Production of 1,3-propanediol from glycerol using a new isolate Klebsiella sp. AA405 carrying low levels of virulence factors

ABSTRACT Klebsiella pneumoniae is a promising industrial species for the production of chemicals, yet its opportunistic pathogenicity restrains real-world applications. In this work, we identified a novel isolate designated Klebsiella sp. AA405 which carried low levels of virulence factors. No capsule, fimbriae and flagella were observed around colonies. Genome alignment between Klebsiella sp. AA405 and other 186 Klebsiella species revealed 3421 homologous genes. Notably, the endotoxin of Klebsiella sp. AA405 was only half of E. coli BL21. This strain matched well with common cloning and expression vectors and was sensitive to antibiotics. More importantly, this strain metabolized diverse carbon sources. Under micro-aerobic conditions, it produced 51.66 g/L 1,3-propanediol, 3.82 g/L 3-hydroxypropionic acid and 6.86 g/L 2,3-butanediol in a 5 L bioreactor using glycerol as the sole carbon source. Overall these results indicate that Klebsiella sp. AA405 is a promising strain for the production of bulk chemicals.


Introduction
Biorefinery has garnered much attention due to depletion of oil reserves and deterioration of environment. To date, series of chemicals, including citric acid, succinic acid, lactic acid and ethanol, have been produced through large-scale microbial fermentation [1][2][3]. Among recently reported industrial species, Klebsiella pneumoniae is of great attractiveness due to the innate ability to produce chemicals and efficient utilization of glucose and glycerol. When glucose is supplied as carbon source, K. pneumoniae can synthesize a panel of value-added chemicals, including 2,3-butanediol (2,3-BDO), acetoin and pyrroloquinoline quinon (PQQ) [4,5]. The titers and productivities can be enhanced by reprogramming metabolic flux because biosynthesis pathways have been well documented [6,7]. On the other hand, when glycerol is provided as carbon source, K. pneumoniae turns to generate a range of bulk chemicals, including 1,3-propanediol (1,3-PDO), 3-hydroxypropionic acid (3-HP) and 1-butanol [8][9][10]. In K. pneumoniae, glycerol metabolism is mediated by the dha regulon [11]. Compared with glucose-utilizing pathways, glycerol-based biosynthesis is more appealing because of inexpensive glycerol, which is the major byproduct of biodiesel industry [12].
For biological production of 1,3-PDO, 3-HP and 2,3-BDO, K. pneumoniae outperforms E. coli owing to its striking biochemical attributes, including fast growth, extraordinary capability to metabolize glycerol, and particularly the innate ability to synthesize vitamin B 12 , which is the cofactor of glycerol dehydratase [9,11]. Despite the aforementioned advantages of K. pneumoniae, its pathogenicity to human restricts real-world applications [13]. To our knowledge, at least four aspects contribute to pathogenicity. The first is fimbriae and flagella, as both participate in bacterial movement and adhesion to human cells [14][15][16]. The spatial proximity between K. pneumoniae and human cells is necessary for subsequent infection and immunity. The second aspect for pathogenicity is the capsule by which bacteria survive in harsh conditions [17]. The third pathogenic factor is the endotoxin or lipopolysaccharides (LPS) in cell wall. When bacteria are decomposed, LPS is movable and can elicit immune reactions of human cells [18,19]. The fourth is bacterial insensitivity to antibiotics. Namely, bacteria evolve insensitivity to cope with the immune reactions of host cells. The underlying mechanisms are multifaceted, including the recruitment of resistance genes, overproduction of capsule and deficiency of porin proteins in cytomembrane [20]. Presumably, it is hard to circumvent all virulence factors.
In addition to pathogenicity, other aspects also affect industrial applications of K. pneumoniae. The first is the compatibility between K. pneumoniae and expression vectors. The second is the cost of fermentation medium, which is particularly important for large-scale fermentation. Ideally, strains can efficiently metabolize low-cost carbon sources such as xylose, cellulose or even food waste. Besides, strains are expected to grow aggressively on medium because cell growth usually contributes to the production of desired chemicals. Apart from compatibility and medium cost, the inherent capacity of strains to produce wanted chemicals is highly desirable. In this case, gene manipulation of codon optimization is not required. For example, lactobacillus species is regarded as an ideal host strain for the production of lactic acid because it can naturally synthesize a large quantity of lactic acid. Similarly, K. pneumoniae is a promising host for production of 1,3-PDO and 2,3-BDO because it harbours native synthesis pathways. Presumably, if gene transformation is tractable, native host strains might outperform model micro-organisms (e.g. E. coli and Saccharomyces cerevisiae) in overproduction of desired metabolites.
Given that K. pneumoniae is a promising industrial strain despite pathogenicity, we therefore sought to screen a novel Klebsiella species with low pathogenicity. To this end, large amounts of soil samples were collected from relatively uncontaminated sites. Glycerol was added into medium as the sole carbon source to isolate Klebsiella species. After multiple rounds of purification and identification by PCR, a novel Klebsiella species was acquired and the genome was sequenced. Morphological observation with scanning electron microscope (SEM), investigation of the compatibility between vector and recipient host, and assay of endotoxin content were to determine whether the screened species harboured low levels of virulence factors compared with previously reported Klebsiella species. Finally, fermentation using different substrates was to examine the production of bulk chemicals.

Isolation and identification of klebsiella species
All samples were collected from relatively uncontaminated river and soil to acquire a novel Klebsiella species containing low levels of virulence factors. Each solid sample (10 mg) was diluted with 100 mL sterile ddH 2 O in a 1.5 mL Eppendorf tube. After oscillation, Eppendorf tubes were incubated for 10 min, and the supernatant was added into 4 mL glycerol medium at 30 C and shaken at 200 rpm. For liquid sample, 5 mL supernatant was added into 4 mL medium. The glycerol-containing medium included 3.4 g/L K 2 HPO 4 3H 2 O, 1.3 g/L KH 2 PO 4 , 4 g/L (NH 4 ) 2 SO 4 , 0.5 g/L MgSO 4 7H 2 O, 0.1 g/L CaCO 3 and 20 g/L glycerol. Since glycerol was the sole carbon source in medium, only the microbes able to consume glycerol would survive. After successive cultivation of 10 generations, the strains were grown in LB solid medium for isolation. The colonies devoid of visible capsule were retained for further identification through 16S rRNA sequencing. 16S rRNA was amplified with primers 27F and 1492R, and DNA fragment was sequenced in Beijing BioMed Co., Ltd. Multiple sequence alignments were performed using Clustal W program [21]. Phylogenetic tree was constructed using neighbour-joining and maximum-likelihood method in MEGA 5 [22].
SEM of HITACHI S4700 was recruited for morphological observation. The samples were treated according to the reported method [23]. Briefly, samples were fixed by 2.5% glutaraldehyde for 3 h, and washed with sodium cacodylate buffer. After dehydration, samples were subjected to critical point drying and mounted on stubs. After gold coating, the samples were subjected to SEM analysis. To explore genetic background, de novo genome sequencing was carried out in Beijing YuanQua-nYiKe Co., Ltd. The sequencing results were deposited in GenBank.

Testing of sensitivity to antibiotics and compatibility with vectors
Towards industrial applications, we investigated the sensitivity of strain to antibiotics and compatibility with common vectors. Prior to Klebsiella sp. inoculation, kanamycin (25 mg/mL), ampicillin (50 mg/mL), chloramphenicol (34 mg/mL), tetracycline (30 mg/mL) and carbenicillin (50 mg/mL) were individually mixed with LB solid medium. The plates were incubated at 37 C for 24 h. The sensitivity to antibiotics was determined by counting colony number. The LB solid plate without antibiotic was used as the control. Vectors including pUC19, pET-28a, pBR322, pLysS and pCP20 that harbour different replicons and antibiotic markers were electro-transformed into Klebsiella species. The colonies were analysed by PCR and the vectors were extracted. The strains harbouring vectors were considered compatible with the corresponding vectors.
The endotoxin (lipopolysaccharide) content of newly isolated Klebsiella species was measured and compared with that of E. coli BL21 (DE3). Same amount of Klebsiella species and E. coli cells were taken out by measuring OD 600 after cultivation in LB medium for 12 h. The Klebsiella species and E. coli cells were harvested by centrifugation and washed twice with sterile 50 mmol/L PBS buffer at pH 7.0. Next, the cells were sonicated (80 W, working 3 s, pause 2 s, 50 times) and the resultant solution was centrifuged at 17,000 rpm for 15 min. The endotoxin was measured in Beijing Center for Physical and Chemical Analysis using Tachypleus Amebocyte Lysate reaction method.

Carbon distribution analysis
In addition to glucose and glycerol, other inexpensive carbon sources including sucrose, lactose, starch and cracked food waste were individually recruited as carbon sources to examine carbon utilization by newly isolated strain. The strain was grown in LB medium at 37 C for 12 h. One per cent overnighted cells were washed with sterile saline water twice, and incubated in testingmedium. The testing-medium was the same as previous screening medium except the replacement of glycerol by 5 g/L aforementioned carbon sources. The flasks were shaken at 200 rpm and 37 C, and samples were taken out at 24 h.
Glucose, glycerol and sucrose were individually used as carbon sources for cultivating the strain in a 5 L bioreactor. The fermentation medium contained:  [8]. The strain was pre-cultivated in 100 mL fermentation medium overnight at 37 C and then moved to a 5 L bioreactor. The agitation speed was 150 rpm and the air was supplied at 0.4 vvm. The temperature was 37 C and pH value was maintained at 7.0 by adding 5 mol/L NaOH. Initial glycerol concentration was 40 g/L and residual glycerol was maintained at 10 g/L. When glucose or sucrose was applied as carbon source, the initial concentration of carbon source was 40 g/L and further supplementation was not necessary, because glucose and sucrose were consumed less than glycerol. Dissolved oxygen was monitored automatically. Samples were taken out every 6 h to examine cell concentrations and metabolite titers.

Analytical methods
Cell concentrations were measured by using microplate reader at 600 nm with 200 mL fermentation broth added in a cuvette. The metabolites 3-HP, lactic acid and acetic acid were determined by high performance liquid chromatography (HPLC) system (Shimazu, Kyoto, Japan) equipped with a C 18 column and a SPD-20A UV detector at 210 nm. Column temperature was 25 C and mobile phase was 0.05% phosphoric acid at a flow rate of 0.8 mL/min. 1,3-PDO, glycerol, glucose, sucrose and 2,3butanediol (2,3-BDO) were quantitatively analysed by HPLC (Shimazu, Japan) equipped with a column of Aminex HPX-87H Ion Exclusion particles (300 mm £ 7.8 mm, Bio-Rad, Hercules, CA, USA) using a differential refractive index detector. The column was maintained at 65 C and mobile phase was 5 mmol/L sulphuric acid (in Milli-Q water) at 0.6 mL/min. Residual glycerol concentration was measured every 3 h by a titration method with NaIO4 (for control of glycerol). All samples were filtered through 0.22-mm membrane filter.

Isolation and characterization of isolates
To obtain a Klebsiella species with low pathogenicity, the colonies devoid of visible capsule were screened, given that most virulence factors reside in capsule area. Briefly, the colonies manifesting clear boundary were retained and subjected to further identification. After successive purification and characterization, a strain designated Klebsiella sp. AA405 was acquired. Alignment of 16S rRNA sequences revealed high homology between Klebsiella sp. AA405 and K. pneumoniae, and a phylogenetic tree was constructed (Figure 1, Supplementary Table 1). Similar to E. coli BL21, Klebsiella sp. AA405 demonstrated faint colony with clear boundary (Figure 2(A)). For further morphological observation, Klebsiella sp. AA405 was immobilized with glutaraldehyde and analysed with SEM. As shown in Figure 2(B,C), Klebsiella sp. AA405 was a rod-shaped bacterium, and neither fimbriae nor flagella was observed, indicating low pathogenicity relative to species harbouring fimbriae and flagella. Collectively, Klebsiella sp. AA405 is morphologically different from other Klebsiella species, especially the cell surface where virulence factors reside.
Capsule contributes largely to pathogenicity [24]. There are at least three modes of action. The first is bacterial adhesion to human cells, which is mediated by the polysaccharides surrounding the cell membrane [25]. The second is the physical barrier of the capsule, by which bacteria cope with harsh conditions such as antibiotics and phagocyte cells [26]. The third is capsule which serves as an antigen and trigger immunoreaction [27]. Apart from the capsule, both fimbriae and flagella also act as virulence factors and are responsible for the adhesion and motion [14,28]. In fact, capsule, fimbriae and flagella are ubiquitous in pathogenic micro-organisms and make up the majority of virulence factors [28][29][30][31].
To better understand Klebsiella sp. AA405, its genome was sequenced. Results showed that a total of 97 scaffolds contained 5473 542 nucleotides, and DNA GC content was 57.27%. Genome annotation revealed 5001 protein-coding genes and 113 non-coding genes, including 102 tRNA coding genes, five 23S rRNA coding genes, two 16S rRNA coding genes and four 5S rRNA coding genes. All data were deposited in NCBI (BioSample: SAMN04094116; Bioproject: PRJNA295946; WGS: LKKX00000000). Genome alignment between Klebsiella sp. AA405 and 186 Klebsiella genomes was performed via blat program in Python. A total of 3421 homologous genes were found in 187 species. Namely, the 3421 genes were relatively conserved in Klebsiella genera. The 3421 homologous genes accounted for 68.4% of the entire genes in Klebsiella sp. AA405 (Supplementary Table 1). Considering this percentage (68.4%) is higher than that of other species, we speculate that Klebsiella sp. AA405 genome could be minimized.

Plasmid compatibility, antibiotic sensitivity and endotoxin content
For genetically engineered strains, vectors are critical for overexpression of key enzymes and reallocation of metabolic flux. Thus, we investigated the compatibility between Klebsiella sp. AA405 and common vectors. We found that Klebsiella sp. AA405 matched well with vectors pET-28a, pBR322, pUC19 and pLysS, but not with pKD46 and pCP20 (Table 1). This is because the latter  two vectors harbour the temperature-sensitive replicon of plasmid pSC101. The compatibility with common vectors indicates that Klebsiella sp. AA405 could be a genetically tractable host strain. In contrast, the incompatibility with vector pKD46 (carrying lambda Red recombinases) implies that developing Red recombination system in Klebsiella sp. AA405 is challenging.
Considering that antibiotic resistance genes usually reside in vectors, we investigated the sensitivity of Klebsiella sp. AA405 to several antibiotics. As described in Table 1, the growth of Klebsiella sp. AA405 was retarded by common antibiotics including kanamycin, chloramphenicol, tetracycline, ampicillin and carbenicillin. Despite sensitivity to antibiotics, Klebsiella sp. AA405 usually grows faster than E. coli. As shown in Figure 3, the cell dry weight of Klebsiella sp. AA405 in LB medium reached 1.3 g/L at 12 h, which was 30% more when compared to E. coli BL21. This finding implies a shorter fermentation time of Klebsiella sp. AA405 relative to E. coli.
In addition to aforementioned compatibility with vectors and sensitivity to antibiotics, Klebsiella sp. AA405 contained less endotoxin (4.75E+08 Eu/g cell dry weight) than E. coli BL21 (Figure 3). Since E. coli BL21 has been allowed for large-scale fermentation, Klebsiella sp. AA405 should also be permitted for industrial applications upon further genetic modification. Although Klebsiella species harbour endotoxin which may serve as antigen to elicit immunoreaction of host cells [18,19], the expanding synthetic biology methods raise the possibility of attenuating virulence factors.
Fermentation is sometimes hampered by the microbial resistance to antibiotics. The reasons may involve at least three aspects. The first is the antibiotic resistance genes in chromosome or plasmids, which cause microbial insensitivity to antibiotics. For example, b-lactamase confers resistance to penicillin [32]. The second is thick capsule or lipopolysaccharide in bacteria, which intercepts antibiotics on their way to microbial cells [33]. The third is the lack of porin in bacterial plasma membrane. In this case, antibiotics cannot enter bacteria. Relying on these mechanisms, bacteria are resistant to antibiotics and thus survive [34]. For Klebsiella sp. AA405, no capsule was observed around cell surface, and only one b-lactamase gene was found in Klebsiella sp. AA405 genome. The copy number was much fewer than other K. pneumoniae species (Supplementary Table 2). The above findings indicate low pathogenicity of Klebsiella sp. AA405.

Utilization of carbon sources
In this study, glycerol, glucose and sucrose were used as the carbon sources of Klebsiella sp. AA405 and the major metabolites were analysed by HPLC. As shown in Figure 4, Klebsiella sp. AA405 produced 9.69 g/L 1,3-PDO in shake flask when glycerol was carbon source. Interestingly, 1,3-PDO was not detected when glucose and sucrose were provided. This finding may indicate that glycerol instead of glucose or sucrose is appropriate for the biosynthesis of 1,3-PDO. Consistent with this finding, 1,3-PDO was found to be the major metabolite in glycerol reduction pathway [11]. Considering that K. pneumoniae synthesizes 2,3-BDO in glycerol oxidation pathway, we anticipated that Klebsiella sp. AA405 might also   generate 2,3-BDO. As expected, when glucose, glycerol and sucrose were supplied, Klebsiella sp. AA405 produced 5.87, 1.75 and 6.04 g/L 2,3-BDO, respectively. Interestingly, when lactose, starch and food waste were individually used as carbon sources, 1,3-PDO and 2,3-BDO were far less than produced when glycerol, glucose or sucrose was utilized. This might be ascribed to the long pathways from lactose, starch and food waste to desired chemicals 1,3-PDO and 2,3-BDO. Overall these results indicated that Klebsiella sp. AA405 harboured a dha regulon similar to that in K. pneumoniae.
To further validate the presence of dha regulon in Klebsiella sp. AA405, fed-batch cultivation was performed in a 5 L bioreactor (Baoxing, China) containing 3 L fermentation medium. When glucose was used as carbon source, this strain produced 14.37 g/L 2,3-BDO and 8.64 g/L succinic acid at 48 h ( Figure 5(B)). When sucrose was provided, this strain synthesized 12.29 g/L 2,3-BDO ( Figure 5 (C)) and 13.44 g/L lactic acid at 48 h. The lactic acid level was much higher than that of using glucose as carbon source. Remarkably, when glycerol was used as carbon source, this strain produced up to 51.66 g/L 1,3-PDO and 3.82 g/L 3-HP at 48 h ( Figure 5(A)), with 49% of glycerol conversion rate. Fortunately, the byproducts lactic acid, acetic acid and formic acid only accounted for a small portion of the overall metabolic flux. It is clear that glycerol is an appropriate carbon source for the production of 1,3-PDO. Overall, these results indicate that Klebsiella sp. AA405 can utilize diverse carbon sources and guarantees industrial production of chemicals aforementioned.
Although glucose is a commonly used carbon source in fermentation, over-consumption of glucose may cause food crisis. Therefore, alternative carbon sources are required. In this study, Klebsiella sp. AA405 was shown to consume glycerol, glucose, sucrose, lactose and food waste. Since K. pneumoniae can also metabolize glycerol, glucose and sucrose [4,37,39], we reasoned that Klebsiella sp. AA405 might have similar genetic background with K. pneumoniae. Compared with glycerol, other carbon sources led to excessive byproducts including lactic acid and acetic acid. Hence, glycerol seems to be suitable for Klebsiella sp. AA405. Since glycerol is the major byproduct in biodiesel industry, Klebsiella sp. AA405 may be a competitive strain for glycerol-based biosynthesis of 1,3-PDO, 2,3-BDO, 3-HP, polyhydroxyalkanoates, or beyond.

Conclusion
We isolated a novel strain named Klebsiella sp. AA405. No capsule, fimbriae and flagella were observed around bacterial surface. The endotoxin level was only half of that in E. coli BL21. This strain matched well with the common cloning and expression vectors and was sensitive to a series of antibiotics. Importantly, this strain could metabolize different carbon sources. In a 5 L bioreactor, this strain produced 51.66 g/L 1,3-PDO, 3.82 g/L 3-HP and 6.86 g/L 2,3-BDO using glycerol as the sole carbon source. Overall, these results indicate that Klebsiella sp. AA405 is a promising strain for the production of bulk chemicals.

Disclosure statement
None.