Figure 2. a) Illustration of the plasmid used in geraniol production in GE01 (pIG_01 ). b)Illustration of the plasmid used in converting geraniol into geranate inGA01 (pGA_01 ). Production of c) geraniol withGE01 and d) geranate with GA01 with and without hexadecane within 24 h. No geranate was detected in the culture of GE01 and no geraniol was detected in the culture ofGA01. e) Isopentenol consumption by GE01 orGA01 within 24 h with hexadecane. f) Formation of by-products (3-MC and 3-MB) by GE01 or GA01 within 24 h. n.d., not detected. g) Extracted ion chromatogram (EIC) of 3-MC (m/z=101.06) and geranic acid (m/z=169.12) standards. 3-MC was eluted at 4.8 min and geranic acid was eluted at 9.2 min. h ) mass spectra of 3-MC and geranic acid. Error bars indicate standard error (n=3).

Pathway construction for conversion of geraniol into geranate in E. coli

To convert geraniol into geranate, we co-expressed CdGeDH andCdGaDH from C. defragrans in one operon in a separate plasmid (pGA01 ) under the control of a strong inducible promoter PcymO (Figure 2b ). We named the geranate production strain GA01 . After the fermentation, Liquid Chromatography Mass Spectrometry (LC/MS) was used to quantify geranate in the aqueous phase (Figure 2g, h ); Gas Chromatography Mass Spectrometry (GC/MS) was used to quantify geranic acid in the organic overlay. The fermentation produced 298 mg/L geranate with GA01(Figure 2d ). Geraniol and geranial were not detected in the culture of GA01 , indicating that CdGeDH and CdGaDH have high catalytic efficiency with geraniol and geranial, respectively.
While geraniol was partitioned more favourably into the organic phase of the GE01 culture (Figure 2c ), most of the geranate accumulated in the aqueous phase of the GA01 culture (Figure 2d ). This was expected because geraniol (LogP = 3.56) (Griffin et al., 1999) is more hydrophobic than geranate (LogP = 2.8) (Ko et al., 2021). Hexadecane is frequently used to prevent loss of volatile products (such as geraniol). Using hexadecane, however, complicates downstream product purification (it is challenging to separate geraniol from hexadecane). When we removed the hexadecane from the GE01 culture, the geraniol titer was reduced to 8% of the control (with hexadecane, Figure 2c ). We, however, found that a similar geranate titer (268 mg/L) was achieved with or without hexadecane (Figure 2d ). Hexadecane was not used in the rest of this study to simplify the fermentation process.
Next, we determined the consumption of isopentenols in the cultures, and found that GA01 consumed more isopentenols than GE01(Figure 2e ). Since the titer of geranate in the GA01culture was much lower than that of geraniol in the GE01culture (Figure 2c,d ), this was an indication that by-products were formed along with geranate by the geranate-producing GA01 . We hypothesized that the hydroxyl group of isoprenol and prenol may also be oxidized by CdGeDH and CdGaDH to form corresponding carboxylates (3-methyl-crotonate [3-MC] and 3-methyl-3-butanoate [3-MB],Figure 1 ), as these oxidation reactions would compete with the downstream reaction of geranate biosynthesis. We confirmed this hypothesis by analysing the aqueous phase of GE01 andGA01 cultures with LC/MS (Figure 2f ). As shown inFigure 2f , 773 mg/L of by-products (3-MC and 3-MB) were produced by GA01 while no by-products were found in the culture of GE01 (Figure 2f ).

Tuning expression of CdGeDH and CdGaDH to reduce by-product formation

It is possible that the expression level of CdGeDH and CdGaDH was too high in GA01 , rapidly oxidizing a large amount of isopentenols into the by-products and competing with EcthiM (the first kinase in IUP) for substrate. This suggests a reduction in the expression level of CdGeDH and CdGaDH so that more isopentenols could enter the IUP for production of geranate. To test this hypothesis, we constructed a new plasmid (pGA02 ) with a low copy number replication origin (pSC101; copy number: ~5) to weaken the expression of CdGeDH and CdGaDH (pGA01 used pAC replication origin; copy number: 10-15). We replaced pGA01 with pGA02 inGA01 , creating a new strain (GA02 ). The geranate produced by GA02 (764 mg/L) was almost three times that produced by GA01 (Figure 3a ), and the concentration of the by-products (3-MC and 3-MB) was substantially reduced (from 865 mg/L to 436 mg/L) (Figure 3b ). Then we replaced the PcymO promoter in GA02 with PT7or PthrC3 (Anilionyte et al., 2018), creating E. coli GA03 and GA04 , respectively. These two strains produced a similar amount of geranate as GA02 (Figure 3a ), but with significantly reduced amount of by-products (68 mg/L and 190 mg/L 3-MC/3-MB by GA03 and GA04 , respectively,Figure 3b ). A low IPTG concentration (0.05 mM) was used in this study, so PT7 could be weaker than PthrC3 and PcymO.
A previous study successfully engineered Acinetobacter sp. Tol 5to transform geraniol into geranate (Usami et al., 2020). Their geranate titer was similar to what we achieved in this study. Our process, however, used a cheaper substrate (isopentenols instead of geraniol) and was much faster (24 h instead of 144 h).