OUP user menu

Powdery mildew susceptibility and biotrophic infection strategies

Ralph Hückelhoven
DOI: http://dx.doi.org/10.1016/j.femsle.2005.03.001 9-17 First published online: 1 April 2005

Abstract

Plants are resistant to most potentially pathogenic microbes. This forces plant pathogens to develop sophisticated strategies to overcome basic plant resistance, either by masking intrusion or by suppression of host defences. This is particularly true for fungal pathogens, which establish long lasting interactions with living host tissue, without causing visible damage to invaded cells. The interactions of cereal crops and Arabidopsis with powdery mildew fungi are model systems for understanding host resistance. Currently, these systems are also promoting the understanding of fungal infection by identifying fungal pathogenicity and virulence factors and host target sites. This minireview focuses on recent findings about host susceptibility and the way powdery mildew fungi might induce it.

Keywords
  • Cell death regulation
  • Compatibility
  • Cytoskeleton
  • MLO
  • Pathogenesis associated molecular patterns

1 Introduction

Plants are immobile and face a regularly changing environment. This forces plants to cope with many kinds of stresses, including pathogenic microbes. Nevertheless, plants are resistant to the majority of potential pathogens that they are in constant contact with. This kind of plant disease resistance is called basic resistance, basic incompatibility or nonhost resistance. The latter term reflects the fact that incompatibility is most often provoked by specialization of a pathogen species to a narrow host range [1]. It is possible that specialization is a consequence of co-evolution with plants that were constantly forced to improve their resistance qualities. To date, it is not fully understood how nonhost resistance is constituted [2]. It works via several mechanisms, and the degree to which they contribute to resistance may depend on the individual plant–microbe interaction. Major mechanisms of basic incompatibility are non-recognition of a plant by a pathogen, defects in the pathogen that limit its ability to overcome preformed penetration barriers, and recognition of an inappropriate pathogen by the plant followed by effective defence reactions. Alternatively, the latter can be interpreted as an inability of the pathogen to avoid recognition or to suppress plant defence. This interpretation gained acceptance once it was understood that apparently all pathogens invading a plant are accompanied by pathogenesis-associated molecular patterns (PAMPs). Such PAMPs are generally recognized by the plant as non-self molecules, which is comparable to innate immunity in mammals. PAMPs can be conserved vital pathogen structures such as bacterial flagella peptides or fungal cell-wall polysaccharides [35]. Since PAMP recognition seems to be common in plants, a pathogenic microbe either has to (i) mask its PAMPs, (ii) bypass recognition, or (iii) suppress defence, in order to be pathogenic on its host plant.

Little is known about the way pathogens overcome basic resistance of their host plants. It is thought that pathogens have evolved various strategies. Pathogens might avoid PAMP recognition by release of so-called suppressors that can derive from the plant or the pathogen. Such suppressors might directly impair PAMP recognition by competing with a PAMP for binding by a PAMP receptor. Alternatively, suppressors might bind an independent target, which subsequently interferes with PAMP signal transduction [6,7]. A second strategy of pathogenic microbes to achieve and maintain compatibility appears to be targeted suppression of active host defence [810].

Given that microbes need to overwhelm basic resistance mechanisms to be infectious, understanding both sites of compatibility, the pathogenic strategy and the host target site, ought to be useful in re-establishing resistance by pharmacological or biotechnological means.

2 Powdery mildew infection

Powdery mildew (PM) fungi are pathogens infecting aerial parts of higher plants. They cause the PM disease of wild and crop plants that may depress yield by up to 30%. PM fungi are true ascomycete fungi, forming the order of Erysiphales with only one family, the Erysiphaceae. They are subdivided into five tribes (Erysipheae Golovinomycetinae Cystotheceae Phyllactinieae Blumerieae) and further subtribes, and more than 10 genera [11].

The PM fungi are biotrophic parasites invading only epidermal cells. After a conidial spore lands on the host surface, the PM fungus needs to attach to the surface and to penetrate the host cuticle and cell wall. The fungi build one or two germ tubes, depending on the genera [12]. The so-called appressorial germ tube swells at its tip to build an appressorium that is the essential penetration organ. The fungus directly penetrates the host cell wall apparently by means of both enzymatic and mechanical power [12]. After breakdown of the cell wall barrier, PM fungi develop a haustorium. This organ invaginates the host plasma membrane, which serves to supply the invader with nutrients, whilst the invaded cell remains intact. During penetration, the fungus has to cope with cell wall-associated defence of the host. Hence, even on a susceptible host, only a certain portion of germinated spores succeed in penetration. This also suggests that speed of penetration, or individual spore ability to locally suppress host defence, or both, are crucial for access to host cells. There is accumulating evidence for the ability of cereal PM fungi to suppress host defence gene expression [1315]. Importantly, living haustorium-containing cells show enhanced accessibility to subsequent penetration by normally incompatible PM fungi [16]. Additionally, different host cell types differ remarkably in their ability to prevent fungal penetration [17]. Together, the outcome of a fungal penetration attempt on a susceptible host depends on fungal aggressiveness, its ability to suppress host defence and the physiological status of the attacked cell.

The fact that leaf cell-death regulation and powdery mildew development are linked is evident from the observation that PM fungi induce a ‘green island effect’ on infected leaves. The leaf tissue surrounding a fungal colony remains green whilst the rest of the leaf shows chlorosis. This indicates semi-systemic cell death suppression at infection sites and redefinition of invaded tissues as a nutrient sink. Accordingly, an Arabidopsis invertase and a monosaccharide sugar transporter normally expressed in sink tissues are activated in source leaves by PM infection [18]. Monosaccharides are most likely to be taken up by PM fungi [19].

Besides these principal and structural prerequisites for plant infection, the development of highly specialized infection structures is regulated by sophisticated signalling cascades. Since PM fungi do not grow in axenic culture, they are difficult to study on the molecular level. Transformation of PM fungi is possible [20], but still challenging. Hopefully, future gene disruption strategies will help to identify pathogenicity and virulence factors of PM fungi. Random EST sequencing and differential gene expression studies have provided sequence information for Blumeria graminis f.sp. hordei (Bgh), the barley PM fungus, and candidate genes for functional analyses [2127]. Cyclic AMP-dependent protein kinase A signalling appears to be involved in early appressorial germ tube differentiation [22]. Best evidence for involvement of protein kinase A in appressoria formation stems from heterologous complementation experiments performed in the non-obligate plant pathogenic fungus Magnaporthe grisea. A M. grisea disruption mutant lacking cAMP-dependent protein kinase A subunit c shows delayed and ineffective appressorium development, and could be complemented to full pathogenicity by transformation with the homologous Bka1 gene from Bgh[28]. Furthermore, gene expression and pharmacological evidence suggest that a mitogen-activated protein kinase (MAPK) pathway is also crucial for appressoria formation. Pharmacological analyses suggest a signalling branch downstream of a heterotrimeric G-protein, and parallel but cooperative action of MAP kinase and cAMP in appressoria formation [29].

Bgh accumulates the redox-active substance 3-hydroxykynurenine during interaction with barley. The role of this chemical is not clear, but it could participate in either cross-linking the fungus with the host surface or it might have redox activity [30]. Additionally, Bgh expresses an apparently secreted catalase during host infection, which potentially has a role in removing H2O2 produced by the host, probably to cross-link its cell wall for penetration resistance [27,31].

Fungal avirulence factors (AVR) determine the ability of certain host cultivars to recognize the attacking fungus in a race-specific manner by corresponding resistance proteins. AVRs are also believed to be involved in pathogenicity on susceptible hosts. This dual role makes cloning of AVR-genes from biotrophs one of the greatest challenges in plant pathology [32]. Recently, Christopher Ridout from James Brown's group [33] has cloned candidate AVRK1 and AVRA10 from Bgh, which trigger resistance gene dependent host defences. Since Bgh AVRs lack typical signal peptides and the corresponding host resistance proteins seem to be expressed in the cytoplasm, it will be interesting to learn whether and how AVRs cross the plasma membrane of the parasite and the host.

3 Genetics of plant resistance to powdery mildew fungi

Once a microbe has overcome basic resistance, basic compatibility is achieved and the virulent pathogen can amplify on the susceptible host. However, normally not every host genotype is equally susceptible to a certain race of a pathogen. The aggressiveness of a pathogen genotype, together with the background resistance of a susceptible host plant, determine the severity of disease. Little is known about the molecular basis of these quantitative traits [34,35]. Quantitative (synonym: partial, horizontal) resistance is little understood and is usually expressed as a polygenic trait. Single genes contributing to quantitative resistance usually act additively [36].

Complete host powdery mildew resistance is monogenic and can be dominantly or recessively inherited. Major race-specific barley MLA and non-specific Arabidopsis RPW8 powdery mildew resistance genes have been isolated, and currently resistance proteins and downstream signalling are under intensive investigation (for a comprehensive review see [5]). Barley Mlo-mutant alleles (mlo) mediate broad spectrum-resistance to PM, which has been durable in spring-barley breeding for more than 20 years [36]. Recessive mlo-mediated resistance is extremely efficient. However, it is accompanied by spontaneous leaf cell death in older leaves, and is sometimes limited in extreme environments.

4 Plant defence mechanisms

Effective plant defence against PM fungi is usually organized in different sequential steps. Early defence prevents penetration, and a second line of defence inhibits nutrient uptake after haustoria formation. A few model systems for the interaction of plants and PM fungi are established. Among them, the interaction of cereals and Arabidopsis thaliana with PM fungi are the best studied in several aspects. Penetration resistance to the PM fungi is normally characterized by formation of cell wall appositions (CWAs) that are believed to build mechanical and chemical barriers against hydrolytic and osmotic pressure from fungal appressoria. CWAs are composed of altered cell wall material, which comprises inter alia 1,3-glucans (callose), silicon, phenolics, and diverse cell wall proteins [37]. The particular role of these constituents in penetration resistance is not fully understood. Likewise, the role of low molecular weight antibiotics in penetration resistance is not clear. Barley mlo-genotypes accumulate comparatively high contents of the phytoalexin p-coumaroyl-hydroxyagmatine in response to PM inoculation. Since p-coumaroyl-hydroxyagmatine has antifungal activity on Bgh, it could contribute to stopping early fungal development [38]. It is currently believed that PM fungi cannot dissolve lignin-like material. Blue and yellow autofluorescent lignin-like material is generally integrated in grass CWAs, and it has been observed that such material is insensitive to saponification earlier in resistant barley mlo -mutant genotypes than in susceptible Mlo-genotypes [38]. Additionally, H2O2 accumulation, protein cross-linking and immobilization was observed in CWAs induced by Bgh[31]. Both lignification and protein cross-linking depend on H2O2 as an oxidant [39,40]. H2O2 accumulation in CWAs is observed extremely frequently in resistant mlo-barley, whereas susceptible Mlo-genotypes accumulate little H2O2[4143]. Intriguingly, in contrast to H2O2 seems to play a much more complex role and might even be involved in cellular accessibility [4446]. This is supported by the observation that single cell dsRNA interference to knock down a potentially generating barley NADPH oxidase homologue led to enhanced background resistance in a susceptible host [47]. This might be explained by the contribution of to cell wall polymer loosening [45,46,48]. Alternatively, NADPH oxidase might have a role in negative control of independent defence pathways. In Arabidopsis, is monitored via LSD1 for negative control of cell death and defence gene expression [49].

Local formation of CWAs and secretion of toxic compounds requires polarization of the cellular defence machinery, which involves cytoskeleton reorganization. It has been shown that the actin cytoskeleton massively rearranges in barley cells under attack from PM fungi to form a polarized pattern focused towards the site of attempted penetration [5052]. This appears crucial in penetration defence, because plants show less actin polarization during compatible interactions [51]. When cowpea was treated with the actin polymerization inhibitor cytochalasin E after inoculation with the inappropriate plantain PM fungus, local H2O2 accumulation was reduced and, simultaneously, fungal penetration efficiency was greatly enhanced [53]. Similarly, nonhost resistance of barley coleoptiles to Erysiphe pisi and of Arabidopsis enhanced disease susceptibility 1 mutants to B. graminis f.sp. tritici could be partially breached by cytochalasins [50,54].

ROR1 and ROR2 are two barley host resistance factors that are required for full mlo-mediated resistance (identified as independent loci in a mlo-re-mutagenesis screen) and additionally are involved in nonhost resistance and background resistance to B. graminis but not in race-specific resistance [5558]. While the ROR1-gene is not yet characterized, ROR2 and its Arabidopsis ortholog PEN1, which was identified in a screen for nonhost resistance to Bgh, were recently isolated [55]. ROR2 and PEN1 proteins are SNARE-family plasma membrane syntaxins, which seem to be involved in local vesicle dynamics. Barley ROR2 interacts with barley SNAP34 (a SNAP25 homologue), possibly building a binary SNARE complex for vesicle-mediated exocytosis [55]. Green fluorescent protein (GFP)-PEN1 and yellow fluorescent protein (YFP)-ROR2 fusions focally accumulate at CWAs induced by Bgh[59,60]. Hence, both actin and membrane polarization appear to build a subcellular domain crucial for pre-haustorial defence against PM fungi.

As an additional or second line of defence, the host can prevent nutrient uptake by disturbing haustorial function. The most prominent way to achieve this is by the hypersensitive reaction (HR), including programmed cell death (PCD) of the attacked cell and/or a few surrounding cells. As in penetration resistance, H2O2 accumulates during execution of HR [31,41,46]. In fact, H2O2 might contribute in two ways to HR. First, H2O2 acts as a signal for PCD. Second, H2O2 might be fungitoxic and hence prevents the fungus from suppressing cell death.

Post-penetration defence appears not to be restricted to HR. Quantitative resistance of mildew host plant genotypes is often observed as a slow-disease phenotype or mild disease severity without leaf necrosis e.g. [61]. The factors restricting fungal development in these cases are largely unknown, and corresponding quantitative trait loci have not been isolated yet.

5 Host susceptibility factors – forward and reverse genetics

Host proteins apparently contribute to PM susceptibility (Table 1). This has been shown by barley and Arabidopsis mutagenesis leading to isolation of mutants, which have broad spectrum resistance to PM fungi, but not to other pathogens. Some of these mutants do not show constitutive expression of defence but instead show post-inoculation defence or fail to support fungal development [5,6265]. This indicates that target gene products function as negative controls of PM defence or are required to provide substrates for PM fungi. For instance, the Arabidopsis powdery mildew resistant 6 (pmr6) mutant shows a defence phenotype against PM fungi that is independent of HR, and common defence pathways which involve salicylic acid (SA), jasmonate or ethylene signalling. The PMR6 protein is similar to pectate lyases, and is likely to be associated with the extracellular site of the plasma membrane via a glycosylphosphatidylinositol anchor. PMR6 seems to function in cell-wall modelling since pmr6 has altered cell wall components. PMR6 is considered to be a PM specific host susceptibility factor, although its biochemical function has still to be determined [64]. The phenotype of Arabidopsis pmr5 mutants, affected in a gene encoding a plant specific predicted ER- and/or plasma membrane associated protein of unknown function, is very similar to that of pmr6, highlighting the role of cell wall composition in susceptibility to PM fungi [65]. This is also supported by the finding that the broad spectrum and PM resistant Arabidopsis mutant constitutive expressor of vegetative storage protein 1 cev1, is affected in the cellulose synthase gene A3[66]. Arabidopsis pmr4 mutants differ substantially from pmr5 and pmr6 mutants in that resistance includes late cell death reactions, defence gene expression and SA signalling [62,67]. PMR4 is a glucan synthase, also independently identified as GLS5. PMR4 is responsible for callose deposition at wound sites and in CWAs. Thus, PMR4 itself seems to down-regulate SA-dependent defence. It might be that appropriate PM fungi have evolved a strategy to utilize PMR4 to suppress intracellular defence signalling, which is derepressed in the absence of PMR4 [67,68].

View this table:
Table 1

Host proteins allowing or enhancing PM susceptibility identified by either forward or reverse genetics

GeneProven or potential protein functionForward geneticsReverse geneticsOver-expression induces enhanced susceptibilityMutation or knock-down induces resistanceRef.
AtPMR1Unknown+n.d.a.+[63]
AtPMR2; AtMLO2MLO-family member++n.d.a.+[63,78]
AtPMR3Unknown+n.d.a.+[63]
AtPMR4; AtGLS5Glucan synthase 5++n.d.a.+[63,67,68]
AtPMR5Cell expansion+n.d.a.+[63,65]
AtPMR6Pectate-lyase like protein+n.d.a.+[63,64]
AtEDR1MAP kinase kinase kinase++n.d.a.+[62,69,70]
AtCEV1Cellulose synthase A3+n.d.a.+[66]
HvMLOReceptor-like membrane protein++++[71,73,75,87]
HvBI-1Potential cell death inhibitor++aa[77,83]
HvRACBsmall RAC/ROP GTPase++b+[85,86]
HvRBOHaNADPH oxidase+n.d.a.+[47]
HvWRKY (WRKY9-10)Transcription factor++c+[80]
HvRLK (Hv8a14)Receptor-like kinase (LRK1-like)+n.d.a.+[80]
HvWIR1Unknown++n.d.a.[87]
HvPRX7Vacuolar peroxidase++n.d.a.[81]
  • n.d.a., presently no data available.

  • a HvBI-1 over-expression also breaches mlo-mediated resistance and nonhost penetration resistance to wheat PM fungus [77,83]. Standard transient BI-1 RNAi has low resistance inducing capability [83].

  • b Over-expression of wild type RACB has no effect whereas the constitutively activated mutant RACB-G15V induces super-susceptibility.

  • c Christina Eckey, University Giessen, Germany, personal communication.

Figure 1

Host and PM fungus proteins potentially involved in compatibility. Fungal proteins are likely to be required for appressorium development whereas host proteins were shown to be either required for full susceptibility or to promote it when over-expressed or both (see Table 1). Host proteins are depicted at their subcellular location that was either predicted or experimentally demonstrated: BI, BAX inhibitor-1; CAM, calmodulin; ER, endoplasmatic reticulum; Gαβσ, heterotrimeric G-protein; for further abbreviations see Table 1 and text.

The Arabidopsis enhanced disease resistant 1 (edr1) mutant also shows SA-dependent late acting mildew resistance accompanied by cell death. The EDR1 protein is a putative mitogen activated kinase kinase kinase that possibly acts early in negative regulation of SA-dependent defence [62,69,70]. However, edr1, like cev1, is not PM-specific but a more generally resistant mutant, which distinguishes edr1 from most pmr- and barley mlo-mutants [5,62]. The specificity of some mutation induced resistance against PM fungi supports the view that the corresponding genes encode host susceptibility factors required for basic compatibility [5]. Further evidence for this assumption comes from experiments with barley mlo. The dominant Mlo gene was isolated by Büschges et al. [71]. The gene encodes a deduced 60 kDa protein with seven transmembrane domains reminiscent of a G-protein coupled receptor [71,72]. Despite this topology, MLO function in susceptibility to Bgh appears to be independent from heterotrimeric G-proteins. Instead, MLO interacts with calmodulin to fulfill its function in susceptibility [60,73,74]. Transient single cell over-expression of the dominant Mlo cDNA provokes breakdown of mlo-resistance and of nonhost penetration resistance of both barley and wheat to the respective inappropriate forma specialis of B. graminis[7577]. Astonishingly, the barley mlo-phenotype, upon PM challenge, is mimicked in the Arabidopsis Atmlo2-mutant pmr2 and independently identified Atmlo2 insertion mutants, demonstrating that MLO is a PM susceptibility factor both in monocots and dicots. In barley and Arabidopsis mlo-mutants, fungal penetration is totally restricted [5,63,78, and Chiara Consonni, Matt Humphrey, Andreas Hartmann, Paul Schulze-Lefert, Ralph Panstruga, and Shauna Somerville, personal communication of unpublished results]. Interestingly, barley mlo-mutants show some pleiotropic effects in older plants under sterile conditions. This includes spontaneous formation of CWAs and an early senescence-like phenotype finally leading to leaf cell death [43,57,79]. Thus, MLO is both a host susceptibility factor and a cell death control element underscoring a link between host cell survival and susceptibility to the biotrophic fungus Bgh. However, the molecular basis of how the barley PM fungus takes advantage of MLO is not understood [5]. Together with other findings, the fact that Bgh grows on null-mlo ror genotypes indicates that MLO is not simply a factor needed by Bgh to recognize its host, but rather a negative regulator of local defence responses triggered by Bgh[56]. Recently, barley YFP-labelled MLO was shown to accumulate focally beneath fungal appressoria, where it interacts in planta with calmodulin. MLO accumulation appeared independently of actin, supporting the view that it accumulates early in plasma membrane microdomains, thus defining an entry gate for Bgh[60].

Additionally, some reverse genetics strategies have been successful in identifying potential host susceptibility factors. Candidate genes (Table 1Fig. 1) were either identified by differential expression analyses in plants, or intuitively. Candidates were either transiently knocked down or over-expressed in susceptible barley or wheat backgrounds. This led to modulation of accessibility to B. graminis. Some genes, including a vacuolar peroxidase, a WRKY transcription factor, a potential membrane protein (WIR1), and a receptor-like kinase, were found to enhance accessibility when over-expressed or to reduce accessibility when knocked down, or both [8083,8587]. Since the majority of genes tested have no effect in this complex system, these genes can be considered as potential host susceptibility factors (Table 1). Out of the genes tested, barley RACB and BAX Inhibitor-1 (BI-1) are the ones studied most intensively with regard to the mechanism of induced susceptibility. Barley BI-1 cDNA was isolated in an approach seeking cell-death suppressor proteins, which might regulate PM resistance similar to MLO. Since post-penetration cell survival is linked with induced accessibility even in mlo-genotypes [16], cell survival factors such as BI-1 might be involved in cellular accessibility. Leaf BI-1 expression was found to be up-regulated by Bgh and systemically down-regulated after root treatment with the chemical resistance inducer 2,6-dichloroisonicotinic acid [83]. Single cell transient over-expression of BI-1 rendered cells more accessible to penetration by Bgh and restored accessibility in resistant mlo-lines [83]. Until now, direct evidence of a role for endogenously expressed BI-1 in susceptibility is lacking, since transient BI-1 knock down is only slightly effective [83]. However BI-1, like MLO over-expression, partially breached barley nonhost resistance to the wheat PM fungus B. graminis f.sp. tritici, and this could not be further enhanced by simultaneous over-expression of both proteins [77]. Together, this led to the assumption that BI-1 over-expression substitutes for MLO functions in barley B. graminis interactions, and might act independently or downstream of MLO as a suppressor of penetration resistance [77,83].

Small RAC/ROP GTPases, such as barley RACB, are involved in hormone signalling, regulation of generating NADPH oxidases, Ca2+ signalling and actin remodelling [84]. Transiently induced gene silencing of barley RACB enhanced penetration resistance to Bgh[85,86]. In contrast, single-cell over-expression of constitutively activated RACB mutant RACB-G15V enhanced accessibility to Bgh[86]. Both knock down and RACB-G15V mutant effects on susceptibility were dependent on either ROR1 or MLO, which suggests that RACB modulates an intrinsic defence pathway [85,86]. Furthermore, activated RACB partly inhibited actin filament focusing towards sites of attempted penetration whereas RACB knock down and mlo-mutations promoted it [51]. Since actin remodelling takes part in CWA formation and haustoria establishment, Bgh might have corrupted RACB to antagonize local defence or to promote haustorium establishment in susceptible backgrounds, or both [51].

Studying the complex interaction of plants with PM fungi draws a fascinating picture of cell biology and molecular interplay of host and parasite. In particular, host genetics and in planta functional analyses have recently provided fascinating views of susceptibility factors. Future research will hopefully provide more insights into the mechanisms of susceptibility and the parasite's strategy to target host factors for establishment and maintenance of compatibility.

Acknowledgements

I am grateful to Holger Schultheiss for critical reading of the manuscript and to Ralph Panstruga (Max Planck Institute for Plant Breeding Research, Cologne, Germany) John Vogel (USDA Western Regional Research Center, Albany, USA), Shauna Somerville (Department of Plant Biology, Carnegie Institution, Stanford, USA,), Christopher Ridout (John Innes Centre, Norwich, UK) and Christina Eckey (University of Giessen, Germany) for communicating unpublished results. I apologize for the arbitrary selection of literature cited but limited space forced me to restrict references to some recent publications. I refer the willing reader to more general surveys as published, for instance, in the book “The Powdery Mildews” edited by R.R. Bélanger, W.R. Bushnell, A.J. Dik, and T.L.W. Carver, and published in St. Paul, Minnesota, USA.

References

  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].
  36. [36].
  37. [37].
  38. [38].
  39. [39].
  40. [40].
  41. [41].
  42. [42].
  43. [43].
  44. [44].
  45. [45].
  46. [46].
  47. [47].
  48. [48].
  49. [49].
  50. [50].
  51. [51].
  52. [52].
  53. [53].
  54. [54].
  55. [55].
  56. [56].
  57. [57].
  58. [58].
  59. [59].
  60. [60].
  61. [61].
  62. [62].
  63. [63].
  64. [64].
  65. [65].
  66. [66].
  67. [67].
  68. [68].
  69. [69].
  70. [70].
  71. [71].
  72. [72].
  73. [73].
  74. [74].
  75. [75].
  76. [76].
  77. [77].
  78. [78].
  79. [79].
  80. [80].
  81. [81].
  82. [82].
  83. [83].
  84. [84].
  85. [85].
  86. [86].
  87. [87].
View Abstract