Reconstruction of a pathway for 3-HP synthesis via malonyl-CoA by coexpression of MCR-C* and MCR-N
To develop a S. pombe strain capable of malonyl-CoA-mediated production of 3-HP, we introduced two genes: C. aurantiacus mcr, which encodes malonyl-CoA reductase (MCR); and the endogenous acc (cut6) gene, which encodes acetyl-CoA/biotin carboxylase. MCR converts malonyl-CoA to 3-HP via malonate semialdehyde. ACC has a critical role in both malonyl-CoA supply and fatty acid biosynthesis. Earlier work suggested that strains deficient for intracellular proteases may exhibit improved protein expression [26, 27], a property that might also enhance 3-HP production. In the present work, mcr was introduced by complementation of a Leu- auxotrophy with a leu marker and cut6 (expressed under the control of ef1a-c constitutive promoter) was introduced by gene replacement at the protease-encoding isp locus. After a 48-h cultivation in EMM medium supplemented with 50 g/L glucose, a wild-type S. pombe strain did not produce 3-HP to a detectable level, whereas the mcr acc strain (designated SPO-01) generated 3-HP to a titer of 0.016 g/L (Fig. 1a).
The C. aurantiacus MCR is a bi-functional enzyme that catalyzes a two-step reduction from malonyl-CoA to 3-HP. This enzyme can be separated into two fragments, MCR-C and MCR-N (i.e., the C- and N-terminal subdomains). The MCR-C fragment reduces malonyl-CoA to malonate semialdehyde, while the MCR-N fragment reduces malonate semialdehyde to 3-HP [15]. The activities of MCR-C and MCR-N are unbalanced, such that malonyl-CoA reduction by MCR-C is the rate-limiting aspect of the two-step reaction [15]. In E. coli, this imbalance was minimized by increasing the expression ratio of MCR-C/MCR-N after saturation mutagenesis of MCR-C, yielding a mutant enzyme referred to here as MCR-C*. To recreate the example from E. coli, we introduced both fragments [one encoding MCR-N (mcrN) and the second encoding MCR-C* (mcrC*)] into the genome of ACC-expressing S. pombe. The resultant strain generated 3-HP to a titer of 0.18 g/L, which was improved compared to SPO-01 (data not shown). We then introduced an additional copy of mcrC*. The resultant strain (SPO-02) harbors 2 copies of mcrC* under the control of ef1a-c promoter, 1 copy of mcrN under the control of the hsp promoter, and 1 copy of acc under the control of the cam1 constitutive promoter. SPO-02 generated 3-HP to a titer of 0.53 g/L (Fig. 2a), a 30-fold improvement compared to SPO-01. Both strains consumed almost all of the supplied glucose after 24 h, producing 0.5 g/L of acetic acid as a by-product. Introduction of the mcrN and mcrC* fragments did not adversely affect cell growth (Fig. 2d). Introduction of a third copy of mcrC* into SPO-02 did not further enhance 3-HP production (data not shown), suggesting that further improvement of 3-HP titer would require the malonyl-CoA and acetyl-CoA supply.
Improvement of 3-HP production by increasing the supply of acetyl-CoA
To increase the malonyl-CoA supply, we sought to enhance the acetyl-CoA supply. Notably, engineering of the availability of acetyl-CoA has been reported to yield a positive effect on 3-HP production, presumably via the increase in the supply of malonyl-CoA. This strategy has been employed successfully in S. cerevisiae to improve the production of 3-HP [11, 28], butanol [29, 30], and isoprenoids [31, 32]. Specifically, these laboratories enhanced the flux towards acetyl-CoA by co-expression of a Salmonella enterica gene encoding acetyl-CoA synthase harboring a point mutation (referred to here as acsSE*) and a gene encoding an intact aldehyde dehydrogenase (S. cerevisiae ald6). To apply this approach in S. pombe, we introduced acsSE* (under the control of the cam1 promoter) into the protease-encoding atg4 locus of strain SPO-02. We then introduced a copy of the endogenous atd1 gene (a S. pombe homolog of the S. cerevisiae ald6 aldehyde dehydrogenase; also under the control of the cam1 promoter) into the adh4 locus (which encodes an alcohol dehydrogenase) of the resulting strain. We chose to target an alcohol dehydrogenase-encoding gene based on the observation that the loss of alcohol dehydrogenase activity permits accumulation of acetoaldehyde, a substrate of aldehyde dehydrogenase. However, given a previous report [17] that disruption of the major S. pombe alcohol dehydrogenase locus adh1 significantly impaired growth, we chose to target the adh4 locus, which encodes a minor alcohol dehydrogenase.
The resulting SPO-03 strain (which harbors acsSE* and atd1) did not exhibit an improved 3-HP titer (compared to strain SPO-02) (Fig. 3a). However, SPO-03 accumulated acetate to 2.4 g/L after 48 h of cultivation (Fig. 3b). Introduction of additional copies of acsSE* or of the endogenous acs gene (acs1) could not yield improved 3-HP production (data not shown).
Engineering the CoA supply is an alternative to for improving the acetyl-CoA supply. Pantothenate kinase (PTK) is a key enzyme for improving CoA production [33]. For example, the overexpression of mpanK1b (encoding a mouse PTK homolog) yielded a 13-fold increase of the intracellular CoA concentration in mammalian cells [34]. Similarly, a fivefold increase of acetyl-CoA levels (with tenfold higher CoA production) was obtained by overexpression of the endogenous PTK in E. coli [35]. In S. cerevisiae, overexpression of the endogenous PTK in combination with the introduction of acs and ald6 was reported to yield ~3.6-fold increased production of naringenin (a flavonoid) compared to the level obtained upon introduction of acs and ald6 alone [20]. To implement this strategy, we introduced a copy of the endogenous gene encoding pantothenate kinase (ptk1) into SPO-03 by leu auxotrophic marker. The resultant strain (SPO-04) produced 3-HP of 1.1 g/L in after 48 h cultivation (Fig. 3a), a doubling of titer compared to SPO-03. These results indicated that improving the CoA supply is an effective approach for enhancing the production of malonyl-CoA-derived chemicals. Although intracellular amount of acetyl-CoA could not be detected by acetyl-CoA kit (data not shown), acetate accumulation at 48 h was increased from 2.4 g/L (SPO-03) to 4.7 g/L (SPO-04) (Fig. 3b; yellow) in the new strain. The introduction of additional copies of ptk1 did not further improve 3-HP production (data not shown), suggesting that acetyl-CoA is sufficient in strain SPO-04.
Improvement of 3-HP production by controlling the carbon supply
Avoiding ethanol accumulation is a challenge for 3-HP production in budding yeast. Disruption of the ethanol synthesis pathway is a straightforward approach, but halts growth. Alternatively, 3-HP production under carbon-limited cultivation has been demonstrated in S. cerevisiae using feed-in-time medium, which imitates a fed-batch process [12]. This feed-in-time medium contains polysaccharides as well as a degrading enzyme, providing a gradual release of glucose. Here, we utilized our previously described cell surface-display technique for producing BGL on the surface of S. pombe [23]. BGL was expressed as a genetic fusion with an anchor protein, SPBC359.04c [23]. The resulting BGL localizes to the cell surface, where the enzyme hydrolyzes cellobiose into glucose. Therefore, we introduced the gene encoding the BGL-anchor fusion protein into SPO-04. The resultant strain (SPO-05) produced 3-HP at 1.0 g/L starting from 50 g/L of glucose (Fig. 4a), a titer similar to that obtained with SPO-04. That is the expression of BGL had no adverse effect on 3-HP production. When 50 g/L of cellobiose was used as a carbon source, the BGL-displaying SPO-05 strain produced 3-HP at 1.6 g/L after 48 h of cultivation, a titer was 1.5-fold that obtained with glucose. These results showed that providing cellobiose (or the cello-oligosaccharide form) as the carbon source permits an on-going release of glucose by gradual degradation via the cell surface-displayed BGL, and cell surface display was effective for 3-HP production. During growth on glucose, strain SPO-05 consumed 45.3 g/L of glucose after 20 h; almost all glucose was consumed by 24 h. After 20 h of cultivation on cellobiose, cellobiose and glucose were 14 g/L and 7.7 g/L (respectively) in the medium, corresponding to the consumption of 28.3 g/L of cellobiose. The residual amounts of cellobiose and glucose after 24 h of cultivation were 7.2 g/L and 6.0 g/L, respectively. Thus, SPO-05 cultivated in cellobiose-containing medium appeared to maintain a culture medium glucose concentration of approximately 6–7 g/L. Cell growth on cellobiose was slightly slower than that on glucose, presumably due to slower glucose supply. Acetate accumulation was notably decreased (from 4.2 to 2.8 g/L after 48 h) when cellobiose was used as the carbon source. Maintaining a lower glucose concentration in the culture medium by use of BGL display may prevent overflow metabolism, resulting in reducing acetate accumulation [36].
In an attempt to further improve 3-HP production, we introduced additional copies of acc and acsSE* into SPO-05. The resultant strain (SPO-06) produced 3-HP at 1.8 g/L of after 48 h of cultivation, a slight improvement compared to SPO-05. Although SPO-06 exhibited decreased acetate accumulation (1.6 g/L in 48 h cultivation), 3-HP production was not improved. This observation suggested that an improved malonyl-CoA supply was failed to enhance 3-HP production. Therefore, further improvement of MCR activity is desirable.
Re-tuning of MCR-C expression level
MCR activity was improved in the SPO-02 strain, and the acetyl-CoA supply then was further strengthened in the SPO-04 strain. By employing cellobiose as a carbon source with BGL display, 3-HP titer was augmented to 1.8 g/L in the strain SPO-05. Further enhancement of the flux toward acetyl-CoA decreased acetate accumulation, and the resultant strain SPO-06 exhibited a slightly improved 3-HP titer. Remarkably, SPO-06 exhibited a significantly strengthened malonyl-CoA supply compared to that in SPO-02. Hence, we anticipated that the further augment of MCR activity (to utilize the increased malonyl-CoA supply) might further enhance 3-HP production. Hence, additional copies of the mcrC* gene were inserted into the chromosomal sxa2 locus of SPO-06. It is difficult to evaluate the activities of mcrC and mcrN because both enzymes use NADPH as a cofactor. However, enzyme expression was confirmed by SDS-PAGE (Additional file 1: Figure S1A) and qPCR analysis of the resulting strain (SPO-07) revealed that the relative mRNA expression level of mcrC* increased with gene copy number (Additional file 1: Figure S1B). Figure 5 shows that SPO-07, which carries 3 copies of mcrC* in its genome, produced 2.0 g/L of 3-HP from 50 g/L glucose after 48 h of cultivation, a titer twofold higher than that obtained with SPO-05 (1.0 g/L of 3-HP from 50 g/L of glucose). When cellobiose (50 g/L) was used as the carbon source, SPO-07 produced 3-HP at 3.5 g/L after 48 h of cultivation, a titer twofold higher than that obtained with SPO-06 (as described above). This level corresponded to a yield of 0.063 g-3-HP/g-glucose (given that 50 g cellobiose corresponds to 55.5 g glucose upon digestion). Introduction of a 4-copies of mcr-C* or 2-copies of mcr-N did not further increase the 3-HP titer (data not shown).
Fed-batch cultivation using cellobiose as a carbon source
Production of 3-HP under aerobic fed-batch conditions with the limited feeding of cellobiose was evaluated. Specifically, EMM medium supplemented with cellobiose as a carbon source was used. After 24 h of cultivation, when almost all of the original cellobiose was exhausted, further cellobiose feeding was initiated. Under these conditions, strain SPO-07 produced 11.2 ± 0.6 g/L of 3-HP (Fig. 6a) after 96 h of cultivation, corresponding to a yield of 0.12 ± 0.007 g-3-HP/g-glucose. The volumetric production rate in the fed-batch phase was 0.14 g/L/h, corresponding to a specific yield of 1.22 g/g-dry cell weight (DCW). Low levels of acetate and glycerol accumulated under fed-batch culture conditions (Fig. 6b). Glucose concentration increased gradually from the start of feeding, and virtually all of the remaining residual glucose was consumed after feeding was stopped (after 65 h). Notably, the results reported here [titer 11.2 ± 0.6 g/L, yield 12 ± 0.7% (g-3-HP/g-glucose; 1 g cellobiose is equivalent to 1.1 g glucose)] are superior to those previously reported for malonyl-CoA-dependent 3-HP production using S. cerevisiae. When glucose was used as a carbon source, 3-HP production under fed-batch conditions was also evaluated. After 24 h cultivation, when all of the original (20 g/L) glucose was consumed, further glucose (200 g/L of glucose in feeding medium) feeding was started. Strain SPO-07 also produced 9.2 ± 0.9 g/L of 3-HP (Fig. 6c) after 96 h of cultivation, corresponding to a yield of 11 ± 0. 11 g-3-HP/g-glucose. Similarly, low levels of acetate and glycerol accumulated under fed-batch culture conditions (Fig. 6d). During feeding phase, ethanol production was kept at low level in spite of glucose accumulation (Fig. 6c, d). Enhancement of glucose uptake may be one of the ways for improving 3-HP titer, as well as shutoff by-products producing pathways. To our knowledge, this work is the first report of chemical production using beta-glucosidase-displaying fission yeast cells under fed-batch condition with cellobiose or glucose feeding.