Extracellular targeting of an active endoxylanase by a TolB negative mutant of Gluconobacter oxydans

Abstract

Gluconobacter (G.) oxydans strains have great industrial potential due to their ability to incompletely oxidize a wide range of carbohydrates. But there is one major limitation preventing their full production potential. Hydrolysis of polysaccharides is not possible because extracellular hydrolases are not encoded in the genome of Gluconobacter species. Therefore, as a first step for the generation of exoenzyme producing G. oxydans, a leaky outer membrane mutant was created by deleting the TolB encoding gene gox1687. As a second step the xynA gene encoding an endo-1,4-β-xylanase from Bacillus subtilis was expressed in G. oxydans ΔtolB. More than 70 % of the total XynA activity (0.91 mmol h⁻¹ l culture⁻¹) was detected in the culture supernatant of the TolB mutant and only 10 % of endoxylanase activity was observed in the supernatant of G. oxydans xynA. These results showed that a G. oxydans strain with an increased substrate spectrum that is able to use the renewable polysaccharide xylan as a substrate to produce the prebiotic compounds xylobiose and xylooligosaccharides was generated. This is the first report about the combination of the process of incomplete oxidation with the degradation of renewable organic materials from plants for the production of value-added products.

Introduction

Gluconobacter oxydans is a member of the alpha proteobacteria and belongs to the family Acetobacteraceae [14]. The organism has the ability to grow at low pH values and high sugar concentrations [15, 18, 51]. Gluconobacter spp. catalyze the incomplete and regioselective oxidation of alcohols and carbohydrates, especially monosaccharides, at high rates using several flavin- and pyrroloquinoline quinone-dependent dehydrogenases. These enzymes are located in the cytoplasmic membrane with their active site oriented towards the periplasm where the incomplete oxidation of substrates takes place [15]. Thus, compounds that are used as energy sources only have to cross the outer membrane via porins, and the oxidation products are easily released into the medium via porins. Electrons derived from these reactions are channeled into the aerobic respiratory chain to establish a transmembrane electrochemical potential for ATP synthesis, active transport, and flagellar rotation. The process of incomplete oxidation by Gluconobacter spp. is of great importance for biotechnological and industrial applications, which include production of l-sorbose from d-sorbitol for vitamin c production [45], the synthesis of 1-amino-l-sorbose from 1-amino-d-sorbitol as precursor for antidiabetic drugs [48], or the production of dihydroxyacetone as tanning agent. Furthermore, the organisms are used for the synthesis of gluconate (e.g. Bionade production) and ketogluconates [23]. Moreover, G. oxydans has applications in food additive production by synthesizing flavoring ingredients from aromatic and aliphatic alcohols [43]. Enzymes, membranes, and whole cells are also used as sensor systems for the detection of polyols, sugars, and alcohols [4, 33, 49], for bioelectrocatalytic oxidation of ethanol [46], and for high-yield biocatalysis of xylonic acid from xylose [62].

G. oxydans is specialized in the oxidation of monosaccharides, while growth with disaccharides as sole carbon source is either very low (e.g. sucrose) or impossible (e.g. lactose) because disaccharide hydrolyzing enzymes are missing. In addition, growth on polysaccharides (e.g. xylan, cellulose) is not possible because extracellular hydrolases are not encoded in the genome of Gluconobacter strains [42]. Owing to the great biotechnological and industrial significance of G. oxydans and the fact that microbial hydrolysis of renewable raw materials as inexpensive and abundant substrates for production of high value chemicals is becoming increasingly important, an extension of the substrate spectrum of this organism for polysaccharides would be highly desirable. In a first step, G. oxydans was metabolically engineered to hydrolyze trehalose in the periplasmic space [25]. This engineered strain was able to cleave trehalose by heterologous production of a periplasmic trehalase. The released glucose could then be used as a growth substrate. In this study, we took advantage of the same expression system for heterologous expression, production, and translocation of the Bacillus subtilis endoxylanase XynA in G. oxydans. To overcome the outer membrane as a barrier between active XynA in the periplasm and the polysaccharide substrate xylan, a leaky outer membrane strain of G. oxydans was generated that released more than 70 % of the active XynA into the medium.

G. oxydans is naturally designed for industrial applications by forming high-value products because of its ability to catalyze rapid stereo- and regiospecific incomplete oxidation reactions of many sugars and sugar derivatives. This is the first report to describe combining this unique biochemical feature with the ability to produce exoenzymes for the hydrolysis of polysaccharides to produce value-added products starting from abundant and inexpensive polysaccharides in acetic acid bacteria. This approach makes production more sustainable as well as cost effective.

Materials and methods

Materials

All chemicals and reagents were purchased from Sigma-Aldrich (Munich, Germany) and Carl Roth GmbH (Karlsruhe, Germany). Restriction endonucleases, Taq DNA polymerase, T4 ligase, and PCR reagents were obtained from Fermentas (St. Leon-Rot, Germany) and Phusion DNA polymerase from New England Biolabs (Frankfurt am Main, Germany). All oligonucleotides were synthesized by Eurofins (Ebersberg, Germany).

Strains and culture conditions

G. oxydans 621H ΔhsdR (hereto referred to as G. oxydans; Table 1) was used for all experiments and was grown in yeast mannitol (YM) medium consisting of 2 % d-mannitol and 0.6 % yeast extract. Escherichia coli DH5α was grown in lysogeny broth [34]. For plasmid maintenance 50 µg/ml kanamycin was added when appropriate [19].

Strains, plasmids and primers

Strain, plasmid, primer	Description or sequence	Source or restriction site
Strains
E. coli DH5α	F-, ø80d lacZΔM15, Δ(lacZYA-argF) U169, deoR, recA1, endA1, hsdR17 (rk−, mk+), phoA, supE44, λ−, thi-1, gyrA96, rel A1	DSM 6897 (17)
G. oxydans 621H ΔhsdR	ΔhsdR (Δgox2567) derivative of G. oxydans 621H (DSM 2343). Referred to as G. oxydans throughout the paper	S. Bringer-Meyer, Forschungszentrum Jülich GmbH
G. oxydans 621H ΔhsdR ΔtolB	ΔtolB (Δgox1687) derivative of G. oxydans 621H ΔhsdR Referred to as G. oxydans ΔtolB throughout the paper	This study
G. oxydans phoA	G. oxydans expressing phoA (b0383) from E. coli	This study
G. oxydans ΔtolB phoA	G. oxydans ΔtolB expressing phoA (b0383) from E. coli	This study
G. oxydans xynA	G. oxydans expressing xynA (BSU18840) from B. subtilis	This study
G. oxydans ΔtolB xynA	G. oxydans ΔtolB expressing xynA (BSU18840) from B. subtilis	This study
Plasmids
pKOS6b	pAJ63a derivative with codBA, mob, KanR, FCS, lacZα	[26]
pKOS6b-UpDogox1687	pKOS6b derivative containing the fusion product of up- and downstream region of gox1687	This study
pBBR1p264-SPpelB-Streplong	pBBR1MCS-2 derivative containing the 5′UTR of gox0264, a strep-tag and the signal sequence of pelB, OriT, mob, KanR, OriRep, MCS	[59]
pBBR1p264-SPpelB-phoA	pBBR1p264-SPpelB-Streplong derivative expressing phoA (b0383) from E. coli	[25]
pBBR1p264-SPpelB-xynA	pBBR1p264-SPpelB-phoA derivative expressing xynA (BSU18840) from B. subtilis	This study
Primer
XynA.fw	ATATTACGTAGCTAGCACAGACTACTGGC	SnaBI
XynA.rev	ATATGGCGCGCCCCACACTGTTACGTTAG	AscI
pKOS6b1687.fw	TCCCAGTCACGACGTTGTAAAACGAC
pKOS6b1687.rev	ATGATTACGAATTCGAGCTCGGT
Upgox1687_3.fw	ATAGAAGCTTAGATGGAATTCGAGCAGGCA	HindIII
Upgox1687.rev	TGAAAGCGCGGAGCCTGACCGAGACTTCCTCGTGGATTAC
Downgox1687.fw	GTAATCCACGAGGAAGTCTCGGTCAGGCTCCGCGCTTTCA
Downgox1687.rev	TAGTTCTAGAATCATCCCAGTCCGT	XbaI
codA.F	TGTGGAGGCTAACAGTGTCG
codA.R	GATAATCAGGTTGGCGCTGT
p264SP.f	GTGCAGAGTCTCTGCAGGAT
pASK_R	CGCAGTAAGCGGTAAACG

Construction of G. oxydans ΔtolB

For the deletion of gox1687 we used the codAB markerless deletion method as previously described [26]. The up- and downstream regions of gox1687 (900 and 1,000 bp) were amplified using primers Upgox1687_3.fw/Upgox1687.rev and Downgox1687.fw/Downgox1687.rev. The amplified fragments were used as template for a fusion PCR reaction using Upgox1687_3.fw/Downgox1687.rev to fuse both fragments. After digestion with HindIII and XbaI and ligation in the corresponding sites in pKOS6b, the resulting vector pKOS6b-UpDogox1687 was transferred into competent E. coli DH5α cells. Vector pKOS6b-UpDogox1687 was electroporated into competent G. oxydans as described below. The markerless deletion of gox1687 was generated by sequential screening on kanamycin containing and 5-fluorocytosine containing medium as described by Kostner et al. [26]. All plasmids and clones were screened for proper insertion or deletion by PCR using primers described in Table 1. Positive clones and mutants were verified by sequencing (StarSeq, Mainz, Germany).

Construction of a xynA expression system

The gene xynA (BSU18840) from B. subtilis, encoding the exoenzyme endo-1,4-β-xylanase A, was amplified by PCR without its native signal site and with addition of the endonuclease restriction sites AscI and SnaBI using primers XynA.fw/XynA.rev. The amplicon was ligated into the SnaBI and AscI digested pBBR1p264-SPpelB-phoA [25] in-frame to the signal sequence of pelB and the strep-tag resulting in the expression vector pBBR1p264-SPpelB-xynA.

Standard molecular techniques

All routine molecular techniques were done as previously described [47]. Transformation of G. oxydans strains was done via electroporation [35]. Briefly, 4 ml of cells was harvested between an OD₆₀₀ of 0.9 and 1.0 (centrifugation 13,000×g, 5 min at 4 °C) and the pellet was washed twice with 1 ml 1 mM sterile HEPES buffer. All steps were done on ice. The pellet was resuspended in 40 µl HEPES and glycerol was added to a final concentration of 20 % (w/w) and used as electrocompetent cells.

Overexpression and purification of XynA

Overnight cultures of G. oxydans (5 ml) containing pBBR1p264-SPpelB-xynA were used to inoculate 500 ml YM medium. Cells were harvested at an optical density between 0.6 and 0.8 by centrifugation at 8,000×g and resuspended in buffer W (100 mM Tris–HCl, 150 mM NaCl, pH 8.0). Cells were lysed via sonication and centrifuged at 13,000×g at 4 °C to separate cell debris. The supernatant was applied to a gravity flow Strep-Tactin Superflow^® affinity column (IBA GmbH, Goettingen, Germany). Protein was eluted by application of buffer E (100 mM Tris–HCl, 150 mM NaCl, 2.5 mM desthiobiotin, pH 8.0). Purification of XynA from culture supernatants was done in the same way, except cells were pelleted before lysis and the supernatant was applied to a Strep-Tactin Superflow affinity column.

Polyacrylamide gel electrophoresis

Samples were diluted 1:2 in sample loading buffer (50 % [v/v] glycerol, 20 % [v/v] collecting buffer [pH 6.8], 5 % [v/v] β-mercaptoethanol, 2 % [w/v] SDS, 0.001 % [w/v] bromophenol blue) and boiled for 10 min prior to separation. SDS-PAGE was performed as previously described [27] with a 5 % (w/v) polyacrylamide stacking gel and a 12.5 % (w/v) separating gel. Protein visualization was done by silver stain [7].

Cell fractionation

For the measurement of activity distribution in different cell fractions the cells were grown to an OD₆₀₀ of 0.6–0.8, and the preparation of periplasmic and cytoplasmic fractions was done as previously described by Kosciow et al. [25] using a modified version of Tsukada and Perlman [55]. For final analysis, the activities of PhoA or XynA found in the cytoplasmic and periplasmic fraction were combined and referred to as activities in the cellular fraction.

Enzyme assays

Alkaline phosphate activity was monitored by the method of Brickman and Beckwith [9]. Briefly, 50 µl of each cell fraction was diluted in 940 µl 1 M Tris–HCl buffer (pH 8) and the substrate p-nitrophenyl phosphate (pNPP) was added to a final concentration of 1 mM. The increase of absorbance was monitored at 405 nm at 30 °C. Endoxylanase activity was analyzed essentially as described [5]. A 250-µl aliquot of cell culture, cell fraction, or purified protein was mixed with 250 µl 0.1 M acetate buffer (pH 6.8) containing 11.5 mg/ml Remazol brilliant blue-birchwood xylan (RBB-xylan). After incubation at 30 °C, 1 ml ethanol was added and the reaction was centrifuged at 13,000×g for 2 min at room temperature to separate cell debris and insoluble RBB-xylan. The absorbance of the supernatant was measured at 595 nm. The amount of released RBB dye was quantified using the molar extinction coefficient (ε = 8,266 mM⁻¹ cm⁻¹) [44] with adjustment for a 17 % labeling efficiency of the RBB-xylan (Sigma-Aldrich). All enzymatic activities were determined in triplicate. One unit of purified enzyme activity corresponded to the release of 1.0 μmol of RBB dye per minute. Activity of cultures and cell fractions is expressed as mmol RBB dye released per h and 1 l culture.

To analyze the amount of contaminating cytoplasmic proteins in the supernatant a routine method for the NADP-linked glucose-6-phosphate dehydrogenase reaction was used. Assays were performed at ambient temperature in 50 mM Tris–HCl (pH 7.5) and contained 250 µM NADP⁺ and 2.5 mM glucose-6-phosphate as substrates [25]. Formation of NADPH was monitored at 340 nm and activities were calculated using the molar extinction coefficient (ε = 6.22 mM⁻¹ cm⁻¹). 97–99 % of NADP-linked glucose-6-phosphate dehydrogenase was found in cell extracts of the xynA or phoA expressing strains and only 1–3 % was in the culture supernatants (data not shown).

Plate assays

The 5-bromo-4-chloro-3-indolyl phosphate (BCIP) plate assay was done according to Brickman and Beckwith [9]. YM plates containing 1.5 % agar and 0.1 % BCIP solution (20 mg/ml N,N-dimethylformamide) were used to verify the secretion of PhoA into the medium resulting in a blue halo surrounding the colony. The toluidine blue indicator plate assay was used to analyze the diffusion of periplasmic RNases into the medium, indicating leaky membranes [24]. For this assay YM plates containing 1.5 % agar, 75 mg toluidine blue O per liter, and 0.2 % torula yeast RNA were used.

Microscopy

For observation of G. oxydans cells, a Zeiss Axio Observer Z1 microscope with an Axiocam MRM camera and the ZEN 2012 software for image capture and analysis was used. For phase contrast imaging, 1 µl of cell cultures was placed on an agarose layer (1 %) in phosphate-buffered saline mounted on a microscope slide and examined using a 100× objective (phase contrast 3, 1.46 oil).

Results

Construction of G. oxydans ΔtolB and characterization of its phenotype by microscopy and growth behavior

The aim of this study was to construct a G. oxydans strain that was able to secrete exoenzymes into the supernatant of the culture medium for the production of value-added compounds from polysaccharides such as xylan. A potential route is engineering strains with novel biochemical abilities by incorporating genes encoding polysaccharide-hydrolyzing exoenzymes. Furthermore, mutants have to be generated that are able to secret those enzymes into the medium to hydrolyze polysaccharides that cannot easily pass through outer membrane porins.

It is known that the deletion of lipoprotein encoding genes, such as lpp or parts of the tol–pal complex, result in leaky outer membranes and periplasmic protein loss in E. coli and Pseudomonas aeruginosa [10, 29, 32, 39, 50]. For E. coli, various lipoproteins knock-outs were constructed and the total release of periplasmic alkaline phosphatase was measured. The two most effective deletions were ΔtolA and ΔtolB, resulting in a release of up to 100 % of PhoA into the extracellular space, respectively [29]. Homology studies with TolB from E. coli showed that this protein is encoded by gox1687 in G. oxydans (28 % identity, 45 % similarity and 90 % coverage, BLASTp NCBI, http://blast.ncbi.nlm.nih.gov/Blast.cgi). Therefore, the first step in the generation of exoenzyme producing G. oxydans was the creation of a leaky outer membrane mutant by deleting gox1687 using the codAB markerless deletion method [26]. Phase contrast microscopy of G. oxydans ΔtolB showed chains of cells of various sizes, whereas the wild-type typically formed pairs of short rods (Fig. 1). This suggests that the ΔtolB mutant strain had defects in the cell division process. To investigate the influence of the tolB gene on growth dynamics and the division defects in the tolB deletion mutant, growth experiments were performed using 50 mM YM medium (Fig. 2). G. oxydans reached a final OD₆₀₀ of 1.6 and had a doubling time of approximately 1.0 h in the first part of the exponential growth phase that increased to 1.5 h in the second part. The TolB-deficient strain grew to a final optical density of 1.45 with a doubling time of 1.5 h. Hence, the growth characteristics of the wild-type and TolB-mutant strain were similar with the mutant showing only slight reduction in final optical density and initial growth rate.

Fig. 1

Morphology G. oxydans and G. oxydans ΔtolB. Phase contrast images of G. oxydans (a) and G. oxydans ΔtolB (b)

 

Fig. 2

Growth behavior of G. oxydans in comparison to the TolB-deficient mutant strain. G. oxydans (open triangle) and G. oxydans ΔtolB (filled triangle)

Analysis of the outer membrane permeability of G. oxydans ΔtolB

After phenotypical analysis of G. oxydans ΔtolB, the increased permeability of the outer membrane was characterized to verify suitability for exoprotein production. The free diffusion of periplasmic ribonucleases into the surrounding medium due to the leakiness of the outer membrane was used to assess the increase in outer membrane permeability. This system was previously applied to identify leaky mutant strains of E. coli and Salmonella typhimurium [28, 30, 31, 57]. The G. oxydans genome contains nine genes encoding predicted ribonucleases. Bioinformatic analysis using SignalP 4.1 [41] and Phobius [22] revealed that only ribonuclease I, encoded by gox2020, possesses a signal peptide and is predicted to be periplasmic (PSORTb v3.0.2, http://www.psort.org/psortb/results.pl) [60]. The toluidine blue indicator plate assay indicated that the wild type degraded only a small amount of yeast RNA in the agar. As these are actively growing cultures, the low activity is likely the result of a small amount of lysed cells present that released cytoplasmic and periplasmic RNases (Fig. 3a). In contrast, G. oxydans ΔtolB formed an intense halo around the colonies, suggesting a massive release of the periplasmic ribonuclease I (Fig. 3b).

Fig. 3

Outer membrane permeability plate assays. a RNase activity of G. oxydans (toluidine blue indicator plate assay). b RNase activity of G. oxydans ΔtolB (toluidine blue indicator plate assay). c BCIP plate assay with G. oxydans phoA. d BCIP plate assay with G. oxydans ΔtolB phoA

To examine the functionality and usefulness of the previously described periplasmic expression system [25] for exoprotein production in G. oxydans, the plasmid pBBR1p264-SPpelB-phoA was transformed into G. oxydans and G. oxydans ΔtolB. This vector contains a strong constitutive G. oxydans promotor p264 [21], a pectate lyase B signal sequence (SPpelB) [61], as well as the phoA gene fused to a strep-tag affinity tag. The phoA gene encodes for the alkaline phosphatase (PhoA) from E. coli that functions as reporter enzyme for translocation efficiency analysis. On BCIP plates only a very small blue halo appeared around wild-type G. oxydans colonies harboring the PhoA reporter system (Fig. 3c). Again it is tempting to speculate that small amounts of PhoA were released by lysed cells. In contrast, G. oxydans ΔtolB containing the vector pBBR1p264-SPpelB-phoA for expression of periplasmic PhoA showed a very strong halo as a clear indication for large amounts of PhoA secretion into the medium (Fig. 3d).

For quantification of PhoA activity, G. oxydans phoA and G. oxydans ΔtolB phoA were grown to a final OD_600nm of 0.6–0.8. In G. oxydans phoA, 97.4 % of the PhoA activity was found in the cellular fraction. The culture supernatant contained only 2.6 % of the total PhoA activity (Fig. 4). In contrast, G. oxydans ΔtolB phoA had only 50.4 ± 3.0 % of PhoA activity in the cellular fraction, and 49.6 ± 3 % in the supernatant (Fig. 4). Hence, PhoA activity in the TolB-deficient strain was 19-fold higher in the culture supernatant compared to G. oxydans phoA.

Fig. 4

Distribution of alkaline phosphatase activity. PhoA activity was measured after fractionation of G. oxydans phoA and G. oxydans ΔtolB phoA, respectively. Total activity corresponded to 4.6 ± 0.7 mmol h⁻¹ l culture⁻¹ for the wild-type and 6.3 ± 0.6 h⁻¹ l culture⁻¹ for the ΔtolB mutant, respectively. Dark gray bars culture supernatant. Light gray bars cell extract

Secretion of active endoxylanase A in G. oxydans ΔtolB

To verify the potential of G. oxydans ΔtolB as a producer of exoenzymes and to analyze its ability for polysaccharide degradation, the phoA expression vector pBBR1p264-SPpelB-phoA was digested with SnaBI and AscI and phoA was replaced with xynA, which encodes an endo-1,4-β-xylanase A (BSU18840) from B. subtilis [3]. After electroporation, G. oxydans ΔtolB harboring the expression system pBBR1p264-SPpelB-xynA was fractionated as described above and enzyme activity in the cell extract and supernatant fractions was determined (Fig. 5). The two fractions comprised a total activity of 1.18 mmol h⁻¹ l culture⁻¹. Only 29.1 ± 7.8 % of the total XynA activity was detected in the cellular fraction, while 70.9 ± 8.0 % was in the supernatant of the TolB mutant. Hence, the efficiency of protein export was even higher compared to translocation of PhoA into the culture supernatant (Fig. 4). In contrast, G. oxydans xynA excreted only 9.9 ± 5.0 % of the total endoxylanase activity into the supernatant and 90.1 ± 4.8 % was found in the cellular fraction (Fig. 5a).

Fig. 5

Distribution of endoxylanase A activity (a) G. oxydans xynA and G. oxydans ΔtolB xynA. b SDS-PAGE of purified XynA. Culture supernatant and cell extract were prepared from 250 ml culture. Total RBB-releasing activity corresponded to 0.47 ± 0.05 mmol h⁻¹ l culture⁻¹ for G. oxydans xynA and 1.18 ± 0.21 mmol h⁻¹ l culture⁻¹ for G. oxydans ΔtolB xynA. Dark gray bars culture supernatant. Light gray bars cell extract

Since the xynA gene was cloned in-frame to a 3′ strep-tag sequence, the protein contained a C-terminal strep-tag that allowed us to purify XynA using a Strep-Tactin affinity column. Purified XynA showed the expected protein band with a molecular mass of 21.4 kDa (including the strep-tag but without the PelB signal peptide) by SDS-PAGE (Fig. 5b). The protein was purified from the culture supernatants of G. oxydans xynA and G. oxydans ΔtolB xynA, and yielded about 0.9 and 14 mg/l XynA, respectively. These data clearly indicated the efficient release of the enzyme into the extracellular space by the ΔtolB mutant.

Xylanase activity of purified XynA was assayed with RBB-xylan (Fig. 6). The activity increased linearly with increasing protein concentration from 0 to 141 mU/ml when 0–5 µg of protein was added to the assay (Fig. 6, inset). The specific activity of purified XynA was 33 ± 7 U/mg protein. These data indicate a direct relationship between enzyme activity and protein concentration in the XynA producing G. oxydans ΔtolB strain.

Fig. 6

Activity of purified XynA. XynA activity per ml reaction mixture with RBB-xylan as substrate and different protein amounts (0–5 µg). Assays consisted of 250 µl protein solution and 250 µl 0.1 M acetate buffer pH 6.8 containing 11.5 mg/ml RBB-xylan. (Open square) 5 µg protein; (filled square) 4 µg protein; (open diamond) 3 µg protein; (filled triangle) 2 µg protein; (open triangle) 1 µg protein; (filled circle) 0.5 µg protein; (open circle) no protein. Inset XynA activity expressed as RBB released (U ml⁻¹) with different amounts of protein (µg)

Whole-cell hydrolysis of RBB-xylan

After creation of a leaky outer membrane strain of G. oxydans that is able to secrete active xylanase into the culture supernatant, in vivo experiments were performed to investigate the degradation of xylan under normal growth conditions. G. oxydans xynA and G. oxydans ΔtolB xynA were grown to an OD₆₀₀ of 0.7–1. Cultures mixed with RBB-xylan solution in a 1:1 ratio at 30 °C were assayed temporally. Absorbance at 595 nm was measured after addition of ethanol and centrifugation to separate cells and non-hydrolyzed RBB-xylan. No significant endoxylanase activity was observed in G. oxydans wild type without xynA expression (Fig. 7). In contrast, G. oxydans xynA showed low hydrolysis activity (0.24 mmol h⁻¹ l culture⁻¹). This minor level of activity is likely due to xylanase A released by lysed cells. G. oxydans ΔtolB xynA exhibited a 4.6-fold higher xylan hydrolysis rate (1.1 mmol h⁻¹ l culture⁻¹) compared to G. oxydans xynA. These results demonstrate that a G. oxydans strain with an increased substrate spectrum was generated and is able to use an inexpensive and renewable polysaccharide as substrate for high-value chemical production.

Fig. 7

Whole-cell hydrolysis of RBB-xylan. Time course of RBB-xylan cleavage by G. oxydans wild type (open circle), G. oxydans xynA (filled circle) and G. oxydans ΔtolB xynA (filled triangle). 250 µl aliquots of actively growing cultures (OD₆₀₀ = 0.8–1.0) were incubated in Eppendorf tubes for different time periods at 30 °C in the presence of RBB-xylan. Reaction was stopped by the addition of 1 ml ethanol. After centrifugation the amount of released RBB dye was quantified as described in “Materials and methods”. Total RBB-releasing activity corresponded to 0.24 mmol h⁻¹ L culture⁻¹ for G. oxydans xynA and 1.1 mmol h⁻¹ L culture⁻¹ for G. oxydans ΔtolB xynA

Discussion

G. oxydans strains are generally regarded as safe (GRAS) because they are non-toxicogenic and non-pathogenic. Their physiology is useful for commodity product synthesis because of their natural ability to incompletely oxidize many hydroxylated carbohydrates stereo- and regioselectively [14]. G. oxydans has great biotechnological and industrial applications, but continuing strain improvements and development of advanced expression systems for optimized protein production are required to fully realize its potential. While construction of efficient vectors for heterologous and homologous expression was already successful [21, 61], optimization of industrially used strains for improved application is still an important topic. In particular, the limited substrate spectrum of Gluconobacter spp. needs to be improved to allow a much cheaper and more sustainable production of high-value chemicals by taking advantage of the membrane-bound dehydrogenases of G. oxydans that are able to regio- and stereoselectively produce building blocks for chemical synthesis. Previously, the construction of an expression system was reported that enables G. oxydans 621H to efficiently produce and translocate an active E. coli trehalase (TreA) into the periplasm [25]. This model periplasmic hydrolase allowed growth with trehalose as sole carbon and energy source. Even more interesting than an extension of the substrate spectrum for disaccharides is the creation of G. oxydans strains with the ability to hydrolyze polysaccharides and to use these polymers for the formation of value-added sugar derivatives. One prerequisite for secretion of desired recombinant enzymes is protein transport into the periplasm. A simple method for such transport is the fusion of the desired protein with a suitable signal peptide. For efficient secretory production of recombinant proteins in E. coli the signal sequences of PelB, OmpA and PhoA are applicable [12]. Nevertheless, the efficiency of protein secretion varies depending on host strain, signal sequence and type of protein secreted [12]. In G. oxydans the signal sequences of PelB (pectate lyase B from Erwinia carotovora) and Gox2219 (ribose ABC transporter periplasmic-binding protein from G. oxydans) resulted in the transport of PhoA into the periplasm [25]. In this work the expression system pBBR1p264-SPpelB was used for translocation of PhoA, as a reporter enzyme, and XynA, as suitable exoenzyme into the periplasm of G. oxydans. The PelB signal peptide was used due to its compatibility with the G. oxydans Sec translocation system and the fact that the native signal peptide of XynA was not recognized in E. coli, resulting in a lack of transport through the inner membrane [40].

In Gram-negative bacteria, localization of proteins destined for the outer membrane or extracellular space is a formidable obstacle. This is especially true for G. oxydans because known systems for protein export across the bacterial outer membrane are absent or incomplete [42]. In general, methods for generating a periplasm secretion mutant are based on the destabilization of the outer membrane. Lazzaroni et al. [29] were able to cause such effects in E. coli K-12 with the deletion of the tolA, tolQRA, rfa, pal or lpp genes encoding outer membrane proteins. In other cases, periplasmic secretion mutants were produced by random mutagenesis [28, 31, 57]. Another possibility for the induction of extracellular protein production in E. coli is the addition of glycine or Triton X-100 into the culture medium [59]. Recently, an alternative method for the increase of extracellular release of recombinant proteins in E. coli was published that is based on the coexpression of a cutinase from Thermobifida fusca, which hydrolyzes phospholipids in the cell membrane increasing membrane permeability [58]. In addition to these methods of protein secretion that rely on altered membrane stability, there are targeted methods for excretion of only the desired protein(s). One possibility is secretion using the E. coli hemolysin secretion system HlyABD and TolC [6, 17]. G. oxydans encodes for hlyA (gox0253), hlyD (gox0714) and tolC (gox2487). However, hlyB seems to be absent. Consequently, the α-hemolysin system is likely not functional. Therefore, a secretor mutant of G. oxydans was generated by deletion of gox1687, encoding TolB. This deletion causes secretion of periplasmic proteins across the outer membrane and a higher sensitivity against antibacterial agents in E. coli K-12. Thus, TolB seems to have a stabilizing function in the outer membrane and is a suitable target for generation of a secretor mutant [29]. As indicated above, the deletion of gox1687 in G. oxydans also led to the secretion of periplasmic proteins into the extracellular space.

TolB is part of the Tol–Pal system and is conserved in Gram-negative bacteria. The Tol–Pal system consists of a set of integral cytoplasmic membrane proteins (TolR, TolA, TolQ), the outer membrane protein Pal, and soluble periplasmic linker protein TolB [52]. The Tol–Pal complex has a myriad of functions, for instance colicin import [8], uptake of filamentous phage DNA [13], and the maintenance of outer membrane integrity [29]. We have shown that TolB-deficient G. oxydans formed cell chains, indicating problems with cell division and cell elongation. These effects have been previously described for other TolB-deficient Gram-negative bacteria [16, 32]. However, growth rate and final OD₆₀₀ of G. oxydans ΔtolB were similar to the wild-type strain, making the mutant strain suitable for potential future industrial and biotechnological applications. RNase and BCIP plate assays indicated successfully secretion of native RNase and recombinant PhoA into the extracellular space due to a leaky outer membrane phenotype [9, 24, 28, 57]. Permeability of the outer membrane of G. oxydans ΔtolB was quantified by cell separation and calculation of alkaline phosphatase activity distribution in the cellular fraction and the culture supernatant. Up to 50 % of the PhoA total activity was localized in the culture supernatant in the TolB mutant, while the wild type only showed 3 % of total activity, confirming the highly increased permeability of the outer membrane for periplasmic proteins in the TolB mutant. A similar distribution was observed for the XynA expression strains. While only 5 % of the enzyme was found in the supernatant of G. oxydans xynA, 64 % was detected in the culture supernatant of the ΔtolB mutant. Furthermore, total XynA activity was much higher in the tolB deletion strain compared to the parental strain. Both strains contained the expression vector pBBR1p264-SPpelB-xynA indicating that the expression rate of xynA gene was identical. Therefore, it is tempting to speculate that the XynA protein is slowly degraded in the periplasm while it is stable outside the cell. Since XynA is continuously secreted only by the tolB deletions strain, the enzyme accumulates in the culture supernatant. Interestingly, the ΔtolB mutant revealed a significant shift of XynA activity towards the supernatant fraction in comparison to the PhoA producing ΔtolB mutant. This effect might be based on the much lower molecular mass of XynA (~21 kDa) in comparison to PhoA (~48 kDa). It is known that the periplasmic peptidoglycan layer has a cut-off value of about 50 kDa for the passage of proteins [29]. Furthermore, PhoA is active as a homodimer that is formed in oxidizing environments, such as the periplasm, due to the formation of intramolecular disulfide bonds [54].The dimerization further increases the molecular size and would greatly inhibit diffusion through the peptidoglycan, which might explain the lower secretion efficiency of PhoA in comparison to XynA.

The overexpressed XynA protein contained a C-terminal strep-tag allowing the purification of the protein. After purification from the culture supernatant, the amount of XynA was 15 times higher in the ΔtolB mutant in comparison to the wild type. XynA-like proteins from many organisms were already heterologously produced in E. coli [2, 3]. However, in most cases the xylanases were not transported into the extracellular space and got stuck in the cytoplasm or periplasm. This fact illustrates the general problem for an effective utilization of heterologously produced xylanases in E. coli, which is the cells’ need to be lysed and prohibits continuous degradation of xylan during growth.

The purified XynA protein from G. oxydans allowed enzymatic activity assays with RBB-xylan as substrate, indicating the endohydrolytic activity of the enzyme. While the wild-type strain was not able to hydrolyse xylan, G. oxydans xynA showed low endoxylanase activity that was probably due to cell lysis releasing XynA into the culture supernatant. In contrast, the TolB-deficient strain secreting XynA showed at least a five times higher xylan degradation activity, indicating the feature of this strain for active exoenzyme production and polysaccharide cleavage.

Hemicelluloses are the second most prevalent biopolymer worldwide. Xylan is the major component in hardwood hemicelluloses, making this renewable polysaccharide an excellent model for adapting G. oxydans to various renewable feedstocks for bio-based production of commodity products. XynA from B. subtilis was used as model enzyme in this work because it has been thoroughly characterized and was suitable to the physiological growth conditions of G. oxydans [3, 53]. Furthermore, active recombinant XynA was secreted into the medium of a lpp deficient E. coli strain [3] making the attempt of xylan degradation with G. oxydans promising. A further desirable feature of XynA is that it produces mainly xylobiose, which can be further hydrolysed by to xylose by xylobiases, and minor amounts xylotriose and xylotetraose [3]. There has been an increased interest in xylooligosaccharides due to their potential for application in pharmaceuticals, cosmetics, food products, and agricultural products [1, 56]. Xylobiose is considered a xylooligosaccharide in food products and has been shown to increase the growth of beneficial intestinal flora, such as bifidobacteria and lactobacilli, suggesting its use as a prebiotic [11, 37]. Additionally, xylobiose is approximately 40 % as sweet as sucrose but does not increase blood sugar, indicating it can be used as a low-caloric sweetener for anti-obesity and diabetic diets [36, 38]. However, the production of analytical grade xylobiose is time-consuming and costly [20]. Consequently, a high xylobiose production strain is needed to shorten production time and decrease cost. Unlike natural endoxylanase producers, G. oxydans ΔtolB expressing recombinant XynA from B. subtilis cannot metabolize xylobiose, xylotriose, or xylotetrose. Therefore, this strain might be an ideal host for the production of di-, tri-, and tetra-saccharides from xylan. It is tempting to speculate that other glycosyl hydrolase-producing G. oxydans strains could be engineered for the effective and sustainable synthesis of desirable sugar derivatives using the methods described here. This strategy opens the door for the development of novel and innovative approaches for hydrolysis of polysaccharides and simultaneous production of value-added products.

Acknowledgments

This project was supported in part by funds from the NRW-Strategy project BioSC (project GLUFACT). We thank Dr. Fabian Grein for help with the microscopy pictures and are grateful to Elisabeth Schwab for technical assistance.