Background

Want to treat your baby cat better? I bet you are using catnip. Or even if you don’t own a cat, you must have heard of this magical herb. Catnip refers to the plant Nepeta cataria renown for its interesting behavioural effects on felines. Upon smelling the catnip, not only domestic pet cats, but also larger feline species like tiger, leopard, lynxes, and cougars^[1], all exhibit behaviors including sniffing, licking and chewing, chin and cheek rubbing, body rolling, vocalizing and salivating^[2], similar to those of females in heat.

Figure. 1 Cats and catnip

Although it has a drug-like effect, catnip is non-addictive and harmless to cat and has been used in many recreational toys made for cats. The secret behind the magic of catnip is nepetalactone, a monoterpene found in catnip oil similar to cats’ pheromones. The binding of nepetalactone to specific olfactory receptors at the olfactory epithelium of cats, which is hypothesized to stimulate the medial amygdala and medial preoptic area associated with sexual behaviors.

Apart from being a powerful cat-attractant, nepetalactone also exhibits potent repellent against a large number of insects including mosquitoes, fly, cockroaches, and termites.^[3][4] Research suggests that nepetalactone repels mosquitoes ten times more effectively than DEET.^[5]

However, the complete biosynthesis pathway of nepetalactone is not yet elucidated, with the last enzyme responsible for the conversion of nepetalactol to nepetalactone unidentified.^[6]

The precursor of nepetalactone, nepetalactol is also an intermediate in the synthesis pathway of strictosidine, the universal biosynthetic precursor of all monoterpene indole alkaloids (MIAs). ^[12]

Figure. 2 The reconstituted S. cerevisiae strictosidine biosynthetic pathway.^[12]

MIAs are a family of plant-derived metabolites^[7] with significant therapeutic value and important biological activities, such as the anticancer medicine vinblastine and vincristine,^[8][9] and the antimalarial drug quinine. However, the structural complexity of MIAs has made them difficult to be chemically synthesised.^[10] Moreover, the natural yield of MIAs by plants is extremely low — vinblastine for example only makes up about 0.001% dry weight of Catharanthus roseus,^[11] consequently making the price of it incredibly high.

Figure. 3 Price of vinblastine provided by www.drug.com

Synthetic biology offers a potential solution to this puzzle. Recently, progress has been made in the microbial production of strictosidine in engineered yeast, producing a titre of 0.53mg/L.^[12] But this could still be improved as the unsatisfactory flux to the early intermediate nepetalactol was limiting the yield. Thus, optimisation of nepetalactol production becomes a crucial bottleneck to overcome in order to enable efficient microbial production of strictosidine and downstream MIAs^[13] — the reason why we decided to work on this project.

The Synthesis Pathway of Nepetalactol

Figure. 4 The biosynthesis pathway from simple sugar to nepetalactol

Obtaining nepetalactol from a simple carbon source involves the participation of a considerable sum of 23 enzymes.^[14] Energy sources like glucose is first metabolised to acetyl-coA which is then converted into IPP and DMAPP, the two universal precursor of all terpenes, through a mevalonate-dependent (MVA) isoprenoid pathway in the case of animals, archaea, and yeast, or a methylerythritol 4-phosphate (MEP) pathway as in most prokaryotes,^[15][16] or both for certain species. IPP and DMAPP would be catalysed into geranyl pyrophosphate (GPP) by a geranyl pyrophosphate synthase (GPPS) followed by a conversion of GPP to geraniol facilitated by a geraniol synthase (GES). The conversion of geraniol to nepetalactol involves three cytochrome P-450 catalysing three metabolic reactions: (a) hydroxylation of geraniol by geraniol 8-hydroxylase (G8H) to form 8-hydroxygeraniol; (b) oxidation of 8-hydroxygeraniol to 8-oxogeranial by geraniol oxidoreductase (GOR); (c) the reductive cyclisation to produce nepetalactol by iridoid synthase (ISY).^[17] Besides reacting with the carbon eight, an alternative biosynthesis route exist on carbon ten involving geraniol 10-hydroxylase (G10H), 10-hydroxygeraniol oxidoreductase (10HGO) and iridoid synthase (ISY).

Biosynthesis of Nepetalactol in S. cerevisisae is subjected to Enzyme Promiscuity and Endogenous Reduction

To our knowledge, there are two precious studies aiming at engineering nepetalactol-producing platform strains: A. Campbell’s in 2016^[18] and J. Billingsley’s in 2017^[13], both used the eukaryotic host S. cerevisiae because expression of cytochrome P-450 is required.

A. Campbell and his colleagues tried to create a nepetalactol-producing yeast strain based on an existing yeast strain that could produce 11.4mg/L of the precursor geraniol. They have shown the production of 10-oxogeraniol and nepetalactol in vitro. Upon introduction of G10H, the strain successfully produced 5.3 mg/L of 10-hydroxygeraniol. But after expressing 10HGO and ISY, none of geraniol, 10-hydroxygeraniol, 10-oxogeraniol, or nepetalactol was detectable by GC-MS, instead, striking peaks of the shunt products like citronellol and 10-hydroxycitronellol were observed. Later, they found ISY was the promiscuous enzyme which has diverted the flux of carbon source away to the shunt products.

Figure. 5 Production of shunt products due to enzyme promiscuity

J. Billingsley and his co-workers have improved based on A. Campbell’s study. They only explored the biosynthesis of nepetalactol from 8-hydroxygeraniol by directly supplying chemically synthesised 8-hydroxygeraniol to yeast. This study has clarified two distinctive mechanisms of shunt product formation involving yeast endogenous ‘ene’-reduction and alcohol dehydrogenation. They succeeded in obtaining a nepetalactol yield of 45 mg/L from 8-hydroxygeraniol upon deletion of 5 endogenous yeast genes: oye2, oye3, ari1, adh6, adh7.

Figure. 6 Two distinctive routs towards shunt product formations

Our approach: a division of labour

Figure. 7 An overview of the specialisation between E. coli and S. cerevisiae

Although some progress has been made by previous works, there are still limitations in the current methods used. We want to establish an one-stop production of nepetalactol without having to chemically synthesise any precursors. The solution lies behind the employment of a division of labour between different chassis: S. cerevisiae, as the most widely-used eukaryotic model organism, is often used for the production of complicated molecules like artemisinic acid^[19] and even cannabinoid^[20], and was once used for nepetalactol production; E. coli though being a lower organism than yeast, exhibits much better efficiency in producing simpler products like the precursor geraniol^[21]. So our strategy is to split the pathway at the position of geraniol to have E. coli generating geraniol and yeast converting it to nepetalactol, and to co-culture E. coli and yeast.

Figure. 8 Comparison between different host for monoterpene/monoterpenoid synthesis^[21]

To put in a nutshell, we want to create a co-culture system with synthetic E. coli and S. cerevisiae able to perform the production of geraniol and conversion of geraniol to nepetalctol respectively. E. coli would first convert the carbon source fed to geraniol, the semi-finished product. Next, geraniol would be transported out of E. coli into the culture medium by diffusion and native E. coli transporters as geraniol are inhibitory to its growth. Geraniol would then enter into S. cerevisiae and it’s metabolised into nepetalactol. Together, E. coli and yeast are like two compartments in a nepetalactol-manufacturing factory, and they are designed, characterised for their individual functions before being cultured in a system.

Firstly, we need to engineer E. coli to make it synthesise geraniol at the higher titre possible. Excretion of geraniol into the culture medium would follow its production simultaneously by diffusion and native E. coli transporters.^[22] Since E. coli naturally produces acetyl-CoA and possesses a MEP pathway, theoretically with addition of only GPPS and GES, it would acquire the capability of synthesising geraniol. However, for E. coli to generate substantial geraniol, it is probable that the endogenous MEP pathway couldn’t supply enough IPP and DMAPP. Previously, engineering effort aimed at terpenoid/terpene synthesis in prokaryotes attempted to enhance the intracellular availability of precursors IPP and DMAPP through the optimisation of the native E. coli MEP pathway^[23], which was proven to be restricted by some not-yet elucidated physiological mechanism in the host E. coli. The subsequent research has altered the approach, supplementing the terpenoid/terpene production by introduction of a heterologous S. cerevisiae mevalonate-dependent pathway in E. coli.^[24] This method appeared to establish a more stable and efficient supply of IPP and DMAPP.

Then, we would engineer yeast for converting geraniol to nepetalatcol. Besides introduction of G8H, GOR and ISY, knocking out native yeast gene which crosstalk with nepetalatol synthesis pathway is also necessary to avoid shunt product production as indicated in J. Billingsley et. al’s study. The hydroxylation of geraniol to 8-hydroxygeraniol was reported to be a limiting step in nepetalactol synthesis because of poor activity of G8H in yeast, making the over-expression of G8H a crucial requisite. ^[12]