Physiology of deletion mutants in the anaerobic β-myrcene degradation pathway in Castellaniella defragrans

Background Monoterpenes present a large and versatile group of unsaturated hydrocarbons of plant origin with widespread use in the fragrance as well as food industry. The anaerobic β-myrcene degradation pathway in Castellaniella defragrans strain 65Phen differs from well known aerobic, monooxygenase-containing pathways. The initial enzyme linalool dehydratase-isomerase ldi/LDI catalyzes the hydration of β-myrcene to (S)-(+)-linalool and its isomerization to geraniol. A high-affinity geraniol dehydrogenase geoA/GeDH and a geranial dehydrogenase geoB/GaDH contribute to the formation of geranic acid. A genetic system was for the first time applied for the betaproteobacterium to prove in vivo the relevance of the linalool dehydratase-isomerase and the geraniol dehydrogenase. In-frame deletion cassettes were introduced by conjugation and two homologous recombination events. Results Polar effects were absent in the in-frame deletion mutants C. defragrans Δldi and C. defragrans ΔgeoA. The physiological characterization of the strains demonstrated a requirement of the linalool dehydratase-isomerase for growth on acyclic monoterpenes, but not on cyclic monoterpenes. The deletion of geoA resulted in a phenotype with hampered growth rate on monoterpenes as sole carbon and energy source as well as reduced biomass yields. Enzyme assays revealed the presence of a second geraniol dehydrogenase. The deletion mutants were in trans complemented with the broad-host range expression vector pBBR1MCS-4ldi and pBBR1MCS-2geoA, restoring in both cases the wild type phenotype. Conclusions In-frame deletion mutants of genes in the anaerobic β-myrcene degradation revealed novel insights in the in vivo function. The deletion of a high-affinity geraniol dehydrogenase hampered, but did not preclude growth on monoterpenes. A second geraniol dehydrogenase activity was present that contributes to the β-myrcene degradation pathway. Growth on cyclic monoterpenes independent of the initial enzyme LDI suggests the presence of a second enzyme system activating unsaturated hydrocarbons.


Background
Monoterpenes represent a prominent group of volatile organic compounds (VOC), with an estimated mean global emission of 117 Tg C yr -1 into the atmosphere [1] and a fast photochemical turnover [2]. Especially coniferous plants are considered to be main producers of monoterpenes, e.g. for thermotolerance or for communication between plants or the interaction between plants and insects [3][4][5]. Monoterpenes also enter the soil by the rhizosphere or by rotten leafs [6], where they inhibit growth of microorganisms as well as of seedlings [7][8][9], but also stimulate the bacterial activity detectable in higher biomass and CO 2 -production [5,10,11]. By definition, monoterpenes possess a carbon skeleton based on two C 5 units originating from isopentenyl pyrophosphate (IPP), which is synthesized via the mevalonate (in eukaryotes) or the mevalonate-independent pathway (in prokaryotes and plant plastids) [12][13][14]. Mainly, plant monoterpenes are produced via the latter pathway, but the metabolic cross linkage between both has been reported in several species [15,16].
Under anaerobic conditions, the biochemistry for the activation of these natural abundant alkenes seems to follow a totally different mechanism. The first evidence for the anaerobic degradation of monoterpenes were seven nitrate-reducing enrichment cultures with monoterpenes as sole carbon source [39]. Isolation led to the description of four Alcaligenes defragrans strains, including strain 65Phen isolated with α-phellandrene [40]. A taxonomic study transferred these strains in the novel genus Castellaniella within the Alcaligenaceae, as C. defragrans [41]. The betaproteobacterium is capable of degrading a broad substrate range of a-, mono-, and bicyclic monoterpenes ( Figure 1) [40]. Initial metabolite studies on the anaerobic monoterpene degradation pathway in C. defragrans elucidated the demand for a sp 2hybridized C1-atom as structural prerequisite for monoterpenes utilization [42] as well as the formation of geranic acid as intermediate [43], which is likely degraded on a modified β-oxidation pathway [44,45]. These findings proposed the degradation of β-myrcene via hydration to linalool, followed by isomerisation to geraniol, and then two oxidations to geranial and to geranic acid [43]. The genes and proteins involved this pathway were recently identified [46,47] (Figure 2). The bifunctional linalool dehydratase-isomerase ldi/LDI catalyzes the first two steps, the highly enantiospecific hydration of β-myrcene to (S)-(+)-linalool and its isomerisation to geraniol [46,48]. Subsequently, two dehydrogenases oxidize the allylalcohol geraniol and geranial. The geraniol dehydrogenase geoA/GeDH (E. C. 1.1.1.183) is a member of the medium-chain dehydrogenase/reductase superfamily [49] with high affinity for its substrate geraniol [47]. In vitro studies confirmed the activity of a geranial dehydrogenase Figure 1 Monoterpene substrate range of C. defragrans [40].
geoB/GaDH. Both dehydrogenases were expressed in cells growing with monoterpenes [47]. So far, the evidence for the anaerobic β-myrcene degradation pathway was rather biochemically based on metabolite and enzyme studies. To prove the physiological role in vivo, we created deletion mutants of C. defragrans missing the gene ldi and geoA, respectively. The previous findings, i.e. the geranic acid formation and the induced dehydrogenase activities, were observed in both acyclic and monocyclic monoterpenes grown cells and suggested the existence of a common degradation pathway. To clarify whether there is one defined metabolic route or multiple pathways present for the anaerobic degradation of monoterpenes in C. defragrans, we deleted the initial, β-myrceneactivating enzyme, the LDI. The deletion of the GeDH was of interest due to the frequent presence of multiple alcohol dehydrogenases in genomes, often with a broad substrate range.

Results and Discussion
Construction of the in-frame deletion mutant C. defragrans Δldi and ΔgeoA Growth of C. defragrans as single colony under denitrifying conditions was achieved on acetate in a defined, solidified medium. A spontaneous mutant strain resistant to rifampicin (150 μg/ml) was obtained showing the phenotype of the wildtype with respect to growth on monoterpenes (Additional file 1: Table S1). Conjugation was established with the broad host range plasmid pBBR1MCS-2, proceeding with a frequency of 1.8 × 10 -4 transconjugants cell/ donor cells in 8 h (Additional file 1: Table S2). The plasmid was maintained in C. defragrans. For genomic deletion mutants, we constructed pK19mobsacBΔldi and pK19mobsacBΔgeoA that carried the start and stop codon of the ldi (ORF26) or geoA (ORF31) separated by a specific restriction site and the upstream and downstream located regions (Additional file 1: Figure S1). The sequence information was obtained from a 50 kb contig (Acc. no. FR669447.2) with the following annotation for ORFs adjacent to ldi or geoA: ORF27 as a thioesterase, ORF29 as a putative subunit of cytochrome c oxidase, ORF30 as a secretory protein and ORF32 as a long-chain-fatty-acid CoA ligase, while for ORF25 only hypothetical proteins were found in database queries (Additional file 1: Figure S2). Conjugation and homologous recombination yielded genomic in-frame deletions, with a second recombination frequency of 0.5% and 1.25% for the deletion of ldi and of geoA, respectively. Analysis by PCR revealed in the deletion mutants the expected, shortened amplicons with primer pairs spanning the deleted gene in comparison with the wild type (Additional file 1: Figure S3). Polar effects due to the deletion of ldi or geoA were not detected in mRNA analyses (Additional file 1: Figure  S4). The genes ldi or geoA and their native ribosomal binding site were cloned in the MCS of pBBR1MCS plasmids. Conjugation into C. defragrans deletion mutants yielded ampicillin-resistant transconjugants named C. defragrans Δldicomp and kanamycin-resistant transconjugants named C. defragrans ΔgeoAcomp.
In previous studies, β-myrcene as well as αphellandrene supported the formation of geranic acid in cell suspension experiments. The geranic acid pool was 10fold larger in β-myrcene experiments than with the cyclic monoterpenes α-pinene, α-phellandrene, and limonene [43]. We assayed the geranic acid pools in C. defragrans mutant strains under nitrate-limited conditions in liquid cultures on 6 mM monoterpene in HMN (Table 1). This metabolite was only detectable in myrcene-grown C. defragrans cultures with the ldi either present in the genome or in trans, in concentrations of 8.85 μM and 6.61 μM, respectively. In α-phellandrene grown cultures, geranic acid was detectable in media of these C. defragrans strains in concentrations of 0.24 μM and 0.33 μM. Geranic acid formation was not detectable in cultures of the mutant lacking the gene ldi. The RP-HPLC detection limit was 6.4 nM, thus geranic acid formation in C. defragrans Δldi was below a thousandth of that in the wild type. Growth on α-phellandrene clearly does not involve the formation of geranic acid suggesting the presence of another monoterpene degrading pathway that circumvents the activation of the substrate by LDI as well as geranic acid formation.
Under aerobic conditions microbial biotransformation of (−)-limonene and β-myrcene revealed the formation of enantiopure (−)-perillyl alcohol, perillyl acid and myrcenic acid [30,[50][51][52]. Anaerobic hydroxylations catalyzed by molybdenum enzymes have been recently reported, e.g. the hydroxylation of ethylbenzene to (S)-phenylethanol in Aromatoleum aromaticum [53] and of cholesterol to cholest-1,4-diene-3-one in Sterolibacterium denitrificans [54]. Whether the degradation of cyclic monoterpenes proceeds via a homologue pathway is subjected in ongoing research. To our knowledge, this is the first report on the existence of different activation mechanisms for cyclic and acyclic monoterpenes in one bacterial strain.

Physiological and enzymatic characterization of C. defragrans ΔgeoA
The deletion of geoA resulted in an increased generation time and reduced biomass yields, e.g. on α-phellandrene, limonene and β-myrcene ( Figure 3A-C, Table 2). Nitrate was completely consumed, but the generation time was always prolonged, e.g. 3.5-fold for α-phellandrene. The biomass formed as determined by protein analyses was decreased by 32% to 48% in the deletion mutant ( Table 2). Most likely, geraniol was oxidized at slower rate in the deletion mutant. This seems to have an inhibitory effect on the growth due to the known geraniol in vivo toxicity of above 5 μM in the aqueous phase [47]. The intracellular geraniol concentrations were below the detection threshold of gas chromatographical analysis, but we observed physiological evidence for increased geraniol pools. In the cultivation system with HMN, 4 mM geraniol stopped monoterpene utilization completely [47]. In the wild type, addition of 16 mM acetate supported growth in the presence of 4 mM geraniol and 20 mM nitrate to an OD 660 of 0.15 (± 0.002; n = 2). The deletion mutant C. defragans ΔgeoA also grew after acetate addition, but reached only an OD 660 of 0.061 (± 0.01; n = 2), although both strains consumed the same nitrate amount. In conclusion, C. defragans ΔgeoA reacts more sensitive towards geraniol than the wild type.
The growth phenotype of the wild type was recovered in the mutant strain by complementation with the geoA gene located on a broad-host range plasmid. The in trans complemented mutant C. defragrans ΔgeoAcomp revealed physiological characteristics similar to C. defragrans 65Phen: growth rate and yield, monoterpene consumption and nitrate reduction were almost identical suggesting that the wild type phenotype was restored by GeDH constitutively expressed from the plasmid pBBR1MCS-2geoA (Table 2, Figure 3).
The absence of GeDH was expected to reduce the rate of geranic acid formation. In this study, geranic acid was detected in cultures grown on 6 mM monoterpene in  the presence of HMN and 10 mM nitrate (Table 1). Cultures were sampled after nitrate depletion. Geranic acid concentrations of acidified and lysed cultures were 9 ± 1 μM in the medium of the wild type and 12 ± 1 μM in the medium of the complemented mutant, but only 5 ± 2 μM in the medium of C. defragrans ΔgeoA, thus revealing a limited capacity to form geranic acid in the absence of GeDH.
The ΔgeoA phenotype has still the capacity to degrade monoterpenes, an indication for the presence of another alcohol dehydrogenase that catalyzes the geraniol oxidation. Thus, we tested the GeDH activity spectrophotometrically in cell-free, cytosolic extracts of C. defragrans strains 65Phen, ΔgeoA and ΔgeoAcomp. Under standard conditions, with 0.8 mM geraniol as substrate and identical protein concentrations in the assay, the geraniol oxidation rates were 5.8 nkat mg -1 protein for C. defragrans 65Phen and 1.05 nkat mg -1 protein for C. defragrans ΔgeoA. Complementation restored the activity to 9.4 nkat mg -1 protein in C. defragrans ΔgeoAcomp. The in vivo concentration of geraniol inside the cell is expected to be in the micromolar range [47]. The GeDH activity in the extracts of C. defragrans ΔgeoA catalyzed the reaction with a high affinity, the apparent concentration for half-maximal rate was below 10 μM geraniol (Figure 4). This indicated an activity of the second alcohol dehydrogenase at physiological conditions.
In summary, the presented data argue for a reduced geraniol flux to geranic acid in the metabolism of the deletion mutant. We suggest that a geraniol accumulation or increased pools of metabolites derived from geraniol on other pathways cause a reduced growth rate as indicated by prolonged generation time, decreased biomass production, and reduced geranic acid formation. The accumulation of a toxic intermediate in monoterpene catabolism causing reduced growth rate has also been seen for deletion mutants of P. putida M1 in ß-myrcene degradation [24,55]. Accumulation of geraniol is known to be toxic for cells: due to its hydrophobic properties it can integrate into bacterial membranes causing disintegrations followed by failure of the proton motive force [56,57].
The presence of several ADHs in a genome is not unusual. In microorganisms, alcohol dehydrogenases possess a wide variety of substrate specificities and are involved in different physiological functions [58]. For various ADHs deficient mutants, retarded growth on the prevailing substrate and reduced ADH activity was observed [59][60][61]. Also in plants the existence of additional ADHs capable of oxidizing geraniol was suggested [62].

Conclusions
We developed a genetic system for Castellaniella defragrans and constructed in-frame deletion mutants that allows for insights into the physiology of the anaerobic degradation of monoterpenes.
C. defragrans ΔgeoA lacking the gene for a geraniol dehydrogenase was physiologically analysed. The geoA deficient strain exhibited reduced growth on monoterpenes and slower geraniol oxidation rates in soluble extracts, in comparison to the wild type. The original phenotype was restored in trans with an episomal geoA in the C. defragrans ΔgeoAcomp. One explanation for the reduced growth is a higher steady-state level of  geraniol in the cell causing toxic effects. These observations together with reduced geranic acid formation demonstrate clearly a participation of GeDH in the anaerobic degradation of β-myrcene. However, the geoA deletion is not mortal. A second GeDH activity is present in soluble extracts. This suggests a need for both GeDHs to balance the geraniol formation by oxidation during fast growth of the wild type. The physiological characterization regarding growth with acyclic and cyclic monoterpenes exhibited an unexpected effect of the ldi deletion that caused a phenotype dependent on the substrate structure in C. defragrans Δldi: the cyclic monoterpenes α-phellandrene and limonene were metabolized, but not the acyclic β-myrcene. Thus, the degradation of the acyclic β-myrcene required the activity of a linalool dehydratase-isomerase that was not necessary for the degradation of cyclic monoterpenes. This observation indicates for the presence of a second hydrocarbon activating system in C. defragrans.
Culturing conditions and growth media E. coli strains were cultured according to established methods [66]. For propagation of plasmids, additional antibiotics were supplemented in the indicated concentrations [66]. Maintenance and growth experiments in liquid cultures with C. defragrans 65Phen and mutants were performed as described previously [40]. Growth in liquid cultures was monitored by turbidity measurements at 660 nm.
Minimal medium for plates contained 50 mM sodium acetate in medium solidified with 18 g/L agar and additionally buffered with 50 mM HEPES, pH 7.2. Incubation took place in anaerobic jars for 4 to 5 days under N 2 atmosphere at 28°C. Biomass production of C. defragrans strains was performed according to [46].
Antibiotics were used at following concentrations (unless indicated otherwise): 50 μg/mL ampicillin, 50 μg/mL kanamycin, and 150 μg/mL rifampicin. Plating efficiency was determined by plating decading dilutionto-extinction series of cell suspensions with known optical density (OD) at 660 nm in duplicates.

Preparation and manipulation of genetic material
Genomic DNA was isolated from C. defragrans 65Phen using the DNeasy Tissue Kit (Quiagen, Hilden, Germany). Plasmid DNA was isolated from E. coli strains and C. defragrans 65Phen using mini-plasmid preparation kits (Quiagen). Gel-excised PCR products and plasmid fragments were purified with the QIAquick gel extraction kit (Quiagen). PCR amplification was usually performed using Taq polymerase (Promega, Madison, USA). For cloning purposes a mixture of Taq polymerase and a thermostable polymerase with proofreading activity (Fermentas, St. Leon Rot, Germany) were applied.

Transcription analyses with Reverse Transcriptase-PCR
Preparation of total RNA from C. defragrans strains after growth on α-pellandrene was performed with RNeasy Mini Kit (Qiagen) according to manufacturer's instructions, followed by cDNA synthesis using the Revert Aid™ First Strand cDNA Synthesis Kit (Fermentas). For transcriptional analyses, RT-PCR was performed with 35 cycles with primer pairs listed in Table 4. Negative controls included RT-PCR without reverse transcriptase. Table 4 lists primers used for the different amplification purposes.

Ligation and transformation of plasmid constructs
Subcloning of PCR products into pCR4-TOPO W vector (Invitrogen, Darmstadt, Germany) was performed corresponding to manufacturer's instructions. PCR products with inserted restriction sites and purified plasmids were digested with the appropriate restriction enzymes and separated by gel electrophoresis. Both digested plasmids and PCR products were gel excised and purified. For ligation reactions, an insert-vector ratio of 1:1, 3:1 or 10:1 was chosen. To this mixture, T4-ligase buffer (1x), ATP (25 μM) and T4-ligase (2.5 U) were added. Incubation was for 12-16 h at 12°C. Transformation of 5 or 10 μL of the ligation reaction to chemical competent E. coli strains S17-1 or Top10 was performed as described [67]. Single colonies growing on selective solid medium were picked and screened for the correct insert size by PCR applying M13 or T7 primers. Plasmids of positive tested clones were purified and served as sequencing templates.

Construction of suicide plasmids
The 5`-and 3`-flanking regions of ldi or geoA and the start and stop codons of the deleted gene separated by an appropriate specific restriction site were inserted into the suicide vector pK19mobsacB [64]. Oligonucleotide sequences are listed in Table 4.
Initially, the flanking regions were amplified from genomic C. defragrans 65Phen DNA with primers adding restriction enzyme sites to the PCR-product. The 5`-flanking region to the ldi was obtained with the primer pair ORF25_EcoRI_F and ORF25_XhoIATG_R. During amplification of the 3`-flanking region with primer pairs ORF27_XhoI_TAA_F and ORF27_HindIII_R difficulties occurred due to a terminator structure in the genome sequence that was solved with a nested PCR approach. A 2.2 kb amplicon comprising ORF 27 was obtained with the primer pair p27plus_F and p27plus_R that served as template for the initial named primer with an increased initial denaturation time (from 4 min to 10 min). Sequencing of the 763 bp amplicon revealed a base exchange at position 373 from guanine to adenine causing an amino acid replacement from proline to threonine. This shift was revoked by a site directed mutagenesis approach using primer p27_ mismatch_F and p27_mismatch_R in combination with ORF27_XhoI_TAA_F and ORF27_HindIII_R, respectively [68]. The particular amplicons were bond to each other in another reaction with the exterior primer pair. The 5`-flanking region of the geoA was obtained with the primer pair ORF2930_XbaI_F & ORF2930_XhoI_R and the geoA 3`-flanking region ORF32_XhoI_F & ORF32_HindIII_R.
Subcloning vectors were double digested with the prevailing added recognitions site for restriction enzymes. The flanking regions were excised, purified and ligated via a three-piece-ligation into the suicide vector pK19mobsacB [64]. Sequencing of the obtained plasmids pK19mobsacBΔldi and pK19mobsacBΔgeoA was performed to ensure correct sequence of the flanking regions including the start and stop codons of the deleted genes.

Construction of complementation plasmids
For construction of the in trans vector both, the ldi and the geoA was amplified from genomic DNA of C. defragrans 65Phen with primer pair encompassing the entired ORF, i.e. for the ldi primer pair ldi_EcoRI & ldi_BglII, and for geoA geoA_XbaI_F & geoA_HindIII_R (Table 4). Via the added restriction enzyme recognition sites the amplicon was inserted into the multiple cloning site of two different derivatives of the broad-host range vector pBBR1MCS [69]. For confirmation of correct gene insertion the obtained plasmids pBBR1MCS-4ldi and pBBR1MCS-2geoA was sequenced.

Conjugational plasmid transfer
The donor strain, an overnight culture of E. coli S17-1 carrying the appropriate plasmid, and the recipient C. defragrans RIF were grown to late exponential phase and were mixed in several ratios (1:1, 1:5, 1:10) in a total volume of 20 μL and spread as a single drop on minimal agar. After incubation for 24 h at 28°C under oxic conditions the bacteria were resuspended in 1 mL liquid minimal medium. Dilution-to-extinction series were streaked out onto solid minimal medium supplemented with kanamycin and rifampicin and anaerobically incubated at 28°C for four days.

Preparation of cell-free extracts and determination of enzyme activities
Soluble extract preparations of C. defragrans strains 65Phen, ΔgeoA and ΔgeoAcomp were performed as described [46]. The geraniol dehydrogenase activity was monitored in a standard assay following the reduction of NAD + to NADH at 340 nm as described [47]. Equal total protein amounts were applied as certified in a 200μl aliquot by the method of Bradford [70] with BSA as standard protein; concentrations were corrected for the unusual high binding of the Coomassie stain to albumin [71].

Chemical analyses of biomass, educts and products
Nitrate and nitrite was measured by HPLC as described by [72]. Based on the fact that protein accounts for 50% of the cell mass, the Bradford assay was applied in duplicates with two different dilutions to determine the total biomass yield [72]. Geranic acid formation was assayed in liquid cultures of C. defragrans strains after confirmed nitrate depletion (Merckoquant W test strips (Merck, Darmstadt, Germany)). 10 mL cell culture was acidified with H 3 PO 4 (final concentration 0.1 M) and extracted with tert-butyl methyl ether in a 1:0.4 ratio (two biological replicates per strain). The ether extract was extracted with 0.1 M NaOH (1:1) and the aqueous phase was subjected twice to reverse-phase HPLC on a Nucleodur W C18 ISIS (4.6 mm × 250 mm, Macherey Nagel, Düren, Germany). Separation of the organic acid was performed with 1 mM H 3 PO 4 in an isocratic water-acetonitrile eluent (45/55 (v/v)) at 1 mL/min and 25°C. Intermediary, the column was cleaned with water-acetonitrile (20/80 (v/v)). UV detection was performed at 215 nm.

Additional file
Additional file 1: Additional Material.