Although widespread investigations have been carried out, the mechanism of action of artemisinin is still incompletely clarified. It has been widely accepted that the anti-malarial activity of artemisinin is largely dependent on the unusual endoperoxide since derivatives lacking the endoperoxide bridge are discovered to be devoid of antimalarial activity, and the activity could be enhanced by high oxygen tension and by the addition of other free-radical-generating compounds, while some radical scavengers could block the antimalarial activity
[53]. Considerable evidence has proven that the killing parasite’s ability of artemisinin-based combination therapies is mediated by free radicals, which are produced from the endoperoxide bridge
[54]. The degradation of the endoperoxide bridge in a heme-dependent process could form carbon-centered radicals, which then alkylate multiple targets including heme and proteins at the pathogenic
Plasmodium blood stage and lead to the conversion of heme to hemozoin and finally lead to the death of the parasite
[55].
3.2. Key Enzymes Involved in The Biosynthesis Pathway of Artemisinin
The large demand for artemisinin-based combination therapies has caused artemisinin to fall in short supply. To provide more alternative sources, the biosynthesis pathway of artemisinin has been investigated for many years and remarkable achievements have been obtained. Like other regular sesquiterpenes, artemisinin’s biosynthesis precursor is farnesyl pyrophosphate (FPP), which is formed by the condensation of three IPP molecules by either the MVA or plastid non-MVA pathway, respectively
[56]. To verify the origin of the precursors, a plant of
A. annua was grown in an atmosphere containing labeled
13CO
2 for 100 min. Following a chase period of 10 days, artemisinin was isolated and analyzed by
13C NMR spectroscopy. The result shows that the precursor IPP can be provided by both the MVA pathway and the non-MVA pathway. As shown in
Figure 2, DMAPP was initially provided by MVA origin and then transferred to the plastid, where an IPP unit of non-MVA origin is used for elongation to form geranyl diphosphate (GPP). In the subsequent step, GPP is exported to the cytosolic compartment and converted into FPP using IPP from the MVA pathway
[57] (
Figure 2). After FPP is formed, the first committed step of artemisinin biosynthesis is the conversion of FPP to amorphadiene by the terpene synthase enzyme amorphadiene synthase (ADS). To explore the catalysis mechanism of ADS, deuterium-labeled FPP at H-1 position was used as the substrate to trace the H-1 hydrogen migration of FPP during cyclization.
1H NMR results of amorphadiene showed that one of the hydrogen Ha-1 of FPP migrated to H-10 of amorphadiene, while the other hydrogen Hb-1 remained at its position to label amorphadiene H-6. These observations indicated that ADS may act through an initial formation of a bisabolyl cation intermediate through 1,6-ring closure and one 1,3-hydride shift. Bisabolyl carbocation intermediate would then undergo hydride shift through one direct suprafacial 1,3-shift of axial Ha-1 to C-7 (
Figure 2), resulting in the correct
cis-decalin configuration at C-1 and C-6 of amorphadiene
[58][59][60][61]. Following the formation of amorpha-4,11-diene, a cytochrome P450, CYP71AV1, was cloned from
A. annua and characterized by expression in
Saccharomyces cerevisiae. CYP71AV1 could catalyze the multiple oxidation steps of amorpha-4,11-diene to produce artemisinic alcohol and artemisinic aldehyde, and finally yield artemisinic acid
[62]. In addition, two genes encoding putative artemisinic alcohol dehydrogenase (ADH1) and artemisinic aldehyde dehydrogenase 1 (ALDH1) were characterized from
A. annua glandular trichomes
[63]. ADH1 is a NAD-dependent alcohol dehydrogenase of the medium-chain dehydrogenase/reductase superfamily, with specificity towards artemisinic alcohol. ALDH1 could effectively convert artemisinic aldehyde to artemisinic acid
[64].
It is obvious that the Δ11 (13) double bond in amorpha-4,11-diene is reduced during the biosynthesis of artemisinin, which is assumed to occur in artemisinic aldehyde. A corresponding gene,
Dbr2, was cloned and characterized from
A. annua [65]. It could specifically reduce artemisinic aldehyde to produce dihydroartemisinic aldehyde, which could be then converted to dihydroartemisinic acid by ALDH1. Further study showed that ALDH1 could also catalyze the oxidation of artemisinic aldehyde as CYP71AV1 did
[31]. Conversely, CYP71AV1 cannot catalyze the oxidation of dihydroartemisinic aldehyde. Meanwhile, experimental results showed that there was no direct enzymatic conversion of artemisinic acid into dihydroartemisinic acid. Therefore, there should be two branches that exist during artemisinin biosynthesis
[66]. It is well accepted that the primary route is through dihydroartemisinic acid, and the route through artemisinic acid is a side pathway
[67][68][69]. From dihydroartemisinic acid, biosynthesis of artemisinin still requires a photooxidative formation of the endoperoxide ring. However, the details of this process, such as the potential involvement of additional enzyme activities, are currently unclear. In 2004, there was a report that, through using the cell-free extracts of
A. annua, realized the bioconversion of artemisinic acid to artemisinin, but the activity was not observed when using artemisinic acid as the only substrate
[70]. Thus, the enzyme in charge of this reaction is still a question. One possibility is that artemisinic acid could be converted into several other compounds such as arteannuin B non-enzymatically, which is later transformed into artemisinin
[71]. Another possibility is that dihydroartemisinic acid could undergo rapid plant pigment photosensitized oxidation, followed by subsequent spontaneous oxidation to form artemisinin
[72].
3.3. Microbial Production of Artemisinic Acid
On the basis of the biosynthesis pathway elucidation, increasingly more attention on artemisinin is now shifting to its microbial production. Particularly represented by Dr. Jay D. Keasling, his team has made great achievements in this field
[73]. They combined the biological synthesis of the earlier steps to produce the precursor artemisinic acid and the organic synthetic steps of artemisinic acid to produce artemisinin together and realized the industrial production of semi-synthetic artemisinin for commerce needs. They first constructed the biosynthesis pathway of amorphadiene in
E. coli. Compared with the expression of DXP pathway genes, a dramatic increase in isoprenoid precursor production was observed when the
S. cerevisiae MVA pathway was heterologously expressed in
E. coli. Thus, two plasmids were correspondingly designed. One encoded the MevT operon (known as the ‘top pathway’), which comprises three genes (
atoB,
ERG13, and
tHMG1) that are needed for the conversion of acetyl-CoA to MVA. The second plasmid encoded the MevB operon (known as the ‘bottom pathway’) comprising five genes (
idi,
ispA,
MVD1,
ERG8, and
ERG12) for the conversion of MVA to FPP. These two plasmids were subsequently expressed in
E. coli with the codon-optimized amorphadiene synthase (ADS) gene together. Combined with the optimization of the fermentation conditions, the production of amorphadiene could reach 0.5 g per liter in
E. coli [74][75][76]. Following this is the next stage: after the identification of CYP71AV1, this project meets a quandary that although the amorphadiene was produced with a higher yield in
E. coli than in
S. cerevisiae,
E. coli is unsuitable for the expression of the P450 enzyme CYP71AV1, which is crucial for the following steps. Thus, in this stage, Keasling’s team switched the expression system of artemisinin to
S. cerevisiae. Following this, a series of gene manipulations were performed, including: (1) The
S. cerevisiae strain was engineered to overexpress the MVA pathway, and all genes were integrated into the genome; (2)
ADS and
CYP71AV1 genes were constructed as plasmid borne; (3) overexpression of a 3-hydroxy-3-methylglutaryl-CoA reductase (tHMGR) occurred to improve the production of amorphadiene; (4) downregulation of
ERG9 occurred, which encodes squalene synthase, catalyzing the first step in the sterol biosynthetic pathway to inhibit the flux from FPP to sterol; (5) a methionine repressible promoter P
MET3 was used to increase amorphadiene production; (6) the
ADS gene was expressed under the control of the
GAL1 promoter; (7) the
CYP71AV1 gene was expressed along with its cognate reductase (CPR1); (8) yeast strain CEN.PK2 was chosen as the host, which is capable of sporulating sufficiently; (9) every enzyme of the MVA pathway including
ERG20 (the final step for the production of FPP) was overexpressed in CEN.PK2 in an effort to increase the production of amorphadiene; (10) the
GAL80 gene was deleted to ensure constitutive expression of the overexpressed MVA pathway enzymes and the
A. annua-derived genes; (11) the much cheaper glucose was used as the carbon source instead of galactose; (12) another two enzymes, aldehyde dehydrogenase (ALDH1) and artemisinic alcohol dehydrogenase (ADH1), were combinedly expressed with CYP71AV1, which resulted in the highest production yield of artemisinic acid. With all the above manipulations coupled with the development of the fermentation process, the production of artemisinic acid in the engineering yeast strain was finally as high as 25 g per liter
[63][77][78].
3.4. Chemical Conversion to Produce Artemisinin
The final stage for artemisinin chemo-enzymatic synthesis is the chemical conversion of artemisinic acid to artemisinin (
Figure 2). The chemical process involves a four-step conversion that begins with the reduction of artemisinic acid to dihydroartemisinic acid. Then, the esterification of the carboxylic acid moiety will be performed to block the subsequent formation of side products. The third step is an ‘ene-type’ reaction of the C4–C5 double bond with singlet oxygen (
1O
2) to produce an allylic 3-hydroperoxide. Moreover, in the final step, the allylic hydroperoxide undergoes an acid-catalyzed hock fragmentation and rearrangement to afford a ring-opened keto-aldehyde enol. Trapping of this enol with
3O
2 produces a vicinal hydroperoxide aldehyde, followed by a cascade reaction of acid-catalyzed cyclization that could form an endoperoxide bridge to provide artemisinin at last
[63]. Finally, through the metabolic engineering of the earlier steps using multiple gene manipulations and following synthetic organic chemistry, the anti-malaria drug artemisinin production system was successfully established and effectively used for industrial production by Sanofi company as the worldwide supplement
[73]. Artemisinin is by far the most successful and representative example of the perfect combination of biosynthetic pathway research and industrial production.
4. Biosynthesis Pathway Investigation of the Terpenoid Antimicrobial Agent—Oleanolic Acid
Oleanolic acid (
Table 1) is a pentacyclic triterpenoid originating from a number of medicinal plants. It has desired antimicrobial activity against various bacterial pathogens and viruses
[15][79][80][81][82]. Furthermore, the study on this antimicrobial agent is of importance because as a natural source product, there has been no resistance case toward oleanolic acid found yet
[83]. The biosynthesis pathway of oleanolic acid has been relatively clear
[84].
In plant cells, acetyl CoA generates DMAPP and IPP through the MVA pathway in the cytosol. IPP and DMAPP are isomerized into FPP under the action of farnesyl pyrophosphate synthase (FPS), and FPP is then converted into squalene under the action of squalene synthase (SQS). Squalene cyclooxygenase (SQE) then oxidizes squalene into a precursor molecule for primary sterol metabolism, 2,3-oxsqualene
[85]. From this step, the different cyclizations of 2,3-oxidized squalene become a branching point between primary sterol and secondary triterpene metabolism. For the biosynthesis of plant sterols, the cyclization of 2,3-oxysqualene to the tetracyclic plant sterol precursor cycloartenol is mainly catalyzed by cycloartenol synthase (CAS)
[86]. Conversely, the oleanolic acid biosynthetic pathway of the researchers' interest, 2,3-oxysqualene, is cyclized by
β-amyrin synthase (BAS), which was first cloned from the medicinal plant ginseng and subsequently from a variety of other plants
[86][87]. This pentacyclic carbon skeleton is assumed to be formed from (3
S)-2,3-oxidosqualene folded in pre-chair–chair–chair conformation
[88]. Opening of the epoxide ring followed by cation–
π cyclization initially produces a tetracyclic dammarenyl cation. Following ring expansion and the formation of fifth ring, the lupenyl cation is formed
[87]. Another ring expansion followed by a series of stereospecific 1,2-hydride shifts and the final abstraction of 12α proton produces β-amyrin
[89] (
Figure 3). The
C-28 position of β-amyrin is then oxidized in three consecutive steps by a single cytochrome P450 enzyme, CYP716A12, to produce oleanolic acid. The key enzyme for this step—CYP716A12—was first identified in
Medicago truncatula, and the study found that erythrodiol, oleanolic aldehyde, and oleanolic acid production were detected in the reaction solution catalyzed by this enzyme
[90][91]. Thus, it is suggested that CYP716A12 is a C-28 oxidase of β-amyrin, catalyzing three sequential oxidation reactions of oleanane main chain C-28 rather than a one-step generation. The oleanolic acid biosynthetic pathway is shown in
Figure 3.
Figure 3. Biosynthesis pathway of terpenoid antimicrobial agent oleanolic acid. FPS, farnesyl diphosphate synthase; SQS, squalene synthase; SQE, squalene epoxidase; CAS, cycloartenol synthase.
With the development of synthetic biology, some conventional biosynthetic pathways were interfered with using genetic engineering to improve the target compound’s production. For example, limonene, a cyclic monoterpene of plant origin, is antimicrobially sensitive to
Listeria monocytogenes and can damage its cell integrity and wall structure
[92]. The most classical biosynthetic pathway of limonene is the condensation of IPP and DMAPP to form GPP by the action of geranyl pyrophosphate synthase, and limonene synthase (LS) uses GPP as a substrate to synthesize limonene. However, GPP can also subsequently condense with a molecule of IPP to form FPP, and studies have shown that the synthesis of excessive FPP hinders the efficient synthesis of monoterpenes. According to a recent report, researchers have developed an FPPS mutant (F96W, N127W; FPPS
F96W, N127W) that can selectively produce GPP without further extension to FPP. In the yeast strain with high isoprene production, fpps
F96W, N127W genes were combined with nine plant LS genes, and the
N-terminal sequence of plasma-membrane-targeted transport peptide (TLS) was truncated. The best effect of 15.5 mg L
−1 limonene on
Citrus lemon tls1 (cltls1) was achieved. Moreover, an orthogonal engineering pathway was constructed. In this pathway, limonene could be produced through the condensation of IPP and DMAPP by neryl pyrophosphate (NPP) synthase to form NPP, and limonene synthase can also use NPP as a substrate to synthesize limonene. The expression of
Solanum lycopersicum nerolidyl diphosphate synthase (SlNDPS1) and
Citrus limon tLS2 (CltLS2) genes in the same yeast strain made the limonene yield higher than that of traditional methods (28.9 mg L
−1). Under the action of glucose-induced promoter HXT1, the production of limonene can be increased to more than 900 mg L
−1 by extensive pathway engineering using the FPPS competitive gene
[93].