Introduction

As aromatic compounds are structurally diverse and abundant constituents of naturally occurring biomacromolecules (de Leeuw et al. 2006) and thermogenic kerogen (Vandenbroucke and Largeau 2007), their biomineralization (to CO2) significantly contributes to the global carbon cycle. Due to their high energy content, aromatic compounds are in principle attractive substrates, but their degradation is challenged by the chemical stability of the resonance-stabilized aromatic ring (Wilkes and Schwarzbauer 2010). In the presence of molecular oxygen (O2), highly reactive oxygen species are employed by oxygenases for activation and cleavage of the aromatic ring (Pérez-Pantoja et al. 2010; Widdel and Musat 2010). Despite this pivotal role of molecular oxygen, already in the 1970s evidence for respective anaerobic (in the absence of O2) microbial capacities emerged (e.g. Evans 1977).

Since then, diverse bacteria capable of anaerobic aromatic compound degradation coupled to various modes of energy generation (denitrification, Fe(III) reduction, sulfate reduction, methanogenesis and phototrophy) have been newly isolated (for overview, see Widdel and Rabus 2001; Widdel et al. 2010; Kaser and Coates 2010). During the last two to three decades, some of these isolates became valuable model organisms for elucidating biochemical key reactions of O2-independent aromatic compound degradation. Particularly fruitful proved the denitrifying betaproteobacterium Thauera aromatica K172 (Anders et al. 1995). Seminal discoveries with this bacterium include phenol phosphorylation/carboxylation (Schmeling et al. 2004; Schühle and Fuchs 2004), addition of toluene to fumarate (Biegert et al. 1996) and reductive dearomatization of benzoyl-CoA (e.g. Boll and Fuchs 1995). Overall, a large variety of intriguing reactions has been elucidated over the last two decades that are employed for functionalization and conversion to central intermediates (mainly benzoyl-CoA, also resorcinol and phloroglucinol) as well as their reductive dearomatization and further degradation via central pathways mainly to acetyl-CoA (for overview, see Gibson and Harwood 2002; Carmona et al. 2009; Boll and Heider 2010; Widdel and Musat 2010; Fuchs et al. 2011; Philipp and Schink 2012; Boll et al. 2013). Further interest in these novel catabolic capacities arises from bioremediation of sites contaminated with toxic aromatic compounds (Dejonghe et al. 2001) as well as from biotechnological prospects in (stereospecific) functionalization of aromatic molecules (López-Cortés et al. 2010).

With the rapid propagation of genomics during the past 10 years (Shendure and Lieberman Aiden 2012), an unprecedented dimension of sequence data became accessible also to the field of anaerobic biodegradation. To crystallize functional and sustainable insights from the ever more affluent genome-derived “blueprint” information, differential-quantitative proteomics proved to be highly useful (Hufnagel and Rabus 2006; Wöhlbrand et al. 2013b). Combining proteogenomics with classical physiology enables a causal-holistic approach to investigate regulatory networks and adaptation strategies. Understanding of those will serve as gateway to ultimately decipher the “modus operandi” of aromatic compound-degrading anaerobes in the natural habitat.

The study organism of this review, “Aromatoleum aromaticum” EbN1, was the first isolate demonstrated to anaerobically degrade ethylbenzene. Moreover, strain EbN1 utilizes a wide range of aromatic compounds under nitrate-reducing conditions (Rabus and Widdel 1995) and grows anaerobically with crude oil (Rabus and Widdel 1996). “A. aromaticum” EbN1 belongs to the betaproteobacterial “Aromatoleum”/Azoarcus/Thauera cluster within Rhodobacterales. Harbouring 22 novel species this cluster represents a large share of currently known anaerobic aromatic compound-degrading bacteria (Widdel et al. 2010). Initial substrate utilization studies with A. aromaticum EbN1 (Rabus and Widdel 1995; Champion et al. 1999) already indicated anaerobic ethylbenzene degradation to proceed via 1-phenylethanol and acetophenone, which was later on demonstrated with the purified enzymes (Kniemeyer and Heider 2001a, b; Jobst et al. 2010). To date, crystal structures of the first two enzymes of the ethylbenzene pathway have been obtained, i.e. ethylbenzene dehydrogenase (Kloer et al. 2006) and (S)-1-phenylethanol dehydrogenase (Höffken et al. 2006). Enzymatic studies with substrate analogues revealed application potentials of ethylbenzene dehydrogenase (Knack et al. 2012, 2013) and (S)-1-phenylethanol dehydrogenase (Breuer et al. 2008) for enantioselective formation of secondary alcohols from a diverse range of aromatic compounds. The recent development of cultivation conditions for anaerobic growth of strain EbN1 on agar plates and of a procedure to introduce foreign DNA via conjugation allowed establishing a system for directed, unmarked deletion mutagenesis by homologous recombination and subsequent complementation in this bacterium (Wöhlbrand and Rabus 2009). First genomic insights were achieved by assembling complete gene clusters for anaerobic degradation of ethylbenzene (Rabus et al. 2002) and toluene (Kube et al. 2004) from a genomic shotgun database. Subsequently, the complete genome sequence of A. aromaticum EbN1 was determined as the first of an anaerobic hydrocarbon degrader and representative of the Aromatoleum/Azoarcus/Thauera cluster (Rabus et al. 2005). Only recently, the first complete genome sequence of a Thauera genus member, Thauera aminoaromatica MZ1T, was reported (Jiang et al. 2012). Recent habitat-oriented, methodological studies with A. aromaticum EbN1 included Protein-SIP of anaerobic mixed cultures (Jehmlich et al. 2008), stable isotope fractionation of nitrate during denitrification (Wunderlich et al. 2012) and competition with aerobic Pseudomonas putida F1 in 2D bench-scale flow-through tanks (Bauer et al. 2009; Ballarini et al. 2013). Environmental relevance of Aromatoleum/Azoarcus/Thauera cluster members is supported by their detection in diverse soils and sludges (Sun et al. 2013), oil sand tailing ponds (Penner and Foght 2010; Golby et al. 2012), petroleum refinery wastewater treatment systems (Silva et al. 2012), coke-oven wastewater (Sueoka et al. 2009) and wastewater treatment plants (Thomsen et al. 2007).

The synopsis of its catabolic breadth, robust growth (including large scale), genetic accessibility, phylogenetic representativeness, biotechnological prospects and ecophysiological relevance qualifies A. aromaticum EbN1 as a promising model organism for facultative anaerobes degrading aromatic compounds. This would complement genome-based studies with aerobic degraders such as P. putida KT2440 (Nelson et al. 2002) and Burkholderia xenovorans LB400 (Chain et al. 2006), phototrophic anaerobic degraders such as Rhodopseudomonas palustris CGA009 (Larimer et al. 2004) as well as obligate anaerobic degraders such as Geobacter metallireducens GS-15 (Butler et al. 2007) and Desulfobacula toluolica Tol2 (Wöhlbrand et al. 2013a). Since publication of the A. aromaticum EbN1 genome in 2005, we have conducted comprehensive physiological–proteogenomic studies to reconstruct the catabolic network of this bacterium and to resolve its substrate-specific regulatory potential and its response to in situ relevant substrate conditions. This overview summarizes the post-genomic insights in A. aromaticum EbN1 that could substantiate its role as systems biology model and ultimately prepare the ground for habitat-oriented, systems biology-based understanding of anaerobic aromatic compound degradation.

Differential-quantitative proteomic dataset

A. aromaticum EbN1 was studied under more than 50 different growth conditions (i.e. adapted to different types, concentrations and mixtures of substrate(s)) to determine context-specific proteome signatures. For this purpose, two-dimensional difference gel electrophoresis (2D DIGE) was mainly applied. Distinctive features of 2D DIGE are pre-electrophoretic covalent labelling of proteins with different fluorescent dyes, multiplexing and application of an internal standard. Taken together, these features allow for relative quantitation of protein abundance changes with high statistical confidence across large datasets (Unlü et al. 1997; Gade et al. 2003; Rabus and Trautwein 2010; Zech et al. 2011). The context-specific protein abundance profiles were instrumental for metabolic and regulatory reconstruction of anaerobic aromatic compound degradation in strain EbN1. The range of analysed substrate conditions and extent of proteomic coverage is unique in the field of anaerobic biodegradation and environmental microbiology in general.

The genome of strain EbN1 (single circular chromosome of 4.3 Mbp, two plasmids of 224 and 207 kbp) encodes 4,603 proteins, of which 1,391 (30.2 % coverage) were identified in the past few years. In most cases, protein identification was based on MALDI-TOF-MS/MS analysis of 2DE-separated protein spots; in selected cases, also shotgun and membrane proteomics by 1DE combined with nanoLC-ESI-MS/MS were applied (Wöhlbrand et al. 2013b). Notably, the abundance changes of 604 identified proteins were quantified by 2D DIGE analysis. Catabolic proteins (degradation to benzoyl-CoA and acetyl-CoA, respectively) usually showed 3–50-fold increased abundances during growth of strain EbN1 with the respective substrate. In some cases, abundance increases of even ≫50-fold were observed (e.g. 120-fold for the C-subunit of phenylphosphate carboxylase) (Wöhlbrand et al. 2007). While pathway-specific proteins are specifically and abundantly formed (comprising 20–40 % of the protein load separated on a 2DE gel), >90 % of the detected protein spots revealed insignificant abundance changes (i.e. <│2.5│-fold). These rather constitutively formed proteins are mostly involved in general metabolism (e.g. tricarboxylic acid (TCA) cycle, biosynthesis or nucleotide metabolism) and overall displayed only marginal abundance changes of ∼│1.3│-fold (Wöhlbrand et al. 2007).

Metabolic reconstruction by proteogenomics

Catabolic network—overview

Functional assignments during initial annotation of the A. aromaticum EbN1 genome benefited greatly from the biochemical groundwork on phylogenetically related T. aromatica K172 (Heider and Fuchs 1997; Heider et al. 1998). The reconstructed network for anaerobic degradation of aromatic compounds in strain EbN1 was originally composed of 71 proteins predicted from the genome (Rabus et al. 2005). Subsequently, differential proteome profiles of substrate-adapted strain EbN1 cells confirmed 45 of the originally predicted proteins and added 20 new proteins, allowing proposal of 14 new reactions and recognition of 11 hitherto unknown aromatic growth substrates of strain EbN1 (Kühner et al. 2005; Wöhlbrand et al. 2007, 2008; Trautwein et al. 2012c). At present, the catabolic network consists of 10 peripheral degradation routes, all of which (except for meta-hydroxylated aromatic compounds) are channelled into the central anaerobic benzoyl-CoA pathway (Fig. 1). In total, 65 of the currently 85 predicted network protein constituents were identified, corresponding to 76.5 % coverage.

Fig. 1
figure 1

Catabolic network for the anaerobic degradation of aromatic compounds in A. aromaticum EbN1. Compound colouring/labelling: black, originally determined growth substrate (Rabus and Widdel 1995); green, newly recognized growth substrate; grey, postulated intermediate; asterix, identified intermediate. Protein colouring: blue, originally predicted (Rabus et al. 2005) and identified; green, newly predicted and identified; grey, predicted only. Compound names: (1) toluene, (2) (R)-benzylsuccinate, (3) (R)-benzylsuccinyl-CoA, (4) (E)-phenylitaconyl-CoA, (5) 2-[hydroxy(phenyl)methyl]-succinyl-CoA, (6) benzoylsuccinyl-CoA, (7) ethylbenzene, (8) (S)-1-phenylethanol, (9) acetophenone, (10) benzoylacetate, (11) benzoylacetyl-CoA, (12) 4-ethylphenol, (13) 1-(4-hydroxyphenyl)-ethanol, (14) 4-hydroxyacetophenone, (15) 4-hydroxybenzoylacetate, (16) p-cresol, (17) 4-hydroxybenzaldehyde, (18) 4-hydroxybenzoate, (19) 4-hydroxybenzoyl-CoA, (20) phenol, (21) phenylphosphate, (22) hydrocinnamate or 3-(4-hydroxyphenyl)propanoate, (23) hydrocinnamoyl-CoA or 3-(4-hydroxyphenyl)propanoyl-CoA, (24) cinnamoyl-CoA or p-coumaroyl-CoA, (25) 3-hydroxy-3-phenylpropanoyl-CoA or 3-hydroxy-3-(4-hydroxyphenyl)propanoyl-CoA, (26) benzoylacetyl-CoA or 4-hydroxybenzoylacetyl-CoA, (27) cinnamate or p-coumarate, (28) benzyl alcohol, (29) benzaldehyde, (30) benzoate, (31) phenylalanine, (32) phenylpyruvate, (33) phenylacetaldehyde, (34) phenylacetate, (35) phenylacetyl-CoA, (36) phenylglyoxylate, (37) 2-aminobenzoate, (38) 2-aminobenzoyl-CoA, (39) 3-hydroxybenzoate, (40) 3-hydroxybenzoyl-CoA, (41) 2-ene-5-oxo-cyclohex-1-carbonyl-CoA, (42) 2-hydroxy-5-oxo-cyclohex-1-carbonyl-CoA, (43) 2,5-dioxo-cyclohex-1-carbonyl-CoA, (44) 4-oxopimelyl-CoA, (45) benzoyl-CoA, (46) cyclohex-1,5-diene-1-carbonyl-CoA, (47) 6-hydroxycyclohex-1-ene-1-carbonyl-CoA, (48) 6-oxocyclohex-1-ene-1-carbonyl-CoA, (49) 3-hydroxypimelyl-CoA. Enzyme names (in alphabetical order): (AdhB) benzyl alcohol dehydrogenase, (Ald) benzaldehyde dehydrogenase, (AldB) benzaldehyde dehydrogenase, (Apc1–5) acetophenone carboxylase, (Bal) benzoylacetate-CoA ligase, (BbsEF) succinyl-CoA:(R)-benzylsuccinate-CoA transferase, (BbsG) (R)-benzylsuccinyl-CoA dehydrogenase, (BbsH) phenylitaconyl-CoA hydratase, (BbsCD) 2-[hydroxy(phenyl)methyl]-succinyl-CoA dehydrogenase, (BbsAB) benzoylsuccinyl-CoA thiolase, (BclA) benzoate-CoA ligase, (BcrCBDA) benzoyl-CoA reductase, (BssABC) benzylsuccinate synthase, (ChnA) predicted cyclohexanol dehydrogenase, (Dch) cyclohex-1,5-diene-1-carbonyl-CoA hydratase, (EbA720/722) putative 3-oxo-5-ene-cyclohexylcarbonyl-CoA hydratase and 3,5-dioxo-cyclohexylcarbonyl-CoA hydrolase, (EbA309) putative alcohol dehydrogenase, (EbA723/738) putative 3-oxo-5-hydroxy-cyclohexylcarbonyl-CoA dehydrogenase, (EbA5004) iron-sulphur cluster binding protein similar to 4Fe-4S ferredoxin, (EbA5005) predicted aldehyde:ferredoxin oxidoreductase, (EbA5007) ferredoxin:NADH oxidoreductase, (EbA3118) Zn-containing alcohol dehydrogenase, (EbA5317) fatty acid-CoA ligase, (EbA5318) enoyl-CoA hydratase/isomerase, (EbA5319) β-ketothiolase, (EbA5320) 3-hydroxyacyl-CoA dehydrogenase, (EbA5321) probable acyl-CoA dehydrogenase, (EbA5380) putative p-cresol methylhydroxylase, (EbA5381) predicted 4-hydroxybenzaldehyde dehydrogenase, (EbdABC) ethylbenzene dehydrogenase, (Had) 6-hydroxycyclohex-1-ene-1-carbonyl-CoA dehydrogenase, (HbcL) 3-hydroxybenzoate-CoA ligase, (HbcL-1) 4-hydroxybenzoate-CoA ligase, (HbrCBDA) 3-hydroxybenzoyl-CoA reductase, (HcrCAB) 4-hydroxybenzoyl-CoA reductase, (Oah) 6-oxocyclohex-1-ene-1-carbonyl-CoA hydrolase, (PadBCD) phenylacetyl-CoA:acceptor oxidoreductase, (PadEFGHI) phenylglyoxylate:acceptor oxidoreductase, (PadJ) phenylacetate-CoA ligase, (Pat) phenylalanine aminotransferase, (PchA) 4-hydroxybenzaldehyde dehydrogenase, (PchCF) predicted 4-ethylphenol methylhydroxylase, (Pdc) phenylpyruvate decarboxylase, (Pdh) phenylacetaldehyde dehydrogenase or putative 4-hydroxybenzaldehyde dehydrogenase, (Ped) (S)-1-phenylethanol dehydrogenase, (PpcABCD) phenylphosphate carboxylase, (PpsABC) phenylphosphate synthetase, (TioL) predicted thiolase, (XccABC) predicted 4-hydroxyacetophenone carboxylase

Full size image

Catabolic network—new additions

New proteogenomics-inferred additions to the catabolic network are marked in green in Fig. 1 and summarized in the following:

2-Aminobenzoate

A degradation pathway for the newly recognized substrate 2-aminobenzoate was predicted (Wöhlbrand et al. 2007).

Phenylacetate

Formation of phenylacetate from phenylacetaldehyde in the phenylalanine pathway does apparently not involve the originally predicted phenylacetaldehyde dehydrogenase, but an abundantly formed aldehyde:ferredoxin oxidoreductase (EbA5005). Furthermore, also the specifically formed and adjacently encoded ferredoxin-like protein (EbA5004) and ferredoxin:NADH oxidoreductase (EbA5007) were suggested to serve as electron-accepting cofactor and in cofactor regeneration, respectively (Wöhlbrand et al. 2007). Notably, this proteogenomic-inferred hypothesis was most recently verified on the biochemical level (Debnar-Daumler et al. 2013).

p-Cresol

Initial oxidation of p-cresol could be performed by a predicted flavin adenine dinucleotide (FAD)-linked oxidase (EbA5380) instead of a predicted p-cresol methylhydroxylase (PchCF). Subsequent formation of 4-hydroxybenzoate should involve an adjacently encoded aldehyde dehydrogenase (EbA5381) rather than originally predicted 4-hydroxybenzaldehyde dehydrogenase (PchA) (Wöhlbrand et al. 2007).

3-Phenylpropanoids

A joint β-oxidation pathway for eight newly recognized substrates of plant-derived non-hydroxylated (e.g. cinnamate), mono- (e.g. p-coumarate) and dihydroxylated (e.g. caffeate) 3-phenylpropanoids involves proteins (EbA5316–5321) encoded directly downstream of the genes for anaerobic degradation of benzoate (Trautwein et al. 2012c).

4-Ethylphenol

A joint pathway for the so far unrecognized substrates 4-ethylphenol and 4-hydroxyacetophenone was discovered (Wöhlbrand et al. 2008). Interestingly, activation and oxidation of the ethyl moiety, yielding 4-hydroxybenzoyl-CoA, resemble that of the ethylbenzene pathway, but involves a different set of enzymes. Initial dehydrogenation is performed by a predicted 4-ethylphenol methylhydroxylase (PchCF), a reaction previously only described for aerobic 4-ethylphenol degradation (Reeve et al. 1989). Subsequent oxidation of 1-(4-hydroxyphenyl)ethanol involves a cyclohexanol dehydrogenase (ChnA) that is distantly related to (S)-1-phenylethanol dehydrogenase (Ped) of the ethylbenzene pathway (Büsing, I., Wöhlbrand, L., Rabus, R., unpublished). The following decarboxylation is carried out by a biotin-dependent carboxylase (XccABC) that is distinct from acetophenone carboxylase (Apc1–5) (Jobst et al. 2010), acetone carboxylase (AcxABC) (Schühle and Heider 2012) and phenylphosphate carboxylase (PpcABCD) (Schmeling and Fuchs 2009) of strain EbN1. Final activation and thiolytic cleavage may involve an acetoacetyl-CoA synthetase-like protein (AcsA) and a predicted thiolase (TioL), respectively (Wöhlbrand et al. 2008).

Catabolic network—paralogues

Analysis of the strain EbN1 genome revealed several paralogous genes (or clusters) for aromatic compound catabolism. In case of anaerobic degradation, the following three cases were concerned:

Paralogous ebd/ped genes

Ethylbenzene and (S)-1-phenylethanol degradation involves products of the first ebd/ped cluster (at 1.39 Mbp) (Kühner et al. 2005). However, the catabolic function of the second gene cluster (ebd2/ped2 at 3.44 Mbp) remains unclear since an unmarked ΔebdC2 knockout mutant did not lead to an impaired phenotype (Wöhlbrand and Rabus 2009).

Paralogous pch genes

While p-cresol degradation apparently involves a FAD-linked oxidase (Wöhlbrand et al. 2007), one paralogue of the two originally predicted p-cresol methylhydroxylases is involved in 4-ethylphenol degradation (pchCF at 0.19 Mbp) (Wöhlbrand et al. 2008).

Paralogous bcr genes

The highly similar paralogues of Bcr-type benzoyl-CoA reductases (100 % amino acid sequence identities were found among paralogous BcrBs as well as BcrDs) could be assigned to reduction of benzoyl-CoA (bcrCBDA at 3.14 Mbp) and 3-hydroxybenzoyl-CoA (reannotated to hbrCBDA at 0.43 Mbp), respectively (Wöhlbrand et al. 2007).

Transporter complement

A comprehensive bioinformatic analysis of the A. aromaticum EbN1 genome revealed that the encoded transport proteins reflect the versatile nutritional lifestyle of this degradation specialist for aromatic compounds (Tamang et al. 2009). While only few transporters for commonly used substrates (e.g. sugars) could be predicted, strain EbN1 devotes 28 % of its 274 predicted transporter proteins to the active uptake (mainly acids) and efflux of hydrophobic and aromatic compounds. Notably, aromatic hydrocarbons (e.g. toluene) and related aromatic compounds (e.g. phenol) generally do not require active transport into the cell, since their hydrophobic character allows free passage across the cytoplasmic membrane. In contrast, active removal of these compounds from cytosol and cytoplasmic membranes by solvent efflux systems is required to counteract their toxicity.

Pathway regulation by chemical sensing of aromatic compounds

Substrate-dependent formation of pathway-specific subproteomes

Evidence for substrate-specific regulatory circuits controlling the peripheral pathways of the catabolic network (Fig. 1) was obtained from differential proteome signatures. These include 53 catabolic and 13 efflux/uptake proteins from 13 degradation modules (constituting 10 peripheral pathways) profiled across 18 different substrate adaptation conditions (Fig. 2).

Fig. 2
figure 2

Abundance profiles of catabolic proteins for anaerobic aromatic compound degradation across different substrate adaptation conditions in A. aromaticum EbN1. Fold changes in protein abundances were determined by 2D DIGE using benzoate-adapted cells as reference state in all cases. The fold change data were compiled from previous publications (Kühner et al. 2005; Wöhlbrand et al. 2007, 2008; Trautwein et al. 2012c). Enzyme abbreviations are as detailed in legend to Fig. 1

Full size image

Toluene and ethylbenzene

One of the unique properties of strain EbN1 is its capacity to anaerobically degrade toluene and ethylbenzene via two biochemically distinct pathways characterized by the key intermediates benzylsuccinate and 1-phenylethanol, respectively. This was initially inferred from substrate utilization patterns (Rabus and Widdel 1995) and enzyme activities (Rabus and Heider 1998) and proteome profiles (Champion et al. 1999). The regulation of the pathways was studied in detail by combining transcript (RT-PCR and DNA microarrays) and protein profiling (2D DIGE), indicating differing degrees of regulatory specificity (Kühner et al. 2005). The two gene clusters for toluene degradation (bss and bbs) were only expressed and respective proteins formed in the presence of toluene. In contrast, transcripts and proteins of the ethylbenzene degradation pathway (ebd and apc clusters) were not only detected in ethylbenzene-adapted cells, but also in those adapted to toluene, even though at lower level. Transcript and protein profiles for both pathways agreed with the earlier genome-inferred predictions of coordinated (toluene; Kube et al. 2004) versus sequential regulation (ethylbenzene; Rabus et al. 2002). Correspondingly, ethylbenzene dehydrogenase subunits were detected in ethylbenzene-adapted cells, but not in cells adapted to the pathway intermediate acetophenone, whereas subunits of acetophenone carboxylase were formed under both substrate adaptation conditions.

4-Ethylphenol

The anaerobic degradation pathway for 4-ethylphenol and 4-hydroxyacetophenone displayed coordinated abundance profiles of all pathway proteins, in contrast to the biochemically analogous ethylbenzene/acetophenone pathway (Fig. 1), i.e. protein abundance profiles were highly similar in cells adapted to growth with 4-ethylphenol or 4-hydroxyacetophenone (Wöhlbrand et al. 2008). No indications for gratuitous induction by the chemically highly similar ethylbenzene or acetophenone were observed.

Phenylalanine

Sequential regulation of the upper and lower modules of phenylalanine/phenylacetate degradation (Wöhlbrand et al. 2007) is analogous to the above-described modular ethylbenzene/acetophenone pathway.

3-Phenylpropanoids

Eight structurally similar 3-phenylpropanoids, sharing a common pathway, lead to formation of essentially identical catabolic subproteomes (Trautwein et al. 2012c).

Benzoate

Congruent abundances of the protein constituents of the anaerobic benzoyl-CoA pathway across adaptation to 19 different aromatic compounds agree with its central role in the catabolic network (Fig. 1). Correspondingly, benzoyl-CoA is the assumed common single effector controlling expression of the anaerobic benzoyl-CoA pathway, as previously suggested for R. palustris CGA009 (Egland and Harwood 1999; Peres and Harwood 2006), T. aromatica K172 (Schühle et al. 2003) and Azoarcus sp. CIB (Barragan et al. 2005; Durante-Rodriguez et al. 2013).

Moreover, all known catabolic routes for anaerobic degradation of aliphatic compounds and aerobic degradation of aromatic compounds employed by strain EbN1 also displayed a highly substrate-specific regulation. In contrast, no or only minor abundance changes were observed for protein constituents of general metabolism (e.g. gluconeogenesis, TCA cycle) across all investigated substrate adaptation conditions (Wöhlbrand et al. 2007, 2008; Trautwein et al. 2012c).

Chemical sensing by dedicated sensory/regulatory proteins

A prerequisite for the above-described substrate-specific expression of “catabolic” gene clusters in strain EbN1 is the presence of molecular sensors capable of differentiating aromatic compounds with highly similar structures. Conspicuous examples of such substrates are alkylbenzenes (toluene, ethylbenzene), phenolics (phenol, p-cresol and 4-ethylphenol) and aromatic carboxylates (benzoate, phenylacetate, 3-phenylpropanoids). Allocated sensory/regulatory proteins predicted from the genome of strain EbN1 are illustrated in Fig. 3 with respect to the predicted relevant effector-sensing, signal-transmitting and DNA-binding domains as well as their binding sites on the DNA and are described in the following section.

Fig. 3
figure 3

Predicted domain architecture and DNA-binding sites of sensory/regulatory proteins related to anaerobic degradation of aromatic compounds in A. aromaticum EbN1. a Toluene. b Ethylbenzene. c Acetophenone. d p-Cresol. e 4-Ethylphenol and 4-hydroxyacetophenone. f Phenol. g 3-Phenylpropanoids. h Benzoate. Proposed effectors and gratuitous inducers are indicated in black and grey, respectively. Sensory/regulatory proteins are drawn to scale, which does not apply for the shown gene clusters (chromosomal position indicated)

Full size image

Toluene

Coordinated expression of the two operon-like structures involved in anaerobic degradation of toluene (bss genes, benzylsuccinate synthase; bbs genes, β-oxidation of benzylsuccinate) and of a third one, encoding a potential, newly assigned solvent efflux system (ebA1936–1926) is supposed to be mediated by the presumably toluene-specific, predicted TdiSR two-component regulatory system (Fig. 3a; Kube et al. 2004; Kühner et al. 2005; Wöhlbrand et al. 2008). The latter is encoded directly upstream of the bss gene cluster, with its localization and amino acid sequence being similar to those of other toluene-degrading denitrifiers (e.g. Coschigano and Young 1997; Hermuth et al. 2002). The two N-terminal Per-Arnt-Sim (PAS) domains of the sensory histidine kinase TdiS are predicted to account for the assumed toluene-responsive TdiR activation by phosphorelay. A conserved sequence within the promoter region of the three gene clusters may serve as binding site of phosphorylated response regulator TdiR (Fig. 3a; Kube et al. 2004).

Ethylbenzene

In contrast to the toluene pathway, the one for ethylbenzene is proposed to be sequentially regulated by two distinct two-component sensory/regulatory systems encoded in between the ebd and apc gene clusters (Fig. 3b, c) (Rabus et al. 2002; Kühner et al. 2005). Expression of genes for the upper pathway (from ethylbenzene to acetophenone) is supposed to be controlled by the predicted Tcs2/Tcr2 system. Similar to toluene-sensing TdiS, the Tcs2 sensory protein contains two N-terminal PAS domains predicted to be responsible for ethylbenzene sensing and, hence, ethylbenzene-dependent control of ebd gene cluster expression (Wolfsheimer et al. 2012). The second sensory histidine kinase, Tcs1, harbours a single PAS domain possibly recognizing acetophenone. Tcs1 together with its response regulator Tcr1 is assumed to activate apc gene cluster expression.

Phenolic compounds

A common regulatory principle of anaerobic degradation of phenolic compounds (i.e. phenol, p-cresol and 4-ethylphenol/4-hydroxyacetophenone) is the assumed σ54-dependent gene expression. This assumption is based on the following two observations: (a) A predicted AAA+ domain-containing sensory/regulatory protein (so-called bacterial enhancer binding protein, bEBP) (Bush and Dixon 2012) is encoded in juxtaposition to each catabolic gene cluster. (b) Nucleotide sequences resembling the conserved DNA-binding consensus of bEBPs (Helmann and Chamberlin 1988) are present upstream of all three catabolic gene clusters as well as the gene cluster encoding a predicted solvent efflux system for 4-ethylphenol/4-hydroxyacetophenone (Fig. 3d–f). p-Cresol-dependent activation of gene expression presumably involves the EbA5371/EbA5375 two-component sensory/regulatory system. Similar to TdiS (toluene) and Tcs2 (ethylbenzene), the predicted sensor EbA5371 contains two PAS domains, but also an additional N-terminal XylR and 4-vinyl reductase (4VR) domain (Fig. 3d). The XylR and 4VR domains (but no PAS) are also present in the presumably phenol- and 4-ethylphenol/4-hydroxyacetophenone-responsive one-component regulators PdeR and EbA324, respectively (Fig. 3e, f), suggesting XylR/4VR domains to mediate phenolic compound sensing in strain EbN1. However, 4VR-domains are known to be responsive to aromatic effectors other than phenolics, e.g. to toluene and xylenes in characterized XylR of P. putida (Delgado et al. 1995). In general, effector binding to bEBPs facilitates the highly efficient transcriptional initiation characteristic for σ54-dependent promoters (Bush and Dixon 2012).

Gratuitous induction

The differences in sensory domain composition of the predicted alkylbenzene- and phenolic compound-related sensory proteins in strain EbN1 may be responsible for the observed instances of gratuitous induction. In all cases, presence of a non-cognate effector/substrate resulted in markedly lower abundances of the catabolic proteins than observed with their cognate counterparts (Figs. 2 and 3), as exemplified in the following: (a) Non-cognate p-cresol in addition to cognate toluene leads to induction of bss/bbs gene cluster expression. (b) Toluene apparently induces the ebd gene cluster for ethylbenzene degradation. (c) Phenol induces genes for p-cresol degradation and vice versa. Notably, alkylbenzenes do not induce expression of phenolic compound degradation-related gene clusters under σ54-dependent control, while phenolic compounds may induce those of alkylbenzene degradation-related genes (Figs. 2 and 3). In addition to sensory promiscuity, cross-talk between autophosphorylated sensory histidine kinases and non-cognate response regulators, or relaxed specificities of response regulator/promoter interactions, could also be responsible for the observed instances of gratuitous induction.

Transcriptional repressors

Unlike the aforementioned transcriptional initiation by activator proteins, co-localized gene clusters for anaerobic benzoate and 3-phenylpropanoid catabolism are possibly controlled by specific repressor proteins (Fig. 3g, h). Upstream of the 3-phenylpropanoid (3Pp) gene cluster, a TetR-type transcriptional regulator (EbA5314), is encoded on the opposite strand. One may speculate that upon effector binding (possibly a phenylpropanoyl-CoA) to its C-terminal TetR-like domain, EbA5314 dissociates from the promoter region of the 3Pp gene cluster, thereby rendering the latter accessible for initiation of transcription (Fig. 3g). The first gene of the large benzoate (Bz) gene cluster codes for a BzdR-type repressor (EbA5278) (Fig. 3h), which is highly similar to the functionally characterized homologue of related Azoarcus sp. strain CIB. Here, effector binding (most likely benzoyl-CoA) was demonstrated to relieve BzdR-mediated transcriptional repression of the catabolic P N promoter (Barragan et al. 2005; Durante-Rodriguez et al. 2010). BzdR is a hybrid protein composed of an N-terminal DNA-binding domain (HTH-XRE family) and a C-terminal benzoyl-CoA-binding shikimate kinase domain (Durante-Rodriguez et al. 2013). While the DNA-binding domain of the strain EbN1 protein shows very high similarity to that of BzdR of strain CIB (87 % identity; 94 % positive), strain EbN1 possesses only a single palindromic BzdR-binding site in the promoter region (three in case of strain CIB, translating into a strong repression Durante-Rodriguez et al. 2010). The putative effector-binding domain of BzdR from strain EbN1 shares also high (75 %) sequence identity with that from strain CIB.

Response to in situ relevant growth conditions

Proteogenomic reconstruction of the catabolic network and regulatory potential of strain EbN1 established a solid basis to adjust our system-wide investigations to more eco-physiological research questions, addressing, e.g. the in situ availability of substrates. In natural habitats, bacteria do not encounter substrates as single compounds at optimal concentrations, but rather as mixtures with changing concentrations (from limiting to inhibiting) along biogeochemical gradients (Cozzarelli et al. 2001; Bauer et al. 2009; Winderl et al. 2008).

Substrate mixtures (no or unusual preferences)

Initial studies on the response of strain EbN1 to substrate mixtures were concerned with toluene and ethylbenzene. Benzoate-adapted cells, which were shifted to a mixture of both alkylbenzenes, induced the expression/formation of the corresponding catabolic genes/proteins to simultaneously utilize the two provided substrates (Kühner et al. 2005). Both alkylbenzenes were also co-utilized during anaerobic growth of strain EbN1 with crude oil (Rabus and Widdel 1996). This apparent absence of a regulatory control such as carbon catabolite repression (CCR) under the applied cultivation conditions could be interpreted as an adaptation to the low bioavailability of alkylbenzenes usually encountered in the natural environment, e.g. ethylbenzene and toluene in petroleum-contaminated groundwater co-occur often at concentrations only in the micromolar range (Gülensoy and Alvarez 1999; Griebler et al. 2004). Since hydrocarbons dissolve in and freely pass cellular membranes, their instantaneous and parallel degradation also avoids intracellular accumulation and, hence, cytotoxicity.

Contrasting the apparent absence of CCR described above, C4-dicarboxylate-adapted cells of strain EbN1 preferentially utilized benzoate from a mixture with a C4-dicarboxylate (succinate, fumarate or d,l-malate), resulting in diauxic growth (Fig. 4a). Benzoate preference was unexpected, since usually more easily degradable substrates (e.g. succinate) that also allow for higher growth rates are preferred. The latter was previously observed, e.g. for aromatic compound-degrading P. putida KT2440 (Morales et al. 2004) and Azoarcus sp. strain CIP (López Barragán et al. 2004). In case of strain CIB, the response regulator AccR was demonstrated to control succinate preference from mixtures with aromatic acids (benzoate, 3-methylbenzoate or 3-hydroxybenzoate) (Valderrama et al. 2013). Targeted transcript analysis and differential-quantitative proteomics with related strain EbN1 indicated repression of C4-dicarboxylate uptake via a predicted tripartite ATP-independent periplasmic (TRAP) transporter (DctPQM), as the main determinant for the observed diauxie. One may speculate that the presence of benzoate negatively controls the transcriptional activation of dctPQM expression by the assumed C4-dicarboxylate-responsive DctSR two-component sensory/regulatory system via a presently unknown mechanism (Fig. 4b) (Trautwein et al. 2012a). Notably, diauxic growth with benzoate preference was also observed in the distantly related alphaproteobacterium Magnetospirillum sp. strain pMbN1 (Lahme, S., Trautwein, K., Rabus, R. unpublished). In contrast, several close relatives of strain EbN1 (Aromatoleum strains ToN1 and HxN1, as well as T. aromatica K172) displayed monophasic growth and strain-specific utilization preferences (Trautwein et al. 2012a). Unlike the well-known control mechanisms for hierarchical utilization of carbohydrates in enterobacteria and bacilli (Deutscher 2008; Görke and Stülke 2008), only little is known about carbon catabolite control in (anaerobic) aromatic compound-degrading bacteria (Fuchs et al. 2011). In case of aerobic, aromatic compound-degrading Pseudomonas spp., mechanisms of CCR are distinct from those of Enterobacteriaceae and Firmicutes (for review, see Rojo 2010). The abovementioned first insights into the distinct substrate preferences among anaerobic degraders of aromatic compounds, however, require to individually assess substrate preferences, their evolution and underlying regulatory principles.

Fig. 4
figure 4

Anaerobic diauxic growth of A. aromaticum EbN1 with a mixture of benzoate (preferred) and succinate is mediated by repression of succinate uptake. a Diauxic growth curve, modified from Trautwein et al. (2012a). b Scheme of transcriptional control of the C4-dicarboxylate uptake system DctPQM

Full size image

Solvent stress

Alkylbenzenes and phenolic compounds are special substrates as they are attractive due to energy richness and at the same time potentially deleterious due to membrane toxicity (Sikkema et al. 1995; Heipieper and Martinez 2010). Microbial persistence under solvent stress conditions is relevant in naturally hydrocarbon/phenolic-rich environments (carbon cycle) or at contaminated sites (bioremediation), but can also be key to biotechnological processes employing two-phase fermentation systems (Heipieper et al. 2007).

The physiological and proteomic solvent stress response of strain EbN1 was studied during growth at semi-inhibitory (about 50 % growth inhibition) concentrations of toluene (Fig. 5) or ethylbenzene (Trautwein et al. 2008). The applied concentrations were well within or even exceeded the range of benzene, toluene, ethylbenzene and xylene concentrations observed in situ (for references, see Zink and Rabus 2010). Strain EbN1 responded to alkylbenzene stress with decreased growth rates, impairment of denitrification (i.e. reduced nitrate consumption and nitrite turnover) and increased levels of intracellular poly(3-hydroxybutyrate) (PHB). Formation of PHB contributes to consumption of surplus reducing equivalents generated during alkylbenzene degradation, if alkylbenzene-derived acetyl-CoA is re-routed from the TCA cycle to PHB synthesis (Fig. 5). Such a tuning of the metabolic flux is supposed to facilitate continuous alkylbenzene degradation despite impaired denitrification. Due to their hydrophobic nature, PHB granules could, as a side effect, trap alkylbenzenes and thereby contribute to alleviate cytotoxic effects. Another level of solvent stress adaptation is the compositional change of phospholipids to prevent maceration of cellular membranes. To counteract an uncontrolled increase in membrane fluidity at semi-inhibitory alkylbenzene concentrations, strain EbN1 increased the share of more bulky head groups (phosphatidylglycerol, phosphatidylcholine, acyl-phosphatidylethanolamine) and saturated phospholipid-linked fatty acids (Fig. 5) (Zink and Rabus 2010). Similarly, other anaerobic beta- and deltaproteobacterial strains increase the degree of saturation in their membrane fatty acids when exposed to semi-inhibitory concentrations of organic solvents (Duldhardt et al. 2010). Furthermore, solvent efflux systems (Ramos et al. 2002) may also contribute to solvent stress tolerance in strain EbN1. Cells adapted to toluene (Fig. 5), 4-ethylphenol and phenol revealed the formation of apparently compound-specific putative solvent efflux systems (Fig. 2) (AcrAB/TolC homologues). Notably, in case of toluene and 4-ethylphenol/4-hydroxyacetophenone, these potentially new types of efflux systems are encoded in juxtaposition with the catabolic gene cluster of the respective solvent (Wöhlbrand et al. 2008). Remarkably, solvent-unadapted cells of strain EbN1 tolerated at least 2.5-fold higher alkylbenzene concentrations as compared to other anaerobic degraders (e.g. T. aromatica K172, Geobacter sulfurreducens) (Duldhardt et al. 2007).

Fig. 5
figure 5

Solvent stress response of A. aromaticum EbN1 involves re-routing of carbon flow from TCA cycle to PHB synthesis during impaired denitrification, possible use of a novel type of efflux system and modifications of phospholipid head groups and phospholipid-linked fatty acids to stabilize the cytoplasmic membrane. Abbreviations: PC phosphatidylcholine, PG phosphatidylglycerol, PE phosphatidylethanolamine

Full size image

Carbon limitation (submaximal growth rates)

As many natural environments are oligotrophic (i.e. organic substrates and other nutrients occur only at nanomolar to micromolar concentrations) (Egli 1995), heterotrophic microorganisms usually face limited rather than ample supply of substrates. Thus, survival and success in the habitat are often determined by an organism’s capacity to adapt to nutrient limitation, which can be best studied in the constant and controlled environment of chemostats (e.g. Harder and Dijkhuizen 1983). Here, the constant supply of a growth-limiting nutrient results in microbial populations to attain an equilibrium (steady state), which is characterized by constant physiological properties at a defined specific growth rate (μ).

Benzoate is a representative model substrate supporting heterotrophic growth, since aromatic compounds in general are abundant and widespread in nature (de Leeuw et al. 2006) and benzoyl-CoA (activated benzoate) is the central intermediate of most currently known anaerobic degradation pathways for aromatic compounds (Fig. 1). This agrees with essentially all known facultatively and strictly anaerobic aromatic compound-degrading bacteria being able to utilize benzoate (Fuchs et al. 2011).

The adaptive response of strain EbN1 to benzoate limitation was recently studied in anoxic chemostats at submaximal (low to high, 0.036 to 0.200 h–1) growth rates on the physiological and proteomic level (Fig. 6; Trautwein et al. 2012b). Growth rate positively correlated with measured cellular parameters, such as cell size and DNA/protein content as expected from seminal studies with Escherichia coli and related bacteria (Schaechter et al. 1958; Dennis and Bremer 1974). Growth rate also affected the distribution of available free energy that is derived from benzoate catabolism. First, the free energy consumed for biomass formation steadily increased with growth rate. Second, in addition to the synthesis of new biomass (growth), different non-growth duties (defined as maintenance) (Pirt 1965) consume a high share of free energy, e.g. sustainment of membrane integrity, turnover of macromolecules or motility. We used a thermodynamics (ΔG)-based approach (Tijhuis et al. 1993; Harder 1997) to calculate the growth-dependent (m(μ)) and growth-independent (m e; at μ = 0.0 h–1) maintenance energy consumption rates in strain EbN1. From the total free energy available, the share of energy consumed for maintenance was largest at low growth rates (Fig. 6a). The observed influence of growth rate on energy distribution suggests a fine-tuned control of the coupling between anabolism and catabolism by yet unknown mechanisms.

Fig. 6
figure 6

Response of A. aromaticum EbN1 to low growth rates (μ low) in benzoate-limited, anoxic chemostats (modified from Trautwein et al. 2012b). a Selected steady-state (dilution rate, D = μ) physiological characteristics of cells at μ low (0.036 h–1) and μ high (0.180 h–1). b At μ low, benzoate-limited cells of strain EbN1 formed proteins for uptake and catabolism of absent (aromatic/aliphatic) substrates (grey), with some of them (i.e. gentisate) being degraded only aerobically. Colour code of proteins: green or red, increased or decreased abundance at μ low, respectively (as compared to μ high by 2D DIGE); black, proteins detected (in most cases only) at μ low by nanoLC-ESI-MS(/MS); grey, not detected. Names of aromatic compounds and enzymes: see legend to Fig. 1. Additional compounds: (50) gentisate. Additional enzymes: (PaaK) aerobic phenylacetate-CoA ligase, (PaaFH) 3-hydroxyacyl-CoA dehydrogenase, (BclA-1) aerobic benzoate-CoA ligase (BoxAB) benzoyl-CoA oxygenase, (BoxC) enoyl-CoA hydratase/isomerase, (EbA2768) β-ketoadipyl-CoA thiolase, (Hbh) hydroxybenzoate hydroxylase, (NagI) gentisate 1,2-dioxygenase, (NagL) maleylpyruvate isomerase, (NagK) fumarylpyruvate hydrolase, (AcsA) acetate-CoA ligase, (AceEF, LpdA) pyruvate dehydrogenase complex, (PrpE) propionate-CoA ligase, (PccAB) propanoyl-CoA carboxylase, (AccBC) acetyl-CoA carboxylase, (SbmAB) methylmalonyl-CoA mutase. Transporter: (BenK) benzoate:H+ symporter, (EbA5303–09) predicted benzoate uptake ABC transporter

Full size image

Comparing the proteomes of slowly (μ low; 0.036 h–1) and fast (μ high; 0.180 h–1) growing cells of strain EbN1 revealed multi-faceted adaptations to substrate limitation. At μ low, benzoate uptake is suggested to primarily involve the high affinity ABC transporter (EbA5303-5309). Fitting of growth functions and population models to experimentally determined data (cell numbers, benzoate concentrations) supported the originally genome-inferred participation of a second transporter, i.e. the benzoate:H+ symporter BenK, in benzoate uptake. The assumed growth rate (benzoate concentration)-dependent contribution of each transporter to bulk benzoate uptake remains to be demonstrated. Reduced enzyme abundances sufficed for benzoate catabolism under these conditions of slow growth. Remarkably, uptake systems and catabolic enzymes for a variety of aerobic and anaerobic degradation routes were additionally present at μ low (Fig. 6b), despite absence of the respective aromatic substrate and oxygen, respectively. This contrasts the substrate-specific regulation of individual pathways, previously resolved on the basis of substrate-replete batch cultures growing at μ max (see “Pathway regulation by chemical sensing of aromatic compounds” section). The observed derepression and/or passive induction of multiple catabolic pathways under conditions of strong limitation of a single substrate may prepare the cells to instantly utilize a broad range of substrates upon becoming available (Trautwein et al. 2012b). Such a “multivorous” lifestyle (Harder and Dijkhuizen 1982; Ihssen and Egli 2005; Egli 2010) may result from adaptation of strain EbN1 to substrate and redox conditions encountered in its natural habitat. Apparently, the “preparedness for future opportunities” could be (at least) one reason for higher maintenance energy demands during slow growth (see above).

Perspectives

Future studies with A. aromaticum EbN1 should strive for a holistic (eco-)physiological understanding pursuing the following lines of research: (a) The regulatory network should be dissected with respect to the molecular mechanisms of chemical sensing and the overarching circuits for integrating aerobic and anaerobic regulons. (b) Adaptation to low growth rates (energy limitation/dissipation) should be elaborated with various model substrates (single or in mixtures) as well as adaptation to stress conditions. (c) Molecular genetics should be applied to delineate functions of proteins newly hypothesized to be involved in substrate uptake, catabolism and regulation. (d) Gel-free, mass spectrometry-based quantitation of protein abundances should be used in addition to 2D DIGE (Wöhlbrand et al. 2013b). (e) The biotechnological potential of strain EbN1 and its close relatives should be further explored. (f) The knowledge on strain EbN1 can be used as crystal seed for a general physiological and evolutionary understanding of the widely unexplored Aromatoleum group. (g) Defined communities of degradation specialists should be established to investigate competition, coexistence and cooperation between different species of the same genus or different genera.