Terpene synthases catalyze the cyclization of the hydrocarbon moiety of isoprenoids to yield cyclized hydrocarbons. While terpene synthases generally have one major product, they are fairly promiscuous and can produce several different terpenes with one enzyme. For example, the humulene synthase from Abies grandis species mainly produces humulene, but also produces 51 additional sesquiterpenes at lower levels. The benefit of producing 52 different compounds with one enzyme is to maximize the diversity of secondary metabolites made from one bio-synthetic pathway, or better yet one enzyme. Due to the myriad of beneficial properties of terpenes, a great deal of research has been dedicated to engineering microbial strains to produce terpenes. Limonene is a rather simple, well characterized monoterpene that is commonly used in the fragrance industry and as a food additive. While limonene has a relatively high natural abundance in the rind of citrus fruit, several groups have engineered E. coli to produce this molecule as proof of concept. Additionally, a derivative of limonene, perillyl alcohol, is reported to have anticancer properties, and a low natural abundance making it an interesting target for an engineered microbe. Alonso-Gutierrez et al engineered E. coli to produce limonene at 435 mg/L from glucose, and then introduced a cytochrome P450 to hydroxylate limonene to produce perillyl alcohol at 100 mg/L.However, this level of limonene production still falls short of the theoretical maximum, 3.2 g/L, cannabis dryer indicating the system is not an efficient producer of limonene. Willrodt et al improved on the E. coli production system by introducing different mevalonate pathway enzymes, and reached a titer of 2.7 g/L of limonene.
While the authors were not able to determine the limiting factor in their system, they identify high acetate levels in the high limonene producing strain. While the authors determine this to mean there was sufficient levels of acetyl-CoA, they do not discuss how a competing pathway, like phosphotransacetylase and acetate kinase, could limit the flux of acetyl-CoA into the desired mevalonate pathway. Engineering microbes to produce monoterpenes is further complicated due to the effects monoterpene toxicity. To address this issue both studies utilized an organic overlay to extract the secreted product. Amorphadiene is a sesquiterpene and a precursor to the essential anti-malarial drug artemisinin. The highest titer of amorphadiene was achieved by Westfall et al. They reached a titer of 40 g/L of amorphadiene in yeast by overexpressing enzymes in the mevalonate pathway and the amorphadiene synthase, and limiting competing pathways. In an attempt to produce artemisinic acid in an engineered microbe, the strain developed by Westfall et al was further engineered to express a cytochrome P450 to produce artemisinic acid. However, despite the high titers of amorphadiene, the artemisinic acid levels remained significantly lower at ~150 mg/L. Amyris improved the titer substantially to 25 g/L of artemisinic acid. The improvement in artemisinic acid was due to the discovery of an aldehyde dehydrogenase from A. annua that improved the conversion of artemisinic aldehyde into artemisinin. Additionally they increased the viability of the strains by introducing an organic overlay, isopropyl myristate, which helped solubilize the product.
While amorphadiene is a precursor in the biosynthesis of artemisinin, artemisinic acid is not. Therefore, Amyris designed a four step synthetic approach to convert artemisinic acid into artemisinin. The combination of synthetic and bio-based approaches is known as semi-synthesis, and is another approach for the production of natural products. This approach can reduce the difficulty of chemical synthesis, and lead to a more cost effective option. Amyris demonstrated this by chemically converting artemisinic acid derived from microbial fermentation into artemisinin in a 4 step chemical synthesis with an overall yield of 40%. This semi-synthetic process is currently being used by Sanofi to supplement the world’s supply of artemisinin. Taxadiene is a diterpene and the first intermediate in the biosynthesis of paclitaxel. Several groups have engineered microbes to produce taxadiene and one study produced taxadien-5a-ol, the second intermediate in the pathway. While taxadiene has been produced at moderate titers in E. coli , the production in other organisms, such as S. cerevisiae, B. subtilis and the fungi A. alternata TPF6 remains low . The E. coli system that generated 1 g/L taxadiene was also engineered to produce taxadien-5a-ol, but the titers of taxadiene remain significantly higher indicating inefficiencies with the hydroxylation of taxadiene. This is most likely because cytochrome P450s are challenging to express in E. coli. Biggs et al were able to improve hydroxylation of taxanes by optimizing for P450 expression, reductase partner interactions and modifications of the N-terminus of the P450. While it may be possible to complete the paclitaxel biosynthetic pathway in a microbe, it would be extremely challenging. Therefore, plant cell culture is still the best route for paclitaxel production. While there are some success stories of engineering microbes to produce terpenes and terpene derived molecules , several challenges still remain. Monoterpenes are toxic to both E. coli and yeast at relatively low concentrations . While cell viability can be improved by using an organic solvent to extract the product, slow diffusion out of the cell can still have an impact. There are also issues with precursor availability. The data from Willrodt et al indicates that competing pathways reduce the availability of the precursor acetyl-CoA.
Additional challenges, like functional enzyme expression create issues with the production of complex natural products like artemisinin and paclitaxel. While yeast and E. coli are relatively easy to engineer, they do not always possess the cellular machinery needed for the functional expression of necessary enzymes.Alkaloids are a broad group of natural products that are produced by plants, fungi and bacteria, they are loosely classified as molecules that include one or more nitrogen atoms. Some examples of alkaloids are caffeine , opiods and psilocybin . Due to the broad classification of alkaloids, their biosynthesis is rather diverse . For example, the molecules listed above are derived from different precursors, caffeine is derived from xanthosine, opioids are derived from tyrosine and is derived from tryptophan. Several alkaloids found in plants have been produced in engineered microbes. Although caffeine is naturally abundant, Jin et al engineered S. cerevisiae to produce the stimulant. The conversion of xanthosine to caffeine is four enzymatic steps requiring three enzymes, however the nucleosidase is naturally present in yeast, so they only introduced two genes into the strain, a caffeine synthase and a methyl transferase. The strain produced 0.38 mg/L of caffeine, when supplemented with xanthosine. The authors hypothesized the low titer could be due to low nucleosidase activity and screened several non-native nucleosidases, however this did not increase titers. Additionally, Jin et al demonstrate that substrate consumption is not correlated with caffeine production, and that no caffeine is produced if the culture is not supplemented with xanthosine. Because xanthosine is required for primary metabolism of yeast, it is likely that the added xanthosine is being diverted into primary metabolic pathways instead of caffeine biosynthesis, limiting the titers of caffeine. As mentioned previously, vertical farming systems many opioids are still isolated from their natural source. In attempt to find a better production method yeast was engineered to produce the opioid, hydrocodone. It required introducing over twenty genes into the yeast genome . The resulting strain produced 0.3 µg/L of hydrocodone. At this titer it would require nearly 17,000 L of yeast to produce one dose of hydrocodone. The low titer is likely due to the length and complexity of the biosynthetic pathway, but it is a proof of concept suggesting that long, complex exogenous pathways can be functional in yeast. The remaining challenge is identifying the bottlenecks and adjusting flux accordingly. Aspergillus nidulans was engineered to produce psilocybin, a psychotropic molecule under FDA investigation as a treatment for anxiety, depression and substance abuse. Because psilocybin is naturally produced by fungi, it is logical to use the fungal model organism A. nidulans to produce this molecule. The four step pathway, shown in Figure 1-4, was transformed into A. nidulans, which resulted in a final titer of 110 mg/L, or 1.5% of mycelium dry weight, similar to the amount naturally produced in mushrooms. However, besides using a fluorescence based assay to identify positive transformants, the authors did not make any additional changes to the fungi. Therefore, by engineering the fungi to limit competing pathways and direct flux into the desired pathway it is possible to reach higher titers. The microbial production of alkaloids has similar challenges to that of terpenes. It is a challenging to balance competing pathways with a target pathway, like for the production of caffeine. Xanthosine is required for primary metabolism, and so most of the added substrate is diverted into primary metabolism instead of caffeine production. A similar challenge is presented for opioids with tyrosine being an essential metabolite that is required for opioid biosynthesis as well. In addition, longer more complex pathways are generally extremely difficult to engineer into a heterologous host. First, over expressing twenty enzymes can create a metabolic burden, and it can be difficult to express the enzymes at appropriate levels. Additionally, the enzymes used in the study were sourced from mammals, plants, bacteria in addition to yeast enzymes, which means there could be issues with functional expression.
The opioid pathway also produces several non-native intermediates which could have an impact on cellular metabolism. Interestingly, the baseline production of psilocybin in Aspergillus nidulans was 110 mg/L, however it may be possible to reach significantly higher titers by engineering the organism further. Previous studies have indicated that altering central metabolism to increase precursor pools can have a dramatic improvement on overall titer. Polyketides are produced by all kingdoms of life, and are a rich source of medicinal compounds. In fact, several are FDA approved antibiotics, immunosupressants, anticancer and cholesterol reducing medications. The biosynthesis of polyketides is similar to fatty acid biosynthesis, where small extender units are added to a growing acyl chain; however the subsequent steps are less uniform than FA biosynthesis allowing for more diversity in the final polyketide product. While the chemistry for the biosynthesis of polyketides is highly conserved, the types of polyketide synthases involved are very different. There are three general types of PKS reviewed previously by Hertweck in 2009. While several bacteria have been engineered to increase natural production of FDA approved antibiotics, herein the focus will be on producing plant polyketides in microbial hosts. Stilbenoids and flavonoids are a classes of aromatic polyketides produced by plants, and they have an array of useful bio-activities. These molecules can further modified to yield prenylflavonoids and prenyl-stilbenoids , which has been shown to increase the potency of flavonoids and resveratrol. Due to the low natural abundance of flavonoids, stilbenoids and their prenyl derivatives, they are an interesting target for microbial engineering. The molecule is potent antioxidant cited to have neuroprotective, cardioprotective and anti-aging properties. The molecule is sold as a supplement, and is also used as an additive in food products and cosmetics. The market for resveratrol in 2017 was valued at $69.1 million based on a report published by Coherent Market Insights. Several groups have engineered microbes to produce this molecule . The most successful was Lim et al, reaching a titer of 2.3 g/L. They attribute the high titer to directing carbon flux into the malonyl-CoA precursor, and limiting pathways, like fatty acid biosynthesis that would compete with the resveratrol pathway. Naringenin is a flavonoid produced by an array of plants. It is a fairly potent estrogen mimic, and is sold as a supplement to ease the symptoms of menopause. In addition it is a potent antioxidant with potential antimicrobial and antiviral properties. There are numerous studies dedicated to engineering E. coli and S. cerevisiae to produce naringenin, however the best titer of 474 mg/L was reported by Xu et al in 2011. Unlike the number of other studies, Xu et al did more than simply transform the biosynthetic pathway for naringenin into E. coli. They sought to increase carbon flux into the pathway by increasing the precursor, malonylCoA. To do this they overexpressed key enzymes in the native glycolysis pathway to increase the available pool of acetyl-CoA. Then they introduced acetyl-CoA carboxylase from Photorhabdus luminescens to convert the acetyl-CoA into the precursor malonyl-CoA. Additionally, they downregulated genes that would direct carbon into the citric acid cycle. Although Xu et al were able to increase the pool of malonyl-CoA, they did not address the issue of native E. coli pathways that siphon off the available malonyl-CoA. This data along with data for resveratrol production would indicate that one of the limiting factors in the production of polyketides in E. coli is the availability of the precursor malonyl-CoA, and that increasing malonyl-CoA levels increases product titers. However, there is potential for other issues that may not be as clear. For example, it is possible that some of intermediates or the product of the target pathway could negatively impact cellular metabolism.