of the same solution. Plasmid DNA (1 µg) was added to 120 µl of the cell suspension in a 0.4 cm electroporation cuvette, and 800 µl of buffer were added before electroporation (Gene Pulser, Bio-Rad, Hercules, CA, USA; 1.8 kV, 400 
resistance and 25 µF capacitance). LB medium (1 ml) was added to the cells before incubation at 30°C, 200 rpm for 1 h and spreading onto LB agar plates containing 15 µg/ml erythromycin.
Crystal purification
Sporulated cultures of B. sphaericus strains or recombinant B. thuringiensis subsp. israelensis 4Q7, carrying plasmids containing genes encoding crystal toxin protein, were harvested by centrifugation and washed once in 1 M NaCl and 10 mM EDTA and twice in 10 mM EDTA before final resuspension in 1/30 original culture volume of 10 mM EDTA. The suspension was sonicated (Vibra-Cell VCX500 Ultrasonic Processor (Sonics, Newtown, CT, USA), 50% amplitude, 3x30 s) and layered onto a discontinuous sucrose gradient 67%/72%/79%/84% (w/v) before centrifugation (110,000 g, 15°C, 16 h). Crystals were collected from the interfaces and washed thoroughly in sterile distilled water.
Bioassays
Qualitative bioassays
Ten second or third instar C. quinquefasciatus mosquito larvae were exposed to crystal protein (20 µg) or sporulated cultures of recombinant B. thuringiensis subsp. israelensis 4Q7 (100 µl) in 10 ml of distilled water. Bioassays were performed in triplicate and incubated at 28°C. Mortality was assessed after 24 h by counting the number of living larvae. Assays were carried out using SLCq larvae sensitive to the B. sphaericus Bin toxin and RLCq/C3–41, a strain with Bin toxin resistance, developed under laboratory conditions by selection with B. sphaericus strain C3–41 that only produces Bin toxin on sporulation (16)
.
Quantitative bioassays
Third instar C. quinquefasciatus larvae were exposed to serial dilutions of crystal protein in plastic cups containing 25 larvae in 100 ml of distilled water for each concentration of crystal protein tested. At least five concentrations yielding between 5–95% mortality were tested, and mortality was recorded after 48 h. Each experiment, performed in triplicate, contained a control group of larvae exposed only to water and was repeated three times on different days. The 50% lethal concentration was determined using probit analysis (27)
with the software package Micro Probit 3.0. Crystal proteins for bioassay were prepared from recombinant B. thuringiensis using sucrose density gradients, and the amount of crystal protein was estimated by densitometric scanning of SDS-PAGE gels on which a range of concentrations of BSA had also been run for comparison.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 18)
, a protein of 49 kDa (therefore given the interim name of P49) was seen in SDS-PAGE gels of lysed sporulated culture preparations from B. sphaericus strains IAB59 and NHA15b. A second band of 43 kDa (P43) was also observed in these strains (Fig. 1
A). Since insecticidal toxins are frequently produced as parasporal crystals, strain NHA15b was examined and showed parasporal bodies that appeared to be located outside the exosporium. This strain, which does not produce Bin toxin crystals, was, therefore, used to purify parasporal bodies from sucrose density gradients. The P49 was predominant at the 72%/79% sucrose interface, and a larger protein of 135 kDa (given the interim name P135) was also seen in the crystal enriched fraction from the 79%/84% sucrose interface from the gradients (Fig. 1B
shows the P135 enriched fraction). The yield of this protein was much lower than that of P49, which accounts for the absence of a P135 band in Fig. 1A
. N-terminal sequencing of the P43, P49, and P135 bands yielded the sequences NNDPG, MENQIKEEFNKNNHGIPSDCSCIKE and XDINN, respectively (where X represents ambiguous sequence). From the P49 sequence, two degenerate oligonucleotide sequences were designed: DP491 and DP492 (Table 1)
, which were used to probe Southern blots of B. sphaericus IAB59 total DNA digested with HindIII. A fragment of 15.5 kb hybridized to the labeled oligonucleotide probe (results not shown), and fragments of this size were eluted from an agarose gel and ligated into HindIII cut, phosphatase treated pUC18 vector. The entire fragment was also recloned into the HindIII site of the pHT304 vector and transformed into B. thuringiensis 4Q7. A single point bioassay using a 1/100 dilution of a 72 h sporulated culture of this transformant grown in Embrapa medium caused 100% mortality in C. quinquefasciatus larvae, whereas control, nontransformed B. thuringiensis 4Q7 produced no larval death. This established the presence of mosquitocidal toxin genes within the cloned fragment.
|
The inserted DNA was subjected to sequencing using vector-based primers and a series of internal primers designed to allow "sequence walking" through the entire 15,649 bp fragment. The sequence of this cloned region has been deposited in the EMBL database with accession number AJ841948.
Sequence analysis
Sequence analysis was performed using the visualization and annotation tool Artemis (28)
and AMIGene (29)
to identify coding sequences (CDSs) in the DNA. BLAST (30)
and conserved domain (31
, 32)
searches were performed in attempts to assign putative functions to CDSs. The schematic diagram of the region (Fig. 2
) shows the 16 CDSs that were identified, 11 of which showed similarity with transposase sequences (similarities described in more detail in Supplemental Table 1). Of particular note, however, were CDS3 and CDS7, which appear to encode P49 and P135 respectively. The predicted product of CDS15 also appears to be related to known B. thuringiensis toxins but seems to be a deletion pseudogene with similarities to the C termini of Cry20Aa and Cry11Bb (see Supplemental Table 1).
|
CDS3 commences with a TTG initiation codon and encodes a 464 amino acid protein with a predicted molecular mass of 53.3 kDa. The first 25 residues of the deduced amino acid sequence of this protein exactly match those determined by N-terminal sequencing of the P49 protein. In addition, residues 76–80 were an exact match to the sequence derived from the N terminus of P43, indicating that this is a processed form of P49. Comparison of the protein with sequences in the publicly available databases showed the only significantly related proteins to be the BinA & BinB proteins of the B. sphaericus binary toxin (30% pairwise identity with each Bin protein) and Cry35 and Cry36 from B. thuringiensis (20 and 34% pairwise identity, respectively) and a putative protein from Chlorobium phaeobacteriodes (accession number YP_911930; 15% identity; Fig. 3
).
|
CDS7 is transcriptionally convergent with CDS3 and encodes a second toxin homologue with a molecular mass of 135.6 kDa and the N-terminal sequence MDINN, thus establishing CDS7 as the gene encoding P135. Conserved domain searches revealed the characteristic features of three-domain Cry toxins of Bacillus thuringiensis (CDD entries: pfam03945, pfam03944, and pfam00555), and the protein was most closely related (33% identity) to the sequences of the known mosquitocidal toxins Cry4Aa1 and Cry4Ba1 (alignment shown in Supplemental Fig. 1
). Sequences of CDSs 3 and 7 were submitted to the Bacillus thuringiensis delta-endotoxin nomenclature committee and were assigned the cry gene designations cry48Aa1 (CDS7) and cry49Aa1 (CDS3), used hereafter.
Recombinant expression of Cry49Aa and Cry48Aa
A fragment of 1.6 kb encompassing the entire cry49Aa gene along with 201 bp of upstream sequence containing its putative promoter was amplified from the HindIII clone by PCR using primers P49ProF and P49R (Table 1)
. A BamHI clone in the vector pHT304 was constructed and designated pHTP49.
A fragment of 3.6 kb containing the entire cry48Aa gene along with a putative promoter sequence was amplified from the HindIII clone by PCR using primers BamP135F and BamP135R (Table 1)
. A BamHI ligation into pHT304 produced plasmid pHTP135. As a result of the very low level of production of Cry48Aa from pHTP135, the cry48Aa gene was amplified, along with 39 nucleotides of upstream sequence containing its ribosome binding site, using primers PSTABF and PSTABR and ligated into the Bacillus expression vector pSTAB to form plasmid pSTABP135.
B. thuringiensis 4Q7 was transformed with the above constructs and grown in Embrapa medium to >99% sporulation, as assessed by phase contrast microscopy (72 h). SDS-PAGE of the lysed sporulated culture preparation showed high-level production of Cry49Aa from its own promoter in the pHT304 vector, but Cry48Aa yield from pHTP135 was very low even after sucrose density gradient purification (Fig. 4
, lane 2). Production of Cry48Aa from the pSTAB vector was greater (Fig. 4
, lane 3) but still much less than for Cry49Aa (Fig. 4
, lane 1). In the latter case, a band with an apparent molecular mass a little over 83 kDa was seen in the crystal preparations (marked with a star, Fig. 4
, lane 1). Analysis of this band by Edman degradation revealed an identical N-terminal sequence to Cry49Aa, indicating that this band represents a dimer, stable under the conditions of SDS-PAGE. Both Cry48Aa and Cry49Aa proteins were produced in the form of small (<1 µm) spore-associated crystals that showed a bipyramidal morphology (Cry49Aa) or formed amorphous crystals (Cry48Aa). These were located outside the exosporium (as seen by electron microscopy; Fig. 5
) as is usual for B. thuringiensis-produced Cry toxins, and analysis of B. sphaericus NHA15b indicated that in this strain too Cry48 and Cry49 are located outside the exosporium (results not shown), which is not the case for Bin toxin crystals produced in B. sphaericus, which are found inside the exosporium (although on production in recombinant B. thuringiensis Bin is deposited outside the exosporium; ref. 33
).
|
|
Bioassay
Despite their similarities to previously identified insecticidal toxins, neither Cry48Aa nor Cry49Aa was toxic to C. quinquefasciatus larvae at up to 2.8 and 6.0 µg/ml, respectively (results not shown). Since Cry49Aa is similar to the Bin toxins of B. sphaericus, we speculated that it might act in concert with existing Bin proteins. Therefore, its toxicity was tested in combination with B. sphaericus strain C3–41 that produces Bin toxins but not Cry48Aa or Cry49Aa. As expected, this combination showed toxicity to Bin-sensitive SLCq mosquitoes (through the action of the Bin toxins) but no toxicity toward the Bin-resistant C. quinquefasciatus strain RLCq/C3–41 (16)
was observed, thus indicating that Cry49Aa does not act with Bin proteins to overcome Bin resistance. In separate bioassays of Cry48Aa in combination with individual BinA or BinB proteins, there was also no toxicity to RLCq/C3–41 larvae.
Assay of Cry48Aa and Cry49Aa in combination did, however, produce toxicity to both Bin-sensitive (SLCq) and Bin-resistant (RLCq/C3–41) C. quinquefasciatus larvae. Quantitative bioassays using a 1:1 M ratio of both proteins were then performed, and an LC50 value at 48 h for this mixture was calculated by Probit analysis. This LC50 value for the equimolar mixture corresponded to 6.3 ng/ml (confidence limits 5.0–7.7 ng/ml) Cry49Aa and 15.9 ng/ml (confidence limits 12.7–19.6 ng/ml) Cry48Aa. To analyze the contribution of each component to the activity, bioassays were repeated with Cry49Aa:Cry48Aa molar ratios of 1:10 and 10:1. These experiments produced LC50 values corresponding to 4.4 ng/ml Cry49Aa (confidence limits 3.9–5.6 ng/ml) and 112 ng/ml (confidence limits 99.5–129 ng/ml) Cry48Aa and 74.4 ng/ml (confidence limits 57.3–109 ng/ml) Cry49Aa and 18.8 ng/ml (confidence limits 14.5–27.6 ng/ml) Cry48Aa respectively. In each case, therefore, the LC50 is attained when the protein present in the lowest molar quantity reaches that seen in the 1:1 bioassay. This indicates that the optimal toxicity is likely to arise from a ratio of 1:1.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. Despite 33% identity with known mosquitocidal toxins such as Cry4Aa, Cry48Aa alone is unable to produce toxicity in C. quinquefasciatus mosquitoes. Instead, its toxicity appears to be reliant on the presence of a second protein (Cry49Aa), with optimal activity occurring at a 1:1 M ratio of the two proteins. Cry49 is most closely related to Cry36 from B. thuringiensis and also to both components of the Bin toxin of B. sphaericus and their other previously described relative, Cry35 from B. thuringiensis as well as a putative protein from C. phaeobacteriodes. An alignment of these protein sequences (Fig. 3)
indicates that the blocks of the conserved sequence, previously noted between BinA and BinB proteins (34
, 35)
, are also well conserved throughout this whole family of proteins. Conservation is far lower toward the N and C termini of the family members, but this is consistent with the fact that BinA and BinB are still active after deletions from their N termini of up to 16 and 34 residues, respectively, and up to 17 and 54 residues from their respective C-termini (36
37
38)
.
Cry36 can act alone to produce weak activity against coleopteran larvae (39)
, whereas other members of this family require partner proteins to produce toxicity. Cry35 is part of a binary toxin encoded on a single operon with its 14 kDa Cry34 partner protein (itself not related to other Cry or Bin toxins; refs. 35
, 40
). No Cry34 homologues appear to be encoded in association with Cry49Aa. It is notable that the optimum pore-forming ability in lipid bilayers is achieved at a 3:1 M ratio of Cry34/Cry35, in contrast to the optimum 1:1 ratio for maximum toxicity for both the BinA/BinB (12)
and Cry48Aa/Cry49Aa combinations. Cry49Aa exhibited dimers on SDS-PAGE (Fig. 4)
, a phenomenon that has also been reported for the Bin toxins (41)
, whereas Cry35Ab has been reported to form much larger multimers (>50-mers; ref. 42
).
It is interesting to speculate on the mode of action of the Cry48Aa/Cry49Aa pair. Both 3-domain toxins (43)
and the BinA component of the B. sphaericus binary toxin (9)
are able to form pores in lipid bilayers. However, BinB (9)
and Cry35Ab (42)
are poor pore formers. BinB is the receptor binding component in the Bin toxin in C. quinquefasciatus (6)
, whereas Cry35Ab appears to destabilize membranes (42)
. At this point, it is not clear which of the Cry48Aa/Cry49Aa pair provides receptor binding and putative pore-forming properties. However, a novel interaction would seem to be required since the requirement for a 3-domain protein for an interaction with an accessory protein to confer toxicity has never been described previously. Similarly, a binary-like protein such as Cry49Aa has not been shown to require a 3-domain toxin to produce insect mortality. Our results show that the presence of both Cry48Aa/Cry49Aa in strain IAB59 may explain why this strain can overcome resistance toward Bin toxin in C. quinquefasciatus colonies that are resistant to B. sphaericus due to an abolition of toxin binding (44)
.
This new Cry48Aa/Cry49Aa toxin pair has a high potency when administered in purified protein form (LC50: 15.9 and 6.3 ng/ml, respectively), which is comparable to that of the Bin toxins from B. sphaericus (LC50: 30 ng/ml each of BinA and BinB; ref. 7
). Despite this fact, the contribution of this toxin to B. sphaericus spore associated toxicity appears to be rather low since the potency of strains such as IAB59 producing both Bin and Cry48/Cry49 is 10 fold lower to Bin-resistant mosquitoes than to Bin-sensitive ones (16
, 17)
. A likely cause for this phenomenon is the low level of accumulation of Cry48Aa in B. sphaericus (as also seen in our B. thuringiensis recombinants). This potential "underuse" of the Cry48/Cry49 toxin leads us to speculate that the Cry48Aa/Cry49Aa pair may represent a relatively recent toxin combination. The convergent orientation of the cry48Aa and cry49Aa genes with several intervening CDSs rather than the operon structure seen with the bin and cry34/cry35 genes (34
, 35)
also supports this conjecture. Like many insecticidal toxin genes in bacilli (45)
, cry48Aa and cry49Aa are associated with neighboring transposase sequences that may provide a clue to their evolutionary acquisition via transposition. Genetic movement between B. thuringiensis strains (46
, 47)
and from B. thuringiensis to B. sphaericus (48)
has been shown to be feasible and the potential coexistence of both species in mosquito breeding habitats may have led to the combination of a B. thuringiensis–like 3-domain protein in association with a B. sphaericus Bin-like protein. Synergistic interactions between Bin and Cry toxins from B. thuringiensis subsp. israelensis have been demonstrated previously (49)
and although we observed no toxicity for individual Bin proteins in combination with Cry48Aa, perhaps the potential interactions between the two protein families have, in the case of Cry48Aa/Cry49Aa, developed further to produce a new toxin partnership.
Thus, the novel toxin pair described in this work may illustrate the adaptability of the toxin arsenals carried by the bacilli and their ability to recruit new combinations of proteins to develop new and potent insecticidal agents.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication April 27, 2007. Accepted for publication June 14, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H.-W. Park, M. Tang, Y. Sakano, and B. A. Federici A 1.1-Kilobase Region Downstream of the bin Operon in Bacillus sphaericus Strain 2362 Decreases Bin Yield and Crystal Size in Strain 2297 Appl. Envir. Microbiol., February 1, 2009; 75(3): 878 - 881. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X. Hu, W. Fan, B. Han, H. Liu, D. Zheng, Q. Li, W. Dong, J. Yan, M. Gao, C. Berry, et al. Complete Genome Sequence of the Mosquitocidal Bacterium Bacillus sphaericus C3-41 and Comparison with Those of Closely Related Bacillus Species J. Bacteriol., April 15, 2008; 190(8): 2892 - 2902. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
