Recently the purification and characterization of a novel FAD-dependent glucose dehydrogenase produced by the plant pathogenic fungus G. cingulata and its proposed role in plant pathogenicity were published . The reported features of this GDH are of interest in two respects: (i) to elucidate the role in the mechanism of plant-pathogen interactions during the infection process and (ii) in electrochemical applications [13, 14]. To facilitate biochemical and structural studies as well as engineering of G. cingulata FAD-dependent GDH, the heterologous expression of GcGDH was investigated. To target potential problems with the expression of a heavily glycosylated eukaryotic flavoprotein in a prokaryotic host several approaches were taken. Along with expression of GcGDH with varying N-termini under mild conditions (auto inducing minimal media, 20°C) we also tested different E. coli expression strains for their suitability to express soluble and catalytically active GcGDH.
The effect of the N-terminal amino acids on the expression levels of a fungal FAD-dependent GDH in E. coli was shown in the US patent 7,741,100 . Expression levels could be increased approximately 10-fold by deletion of the signal sequence of A. oryzae GDH. Therefore, GcGDH was expressed in full length and with the native signal sequence removed. A third, truncated N-terminus was designed according to a sequence alignment of closely related members of the GMC oxidoreductase family. The N-terminal sequences that were successfully used for the expression of A. oryzae GDH  and the flavin domain of Phanerochaete chrysosporium cellobiose dehydrogenase (CDH) in E. coli  seem to be highly conserved in these closely related proteins. The analogous sequence MTAYDYIVI was therefore chosen as N-terminal sequence for the third variant of GcGDH. Surprisingly, although in a prokaryotic expression host, expression levels of GcGDH were highest with the full-length protein, which included its own signal sequence. For the variant lacking the signal sequence the volumetric activity decreased three-fold, and no activity was detected for the third and shortest construct. For all tested expression constructs the fraction of GDH protein found in inclusion bodies (as judged by SDS-PAGE) was high. For the rather closely related P. amagasakiense glucose oxidase (GOX) refolding experiments from inclusion bodies were successful, retrieving ~10% of the totally aggregated GOX in an active form . Although the same or slightly modified conditions were used, the same result could not be reproduced for GcGDH. We conclude, that although GOx is the phylogenetically closest relative of GDH , the structure of GDH is different enough not to favour cofactor reconstitution under the same or similar conditions.
In addition to in vitro refolding of incorrectly folded protein several other methods have been described in literature for promoting the synthesis of active recombinant protein in the soluble cytoplasmic fraction rather than as inclusion bodies [18, 19]. Increased amounts of the chaperone system GroEL/GroES in the cytoplasm apparently reduces the accumulation of aggregated GcGDH in the cell, leading to small amounts of active soluble GcGDH. The supply of tRNAs for 7 rare codons by the strain Rosetta 2, showed no beneficial effect on the expression of GcGDH. This, however, was to be expected since codon analysis of the gcgdh gene revealed no sequences that could affect the transcriptional or translational efficiencies.
A further strategy to reduce the in vivo aggregation of recombinant GcGDH in E. coli was to use slow growth and weak inducing conditions. To this end, the cultivation temperature was lowered to 20°C and an auto-inducing medium (MagicMedia) was used. It was shown previously that yields of a target protein as well as cell mass can be increased substantially by using such mild conditions . Cell densities were increased up to 30 g L-1 compared to 10 g L-1 obtained by the standard LB medium. Even though all these considerations were taken into account for the expression of GcGDH in E. coli a volumetric activity of10 U L-1 could be produced under optimized conditions. Since expression rates in P. pastoris were much higher no effort was made to purify GcGDH from E. coli cultures.
When using the eukaryotic expression system, GcGDH could be expressed extracellularly in high yields using the native signal sequence, which indicates that this signal sequence is properly recognized and processed by the yeast. A final volumetric activity of 48,000 U L-1 and a space-time yield of 24 mg L-1 d-1 could be achieved by P. pastoris. This is a 70-fold improvement of the space-time yield compared to the wild-type producer. The cultivation yielded a total of 57 mg L-1 of recombinant protein, which corresponds to ~20% of total extracellular protein. The purification protocol resulted in a protein preparation of high purity (as checked by SDS-PAGE) with a specific activity of 836 U mg-1, which is comparable to the wild type preparation (840 U mg-1,,). Since the first purification step already yielded a protein of high specific activity (833 U mg-1) the procedure might be reduced to a one-step purification. All (bio)physical and catalytic properties studied for recGcGDH are essentially identical to those of the wild-type enzyme isolated from the original source G. cingulata (Table 3, ). The high degree of glycosylation of recombinant GcGDH (approx. 65% as judged from SDS-PAGE, Figure 2) is also found in native GcGDH (approx. 70%, ). These values are certainly an overestimation by SDS-PAGE, which is known to smear bands of glycosylated proteins, but the range of the bands of native (95-135 kDa) and recombinant (88-131 kDa) GcGDH are nearly identical. The temperature optimum for recGcGDH is 46°C and close to the previously reported value for an FAD-dependent glucose dehydrogenase from A. terreus (50°C) .
This study reports and compares the successful heterologous expression of Glomerella cingulata GDH in P. pastoris and E. coli. The glycosylation of this protein seems to play an important role for folding into the correct conformation, as already shown for other proteins as well . This makes the eukaryotic host more suitable for the production of recGcGDH, which displays properties that are essentially identical to those of the wild-type enzyme . The expression in E. coli has the advantage that glycosylation-free GcGDH can be obtained, which is useful for e.g. crystallization studies. However, for this application the production in the prokaryotic host has to be optimized further to provide sufficient amounts of protein.