OUP user menu

N-hexanoyl-l-homoserine lactone, a mediator of bacterial quorum-sensing regulation, exhibits plant-dependent stability and may be inactivated by germinating Lotus corniculatus seedlings

Laurie Delalande, Denis Faure, Aurélie Raffoux, Stéphane Uroz, Cathy D'Angelo-Picard, Miena Elasri, Aurélien Carlier, Romain Berruyer, Annik Petit, Paul Williams, Yves Dessaux
DOI: http://dx.doi.org/10.1016/j.femsec.2004.10.005 13-20 First published online: 1 March 2005


The half-life of N-hexanoyl-l-homoserine lactone (C6-HSL) was determined under various pH and temperature conditions, and in several plant environments. C6-HSL was sensitive to alkaline pH, a process that was also temperature-dependent. In addition, C6-HSL disappeared from plant environments, i.e. axenic monocot and dicot plants cultivated under gnotobiotic, hydroponic conditions, albeit with variable kinetics. The disappearance was rapid at the root system of legume plants such as clover or Lotus, and slow or non-existent at the root system of monocots such as wheat or corn. These variable kinetics were not dependent upon pH changes that may have affected the growth media of the plants. Furthermore, C6-HSL did not accumulate in the plant, and the plant did not produce inhibitors of the C6-HSL signal. HPLC analyses revealed that C6-HSL disappeared from the media, and hence, Lotus exhibited a natural C6-HSL inactivating ability. This ability was not specific for C6-HSL and allowed the degradation of other N-acyl-homoserine lactones such as 3-oxo-C6-HSL, 3-oxo-octanoyl-HSL and 3-oxo-decanoyl-HSL. Preliminary investigation revealed that the inactivating ability is temperature-dependant and possibly of enzymatic origin.

  • N-acyl homoserine lactone
  • Quorum sensing
  • Quenching
  • Legume
  • Rhizosphere

1 Introduction

Most microbes have evolved the ability to respond to environmental changes. Amongst the regulatory mechanisms involved in the microbial response, one, termed quorum sensing (QS), allows the transcription of the regulated genes in a population density-dependent fashion. This regulatory mechanism is characterized by the involvement of low molecular weight signal molecules that accumulate in the environment as a function of the cell density of the producer organism. Once a threshold concentration of the signal molecules – hence a threshold population density is reached, the signal is perceived allowing the microbes to respond in a concerted way [14].

Different families of QS molecules have been identified in Gram-positive and Gram-negative bacteria. Several of these molecules belong to a class termed N-acyl-homoserine lactones (or AHLs). A broad range of functions essential to the competitiveness of plant-associated bacteria are regulated by QS via the production of AHLs. Amongst these, biofilm maturation, pathogenicity, plasmid conjugal transfer, swarming motility and secondary metabolite production have been extensively documented [2,3,5].

As signaling molecules AHLs have to be produced at the appropriate time to ensure a desirable biological effect. This implies de facto that the AHL signal must have a limited lifetime in natural environments since a highly stable molecule would be present at inducing amounts at inappropriate times. Previous data indeed indicate that AHLs undergo degradation, for instance as a function of pH [6,7]. However, limited numerical data on the half-life of these molecules have been reported, especially in plant environment. To investigate this question, the stability of the AHL molecule N-hexanoyl-l-homoserine lactone (C6-HSL), was determined in vitro as a function of pH and temperature. C6-HSL was chosen for this study because it is produced – and used as a QS signal molecule – by many different Gram-negative bacteria including Pseudomonas, Aeromonas, Serratia, Yersinia, Chromobacterium, and Rhizobium (reviewed in [3]). This molecule is also commercially available and easily detectable in liquid culture media. In the second part of the work, the stability of C6-HSL at the root system of various monocot and dicot plants was investigated. This series of experiments was performed under axenic conditions to prevent C6-HSL from being degraded by some root-associated bacteria, a phenomenon that has already been reported for members of several genera such as Bacillaceae, Pseudomonas, Ralstonia, Variovorax, Comamonas, Arthrobacter and Rhodococcus (see [812]). Overall, the data demonstrate that the half-life of C6-HSL is strongly dependent upon pH and temperature and also plant species. Thus, while C6-HSL appears to be stable in the gnotobiotic root system of wheat and corn, it was rapidly eliminated from that of developing Lotus plantlets, suggesting that this legume species remarkably inactivates this QS signal molecule. Similar experiments performed with other AHLs molecules indicated that this ability is not restricted to C6-HSL as 3-oxo-C6-HSL, 3-oxo-octanoyl-homoserine lactone (3-oxo-C8-HSL) and 3-oxo-decanoyl-homoserine lactone (3-oxo-C10-HSL) were also degraded by germinating Lotus plants.

2 Materials and methods

2.1 Media and bacterial strains

Modified Luria–Bertani (i.e. LB with 5 g L−1 NaCl) was used as rich, non-selective media. The bacterial strains used in this study were the AHL biosensors Chromobacterium violaceum strain CV026[13] and Agrobacterium tumefaciens strain NTLR4[14], which were grown while shaking at 28 °C. Colonies of CV026 are white but turn purple upon exposure to AHLs such as C6-HSL, producing the pigment violacein. Agrobacterium colonies are white but produce a β-galactosidase reporter activity that can be detected using chromogenic substrates upon exposure to AHLs such as 3-oxo-C6-HSL, 3-oxo-C8-HSL and 3-oxo-C10-HSL.

2.2 Detecting C6-HSL and other AHLs

C6-HSL and other AHLs were detected using the above-mentioned AHL biosensors, and agar and silica plate assays (KC18 silica gel, 200 μm, Whatman International, Maidstone, England), as described by McClean et al.[13] and Shaw et al.[14]. Quantification was performed from the measurement of violacein production or β-galactosidase activity halos that were converted into residual concentrations of AHLs via a calibration curve obtained from the spotting of known amounts of authentic synthetic AHLs. Alternatively, C6-HSL was quantified using HPLC following elution with acetonitrile/water combinations of a Kromasil C8 5 μM column (2.1 × 250 mm, Jones Chromatography Ltd., Glamorgan, UK) fitted on a Waters 625 system coupled with a Waters 996 PDA photodiode array detector [7,15].

2.3 Analyzing C6-HSL degradation as a function of pH and temperature

Stability of C6-HSL towards pH was investigated by incubating 40 μg of C6-HSL at −20 °C, +4 °C, +20 °C, +28 °C and +37 °C in the dark in a final volume of 100 μl (final C6-HSL concentration: 2 μM) of 10 mM Tris–maleate or 10 mM Tris–HCl buffered at various pHs (Tris–maleate pH 5.5, 6.0, 6.5, 7.0; Tris–HCl pH 7.0, 7.5, 8.0, 8.5, 9.0). Samples, taken at times ranging from hours to days, were neutralized (final pH 6.8) and the residual concentration of C6-HSL in the incubation media was determined using the semi-quantitative assay presented above. Sextuplate independent evaluations were obtained for each condition at each sampling time. These values were averaged, converted into residual concentrations of C6-HSL via a calibration curve and plotted as a function of time. A logarithmic regression analysis was performed on these values using the ad hoc routine of the Excel software (Microsoft-France) from which the various C6-HSL half-life values were deduced. Half-life is the period of time in which the concentration of the assayed AHL decreases by a factor of two. Several conditions that yielded half-life values shorter than 1 day were re-analyzed in a second series of experiments. For these, 800 μg of C6-HSL were incubated at 20, 28 and 37 °C in the dark in a final volume of 200 μl (final C6-HSL concentration 20 μM) of 10 mM Tris–HCl buffered at various pH levels (pH 7.0, 7.5, 8.0, 8.5, 9.0). Samples, taken at times ranging from minutes to hours, were processed as indicated above.

2.4 Plants and plant-growth conditions

The plants investigated belonged to two monocot species, i.e. Zea mays L. cv. Banguy (Poaceae) and Triticum aestivum L. cv. Isengrain (Poaceae), and to two dicot legume species, i.e. Lotus corniculatus cv. Rodéo (Fabaceae) and Trifolium pratense L. (Fabaceae). Seeds from these plants were surface-sterilized by immersion first in 70% (v/v) ethanol and then in 90 g L−1 calcium hypochlorite (both for 15 min). They were rinsed extensively with ultrapure water. Seeds were germinated in 30 ml liquid, Murashige and Skoog (MS) medium with no sucrose added at 24 °C in a growth chamber under 12 h daylight conditions. Two non-buffered formulations of the MS medium were used: (i) the regular (termed MS) basal mixture of macro- and micro-nutrients, which contained 20.61 mM ammonium nitrate and 18.79 mM potassium nitrate; (ii) the half-concentrated MS (termed MS 1/2), which contained ammonium nitrate (10.30 mM) and potassium nitrate (9.40 mM). Buffered MS/2 consisted of regular MS/2 media supplemented with MES (10 mM; pH 5.6).

2.5 Analysis of C6-HSL degradation as a function of plant species and growth conditions

At time zero, t0 (germination), the various MS formulations (30 mL) were supplemented (or not) with 20 μM C6-HSL. Control media sustaining no plant growth but supplemented with C6-HSL were set up and processed as the plant growth media. The pH of the media was monitored at 0, 72 h (3 days) and up to 144 h (6 days). The apparent residual concentration of C6-HSL in the media was determined on samples taken at the same time pH was measured, using the assay described above. Determinations of pH and C6-HSL concentration were performed each time in triplicate. The whole experiment was independently performed twice. Values obtained from the measures of violacein production zones (spot diameter) were converted into residual concentrations of C6-HSL via a calibration curve, averaged, and plotted as a function of time, plant, and growth-medium used.

Detailed analysis of the disappearance of C6-HSL from the incubation medium of L. corniculatus was performed in a micro-titration plate as follows. Per micro-well, four surface-sterilized seeds were germinated in 100 μL of MS/2 medium that was buffered (or not) with MES (pH 5.6; 10 mM). The culture media were supplemented with 25 μM C6-HSL when appropriate. The presence of C6-HSL was evaluated at sampling times in 10 μL (i.e. representing 0.4 plantlets) of the culture medium, taken from four independent series of 10 micro-wells, for each experimental condition and as described in Section 2.2. Similar experiments were performed using AHLs with longer acyl side chain or 3-oxo substitutions, i.e. 3-oxo-C6-HSL, 3-oxo-C8-HSL (both at 25 μM) and 3-oxo-C10-HSL (40 μM).

The putative presence of C6-HSL inhibitors in the plant growth media was investigated in the absence of C6-HSL by directly subjecting 10 μL (i.e. representing 0.4 plantlets) samples taken from the growth media of plants at 144 h, to a reverse test performed using biosensor C. violaceum CV026 and 0.25 μM C6-HSL, as described by McClean et al.[13]. The presence of compounds inhibiting the sensing of C6-HSL by biosensor C. violaceum CV026 in plantlet extracts was investigated similarly, i.e. by subjecting series of 10 μL aliquots of plantlet extract generated from plants grown in the absence of C6-HSL, and taken at 144 h to the above-described reverse test. Plantlet extracts were obtained by crushing four plantlets (grown in 100 μL buffered MS/2 medium, supplemented or not with 20 μM C6-HSL) in 50 μL water. Moreover, the possible presence of C6-HSL in plants, as a result of sequestration of the molecule by the Lotus roots, was investigated by subjecting 10 μL (0.8 equivalent plantlet) of plantlet extracts obtained as described above from plantlets grown in the presence of C6-HSL to direct TLC analysis using the CV026 biosensor (see Section 2.2).

The disappearance of C6-HSL from the L. corniculatus growth medium was also confirmed through separation of compounds extracted by ethyl acetate (v/v) from the Lotus growth medium by HPLC analysis of concentrated extracts (see Section 2.2). Prior to HPLC separation, the ethyl acetate extracts were evaporated to dryness and resuspended in acetonitrile in a volume allowing a 40-fold concentration of the extract.

2.6 In vitro assays for C6-HSL or 3-oxo-C6-HSL degradation

In vitro assays for C6-HSL or 3-oxo-C6-HSL degradation were performed using L. corniculatus crude extracts. The extracts were obtained by crushing to homogeneity in a mortar 12 plantlets (6 days-old) in 1 mL of extraction buffer that consisted of 15 mM potassium phosphate, pH 6.7. The degradation assays were performed by incubating C6-HSL or 3-oxo-C6-HSL (25 μM final) with 100 μL of the Lotus extract in a final volume of 200 μL for up to 24 h at 24 °C. A control experiment was performed using extracts that were heat-processed (15 min., 100 °C). The residual amounts of C-6HSL and 3-oxo-C6-HSL were determined using the biosensors Chromobacterium CV026 and Agrobacterium NTLR4, as indicated in Section 2.2.

3 Results and discussion

3.1 Analysis of pH- and temperature-dependent half-life of C6-HSL

C6-HSL was incubated for up to 21 days in aqueous buffers at various pH levels and temperatures. The residual amounts of C6-HSL were determined in the incubation media using a spot test assay and the biosensor C. violaceum CV026, as indicated in Section 2. Half-life values were calculated from these data (Table 1). Half-life values ranged from less than 4 h to over 21 days. C6-HSL was more stable at −20 °C than at 37 °C and more stable under acidic pH conditions than under alkaline conditions. At −20 °C, the stability of C6-HSL was “constant” up to pH 8, i.e. >21 days, but was reduced to 10 days at pH 9. In contrast, at 37 °C, C6-HSL exhibited a very limited half-life, even under moderately acidic pH. Under both alkaline pH (>8.0) and physiological temperatures (20–37 °C), very short half-life values – ranging from 16 to 4 h were calculated. Additionally, at neutral pH and at temperatures ranging from 4 to 37 °C, the half-life of C6-HSL varied from less than 1 up to 7 days.

View this table:

Half-life of C6-HSL as a function of pH and temperature

TemperaturepH and buffers
Tris–maleate bufferTris–HCl buffer
−20 °C>21a>21>21>21>21>21>211710
+4 °C>21>2114976332
+20 °C877<133<1(12)b<1(6)<1(4)
+28 °C854422<1(16)<1(6)<1(4)
+37 °C544<132<1(10)<1(4)<1(<4)
  • aHalf-life values are given in days.

  • bFor short time half-life (i.e. <1 day), values were re-examined as explained in the text. Values between parentheses are half-life in hours.

The above results are overall consistent with previous studies on the persistence of AHLs that demonstrated that the lactone ring is highly sensitive to elevated temperatures and alkaline hydrolysis. Byers et al.[6] reported that another AHL, namely 3-oxo-C8-HSL, rapidly disappears from a reaction medium upon boiling but not upon freezing. They also observed the depletion of 3-oxo-C8-HSL from a supernatant of Erwinia culture in late log and stationary phases under conditions leading to the alkalinization of spent supernatants. A similar depletion of AHLs was observed in the culture supernatants of both Pseudomonas aeruginosa and Yersinia pseudotuberculosis and correlated with alkalinization of the media [7]. The latter authors confirmed that the disappearance of the signal is related to lactonolysis, i.e. the opening of the lactone ring. Lactonolysis leads to the conversion of AHLs to the cognate N-acyl homoserine derivatives, which are not recognized as QS signals. In addition, short chain AHLs are more prone to temperature-dependent lactonolysis than long chain AHLs. From the previous work of Byers et al.[6] and Yates et al.[7], the half-life of 3-oxo-C8-HSL may be estimated to be ca. 7 h at pH 8.0 and 30 °C, a value comparable to that reported here for C6-HSL.

3.2 Stability of C6-HSL in the plant environment

The above findings and related reports indicate that pH and temperature influence the stability of the C6-HSL signal. In the rhizosphere, pH level is determined in part by edaphic factors (i.e. abiotic soil parameters) and also by the metabolic activity of plants and microorganisms. To investigate whether the latter parameter may have a significant effect on the stability of C6-HSL, various plants were grown under gnotobiotic and axenic conditions in the presence of different nitrogen sources, and in the presence of C6-HSL. Residual concentrations of C6-HSL and pH values were determined in the incubation medium as a function of incubation time. The results of these experiments (Fig. 1) reveal that the kinetics of C6-HSL disappearance was comparable whatever the growth medium (MS or MS 1/2). However, C6-HSL disappeared from the plant environment with kinetics that varied as a function of the plant. The stability of the AHL signal at the root system of monocots was comparable to that observed in the control experiment (with no plant). However, disappearance appeared to be much faster at the root system of legumes, especially at the root system of Lotus from where C6-HSL was undetectable after 72 h of incubation. The analysis of the pH changes affecting the growth media revealed that pH tended to decrease when monocots were grown in MS or MS 1/2 medium, but to remain stable or rise when legumes were grown in the same medium. This is likely to be related to the preferential metabolism of cations and nitrogen sources by monocots and dicots.


Evolution of C6-HSL concentration and pH in the growth media of plants cultivated under gnotobiotic conditions. Plants were germinated in liquid growth media supplemented or not with C6-HSL at time zero. Residual C6-HSL concentrations were determined using spot assays and C. violaceum biosensor CV026 at 72 and 144 h. Assays were performed in triplicate in two independent experiments. Averaged values are shown and expressed as residual concentration of C6-HSL in the media (standard error estimated: ±10%). At the time of sampling, pH values were also determined under the same conditions (standard error: ±0.1 pH unit). Upper two panels: experiments performed in MS medium; lower two panels: experiments performed MS 1/2 medium.

Though the observed pH changes were limited, they may contribute in part to the differential stability of C6-HSL in the plant environment (see Section 3.1). To address this question, the above experiment was repeated in MS 1/2 medium buffered at pH 5.6 by addition of MES (10 mM; see Section 2). The results (Fig. 2) confirmed that C6-HSL rapidly disappeared from the growth media of Lotus plantlets, but rates were comparable whatever the medium (MS 1/2 or buffered MS 1/2). Importantly, the pH of the buffered medium remained absolutely constant over the duration of the experiment. These results indicate that pH change did not contribute to the disappearance of C6-HSL from the plant environment, as initially hypothesized.


Evolution of C6-HSL concentration in the growth media of Lotus plants cultivated in buffered and non-buffered media. Plants were germinated in liquid MS 1/2 growth media with or without buffering at pH 5.6 using MES, supplemented or not with C6-HSL at time zero. Residual C6-HSL concentrations were determined using spot assays and C. violaceum biosensor CV026 at 72 and 144 h. Assays were repeated ten times. Averaged values are shown and expressed as residual concentration of C6-HSL in the media (standard error: ±5%).

3.3 L. corniculatus plantlets naturally degrade C6-HSL

Three hypotheses could explain the disappearance of C6-HSL from the Lotus growth medium. First, Lotus may produce compounds such as those detected in Delisea pulchra, a red algae [16] and other plants [17], that inhibit the sensing of the C6-HSL signal by the biosensor C. violaceum CV026. Second, Lotus plantlets may cause the disappearance of C6-HSL from the growth medium by sequestering C6-HSL within its tissues. Third, one or more enzymatic activities found in Lotus plantlets may actively degrade C6-HSL, as this has been reported for Prokaryotes (see [1012,18]). The search for inhibitors was performed on 10 μL samples of plant growth media as indicated in Section 2.5. The reverse assay performed before and after TLC separation of the compounds present in at least three independent series of 10 μL samples did not reveal any activity inhibiting the sensing of C6-HSL by C. violaceum in these samples (data not shown). Similarly, no activity inhibiting the sensing of C6-HSL by C. violaceum was detected in three independent 10 μL aliquots of plantlet extracts. Additionally, no C6-HSL was detected in the extracts obtained from plants grown in the presence of this molecule (experiment performed in triplicate, data not shown), suggesting that Lotus plantlets did not accumulate C6-HSL. However, HPLC analysis of the plant growth medium revealed that C6-HSL had indeed disappeared from this medium upon growth of the plantlets (Fig. 3) that therefore, appear to be able to degrade C6-HSL.


HPLC analysis of the C6-HSL content of the culture supernatants of Lotus plantlets. Plants were germinated in MS medium supplemented with C6-HSL at time zero. The media were extracted at 144 h with ethyl-acetate and concentrated 40 times. The C6-HSL content of 15 μL of concentrated extracts (corresponding to 600 μL of media) was analyzed by HPLC. The peak with a retention time of 10.5 min has characteristics identical to that of synthetic C6-HSL. Dotted grey line, medium extracted at time zero (i.e. at the time the seeds were added); continuous black line, medium extracted at 144 h. The elevated base line in this latter case is due to the release of UV-absorbing compounds by germinating seeds and developing plantlets.

3.4 L. corniculatus plantlets degrade other AHLs and degradation ability is heat-sensitive

Two questions arose from the results above. Is the activity enzymatic, and is it specific for C6-HSL? To answer the second question, an experiment similar to that reported above in Section 3.2 was set up in buffered medium (see Section 2), but C6-HSL was omitted and replaced with 3-oxo-C6-HSL, 3-oxo-C8-HSL or 3-oxo-C10-HSL. Under those conditions, experimental data revealed that over 99% of input-AHLs were readily degraded by Lotus plantlets in less than 144 h (6 days; data not shown). These results suggest that the activity is not specific for unsubstituted AHLs as both C6-HSL and 3-oxo-C6-HSL were degraded. Similarly, the activity is not specific fort short-chain AHLs as both 3-oxo-C6-HSL and 3-oxo-C10-HSL were degraded.

The putative enzymatic nature of the activity was assessed using Lotus crude extracts. These were incubated in the presence of C6-HSL and 3-oxo-C6-HSL. Both AHLs appeared to be degraded very rapidly by the crude extracts, even diluted at 1/10, unless these were boiled. Under the latter condition no degradation was observed (Fig. 4).


Evolution of C6-HSL and 3-oxo-C6-HSL concentration during incubations with Lotus crude extracts. Plants were germinated in liquid MS 1/2 growth media. Crude extract were obtained by crushing plantlets in a potassium phosphate buffer (pH 6.7). The degradation assay was performed by incubation of these extracts with C6-HSL and 3-oxo-C6-HSL. Residual concentrations were determined using spot assays, and C. violaceum and A. tumefaciens biosensors at the times indicated on the graphs. Extracts were used undiluted (Ext/1) or diluted as indicated on the graph (e.g. Ext/10). “Boiled” indicates the results of the experiments performed with the heat-processed extracts, while “control” indicates the results of the incubation assay performed without extract (AHL + buffer only). Averaged values are shown and expressed as residual concentrations of input AHL in the media (standard error: ±5%).

The above results demonstrate that plants from the legume species L. corniculatus are able to inactivate C6-HSL, as well as 3-oxo-C6-HSL, 3-oxo-C8-HSL and 3-oxo-C10-HSL, via a degradative process. The heat-sensitivity experiments reported above strongly support the view that the degradation of AHLs by the Lotus plants may depend on one or more enzymatic activities. Two degradative processes and relevant enzyme activities have been described in bacterial species. The first one was a lactonase, responsible for the opening of the lactone ring [8,18]. More recently, another activity, an acylase-like, has been identified in several Gram-negative and Gram-positive bacteria [10,11,19]. From the results reported here, it is not possible, however, to determine which type of activity is involved in the degradation of the C6-HSL by the Lotus plantlets.

4 Concluding remarks

This is the first report demonstrating that a germinating plant may inactivate AHLs. Recently, mammalian cells have been shown to degrade the long-chain AHL, N-(3-oxododecanoyl)-homoserine lactone (3-oxo-C12-HSL)[20], which is produced by the human pathogen, P. aeruginosa (for a review see [21]). This observation and others related to the quenching of the QS signals suggest that degradation systems targeting AHLs may play a role as a defense mechanism, naturally or upon engineering. Such a mechanism could help individual bacterial cells or eukaryotes to combat competitor microorganisms or disease-causing pathogens, which rely upon AHL-dependent QS regulation to control antibiotic production and virulence, respectively [8,10,12,20,22,23]. It is tempting to speculate that the natural AHL degradation capability of the Lotus plants described may also constitute a defense mechanism. Alternatively, a role in the molecular dialogue that takes place between the plant and the beneficial symbiont Mesorhizobium loti[24] cannot be excluded. This assertion is strengthened by recent results demonstrating that in a closely-related legume/Rhizobiaceae interaction, both the bacterial symbiont [25] and the host plant [26] respond to AHL signals. Elucidation of the ecological role of the AHL degradative ability of Lotus now requires the identification of both biochemical activity and corresponding genes.


This work was made possible by E.U. grants “Ecosafe” (QLK3-2000-01759) to Y.D. and P.W. C.d'A.-P. was also supported by the E.U. grant “Ecosafe”, while S.U. was a Ph.D. fellow student from French Ministère de la Recherche.


  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]
View Abstract