OUP user menu

Role of the stress sigma factor RpoS in GacA/RsmA-controlled secondary metabolism and resistance to oxidative stress in Pseudomonas fluorescens CHA0

Stephan Heeb, Claudio Valverde, Cécile Gigot-Bonnefoy, Dieter Haas
DOI: http://dx.doi.org/10.1016/j.femsle.2004.12.008 251-258 First published online: 1 February 2005


In Pseudomonas fluorescens biocontrol strain CHA0, the two-component system GacS/GacA positively controls the synthesis of extracellular products such as hydrogen cyanide, protease, and 2,4-diacetylphloroglucinol, by upregulating the transcription of small regulatory RNAs which relieve RsmA-mediated translational repression of target genes. The expression of the stress sigma factor σS (RpoS) was controlled positively by GacA and negatively by RsmA. By comparison with the wild-type CHA0, both a gacS and an rpoS null mutant were more sensitive to H2O2 in stationary phase. Overexpression of rpoS or of rsmZ, encoding a small RNA antagonistic to RsmA, restored peroxide resistance to a gacS mutant. By contrast, the rpoS mutant showed a slight increase in the expression of the hcnA (HCN synthase subunit) gene and of the aprA (major exoprotease) gene, whereas overexpression of σS strongly reduced the expression of these genes. These results suggest that in strain CHA0, regulation of exoproduct synthesis does not involve σS as an intermediate in the Gac/Rsm signal transduction pathway whereas σS participates in Gac/Rsm-mediated resistance to oxidative stress.

  • Pseudomonas fluorescens
  • RpoS
  • GacA
  • RsmA
  • Secondary metabolism
  • Oxidative stress

1 Introduction

The regulation and functions of the stress and stationary phase sigma factor σS (also known as σ38) encoded by rpoS, have been studied in a variety of Gram-negative bacteria, especially in Escherichia coli[1,2] and Pseudomonas spp. [37]. In E. coli, σS induces more than 40 genes or operons during the transition from exponential to stationary phase, producing proteins usually associated with resistance to stress [1,2]. The regulation of rpoS is very complex and involves transcriptional, translational and post-translational control mechanisms. At the transcriptional level, rpoS expression appears to be modulated by the sensor kinase BarA (a GacS homologue) [8], intracellular levels of cAMP and its receptor protein CRP, guanosine-3′,5′-bispyrophosphate (ppGpp), inorganic polyphosphate, and UDP–glucose [911]. Translational rpoS control responds to changes in osmolarity, temperature, and nutrient deprivation, e.g. during transition to stationary phase, and is exerted on the 5′ untranslated leader of rpoS mRNA [2]. Translation of rpoS is modulated by the RNA-binding protein Hfq [12], the histone-like protein H-NS [13], and the small regulatory RNAs DsrA, OxyS, and RprA [1416]. Post-translational control involves σS turnover and is mediated essentially by the intracellular ClpXP protease [2], the heat shock chaperone DnaK [17], and the response regulator RssB [18]. Additionally, the activity of σS is subject to competition with other sigma factors for association with core RNA polymerase to form the RNA polymerase holoenzyme [19].

Certain functions of RpoS are similar in Pseudomonas spp. and in enteric bacteria. In particular, survival during osmotic, heat or oxidative stresses is reduced in rpoS mutants of P. aeruginosa, P. putida, and P. fluorescens[47,20]. However, in Pseudomonas spp., rpoS appears to be more extensively regulated at the transcriptional level than in E. coli. The TetR-like regulator PsrA, which strongly activates rpoS transcription by binding to a palindromic sequence in the rpoS promoter, has been found so far only in Pseudomonas spp. [2123]. In P. fluorescens Pf-5 and P. chlororaphis O6, but not in P. aeruginosa, rpoS expression is positively controlled by the GacA/GacS two-component system and an rpoS null mutation negatively affects the survival under conditions of oxidative stress [4,22,2426].

In the biocontrol strain P. fluorescens CHA0 [27], the rpoS gene is located immediately upstream of rsmZ, a GacA/GacS-controlled gene encoding a small regulatory RNA that antagonizes translational repression of target genes by the RNA-binding protein RsmA [28]. Typical Gac/Rsm-controlled target genes encode exoproducts involved in biocontrol, e.g. hcnA (for HCN synthesis), phlA (for 2,4-diacetylphloroglucinol [Phl] synthesis) and aprA (for the major exoprotease) [2931]. In Erwinia carotovora, RpoS negatively affects the production of extracellular enzymes, apparently by upregulating rsmA expression [32]. In Azotobacter vinelandii, GacA and RpoS are part of a cascade controlling alginate production [33]. These observations prompted us to investigate the question whether RpoS could be an intermediate in the Gac/Rsm signal transduction pathway of P. fluorescens CHA0, in the regulation of exoproduct formation and in resistance to oxidative stress.

2 Materials and methods

2.1 Bacterial strains and growth conditions

P. fluorescens CHA0 derivatives [27] and E. coli DH5α[34] were routinely grown in nutrient yeast broth (NYB; 2.5% [w/v] nutrient broth, 0.5% [w/v] yeast extract) with shaking, or on nutrient agar [28] amended with the following antibiotics, when required: ampicillin, 100 μg ml−1 (for E. coli); tetracycline, 25 μg ml−1 (125 μg ml−1 for P. fluorescens); and gentamicin, 10 μg ml−1. Routine incubation temperatures were 37 ° C for E. coli and 30 °C for P. fluorescens. When P. fluorescens was to receive heterologous DNA (e.g., in electroporation [28,35]), the incubation temperature was raised to 35 °C. Expression of a phlA′–′lacZ fusion in P. fluorescens was monitored in minimal medium OS amended with glucose and ammonium [35].

2.2 DNA manipulations

Small-scale plasmid extractions were performed by the cetyltrimethylammonium bromide method [36], and large-scale preparations were done by using the Qiagen Plasmid Midi kit (Qiagen Inc.). DNA fragments were purified from agarose gels with the Geneclean II Kit (Bio101, La Jolla, CA). DNA sequencing was performed with the Big Dye Terminator Cycle sequencing kit and an ABI-Prism 373 automatic sequencer (Applied Biosystems). PCRs were typically carried out with 2.5 U of thermostable DNA polymerase (Extra-Pol II; Eurobio) in a reaction mixture containing 100 ng of target DNA, 250 μM of each of the four dNTPs (Roche), 10 pmol of each primer, 5 mM MgCl2, and 1× Extra-Pol buffer in a final volume of 20 μl. For the amplification reaction, 25 cycles (1 min at 94 °C, 1 min at 50–55 °C (depending on the G + C content of the primers), and 1 min at 72 °C) were followed by a final elongation step of 5 min at 72 °C.

2.3 Overexpression of rpoS

The 2.6-kb HindIII-EcoRI rpoS rsmZ fragment from pME6087 [28] was inserted into pBluescript II KS+ [Stratagene] to produce pME6088 (Fig. 1), and a PCR with oligonucleotides RPOSCHA04 and RPOSCHA05 was done to amplify a 1.1-kb EcoRI–BglII rpoS fragment from this plasmid. RPOSCHA04 (5′-CCGGAATTCGAACTCACCAAAGGACTATAAC-3′) anneals upstream of the rpoS gene and creates an artificial EcoRI site (underlined) 28 nucleotides upstream of the start codon. RPOSCHA05 (5′-CGAGTAGATCTGGGCTCTTGTGAATCGATC-3′) anneals downstream of the coding region and creates an artificial BglII site (underlined) 26 nucleotides behind the stop codon. The EcoRI–BglII fragment was inserted into EcoRI–BamHI-digested pME6001 [37] to produce pME6354, placing rpoS under control of the lac promoter (Fig. 1).

Figure 1

Maps of constructs involving rpoS from P. fluorescens CHA0 used in this study. Vectors used for the construction of pME6088, pME6354, and pME6355 are shown in brackets. In strains CHA813 (chromosomal aprA′–′lacZ fusion), CHA814 (chromosomal hcnA′–′lacZ) and CHA815 (wild-type background) rpoS codons 8–322 (out of 336) were deleted in-frame [28]. In pME6354, rpoS was placed under the lac promoter (Plac) of vector pME6001. Plasmid pME6355 carries a translational rpoS′–′lacZ fusion at the 8th codon. PrpoS: rpoS promoter; T1, T2 and T3: rho-independent transcriptional terminators. Restriction sites that were used for the constructions are shown together with the positions and orientations (5′→ 3′) of oligonucleotides used to amplify fragments. Nucleotide sequence of the 2.6-kb HindIII–EcoRI fragment has been deposited in GenBank under Accession No. AF245440.

2.4 Construction of an rpoS′–′lacZ translational fusion

PCR with oligonucleotides RPOSCHA03 and T7 [28] was done to amplify a 0.96-kb EcoRI–XhoI rpoS′ fragment from pME6088. RPOSCHA03 anneals to the 5′ end of rpoS and creates an artificial XhoI site at the 8th codon [28]. The PCR product was digested with EcoRI and XhoI and inserted into pBluescript II KS+ to produce pME6350, where it was sequenced to confirm the absence of mutations. This insert was then subcloned into EcoRI–SalI-cut pNM482 [38], and the resulting 1.8-kb EcoRI–ClaI rpoS′–′lacZ′ fragment was finally inserted into EcoRI–ClaI-cut pME6014 [35] to produce pME6355 (Fig. 1).

2.5 β-Galactosidase assays

β-Galactosidase activities in P. fluorescens carrying lacZ fusion constructs were quantified by the Miller method [39], using cells permeabilised with 5% [v/v] toluene. All measurements were done after inoculation of 30 ml of NYB with 300 μl of overnight cultures and incubation at 30 °C with shaking.

2.6 Detection of RsmA by Western blotting

Strains were grown in 20 ml of NYB containing 0.05% [v/v] Triton X-100 in 125-ml Erlenmeyer flasks at 30 °C with agitation. An equivalent of 0.4 OD600 units from cultures in stationary phase cells was centrifuged. Cells were washed with 0.9% NaCl [w/v], resuspended in 20 μμl of loading buffer (50 mM Tris–HCl pH 6.8, 2% [w/v] SDS, 0.1% [w/v] bromophenol blue, 15% [w/v] glycerol, 5% [v/v] 2-mercaptoethanol) and immediately treated at 100 °C for 10 min. For each sample, 15 μl of cell lysate was loaded per well. Samples were fractionated by Tricine–SDS–PAGE in 16% [w/v] acrylamide:bisacrylamide 29:1 gels at 75 V during 4 h [40]. After electrophoresis, proteins were electroblotted onto polyvinylidene difluoride (PVDF) membranes (Immobilon P, Millipore) at 50 mA and 4 °C for 1 h. RsmA was detected using polyclonal antibodies raised against purified Yersinia enterocolitica RsmA as previously described [41]. Membranes were developed with the ECL Western Blotting analysis system (Amersham-Pharmacia) following the manufacturer's instructions.

2.7 Assay of oxidative stress survival

Stationary phase cells of P. fluorescens grown in NYB were washed and resuspended in 0.9% [w/v] NaCl to an OD600 of 0.8. This suspension (5 ml) was exposed to 40 mM H2O2 at 30 °C with agitation, and the number of culturable cells per ml was determined on nutrient agar before and after 1 h of exposure.

3 Results

3.1 Expression of rpoS is regulated by GacA and RsmA

In P. fluorescens CHA0, like in most pseudomonads, the rpoS gene is flanked by the nlpD (for a lipoprotein) and rsmZ genes [28] (Fig. 1). The position of the rpoS promoter, located in the nlpD coding region, was deduced by sequence comparison with the highly conserved rpoS promoters previously described in P. aeruginosa[42] and P. putida WCS358 [21]. A fragment containing the rpoS promoter and the first 8 codons of rpoS was joined to a ′lacZ fragment in pME6355, creating an rpoS′–′lacZ translational fusion (Fig. 1). In the wild-type CHA0, the expression of the rpoS′–′lacZ fusion was induced at high cell population densities and was about 4-fold higher than in the gacA mutant CHA89 [43] (Fig. 2(a)). A similar positive effect of the GacS/GacA system has been observed with a chromosomal rpoS lacZ transcriptional fusion in P. fluorescens strain Pf-5 [24]. In the rsmA mutant CHA809 [28], β-galactosidase activities were elevated about 2-fold and occurred at lower cell population densities compared with the wild-type CHA0 (Fig. 2(a)), suggesting that RsmA acts as a negative control element in the expression of rpoS.

Figure 2

Regulation of rpoS, hcnA, aprA, and phlA in the Gac/Rsm system. (a) β-Galactosidase activities specified by pME6355 (translational rpoS′–′lacZ fusion) were assayed in the P. fluorescens wild-type CHA0 (○), the gacA mutant CHA89 (△) and the rsmA mutant CHA809 (◻). (b) β-Galactosidase activities were determined in the RpoS-negative strains CHA814 (hcnA′–′lacZΔrpoS; ▲) and CHA813 (aprA′–′lacZΔrpoS; ●) as well as in the rpoS+ strains CHA207 (hcnA′–′lacZ; △) and CHA805 (aprA′–′lacZ; ○). (c) β-Galactosidase expression of chromosomal translational aprA′–′lacZ (●,○) or hcnA′–′lacZ (▲,△) fusions were tested in a wild-type background carrying the control vector pME6001 (○,△) or the rpoS-overexpressing plasmid pME6354 (●,▲). (d) β-Galactosidase activity produced by a translational phlA′–′lacZ fusion on pME6259 was tested in the P. fluorescens wild-type strain CHA0 (●) and the rsmA mutant CHA809 (○). Activities (Miller units) are mean values of triplicate experiments ± SD.

3.2 Effects of deletion and overexpression of rpoS on hcnA, aprA, and phlA expression

In strain CHA0, the expression of typical exoproduct genes such as hcnA, aprA, and phlA is 30- to 60-fold higher than in gacS or gacA mutants [28,31,35,37]. If these positive effects of the GacS/GacA system were mediated, at least in part, by RpoS, we would expect to see reduced expression of hcnA, aprA, and phlA in an rpoS-negative background. However, in the rpoS deletion strains CHA814 [28] (hcnA′–′lacZΔrpoS) and CHA813 [28] (aprA′–′lacZΔrpoS) a slight (≤30%) increase in β-galactosidase activities was found with both reporter fusions when compared with the rpoS+ strains CHA207 [37] (hcnA′–′lacZ) and CHA805 [37] (aprA′–′lacZ) (Fig. 2(b)). This small effect appeared late in growth (Fig. 2(b)). The growth rates were unaltered by the rpoS mutation under the conditions used. The result obtained with thehcnA′–′lacZ fusion is in harmony with the observation that an rpoS mutant of P. fluorescens Pf-5 overproduces the extracellular metabolite HCN [24]. To monitor phlA expression, the phlA′–′lacZ translational fusion carried by pME6259 [35] was introduced into the wild type and the rpoS mutant CHA815 [28]. No significant difference was found with this fusion between the two backgrounds (data not shown).

When plasmid pME6354, in which rpoS is under the control of the lac promoter (Fig. 1), was introduced into strains CHA207 and CHA805, β-galactosidase activities produced by the hcnA′–′lacZ and aprA′–′lacZ translational fusions were reduced ≥3-fold during growth, compared to the control strains containing the vector pME6001 (Fig. 2(c)), suggesting a negative control of HCN and protease production by RpoS.

3.3 Negative control of phlA expression by RsmA

In our previous studies we have shown that RsmA negatively controls the expression of the hcnA and aprA genes at a post-transcriptional level in strain CHA0 [28,37]. The apparent lack of regulation of the phlA′–′lacZ fusion by RpoS prompted us to verify that the expression of the phlA gene is indeed negatively controlled by RsmA. The expression of the phlA′–′lacZ fusion was 2- to 3-fold higher in the rsmA mutant CHA809 [28] than in strain CHA0 throughout growth (Fig. 2(d)). This result confirms that GacS/GacA-controlled phlA expression [35] involves RsmA as an intermediate in the signal transduction pathway.

3.4 RpoS does not control RsmA levels

In E. carotovora, RpoS positively controls the expression of RsmA [32]. We therefore monitored the amount of RsmA protein in the wild-type CHA0, in the rpoS mutant CHA815 [28], as well as in the gacS mutant CHA19 [31] and in the gacA mutant CHA89 [43]. However, no significant differences in RsmA levels could be detected (Fig. 3), suggesting that in strain CHA0, regulators of stationary phase-dependent phenotypes like GacS/GacA, or RpoS, do not influence RsmA concentration.

Figure 3

RsmA levels are not affected by GacS, GacA or RpoS. An equivalent amount of cells from strains CHA0 (wt), CHA19 (gacS), CHA89 (gacA), CHA815 (rpoS) and CHA809 (rsmA) was taken from stationary phase cultures, subjected to SDS–PAGE, and analysed by Western blotting with antibodies raised against Y. enterocolitica RsmA [41]. The optical density of the cultures at 600 nm was 4.6–5.4 units at the moment of sampling. The band migrating below RsmA represents the RsmA homologue RsmE [48].

3.5 Oxidative stress survival is under Gac/Rsm/RpoS control

The aim here was to determine whether resistance to oxidative stress is controlled by the postulated Gac/Rsm/RpoS cascade. Thus, we assayed the survival of stationary-phase cells after exposure to hydrogen peroxide in the wild-type strain CHA0, the gacS mutant CHA19, the rsmA mutant CHA809, and the rpoS mutant CHA815. Resistance to H2O2 is caused probably by catalase(s) and/or peroxidase(s) which are under GacS/GacA control [26]. Loss of gacS or rpoS function both caused a drastic reduction in survival, whereas an rsmA mutation had little effect (Table 1). The sensitivity of the gacS mutant to hydrogen peroxide could be suppressed to wild-type levels by overexpressing either rpoS with pME6354 or rsmZ with pME6359, when comparing to the wild-type strain carrying the empty expression vectors pME6001 [37] or pME6032 [28], respectively (Table 1). Overexpression of rsmA in strain CHA0 with plasmid pME6073 [28] mimicked a gacS defect (Table 1). Together, these results support the existence of a linear GacS/GacA → RsmZ → RsmA → RpoS → oxidative stress resistance pathway (Fig. 4).

View this table:
Table 1

Survival of P. fluorescens strains following exposure to H2O2

StrainGenotypeViable counts (CFU ml−1) atSurvival (%)
t= 0t= 1 h
CHA0Wild type6.2 ± 0.2 × 1085.0 ± 0.5 × 10880.6
CHA815ΔrpoS4.8 ± 0.6 × 1088.7 ± 1.6 × 1061.8
CHA19/pME6001ΔgacS3.0 ± 0.1 × 1081.8 ± 2.8 × 1050.1a
CHA19/pME6354ΔgacS rpoS++3.0 ± 0.6 × 1082.8 ± 0.3 × 10893.3
CHA19/pME6032ΔgacS1.8 ± 0.6 × 1082.3 ± 0.3 × 1061.3
CHA19/pME6359ΔgacS rsmZ++4.6 ± 1.7 × 1084.7 ± 2.6 × 108102.2
CHA809rsmA:: Ω KmR7.5 ± 1.7 × 1084.6 ± 1.5 × 10861.3
CHA0/pME6001Wild type3.0 ± 0.5 × 1081.1 ± 0.9 × 1073.8a
CHA0/pME6073rsmA ++3.5 ± 1.0 × 1084.7 ± 1.0 × 1050.1
  • Strains were grown to early stationary phase. At t= 0, when 40 mM H2O2 was added, and after 1 h of exposure (t= 1 h), viable counts were determined in triplicate (±SD).

  • aThe gentamicin resistance plasmid pME6001 was included as a control as it had a strong negative effect on viability, for unknown reasons. Plasmids pME6354 and pME6073 are pME6001 derivatives, whereas pME6359 is a derivative of the vector pME6032 [28].

Figure 4

Proposed pathway depicting the role of RpoS in Gac/Rsm-controlled secondary metabolism and resistance to oxidative stress in P. fluorescens CHA0. Signal transduction involving the GacS/GacA two-component system positively controls the transcription of small regulatory RNAs such as RsmZ and RsmY. These regulatory RNAs antagonise the RsmA-mediated post-transcriptional regulation of target genes such as the hcn, apr and phl operons. RpoS, being also controlled by this pathway, positively controls the resistance to oxidative stress, and negatively affects the expression of the hcnA and aprA genes, presumably indirectly by competition with RpoD (σ70) for RNA polymerase core enzyme.

4 Discussion

The observation that in some, but not all, fluorescent pseudomonads the expression of RpoS is positively controlled by the GacS/GacA two-component system [22,2426] incited us to investigate two aspects of the Gac regulon in the biocontrol organism P. fluorescens CHA0: first, the question whether the two-component system GacS/GacA and the RNA-binding protein RsmA, a key regulatory element in the Gac regulon [37], regulate RpoS expression and hence RpoS-controlled resistance to oxidative stress, and second, the question whether RpoS is an intermediate element in Gac/Rsm-controlled expression of secondary metabolism and exoenzymes.

The fact that an rpoS′–′lacZ fusion was poorly expressed in gacA background and overexpressed in an rsmA mutant, by comparison with the wild type (Fig. 2(a)), suggests that the Gac/Rsm cascade activates RpoS expression in P. fluorescens CHA0. This is in agreement with the data of Whistler et al. [24] who showed that in gacS or gacA mutants of P. fluorescens Pf-5 RpoS protein levels were 20–50% of those present in the wild type. Whereas the mechanism by which the Gac/Rsm cascade regulates rpoS expression is still unknown, the effects of this cascade could be clearly seen at the level of resistance to hydrogen peroxide (Table 1) in that both a gacS and an rpoS mutation resulted in a 50-fold decrease of viability within 1 h of incubation with hydrogen peroxide, and a similar effect occurred in an rsmA overexpressing background. An rsmA-negative mutant manifested a slight reduction (rather than an increase) in resistance to oxidative stress, suggesting that the rsmA mutation may have some negative side effects on the viability of the strain (Table 1). By contrast, the presence of either rsmZ or rpoS on a multi-copy plasmid entirely suppressed a gacS defect, confirming the model (Fig. 4) according to which rsmZ overexpression would relieve RsmA-mediated repression of rpoS translation and rpoS overexpression would override the gacS signalling defect. It will be interesting to find out whether the extensive rpoS 5′ leader transcript [21,23] is able to bind RsmA and other regulatory elements such as small RNAs [44].

RpoS is unlikely to be an intermediate in Gac/Rsm-dependent control of hcnA and aprA expression, as in both cases the effect of RpoS was clearly negative (Figs. 2(b) and (c)). We interpret these observations in terms of competition of different σ factors for RNA polymerase core enzyme: the house-keeping σ70 has the strongest affinity and σ38 (RpoS) has the weakest affinity for the core enzyme [45]. If, as it appears likely, the transcription from the hcnA, aprA and phlA promoters is essentially driven by σ70 RNA polymerase, an excess of σ38 would be expected to have a negative effect on the transcription of these genes, and this is compatible with our results as well as with observations made on HCN and Phl biosynthesis in P. fluorescens Pf-5 [4,24,46].

Although the Gac/Rsm cascade is conserved in many Gram-negative bacteria, the target genes regulated vary widely and so does the fine-tuning of the cascade [29,47]. For instance, RpoS positively controls RsmA synthesis in E. carotovora, whereas we see no measurable effect of RpoS on RsmA levels in P. fluorescens CHA0 (Fig. 3) but, on the contrary, we have obtained evidence for negative control of rpoS expression by RsmA (Fig. 2(a)). In strain CHA0, the regulatory effects of an rsmA mutation on the expression of target genes are less pronounced than are those of gacS and gacA mutations [37], essentially because this bacterium has a second RsmA-like regulatory protein, RsmE, whose expression is under GacA control [48]. The role of small regulatory RNAs, i.e. RsmZ [28] and RsmY [30], in the regulation of rpoS is currently under investigation.


We thank Paul Williams for the anti-RsmA antibodies, Karin Heurlier for her assistance in the detection of RsmA by Western blotting, and Kirsten Bang Lejbølle for the construction of pME6354. Support from the Swiss National Foundation for Scientific Research (projects 31-64048.00 and 3100A0-100180) is gratefully acknowledged.


  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].
View Abstract