De novo biosynthesis of aromatic compounds from carboxymethyl-cellulose by microbial co-culture strategy

Miao Cai Ministry of Education, Nankai University Jiayu Liu Ministry of Education, Nankai University Xiaofei Song Zhejiang University of Technology Hang Qi Ministry of Education, Nankai University Yuanzi Li Beijing Technology and Business University (BTBU) Zhenzhou Wu Ministry of Education, Nankai University Haijin Xu Ministry of Education, Nankai University Mingqiang Qiao (  qiaomq@nankai.edu.cn ) Ministry of Education, Nankai University

Although there are many studies on the production of aromatic compounds, most of them have utilized glucose as the sole or main carbon source. Unlike starch materials, which are costly and occupy limited agricultural land, lignocellulose is the most abundant renewable resource in the world [14,15].
Lignocellulose is environmentally friendly and greatly reduces costs, if the bioconversion of lignocellulose into high value-added products is an e cient process. However, as a biomass with high polymer content, the biodegradation of lignocellulose is restricted unless it is pretreated to reduce cellulose crystallinity and lignin content. Following the advances in metabolic engineering, enzyme engineering, and synthetic biology, several microbial strains were engineered and recombined to possess or enhance the function of cellulose degradation and to further yield industrial products, especially for biofuels. In our previous laboratory studies, we constructed a series of cellulase-expressing yeast strains through a POT1mediated δ-integration strategy to yield bioethanol in a cellulose-based medium [16]. Liu et al. improved the cellulolytic ability of a recombinant S. cerevisiae strain by optimizing the ratio of four cellulases in a cell surface display system, and the ethanol titer was increased to 60% [17]. Gaida et al. achieved the biotransformation from crystalline cellulose to n-butanol by introducing the n-butanol synthesis pathway into the cellulose-degrading bacteria Clostridium cellulolyticum [18].
While, the complex processes of cellulase expression, cellulose decomposition, and fermentation of high value-added products will increase the metabolic burden of a single strain. Therefore, we used a coculture system to achieve the bioconversion of carboxymethyl-cellulose (CMC) to aromatic compounds.
The simultaneous cultivation of two or more cell populations is termed as co-culture technology. Different from natural mixed culture, co-culture technology is used to study interactions between species, generate new products, or improve yield through a purposeful and conscious use of high-throughput technology and bioinformatics platform [19][20][21]. For example, Sun et al. constructed a bacterial co-culture of Rhodococcus sp. WB9 and Mycobacterium sp. WY10; the co-culture showed improved degradation and mineralization of phenanthrene because of the metabolic interactions between these two bacterial species [22]. Oh et al. isolated two new antimicrobial cyclic depsipeptides, namely emericellamides A and B (1 and 2), during co-cultivation of the marine-derived fungus Emericella sp. and the marine actinomycete Salinispora arenicola [23]. The cellulolytic bacterial species Clostridium thermocellum was co-cultured with the butanol-producing strain Clostridium saccharoperbutylacetonicum to achieve the conversion from crystalline cellulose to butanol e ciently [24].
Thus, in the present study, to achieve the biotransformation of CMC to multi aromatic compounds, the coculture system was used that involved the CMC-degrading strain S. cerevisiae SK10-3 [25] and an engineered S. cerevisiae strain NK-B2 [8], which could produce a high amount of tyrosine, a precursor of many aromatic compounds. Here, p-coumaric acid (p-CA) was obtained, which is an important precursor of many avonoids and stilbenes. After a series of optimization, the bioconversion of CMC to p-CA was achieved, and the titer of p-CA was 71.71 mg/L from 30 g/L CMC. Subsequently, the synthesis of caffeic acid in this co-culture system was attempted. Caffeic acid is a natural phenolic compound and can be derived from p-CA. It has received increasing attention because of its many pharmacological activities [26-28]. Moreover, the derivatives of caffeic acid, such as chlorogenic acid [29], rosmarinic acid [30], and caffeic acid phenethyl ester [31], also possess important medicinal values. However, the long growth cycle of plants, complicated extraction procedures, and ine cient puri cation process have severely affected the yield of caffeic acid [32,33]. In the last decade, many studies have reported on heterogeneous biosynthesis of caffeic acid in microorganisms. Rodrigues et al. obtained 280 mg/L caffeic acid from tyrosine in Escherichia coli by using tyrosine ammonia lyase (TAL) from Rhodotorula glutinis and CYP199A2 from Rhodopseudomonas palustris [34]. S. cerevisiae is also an excellent host for the heterogeneous biosynthesis of caffeic acid. Li et al. achieved the de novo biosynthesis of caffeic acid from glucose in a tyrosine pathway-engineered S. cerevisiae [9]. Recently, Zhou et al. maintained a stable production of precursors by the genomic integration of caffeic acid synthetases under the regulation of a modi ed GAL system and further eliminated the feedback inhibition to ensure that a su cient amount of carbon source is directed toward caffeic acid synthesis. Finally, they reported the highest titer of caffeic acid biosynthesis in S. cerevisiae [35]. Besides, Zhang et al. reported the production of caffeic acid from glucose and xylose by engineered E. coli, wherein glucose was the main carbon source [36].
In the present study, 16.91 mg/L caffeic acid was synthesized from a medium containing 30 g/L CMC in a multi-strain co-culture system. Despite the production was still low, the possibility of conversion of lignocellulose to aromatic compounds was con rmed. The present study provided the foundation for the bioconversion of lignocellulose, the most abundant renewable resource in the world, to more value-added compounds.

Co-culture system construction
In previous studies, we performed genetic manipulation on S. cerevisiae wild-type strain BY4741, wherein three codon-optimized cellulase genes encoding Talaromyces emersonii CBHI, Trichoderma reesei EGII, and Aspergillus aculeatus BGLI were integrated into the yeast chromosome by a POT1-mediated δintegration strategy [25]. A series of recombinant strains with high cellulolytic activity to degrade different cellulosic substrates [Avicel, CMC, and phosphoric acid swollen cellulose (PASC)] were screened. To achieve the biosynthesis of high value-added products from lignocellulose, a highly e cient CMCdegrading strain, SK10-3, was chosen for further engineering.
The TAL-encoding gene from Rhodobacter capsulatus was introduced into SK10-3 to synthesis p-CA. However, the degrading pathway of CMC in SK10-3 increased its metabolic burden, and the p-CA titer of SK10-3b were 2.14 mg/L in 20 g/L glucose medium and only 0.46 mg/L in 10 g/L CMC medium, which are much lower than the control strain BY4741b (4.98 mg/L) in glucose medium.
To alleviate the metabolic stress of single strain, the S. cerevisiae strain NK-B2 without an encoding histone H2A gene HTZ1 [8] was selected to introduce multi aromatic compound synthetases and cocultivate with SK10-3. In this co-culture system, CMC was used as the sole carbon source in the medium and was degraded by SK10-3. Following CMC degradation, glucose was released, which was absorbed by SK10-3 and NK-B2, immediately (Fig. 1). To verify whether this strategy is feasible, BY4742a and NK-B2a strains possessing synthases of betaxanthin were rst co-cultured with SK10-3. Due to a longer time period was required for cellulose degradation, we adopted a sequential co-culture strategy. Herein, SK10-3 was rst inoculated into the 10 g/L CMC medium for 24 h. Next, BY4742a or NK-B2a was inoculated into the medium in the ratio of 1:1 to SK10-3. After co-incubation of both strains for 48 h, the fermentation broth turned yellow (Fig. 2a), which indicated that BY4742a and NK-B2a were survived in the co-culture system.
After the above experimental con rmation, further biosynthesis of aromatic compounds from CMC was performed. Similar to the previous experiment, NK-B2b, which contained the TAL gene from R. capsulatus, was inoculated after incubation of SK10-3 for 24 h. During the co-culture process, the growth of both these strains was monitored (Fig. 2b). Although the growth of these strains was limited in the CMC medium, p-CA was detected in co-culture samples (Fig. 2c). In the co-culture system, 3.41 mg/L and 5.93 mg/L p-CA were accumulated by BY4742b and NK-B2b, respectively.

Effect of the inoculum ratio and interval time on p-CA production
Considering that cellulose sacchari cation by SK10-3 is in uenced by incubation time, the changes in glucose content for different inoculum amounts of SK10-3 separately incubated in CMC medium were monitored. As shown in Table 1, the higher the amount of SK10-3 inoculated, the faster was the degradation of CMC. CMC was almost fully sacchari ed after 36 h of incubation, indicating that if the inoculation interval time exceeds 36 h, NK-B2b will not have adequate carbon source for growth. The inoculum ratio is also an important factor to maintain the balance of bacterial growth and product yield in co-culture systems. Thereby, various ratios of SK10-3 to NK-B2b (3:1, 2:1, 1:1, 1:2, and 1:3) and the interval time (0, 12, and 24 h) were investigated simultaneously. The total inoculum OD 600 of the two engineered strains was 0.1. During the co-cultivation period, the growth was recorded and the production of p-CA was detected by HPLC after 120 h of co-culture fermentation (Figs. 3 and 4). Although the biomass was lowest when SK10-3 and NK-B2b were simultaneously inoculated, the highest p-CA titer (29.2 mg/L) was observed when these two strains inoculated at the same time with the ratio of 1:2. Additionally, when the ratio of SK10-3 to NK-B2b inoculated simultaneously was 1:1 or the ratio was 1:3 and the interval time was 12 h, the production of p-CA was considerable (23.4 and 24.3 mg/L, respectively). The glucose content of the co-culture systems was also monitored (Table S1). Table 1 Glucose content during SK10-3 mono-culture in 10 g/L CMC medium with different inoculum doses. To further increase the production of p-CA in the co-cultivation system, the CMC content in the medium was increased. We increased the nal concentration of CMC to 20 and 30 g/L, as adding too much CMC will make the medium thick and almost gelatinous, which is not conducive to medium preparation and fermentation. Subsequently, the three best conditions in the previous result ( Fig. 4) were selected to conduct the co-cultivation experiment in the high-carbon source medium, i.e., SK10-3 and NK-B2b inoculated simultaneously at the ratio of 1:1 or 1:2 or inoculated at an interval time of 12 h at the ratio of 1:3. The growth of these strains and the titer of p-CA were measured simultaneously (Fig. 5).
After 120 h of co-cultivation, the p-CA titer was 46.55 mg/L in 30 g/L CMC medium when SK10-3 and NK-B2b were inoculated simultaneously in the ratio of 1:2 (Fig. 5c), although the lowest level of growth was recorded. This might be probably due to the high consistency of the medium with high CMC content, which limited the e ciency of SK10-3 to degrade CMC. High concentrations of glucose were detected in these co-culture samples after 96 h of culture, and CMC was almost completely decomposed and utilized after 168 h of co-cultivation (Table S2). The p-CA titer after 168 h of fermentation was also monitored, and the results were very optimistic. The highest p-CA titer was detected in the sample of SK10-3 and NK-B2b simultaneously inoculated at the ratio of 1:2. The production of p-CA was increased by 54% (71.71 mg/L) as compared to the titer at 120 h (Fig. 5d). Moreover, to determine the ratio of these two strains after co-culture fermentation, the spotting plate experiment was performed because NK-B2b alone cannot survive in CMC medium. In the optimum co-culture condition, the proportion of NK-B2b was 62% after fermentation.
De novo biosynthesis of caffeic acid from CMC To con rm whether the bioconversion of CMC to more high value-added compounds could be achieved under this co-culture strategy, the production of caffeic acid was assessed; caffeic acid has a high medicinal value and is biosynthesized with p-CA as a precursor. The strain NK-B2c, which possesses the codon-optimized caffeic acid synthase gene HpaB from Pseudomonas aeruginosa and HpaC from Salmonella enterica [35,37] on the basis of NK-B2b, was co-cultured with SK10-3 under the optimum coculture condition we screened before. After 168 h of fermentation, 8.33 mg/L caffeic acid was detected by HPLC (Fig. 6) from 30 g/L CMC medium without any precursor addition. Thus, the de novo biosynthesis of caffeic acid from lignocellulose was achieved.
Improving the production of caffeic acid by a multi-strain co-culture system.
It has been con rmed that the caffeic acid synthases, PaHpaB and SeHpaC, are highly catalytically e cient when expressed in S. cerevisiae [35,37]. However, in the co-culture system we studied, the titer of caffeic acid was only 8.33 mg/L, and there was a large amount of p-CA residues (Fig. 6c). Accordingly, we speculated that under such low-sugar, unfavorable growth conditions, the caffeic acid biosynthesis pathway increases the growth pressure of the NK-B2c strain, whereby limiting the expression of PaHpaB and SeHpaC. Therefore, we also split the caffeic acid biosynthesis pathway into two strains, one is the NK-B2b strain, which only expresses the RcTAL gene, and the other is the NK-B2d strain, which expresses PaHpaB and SeHpaC genes. To alleviate the metabolic stress of NK-B2c, a multi-strain co-culture system was constructed (Fig. 7a). In this system, SK10-3 still accounted for one-third of the total biomass to meet the glucose supply. To nd a balance between the other two strains to maximize the caffeic acid production, we set different inoculation ratios of NK-B2b to NK-B2d (3:1, 2:1, 1:1, 1:2 and 1:3), and detected the nal caffeic acid titers (Fig. 7b). After 168 h fermentation, 16.91 mg/L caffeic acid was accumulated while NK-B2b and NK-B2d was inoculated equally, and the residual amount of p-CA was considerably reduced.

Discussion
The co-culture strategy is a very important method in industrial production and research. This approach usually divides a complete biosynthetic pathway into separate serial modules and introduces these modules into different strains, thereby reducing the metabolic pressure on a single strain [38]. This strategy is highly signi cant in terms of use of cheap substrates, increase in product yield, and development of new substances [39]. Previous studies have reported several applications of cocultivation using different species of microorganisms [40][41][42][43][44]. Several factors such as environmental conditions, nutrient types, and interaction relationships should be considered during the co-cultivation of strains of different species.
In the present study, we rst achieved the bioconversion from lignocellulose to aromatic compounds, like betaxanthin, p-CA and caffeic acid. Here, the co-culture system was utilized to alleviate the metabolic burden of a single strain. In pure culture system, SK10-3b degraded CMC to release glucose as well as fermented to produce p-CA, the p-CA titer of SK10-3b, however, was much lower than the control BY4741b in glucose medium, and there was even almost no p-CA produced in CMC medium. Thus, the co-culture system was considered, we assumed that SK10-3 provides available carbon source for another strains to growth and fermentation (Fig. 1). As shown in Fig. 2a, we can easily see that BY4742b and NK-B2b survived, it indicates that the glucose which SK10-3 released could be absorbed and utilized by other strains.
Obviously, SK10-3 as the carbon source donor strain, the inoculation sequence and ratio are important parameters. However, unexpectedly, the outcome of p-CA was the highest when both engineered strains were inoculated simultaneously (Fig. 4). And the production of p-CA was proportional to the inoculation amount of NK-B2b, except when SK10-3 and NK-B2b were inoculated simultaneously at the ratio of 1:3. This result could be explained by the fact that SK10-3 also absorbed glucose as the carbon source, which consumes part of the glucose pool. The glucose monitoring data shows that glucose will release in large quantities in a short period of time when inoculate more SK10-3 ( Table 2). And the longer the inoculation interval, the less glucose remaining, and the less carbon source NK-B2b can utilize (Table S1). Moreover, when SK10-3 was simultaneously inoculated with NK-B2b, the low concentration of SK10-3 led to the inability to meet the demand of glucose and eventually affected the production. Therefore, nding a balance is crucial. Subsequently, to improve the titer of p-CA in the co-culture system, the nal concentration of CMC was increased to 20 g/L or 30 g/L. And the result was consistent with our objective, the higher concentration of CMC led to more biomass of the strains and higher production; a high titer of p-CA (71.71 mg/L) was obtained after 168 h of co-cultivation in the medium containing 30 g/L CMC. The nal proportion of NK-B2b was 62% after co-culture fermentation, which indicates that NK-B2b predominates in the co-culture system under the optimum condition.
Eventually, the synthesis of caffeic acid in this co-culture system was attempted due to its high medicinal value. In the present study, we co-cultured SK10-3 and a recombinant strain NK-B2c, which carried the caffeic acid biosynthesis pathway, and nally achieved de novo biosynthesis of caffeic acid from CMC without glucose and other precursors added. The titer of caffeic acid was 8.33 mg/L under the optimum co-culture condition we screened earlier. However, such low caffeic acid production was likely due to the metabolic burden of NK-B2c strain. For this reason, we then split the caffeic acid biosynthesis pathway into two strains (NK-B2b and NK-B2d) and constructed a multi-strain co-culture system, to improve the caffeic acid titer (Fig. 7). As expected, the production of caffeic acid was increased, it was 16.91 mg/L, 2.03-fold higher than two strain co-culture system.
Furthermore, considering that many unnecessary pathways consume a large proportion of glucose in this co-culture system, in the future research, an increase in the production of caffeic acid can be achieved through a series of optimization, including weakening the metabolic capacity of SK10-3 to utilize glucose and the ability of producing ethanol to reduce its consumption of glucose. Besides, the fed-batch fermentation method could also be used to optimize the co-culture procedure.

Conclusion
In this study, multi aromatic compounds were biosynthesized from the most abundant renewable and inedible resource lignocellulose via a co-culture strategy. After a series of optimization, the optimum coculture condition was obtained: simultaneous inoculation of SK10-3 and NK-B2b or NK-B2c at the ratio of 1:2, and the multi-strain co-culture system can be used to synthesis more complex compounds. This nding was signi cant for industrial manufacturing, although the nal titer was still low to meet industrial production, the present study has provided the foundation for the application of de novo biosynthesis of a variety of high value-added products from lignocellulose. The co-cultivation strategy also creates more possibilities for synthetic biology research, wherein more than two strains can be cocultured to relieve the metabolic burden and produce multi complexes.

Materials And Methods
Strains, media, and mono-culture conditions The metabolically engineered strains SK10-3 and NK-B2 were constructed in our previous studies [8,25].

Co-cultivation method
Before inoculation into the co-culture system, the strains were incubated in SC medium or drop-out medium for 24 h to prepare seed culture. The seed culture was centrifuged and washed, and the cell pellets were suspended in S solution (0.5% (NH 4 ) 2 SO 4 , 0.17% YNB, and 0.13% amino acid mixture) and inoculated into CMC medium. In the experiment for verifying the effect of the inoculum ratio and interval time on p-CA production, the seed culture of SK10-3 was inoculated rst and mono-cultured for 12, 24, 36, or 48 h, then the NK-B2b seed culture was inoculated with the total inoculum OD 600 of 0.1 and fermented for another 120 h.

HPLC analysis
The aromatic compounds in this study were quanti ed by an HPLC instrument (CoMetro 6000, NJ, USA) equipped with an ultraviolet detector (CoMetro 6000 PVW, NJ, USA) and a C18 column (250 mm × 4.6 mm, 5 μm, Agilent). p-CA and caffeic acid were detected at 310 nm wavelength (Fig 6). A mixture of 5% acetonitrile and 0.1% tri uoroacetic acid in pure water was used as mobile phase A, while 0.1% tri uoroacetic acid in acetonitrile was used as mobile phase B. A sample volume of 10 μL was injected into the detector, and the ow rate was 1 mL/min. The samples were detected under a 35-min gradient program using the following conditions: 6% to 50% phase B for 13 min, 50% to 98% phase B for 13 min, 98% phase B for 3 min, 98% to 6% phase B for 12 min, and washing with 6% phase B for 4 min.

Determination of glucose concentration
During the co-cultivation process, the glucose concentration in the medium was monitored by a glucose assay kit (Solarbio ® BC2500). Samples were collected every 12 or 24 h, and the glucose concentration was then determined according to the manufacturer's protocol. The readings were measured by a UV-vis spectrophotometer (Jinhua 752, Shanghai, China). Every experiment was performed at least three times.
Determination of the ratio of the two S. cerevisiae strains after co-culture fermentation A spotting plate experiment was performed to determine the ratio of SK10-3 and NK-B2b after co-culture fermentation. SK10-3 can grow on SC medium and CMC medium, while NK-B2b can survive only on SC medium. After co-culture fermentation, the yeast culture suspension was diluted and spread on a fresh SC medium plate. After 48 h of incubation, a count of 100~300 colonies per plate was considered to be appropriate. Next, 100 colonies were randomly selected and spotted on a fresh CMC medium plate. Finally, the number of surviving colonies on the CMC plates was counted; these colonies belonged to the SK10-3 strain, and the remaining colonies that did not grow on the CMC plates were of the NK-B2b strain. The ratio of these two strains was then accordingly calculated. Every experiment was performed at least three times. Availability of data and materials All the data and materials supporting the ndings of this article are included within the article and its additional les.
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.

Figure 2
Phenotypes of betaxanthin and p-CA production in the co-culture systems in CMC medium. (a) Comparison of color phenotypes in mono-culture and co-culture systems. (b) Growth curves. (c) p-CA production in mono-culture and co-culture systems. Three replicates of each sample were used.

Figure 3
Growth curves of co-culture systems carried out according to different inoculum ratios of SK10-3 to NK-B2b and inoculation at different interval times. Three replicates of each sample were used.

Figure 4
The p-CA production of co-culture systems carried out according to different inoculum ratios of SK10-3 to NK-B2b and inoculation at different interval times. Three replicates of each sample were used.  HPLC chromatogram of p-CA and caffeic acid. (a) Standards of p-CA and caffeic acid. (b) Co-culture sample of SK10-3 and NK-B2b; peak 1 was p-CA obtained from this co-culture system. (c) Co-culture sample of SK10-3 and NK-B2c; peak 2 was caffeic acid obtained from this co-culture system. Figure 7 (a) Schematic illustration of the multi-strain co-culture system. (b) CA production and p-CA residue after 168 h fermentation in the multi-strain co-culture system with different inoculation ratios of NK-B2b to NK-