Exploring functionality of the reverse β-oxidation pathway in Corynebacterium glutamicum for production of adipic acid

Background Adipic acid, a six-carbon platform chemical mainly used in nylon production, can be produced via reverse β-oxidation in microbial systems. The advantages posed by Corynebacterium glutamicum as a model cell factory for implementing the pathway include: (1) availability of genetic tools, (2) excretion of succinate and acetate when the TCA cycle becomes overflown, (3) initiation of biosynthesis with succinyl-CoA and acetyl-CoA, and (4) established succinic acid production. Here, we implemented the reverse β-oxidation pathway in C. glutamicum and assessed its functionality for adipic acid biosynthesis. Results To obtain a non-decarboxylative condensation product of acetyl-CoA and succinyl-CoA, and to subsequently remove CoA from the condensation product, we introduced heterologous 3-oxoadipyl-CoA thiolase and acyl-CoA thioesterase into C. glutamicum. No 3-oxoadipic acid could be detected in the cultivation broth, possibly due to its endogenous catabolism. To successfully biosynthesize and secrete 3-hydroxyadipic acid, 3-hydroxyadipyl-CoA dehydrogenase was introduced. Addition of 2,3-dehydroadipyl-CoA hydratase led to biosynthesis and excretion of trans-2-hexenedioic acid. Finally, trans-2-enoyl-CoA reductase was inserted to yield 37 µg/L of adipic acid. Conclusions In the present study, we engineered the reverse β-oxidation pathway in C. glutamicum and assessed its potential for producing adipic acid from glucose as starting material. The presence of adipic acid, albeit small amount, in the cultivation broth indicated that the synthetic genes were expressed and functional. Moreover, 2,3-dehydroadipyl-CoA hydratase and β-ketoadipyl-CoA thiolase were determined as potential target for further improvement of the pathway. Supplementary Information The online version contains supplementary material available at 10.1186/s12934-021-01647-7.


Introduction
The detrimental impact of petrochemical-based manufacturing on the environment has increased demand for biomass-derived chemicals, fuels, and consumer products [1,2]. Biomass-based derived compounds now include fuels [3], monomers [4], polymers [5], and pharmaceuticals [2], with a few products already exerting a significant impact in the chemical industry [6,7]. Global ongoing efforts are focused on expanding the portfolio, as well as increasing titers and yield.
Adipic acid, an aliphatic dicarboxylic acid, has gained wide attention from the metabolic engineering community as a six-carbon platform chemical for more than two decades [8]. Adipic acid is used as a building block for nylon-6,6, as well as other fibers, polyesters, and resins [9]. Its industrial production volume amounts to nearly 3 million US tons and is achieved via chemical oxidation of petroleum-derived KA oil (cyclohexanone and cyclohexanol) [10]. The petroleum-dependent nature of adipic acid production, has intensified the search for sustainable adipic acid precursors, derivatives, and analogues [11][12][13].

Assessment of non-decarboxylative Claisen condensation of acetyl-CoA and succinyl-CoA in C. glutamicum by overexpressing paaJ from E. coli
The design of synthetic pathway ( Fig. 1) is motived by the similarities found in the bacterial phenylalanine catabolism pathways. Catabolism of phenylalanine in E. coli and styrene in Pseudomonas proceed via phenylacetate degradation [50][51][52]. In the last steps of phenylacetate catabolism, 2,3-dehydroadipyl-CoA is hydrated to 3-hydroxyadipyl-CoA, which is then oxidized to yield 3-oxoadipyl-CoA. Subsequently, succinyl-CoA and acetyl-CoA formed by thiolytic cleavage of 3-oxoadipyl-CoA are assimilated in the TCA cycle. Overexpression of E. coli paaF, paaH, and paaJ genes involved in these steps previously has demonstrated not only the reversibility of the individual enzymes, but also that of the pathway's direction [16,19]. Additionally, CoA-released byproducts of the abovementioned pathway, including 3-oxoadipate, 3-hydroxyadipate, and trans-2-hexenedioate, have been observed in the cultivation broth of E. coli overexpressing paaJ, paaH, and paaF [19]. While reverse β-oxidation may function through multiple cycles [53], PaaJ was shown to stop after the first cycle when overexpressed in E. coli, thus preventing the formation of C8 or C10 compounds [16]. Accordingly, here, E. coli PaaJ was chosen as the first reaction step (Fig. 1).

Demonstration of 3-hydroxyadipate biosynthesis from engineered C. glutamicum
To biosynthesize and detect 3-hydroxyadipate ( Fig. 1), we cloned paaJ, tesB, and a codon-optimized version of E. coli paaH (KEGG ID: JW1390) in the pZ8-Ptac backbone. 3-hydroxyadipate has previously been detected in recombinant E. coli [19]. To ensure sufficient expression of paaJ, tesB, and paaH from a vector-based operon in C. glutamicum [35], the new pZ8-paaH-tesB-paaJ plasmid included an extra tac promoter [55]. The latter was added to the native tac promoter upstream of paaJ to ensure expression of paaJ and paaH. Owing to possible activity of endogenous C. glutamicum thioesterases, the tesB gene accompanied by an additional ribosomebinding site (RBS) was placed downstream of paaH. C. glutamicum transformed with pZ8-paaH-tesB-paaJ reached OD 600 of 24.3 after 72 h of flask-batch Reverse β-oxidation pathway for adipic acid biosynthesis from TCA metabolites. a Synthetic pathway designed to enable adipic acid biosynthesis from succinyl-CoA and acetyl-CoA. Five enzymatic steps, including condensation, reduction, dehydration, second reduction, and hydrolysis of a thioester bond, are carried out by paaJ, paaH, paaF, ter, and tesB gene products encoding the E. coli PaaJ thiolase, PaaH 3-hydroxyacyl-CoA dehydrogenase, PaaF 2,3-dehydroadipyl-CoA hydratase, Ter trans-2-enoyl-CoA reductase, and TesB thioesterase. PcaF corresponds to a putative acetyl-CoA:acetyltransferase. b Vector design to investigate pathway intermediate compounds cultivation (Fig. 2b). MSTFA derivatization of the cultivation broth followed by GC/MS revealed the presence of 3-hydroxyadipic acid among excreted metabolites (Fig. 2c, d). The retention time of trimethylsilyl-3-hydroxyadipic acid was 13.3 min under our experimental conditions. The m/z pattern of the compound matched that of the MSTFA-reacted 3-hydroxyadipic acid standard (Additional file 4: Fig. S2). These findings demonstrated successful heterologous overexpression of paaJ, paaH, and tesB, resulting in production and excretion of 3-hydroxyadipate from engineered C. glutamicum. Moreover, the presence of 3-hydroxyadipic acid in the cultivation broth indirectly confirmed the correct functioning of PaaJ and condensation of acetyl-CoA and succinyl-CoA in engineered C. glutamicum. Thus, overexpression of PaaJ in C. glutamicum allows manipulation of the thermodynamic equilibrium between acetyl-CoA, succinyl-CoA, and 3-oxoadipyl-CoA, in spite of endogenous β-ketoadipyl CoA thiolase (PcaF), which is thought to promote the catabolic direction of the pathway [56].
The engineered C. glutamicum strain expressing the four genes reached OD 600 of 24.8 after 72 h shake-flask cultivation (Fig. 3a). Concentrating the cultivation broth 5-fold using solid-phase extraction followed by LC/MS revealed presence of trans-2-hexenedioic acid (Fig. 3b). C. glutamicum strain harboring pZ8-paaH-tesB-paaJ did not produce detectable amount of trans-2-hexenedioic acid (Fig. 3b). The results indirectly demonstrated expression of paaJ, paaH, paaF, and tesB genes despite the operon organization with placing paaF and tesB at the downstream of each operon (Fig. 1). Downstream genes are known to be expressed weaker than the upstream genes in operon configurations [57]. However, the results also suggested that either expression of operons was not optimal for downstream genes or there was a bottleneck in the pathway.

Implementation of full-length reverse β-oxidation in C. glutamicum
Addition of Ter from T. denticola [48] completed the reverse β-oxidation pathway for adipic acid production in C. glutamicum (Fig. 1). Overexpression of ter from T. denticola (KEGG ID: TDE0597) was shown to reduce trans-2-enoyl-CoA in engineered E. coli [16]. To express codon-optimized paaJ, paaH, paaF, ter, and tesB on a single vector, the genes were reorganized in two separate operons, each driven by a tac promoter and constructed in such a way that paaF was expressed after paaJ and ter after paaH. Expression of tesB was again given the lowest priority and was placed in the last position of the second operon, downstream of paaF. Translation of paaF, ter, and tesB was enabled by each accompanying synthetic RBS. C. glutamicum transformed with the resulting pZ8-paaH-ter-paaJ-paaF-tesB plasmid reached OD 600 of 25.2 after 78 h of flask-batch cultivation (Fig. 4a). MSTFA derivatization of the cultivation broth followed by GC/ MS revealed the presence of 3-hydroxyadipate (Fig. 4b, c) but not trans-2-hexenedioic acid or adipic acid. However, concentrating cultivation broth 5-fold followed by LC-MS revealed presence of 37.3 µg/L of adipic acid ( Fig. 4d). Analysis of LC-MS/MS further confirmed the identity of adipic acid from the cultivation broth (Additional file 5: Fig. S3). The MS/MS spectra obtained was also in agreement with publicly available database entry (HMDB0000448) [58]. C. glutamicum harboring an empty vector also reached OD 600 of 25.2 (Fig. 4c), and no intermediate compounds were detected in the control strain (Fig. 4b-d). The presence of an intermediate and adipic acid confirmed that even when tesB expression was given the least priority, release of CoA was not hindered by the pZ8-paaH-ter-paaJ-paaF-tesB construct. A possible reason for this is that native thioesterase (i.e., NCgl1600) activity was strong enough to release CoA from 3-hydroxyadipyl-CoA or that expression and functionality of tesB were achieved. However, these results also pointed to the need for a new strategy to assess and improve gene expression of the attempted synthetic pathway.

Discussion
In this study, we aimed to investigate the reverse β-oxidation pathway and its functionality in engineered C. glutamicum. In the final strain, detection of 3-hydroxyadipic acid as major byproduct, small amount of adipic acid, and concomitant absence of trans-2-hexenedioic acid pointed to inefficient PaaF activity, although Ter functionality should be assessed separately as well (Figs. 3, 4). Such pattern indicated that the dehydration of 3-hydroxyadipyl-CoA was the bottleneck step. The paaF and ter genes were each placed downstream in the respective operons (Fig. 1), since placing the upstream pathway genes in the upstream of operon organization has proven advantageous [34,35]. This also might have led to decreased expression of paaF. Hence, a possible strategy to overcome the expression challenge is to switch the operon organization or to place additional promoters for improved paaF and ter expression [57]. A potential strategy to overcome dehydration step is to use another enzyme, such as Ech (enoyl-CoA hydratase) from Ralstonia eutropha, Crt (crotonase) from Clostridium acetobutylicum [14], or Tfu_0067 from T. fusca [15], as it might be that E. coli PaaF is simply less compatible with C. glutamicum for some unknown reason. Lack of 3-oxoadipic acid in the cultivation broth can be potentially explained by the presence of PcaF, which is encoded by cg2625 (EC:2.3.1.9) and involved in degradation of 3-oxoadipic acid [59]. Although paaJ overexpression might have allowed an equilibrium between 3-oxoadipyl-CoA, acetyl-CoA, and succinyl-CoA, PcaF could have represented a potential source of disturbance by reversing the direction of the chemical reaction. We speculate that PcaF acts against PaaJ to actively degrade 3-oxoadipyl-CoA. Furthermore, because 3-oxoadipic acid synthesized in the presence of PaaJ and TesB can be reverted to 3-oxoadipyl-CoA by PcaIJ [59], chromosomal deletion of pcaIJ and pcaF appears necessary to prevent degradation of 3-oxoadipyl-CoA and 3-oxoadipic acid.
Once bottleneck step of the dehydration of 3-hydroxyadipyl-CoA is overcome, further strategies to improve adipic acid production might include expression optimization by engineering the reductive branch of TCA cycle [60], RBS switching [61], alternating reducing equivalents between NADH + and NADPH [62], media engineering [63], eliminating competing pathways [64], using a different thioesterase for CoA removal [24], and removing any pathways leading to the formation of byproducts [65]. A switch in reducing equivalent can be achieved using trans-2-enoyl-CoA reductase (E.C. 1.3.1.38), which accepts NADPH, an abundant cofactor in C. glutamicum. Competing reaction can be eliminated by deleting pcaF, because the reverse degradation flux must be engineered to overcome the catabolic step [66]. Possible strain and pathway engineering strategies to improve titer and productivity have been discussed previously [67][68][69].
C. glutamicum offers various benefits, such as elevated tolerance to acids, strong TCA flux, and natural excretion of acetic and succinic acid upon metabolic stress. Its use of acetyl-CoA and succinyl-CoA in the initial steps was hypothesized to favor adipic acid production. However, while condensation of acetyl-CoA and succinyl-CoA was indeed successful, directing the flux toward adipic acid proved either impossible or insufficient. Hence, several Fig. 4 Shake-flask cultivation profile of C. glutamicum harboring pZ8-paaH-ter-paaJ-paaF-tesB. a Cultivation parameters include OD 600 (filled circles) and residual sugar (empty circles). N = 3; error bars = standard deviation. b Ion (m/z = 363) extracted chromatogram obtained by GC/MS of the cultivation broth (t = 55 h) of C. glutamicum harboring pZ8-paaH-ter-paaJ-paaF-tesB (black line) and an empty vector (grey line). The peak corresponding to 3-hydroxyadipate is indicated with a black arrow. c Cultivation profile of C. glutamicum harboring pZ8-Ptac showing OD 600 (filled circles) and residual sugar (empty circles). N = 3; error bars = standard deviation. (d) Ion (m/z = 145.0506) extracted chromatogram obtained by LC/ MS of adipic acid standard (purple), cultivation broth of C. glutamicum harboring pZ8-paaH-ter-paaJ-paaF-tesB (green), pZ8-paaH-tesB-paaJF (grey), and an empty vector (black) challenges need to be overcome when engineering C. glutamicum for adipic acid production.

Conclusions
The present study demonstrated the feasibility of the synthetic reverse β-oxidation pathway by showcasing adipic acid biosynthesis in C. glutamicum. This work laid the groundwork for future development of C. glutamicum as a microbial factory for adipic acid production. For the subsequent studies, metabolic flux should be engineered to overcome the present bottleneck represented by dehydration of 3-hydroxyadipyl-CoA, while chromosomal deletion of catabolic genes will bring further advantages.

Strains and plasmids
Primers (Additional file 6: Table S3) used in this study were synthesized at Eurofins Genomics (Ebersberg, Germany), Thermo Fisher (Waltham, MA, USA) or Sigma-Aldrich (St. Louis, MO, USA). Codon-optimization for C. glutamicum (Additional file 2: Table S2) and subsequent gene synthesis were carried out at GenScript (Piscataway, NJ, USA). All constructed vectors were sequence-verified at Eurofins Genomics or Macrogen Europe (Amsterdam, the Netherlands). Plasmid pZ8-Ptac (Addgene plasmid number 74,064) was a gift from Timothy Lu (MIT, Cambridge, MA, USA) [54]. General PCR conditions were as follows: 95 °C for 5 min; 28 cycles at 95 °C for 30 s, 55 °C for 30 s, 72 °C for 1 min; and a final extension at 72 °C for 7 min. The PCR was carried out with Phusion High-Fidelity DNA Polymerase (Thermo Fisher) in a total reaction volume of 50 µL. All restriction enzymes were purchased from Thermo Fisher. E. coli NEB ® 5-alpha and NEB ® 10-beta from New England Biolabs (Ipswich, MA, USA) were used for general cloning purposes. The C. glutamicum ATCC 13,032 strain used in this study is a lab stock. Chemicals used in this study were purchased from Merck (Kenilworth, NJ, USA) or Thermo Fisher.
pZ8-paaJ-tesB was constructed by cloning codon-optimized paaJ and tesB into the pZ8-Ptac vector. Primers 134_F_EcoRI and 133_i3R_KpnI v2 were used to amplify paaJ and the product was digested with EcoRI/KpnI. Primers tesB_F_KpnI and tesB_R_BamHI were used to amplify tesB and the product was digested with KpnI/BamHI. pZ8-Ptac was digested with EcoRI/BamHI and subsequently three-way ligated with the two PCR products. A 16-bp RBS (tttcacacaggaaaca) from the synthetic lacUV5 promoter was placed between paaJ and tesB for tesB expression during this step.
Construction of pZ8-HB was carried out by replacing ter from pZ8-Hr with tesB. Codon-optimized tesB was PCR-amplified with primers tesB_F_KpnI and tesB_R_BamHI, and then digested with KpnI/BamHI. This fragment was ligated into pZ8-Hr previously digested with KpnI/BamHI to replace ter. An extra promoter and paaJ were amplified by primers PA146F_Gib and PA146R_Gib using pZ8-paaJ as template, and then inserted into pZ8-HB by Gibson assembly to yield pZ8-paaH-tesB-paaJ.
Vector pZ8-paaH-ter-paaJ-paaF-tesB was constructed in a stepwise manner. A codon-optimized version of E. coli paaH was PCR-amplified using primers 133_i1F_EcoRI and 133_i1R_KpnI (Additional file 6: Table S3), and then digested with EcoRI and KpnI. A codon-optimized version of T. denticola ter was PCR-amplified using primers 133_i2F_KpnI and 133_ i2R_BamHI and digested with KpnI and BamHI. The digested PCR products were three-way ligated with EcoRI/BamHI-digested pZ8-Ptac to yield pZ8_Hr. A 16-bp untranslated region (UTR) sequence (tttcacacaggaaaca) [55] including an RBS was introduced between the two genes for codon-optimized ter expression. To construct pZ8_HrJF, an extra tac promoter, codon-optimized paaJ, and paaF were introduced into pZ8_Hr. The codon-optimized version of paaJ was PCR-amplified using primers pre132_i3F_BamHI and 133_i3R_KpnI v2, and subsequently digested with BamHI/KpnI. The codon-optimized paaF fragment (PCR-amplified with primers 133_i4F_KpnI and 133_ i4R_PstI v2) was digested with KpnI/PstI. Digested paaJ and paaF fragments were subsequently threeway ligated into pZ8_Hr previously digested with BamHI/PstI. As with pZ8_Hr, an additional 16-bp UTR (tttcacacaggaaaca) with RBS was introduced at this point to allow for paaF expression. Thus, an extra promoter was PCR-amplified with primers tac_BamHI_F and tac_BamHI_R, digested with BamHI, and then ligated upstream of paaJ at the BamHI site. The resulting plasmid was named pZ8_HrJF. For pZ8-paaH-ter-paaJ-paaF-tesB construction, the tesB gene was added to pZ8_HrJF. Codon-optimized tesB was PCR-amplified using primers 133_i5F_PstI and 133_i5R_PstI, and then digested with PstI. The resulting fragment was ligated into pZ8_HrJF at the PstI site downstream of paaF.
pZ8-paaJ was constructed by removing tesB from pZ8-paaJ-tesB. This was achieved by digesting pZ8-paaJ-tesB with KpnI and BamHI, followed by ligation, whereby a short linker was placed at the 3′ end of paaJ as KpnI and BamHI were not compatible. The linker consisted of oligos linker_sense and linker_antisense (Additional file 6: Table S3). This step introduced also an XbaI site between KpnI and BamHI for potential future usage.

Media
Pre-culture medium included (per liter): 40 g brain heart infusion, 10 g glucose, and 20 g D-sorbitol. Flask cultivation medium included (per liter): 25  Additionally, 20 g/L of CaCO 3 was added to each flask as buffer and CaCO 3 was removed by diluted HCl post-cultivation.

Cultivation conditions
Cultivation was carried out in 100-mL Erlenmeyer flasks at 30 °C and 200 rpm in a shaking incubator (KS 4000 I control; IKA-Werke GmbH, Staufen im Breisgau, Germany). Each strain was maintained in 60 % glycerol solution and was inoculated in 5 mL pre-culture medium. After 17-19 h, when OD 600 reached 5.5-6.0, 1 mL of pre-culture was spun down at 3000 rcf and the pellet was resuspended in 25 mL of cultivation medium. When applicable, isopropyl β-d-1-thiogalactopyranoside at a final concentration of 1 mM was added at OD 600 0.5-0.8. Neomycin at a final concentration of 50 µg/L was added as a selective marker during each cultivation.

Cell growth and routine metabolite analysis
Cell growth during cultivation was monitored by measuring OD 600 on a Genesys 20 spectrophotometer (Thermo Fisher).
The concentration of glucose and organic acids in cultivation broth was analyzed by a Jasco (Tokyo, Japan) LC-4000 high-performance liquid chromatography system equipped with an autosampler (AS-4150), a pump (PU-4180), a column oven (CO-4061), a UV detector (RI-4030), and an RI detector (UV-4075). The mobile phase consisted of 5 mM H 2 SO 4 and was pumped at constant flow rate of 0.8 mL/min. A Rezex ROA-organic acid H + column (Phenomenex, Torrance, CA, USA) maintained at 80 °C was used to separate the metabolites in the cultivation broth. The concentration of each compound was determined by Jasco ChomNAV software (version 2.03.03).

Confirmation of 3-hydroxyadipic acid by GC/MS analysis
3-hydroxyadipic acid was identified by GC/MS as reported previously with minor modifications [19,71]. Briefly, 40 µL of cell-free cultivation broth was lyophilized, resuspended in 50 µL pyridine, and reacted with 50 µL MeOX (20 g/L in pyridine) for 120 min at 30 °C. The metabolites were then derivatized with 50 µL MSTFA at 60 °C for 1 h prior to injection.
The GC/MS system was equipped with a Focus GC (Thermo Fisher), ISQ MS, and PAL COMBI-xt autosampler. A Zebron ZB-5MS column (Phenomenex) was used, with 1.1 mL/min gas flow and helium as the carrier gas. The column temperature gradient was as follows: initial hold at 60 °C for 1 min; ramp at 10 °C/min to 325 °C; and final hold at 325 °C for 10 min. Equilibration time was 0.5 min and preparatory run time out was 10 min. The inlet temperature was 250 °C. The electron multiplier was operated at 70 eV with scanning acquisition mode. In total, 1 µL of sample was injected for each analysis with split-less mode. The ion (MS) source was set at 250 °C, mass filter (MS Quad) at 150 °C, and MS transfer line at 290 °C.

Solid-phase extraction
For sample clean up, 33 μm polymeric reverse-phased solid-phase extraction (SPE) column (200 mg/3 mL; 8B-S100-FBJ) was used with minor modification from manufacturer's protocols (Phenomenex). The precondition and equilibration steps were according to the manufacturer's protocols. The 5 mL of cell-free cultivation broth was pretreated with final 1 % of formic acid prior to loading to the column. The column was washed with 1 mL water and eluted with 1 mL methanol. Methanol was removed by evaporation prior to next steps.

Detection of trans-2-hexenedioic acid and adipic acid by LC-MS
For detection of adipic acid and trans-2-hexenedioic acid, SPE treated samples were analyzed using LC-MS. A Dionex Ultimate 3000 RS UPLC with a diode array detector (DAD) was coupled to a Bruker Maxis QTOF MS with electrospray ionization (ESI) (Bruker Daltonics, Bremen, Germany). The system was equipped with a reversed phase column (Kinetex; 1.7 μm, F5, 100 Å, 150 mm × 2.1 mm; Phenomenex) and was maintained at 40 °C. Gradient elution was utilized for separation and a constant flow rate of 400 µL/min was held throughout the run. Two eluents were used in the gradient elution: eluent A (H2O, 20 mM formic acid) and eluent B (ACN, 20 mM formic acid). The gradient started with 10 % eluent B rising to 70 % over 6.7 min, followed by 70-100 % eluent B over 1.3 min, then held at 100 % eluent B for 3 min. Analysis was carried out in negative ESI. The scan range was m/z 75-1250 with 2 spectra per second. For the targeted fragmentation a collision energy of 15 eV was used.
Prior to the analysis of samples, the QTOF-MS was calibrated using a sodium formate calibrant. In addition, all data files were recalibrated with an internal standard of sodium formate injected prior to initial sample elution for each sample.