Accelerated pentose utilization by Corynebacterium glutamicum for accelerated production of lysine, glutamate, ornithine and putrescine

Summary Because of their abundance in hemicellulosic wastes arabinose and xylose are an interesting source of carbon for biotechnological production processes. Previous studies have engineered several Corynebacterium glutamicum strains for the utilization of arabinose and xylose, however, with inefficient xylose utilization capabilities. To improve xylose utilization, different xylose isomerase genes were tested in C. glutamicum. The gene originating from Xanthomonas campestris was shown to have the highest effect, resulting in growth rates of 0.14 h−1, followed by genes from Bacillus subtilis, Mycobacterium smegmatis and Escherichia coli. To further increase xylose utilization different xylulokinase genes were expressed combined with X. campestris xylose isomerase gene. All combinations further increased growth rates of the recombinant strains up to 0.20 h−1 and moreover increased biomass yields. The gene combination of X. campestris xylose isomerase and C. glutamicum xylulokinase was the fastest growing on xylose and compared with the previously described strain solely expressing E. coli xylose isomerase gene delivered a doubled growth rate. Productivity of the amino acids glutamate, lysine and ornithine, as well as the diamine putrescine was increased as well as final titres except for lysine where titres remained unchanged. Also productivity in medium containing rice straw hydrolysate as carbon source was increased. Funding Information No funding information provided.


Introduction
Lignocellulosic hydrolysates contain glucose as well as significant amounts of the pentoses xylose (5-20%) and arabinose (1-5%) (Aristidou and Penttila, 2000). Lignocellulosic hydrolysates may be obtained from agricultural wastes such as rice straw and are therefore cheap carbon sources. However, lignocellulosic hydrolysates are not fully capitalized on since several industrially relevant microorganisms are not able to utilize pentose sugars as substrates (Jeffries and Jin, 2000). Metabolic engineering of pentose utilization has been successful in some cases, e.g. of Saccharomyces cerevisiae, while in other cases absent or inefficient pentose utilization is still a major bottleneck to be overcome for industrial processes based on lignocellulosic biomass (Aristidou and Penttila, 2000;Becker and Boles, 2003;Karhumaa et al., 2006;Hahn-Hagerdal et al., 2007a,b).

Comparative analysis of recombinant C. glutamicum strains with different plasmid encoded xylose isomerases
Corynebacterium glutamicum possesses a xylulokinase and heterologous production of E. coli xylose isomerase allowed C. glutamicum to grow with xylose, however, the observed growth rates were low (0.09 h -1 ) as compared to growth rates, e.g. with glucose (0.32 h -1 ), ribose (0.23 h -1 ) or acetate (0.28 h -1 ) (Wendisch et al., 2000;Wendisch, 2003;Netzer et al., 2004). In order to test whether xylose isomerase activity is limiting growth with xylose of C. glutamicum recombinants expressing xylA from E. coli, several recombinants were constructed expressing xylose isomerase genes from different sources. The xylose isomerase genes of well-understood model organisms, plant pathogens and strains closely to C. glutamicum related E. coli, Bacillus subtilis, Xanthomonas campestris and Mycobacterium smegmatis, were cloned into the IPTG-inducible expression vector pEKEx3 and transformed into C. glutamicum WT (Table 1C). Xylose isomerase (XI) activity measured as described in Experimental procedures was not detectable in empty vector control strains (< 0.005 U mg -1 ) ( Table 2). High and comparable XI activities were observed in crude extracts of WT(pEKEx3-xylAEc) (0.095 Ϯ 0.010 U mg -1 ) and WT(pEKEx3-xylAXc) (0.090 Ϯ 0.008 U mg -1 ), while about three times less activity was found for WT(pEKEx3-xylABs) (0.023 Ϯ 0.003 U mg -1 ) and WT(pEKEx3-xylAMs) (0.033 Ϯ 0.007 U mg -1 ). While XI activity increase due to overexpression of xylA could be detected in each case, the values are difficult to compare as a single enzyme assay was used without optimization for the enzymes of various origin.

Comparative analysis of recombinant C. glutamicum strains overexpressing endogenous or heterologous xylulokinase genes
Corynebacterium glutamicum WT contains xylulokinase, however, xylulokinase activities determined as described in Experimental procedures were low in crude extracts of C. glutamicum WT, the empty vector control strain and of all recombinants expressing only a heterologous xylose isomerase gene (between 0.013 and 0.024 U mg -1 ) ( Table 2). Ectopic expression xylA from X. campestris was combined either with overexpression of endogenous xylB or with overexpression of xylulokinase genes from E. coli or B. subtilis (Table 1C). Xylulokinase (XK) activity was not increased significantly in strain WT(pEKEx3-xylAXc-xylBBs) ( Table 2). In contrast, ectopic expression of xylB from E. coli and overexpression of endogenous xylB increased XK activity in crude extracts about 25-fold. To test the effect of xylB overexpression in addition to xylA overexpression, growth of C. glutamicum strains overproducing the different XK's along with XI from X. campestris in CgXII minimal medium with 100 mM xylose as sole carbon source was compared (Fig. 1B). The control strain WT(pEKEx3-xylAXc) reached a lower biomass concentration (3.37 Ϯ 0.12 gCDW l -1 ) and grew with a slower growth rate (0.144 Ϯ 0.001 h -1 ) than the strains overproducing XK gene in addition. Corynebacterium glutamicum WT(pEKEx3-xylAXc-xylBCg) grew fastest (0.199 Ϯ 0.009 h -1 ) and reached the highest biomass concentration (4.87 Ϯ 0.53 gCDW l -1 ) followed by WT(pEKEx3-xylAXc-xylBEc) (0.189 Ϯ 0.001 h -1 ; 4.82 Ϯ 0.33 gCDW l -1 ) and WT(pEKEx3-xylAXc-xylBBs) (0.162 Ϯ 0.001 h -1 ; 5.30 Ϯ 0.22 gCDW l -1 ). Thus, heterologous expression of the endogenous xylulose kinase gene from C. glutamicum in addition to the xylose isomerase gene from X. campestris further improved xylose utilization significantly reducing generation times from 4.8 h to 3.5 h. Amino acid and diamine production from xylose by the improved strain Previously, we showed production of amino acids like L-glutamate and L-lysine as well as the diamine putrescine from xylose minimal medium by strains harbouring the basic xylose utilization plasmid pEKEx3-xylAEc (Gopinath et al., 2011). The improved plasmid pEKEx3-xylAXc-xylBCg was transformed into the model lysine producer DM1729, the model ornithine producer ORN1 and the model 1,4diaminobutane producer PUT21. L-glutamate production in CgXII minimal medium with 100 mM xylose as sole carbon source by C. glutamicum WT(pEKEx3-xylAEc) and by WT(pEKEx3-xylAXc-xylBCg) was triggered by ethambutol addition and the improved strain reached higher titres (14.5 Ϯ 0.1 mM as compared with 0.8 Ϯ 0.1 mM) and exhibited an increased productivity (29.7 Ϯ 0.2 as compared with 1.6 Ϯ 0.3 mg l -1 h -1 , Fig. 2). Lysine production by DM1729(pEKEx3-xylAXc-xylBCg) was characterized by a volumetric productivity improved from 25.5 Ϯ 0.8 to 35.4 Ϯ 1.4 mg l -1 h -1 . The volumetric ornithine productivity by ORN1(pEKEx3-xylAXc-xylBCg) was higher than that of the control (43.2 Ϯ 4.3 as compared with 14.8 Ϯ 2.2 mg l -1 h -1 ) and higher ornithine concentrations were achieved (19.6 Ϯ 1.9 mM as compared with 9.4 Ϯ 1.4 mM). Also putrescine production was faster (27.8 Ϯ 2.0 as compared with 15.7 Ϯ 1.2 mg l -1 h -1 ) and titres rose from 12.9 Ϯ 1.0 mM to 15.1 Ϯ 1.1 mM. Taken together, all recombinants carrying the improved plasmid pEKEx3-xylAXc-xylBCg showed significantly increased volumetric productivities in medium with pure xylose.

Amino acid production on rice straw hydrolysate
To characterize L-glutamate and L-lysine production from hemicellulosic hydrolysates in particular rice straw hydrolysate (52 mM glucose, 203 mM xylose, 55 mM arabinose) derivatives of C. glutamicum WT or L-lysine model producer DM1729 harbouring either empty vectors, pVWEx1-araBAD and pEKEx3-xylAEc or pVWEx1-araBAD and pEKEx3-xylAXc-xylBCg were used. The empty vector control utilized glucose for biomass formation and amino acid production, while the pentose-utilizing recombinants grew to higher biomass concentrations and produced more L-glutamate and L-lysine as they utilized arabinose and xylose in addition to glucose (Fig. 3). In case of L-glutamate the empty vector control reached 16 Ϯ 5.4 mM and a volumetric productivity of 98.1 Ϯ 33.1 mg l -1 h -1 in contrast to the pentose-utilizing strain WT(pVWEx1-araBAD)(pEKEx3-xylAEc) with 39 Ϯ 1.9 mM L-glutamate and a productivity of 79.7 Ϯ 3.9 mg l -1 h -1 . The strain improved for xylose utilization reached a comparable level of L-glutamate at 37 Ϯ 5 mM and the highest productivity at 113.4 Ϯ 15.3 mg l -1 h -1 . As expected for growth-coupled L-glutamate production the specific productivities were similar [around 2.6, 3.0 and 3.1 mg gCDW -1 h -1 for the empty vector control, WT(pVWEx1-araBAD)(pEKEx3-xylAEc) and WT(pVWEx1-araBAD) (pEKEx3-xylAXc-xylBCg) respectively].
As observed in the L-glutamate production experiment, product formation and productivity strongly depends on the ability to utilize the pentose fraction of the rice straw hydrolysate due to the expression of araBAD and xylA and/or xylA along with xylB. Therefore empty vector control reached 8.3 Ϯ 1.5 mM L-lysine and a volumetric productivity of 50.6 Ϯ 9.1 mg l -1 h -1 in contrast to the pentose-utilizing strain DM1729(pVWEx1-araBAD) (pEKEx3-xylAEc) with clearly increased 47 Ϯ 5 mM L-lysine and a productivity of 95.4 Ϯ 10.2 mg l -1 h -1 . The strain improved for xylose utilization reached a similar level of L-lysine at 47.5 Ϯ 2 mM together with the highest productivity at 144.7 Ϯ 6.1 mg l -1 h -1 . As L-lysine production was growth-coupled the specific productivities were similar [around 2.6, 3.0 and 3.1 mg gCDW -1 h -1 for the empty vector control, DM1729 (pVWEx1-araBAD) (pEKEx3-xylAEc) and DM1729 (pVWEx1-araBAD) (pEKEx3-xylAXc-xylBCg) respectively].
Engineering for a better use of second-generation feedstock like rice straw hydrolysate, one possible bottleneck regarding catabolism of those carbon sources was successfully dealt with in this work. Further possible bottlenecks are transport of the carbon sources into the cell and the process of breaking the poly-and oligomeric sugars into their monomeric compounds to make them accessible for the producing microorganisms. Concerning the later point in this study a mild sulfuric acid treatment was used to hydrolysate the rice straw (Gopinath et al., 2011). A potential formation of typical fermentation inhibitors, e.g. 5-HMF or weak acids, may result in slower growth and lower production (Palmqvist et al., 1999;Klinke et al., 2004;Heer and Sauer, 2008;Gopinath et al., 2011) and could be an aim to analyse in more detail in future studies for further optimization. However it was already described that in case of ethanol production by growth-arrested cells, typical inhibitors like organic acids, phenolic inhibitors or furans did not substantially disturb C. glutamicum (Sakai et al., 2007). In principal overcoming inhibition can be achieved by different ways, e.g. by simple resistance to the inhibitory substances due to efflux pump or prevention of uptake, by degradation of the relevant substances (Koopman et al., 2010) or by simply not creating inhibitors during processing of the substrates.
Dealing with the potential bottleneck of transport, the xylose and/or arabinose transporting system in the used C. glutamicum wild-type strain ATCC13032 is still unknown and therefore the heterologous expression of xylose and/or arabinose transport systems, e.g. araE (Sasaki et al., 2009), might result in faster substrate uptake and also higher productivity. AraE from C. glutamicum ATCC31831 might be a promising target for transport optimization because this uptake system accepts both arabinose and xylose and allows growth on even very low arabinose or xylose concentrations (Sasaki et al., 2009) and the donor strain is closely related to the used C. glutamicum ATCC13032.
With the potential for further improvements this study has clearly shown the ability of the already industrially intensively used C. glutamicum (Eggeling and Bott, 2005;Wendisch, 2006) to play a key role in utilization of secondgeneration feedstocks with respect to a wide product spectra reaching from products like amino acids to products like fine chemicals as diamines, e.g. putrescine.

Experimental procedures
Microorganisms and cultivation conditions E. coli strain DH5a (Hanahan, 1985) was used for cloning and was cultivated in lysogeny broth medium (LB) (Sambrook et al., 1989). Corynebacterium glutamicum strains used in this work are wild-type strain ATCC13032 (WT) (Abe et al., 1967), L-lysine producing model strain DM1729 (Georgi et al., 2005), L-ornithine producing strain ORN1 and putrescine producing strain PUT21 (Schneider et al., 2012). Pre-cultures of C. glutamicum strains were inoculated from brain heart infusion (BHI) plates into BHI medium. For growth experiments with C. glutamicum 50 ml of BHI overnight cultures were harvested by centrifugation (10 min; 3220 g), washed in CgXII (Eggeling and Bott, 2005), centrifuged again and inoculated in CgXII medium to a final optical density (l = 600 nm) (OD600) of 0.5. All growth and production experiments were carried out with CgXII medium in baffled shake flasks at 30°C and 120 r.p.m. When appropriate 100 mg ml -1 spectinomycin, 25 mg ml -1 kanamycin and 1 mM isopropyl-b-D-thiogalactopyranosid (IPTG) were added to the medium. L-glutamate excretion was triggered by addition of 500 mg ml -1 ethambutol (Radmacher et al., 2005). Ethambutol triggering of glutamate production was preferred over triggering by biotin limitation to avoid effects of residual biotin in hydrolysates. ORN1 was supplemented by addition of 500 mM L-arginine to minimal medium (Schneider et al., 2011). Growth was followed by OD600 determination until the station-ary phase. OD 600 was measured using a UV-1650 PC photometer (Shimadzu, Duisburg, Germany) in dilutions resulting in an OD600 between 0.05 and 0.25. The plasmids used in this study are listed in Table 1B. DNA from of E. coli, B. subtilis, M. smegmatis and X. campestris was prepared by using DNA isolation Kit (Roche, Mannheim, Germany). For overexpression of xylB, the gene was amplified via PCR from genomic DNA of C. glutamicum WT, which was prepared as described previously (Eikmanns et al., 1995).
Genes were amplified by PCR using the oligonucleotide primer pairs xylA_fw_Eco and xylA_rv_Eco, xylA_fw_Bsu and xylA_rv_Bsu, xylA_fw_Msm and xylA_rv_Msm, xylA_fw_Xcc and xylA_rv_Xcc, xylB_fw_Eco and xyl-B_rv_Eco, xylB_fw_Bsu and xylB_rv_Bsu and xylB_fw_Cgl and xylB_rv_Cgl (oligonucleotide sequences are listed in Table 1A). PCR products encoding XylA were cloned blunt into SmaI-restricted vector pEKEx3  and xylB genes were cloned into Ecl136II-restricted vector pEKEx3-xylAXc resulting in the pEKEx3 derivatives listed in Table 1B, pEKEx3 allows IPTG-inducible gene expression. All resulting vectors were sequenced to confirm their sequence integrity.

Enzyme activity measurements
Enzyme activity measurements were analysed in crude extracts of C. glutamicum (Guyer et al., 1981). Cells were inoculated from LB overnight cultures to an OD600 of 0.5 in 50 ml of LB medium containing 1 mM IPTG. Cells were harvested by centrifugation at a final OD600 of 4 and stored at -20°C until use.
Xylose isomerase and xylulokinase activity was measured by the determination of NADH using sorbitol dehydrogenase in the case of xylose isomerase and pyruvate kinase as well as lactate dehydrogenase in case of xylulokinase. Xylose isomerase assays were carried out at 30°C in a total volume of 1 ml containing 100 mM TRIS/HCl, pH 7.5, 10 mM MgCl2, 0.23 mM NADH and sorbitol dehydrogenase (1 U) (Brat et al., 2009), in the case of xylulokinase the assay contained pyruvate kinase (6.8 U), lactate dehydrogenase (9.9 U), 2 mM phosphoenolpyruvate, 0.2 mM NADH, 1 mM ATP, 2 mM MgCl2 and 50 mM TRIS/HCl, pH 7.5 (Eliasson et al., 2000). Tests were started by addition of D-xylose (2 M) or D-xylulose (167 mM) respectively. Enzymatic activities are displayed in mmol min -1 mg -1 , defined as one unit (U). Continuous measurements were carried out using a Shimadzu UV-1650 PC photometer (Shimadzu, Duisburg, Germany).