This system yielded 24 g/L of isobutanol, close to titers reported in microbes. However, the limiting factor of the system was identified as enzyme instability in high isobutanol, and the authors were able to determine precisely which enzymes were the least stable. Since, the lab has stabilized many of the enzymes in the pathway by either engineering the enzyme or using a more thermostable variant. Current titers in a 15 mL bioreactor have reached nearly 300 g/L isobutanol orders of magnitude higher than microbial production . While the monoterpene and isobutanol pathways required sophisticated methods for cofactor regeneration, that may not always be the case. You et al utilized an 5 step enzymatic system to convert starch into myo-inositol , which is co-factor independent. The enzymes were expressed in E. coli and purified using a simple heating step. The resulting biosynthesis of starch to myo-inositol yielded 95 g/L in 48 hours on an 18,000 L scale. This was a remarkable study because of the scale and final titers achieved. It demonstrates that it is possible to translate the enzymatic systems to an industrial scale. Synthetic biochemistry is still in its infancy, but it has the potential to be a powerful tool for the production of natural products. There are a few obstacles that need to be addressed before more complex systems are industrially relevant. First, since synthetic biochemistry does not use cell lysates or cells, so cofactors like ATP, coenzyme A and NAD+ need to be added to start the system. While recycling these cofactors and using them for several iterations would reduce the associated cost, it is important to find an inexpensive source for these molecules. Additionally, dry racks for weed the enzyme catalysts can also be an expensive component of these systems, however this cost can be overcome by recycling the enzymes and using them for several iterations.
The use of thermostable enzymes is a key factor, demonstrated by You et al. The thermostable enzymes are able to be purified by a simple heating step, reducing the cost associated with protein purification. Additionally, they are able to recycle the enzymes and use them in subsequent bioreactor runs. While the obstacles might take time to solve, they are definitely solvable problems, making synthetic biochemistry an alternative approach for the biosynthesis of natural products. Described above are several bio-based approaches for the production of natural products. When seeking a bio-based approach for natural product production, it is important to recognize that different systems will work better for different molecules. In the case of paclitaxel it is clear that plant cell culture is currently the best method, however for monoterpenes a synthetic biochemistry approach may be a better option based on titer. The focus of this thesis is to evaluate a synthetic biochemistry approach for the production of cannabinoids and other prenylated aromatic polyketides. Due to the low cannabinoid titers seen with metabolic engineering it is possible that an alternative approach may be more successful. The low titers are most likely due to competition for the precursor acetyl-CoA. It requires 9 acetylCoA molecules to produce 1 molecule of THC or CBD. In addition to the 9 molecules of acetylCoA needed for cannabinoid biosynthesis, there are other essential competing pathways like fatty acid biosynthesis. Additionally the 9 acetyl-CoA molecules are split between the isoprenoid pathway and the polyketide pathway , and engineering the yeast to express each enzyme at the level needed to balance the acetyl-CoA flux is extremely difficult. A synthetic biochemistry approach for this pathway may be the better option due to better control over reaction components, flux, and a sufficient supply of the precursor acetyl-CoA. Due to these advantages, synthetic biochemistry could be a useful tool for the sustainable production of cannabinoids.
Prenylated natural products are a large class of bioactive molecules with demonstrated medicinal properties.1 Examples include prenyl-flavanoids, prenyl-stilbenoids and cannabinoids . Cannabinoids in particular show immense therapeutic potential with over 100 ongoing clinical trials as antiemetics, anticonvulsants, antidepressants and analgesics2–6 . Nevertheless, despite the therapeutic potential of prenyl-natural products, their study and use is limited by the lack of cost-effective production methods. Plant-derived prenyl-compounds are difficult to isolate due to the structural similarity of contaminating molecules, and the variable composition between crops. These challenges are further exacerbated when attempting to isolate low abundance compounds. Many chemical syntheses have been developed to address the challenges associated with making prenylated natural products, but they are generally impractical for drug manufacturing due to the degree of complexity and low yields. Microbial production is a useful alternative to natural extraction for prenylated natural products, but comes with many challenges such as the need to divert carbon flux from central metabolism and product toxicity to name a few. For example, prenyl natural products like prenylnaringenin, prenyl-resveratrol and cannabidiol are derived from a combination of the metabolic pathways for fatty acid, isoprenoid, and polyketide biosynthesis. So high-level production requires efficient re-routing of long, essential and highly regulated pathways. Despite the challenges, many groups have engineered microbes to produce unprenylated polyketides, like naringenin, resveratrol and olivetolate, but at relatively low levels . Obtaining prenylated products is even more challenging because GPP is an essential metabolite that is toxic to cells at moderate concentrations, creating a significant barrier for high level microbial production. So, in spite of intense interest, to our knowledge there are no published reports of the complete biosynthesis of prenyl-flavonoids, prenyl-stilbenoids or cannabinoids in recombinant microbes.
Much recent effort has focused on alternative methods for cannabinoid production. Two groups have produced the polyketide cannabinoid intermediate, olivetolic acid at low levels in yeast or E. coli , but did not prenylate OA or produce a cannabinoid from the bio-synthesized OA . In other work, tetrahydrocannabinolic acid THCA was produced in cell extracts from either exogenously added geranyl-pyrophosphate and OA in a two enzyme pathway or from cannabigerolic acid using a single enzyme 17. However, it is unclear how GPP or CBGA could be obtained at sufficient levels for economical production due to the high cost of these molecules. Here we propose an alternative biological approach to prenylated natural product biosynthesis using a cell-free enzymatic platform we call synthetic biochemistry, which has shown great promise for the production of bio-based molecules. The synthetic biochemistry approach frees us from worrying about the toxicity of products and intermediates, affords rapid design-build-test cycles, precise control of all system components, and complete flexibility in pathway design. Nevertheless, building highly complex systems involving dozens of enzymes, associated cofactors and myriad metabolites on a large scale outside the context of the cell is an enormous challenge. One of the keys to making commercially viable cell-free systems is reducing enzyme costs by employing stable enzymes that can last for long periods of time. Recently Zhang and co-workers converted maltodextrin into inositol at a 20,000 L scale in a 5 enzyme system using thermophilic enzymes purified by simple heating step, vertical farming pros and cons demonstrating that at least simple cell-free systems can reach industrial scale. Another key requirement is designing systems that effectively generate and recycle high energy cofactors H so that they can be used many times. We have previously reported a flexible enzymatic purge valve and rheostats for the regulating the supply of reducing equivalents and ATP, allowing us to build systems that run for many days and produce high titers of isobutanol and terpenes. Here we employ these concepts to develop cell-free production of a variety of prenylated compounds. We use glucose as a feedstock to produce GPP and optimize the system for the high-titer production of the cannabinoid compounds CBGA and cannabigerovarinic acid .Our synthetic biochemistry approach is outlined in Figure 2-1 and Figure 2-5 and expands on a system we developed previously for terpene production. First, glucose is broken down via a modified glycolysis pathway to produce high energy cofactors ATP and NADPH in addition to the carbon building block, acetyl-CoA using an alternative pyruvate oxidation pathway. The acetyl-CoA is then assembled into the prenyl-donor compound, GPP, via the mevalonate pathway using the ATP and NADPH produced from glycolysis. Importantly, a purge valve introduced into the glycolysis pathway balances NADPH production and consumption while maintaining carbon flux. The prenylation module then uses the GPP to prenylate exogenously added substrate to yield the desired prenylated product. To expand the capabilities of our synthetic biochemistry platform we developed a prenylating system that employs a non-specific prenylating enzyme such as NphB, AtaPT, or NovQ to produce an array of prenyl-compounds derived from glucose.
We then further engineered NphB using Rosetta to specifically prenylate OA. As a first test of the system, we built the full cell-free system to generate GPP from glucose and employed wild-type NphB to prenylate its preferred substrate 1,6 dihydroxynapthalene . 1,6 DHN was added at the beginning of the reaction along with glucose. Up to ~400 mg/L of prenylated product was obtained from 2.5 mM 1,6 DHN. However, increasing the 1,6 DHN concentration from 2.5 to 5 mM, decreased final titers 2-fold suggesting that 1,6 DHN inhibited one or more enzymes . Enzyme assays revealed that pyruvate dehydrogenase was inhibited by 1,6 DHN, as well as olivetol, resveratrol, and olivetolate . Therefore, to engineer a general prenylation system, we sought to eliminate PDH. To remove the need for PDH, we implemented al PDH bypass . In the PDH bypass, pyruvate is converted to acetyl-CoA using a pyruvate oxidase to produce acetylphosphate followed by the action of acetyl-phosphate transferase . The PDH bypass had two advantages. First, PDH is a large enzyme complex that is difficult to work with, so bypassing PDH streamlines enzyme production. More importantly, initial experiments revealed that the bypass is not subject to the inhibition seen at higher concentrations of 1,6 DHN. Once we confirmed the PDH bypass improved 1,6 DHN titers, we began to optimize the system as a general prenylation system. We varied co-factor concentrations, protein levels, and environmental conditions such as temperature and pH to identify the ideal set of conditions. Throughout this process we found that ATP, NADP+ , phosphate and NphB concentrations had the greatest impact on the final titer. As shown in Figure 2-2A, when we employed the PDH bypass, we found a 4- fold increase in titers of 5-prenyl-1,6 DHN when starting with 5 mM 1,6 DHN . When utilizing the PDH bypass system, approximately 50% of 1,6 DHN was converted in 24 hours, reaching a final titer of 705 ± 12 mg/L . We then tested the ability of the PDH bypass cell-free system to prenylate a variety of aromatic substrates . Thus, it is possible to produce a variety of prenylated natural products using a cell-free enzymatic system to generate the expensive co-substrates GPP and DMAPP. Further, the ease with which an exogenous substrate can be added to a synthetic biochemistry system is a great advantage because it is often not possible to add co-substrates exogenously to microbes since they cannot enter the cell.16 To test whether we could use synthetic biochemistry to produce high levels of therapeutically relevant prenylated products, we focused optimization efforts on cannabinoids due to the growing interest in new ways to make these medically important compounds. As shown in Figure 2-2D the initial system produced the cannabinoid precursor CBGA at a constant rate of 2.1 mg L-1 hr-1 over 72 hours and reached a final titer of only 132 mg L-1 . Although the system produced CBGA, there were two problems. First, the turnover rate of the prenyl transferase NphB for CBGA production is extremely poor . Second, prenylation of OA by NphB is highly non-specific, generating a major side-product, 2-O-geranyl olivetolate16. We therefore sought to improve CBGA production by enhancing the activity and specificity of NphB by design. Redesign of NphB to improve CBGA synthesis Briefly, OA was docked into the active site of the NphB crystal structure 32, then Rosetta was used to predict mutations that would improve OA binding. We narrowed the Rosetta results to a 22 construct library , and screened for CBGA production . We made several key observations during the initial screen, Figure 2-12: Y288A and Y288N by themselves dramatically enhanced activity, as predicted by computation; The presence of Y288N in any construct decreased the enzyme yield suggesting Y288N may be a destabilizing mutation ; The addition of G286S in the Y288N background appeared to improve activity further over Y288N , suggesting that G286S could be another favorable mutation; We noted an activity improvement of Y288A/F213N/A232S over Y288A/F213N suggesting that A232S may also be a favorable mutation.