Growth of S. elongatus cscB+ 2973 and sucrose secretion
S. elongatus UTEX 2973 was engineered to secrete sucrose by expressing the sucrose permease-encoding gene cscB (ECW_m2594) under the strong promoter Pcpc560. Sucrose secretion from S. elongatus cscB+ 2973 is mainly dependent upon the pH and NaCl concentration of the medium, and an alkaline environment was previously reported to be beneficial for sucrose secretion from cyanobacterial cells . We used an alkaline environment (pH≈8.3) with 150 mM NaCl (37℃) to ensure the production and secretion of sucrose from S. elongatus cscB+ 2973 [14, 22]. The sucrose yield and the growth of S. elongatus cscB+ 2973 in different culture media are compared in Fig. 2. The results showed that no sucrose was produced from S. elongatus cscB+ 2973 cells without NaCl in the culture medium. However, sustainable production and secretion of sucrose could be observed for six days when 150 mM NaCl was added, and titers of 612.0 mg/L and 576.5 mg/L sucrose were achieved when S. elongatus cscB+ 2973 was grown in BG-11 and CoBG-11, respectively. To maintain the growth of E. coli in co-culture medium, the effect of different salt concentrations on cell growth was also examined (Suppl. Fig. S1), and the results showed that E. coli was able to grow normally under the tested range of salt concentrations. In this study, a sucrose titer of 576.5-612.0 mg/L (4.00-4.25 mg/L/h) was achieved by S. elongatus cscB+ 2973 cells over six days, which is comparable to the levels observed in similar studies conducted previously (Table 2). For example, although no CO2 aeration occurred during S. elongatus cultivation, sucrose secretion in this study was still higher than the 2.2 mg/L/h value reported in a previous study .
Growth of an engineered E. coli mutant in co-culture medium
To ensure that E. coli BL21 utilizes sucrose as the sole carbon source, we cloned and expressed the essential genes for sucrose metabolism, namely, cscB (ECW_m2594), cscK (ECW_m2595) and cscA (ECW_m2596), into E. coli BL21 to generate an engineered strain, E. coli cscN. In addition, to synthesize 3-HP, the malonyl-CoA reductase-coding gene mcr (Caur_2614) was introduced into E. coli cscN, resulting in the engineered strain E. coli ABKm. A growth comparison of these two strains is shown in Suppl. Fig. S4. As shown, under the same conditions, the final cell density of cscN without the mcr gene was slightly increased. In a previous study, an artificial consortium was constructed by inoculating a heterotrophic bacterium into a S. elongatus PCC 7942 culture with OD750=0.5 . In our study, the sucrose yield of S. elongatus cscB+ 2973 was ~200 mg/L when the cells reached OD750=0.5. Therefore, we selected four concentrations, namely, 50, 100, 150, and 200 mg/L, to examine whether E. coli ABKm could be stably maintained in the system using these levels of sucrose as the sole carbon source in M9 and CoBG-11 media (Fig. 3A and 3B). The growth of E. coli ABKm could be detected under 100, 150, and 200 mg/L sucrose. A previous study showed that the E. coli ΔcscR strain required a minimal sucrose concentration of 1.2 g/L for growth , which is much higher than our result for strain ABKm, suggesting that after expression of cscA, cscB and cscK, the efficiency of sucrose utilization might have improved in strain ABKm [25, 40]. To demonstrate that this effect was not caused by a strain-specific difference, the sucrose utilization pathway was also engineered into E. coli MG1655 and BW25113, and a similar result was observed (data not shown). Additionally, we also determined the 3-HP yield in strain ABKm with the different concentrations of sucrose mentioned above (i.e. 50~200 mg/L) in CoBG-11, and the results showed that strain ABKm was able to produce 3-HP under all the concentrations except 50 mg/L sucrose (Fig. 3C).
Establishing a stable artificial consortium to produce 3-HP
Since the optimal growth temperature for both E. coli and S. elongatus UTEX 2973 is 37℃, we initially set this as the incubation temperature for the co-culture system. However, the analysis showed that E. coli strain ABKm grew poorly after 1~2 days in this system compared with the growth observed in a previous study  (Suppl. Fig. S2). According to the data (Fig. 2D), at 37°C, S. elongatus cscB+ 2973 cells produce a sufficient amount of sucrose, which prompted us to hypothesize that the rapid cell growth of E. coli and utilization of sucrose destroy the balance of the two species in this system. To confirm this hypothesis, we determined the rates of sucrose secretion and sucrose utilization in S. elongatus UTEX 2973 and E. coli, respectively. The results showed that the sucrose utilization rate of E. coli strain ABKm increased gradually with increasing initial sucrose concentration, reaching ~4.20 mg/L/h at an initial sucrose concentration of 200 mg/L with growth at 37℃ for 48 h (Fig. 4B). Although the sucrose secretion rate of S. elongatus cscB+ could reach ~4.11 mg/L/h, we speculated that with the accumulation of E. coli biomass, the sucrose consumption rate of ABKm could be faster than the sucrose secretion rate of S. elongatus cscB+ at 37℃. Therefore, the “production-consumption” balance was disrupted, leading to collapse of the consortium. These results led us to adjust the cultivation temperature from 37℃ to 30℃, aiming to slow down the consumption of E. coli and achieve balanced growth of S. elongatus UTEX 2973 and E. coli in the system. The growth of strain ABKm at 30°C was then observed (Fig. 4A), and the sucrose utilization rate of this strain was determined to be ~2.00 mg/L/h at 30°C at 48 h (Fig. 4B). Interestingly, there was no significant difference between 37℃ and 30℃ in terms of cell growth and sucrose production of the S. elongatus cscB+ 2973 strain (Fig. 4C and 4D). As a result, the artificial consortium with S. elongatus cscB+ 2973 and E. coli strain ABKm was successfully constructed and could be maintained stably for at least 7 days at 30°C (Fig. 4E).
To evaluate the production capacity of the artificial consortium, the 3-HP yield of E. coli strain ABKm was analyzed. As shown in Fig. 5A, 3-HP production reached ~68.29 mg/L in 7 days. In parallel, we also determined 3-HP production in E. coli strain ABKm under pure culture conditions with continuous supplementation of sucrose according to the calculated sucrose secretion rate of S. elongatus cscB+ 2973 (Fig. 5B), and the results showed that the 3-HP yield under pure conditions was at the same level. In addition, we also observed that S. elongatus cscB+ 2973 cultivated in the consortium grew better than the cells cultivated in pure culture conditions (Fig. 6A), consistent with previous findings . In addition, the results showed that almost no free sucrose could be detected in the co-culture medium, suggesting that sucrose produced by the cyanobacterium was completely consumed by the E. coli ABKm strain to support cell growth and accumulate the desired product (Suppl. Fig. S6) .
Effect of oxidative stress on cyanobacteria in an artificial consortium system
ROS are common byproducts of aerobic metabolic processes, such as photoreactions and respiration, in oxygenic photosynthetic organisms , and ROS accumulation could cause oxidative damage to cyanobacterial cells. In addition, previous studies have found that organic buffers in culture media may also contribute to the generation of H2O2 . For example, 1~10 mM 4-(2-hydroxyethyl)1-piperazineethanesulfonic acid (HEPES) in culture medium could produce enough H2O2 to kill Prochlorococcus . Since there was also organic buffer (TES) used to maintain pH in our study, to clarify whether this organic buffer generates H2O2, we determined the titer of H2O2 in blank CoBG-11 under the same culture conditions. The results showed that no H2O2 was detected in blank culture medium, suggesting that the H2O2 in culture medium was mostly synthesized from living cells. Next, we examined the impact of E. coli co-cultivation on the H2O2 level, and the results showed that the H2O2 content was significantly reduced when the heterotrophic partner of E. coli was included in the system (Fig. 6B), which is consistent with a previous study .
To further understand this phenomenon at the molecular level, the expression levels of several H2O2-quenching genes in the E. coli ABKm strain under pure and co-culture conditions were comparatively analyzed by qRT-PCR (Suppl. Table. S1). It is well known that E. coli contains three types of catalases: hydroperoxidase I (HPI) (katG), hydroperoxidase II (HPII) (katF), and hydroperoxidase III (HPIII) (katE) [44-46]. In addition, the synthesis of HPII often increases markedly when cells enter the stationary phase [47, 48]. The transcriptional expression of these three genes was determined. As shown in Fig. 7, the relative expression levels of katG, katF and katE in the E. coli ABKm strain were dramatically upregulated under co-culture conditions compared with those in CoBG-11 under continuous supplementation with sucrose according to the calculated sucrose secretion rate of S. elongatus cscB+ 2973, suggesting that E. coli might be able to remove ROS when co-cultivated with cyanobacterial partners and thus possibly alleviate the overall oxidative stress in the consortium system.