Enhanced Production of ALA by metabolic pathway modification
The biosynthetic pathways of TCA cycle in E. coli, the regulations involved, and the strategies for constructing ALA production strain are shown in Fig. 2. The E. coli BWT7-RSA strain was rationally engineered to produce a higher titer of ALA by the following targeted genetic modifications.
HemF, encoded by hemF gene, is a strictly aerobic enzyme that requires molecular oxygen as the electron acceptor and produces hydrogen peroxide. The knock-out of hemF gene would mightily repress the biosynthesis of protoporphyrinogen IX (PPG) resulting in reduced endogenous loss of ALA. The dipeptide binding protein—dppA, actively import ALA through an interaction with the dipeptide inner membrane ATP-binding cassette transporter, DppBCDF, in E. coli [27]. We theorized that inactivation of dppA genes would reduce ALA assimilation and therefore increase ALA accumulation in the medium.
Acetyl-CoA and pyruvic acid are key precursors of TCA cycle. Therefore, we attempted to increase flux of ALA by reducing endogenous loss of these important substrates. There are two isozymes of malate synthase in E. coli which catalyze the Claisen condensation of glyoxylate and acetyl-CoA [28]. Malate synthase A is a key enzyme in the glyoxylate cycle, which catalyzes the irreversible condensation of acetyl-CoA with glyoxylate to produce (S)-malate and coenzyme A [29]. Its encoding gene aceB is located in the aceBAK operon which is transcriptionally regulated by iclR and fadR. Pyruvate oxidase (encoded by poxB gene) is a peripheral membrane enzyme that catalyzes the oxidative decarboxylation of pyruvate to form acetate and CO2 [30]. We believed that knocking out aceB and poxB would improve the flux of acetly-CoA and pyruvic acid, which would lead to an increased accumulation of ALA.
The "Galactose repressor", GalR, is a DNA-binding transcription factor that represses transcription of the operons involved in transport and catabolism of D-galactose [31–33]. Synthesis of these operons is induced when E. coli is grown in the presence of inducer (D-galactose) and the absence glucose [34]. In particular, in the absence of D-galactose, GalR represses the galTKM operon [35]. The galTKM genes could endogenously convert D-galactose to α-D-glucose-6P (Fig. 3). We hypothesized that silence of galR gene can increase glucose flux, thereby positively affecting ALA biosynthesis.
We managed to over-express glk and ppc genes by CRISPR/Cas9 mediated gene knock-in. Glucokinase, encoded by glk gene, phosphorylates glucose which is produced by amylomaltase. Growth on other carbon sources does not appear to affect glk expression. However, in this study we provide glucose as the main carbon source, which will lead to a reduced expression of glk by 50% [36]. In order to improve expression of glk, we try to make an extra clone of glk gene in genome, which was promoted by J23119 promoter (the strongest promoter from constitutive promoter family). Phosphoenolpyruvate carboxylase (Ppc, encoded by ppc gene) is an anaplerotic enzyme that replenishes oxaloacetate in the tricarboxylic acid cycle (TCA). In previous study, researchers reported that overexpression of Ppc improves the growth yield on glucose [37], increases production of succinate from glucose by fermentation [38], and reduces acetate excretion [39]. In this study, we also make an extra clone of ppc gene (induced by J23119 promoter) in genome to better utilize glucose.
After targeted genetic modifications described in this section, the A2 strain (E. coli BW25113-T7 ΔaceB, ΔdppA, ΔhemF, ΔgalR int ppc with J23119, ΔpoxB int glk with J23119) was constructed. Growth of E. coli A2 in different medium was examined to assess whether CRISPR/Cas9-mediated gene knock-out affected the metabolic characteristics of the bacteria. There were no differences in growth rate among two strains (BW25113-T7 and A2) in LB and CAYE medium (Fig. S1), which indicated that metabolic pathway modification based on known metabolic and regulatory information does not impact growth characteristics of E. coli.
To examine the performance of the A2 strain, batch cultures of this recombinant strain were carried out in CAYE medium containing 10 g/L glucose. The final ALA concentrations obtained with this strain were 2,056.1 mg/L (Table. 2). After pathway was modified, the ALA concentration obtained with the A2-RSA (harbored pET28b-LAA and pACYCD-RSA) was 2.27-fold higher than that obtained with the corresponding recombinant BWT7-RSA strain. The ALA yield achieved with A2-RSA (pET28b-LAA, pACYCD-RSA) was as high as 0.206 g of ALA per gram of glucose.
Modified Single-reporter RGMS
Previous studies have indicated that hemL has the potential to integrate multiple signals to regulate ALA biosynthesis. This critical gene encodes Glutamate-1-semialdehyde 2,1-aminomutase (HemL), which catalyzes the pyridoxal 5'-phosphate-dependent transfer of the amino group from C2 of glutamate-1-semialdehyde (GSA) to C1, thereby forming 5-aminolevulinic (ALA) [40]. HemL forms a tight complex with glutamyl-tRNA reductase, the preceding enzyme in the pathway, suggesting metabolic channeling of the highly reactive pathway intermediate GSA [41]. In this study, we chose sYFP (Yellow Fluorescent Protein) as the reporter for RGMS. In addition, hemL was chosen as the biosensor, while its promoter (PhemL) was used to construct the reporter plasmid (pUC57-sYFP-pHemL). To verify the relation between hemL expression and ALA concentration, we constructed pUC57-sYFP-pHemL, in which the promoter of hemL was placed upstream of sYFP. There was a linear correlation between the fluorescent signal and ALA concentration (Fig. 4), revealing that hemL expression is a useful indicator of ALA over-production in our fermentation environment.
In our previous study, we found that overexpression of ALA exporter (such as rhtA, eamA and so on) would accelerate the export and increase the accumulation of ALA. As well, our results revealed that a higher rate of ALA export will increase ALA concentration in supernatant due to greater expression of exporter gene [25, 26]. Hence we choose eamA gene as the target for RGMS, and its CDS was used to construct the mutation plasmid (pSC-LA-eamA).
In our modified single-reporter RGMS design, our object to develop a better mutant of eamA gene with a more efficient rate of ALA export. Thereafter, the mutation plasmid pSC-LA-eamA which contained the original eamA gene was subjected to RGMS as the starting plasmid. A total of four cycles were conducted in this experiment, named as A, B, C and D cycle, respectively. The results of RGMS were shown in Fig. 5.
Figure 5A showed the results of high-throughput screening of reporter genes. It can be observed that the efficiency of fluorescence screening decreases with the increase of RGMS cycles. It is difficult to screen mutants with enhanced fluorescence signals in D cycle. Figure 5B and Fig. 5C showed the results of twice ALA yield screenings. Similar to the results of fluorescence screening, the optimal efficiency of mutation screening effect was obtained in A cycle and the ALA yield increase could reach nearly 40%. However, the screening efficiency decreased dramatically with the increase of RGMS cycles. In D cycle, the ALA yield was only increased by 4% compared with control group. These results indicated that directed evolution of eamA was close to saturation after four cycles of RGMS without sequencing. Thus, we decided to terminate the cycles and select the optimal mutant gene, which was named eamA(C).
CDS repairing of eamA(C)
To find out why the screening efficiency decreased so dramatically, CDS of eamA(C) was sequenced. We found that eamA(C) had a deletion mutation in the base position 322 bp away from the start codon compared with eamA. This deletion mutation resulted in a lack of nearly 180 amino acids in the coding protein (shown in Fig. 6).
However, our result showed that this mutant eamA(C) still led to a good ALA titer. A total of 12 amino acids are mutated in CDS before the deletion mutation. We speculated the ALA efflux capacity of eamA(C) could be enhanced when the deletion mutation was repaired, thus increased ALA production. Based on the above analysis, original eamA gene was used as template to repair the deletion mutation and subsequent CDS of eamA(C). Thereafter, eamA(C1) was obtained (shown in Fig. 6).
We examined the ALA production of these two mutants, eam(C) and eam(C1), in BW25113-T7 (Table. 2 and Fig. 7A). BWT7-LAA(C1) accumulated 17.9% more ALA than BWT7-LAA, which indicated that eamA(C1) has a better excretion efficiency of ALA than eamA (P < 0.001). Then we replaced pET28b-LAA with pET28b-LA-eamA(C1) in E. coli A2-RSA to obtain E. coli A2-RSA-C1. ALA accumulation in recombinant strain A2-RSA-C1 reached 2,471.3 mg/L (Table. 2 and Fig. 7B). These results indicated that eamA(C1) obtained by RGMS and CDS repairing had a greater capacity for ALA excretion and further improved accumulation of ALA in CAYE medium.
Synergetic effect of sodB and katE on ALA production
Although massive efforts have recently been devoted to building microbial producers of ALA through metabolic engineering, few studies focused on the cellular response and tolerance to ALA. An earlier study demonstrated that high concentration of ALA caused severe cell damage and morphological change of Escherichia coli via generating reactive oxygen species (ROS), which were further determined to be mainly hydrogen peroxide and superoxide anion radical [42]. In their study, they found that ALA might cause stress during incubation with E. coli by generating ROS such as H2O2. In addition, porphyrins produced by ALA metabolism have strong photosensitive activities, which could cause serious oxidative damage to bacteria with excessive accumulation. To protect cells from the damage caused by ROS, microorganisms have developed a variety of defense mechanisms against oxidative stress [43]. Enzymatic systems play major roles in the antioxidant mechanisms. Superoxide anion radical is reduced by superoxide dismutase (SOD) to H2O2, and the latter is further degraded to water and oxygen by catalase (CAT) [44]. In this study, we speculated that overexpressing CAT (encoded by katG and katE) and SOD (encoded by sodA, sodB, and sodC) could improve ALA tolerance and lead to an increase of its production level. Thus, we constructed a new plasmid pACYCD-ASK, which overexpressed hemARS, katE and sodB synergistically. This plasmid and pET28b-LA-eamA(C) were both introduced to E. coli A2 strain to obtain A2-ASK. Notably, co-expression of katE and sodB in an ALA synthase expressing strain (A2-ASK) enhanced final ALA titer 9.4% (2703.8 mg/L) in flask fermentation, compared with A2-RSA-C1 (Table. 2). This result demonstrated that reinforcing the antioxidant defense system would hold promise to improve the bioproduction of chemicals that cause oxidative stress such as ALA.
Suppression of hemB gene
ALA accumulation was regulated by feedback inhibition of heme [45, 46]. With the enforcement of ALA biosynthesis, the intracellular heme level may also be elevated [46, 47]. ALAS is the key enzyme for ALA biosynthesis but its activity is commonly inhibited by heme [48]. Therefore, releasing the feedback inhibition of ALA synthesis by heme is expected to further improve ALA production.
Porphobilinogen synthase (HemB) catalyzes the synthesis of porphobilinogen by an asymmetric condensation of two molecules of ALA via a Schiff-base intermediate [49–51]. A previous study suggested that HemB may be feedback-inhibited by protoporphyrinogen IX, an intermediate in the heme biosynthesis pathway [52]. The inhibition of hemB gene was expected to reduce the flux of heme and improve ALA production. However, mutants of hemB had no 5-aminolevulinate dehydratase and extremely low porphobilinogen deaminase activity [53]. What’s worse, hemB mutants do not accumulate porphyrins [54] and are respiration-deficient [55]. In this case, we assumed that replacing the original promoter with a weaker promoter to inhibit hemB gene instead of knocking out it would benefit ALA production.
In this study, we managed to replace PhemB with constitutive promoter (J23108, J23111 or J23116) via CRISPR/Cas mediated gene editing. The expression levels of these constitutive promoter (Table. S1) were acquired from iGEM, which is available at http://parts.igem.org/Promoters. PhemB in genome of E. coli A2 was replaced with J23108, J23111 or J23116. The mutants were named as E. coli 108, E. coli 111 or E. coli 116, respectively. We also found no differences in growth rate among these strains (BW25113-T7, A2, 108, 111 and 116) in LB and CAYE medium (Fig. S1). Subsequently, pET28b-LA-eamA(C1) and pACYCD-ASK were both introduced to E. coli 108, 111 or 116 strain to obtain E. coli 108-ASK, 111-ASK or 116-ASK. Then ALA accumulations of these strains were examined in flask fermentation (Table. 2 and Fig. 8). Out of our expectation, 116-ASK showed the lowest ALA accumulations among these mutants although expression levels of J23116 was weaker than J23111 and J23108. The other two mutants, 108-ASK and 111-ASK, both demonstrated better production of ALA (P < 0.01). The strains 108-ASK and 111-ASK accumulated 20.8% and 27.0% more ALA, respectively than the strain A2-ASK. To further improve ALA production, these three strains (A2-ASK, 108-ASK and 111-ASK) were fermented in 5 L fermenter and glucose was added at initial with a concentration of 20 g/L. After 42 h cultivation, a high titer of ALA (19.02 g/L) with a yield of 0.208 g ALA per g glucose was achieved by 108-ASK (Table. 3 and Fig. 9).
To confirm our assumption, we compared expression levels between these two promoters (PhemB and J23108) by fluorescent reporter gene system. The expression level of J23108 was weaker than PhemB (shown in Fig. S2). These results revealed that replacing the original promoter with a weaker constitutive promoter (J23108) to inhibit hemB gene would improve ALA accumulation.