One‐step bioconversion of hemicellulose polymers to rhamnolipids with Cellvibrio japonicus: A proof‐of‐concept for a potential host strain in future bioeconomy

The purpose of this study was to evaluate Cellvibrio japonicus as a potential host strain for one‐step bioconversion of hemicellulose polymers to value‐added products. C. japonicus could be cultivated on all main lignocellulose monosaccharides as well as xylan polymers as a sole carbon source. This is particularly interesting as most industrially relevant bacteria are neither able to depolymerize wood polymers nor metabolize most hemicellulose monosaccharides. As a result, lignocellulose raw materials typically have to be degraded employing additional processes while the complete conversion of all lignocellulose sugars remains a challenge. Exemplary for a value‐added product, a one‐step conversion of xylan polymers to mono‐rhamnolipid biosurfactants with C. japonicus after transformation with the plasmid pSynPro8oT carrying the genes rhlAB was demonstrated. As achieved product yields in this one‐step bioconversion process are comparably low, many challenges remain to be overcome for application on an industrial scale. Nonetheless, this study provides a first step in the search for establishing a future host strain for bioeconomy, which will ideally be used for bioconversion of lignocellulose polymers with as little exhaustive pretreatment as possible.


| INTRODUCTION
Lignocellulose is a major resource for a bio-based economy as it is the most abundant biological resource of the planet and not a direct competitor to food production. As the structural framework of woody plant cell walls, it consists mainly of the polymers cellulose, hemicellulose, and lignin (Naik, Goud, Rout, & Dalai, 2010). Cellulose consists of glucose monomers linked by β-1,4 glycosidic bonds and has several established applications such as cellulosic fibers for paper or microcrystalline cellulose for food applications (Nsor-Atindana et al., 2017;Walker, 2006) and bioconversion of its depolymerization product glucose is trivial. Lignin, as a complex macromolecule, is the most important renewable source for aromatic polymers and an important target of research in material science (Upton & Kasko, 2016). Approaches to use microbial conversion of lignin in biotechnological processes will probably not surpass the threshold of feasibility studies in the foreseeable future.
Hemicellulose is the general term for the second polymer in lignocellulose, a group of heteroglycans of several different monomers, such as D-glucose, D-galactose, Dmannose, D-xylose, L-arabinose, as well as sugar acids (Hendriks & Zeeman, 2009). Regarding hemicellulose, xylose is the predominant noncellulosic sugar in hardwoods such as beech, birch or willow and in grasses (Poaceae) such as corn or wheat (Jørgensen, Kristensen, & Felby, 2007). Consequently, xylans are the principal hemicelluloses in these plants (Sjöström, 1993;Willför, Sundberg, Pranovich, & Holmbom, 2005). With a content of 14.9% (total dry weight) in willow (Sassner, Galbe, & Zacchi, 2006) or 18.5% in birch (Hayn, Steiner, Klinger, & Steinmüller, 1993) xylose is an important but mostly underestimated renewable carbon source. It remains mostly unused, as many of the biotechnological important microorganisms do not possess enzymes to break down hemicelluloses such as xylans or to metabolize xylose. Therefore, lignocellulose polymers first have to be degraded by time-consuming and expensive treatments (Hendriks & Zeeman, 2009;Jørgensen et al., 2007;van Dyk & Pletschke, 2012). Due to the reasons listed above, microorganisms which are able to metabolize hemicellulose-related monosaccharides or have the ability to degrade lignocellulose polymers may convey an advantage for efficient bioprocesses. As a huge portfolio of enzymes is necessary, establishing a lignocellulose degrading strain through metabolic engineering is a major challenge. An alternative is the use of organisms which are naturally equipped with a wide range of such enzymes, like the Gram-negative saprophytic soil bacterium Cellvibrio japonicus (former name: Pseudomonas fluorescens subsp. cellulosa). Many studies in the past showed that C. japonicus is able to degrade the main plant cell wall polysaccharides (DeBoy et al., 2008;Gardner, 2016;Gilbert, Jenkins, Sullivan, & Hall, 1987;Hazlewood & Gilbert, 1998;McKie et al., 1997). Previous work demonstrated the genetic accessibility and the possibilities to genetically modify this bacterium (Emami, Nagy, Fontes, Ferreira, & Gilbert, 2002). Expression of recombinant genes in C. japonicus was reported via a conjugation based vector system (Gardner & Keating, 2010).
The purpose of this study was to evaluate C. japonicus as a potential host strain for one-step bioconversion of lignocellulose polymers and its potential application for the production of value-added products using rhamnolipid biosurfactants as an example.

| Chemicals and standards
All chemicals were acquired from Carl Roth GmbH (Karlsruhe, Germany) if not mentioned otherwise. Mono-RL (Rha-C 10 -C 10 ) standard was obtained from Sigma-Aldrich Chemie GmbH (Munich, Germany) and rhamnolipid standard as mixture from Jeneil Biotech Inc. (Saukville, WI, USA).

| Cultivation
For precultures, 25 ml of M9 medium were inoculated with 50 µl of glycerol stock solutions and the main cultures were inoculated with a starting optical density at 600 nm (OD 600 ) of 0.1. The cultivations were performed in 250 ml Erlenmeyer baffled flasks at 30°C and 120 rpm in an incubation shaker (Eppendorf AG, Hamburg, Germany). For storing at -80°C the culture was mixed with glycerol (25% v/v), and frozen in liquid nitrogen.

| Strains and plasmids
Cellvibrio japonicus Ueda107 wild type (formerly classified as Pseudomonas fluorescens subsp. cellulosa) was obtained from the National Collection of Industrial Food and Marine Bacteria NCIMB (Aberdeen, UK) listed under strain number 10,462. For expression of genes rhlAB required for rhamnolipid biosynthesis plasmid pSynPro8oT carrying a tetracycline resistance was used as described previously .

| Cell density
Cell growth was determined by measuring optical density at 600 nm (OD 600 ) using a cell density meter (CO8000, Biochrom Limited, Cambridge, United Kingdom) and the culture was diluted with saline (0.9%) as required.

| Rhamnolipid analysis by highperformance thin-layer chromatography (HPTLC)
Sample preparation and extraction for rhamnolipid determination was performed as described previously (Müller et al., 2010). Briefly, rhamnolipids were precipitated from the cell-free extract using 0.01 vol phosphoric acid, extracted twice with ethyl acetate, evaporated to dryness and resolved in acetonitrile. For quantification of rhamnolipids samples were derivatizated with 2,4´-dibromoacetophenone and trimethylamine according to (Cooper & Anders, 1974) as described previously for HPLC (Müller et al., 2010;Schenk, Schuphan, & Schmidt, 1995). Adjustment for the HPTLC was adapted as follows, the derivatization reagent was composed of a 1:1 mixture of 135 mM 2,4´-dibromoacetophenone and 67.5 mM trimethylamine in acetonitrile. Standard and samples were mixed with derivatization reagent in a ratio 1:10 and incubated at 60°C and 2,000 rpm for 90 min in a thermoshaker. For qualitative evaluation with mass spectrometry (MS), samples were not derivatizated. To localize zones of interest derivatizated samples were applied additionally on the HPTLC plate.
Derivatized samples and standards were applied on 20 x 10 cm plates as 6-mm bands with 8 mm distance from the lower edge and 20 mm distance from the left edge (track distance set to automatic). Application was configured to a filling speed of 15 µl/s and a dosage speed of 150 nl/s. Methanol was used as rinsing solvent. As mobile phase, a mixture of isopropyl acetate/ethanol/water/acetic acid (30:5:2.5:1, v/v/v/v) was used. Before development, chamber saturation was adjusted for 10 min, after development, a drying step was performed for 2 min.
For quantitation, the plate was scanned in the absorption mode at UV 263 nm (deuterium lamp) with a scanning speed of 20 mm/s, a resolution of 100 μm per step and a slit dimension of 3 mm x 0.30 mm. Evaluation was performed with the software winCATS applying "polynomial regression mode."

| Mass spectrometry
Zones of interest from the HPTLC plate were directly eluted via the oval elution head (4 x 2 mm) of the TLC-MS Interface (CAMAG) using methanol/water (9:1, v/v) including 0.1% formic acid at a flow rate of 0.1 ml/min. The eluent was provided by an Agilent (Waldbronn, Germany) 1,100 HPLC pump and the interface was connected to the MSonline. In the tubing from the interface to the MS, a PEEK inline filter with a 0.5 µm frit was integrated. A G1956B MSD single quadrupole MS with a G1946 atmospheric pressure ionization electrospray (ESI) interface was employed (Agilent). The devices were controlled by the software LC/MSD ChemStation B.04.03 (Agilent). For negative ionization, the parameters used were as follows: capillary voltage 4 kV, skimmer voltage 35 V, lens 2.5 V, quadrupole temperature 100°C, drying gas temperature 300°C, drying gas flow rate 10 L/min and nebulizer gas pressure 40 psig. Total ion chronograms were recorded at m/z 100-1,000, using a fragmentor voltage of 100 V, gain 1, threshold 100, and step size 0.1. For evaluation, the spectrum of the plate background at a migration distance comparable to the analyte zone was subtracted from the analyte spectrum.

| Growth of Cellvibrio japonicus on lignocellulose monosaccharides
The wild-type strain of C. japonicus was cultured with the main different lignocellulose related monosaccharides. Growth was observed on all these lignocellulose-derived hexoses: glucose, mannose, and galactose. As delineated in Figure 1 the highest growth was reached using glucose with a maximum OD 600 of 3.2. With mannose, C. japonicus showed significantly slower growth. By adding 0.5 g/ L glucose as "starter" faster growth was observed at the beginning of the cultivation. This is the result of the metabolization of glucose. Furthermore, growth was detected on galactose with a maximum OD 600 of 2.2. Furthermore, C. japonicus was able to metabolize the two hemicellulose pentoses. When xylose was used as a sole carbon source, similar growth behavior as with glucose was noted with maximum OD 600 of 3.1. In contrast, no growth was observed when arabinose was used as a carbon source. However, the addition of 0.5 g/L glucose resulted in cell growth with maximum OD 600 of 2.6. C. japonicus was cultivated using softwood xylans as sole sources of carbon. To account for variability of xylans due to origin from different plants as well as different batches, xylans from maize and beech from different sources were compared. Growth behavior with beech-xylans from Carl Roth GmbH & Co. KG and SERVA Electrophoresis GmbH was quite similar compared to growth on xylose reaching a maximum OD 600 of 2.8 and 2.5, whereas the growth with beech-xylan from abcr GmbH was lower with a maximum OD 600 of 1.8 and in case of corn xylan from Carl Roth GmbH & Co. KG the lag phase was longer (Figure 1).

| Biosynthesis toward rhamnolipid precursor molecules in C. japonicus
The formation of rhamnolipid precursors dTDP-L-rhamnose and 3-(3-hydroxyalkanoyloxy)alkanoic acid (HAA) by C. japonicus was investigated in silico (Table 1) and compared to wild-type rhamnolipid producer Pseudomonas aeruginosa PAO1. A full genome sequence of the applied strain C. japonicus Ueda107 has been deposited to Database Resources of the National Center for Biotechnology Information (NCBI), and all relevant genes have been annotated in the past (NCBI Resource Coordinators, 2017).
However, it should be noted that evidence on protein level is given neither by (NCBI Resource Coordinators, 2017) nor universal protein database (UNIPROT) (The UniProt Consortium, 2017), although both databases list all investigated proteins derived from homology.
Furthermore, the presence of each individual gene from P. aeruginosa PAO1 (NCBI:txid1708767) as a natural producer of rhamnolipids, as well as similarity rating was investigated by BLAST analysis in the genome of C. japonicus Ueda107 (NCBI:txid498211) (Altschul, Gish, Miller, Myers, & Lipman, 1990). Total sequence similarities above 65% were calculated for all relevant genes F I G U R E 1 Cultivation of Cellvibrio japonicus with lignocellulose related hexoses, pentoses and with different xylans (corn from Carl Roth GmbH & Co. KG, beech from Carl Roth GmbH & Co. KG (a), SERVA Electrophoresis GmbH (b) and abcr GmbH (c). Initial sugar concentration was 5 g/L and in case of mannose and arabinose, 0.5 g/L of glucose was added to the respective sugar. * Cultivation was continued until t = 124 hr. OD 600 of mannose as carbon source was 2.0 and for mannose with glucose 2.2 except for dTDP-4-dehydrorhamnose reductase, for which no significant alignment was possible, and beta-ketoacyl-ACP synthase, for which only a truncated fragment was found in the genome (Table 1).

| Heterologous production of rhamnolipids with C. japonicus from xylan polymers
Cellvibrio japonicus was successfully transformed with the pSynPro8oT plasmid carrying the genes rhlA (acyltransferase) and rhlB (rhamnosyltransferase I) required for the biosynthesis of mono-rhamnolipids. The resulting recombinant strain C. japonicus_rhlAB was cultivated using glucose and xylose in comparison with different xylans as carbon sources (Figure 2). The successful production of mono-rhamnolipids was determined in the culture supernatant by high-performance thin-layer chromatography (HPTLC, Figure 3). The usage of monosaccharides as carbon sources resulted in the production of 4.0 mg/L with glucose and 3.6 mg/L with xylose after 48 hr of cultivation. In comparison, a maximum rhamnolipid concentration of up to 4.9 mg/L could be achieved with xylan from beech as a carbon source.
As expected, production of mono-rhamnolipids could be detected in the transformed strain, but not in the negative control without plasmid (Figure 3). Furthermore, two additional spots could be detected in the extract from the transformed strain: one with a hR F value of approx. 25 slightly above di-rhamnolipid (Rha-Rha-C 10 -C 10 ) and another spot with a hR F value of approx. 80. No similarities to common rhamnolipid species from P. aeruginosa regarding separation behavior could be observed using HPTLC. Furthermore, attempts to identify a rhamnolipid species by T A B L E 1 Biosynthesis toward rhamnolipid precursor molecules in Cellvibrio japonicus Ueda107 in comparison with wild-type producer Notes. Annotated sequences and gene names of C. japonicus were extracted from NCBI database. Similarity of each individual gene to Pseudomonas aeruginosa PAO1 (NCBI:txid1708767) as a natural producer of rhamnolipids was investigated by BLAST alignment analysis. a Truncated fragment, calculated for alignment with highest score.

F I G U R E 2 Heterologous production of rhamnolipid with
Cellvibrio japonicus pSynPro8oT_rhlAB with lignocellulose monosaccharides and different xylans. Initial concentration of sugars and polymers was 5 g/L assigning fragment masses in HPTLC-MS spectra did not lead to conclusive results. However, using HPTLC-MS, evidence of formation of Rha-C 10 was detected at an hR F value between 40 and 50, which was not detected in wildtype control samples without plasmid (fragment mass of m/ z 331) (Abdel-Mawgoud, Lépine, & Déziel, 2010;Heyd et al., 2008).

| Hemicellulose as carbon source
Many industrially relevant microorganisms such as Corynebacterium glutamicum and Aspergillus niger have not a high enzyme portfolio in terms of hemicellulose degradation and metabolization. In this study, it is remarkable that the wild type of Cellvibrio japonicus grows on xylose and displays similar growth behavior as with glucose. These results have to be underlined as xylose is the predominant hemicellulose sugar in hardwood and as an example, the total dry weight of willow consists of 14.9% of xylose (Sassner et al., 2006). Furthermore, the wild type of C. japonicus also displays growth on minor hemicellulose components such as galactose. Considering the development of fed-batch processes, which are typically required for high efficient bioprocesses, total consumption of all minor components from crude hemicellulose based substrates is highly favorable. Even though the maximal yield cannot be influenced significantly, this allows to circumvent an accumulation of these components leading to inhibition.
Noteworthy are the results with the hemicellulose polymer xylan as a carbon source. With beech-xylan, similar growth behavior as with glucose or pentoses could be observed. With corn xylan, a longer lag phase could be observed. From the structure, there is a difference between monocotyledons and dicotyledons. Monocotyledons such as corn consist mainly of (glucurono)arabinoxylans, while dicotyledons generally have a significantly lower content of arabinose. It is therefore conceivable that the observed effect of an increased lag phase in case of the culture grown on corn xylan is negatively affected resulting to the more complex enzymatic degradation process required for xylan depolymerization. With regard to these results, C. japonicus is a promising potential candidate for development of a one-step bioprocess. Consolidated bioprocesses (CBP) combine the production of enzymes, saccharification, and fermentation in one step. By harnessing the broad portfolio of extracellular hemicellulose-degrading enzymes, a separate process step comprising the preparation and addition of enzymes can be circumvented, as it is common in simultaneous saccharification and fermentation (SSF) processes.

| Heterologous production of rhamnolipids
Heterologous production of rhamnolipids was reported for different microorganisms with initial concentrations in the range of 10-100 mg/L in the past (Ochsner et al., 1995). Nowadays, a much more efficient production is reported for Pseudomonas putida KT2440 with currently achieved concentrations of about 15 g/L (Beuker, Barth, et al., 2016). With the emerging trend of a bio-based economy, novel microorganisms with a broad metabolic spectrum regarding the utilization of lignocellulose polymers are required. As none of the currently employed microorganisms for rhamnolipid production fulfills these requirements, C. japonicus may provide an alternative.
Achieved concentrations of rhamnolipids as reported in this study can by far not compete with current heterologous high yield strains and processes. However, considering the history of heterologous rhamnolipid production and achieved concentrations this not only outlines the demand for optimization but also the high optimization potential. As a strategy for optimization, a view on the molecular level could be worthwhile. The transcription could be improved by applying other promoters, in particular by F I G U R E 3 HPTLC analysis of rhamnolipid production in C. japonicus carrying plasmid pSynPro8oT_rhlAB. Rhamnolipid standard (left), extract of culture supernatant grown on xylan as carbon source (middle) and negative control without plasmid (right) construction and screening of synthetic promoters optimized for a high efficiency in C. japonicas to circumvent native regulatory systems. Another possibility is to improve the translation efficiency by optimization of the Shine-Dalgarno sequence or the codon usage, as several codons for certain amino acids are more commonly used in some organisms than others (Nakamura, Gojobori, & Ikemura, 2000). A further strategy to increase the amounts of rhamnolipids includes the improvement of educt availability, for example by coexpression of responsible genes for dTDP-Lrhamnose biosynthesis, which seems to be a bottleneck in rhamnolipid biosynthesis (Cabrera-Valladares et al., 2006). Furthermore, it should be noted that the reported rhamnolipid concentrations in this study were obtained from shake flask cultivations; therefore, transfer of the process to a bioreactor with controlled feeding could lead to a signification improvement of the process.
Expression of the rhlAB operon in C. japonicus resulted in the biosynthesis of mono-rhamnolipids. The length of contained 3-hydroxyfatty acids with 10 carbon atoms was expected, as the used rhl-genes originating from P. aeruginosa possess a specificity mainly for these short-chain lengths independent of the physiological background of the host organism (Wittgens et al., 2018) (Figure 3). Furthermore, analysis of fragments in spectra obtained from HPTLC-MS suggests the production of mono-rhamnolipid congener Rha-C 10 . This rhamnolipid congener usually represents only a minority in a mixture of rhamnolipids. As the molecular mechanisms of biosynthesis of Rha-C 10 are not yet fully elucidated (Déziel et al., 1999;Wittgens et al., 2018), the expression system used in this study may be applied as a tool to provide further insight into the mechanisms of biosynthesis for mono-rhamno-mono-lipids. However, it should be noted that obtained signals from HPTLC-MS are comparably weak, yet the fragments could only be detected in extracts from cultures harboring the rhlAB expression plasmid but not in the negative control without plasmid (Supporting Information Figure S1).
In this study, the potential of the nonpathogenic bacterium Cellvibrio japonicus as a host strain for one-step bioconversion of hemicellulose polymers to value-added products was evaluated. The wild-type strain of C. japonicus could be cultivated on all main lignocellulose monosaccharides as well as xylan polymers as sole carbon sources. An example for a value-added product, a one-step conversion of xylan polymers to mono-rhamnolipid biosurfactant with a modified C. japonicus was demonstrated. Achieved rhamnolipid concentrations of 4.9 mg/L as reported in this study are still comparably low, which displays that as a future host strain for bioeconomy, many challenges remain to be overcome. Nonetheless, this study outlines the high potential of C. japonicus and provides a proof-of-concept for one-step production of rhamnolipids on hemicellulose polymers.