β-Glucosidase activities of lactic acid bacteria: mechanisms, impact on fermented food and human health
Christian Doppler Research Laboratory for Innovative Bran Biorefinery, Department of Food Science and Technology, University of Natural Resources and Life Sciences (BOKU), Vienna, Austria
Correspondence: Herbert Michlmayr, Christian Doppler Research Laboratory for Innovative Bran Biorefinery, Department of Food Science and Technology, University of Natural Resources and Life Sciences (BOKU), Muthgasse 18, A-1190 Vienna, Austria. Tel.: +43 (1) 47654 6151; fax: +43 (1) 47654 6251; e-mail: firstname.lastname@example.org
Through the hydrolysis of plant metabolite glucoconjugates, β-glucosidase activities of lactic acid bacteria (LAB) make a significant contribution to the dietary and sensory attributes of fermented food. Deglucosylation can release attractive flavour compounds from glucosylated precursors and increases the bioavailability of health-promoting plant metabolites as well as that of dietary toxins. This review brings the current literature on LAB β-glucosidases into context by providing an overview of the nutritional implications of LAB β-glucosidase activities. Based on biochemical and genomic information, the mechanisms that are currently considered to be critical for the hydrolysis of β-glucosides by intestinal and food-fermenting LAB will also be reviewed.
Lactic acid bacteria (LAB) of the taxonomic order Lactobacillales (phylum Firmicutes) play an indispensable role in most societies and have been exploited throughout recorded history (Makarova et al., 2006). They are an integral part of traditional food processing and preservation technologies, including the fermentation of dairy products, vegetables and meat. The evolution of LAB is tightly connected to the consumption of fermented foods and many LAB species were ‘domesticated’ and passed down the generations in the course of emerging food processing technologies. LAB display an evolutionary trend towards genome reduction and metabolic simplification, which relates to the adaptation to mainly anthropogenic habitats rich in proteins and carbohydrates. Analyses of LAB genomes revealed that acquisition and duplication of genes involved in carbohydrate transport and metabolism occurred early in LAB evolution (Makarova et al., 2006). The high redundancy of such genes observed in LAB genomes is accompanied by a high fraction of pseudogenes, serving as further evidence for the adaptation to nutrient-rich environments and ongoing genome reduction (Schroeter & Klaenhammer, 2009).
The metabolism of LAB appears to be interlaced with human nutrition and LAB are mostly valued for their generally positive impact on sensory appearance and dietary value of fermented foods. Furthermore, LAB have adapted to the mucosal surfaces of the gastrointestinal tract and play a potential role in maintaining human health. Together with the Bifidobacteriaceae, some of their members form an essential part of the beneficial probiotic gut microbiota (Matamoros et al., 2013; van Baarlen et al., 2013). β-d-Glucosidase activity is widespread among LAB and presumably plays a substantial role in the interaction with the human host as well, as β-glucosidases release a wide range of plant secondary metabolites from their β-d-glucosylated precursors. For example, plant metabolite deglycosylation was shown to improve the flavour or fragrance of fermented products. It also increases the bioavailability of health-promoting, antioxidative plant metabolites. However, adverse health effects caused by increased bioavailability of dietary toxins and xenobiotics were also reported. Furthermore, β-glucosidases may play an active role in the carbohydrate metabolism of beneficial and pathogenic LAB species.
Driven by these factors, much effort was undertaken to improve our understanding of LAB β-glucosidases. The advances in genome sequencing have further shown that putative β-glucosidase genes are more common and potentially redundant in LAB genomes than expected. This has also resulted in increased awareness of the importance of the phosphotransferase system (PTS). Given these facts, an overview on the nutritional implications of LAB β-glucosidase activities is provided. Furthermore, the mechanisms that are currently considered to be critical for the conversion of β-glucosides by these bacteria will be discussed.
Plant metabolite glycosylation
Plants produce over 100 000 secondary metabolites with highly diverse chemical structures and functions. Of particular dietary importance are phenols and polyphenols, including cinnamic acids and the important class of flavonoids. Traditionally, dietary phenolic compounds have had a rather negative image due to adverse nutritional effects and their bitterness. However, the recognition of their antioxidant capacities caused a marked change in direction. Thus, the uptake of dietary phenols, especially of flavonoids, is now recommended to prevent cancer and cardiovascular diseases (Gachon et al., 2005; Rodríguez et al., 2009).
Glycosylation is an important chemical modification of such hydrophobic plant metabolites. Besides providing water solubility and chemical stability of the aglycone, it is effective in detoxifying endogenous metabolites as well as xenobiotics, for example, by providing defence against mycotoxins (Gachon et al., 2005; Berthiller et al., 2011). The non-reducing end sugar molecule of a glycoside is also termed the glycone, the remaining compound the aglycone. Glycosylation is catalysed by glycosyltransferases, and most common is the formation of monoglucosides. Sugars such as rhamnose, arabinose, apiose, galactose or xylose can also occur in plant metabolite glycosylation. In such cases, the second sugar moiety is usually conjugated to the primary glucose molecule, forming a diglycoside. Sequential enzymatic hydrolysis of such glycosides requires specific glycosidases that first remove the terminal residues (rhamnosidase, arabinosidase, etc.). The aglycone is liberated from the resulting monoglucoside by a β-glucosidase. The mode of glycosylation also determines the bioavailability of a metabolite. In general, glycosides are less well absorbed from the small intestine than their respective aglycones. In the case of glycosylated flavonoids, only monoglucosides have been reported to be absorbed from the small intestine (Hollman, 2004).
β-Glucosidase classification and functionality
By definition, β-d-glucosidases (EC 126.96.36.199) remove glucopyranosyl residues from the non-reducing end of β-d-glucosides by catalysing hydrolysis of the glycosidic bond. β-d-Glucosidases occur abundantly in all domains of life and exert diverse biological functions (Cairns & Esen, 2010). With regard to their substrate specificities, the main distinction is made between enzymes such as the cellobiases that preferentially hydrolyse di- or short-chain oligosaccharides and alkyl- or aryl-glucosidases (Bhatia et al., 2002). Many EC classes have been assigned to β-glucosidases based upon aglycone specificities (Cairns & Esen, 2010). However, most β-glucosidases hydrolyse a broad range of substrates and to some extent such distinctions seem to be arbitrary. While the hierarchical EC nomenclature does not reflect functional relationships between enzymes, the glycoside hydrolase (GH) family classification system (Henrissat & Davies, 1997), accessible through the Carbohydrate Active enZyme database CAZy (Cantarel et al., 2009), takes into account amino acid sequence, structural and functional relationships between GHs. At present, β-glucosidases are assigned to the GH families 1, 3, 5, 9, 30 and 116. Based upon GH classification, glycone specificity (e.g. whether an enzyme would be a β-galactosidase or a β-glucosidase) can reliably be predicted through sequence comparison. Glycone recognition motifs are usually highly conserved within GH families and subfamilies. Conversely, aglycone specificities of β-glucosidases are determined by less conserved amino acid residues, resulting in the broad functional diversity generally observed with such enzymes. Due to this high variability, it is difficult to predict aglycone specificity and hence the exact function of an enzyme based on sequence or structure alone. However, it was recently demonstrated that molecular dynamics simulation may be a powerful tool to approach this difficulty (Grandits et al., 2013).
Glucosidase activities of food-fermenting LAB
LAB involved in plant food fermentations have been investigated extensively with regard to the release of plant metabolites through β-glucosidase activities. The following recent examples should be considered as a selection demonstrating the relevance of the issue. An exhaustive collection of all existing literature on this topic is beyond the scope of this article.
Oleuropein is a phenolic glucoside responsible for the bitterness of unprocessed olives, and partial degradation of oleuropein is required before table olives can be consumed. The most important species for the fermentation of olives is Lactobacillus plantarum. However, L. pentosus, L. brevis and Pediococcus pentosaceus have been isolated from fermented olives as well. Most of these isolates were able to hydrolyse oleuropein through β-glucosidase activity (Ghabbour et al., 2011).
The isoflavones genistein and daidzein (‘phyto-oestrogens’) are much discussed because of their potential health benefits. Soybeans contain high concentrations of their β-glucosides genistin and daidzin. These were reported to be hydrolysed by β-glucosidase activities of LAB during soy milk fermentations. Strains of Streptococcus thermophilus, L. acidophilus, L. delbrueckii ssp. bulgaricus, L. casei, L. plantarum, L. fermentum and several Bifidobacterium species could increase the concentrations of genistein and daidzein in soy milk (Chien et al., 2006; Donkor & Shah, 2008; Rekha & Vijayalakshmi, 2011). Fermentations with LAB could also increase the concentrations of bioactive isoflavones present in Samso-Eum, a traditional oriental herbal medicine formula (Cho et al., 2012).
Anthocyanins belonging to the flavonoid family are the natural pigments of many fruits and vegetables and are of interest as food additives due to their antioxidant capacities. Ávila et al. (2009) studied the hydrolysis of the β-glucosides malvidin and delphinidin by species of Lactobacillus and Bifidobacterium. All strains involved in the study could hydrolyse these glycosides.
Cassava is an important source of dietary energy in tropical regions. It is traditionally fermented by LAB with L. plantarum as one of the most important species. Other species of the genera Lactobacillus, Leuconostoc and Weissella were also identified in natural cassava fermentations (Kostinek et al., 2007). Cassava contains high concentrations of the toxic cyanogenic glucoside linamarin, and LAB contribute to linamarin degradation by β-glucosidase activities (Nout & Sarkar, 1999).
There is considerable interest in the β-glucosidase activities of LAB that conduct the malolactic fermentation of wine. While Oenococcus oeni is usually the desired species for malolactic fermentation, species of the genera Leuconostoc, Lactobacillus and Pediococcus can propagate in wine as well. However, Pediococcus spp. and heterofermentative Lactobacillus spp. are mainly associated with wine spoilage (Bartowsky, 2009). The main volatile constituents of the primary wine aroma are terpenoid compounds derived from the grapes. As these can be released from glycosylated precursors, the β-glucosidase activities of yeasts and malolactic bacteria are of interest due to their impact on the aroma profile of wines (Maicas & Mateo, 2005). It was shown that β-glucosidase actvities are widespread in strains of O. oeni and in wine-related strains of the genera Lactobacillus and Pediococcus. Furthermore, due to recent and ongoing research in oenology, O. oeni is at present among the best-studied LAB species with regard to its β-glucosidase activities (Michlmayr et al., 2012, and references therein).
Glucosidase and glucuronidase activities of intestinal LAB
The examples given above make a clear case for the beneficial effects of LAB β-glucosidase activities. It is important to stress that harmful metabolites can be converted to the more bioactive and bioavailable form as well, a concern that has been duly expressed by authors conducting research on the human gut microbiota. An example is the Fusarium mycotoxin deoxynivalenol. Glucosylation is a major detoxification mechanism in plant defence and deoxynivalenol-3-β-glucoside is not considered to be harmful. However, there is concern that the glucoside acts as a ‘masked’ mycotoxin, as it was shown that deoxynivalenol can be reactivated by hydrolysis of its glucoconjugated form in the digestive tract (Dall'Erta et al., 2013). While human cytosolic β-glucosidase is not able to hydrolyse deoxynivalenol-3-β-glucoside, several LAB strains are capable of hydrolysing this glucoside (Berthiller et al., 2011).
A related health concern is represented by intestinal β-glucuronidase activities (Dabek et al., 2008). Human glucuronosyltransferases (UGT1 family) expressed in the gastrointestinal tract and in the liver facilitate the excretion of toxic metabolites and xenobiotics by forming highly hydrophilic β-d-glucuronides (Basu et al., 2004). Glucuronidases antagonize this detoxification mechanism by promoting enterohepatic circulation and thus prolonged retention time of toxic/mutagenic compounds in the body (Dabek et al., 2008). It was reported that β-glucuronidase activities are present in LAB but less common than glucosidase activities (Mroczynska & Libudzisz, 2010).
The phosphoenolpyruvate (PEP)-dependent carbohydrate PTS (PEP : PTS) is a key mechanism of the bacterial sugar catabolism that directly couples substrate import with phosphorylation (Fig. 1). The PEP : PTS is mainly encountered in obligately and facultatively anaerobic bacteria where it serves as a bioenergetically efficient alternative to active substrate import and ATP-dependent phosphorylation, as only one ATP equivalent has to be expended for both import and activation (Postma et al., 1993). The phosphate group is transferred from the phosphoryl donor PEP to the sugar substrate by a cascade that involves the cytoplasmic proteins Enzyme I and HPr. The Enzyme II complex (EII), whose membrane-associated components are directly involved in substrate translocation (EIIC permeases) and phosphorylation (EIIB), represents the substrate-specific part of the cascade. Upon phosphorylation, the activated sugar (e.g. glucose 6-phosphate) can enter glycolysis directly (Postma et al., 1993; Barabote & Saier, 2005). The PTS is mainly specific for hexoses. Since PEP does not occur as an intermediate in the phosphoketolase pathway of heterofermentative LAB, PEP : PTS appears solely associated with homolactic fermentation. PEP : PTS also has a key regulatory function in the mechanism of carbon catabolite repression (CCR). CCR ensures that in the presence of preferred substrates such as glucose, the metabolism of less favourable carbon sources is inhibited (Deutscher et al., 2006). As described below, EIIA and HPr serve central regulatory roles in CCR of Gram-negative and Gram-positive bacteria, respectively.
Function and occurrence of phospho-β-glucosidases
β-Glucosides and disaccharides such as cellobiose and lactose can be metabolized via the PEP : PTS as well. In these cases, the phosphorylated glycoconjugates are hydrolysed by cytoplasmic phospho-β-glucosidases (EC 188.8.131.52) or phospho-β-galactosidases (EC 184.108.40.206) that usually do not possess hydrolytic activity towards non-phosphorylated substrates (Fig. 1). Most known phosphoglucosidases/galactosidases have been assigned to GH 1. However, the differences between β-glucosidases and phospho-β-glucosidases of GH 1 are subtle and not yet completely resolved. It was shown that specificity for phosphorylated or nonphosphorylated substrates can depend on the exchange of only a few amino acid residues in the substrate binding pocket (Marana, 2006; Hill & Reilly, 2008).
Together with the phosphoglycosidases, EII components form the selective part of the PTS. PTS-related glycosidase genes are frequently organized in operons with genes encoding EII (ABC) components specific for the corresponding substrate. The published genome of L. plantarum WCFS1 contains 11 genes putatively encoding GH 1 enzymes with phospho-β-glucosidase functionality. Nine of these genes were found adjacent to genes encoding β-glucoside/cellobiose-specific EII components (Table 1). A recent crystallographic study (Michalska et al., 2013) provided detailed insights into the substrate binding site of a phosphoglucosidase of L. plantarum (Pbg1, Table 1). Sequence comparison showed that the glycone (6-phosphate-glucopyranosyl) binding site is highly conserved in all L. plantarum phosphoglucosidases. Considerable sequence variations were reported for the aglycone binding sites and entry regions to the active site. It therefore appears that these operons (Table 1) may not simply represent redundant systems, but are possibly specific for distinct β-glucosides.
Table 1. Putative phospho-β-glucosidase genes of Lactobacillus plantarum WCFS1 (GenBank accession no. NC_004567) and their predicted organization in operons; the information was obtained from the Prokaryotic Operon DataBase (Taboada et al., 2012)
Previous studies have implied an evolutionary relationship between PTS-mediated β-glucoside (i.e. cellobiose) and lactose metabolism. It was shown that significant sequence similarities occur between cellobiose- and lactose-specific PTS components, even between Gram-positive and Gram-negative species (De Vos et al., 1990). Cellobiose-specific PTS permeases (EIIC) were found to be capable of transporting lactose and vice versa (Aleksandrzak-Piekarczyk et al., 2011). Further, phosphoglucosidases of L. lactis and L. gasseri were induced by lactose (Simons et al., 1993; Nagaoka et al., 2008). PTS-related phospho-β-glucosidases and phospho-β-galactosidases are closely related in terms of sequence and structure and have been assigned to the GH families 1 and 4. Only few members of GH 4 were detected in Lactobacillales genomes (Michalska et al., 2013). By contrast, putative GH 1 phosphoglycosidase genes are abundantly present in most Lactobacillales genomes and are often encountered in high genomic redundancy. As indicated by the relationship of the PEP : PTS to homolactic fermentation, GH 1 glycosidases generally do not occur in obligately heterofermentative Lactobacillus species (Michlmayr et al., 2013). An exception appears to be L. brevis, harbouring a putative phospho-β-glucosidase/galactosidase gene in its genome. The L. brevis genome (strain ATCC 367) further contains three cellobiose-specific EIIC permease genes, two of which are putatively organized in operons with hydrolases of unknown function (Taboada et al., 2012). So far, L. brevis is the only obligately heterofermentative LAB species known to possess PTS components; evidence for a fructose-specific PTS in L. brevis has been reported (Saier et al., 1996).
In the case of phosphoglycosidases, the main limitation is the requirement of phosphorylated substrates to make such enzymes accessible to detailed kinetic assays and to determine the range of possible substrates. Consequently, despite the metabolic importance of the PTS, information on the biochemistry of its associated glycosidases is fairly limited, especially in comparison with other hydrolase classes. Table 2 provides an overview on literature reporting biochemical or transcriptional studies on Lactobacillales phospho-β-glucosidases. Several of these studies implied that phosphoglucosidase expression is glucose repressible and therefore most likely CCR controlled.
Table 2. PEP : PTS-related GH 1 phospho-β-glucosidases of Lactobacillales species that have been studied on the transcriptional or biochemical level
Corresponding locus of Lactobacillus plantarum WCFS1 (see Table 1).
Vmax (μmol min−1 mg−1) determined with p-nitrophenyl-β-d-glucopyranoside-6-phosphate.
Cryptic β-glucosidase induced by lactose in β-galactosidase-deficient strains
In Gram-negative enteric bacteria such as Escherichia coli, glucose-specific EIIA (EIIAGlc) plays a central regulatory role in CCR. The cyclic AMP (cAMP) receptor protein Crp is a global regulator that controls expression of a vast number of genes and operons. Crp is activated by cAMP, which is synthesized by adenylate cyclase. Signalling the availability of glucose, dephosphorylated EIIAGlc decreases adenylate cyclase activity and thus affects Crp/cAMP-mediated gene regulation. EIIAGlc is also involved in inducer exclusion, the major CCR mechanism in Enterobacteriaceae. In its dephosphorylated state, EIIAGlc inhibits the activity of non-PTS sugar import systems. Consequently, the transcription of corresponding metabolic genes is prevented.
Most low-G + C Gram-positive bacteria lack adenylate cyclase and in Firmicutes the key regulatory role in CCR is exerted by HPr. The importance of HPr is implicit in its two possible phosphorylation sites. PEP-dependent phosphorylation of the catalytic histidine residue (usually His15) forms P–His-HPr. Hpr kinase/phosphorylase catalyses the ATP-dependent, reversible phosphorylation at a distinct serin residue (Ser46), forming P–Ser-HPr. P–Ser-HPr serves as cofactor for the catabolite control protein CcpA, a global regulator that only exists in Gram-positive bacteria. Active CcpA exerts its regulatory function by binding to a DNA site called the catabolite responsive element (CRE). Owing to its four possible phosphorylation states, HPr is the central processing unit of a sophisticated regulatory system that responds to the cellular metabolic state. Although HPr is also involved in inducer exclusion, this mechanism seems to play a less important role in Firmicutes (Deutscher et al., 2006).
Transcriptional regulation of phosphoglucosidase synthesis
The possibly best understood β-glucoside-specific PTS operon is the bgl locus of E. coli, which is regulated by transcription antitermination (Postma et al., 1993). In the absence of an inducer, transcription is initiated at the constitutive bgl promoter, but interrupted by two terminator sequences downstream of the promoter. bglG encodes the transcriptional antiterminator BglG, which is inactive in its phosphorylated state. During uptake of β-glucosides, phospho groups are transferred to the substrate instead of BglG. BglG is activated through dephosphorylation and as such prevents transcription termination. BglG can also be phosphorylated at the expense of EIIAGlc and regulation of the bgl operon is cross-coupled to Crp-dependent control of the bgl promoter. Equivalent models have been proposed for β-glucoside/cellobiose utilization by L. lactis (Bardowski et al., 1994) and L. plantarum (Marasco et al., 2000). Structurally and functionally similar BglB homologues involved in regulation of β-glucoside utilization through antitermination were identified in these species. A similar mechanism was reported for the lac operon of L. casei (Alpert & Siebers, 1997). The presence of CRE sites further implies that CcpA is involved in the regulation of at least two L. plantarum phosphoglucosidases (Marasco et al., 1998, 2000; Tables 1 and 2). However, the exact mechanisms and the role of HPr need further elaboration in these cases.
A different mode of regulation was proposed for the cellobiose operon of the pathogen S. mutans (Zeng & Burne, 2009). Interestingly, CcpA seems not to play a role in the regulation of this operon. In the absence of cellobiose, EIIAB components phosphorylate the transcriptional regulator CelR, which then represses transcription of the operon. Furthermore, mannose-specific EIIAB (encoded by manL) is responsible for repressing the operon in the presence of glucose and mannose. CCR by glucose was abolished in manL-deficient mutants. Interestingly, S. mutans possesses as second β-glucoside utilization operon (bgl), which is not repressed by glucose (Cote et al., 2000).
Table 3 lists the characteristics of non-phospho-β-glucosidases that have been isolated from LAB species. Most of these enzymes were isolated and characterized with regard to possible applications such as the detoxification of cassava (Gueguen et al., 1997) or the release of wine aroma compounds (Sestelo et al., 2004; Michlmayr et al., 2010a,b).
Table 3. Characteristics of non-phospho-β-glucosidases that have been isolated from Lactobacillales species
Two enzymes were assigned to GH 3, a family that also includes β-d-xylosidases and α-l-arabinosidases. It was shown that the β-glucosidases from L. brevis and O. oeni are broad specific glycosidases with additional xylosidase and arabinosidase activities (Michlmayr et al., 2010a,b). Compared with GH 1, GH 3 glycosidase genes occur rather rarely in LAB genomes with usually only one or two copies per genome. GH 3 genes seem to be generally widespread among intestinal bacteria and are also present in Bifidobacteriaceae, which can encode up to six GH 3 glycosidase genes per genome (Van Den Broek & Voragen, 2008). In view of this, it might be tempting to speculate that especially in heterofermentative organisms, GH 3 glucosidases serve an alternative role to GH 1 enzymes. However, such an assumption can presently not be substantiated for LAB as no increased occurrence of GH 3 genes is evident in heterofermentative species that lack GH 1 phosphoglucosidases (Michlmayr et al., 2013; supplementary material). Furthermore, two GH 1 glucosidases with non-phospho-β-glucosidase activities have been isolated (Lee et al., 2012; Fang et al., 2013). However, there is insufficient information regarding to what extent GH 1 enzymes of LAB may function in a PTS-independent fashion. In both cases, the reported specific activities are low and activities towards phosphorylated substrates are not known. Both enzymes display sequence similarities to putative phosphoglucosidases and the coding gene of the Lactococcus sp. FSJ4 β-glucosidase is located adjacent to PTS transporter genes. Therefore, these enzymes may also represent PTS phosphoglucosidases with low side activities towards nonphosphorylated subtrates.
In general, the physiological role of non-phospho-β-glucosidases in LAB is unclear. Several of the reported enzymes can be classified as aryl-glucosidases (Table 3). Only the enzyme isolated from L. casei was shown to be cellobiose inducible and preferred cellobiose as substrate. However, the reported activities were considerably low (Coulon et al., 1998). It would be highly beneficial to further investigate the function of such enzymes and especially their regulation in connection to the PTS and CCR.
Recent research has demonstrated that LAB exhibit considerable versatility with regard to their β-glucosidase activities. Despite tremendous scientific efforts, the underlying mechanisms are still not fully understood. The metabolic function of bacterial β-glucosidases is primarily considered to be associated with glucose acquisition through the catabolism of cellobiose and other β-glucosides. However, the substrate range of most known β-glucosidases is broad, and especially in the case of non-phospho aryl-glucosidases, the physiological role is not yet clear. Adding to the fact that putative β-glucosidase genes appear redundant in LAB genomes, the high abundance and diversities of plant metabolites further complicate a thorough investigation of this matter. According to present information, the cellobiose/β-glucoside-specific PEP : PTS is probably the major factor determining LAB β-glucosidase activities. Through the PTS and the mechanism of CCR, glucosidase activities are tightly linked to the carbohydrate metabolism of LAB. Unfortunately, the biochemical characteristics and transcriptional regulation of PTS-related and other β-glucosidases are still not sufficiently understood. There is considerable scientific need to gain more detailed insights on such enzymes. Conducting further research in these terms is essential as LAB contribute to our daily diet, and as such play a potential role in human health and disease.
This work was supported by Bühler AG, Switzerland, GoodMills Group GmbH, Austria, and the Christian Doppler Forschungsgesellschaft, Austria. The authors declare that they have no conflict of interest.