OUP user menu

Identification of mature appressorium-enriched transcripts in Magnaporthe grisea, the rice blast fungus, using suppression subtractive hybridization

Jian-Ping Lu, Tong-Bao Liu, Fu-Cheng Lin
DOI: http://dx.doi.org/10.1016/j.femsle.2005.02.032 131-137 First published online: 1 April 2005


We have constructed a fungal subtractive suppressive library enriched in genes expressed during appressorium maturation in Magnaporthe grisea. Sequencing of 250 clones from the subtracted appressorium cDNA library revealed 142 unique genes, represented by 155 non-redundant ESTs (expressed sequence tags). Of these ESTs, 72 have not been previously isolated in M. grisea. RT-PCR analysis of 105 of the genes discovered found transcripts corresponding to 71 of the ESTs only in mature appressoria while transcripts corresponding to a further 34 of the isolated ESTs were expressed both in appressoria and conidia/mycelia. Genes specifically expressed in appressorium identified by SSH included a number that have been previously implicated in appressorium formation or function including GAS1, GAS3, and PTH11.

Key words
  • Appressorium
  • cDNA
  • Suppression subtractive hybridization
  • Gene expression profile
  • Rice blast fungus

1 Introduction

Rice blast disease, caused by Magnaporthe grisea, is one of the most severe diseases of rice throughout the world [1]. This fungal pathogen, as one of the best-studied species among phytopathogenic fungi, has been used as a primary model for elucidating various aspects of the host–pathogen interaction with its host [24]. This fungus as well a number of fungal plant pathogens can penetrate plant tissues by using the appressorium, a specialized cellular structure [5]. Appressorium formation appears to be a complex process from initiation to maturation [46].

Once attached to the rice surface, conidia of the fungus quickly germinate forming short germ tubes in 0.5–1.5 h. The germ tube then stops growing at the tip and a terminal “hooking” of hypha appears within 4 h. At approximately 4–8 h, melanization of the appressorium begins, followed by appearances of abundant glycogen rosettes and an appressorium pore ring within 16–24 h. At 24–31 h, glycogen rosettes are nearly absent, appressorial turgor pressure is at a peak, and penetration pegs emerge, infection hyphae form in 31 h and spread to adjacent epidermal cells in 48 h [6]. In view of the timing of the appressorium formation process, as outlined above, 24 h may be a key time point for production of turgor pressure and synthesis of the materials essential for the function of the M. grisea appressorium. Genes expressed during this stage may be required for appressorium maturation and/or penetration of the plant via the appressorium. For this reason we set out to identify genes expressed specifically in the late appressorium stage, but not expressed, or expressed at a low level, within conidia/mycelia.

Many approaches, such as expressed sequence tags (EST) sequencing [79], serial analysis of gene expression (SAGE) [10], microarray [11] and the isolation of appressorium defective or deficient mutants [1214], have been used to identify appressorium specific genes. To date, several genes involved in fungal infection, specifically in appressorium formation and subsequent plant penetration, have been identified and characterized [4,15,16]. Despite these advances, the genetic basis of turgor generation within the appressoria of M. grisea is still poorly understood [17].

Suppression subtractive hybridization (SSH) technology is a powerful approach for identifying genes differentially expressed by cells or organisms in specific development stages or environmental conditions. For example, 35 transcripts showing a significant increase in expression during early stages of germination in Pyrenophora teres[18] and 12 genes whose transcripts are significantly enriched in Aspergillus nidulans conidia [19] were identified by SSH. A late appressorium cDNA library of rice blast fungus had been constructed in our lab [20]. Here, we report the construction of a subtractive suppressive cDNA library and discovery of transcripts uniquely expressed during appressorium maturation.

2 Materials and methods

2.1 Fungal cultures

M. grisea GUY-11 was used in this study. Conidia were obtained by harvesting plate cultures of the fungus grown on complete medium [21] in a 14-h-light and 10-h-dark cycle at 25 °C for 12–14 days. Conidia were harvested by scraping sporulating cultures with a glass rod in sterile distilled water, followed by centrifugation at 1000g for 10 min and resuspension in distilled water to a concentration of 1–1.5 × 106 spores/ml. Drops (20 μl) of this standard inoculum were applied onto the hydrophobic surface of projection transparency film (terylene resin; Gaoke, China) and incubated at 25 °C for 24 h. M. grisea aerial mycelia were obtained using the method of Talbot et al. [21]. Substrate mycelia of M. grisea were harvested by collecting cultures of the fungus grown in liquid complete medium at 25 °C with shaking at 180 rpm for 3 days.

2.2 RNA isolation and cDNA synthesis

Equal weights of conidia, aerial mycelia and substrate mycelia were mixed and quickly frozen in liquid N2 for RNA extraction. Total RNA of the above mixture and the appressoria harvested from duplicate films was isolated separately by a Trizol method following the manufacturer's procedure (Molecular Research Center, Inc., USA). Three microlitres of total RNA (?15 μg total RNA for conidia/mycelia mixture and ?5 μg total RNA for appressoria) was used for the synthesis of double-stranded cDNA with a SMART cDNA library construction kit (Clontech, USA). The protocol for SMART cDNA synthesis by LD PCR supplied with the kit was followed, until the proteinase K digestion step.

2.3 cDNA subtraction and cloning

For a cDNA library of genes specially expressed during appressorium stage, cDNA subtractive hybridization was carried out by following the protocol of the PCR-select cDNA subtraction kit (Clontech, USA) except for the use of AluI digestion of cDNA instead of RsaI digestion. The subtracted cDNAs were cloned into the pBlueScript II SK (+) Vector (Stratagene, USA).

2.4 DNA sequencing and analysis

The recombinant cDNA clones were directly sequenced on an ABI 377 DNA sequencer (Applied Biosystems, USA) with T7 primer. The sequence data were analyzed using VecScreen program (NCBI) for vector masking, and then the adaptor sequences (adaptor 1 and adaptor 2R) were removed from these cDNA sequences. These EST sequences were processed using software BioEdit [22] for contig assembly. Processed sequences were subjected to similarity searches against M. grisea database (genome ver. 2) (http://www.broad.mit.edu/cgi-bin/annotation/magnaporthe/) using Blast version 2.2.1 [23]. Once the genomic regions corresponding to EST sequences were identified, a representative EST sequence was selected for each gene identified by one or more detached EST sequence(s). All representative EST sequences were subjected to blastn and blastx search against GenBank database using Blast 2.2.8 [23] and against phytopathogenic fungi and Oomycete EST database (version 1.4) in COGEME (http://cogeme.ex.ac.uk/index.html) using Blast 2.2.5.

2.5 RT-PCR analysis

For detection of differential expression by RT-PCR, 3 μl of total RNA (?15 μg RNA for conidia/mycelia mixture and ?5 μg RNA for appressoria) was reverse-transcribed into first-strand cDNA by the protocol for SMART cDNA synthesis using SMART cDNA library construction kit (Clontech, USA). The primer sequences of RT-PCR used to detect mRNAs for the 105 clones were designed according to the EST sequences or predicted protein-coding sequences in M. grisea database (genome ver. 2) (http://www.broad.mit.edu/cgi-bin/annotation/magnaporthe/). After the initial denaturation (2 min at 94 °C), PCR was run for 35 cycles of 60 s at 94 °C, 30 s at 55 °C and 60 s at 72 °C. PCR products were separated on a 1.2% agarose/EtBr gel.

3 Results

3.1 Construction of subtractive cDNA library

Using microscopic assessment it was found that after ?24 h of incubation, melanized appressoria had developed in most of cells. The ratio of appressorium formation (appressoria numbers/conidia numbers) on the hydrophobic surface of duplicate film was 96%.

Total RNA of the conidia/mycelia mixture and the mature appressoria, incubated for 23.5–24.5 h on duplicate films, was isolated separately by a Trizol method. RNA quality was assessed using electrophoretic analysis which indicated that the total RNA was intact and suitable for use. From this RNA, a subtractive appressoria cDNA library, subtracted by conidia, aerial mycelia, and substrate mycelia mixture, was constructed. A total of 338 recombinant cDNA clones, from the subtracted cDNA mixture ligated to vector, were stored in −20 °C.

3.2 Sequence analysis

Among 338 subtractive cDNA clones, 250 clones were partially sequenced using T7 primer, and 155 non-redundant ESTs were generated from these cDNA sequences using BioEdit software [22]. Theses 155 sequences were compared to the draft sequences of the M. grisea genome ver. 2 (http://www.broad.mit.edu/cgi-bin/annotation/magnaporthe/), using the Blast algorithm [23]. 119 contigs, matching 152 ESTs were identified within the M. grisea database (Broad Institute), while no matching contigs could be found for the remaining three ESTs (Table 1). Some non-redundant ESTs, which were classed as unique by alignment among the ESTs, were found to match the same predicted M. grisea gene. This analysis revealed a match to a total of 139 different predicted genes from the M. grisea draft genome annotation. Assuming that the three ESTs which did not match any sequence present in the M. grisea database represent a further three separate genes absent form the draft genome sequence, a total of 142 genes in total have been shown to be expressed within maturing appressorium here.

View this table:
Table 1

Blast similarity alignments of 155 cDNA EST sequences against the Magnaporthe grisea database (Broad Institute)

High?e-28152a (139b)
  • a119 Contigs matching 152 ESTs were identified.

  • bThe values in parenthese indicate the number of genes that could be identified by comparing the EST sequences with the predicted gene sequences from 119 contigs.

A summary of homology search against GenBank using the Blast algorithm [23] and analysis of 142 cDNA sequences, representing 142 unique genes, is shown in Table 2. The source organism of every matching record of all homology searches in GenBank using blastn were checked and counted, and it was found that, among 142 ESTs, 70 ESTs have been correspond to known genes or have been previously sequenced in M. grisea while a further 72 ESTs have been isolated for the first time in the current study. Eight-nine per cent (when compared with protein databases) of the 142 ESTs had significant matches (p < 0.001) to known (or predicted) genes present in GenBank at the time of submission.

View this table:
Table 2

Sequence similarities between ESTs and the best match in the GenBank database using blastn and blastx at the time of submission

ModerateE-3 to e-291537

Among the 142 genes identified in 250 cDNA clones, the ESTs of 40 genes (28%) were detected at least twice. The most abundant 10 ESTs are listed in Table 3. Among these, the most frequently detected EST (16 times, 6.4%) and a further four ESTs show significant similarity to previously characterized genes or genes whose predicted product can be functionally categorized based on homology to known proteins. An EST matching probable keto acyl reductase encoding gene, involved in fatty acid metabolism, was detected nine times while ESTs matching the GAS1 (MAS3) gene, encoding a suspected virulence factor, were also detected seven times.

View this table:
Table 3

List of the 10 most abundant ESTs

ESTsaPutative gene productsNumber of clones
ESTs119 (CK828201)Predicted protein (MG02778.4)b16 (6.4%)c
ESTs155 (CK828227)Probable keto acyl reductase (MG10351.4)9 (3.6%)
ESTs98 (CK828185)Membrane-associated or secreted protein (MG02884.4)7 (2.8%)
ESTs105 (CK828190)Predicted protein (MG08526.4)7 (2.8%)
ESTs56 (CK828275)MAS3 protein (MG07044.4)7 (2.8%)
ESTs9 (CK828294)Cytochrome P-450-alk1 (MG06973.4)6 (2.4%)
ESTs28 (CK828256)Predicted protein (MG02287.4)6 (2.4%)
ESTs134 (CK828214)Predicted protein (MG10355.4)6 (2.4%)
ESTs19 (CK828247)Predicted protein (MG10345.4)5 (2.0%)
ESTs127 (CK828209)Neuronal calcium sensor (MG01550.4)4 (1.6%)
  • aSequences were assembled from 250 ESTs randomly sequenced from a cDNA library constructed from appressorium cDNA (incubating for 23.5–24.5 h on an inductive surface) subtracted by cDNA of conidia/mycelia in PCR-select cDNA subtractive hybridization method. GenBank Accession Nos. are in parentheses.

  • bGene names in Magnaporthe grisea database (Broad Institute) that correspond to ESTs are in parentheses.

  • cPercentage in parentheses was calculated based on 250 cDNA clones analyzed.

3.3 RT-PCR analysis

To confirm differential gene expression in appressoria and conidia/aerial mycelia/substrate mycelia as indicted by SSH subtraction strategy, the transcript abundance of 105 of the genes identified was assessed using RT-PCR. Among 105 genes examined, 71 were expressed only in appressoria while 34 genes were expressed both in appressoria and conidia/mycelia (Table 4).

View this table:
Table 4

Results of RT-PCR analysis of gene expression in mature appressoria or in conidia/mycelia mixture

Expression stagesAppressoriaAppressoria, conidia/myceliaTotal
Gene numbers7134105

Fig. 1 shows the gene expression analysis for nine genes in appressoria or in conidia/mycelia mixture of M. grisea. These genes, which include a predicted glyoxalase I-encoding gene (clone s126: GenBank Accession No. CK828208) and two ESTs predicted to encode a protein of unknown function (clone s134 and clone s119: GenBank Accession Nos. CK828214 and CK828201), seem to be expressed specifically in appressorium while the transcript is absent from conidia/mycelia. The EST matching the gene coding for the GAS1 (MAS3) protein (clone s56: GenBank Accession No. CK828275) was found expressed strongly in appressorium and very weakly in conidia/mycelia mixture. Two other genes, predicted to encode a product with no similarity to any known protein (clone s7: GenBank Accession No. CK828283) and a gene whose hypothetical product is related to acetyl coenzyme A synthetase (clone s117: GenBank Accession No. CK828240) seems to be expressed at a higher level within appressoria. The genes predicted to code for a predicted potassium channel beta chain (clone s20: GenBank Accession No. CK828252), an alcohol oxidase (clone s48: GenBank Accession No. CK828270) and a membrane-associated or secreted protein (clone s98: GenBank Accession No. CK828185) were highly expressed both in appressoria and conidia/mycelia. These results indicated that our SSH approach was successful in revealing differential gene expression during the maturation of M. grisea appressoria.

Figure 1

RT-PCR analysis of gene expression of nine genes in mature appressoria or in conidia/mycelia mixture of M. grisea. The same capital or small letter indicated RT-PCR product from total RNA of product from mature appressoria or conidia/mycelia mixture. M indicated GenRulerTM 100 bp DNA ladder (0.5 μg/lane) (MBI, Lithuania). These nine genes were predicted to code for (the gene names in Magnaporthe grisea database (Broad Institute) that correspond to the following ESTs are in parentheses): A or a, s134 sequence coding for a protein of unknown function (MG10355.4); B or b, s117 sequence coding for a protein related to acetyl coenzyme A synthetase (MG00689.4); C or c, s119 sequence coding for a protein of unknown function (MG02778.4); D or d, s7 sequence coding for a predicted protein of unknown function (MG10345.4); E or e, s98 sequence coding for a membrane-associated or secreted protein (MG02884.4); F or f, s56 sequence coding for MAS3 protein (MG07044.4); G or g, s48 sequence coding for alcohol oxidase (MG09072.4); H or h, s20 sequence coding for potassium channel beta chain (MG06182.4); I or i, s126 sequence coding for glyoxalase I S-D-lactoylglutathione lyase (MG10350.4).

4 Discussion

Because the RNA obtained from appressoria was limited, the protocol for cDNA synthesis of mycelia/conidia and appressoria by LD PCR used in the construction of SSH subtracted cDNA library was different from the protocol provided by clontech PCR-select cDNA subtraction kit (Clontech, USA) but identical, with that used earlier in the construction of appressoria cDNA library [20]. According to the restriction enzyme cut site analysis of 8821 unique EST sequences of M. grisea obtained from COGEME 1.5 (http://cogeme.ex.ac.uk/), the number of AluI cut sites in the cDNAs synthesized with SMART cDNA library construction kit is nearly many as the number of RsaI cut sites in the cDNAs synthesized with clontech PCR-select cDNA subtraction kit (Table 5). If RsaI had been used to digest cDNA as suggested by cDNA subtraction kit (Clontech, USA), the subtracted cDNA library would lose its diversity greatly, because the cDNAs synthesized in this study only have 61%RsaI sites of the cDNAs synthesized by the protocol provided by clontech PCR-select cDNA subtraction kit (Clontech, USA). So, AluI was selected to digest cDNA. The analysis of subtracted cDNA library showed that good results were obtained using AluI digestion instead of RsaI. 155 independent clones were found among 250 recombinant cDNA clones and 71 gene transcripts were found to be unique to appressoria among the 105 genes examined with RT-PCR.

View this table:
Table 5

Numbers of AluI/RsaI cutting sites in 8821 unique EST sequences of M. grisea obtained from COGEME 1.5 (http://cogeme.ex.ac.uk/)

Restriction enzymesTotal number of sites in EST sequencesNumber of sites added by cDNA synthesis primeraNumber of sites added by CDS III/3′ PCR primerbTotal number of sites in cDNAs synthesized by cDNA synthesis primercTotal number of sites in cDNAs synthesized by CDS III/3′ PCR primerd
Rsa 113,7458821022,56613,745
  • aNumber of sites in cDNAs added by cDNA synthesis primer with clontech PCR-select cDNAsubtraction kit.

  • bNumber of sites in cDNAs added by CDS III/3′ PCR primer with SMART cDNA library construction kit.

  • cTotal number of sites in cDNAs synthesized by cDNA synthesis primer with clontech PCR-select cDNA subtraction kit.

  • dTotal number of sites in cDNAs synthesized by CDS III/3′ PCR primer with SMART cDNA library construction kit.

In the last decade, many scientific efforts have been focused on the molecular genetic basis of M. grisea pathogenesis. Analyses of subtracted cDNA libraries, SAGE and microarray have revealed a number of genes specially expressed during the early phase of appressorium formation in M. grisea or during the plant–fungus interaction stage [810]. However, the gene expression profile in the later stages of appressorium development has received very little attention until now. The fact that 72 ESTs (51%) of total 142 non-redundant ESTs identified from 250 recombinant cDNA clones in this study have not previously been isolated, suggests that many genes specific to the maturing appressorium remained to be discovered in M. grisea. The identification and characterization of these genes might provide important clues as to the probable metabolic pathways and/or structural features unique to the maturation of the appressorium.

In this study, we found ESTs matching many genes whose product is probably involved in lipid metabolism within peroxisomes. The predicted products of these genes include an acyl-CoA dehydrogenase (clone s104: GenBank Accession No. CK828189) , a probable keto acyl reductase (clone s155), an alcohol oxidase (clone s48), a glyoxalase I in pyruvate metabolism (clone s126), and two different kinds of acetyl-CoA synthetase involving in pyruvate dehydrogenase bypass (clones s283, s177: GenBank Accession Nos. CN121362, CK828241), four distinct P450 monooxygenases (clones s103, s116, s261, s9: GenBank Accession Nos. CK828188, CK828198, CN193452, CK828294), a D-2-hydroxy-acid dehydrogenase (clone s130: GenBank Accession No. CK828211), a further putative monooxygenase (clone s67: GenBank Accession No. CK828282) and a cyclohexanone monooxygenase (clone s159: GenBank Accession No. CK828229). The purified cyclohexanone monooxygenase is a remarkably versatile oxygenation catalyst that uses the bound flavin adenine dinucleotide (FAD)-4a-OOH oxygenating intermediate to initiate oxygen transfer to both electrophilic substrate sites, such as the carbonyl of ketones and aldehydes, and nucleophilic substrate sites [24]. It is likely that the expression of these genes plays a role in the function of the appressorium. At 24-h post germination the appressorium generates enormous turgor pressure (up to 8 MPa), which is used to rupture the plant cuticle [6]. The turgor inside the appressorium is generated by a rapid increase in intracellular glycerol levels, which is maintained by a specialized cell-wall layer containing melanin [25,26]. Peroxisomes are single-membrane-bound organelles possessing multiple metabolic functions, including β-oxidation of fatty acids, glyoxylate metabolism, and metabolism of reactive-oxygen species. Clapex6-deleted mutants of Colletotrichum lagenarium have a defect in fatty acid β-oxidation in peroxisomes and form small appressoria with severely reduced melanization that failed to produce infectious hyphae [27]. The gene ICL1 encoding isocitrate lyase, involved in the glyoxylate cycle, is required for full virulence by M. grisea and shows elevated expression during development of infection structures and cuticle penetration [28]. It has also been demonstrated that the mobilization and dissolution of glycogen and lipid is under the control of the PMK1 MAPK pathway and the CPKA/SUM1-encoded PKA [29]. Both of these signaling pathways are required for appressorium differentiation (give further references to the relevant papers here). An EST (clone s197: GenBank Accession No. CK828251) whose predicted product shows homology to a serine–threonine kinase Pdd7p, required for pexophagy in Pichia angusta and Hansenula polymorpha[30,31], was also identified in the current work. Taken together these results support the view that lipid metabolism plays an important role in infection biology, and suggest that a major metabolic activity during appressorium maturation is the conversion of fat reserves into glycerol. P450 monooxygenase genes, glyoxalase I and other genes could also be involved in secondary metabolism, especially in detoxification of antagonistic substances secreted by plants. Among these, the transcripts of a predicted glyoxalase I encoding gene (represented by EST clone s126), a probable keto acyl reductase (clone s155), three distinct P450 monooxygenase encoding genes (clones, s103, s116, S261), a D-2-hydroxy-acid dehydrogenase encoding gene homolog (clone s130) and a putative monooxygenase encoding gene (clone s67) were confirmed to be abundant within appressoria but were not detected within conidia/mycelia using RT-PCR.

Among the genes expressed in appressorium identified by SSH, some genes, including GAS1 (MAS3) (clone s56) and GAS2 (MAS1) (clone s266: GenBank Accession No. CN121353), THNR (tetrahydroxynaphthalene reductase, clone s142: GenBank Accession No. CK828220), and PTH11 (clone s247: GenBank Accession No. CN121346), have previously been reported to be involved in appressorium formation and pathogenicity [3234]. GAS1 and GAS2 are preferentially expressed within appressoria in M. grisea compared with mycelial growth stage and although GAS1 and GAS2 deleted mutants have no defect in vegetative growth, conidiation, or appressoria formation, they are reduced in appressorial penetration and lesion development [34]. During the biosynthesis of fungal melanin, tetrahydroxynaphthalene reductase catalyzes the NADPH-dependent reduction of 1,3,6,8- tetrahydroxynaphthalene (T4HN) into (+)-scytalone and 1,3,8- trihydroxynaphthalene into (−)-vermelone [33]. Another pathogenicity gene, PTH11, encodes a transmembrane protein that is an upstream effector of appressorium differentiation in response to surface cues; strains lacking this gene are very reduced in pathogenicity due to a defect in appressorium differentiation [32]. Our RT-PCR analysis showed that transcripts matching GAS1 (clone s56), GAS2 (clone s266) and PTH11 (clone s247) were abundant with appressoria but scare within conidia and mycelia.

The predicted product of clones s248, s80 and s243 (GenBank Accession Nos. CN121347, CK828291, CN121345), have similarity to human oxysterol-binding protein, C14 sterol reductase and 6-phosphogluconate reductase, respectively. In C. gloeosporioides, a sterol glycosyl transferase gene (chip6) is induced by hard-surface treatment at an early stage of appressorium formation, and is required for fungal pathogenicity [35]. In M. grisea, clones AP2E02 and AP1I02, similar to phosphoglucose isomerase and delta-(24)-sterol c-methyl transferase gene involved in sterol biosynthesis, were found constitutively up-regulated during appressorium formation compared to vegetative mycelia [11]. These data are consistent with a role for sterol biosynthesis or sterol modification within pathogenesis by fungal phytopathogens [11,35].

Comparison of our data with results from previous analyses of subtracted cDNA libraries, SAGE and microarray in M. grisea[7,8,10] is of interest. Although some of the genes identified by previous analyses, such as GAS1, GAS2 and THNR, were also found in our study, most of the genes revealed by our experiments matched to predicted open/reading frames within the M. grisea genome sequence database for which no corresponding EST has been sequenced to date. The recovery of so many ESTs which have not been previously isolated in this species is likely due to the time point chosen. The gene expression pattern within mature appressoria after a 24 h incubation, is different from the pattern during the early phase of appressorium formation (after a 3–6 h incubation), a phase which has been much more extensively exploited in EST generation in M. grisea in previous studies (give some more references here). It is not surprising that we found many peroxisome-related genes, as that is the time when turgor pressure is generated from glycerol. Functional analyses of the novel genes identified during this work are underway in our laboratory.


The project was sponsored by the Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry, partly supported by National Natural Science Foundation of China (Grant No. 30270049) and by Ministry of Science and Technology of China (863 Program No. 2002AA245041). We are indebted to the following for help and advice generously given to us during the planning and writing of this article: Dr. Muriel Viaud, Dr. Wei-Liang Chen and Professor De-Bao Li.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
  26. [26].
  27. [27].
  28. [28].
  29. [29].
  30. [30].
  31. [31].
  32. [32].
  33. [33].
  34. [34].
  35. [35].
View Abstract