OUP user menu

The involvement of GSH in the activation of human Sod1 linked to FALS in chronologically aged yeast cells

Aline A. Brasil, Allan Belati, Sérgio C. Mannarino, Anita D. Panek, Elis C. A. Eleutherio, Marcos D. Pereira
DOI: http://dx.doi.org/10.1111/1567-1364.12045 433-440 First published online: 1 August 2013


Mutations in Cu, Zn-superoxide dismutase (Sod1) have been associated with familial amyotrophic lateral sclerosis, an age-related disease. Because several studies suggest that oxidative stress plays a central role in neurodegeneration, we aimed to investigate the role of the antioxidant glutathione (GSH) in the activation of human A4V Sod1 during chronological aging. Transformation of wild-type and A4V hSod1 into a gsh null mutant and in its parental strain of Saccharomyces cerevisiae indicated that during aging, the number of viable cells was strongly influenced by A4V hSod1 mainly in cells lacking GSH. Activity of hSod1 increased in response to aging, although the increase observed in A4V hSod1 was almost 60% lower. Activation of hSod1 (A4V and WT) did not occur after aging, in cells lacking GSH, but could still be observed in the absence of Ccs1. Furthermore, no increase in activity could be seen in grx1 and grx2 null mutants, suggesting that glutathionylation is essential for hSod1 activation. The A4V mutation as well as the absence of GSH, reduced hSod1 activity, and increased oxidative damage after aging. In conclusion, our results point to a GSH requirement for hSod1 Ccs1-independent activation as well as for protection of hSod1 during the aging process.

  • hSod1
  • GSH
  • familial amyotrophic lateral sclerosis
  • A4V hSod1
  • aging
  • Saccharomyces cerevisiae


Amyotrophic lateral sclerosis (ALS) is an age-dependent and progressive neurodegenerative disease eventually leading to paralysis and death (Cleveland & Rothstein, 2001). Among potential mechanisms of motor neuron degeneration, increased oxidative stress appears to be an early and sustained event related to motor neuron death (Bogdanov et al., 1998; Liu et al., 1998, 2002; Shaw & Eggett, 2000; Li et al., 2003). Almost 10% of cases have a genetic component and are termed familial (FALS), from which 25% are linked to dominant mutations in the cytoplasmic copper, zinc superoxide dismutase (Sod1) (Deng et al., 1993; Rosen et al., 1993). This enzyme, from yeast to humans, has a crucial role in cell physiology because its inactivation leads to a severe cellular dysfunction (Carlioz & Touati, 1986; Gralla & Valentine, 1991). Hence, Sod1 appears to be the major enzyme involved in removing superoxide anions from the cytoplasm and possibly also in protecting cells against respiration-derived superoxide (Deng et al., 1993; Rosen et al., 1993; Fukai & Ushio-Fukai, 2011).

Aging is a normal process related to critical challenges in cellular redox status, strengthening the oxidative stress theory of motor neuron death and FALS progression (Dufour & Larsson, 2004). Converging evidence indicates that aberrant aggregation of mutant Sod1 is strongly implicated in FALS (Bruijn et al., 1998). Treatments that deplete the cellular pool of the antioxidant glutathione (GSH) exacerbate mutant Sod1s insolubility, whereas an overload of intracellular GSH or overexpression of glutaredoxin-1, which catalyzes the reduction of protein-GSH-mixed disulfides, significantly increases mutant Sod1s solubility (Cozzolino et al., 2008). Recently, it has been demonstrated that GSH and glutathionylation are fundamental to protect Sod1 sulfhydryl residues under mild oxidative stress, enabling Sod1 activation and lifespan extension (Mannarino et al., 2011). Thus, the antioxidant GSH emerges as an important factor associated with aging and, possibly, neurodegenerative processes in FALS.

The current study aimed to investigate the involvement of GSH in A4V mutant Sod1 activation during chronological aging. A4V is the most common mutation discovered up to this day that accounts for approximately 50% of Sod1-linked FALS cases (Rosen et al., 1994; Cudkowicz et al., 1997). Herein, lifespan extension, Sod1 activity and biomarkers of oxidative damage were analyzed in Saccharomyces cerevisiae strains, expressing human Sod1 (wild-type or A4V mutant), deficient or not in GSH synthesis. Currently, as the best-understood eukaryotic organism, S. cerevisiae is a very attractive model for elucidating the molecular mechanisms underlying neurodegenerative diseases and for helping in developing novel therapeutical strategies (Pereira et al., 2003; Khurana & Lindquist, 2010; Tenreiro & Outeiro, 2010). In fact, S. cerevisiae englobes some of the most important features to be considered as a robust eukaryotic model system: it is nonpathogenic, more than 50% of the yeast genome has human homologs, and it is highly amenable to analysis (Khurana & Lindquist, 2010; Tenreiro & Outeiro, 2010).

Materials and methods

Yeast strains and growth conditions

Yeast cells were transformed using a YEp351, containing either one copy of wild-type (WT) or A4V mutant human SOD1 under the control of the SOD1 yeast promoter, by the lithium acetate method (Pereira et al., 2003). Transformed cells were selected by plating on solid YNB medium (2% glucose, 0.67% yeast nitrogen base without amino acids, 0.01% appropriate auxothrophic requirements and 2% agar). LEU2 was used as selectable genetic marker, in the wild-type S. cerevisiae strain BY4741 (MATa; his3; leu2; met15; ura3) or in its isogenic mutants gsh1, ccs1, grx1, and grx2. Wild-type and mutant yeast strains were acquired from Euroscarf, Frankfurt, Germany. Plasmids containing WT hSOD1 or A4V hSOD1 were a kind gift from Dr. Edith Gralla, UCLA, USA (Gralla & Valentine, 1991; Rabizadeh et al., 1995). In all experiments, yeast cells were grown in liquid minimal synthetic medium (2% SD – 2% glucose, 0.67% yeast nitrogen base without amino acids, and 0.01% nutritional requirements), at 28 °C and 160 rpm, with the ratio of flask volume : medium of 5 : 1.

Western blot analysis

Total protein extracts were obtained by disruption of cells with glass beads in 50 mM sodium phosphate buffer pH 7.0 (Pereira et al., 2003). Protein concentration was determined according to Stickland (1951). Soluble proteins obtained after centrifugation of crude extracts at 21 000 g for 5 min were fractionated by SDS-PAGE, using 15% polyacrylamide gels and a running buffer containing 25 mM Tris-HCl (pH 8.3), 192 mM glycine, and 0.5% SDS. Protein samples were dissolved in Laemmli buffer with or without β-mercaptoethanol (β-ME) and boiled at 95 °C for 5 min before loading (Laemmli, 1970). For Western blot analysis, proteins were electrophoretically transferred to nitrocellulose membranes, incubated for 2 h with a rabbit anti-human Sod1 antibody (Sigma-Aldrich) and a further 2 h with goat anti-rabbit IgG conjugated to horseradish peroxidase (Sigma-Aldrich). In both, Western and native electrophoresis, 30 μg protein were loaded.

Chronological aging and determination of life span

Cells harvested at the early stage of the first exponential phase of growth (0.8 mg dry weight per mL) were harvested at 2500 g for 5 min and washed twice with distilled water. Chronological life span was assayed as previously described (Harris et al., 2003) in washed cells resuspended in water and incubated at 37 °C/160 rpm during 24 h (MacLean et al., 2001; Mannarino et al., 2008). Aged cells were harvested, properly diluted and plated on 2% YPD (2% glucose, 1% yeast extract, and 2% peptone) to determine the chronological life span. Colonies were counted after incubation at 28 °C for 72 h. Life span was expressed as percentage of viable cells that were still alive after the aging process. As a control, cells were plated and colonies counted before aging. All cells were 100% viable before being chronologically aged.

Sod activity

Protein extracts were obtained by disruption of cells, submitted or not to the aging process, with glass beads in 50 mM sodium phosphate buffer pH 7.0 (Mannarino et al., 2011). Protein contents were determined by Stickland (1951). Sod1 activity was measured in situ after native polyacrylamide gel electrophoresis from 30–40 μg of protein extract in the presence of riboflavin and NBT (Flohe & Otting, 1984; Mannarino et al., 2011). After NBT reaction, polyacrylamide gel electrophoresis was digitalized on EC3 imaging system from UVP (UVP Bioimaging systems), and Sod1 bands were analyzed taking into consideration the area density by the use of UVP Vision Works LS 6.2 (Imaging Acquisition and Analysis Software). Activity was expressed as a fold increase in hSod1 activity between BY4741 or gsh1 transformed cells (either with WT or with A4V hSod1) and BY4741 or gsh1 (nontransformed) cells, respectively. In this work, determination of human Sod1 was performed by normalization with the yeast Sod1. In all gel electrophoresis, there was a line for yeast Sod1 used to normalize human Sod1 activities.

Intracellular oxidation analysis

The oxidant-sensitive probe 2′, 7′-dichlorofluorescein diacetate (DCF) was used to assess intracellular oxidation during chronological aging. Fluorescence was measured using a PTI (Photo Technology International) spectrofluorimeter set at an excitation wavelength of 504 nm and an emission wavelength of 524 nm (Brennan & Schiestl, 1996; Davidson et al., 1996). Before and after aging, 50 mg of cells were cooled on ice, harvested by centrifugation and washed twice with 50 mM phosphate buffer, pH 6.0. The pellets were resuspended in 500 μL of the same buffer, and 1.5 g of glass beads was added. The samples were lysed by cycles of 1-min agitation on a vortex mixer followed by 1 min on ice, repeated three times. The supernatant solutions were obtained after centrifugation at 21 000 g for 5 min, diluted sixfold with water and, then, fluorescence was measured. Intracellular oxidation determined as a ratio between fluorescence of gsh1 and BY4741 recombinant cells was expressed as a fold increase in oxidation.

Determination of protein carbonylation

Protein carbonylation, used as a biomarker of protein oxidation, was assayed by slot blot using a primary anti-DNP (dinitrophenyl) followed by treatment with a secondary antibody conjugated with peroxidase (Adamis et al., 2009). Yeast extracts were prepared in 50 mM potassium phosphate buffer, pH 7.0, and 0.1 mM EDTA, containing a protease inhibitor cocktail (Complete, Mini, EDTA-free Protease Cocktail Inhibitor Tablets; Boehringer Mannhein). Protein content was estimated according to Stickland (1951) using bovine serum albumin as standard. Proteins (15 μg) were slot-blotted and derivatized with 2,4-dinitrophenylhydrazine (0.1 mg) into Polyvinylidene fluoride (PVDF) membranes (Hybond-PVDF, GE Healthcare, Europe). PVDF membranes were probed with rabbit IgG anti-DNP (Dako, Glostrup, Denmark) at a 1 : 5000 dilution as the primary antibody, and goat anti-rabbit IgG conjugated to horseradish peroxidase (Sigma, St. Louis, MO) at a 1 : 5000 dilution as the secondary antibody. Finally, the blots were developed with diaminobenzidine and H2O2 plus CoCl2 to enhance sensitivity. Quantification of carbonyls was performed by densitometry and expressed as fold increase in protein carbonyls.

Statistical analysis

The results were expressed as mean ± standard deviation of at least three independent experiments. Statistical differences were tested using anova followed by Tukey–Kramer multiple comparison test. The latter denotes homogeneity between experimental groups at P < 0.05.

Results and discussion

To bring light into the role of glutathione (GSH) in the age-dependent mechanism of hSod1-linked FALS activation, we transformed the human wild-type (WT) and the A4V mutant SOD1 into S. cerevisiae strains, BY4741 and its isogenic mutant gsh1Δ. The mutant lacks γ-glutamylcysteine synthetase, which catalyzes the first step of GSH biosynthesis. Yeast expression of human Sod1 (hSod1) was confirmed by Western blotting using a specific antibody against hSod1 (Fig. 1), which is not able to detect the yeast Sod1 counterpart. This result confirms that both WT and A4V hSod1 were expressed in our experimental conditions. Interestingly enough, a decrease in hSod1 expression, which is under the control of the yeast SOD1 promoter, was observed in the gsh1 mutant strain. This reduction was mainly observed in the wild-type form of hSod1. Recently, it has been published that gsh1-deleted cells, growing in similar conditions as presented here, showed a lack of Yap1 oxidative activation (Kumar et al., 2011). Because yeast SOD1 is up regulated by Yap1 under stress conditions, presumably, the lack of Yap1 activation in the gsh1 mutant might be the cause for the reduction in hSod1 expression before and after aging.


Saccharomyces cerevisiae cells express both WT and A4V hSod1. Protein extracts were prepared from cells, expressing WT or A4V hSod1, collected before and after aging. As control, we used the nonrecombinant BY4741 strain. Prior to denaturing gel electrophoresis, samples containing 30 μg of total protein extract were heated at 95 °C in SDS-buffer containing or not 5% β-mercaptoethanol as a reducing agent. Protein bands were developed with Western blot using an antibody that only recognizes human Sod1.

Next, we investigated the effect of hSod1 expression on the viability of yeast cells under the exposure to chronological aging. Chronological aging was carried out by transferring cells grown on 2% SD to nonproliferating condition (water) and incubated at 37 °C to accelerate the chronological aging process. Chronological aging is characterized by damages in lipids, proteins, and DNA leading to loss of cellular integrity during nonproliferating condition (growth-arrested state), such as stationary phase or water. In this scenario, life span (viability) is measured by the capacity of yeast cells to undergo cell division/proliferation after a long period of quiescence. To the best of our knowledge, transferring exponential cells to water became an alternative for aging cells in nonproliferating conditions and in avoiding production and secretion of acetic acid to the medium. Taken together, we decided to adapt our chronological aging assay submitting exponential cells to a nonproliferating condition, water. However, the process of aging in water can be slow at 28–30 °C, taking some weeks to measure life span. Thus, we decided to submit aging cells to 37 °C to accelerate the process. Indeed, it is important to note that our experimental conditions (water at 37 °C) may introduce some degree of thermal stress which in fact could sensitize cells with increased intensity. It had been described that differences observed in life span measured at 30 °C are reproducible at 37 °C in water (MacLean et al., 2001; Harris et al., 2005; Piper et al., 2006; Mannarino et al., 2008). After multiple time points analysis (4, 18, 24, 30, and 48 h), results not shown, we decided to study the molecular events of chronological aging only after 24 h. Prior to this time, yeast life span was very similar without statistically significant results between strains. On the other hand, the life span was seriously affected when long-term aging was performed. The aging process we used showed to be extremely severe for strain BY4741, which showed only 4.4% of viable cells after 24 h of aging (Fig. 2). This result confirms a previous observation that chronological aging assayed in water at 37 °C is extremely drastic accelerating aging many times if compared with other strategies (Piper, 2006). Increased levels of antioxidants decreased ROS accumulation, a cause of aging according to the oxidative stress theory (Harman, 1981). Thus, as expected, overexpression of hSod1 in a BY4741 background increased life span considerably. In addition, a significant difference between the viability of WT and of A4V hSod1-expressing cells was observed, leading to the conclusion that the A4V substitution is detrimental even for the yeast model. To the best of our knowledge, A4V hSod1 has similar activity presented by the WT hSod1; however, it has been described that this mutation increases the levels of hydroxyl formation and protein destabilization (Borchelt et al., 1994; Yim et al., 1996). Presumably, these characteristics might contribute to the reduced life span observed in A4V hSod1-expressing cells when compared with the WT hSod1 strain. When chronological aging of yeast cells expressing hSod1 was assessed in a gsh1 null mutant, the benefits previously acquired were completely abolished, indicating that GSH is required to achieve a functional human Sod1 and, consequently, a greater viability. As observed in strain BY4741, the number of viable gsh1 cells expressing A4V hSod1 was lower than in cells expressing the WT form. It is worth noting that high susceptibility of the gsh1 mutant strain might also be related to metabolic dysfunctions (Kumar et al., 2011) exerted by the GSH1 deletion; however, in transformed cells, our results strongly suggest that the A4V hSod1 construction was critical for cell viability.


hSod1 expression increases viability of yeast cells under chronological aging. Saccharomyces cerevisiae cells, BY4741 and gsh1, harvested at exponential phase were aged in water at 37 °C for 24 h. Cellular viability was measured by plating cells on YPD 2% and expressed as the percentage of viable cells. The results represent the mean ± standard deviation of at least three independent experiments and *BY4741 hSod1 (A4V vs. WT) and gsh1 hSod1 (A4V vs. WT) mean different results at P < 0.05.

In addition to viability of cells after 24 h of aging, we investigated the involvement of GSH and aging, in drastic conditions, in WT and A4V hSod1 activation. To highlight the contribution of hSod1 to total activity, the results were expressed as a ratio between activity of recombinant and control (nonrecombinant) cells. Overexpression of WT or A4V hSod1 in the BY4741 yeast strain promoted a similar increase in hSod1 activity as before aging (Fig. 3).


Effect of aging on hSod1 activation. Sod1 activity was determined, in exponential-phase yeast cells, before (white bars) and after (black bars) chronological aging. Values are means ± SD of at least three independent experiments and represent the ratio of hSod1 activity between recombinant yeast cells harboring hSod1 and the respective nonrecombinant strain BY4741 or gsh1 * and ** represent statistically different results at P < 0.05 within each group before and after aging process, respectively.

Although the relative activities of WT and A4V mutant Sod1 increased significantly in response to aging, a notable difference between WT and A4V hSod1 activation was observed. Indeed, increased activity was due to the presence of hSod1, because the activity of the yeast Sod1 did not change after aging. In addition, the activity of human Sod1 presented here, may only be an estimation of the actual value, and because we could not resolve human Sod1 from endogenous yeast Sod1 on native gels and similars, all values reported for human Sod1 were obtained by subtracting the predicted contributions of yeast Sod1. As a result, the BY4741 strain expressing WT hSod1 showed a percentage of viable cells approximately 25% higher than mutant hSod1 (Fig. 2), suggesting that A4V mutation on hSod1 could promote a decrease in antioxidant protection either by impairment of enzymatic function or due to gain of a toxic function (Gurney et al., 1994; Bruijn et al., 2004). Sod1 is a homodimeric metalloenzyme, whose monomers contain a structural zinc ion and an independent active site that tightly binds a catalytic copper (Tainer et al., 1982). Most cells employ an accessory protein known as the copper chaperone for Sod1 (Ccs1) to facilitate copper binding (Wong et al., 2000). Yeast Sod1 is totally dependent on Ccs1 for incorporating copper in vivo (Culotta et al., 1997). However, Sod1 from humans, mouse, and C. elegans can incorporate copper by a CCS-independent pathway (Wong et al., 2000; Subramaniam et al., 2002). It has also been described that GSH is required for copper delivery in both yeast and fibroblast (derived from adult mice) ccs1 mutant cells therefore indicating that the presence of GSH is crucial for CCS-independent activation of hSod1 (Carroll et al., 2004). Besides endorsing Carroll's et al. results (Carroll et al., 2004), the data shown in Fig. 3 implement some interesting findings on the activation of hSod1 in response to aging. We noted a higher activity of both WT and A4V hSod1 mutant in aged ccs1 mutant cells, emphasizing a preference for a CCS-independent mechanism of hSod1 activation during aging, which probably is dependent on GSH. Neither WT nor A4V hSod1 were activated under GSH deficiency, explaining why hSod1 transformation was not able to increase longevity of gsh1 yeast cells (Fig. 2).

Confirming the requirement of GSH for proper activation of hSod1 during aging, we observed that the gsh1 mutant expressing WT or A4V hSod1 exhibited both higher levels of ROS and protein carbonyls compared to those observed in BY4741 WT or A4V hSod1 cells (Fig. 4). Interestingly, in our experimental conditions, the absence of GSH synthesis in the gsh1 mutant did not alter the levels of protein carbonylation as well as intracellular oxidation at early exponential growth and 24 h aging when compared with its parental strain BY4741. In fact, lower levels of active hSod1 might explain the increased levels of oxidative biomarkers found in gsh1 mutant cells after aging. Moreover, we found that ROS levels and protein carbonylation were excessively high, after aging, in the gsh1 mutant expressing A4V hSod1. Indeed, the requirement of GSH for activating A4V hSod1 in response to aging was considerably higher than for the WT form. A motor neuron-like cell culture system and a transgenic mouse model were used by Chi et al. (2007) to study the effect of cellular GSH alteration on motor neuron cell death. The authors claim that the exposure of NSC34 motor neuron-like cells to ethacrynic acid dramatically reduces cellular GSH levels and is accompanied by increased production of reactive oxygen species (ROS) (Chi et al., 2007). Taken together, our results strongly support the idea that during chronological aging, the lack of GSH might reduce cellular viability increasing oxidative stress. Therefore, besides being important for viability during aging, GSH seems to be also involved in the protection of WT and A4V hSod1 after aging.


Glutathione deficiency increases oxidative stress biomarkers in chronological aged cells containing A4V substitution. (a) Protein carbonylation and (b) intracellular oxidation were determined in BY4741 and gsh1 recombinant cells expressing WT or A4V hSod1 before (white bars) and after chronological aging (black bars). Data were expressed as a ratio between protein carbonylation and fluorescence of gsh1 and BY4741 recombinant cells. Values are mean ± SD of three independent experiments. *represents statistically different results between after and before aging process at P < 0.05.

Different hypotheses concerning the involvement of GSH in a CCS-independent pathway for hSod1 activation and stabilization have been proposed (Wong et al., 2000; Subramaniam et al., 2002; Carroll et al., 2006). At first, the lack of GSH possibly leads to a redox imbalance causing oxidative modifications in sulfhydryl groups preventing Sod1 from being activated by Ccs1. Furthermore, it was described that GSH in a copper complex fully restores the structure of Sod1 holoenzyme through a Cu(I). GSH protein intermediate which in fact incorporates Cu(I) to the copper free, Zn-Sod1 (Ciriolo et al., 1990). At last, but not least cytosolic glutaredoxins also seem to be critical for controlling the disulfide state and stability of Sod1 (Subramaniam et al., 2002; Carroll et al., 2006). Thus, it was proven that GSH plays a critical role for in vivo Sod1 activation. However, we also suggest a possible hSod1 glutathionylation during aging by forming reversible mixed disulfides with cysteine residues. Under oxidative conditions, this mechanism could prevent oxidation of thiol groups involved with and favoring the apoprotein maturation and activation through a heterodimeric interaction with a copper-bonded Ccs1. According to Mannarino et al. (2011), glutaredoxin-1 and glutaredoxin-2 are fundamental for yeast Sod1 activation and lifespan extension. Glutaredoxins are GSH-dependent oxidoreductases acting in both mixed disulfides between GSH and a polypeptide cysteine, as well as intramolecular disulfides in proteins, targeting specific cysteine residues (Holmgren et al., 2005).

To test the involvement of glutathionylation in hSod1 activation, grx1 and grx2 mutant strains were transformed with WT and A4V hSod1. As expected, both WT and A4V hSod1 were not activated after aging confirming the involvement of glutaredoxins/glutathionylation in hSod1 activation (Fig. 5). Carroll et al. (2006) demonstrated that cytosolic glutaredoxins can reduce Sod1 disulfide cysteines affecting the stability of certain ALS hSod1 mutants and making them susceptible to protein misfolding. Furthermore, it has been observed that treatments that deplete the cellular GSH pool exacerbate mutant hSod1 insolubility, whereas an overload of intracellular GSH or overexpression of Grx-1 rescues solubility in Sod1 mutants (Cozzolino et al., 2008). The authors also reported the role of Cys-111 in affecting in vivo Sod1 mutant aggregation and cell toxicity (Cozzolino et al., 2008). Recently, in vitro experiments showed that glutathionylation at Cys-111 introduced steric clashes in hSod1, hindering association of modified monomers and impairing formation of dimers (Redler et al., 2011). Thus, we believe that under stressful conditions of increased levels of ROS, as occurs during chronological aging, hSod1 monomers might be glutathionylated avoiding formation of aggregates. For proper dimerization and further activation of hSod1, Grxs must reduce protein-glutathione disulfides. In addition, the dependence on GSH and Grx for activation of both WT hSod1 and the FALS mutant A4V in response to aging is the first evidence that in a yeast system, glutathionylation might affect activation of hSod1. In addition, we also suggest that activation and formation of hSod1 oligomers may occur by similar mechanisms in both sporadic and familial ALS cases of motor neuron degeneration. Therefore, GSH emerges as a prominent target in the molecular mechanism of neurodegenerative diseases associated with intracellular protein aggregates such as FALS.


Grx1 and Grx2 are necessary for WT or A4V hSod1 activation after chronological aging. Sod1 activity was determined in exponential-phase yeast cells, before (white bars) and after (black bars) chronological aging. Values are means ± SD of at least three independent experiments and represent Sod1 activity ratio between recombinant yeast cells with hSod1 and the respective nonrecombinant strain grx1 or grx2. * represents statistically different results between after and before aging process at P < 0.05.


This work was supported by grants from FAPERJ, CAPES and CNPq.


  • Editor: Ian Dawes


View Abstract