Biochemical properties and yields of diverse bacterial laccase-like multicopper oxidases expressed in Escherichia coli

Laccases are multi-copper oxidases that oxidize a broad range of substrates at the expense of molecular oxygen, without any need for co-factor regeneration. These enzymes bear high potential for the sustainable synthesis of fine chemicals and the modification of (bio)polymers. Here we describe cloning and expression of five novel bacterial laccase-like multi copper oxidases (LMCOs) of diverse origin which were identified by homology searches in online databases. Activity yields under different expression conditions and temperature stabilities were compared to three previously described enzymes from Bacillus subtilis, Bacillus pumilus and Bacillus clausii. In almost all cases, a switch to oxygen-limited growth conditions after induction increased volumetric activity considerably. For proteins with predicted signal peptides for secretion, recombinant expression with and without signal sequence was investigated. Bacillus CotA-type LMCOs outperformed enzymes from Streptomyces and Gram-negative bacteria with respect to activity yields in Escherichia coli and application relevant biochemical properties. The novel Bacillus coagulans LMCO combined high activity yields in E. coli with unprecedented activity at strong alkaline pH and high storage stability, making it a promising candidate for further development.


Identification and description of analysed LMCOs. An increased use of LCMOs in industrial
processes requires access to a high number of efficiently produced (recombinant) enzymes of this class 29 . Ideally, to match the criteria of platform biocatalysts for technically relevant processes the set of available LMCOs should cover a broad substrate range and preference, show heat, solvent and storage stability and have activity over a broad pH range. In an attempt to find novel bacterial LMCOs which can be produced in E. coli, we performed an extensive database search using protein BLAST. The protein sequences of B. subtilis CotA and Streptomyces coelicolor EpoA were used as search templates, because both enzymes were successfully expressed in functional form in E. coli and oxidised the canonical laccase substrates ABTS (2′ -azino-bis(3-ethylbenzthiazoline-6-sulphonic acid)) and 2,6-DMP (2,6-dimethoxyphenol) 23,24 . Only hits which contained the strictly conserved, consecutive copper binding motifs HXHG, HXH, HXXHXH and HCHXXXHXXXXM/L/F of multi copper oxidases were considered for cloning.
The genome sequence of B. coagulans contained a putative LMCO protein sequence annotated as "multicopper oxidase type 2" (NCBI reference sequence YP_004860005.1). The protein exhibited 59% amino acid sequence identity to the B. subtilis CotA search template. The genome sequence of Gramella forsetii KT0803 contained a putative LMCO protein sequence annotated as "blue copper oxidase" (NCBI reference sequence YP_861212.1). The protein exhibited 47.9% amino acid sequence identity to the B. subtilis CotA search template and contained a predicted TAT signal sequence with two probable cleavage sites after residues A20 and S33. The genome sequence of Streptomyces pristinaespiralis ATCC25486 contained a putative LMCO protein sequence annotated as "copper oxidase" (NCBI reference sequence ZP_06908025.1). The protein exhibited 71% amino acid sequence identity to the S. coelicolor A3 copper oxidase search template and contained a predicted TAT signal sequence with the first, most probable cleavage site after residue A41.
Once we found ABTS oxidising activity for Gramella forsetii LMCO, we used this sequence for a further round of Protein BLAST. The genome of Marivirga tractuosa DSM4126 contained a putative LMCO protein sequence annotated as "multicopper oxidase type 3" (NCBI reference sequence YP_004054188.1). The protein exhibited 54% amino acid sequence identity to the G. forsetii blue copper oxidase search template and contained a predicted TAT signal sequence with a probable cleavage site after position C32. A plasmid sequence pSLIN02 of Spirosoma linguale DSM74 contained a putative LMCO annotated as "bilirubin oxidase" (NCBI reference sequence YP_003391614.1). The protein exhibited 26.4% amino acid sequence identity to the G. forsetii blue copper oxidase search template and contained a predicted TAT signal sequence with a probable cleavage site after position G26. Although the S. linguale sequence contained all conserved copper binding motifs, it exhibited only 17% amino acid sequence identity to B. subtilis CotA.
Proteins with disulfide bridges are often difficult to express in the cytoplasm of E. coli. The crystal structure of B. subtilis CotA revealed that the folded, copper-loaded enzyme contains a disulfide bridge between residues Cys 229 and Cys 322 30 . We aligned B. subtilis CotA with other Bacillus LMCOs and observed that these residues are conserved in the previously characterised B. pumilus and B. licheniformis enzymes, while they were missing in the B. clausii and B. coagulans homologues (Supplementary Figure  S1). The protein sequence of the recently characterised B. clausii LMCO 31 was found to contain only one cysteine residue in addition to the strictly conserved copper ligand motif CotA B. subtilis H 491 C 492 H 493 . The sequence of B. coagulans LMCO exclusively contained the copper ligand cysteine. Therefore, no disulfide bonds can be formed in B. clausii and B. coagulans LMCOs.
The putative LMCOs originated from highly divergent bacterial phyla, encompassing both Gram-positive and Gram-negative bacteria: Firmicutes (B. coagulans), Actinobacteria (S. pristinaespiralis) and Bacteriodetes (G. forsetii, M. tractuosa, S. linguale). The candidate genes were cloned into expression vectors for E. coli by PCR based methods or by gene synthesis (Table 1). In the case of G. forsetti and S. pristinaespiralis LMCOs, both full-length versions and N-terminally truncated versions lacking the predicted signal peptides were constructed. Activity yields and expression levels of phylogenetically diverse bacterial LMCOs in E. coli. Activity yields and expression levels of novel LMCOs were compared to previously described B. subtilis, B. pumilus and B. clausii CotA 23,28,31 . With the exception of S. linguale LMCO, moderate to high ABTS oxidising activity was observed after overnight induction (Table 2). In most cases a switch to static cultivation conditions after induction (imposing oxygen limitation) strongly increased volumetric activity ( Table 2). The increase ranged from 1.4-to 118-fold. Only for S. linguale LMCO this effect was not observed, but here volumetric activities were generally very low. When cells were grown in fully aerated, baffled shake flasks under continuous, vigorous shaking, higher final optical densities were reached (Table 2). During late stages of oxygen-limited cultures with standard LB medium protein synthesis might be limited by restrictions in energy supply caused by a lack of fermentable carbohydrates or by a shift to acidic pH and organic acid accumulation due to mixed acid fermentation. Thus, we investigated whether the use of a medium with high buffer capacity and an additional fermentable carbon-and energy source (LBPG medium) facilitates increased yields of active LMCOs. A substantial  Table 2). Another factor influencing the yield of correctly folded proteins in E. coli are N-terminal signal peptides. The presence or absence of native N-terminal TAT signal sequences in the constructs used for expression of LMCOs influenced activity yields considerably ( Table 2). A G. forsetii LMCO version beginning with the amino acid after the second predicted cleavage site (D34, pGoL3) yielded the highest volumetric activities (Table 2). A version starting with G21 (first predicted cleavage site, pGoL2) did not yield any active protein (data not shown). Activity yields for the full-length version of G. forsetii LMCO were reduced more than 50-fold (Table 2). Thus, the other Bacteriodetes LMCOs (M. tractuosa, S. linguale) were also expressed as versions lacking the predicted signal peptide, i.e. starting with residues corresponding to G. forsetii D34 ( Table 1). The M. tractuosa construct facilitated volumetric activities in the same order of magnitude as pGoL3, while only very low activities were detected for S. linguale LMCO ( Table 2). In the case of S. pristinaespiralis, ABTS oxidising activity was only detected for a construct which included the native TAT signal sequence. A version starting with A42 after the first predicted signal peptide cleavage site (pSpL2) did not yield any activity (data not shown). This finding is in contrast to a recent study of Gunne et al. who showed that a homologous enzyme of Streptomyces sviceus with 73.2% sequence identity to S. pristinaespiralis LMCO expressed well in E. coli without signal peptide 32 . The sequence of the S. sviceus signal peptide strongly differs from the S. pristinaespiralis signal peptide in the region before and after the predicted cutting site (data not shown). Possibly, S. pristinaespiralis LMCO could be expressed without signal peptide in E. coli if another starting residue is chosen, e.g. one after the second or third predicted cleavage site.
Heterologous DNA sequences encoding overexpressed genes often exhibit a different codon bias than that of the expression host. We investigated whether the co-expression of rare tRNAs has a positive effect on volumetric activities under the best expression conditions for each LMCO. Expression vectors were transformed into strains harboring the pRARE2 helper plasmid which supplies tRNAs for seven rare codons and has been shown to improve the expression levels of numerous human recombinant proteins in E. coli 17 . A significantly increased volumetric activity due to the helper plasmid was only achieved for B. pumilus and B. clausii LMCOs ( Table 2). For the B. subtilis, B. coagulans, S. pristinaespiralis and S. linguale enzymes no significant effect was found. A detrimental effect of pRARE2 on volumetric activity was observed for G. forsetii and M. tractuosa LMCOs. Only the DNA sequence of G. forsetii was provided as codon optimised gene, containing only 4% rare codons. All other LMCOs were cloned as wild-type genes and contained 9-12% rare codons. The effect of pRARE2 on volumetric activity did not correlate with the percentage of rare codons in the DNA sequence.
Protein expression levels were analysed by SDS-PAGE using samples of cultures with the highest volumetric activities. Specific bands of the expected size were detected in all cases in total cell protein (TCP) samples (Fig. 1). However, with the exception of B. subtilis and S. linguale LMCOs, the corresponding bands in cell free extracts (CFE) were much weaker, indicating that large fractions of the overexpressed proteins remained insoluble (Fig. 1). Solubility of the recombinant proteins did not correlate with activity yields, in particular with respect to S. linguale LMCO.
w/o= without, SP= signal peptide, S.D.= standard deviation, n.a.= not analysed Heat and storage stabilities. Previously described bacterial Bacillus LMCOs have their highest activities at elevated temperatures and resist thermal denaturation 23,25,28 , a highly interesting property with respect to biotechnological applications. We compared the heat stability of phylogenetically diverse bacterial LMCOs to a commercially available fungal laccase. In agreement with previous reports 23,31 all Bacillus LMCOs retained their full activity during 30 min incubation at 70 °C in cell-free extract (Fig. 2). The observed heat activation of B. clausii LMCO by a factor of two was also described recently by Brander et al. 31 . Two LMCOs derived from the Gram negative phylum Bacteriodetes (S. linguale, M. tractuosa) were also highly heat stable, retaining approximately 90% of their initial activity after 30 min exposure at 70 °C (Fig. 2). By contrast, a third Bacteriodetes LMCO (G. forsetii) lost 50% of the initial activity under the same conditions. Similar to previous reports on Streptomyces griseus and Streptomyces coelicolor LMCOs (actinomycetes) 24,33 , the homologous S. pristinaespiralis enzyme analysed in our study lost some activity, but was more heat stable than the fungal reference enzyme from Trametes versicolor  Technical use of enzymes requires high storage stabilities, preferably without cooling. Therefore, we analysed the stability of B. coagulans LMCO and B. clausii CotA in aqueous solution over prolonged periods of time with purified enzyme (LMCO purification: see below). In the case of B. clausii LMCO 40% of the initial activity was retained after 13 days of incubation at 25 °C in phosphate buffer with neutral pH (Fig. 3A). Inactivation at 45 °C proceeded faster; here 80% of the initial activity was lost within 3 days. The time course of activity loss of B. coagulans LMCO in phosphate buffer at 25 °C was similar; at 45 °C the enzyme exhibited slightly higher stability (Fig. 3A). Interestingly, storage stability of B. clausii LMCO was enhanced in double distilled water at both temperatures (Fig. 3B). By contrast, inactivation of B. coagulans LMCO in double distilled water proceeded faster than in phosphate buffer (Fig. 3B). Stabilities of bacterial LMCOs over a time period of weeks have rarely been studied by others. Brander et al. analysed the stability of B. subtilis and B. clausii LMCOs over a period of 25 h at 50 °C; at pH 8-10 they found 100% retention of activity 31 . We observed a somewhat lower stability of B. clausii LMCO at 45 °C. Activity was reduced by 48% after 27 h of incubation at pH 7 in phosphate buffer. This might be due differences in the experimental setup (lower pH, different buffer) compared to Brandner et al. 31 . In double distilled water, the B. clausii enzyme retained 82% of the initial activity after 27 h at 45 °C. This value and a residual activity of 82% recorded for B. coagulans LMCO after 27 h of incubation at pH 7 in phosphate buffer matched the reported values more closely.
As expected, the analysed Bacillus LMCOs where highly stable when stored either at 4 °C in phosphate buffer or at -20 °C in 50% v/v glycerol; 60%-80% of the initial activities were left after 13 d both for the B. clausii and B. coagulans enzymes (data not shown).
Substrate range, pH optimum and kinetic properties. One advantage of laccases and LMCOs is their broad substrate range. The screening of B. subtilis, B. pumilus, S. pristinaespiralis, G. forsetii, M. tractuosa and S. linguale LMCOs for oxidation of 91 potential laccase substrates has been described elsewhere 4 . In brief, Bacillus enzymes exhibited the broadest substrate range, while G. forsetii and M. tractuosa LMCOs oxidised fewer compounds and the number of transformed substrates was smallest for S. pristinaespiralis and S. linguale LMCOs. Here we provide similar data for the B. clausii and B. coagulans CotA-type LMCOs. The majority of B. subtilis CotA substrates were also oxidised by these enzymes (Table 3). Hence, promiscuity of the B. clausii and B. coagulans LMCOs was also higher than that of non-Bacillus bacterial LMCOs.
Purification of recombinant Bacillus LMCOs is simple because the majority of E. coli proteins can be separated by simple heat treatment 25,28 . We purified B. clausii and B. coagulans LMCOs by the same procedure as used for B. pumilus CotA. Purification tables and SDS-PAGE images of the consecutive purification steps are given in the Supplementary Materials ( Figure S2, Table ST1). The remarkable resistance of B. pumilus CotA to denaturation at 95 °C in SDS sample buffer 28 was also observed for B. coagulans, but not for B. clausii. In the case of B. coagulans LMCO, even 30 min of incubation in the presence of a strong detergent was not sufficient for complete denaturation (Supplementary Figure S2).
Fungal laccases are mostly active only at acidic pH 4 . Yet, some industrial processes require neutral to alkaline pH. The pH optimum of B. subtilis, B. pumilus and B. licheniformis CotA was reported to be 4-5 for the non-phenolic substrate ABTS and 7 for the phenolic substrates 2,6-DMP and syringaldazine (SGZ) 23,25,28 . As described elsewhere, the LMCOs derived from the Gram negative species G. forsetii and M. tractuosa also have a pH optimum of 4 for ABTS, while the typical phenolic laccase substrate  2,6-DMP is not oxidised at all 4 . The pH-activity profiles of the small EpoA-type LMCOs from the actinomycetes S. coelicolor and S. griseus were reported to be highly similar to Bacillus LMCOs in the case of ABTS, while a higher relative activity was reported for 2,6-DMP in the pH range of 8-10 24,33 . We confirmed these observations for the homologous S. pristinaespiralis enzyme elsewhere 4 . The only previously described CotA-type LMCO with high activity at pH 8-10 is the one from B. clausii 31 . We compared pH-activity profiles of the novel B. coagulans LMCO to B. clausii CotA for four different substrates. The optimum for ABTS oxidation was again pH 4 for both enzymes, with a continuous decline of relative activity towards neutral pH (Fig. 4A). Residual ABTS-oxidising activity of B. clausii at neutral pH was significantly increased compared to B. coagulans LMCO (Fig. 4A). In the case of 2,6-DMP, B. coagulans LMCO was as active as B. clausii LMCO at alkaline pH, but retained in addition high relative activity at pH 6-7 (Fig. 4B). In the pH range 8.5-11 B. coagulans LMCO outperformed B. clausii CotA when the phenolic compounds SGZ and guaiacol were used as substrates (Fig. 4C,D). Even at a rather extreme pH of 11, B. coagulans LMCO retained more than 60% of the activity recorded at the optimal pH. B. coagulans LMCO was highly stable at alkaline pH, even when 25% v/v dimethyl sulfoxide was added as co-solvent (Supplementary Figure S3). At room temperature, activity was completely retained for 5 days at pH10 and pH11. At 45 °C, 40-60% of the initial activity was left after the same incubation time.
Kinetic constants for purified B. clausii and B. coagulans LMCOs were determined with the two common laccase substrates ABTS and 2,6-DMP (Table 4) 31 . The values reported for conditions similar to ours (pH 7-8, k cat 4.2-4.5 s −1 , K M 1500-2000 μ M) were lower than our values. This could be due to differences in experimental procedures. In the case of SGZ it was not possible to fit a Michaelis-Menten model to the data; therefore, we did not calculate kinetic parameters ( Supplementary Figures S4C and S5C). Most likely, the unusual activity profiles were due to substrate solubility problems 28 . The maximal specific activities for SGZ oxidation were 7.6 ± 0.5 μ mol min −1 mg −1 for B. clausii LMCO and 145 ± 1.9 μ mol min −1 mg −1 for B. coagulans LMCO at 0.125 and 2 mM substrate concentration, respectively.

Discussion
The full potential of laccases and laccase-like multi-copper oxidases for sustainable production of chemicals and materials can only be realised if a broader set of robust and low-priced enzymes becomes available 34 . In particular with respect to activity at neutral to alkaline pH, resistance to heat denaturation and amenability to expression in simple bacterial expression systems, bacterial LMCOs bear high potential 35 . We expressed five novel LMCOs from diverse phyla of eubacteria in E. coli and directly compared   4 ). Out of 91 tested compounds only those are shown which were oxidised by at least one of the three enzymes.
them to previously described Bacillus LMCOs with respect to activity yields and biochemical properties. LMCOs derived from Bacillus species yielded the overall highest volumetric activities, followed by two enzymes from Bacteriodetes. In agreement with previous reports 27, 28 , all LMCOs with volumetric activities above 100 U L −1 in E. coli shake flask cultures exhibited a strict requirement for static incubation  (oxygen limitation) during the induction phase. Presumably, intracellular copper concentrations during aerobic growth of wild-type E. coli are generally too low for correct incorporation of all four metal ions in overexpressed LMCOs during folding. Thus, a static induction phase should always be used when characterising and screening novel and/or engineered LMCOs. The strict requirement for oxygen-limited conditions during induction hinders a scale-up to higher biomass concentrations. Oxygen saturation in high cell density fed-batch cultures of E. coli is usually kept above 10% in order to avoid a shift to fermentative growth 36,37 , as this would lead to rapid accumulation of toxic concentrations of organic acids. Induction of LMCO expression and shift to oxygen limitation very late in the process may solve the problem. If not, engineering of E. coli for increased intracellular copper concentrations under oxic conditions is necessary. One option is the co-expression of intracellular copper chaperones. Gunne et al. co-expressed the small soluble copper chaperone CopZ with B. licheniformis CotA which resulted in a 26% higher specific activity 38 .
No consistent effects of medium additives and co-expression of rare tRNAs were observed in our experiments. Thus, novel LMCOs should best be expressed initially with and without pRARE plasmids using several types of growth media. Volumetric activities of 2000-4000 U L −1 reached in our experiments under optimal conditions are among the highest reported for bacterial LMCOs. To our knowledge only one study reported values of a similar range: the maximum volumetric activity of B. subtilis CotA achieved with a static induction phase was 5614 U L −1 (ABTS, 37 °C) 27 . The maximal volumetric activity of another well-characterised LMCO in E. coli, wild-type B. licheniformis CotA was 410 U L −1 39 .
Our results indicate that bacterial LMCOs generally suffer from low solubility when overexpressed in E. coli. Solubility of proteins can be enhanced by substituting surface-exposed amino acids. The solubility of B. licheniformis CotA was substantially improved with a combined random and site-directed mutagenesis strategy, facilitating an 11-fold increased volumetric activity of 3400 U L −1 in E. coli 39 . A similar approach could be applied to B. coagulans and G. forsetii LMCOs which were characterised by a high amount of insoluble protein when overexpressed.
Impaired disulfide bond formation in the cytoplasm of E. coli does not seem to be related to misfolding and aggregation. B. clausii and B. coagulans CotA lack cysteine residues which can form disulfide bonds; however, activity yields were not higher than that of the B. subtilis homologue for which a disulfide bond has been observed in the crystal structure 30 .
Stability of enzymes is a key factor for their technical applicability. The high resistance of Bacillus LMCOs to thermal and chemical and inactivation was first described for B. subtilis CotA by Martins et al. 23 and was since then observed for several homologous enzymes derived from, e.g., B. licheniformis, B. pumilus and B. clausii 25,28,31 . Although the LMCOs derived from Gram negative bacteria and actinomycetes in our study also exhibited higher heat stability than Trametes versicolor laccase, the Bacillus enzymes were clearly the most stable ones. In addition we demonstrate that B. clausii and B. coagulans LMCO solutions can be stored for weeks at ambient conditions with activity retention of 50-75%, indicating that cooling is not necessary. It is likely that storage stability can be further increased by adding stabilisers such as sucrose.
LMCOs that are active at alkaline pH increase the range of applications for this class of enzymes, e.g. for lignin modification. The novel LMCO derived from B. coagulans exhibited an astonishingly high relative activity at strongly alkaline pH. With respect to the substrates SGZ and guaiacol, more alkaline pH conditions were tolerated compared to previously described alkaliphilic B. clausii CotA 31 . To our knowledge strong activity at pH 11 has not been reported before for any laccase or LMCO.
Other key factors determining the suitability of an enzyme for biotechnological processes are catalytic efficiency and substrate range. Maximal turnover numbers (k cat, derived from maximal specific activity) and Michaelis constants (K M , half saturation constant) of B. coagulans LMCO recorded for ABTS and 2,6-DMP were in the same order of magnitude as those of previously characterised Bacillus LMCOs 25,27,28,31 . Similar to previously analysed B. subtilis and B. pumilus LMCOs 4 , the homologous enzymes derived from B. coagulans and B. clausii were able to oxidize a very broad range of (mostly phenolic) compounds. This is in contrast to the rather limited substrate range of LMCOs derived from Gram-negative bacteria and Actinomycetes that cannot oxidize many of the typical phenolic substrates of fungal laccases, e.g. guaiacol and acetosyringone 4 . We suggest that such compounds, yielding coloured products upon oxidation, should be used if one wants to specifically screen for bacterial LMCOs with a broad substrate range.
In summary, this study confirms that Bacillus LMCOs offer a variety of beneficial properties, not matched by related enzymes derived from other bacterial species. Due to the high expression levels and activity yields in E. coli, high activity at alkaline pH, a broad substrate range and high stability, B. coagulans CotA seems to be a promising candidate for further development.
E. coli strains were routinely grown in LB medium at 37 °C and 150 rpm. For plasmid selection 100 mg L −1 ampicillin was added to LB agar plates and liquid medium and 34 μ g/mL chloramphenicol when cells were co-transformed with pRARE2.
Database search and construction of plasmids. Genes encoding putative LMCOs were identified in bacterial genomes by Protein BLAST searches (http://blast.ncbi.nlm.nih.gov/Blast.cgi) using sequences of the following bacterial proteins with confirmed ABTS oxidising activity as search templates: B. subtilis CotA 23 (NCBI reference sequence NP_388511.1), S. coelicolor A3 copper oxidase 33 (NCBI reference sequence NP_630785.1) and G. forsetii blue copper oxidase (NCBI reference sequence YP_861212.1, this study). The presence of putative N-terminal signal sequences was analysed with the Signal P 3.0-HMM program using appropriate networks for the analysed organisms (http://www.cbs.dtu. dk/services/SignalP/) 40 . Alignments of amino acid sequences and calculations of amino acid sequence similarities were carried out with the MegAlign TM software (LASERGENE, Madison, USA) using the ClustalW algorithm.
Plasmids used in this study for expression of bacterial LMCOs are listed in Table 1. Primer sequences and details of plasmid construction are described in the Supplementary Material. Genomic and plasmid DNA was extracted with commercial kits according to the instructions of the suppliers. Gene sequences, cloning sites and upstream and downstream regions at the 5′ and 3′ ends of LMCO-encoding inserts were controlled by sequencing. The nucleotide sequences of all genes cloned from genomic DNA matched the corresponding database sequences to 100%. E. coli BL21(DE3) was used as host strain for pET derived plasmids. E. coli JM109 (overexpressing lacI q from the genome) was used as host strain for pQE-60 derived plasmids. In order to test the effect of additional copies of rare tRNAs, E. coli Rosetta 2 (DE3) competent cells (Novagen) harboring pRARE2 were transformed with pGoL3, pSiL, pMtraL, pBuL, pLOM10, pBoL and pBaL. JM109 competent cells were transformed with pRARE2, previously isolated from Rosetta 2 cells, made competent and then transformed with pSpL3.

Expression conditions and cell lysis. For expression of recombinant bacterial LMCOs either LB
(5 g L −1 yeast extract, 10 g L −1 tryptone, 5 g L −1 NaCl) or LBPG (phosphate buffered, LB-like medium with glucose: 5 g L −1 yeast extract, 10 g L −1 tryptone, 7.25 g L −1 Na 2 HPO 4 2H 2 O, 3 g L −1 KH 2 PO 4 , 4 g L −1 glucose, pH 7.2) medium was used. Expression studies were carried out in n = 3 biological replicates per condition. Shake flask cultures were inoculated 1:50 from individual overnight tube pre-cultures (37 °C, 160 rpm) and incubated at 30 °C and 160 rpm until an OD 600 of 0.4-0.5. Subsequently, the cultivation temperature was reduced to 25 °C, the expression of recombinant proteins was induced by the addition of IPTG 1 mM and CuCl 2 0.25 mM final concentration, respectively. A subset of experiments was carried out in baffled flasks (total volume 500 mL, liquid volume 60 mL) which were continuously shaken at 160 rpm until the end of the experiment (SF shaken, aerobic growth). Another subset of experiments was carried out in normal Erlenmeyer flasks (total volume 200-300 mL, liquid volume 60 mL) which were shifted from 160 to 100 rpm agitation at induction and to 0 rpm (static incubation) 4 h after induction (SF static). LBPG cultures were shifted to incubation without agitation directly after induction. Static incubation leading to oxygen-limited growth had been shown previously to increase the yield of fully copper loaded bacterial LMCO in E. coli 27 . Final optical densities were determined at 600 nm with a microplate reader after appropriate dilution of samples. Volumetric activities were determined with OD 600 -normalised, lysed cell suspensions without centrifugation. Cells were harvested after overnight incubation by centrifugation (1 min, ≥ 12000 x g) and cell pellets were frozen at -80 °C. Pellets were thawed and resuspended to a calculated OD 600 of 10 in 1x CellLytic B solution (Sigma-Aldrich) containing 0.2 mg mL −1 lysozyme and 0.5 μ L mL −1 Benzonase® Nuclease (Novagen). Cell lysis was carried out for 20 min at room temperature and 200 rpm. LMCO activity was quantified with the ABTS assay (see below) except for S. linguale LMCO where activity was determined with 0.5 mM 2,6-dimethoxyphenol (2,6-DMP) at pH 7.0. Expression conditions with the highest activity yields were used for experiments with the pRARE2 helper plasmid supplying rare tRNAs.
For heat stability experiments and SDS-PAGE analysis, cell debris was removed from the lysed samples by centrifugation (5 min, ≥ 12000 x g) to recover cell free extract (CFE). Before loading on SDS-PAGE gels, CFE samples were mixed 1:1 with 2x reducing SDS-PAGE sample buffer. Total cell protein (TCP) samples for SDS-PAGE analysis were prepared by resuspending cells to an OD 600 of 10 in 1x SDS-PAGE sample buffer. All SDS-PAGE samples were denatured at 95 °C for 20 min before electrophoresis.
Biochemical assays. Enzymatic activity of lysed cells and CFE was determined at room temperature with 0.5 mM ABTS as substrate in 50 mM tartaric acid buffer at pH 4 41 in transparent polystyrene 96-well microplates (Nunc). The reactions were followed with a microplate reader (Bio Tek Synergy Mx spectrophotometer) at 420 nm. Enzymatic activity was calculated using an extinction coefficient for mono-oxidised ABTS of 36′ 000 M −1 cm −1 41 . Resistance to heat denaturation was determined by incubating 100 μ L CFE aliquots in 1.5 mL Eppendorf tubes at 70 °C and 300 rpm in a Thermomixer (Eppendorf). Samples were periodically taken and directly assayed at room temperature for enzymatic activity (ABTS, pH 4, S. linguale: 2,6-DMP, pH 7). Commercial Trametes versicolor laccase (Sigma-Aldrich) was dissolved to a concentration of 0.05 mg/mL in the same buffer as used for cell lysis (1 x CellLytic B). B. clausii and B. coagulans LMCOs were purified from overnight shake flask cultures with static induction phase as described previously 28 . Briefly, cell free extracts were prepared from 0.7 L induced overnight cultures by a combined freeze-thaw, lysozyme and sonication procedure. Non-heat stable proteins were precipitated by incubation at 70 °C for 20 min. LMCOs were further separated from contaminating proteins by anion exchange chromatography and size exclusion chromatography, which was sufficient for achieving protein samples of high purity as determined by SDS-PAGE. Protein concentrations after each purification step were determined with the Bradford method using premixed reagent (Sigma). Purified LMCOs were quantified at 280 nm with a Nanodrop device (Thermo Fisher Scientific, Waltham, USA) using extinction coefficients calculated from protein sequences (http://web.expasy.org/protparam/). Storage stabilities of purified enzyme preparations were determined after dilution in the respective storage solutions to initial volumetric activities of 0.3-0.4 U mL −1 (ABTS, pH 4, 30 °C). Samples were incubated at 25 °C and 45 °C and aliquots were assayed at regular intervals for ABTS oxidising activity at 30 °C.
The substrate range was analysed in 96-well plates by monitoring spectroscopic changes as described previously 4,28 . Wells contained 1 mM of each substrate in McIlvaine buffer (0.1-0.2 M citrate-phosphate) 42 with pH 6 and reactions were initiated by addition of CFE. Compounds were counted as substrate when the UV/Vis spectrum was significantly changed after 24 h of incubation at 37 °C compared to control wells with CFE of a strain carrying the empty expression vector pET22b.
The relative activity at different pH values was recorded at 30 °C using Teorell-Stenhagen universal buffer because of its wide pH range of 2-12 43 . Substrates were added at concentrations that yielded the highest specific activities during determination of kinetic constants. At high pH values auto-oxidation was observed in enzyme free controls for the substrates 2,6-DMP and guaiacol. In these cases, auto-oxidation rates were subtracted from rates measured in experiments with enzymes before calculating relative activities.
Kinetic constants of B. clausii and B. coagulans LMCOs were determined at 30 °C with purified enzyme preparations as described by Reiss et al. 28