The question asks specifically why certain plant products are not produced commercially in genetically modified micro-organisms. There are some general reasons, illustrated in some of the examples mentioned:
- The product is actually a complex mixture, rather than a single compound. This is the case for saffron, preparations of which contain protein, fat, fibre, sugars and gums, in addition to the the bioactive components — thought to be crocin, crocetin, picrocrocin and safranal (see this review).
- It is much easier and cheaper to obtain the product by growing the plant. Although saffron is expensive because it is obtained by hand from crocus blossoms, other plant products do not have this problem, e.g. cotton, mentioned in another answer.
- The overall market may not justify the effort. Ignoring other considerations the worldwide consumption of saffron, for example, is small, whereas that for vanillin is huge.
- Highly developed chemical production methods may already exist. This is the case for rubber, the chemical production of varieties of which was stimulated by the development of tyred vehicles, and subsequently by war-time demands. Natural rubber is actually a mixed polymer, and specific polymers have been used to develop types of synthetic rubber with individual desirable properties.
Production of Secondary Metabolites
The question mentions “quinine, tetrahydrocannabinol and cocaine” — specific secondary metabolites of plant metabolism which may be of value in medicine, but the complex structure of which renders them difficult to synthesize chemically, much less on a commercial scale. This seems the most pertinent area to consider, especially as the question mentions the successful production in bacteria of another secondary metabolite, vanillin.
I shall take a single example of the successful synthesis in yeast of a pair of secondary metabolites — tropane alkaloids — hyoscyamine and scopolamine. These two compounds (hyoscyamine is shown) have much more complex structures than vanillin:
This is reflected in the fact that their synthesis entails 26 enzymes, compared to the 15 use in the engineered synthesis of vanillin. The synthesis of these more complex secondary metabolites illustrates some problems encountered in the production of such compounds from plant genes cloned into micro-organisms not encountered with vanillin, and not fully covered in other answers thus far:
- Identifying the genes for all the enzymes involved in a uncommon biosynthetic pathway
- Modularizating the pathway between different organelles to mimic the separation that occurs in plants
- Maintaining concentrations of precursors and intermediates that might be removed by the yeast cell’s own metabolism
The example to be described was published in an article in Nature in 2020 by Srinivasan and Smolke, and reports the synthesis of the medicinal alkaloids hyoscyamine and scopolamine starting from simple sugars and amino acids. These compounds occurs in plants of the families nightshade (Solanaceae), coca (Erythroxylaceae) and bindweed (Convolvulaceae), and currently they are produced commercially by extraction from crops grown in tropical and sub-tropical countries.
Identifying the Genes
In many biosynthetic pathways the genes responsible for individual steps can be identified by homology of the predicted protein products to those of known genes. In addition, the genes for biosynthetic pathways are often clustered. However, plants which produce useful products may be relatively obscure and their genomes poorly characterized. In the case of the synthesis of hyoscyamine in the nightshades, Solanaceae, the enzyme responsible for catalysing the final step of the 26 step pathway was unknown. Srinivasan and Smolke had to identify candidate genes from gene-expression patterns under different conditions and then, for the 12 candidates identified, express their gene products and assay for the required hyoscyamine dehydrogenase that catalyses this step.
Modularizing the Pathway
In plants the synthetic pathway does not take place in a single location, but is compartmented between different organelles and, indeed, different tissues. The latter is, of course, not possible in yeast, but was mimicked by directing the products of these 26 genes to six different subcellular locations by engineering appropriate targeting signals onto them. These were the cytosolic fluid, the mitochondrion, the peroxisome, the vacuole, the endoplasmic reticulum, and the vacuolar membranes. This is one reason why yeast rather than bacteria (which lack subcellular organelles) were used in this work.
The purpose of this is explained in a Nature News & Views piece from which the diagram above is taken:
“Subcellular compartmentalization of enzymes can improve product
biosynthesis by enabling proper enzymatic activity and isolating
metabolic intermediates to reduce their toxicity and loss to competing
pathways. By restricting space, compartmentalization also increases
local interactions between the enzymes and their targets. …so each
step can be separately optimized to maximize productivity.”
Maintaining concentrations of precursors and intermediates
Expression (in fact over-expression) of the enzymes of a pathway is not enough to ensure adequate production of the end-product. The precursors for the pathway (here glucose and phenylalanine) must be continually available at a suitable concentration, and reactions which consume key intermediates must be supressed. To this end eight other genes were disrupted in the work in question.
The example described is a major tour de force, showing the power of recombinant molecular genetics to express genes in micro-organisms for production of secondary metabolites. Possible it is, but routine it certainly is not. In addition to overcoming the types of problems described, there remains the task of obtaining an efficiency of production that it commercially viable. The authors of this work write:
“Process improvements to increase productivities from titres reported
here (around 30 to 80 μg per litre)… to commercial production
(approximately 5 g per litre)…we anticipate would take 1–2 years of
focused effort by a professional team.”