Figure 1. A picture of Quillaja saponaria in Norfolk by Phil Robinson (JIC)

QS-21 is a complex molecule made of a triterpene scaffold that is decorated with seven sugars and a long and unusual acyl chain harbouring a further eighth sugar. Its amphiphilic nature, derived from the hydrophobic triterpene and acyl chain moieties and the hydrophilic sugars, is at the root of its immunoadjuvant properties. Indeed, QS-21 is now a key ingredient in the formulation of vaccines targeting human diseases such as shingles, COVID-19 and malaria. The demand for QS-21 is expected to keep rising in the future as further QS-21-containing vaccines are approved. Due to its increasing importance in the vaccine world, we endeavoured to identify the genes responsible for the biosynthesis of QS-21. This was a very challenging project due to the complexity of the molecule and the lack of genomic resources. To tackle the challenge, a multidisciplinary team was assembled in the Osbourn lab, so I joined the project to uncover the genes responsible for the biosynthesis of the acyl chain. This acyl chain is very unusual, as only Quillaja is known to make such a structure. I gathered the help of chemists to predict the biosynthetic route for this branched chain 18-carbon-long (C₁₈) group. The branched structure was reminiscent of isoleucine and the chain is a pseudo-dimer, suggesting that two shorter and identical C₉ chains assemble to form the C₁₈ chain.

In parallel, I sampled different parts of a Quillaja saponaria plant that we obtained from a nursery in the UK and measured their QS-21 content.  I found that different tissues contained different amounts of QS-21, opening up the possibility of searching for candidate biosynthetic genes based on correlations between high QS-21 content and high gene expression levels.  I generated transcriptome data for the different tissues.  We also sequenced the genome of the soapbark tree and used the transcriptomic data to support genome assembly and annotation. In a previous study, published in Science¹, we describe how the first genes of the pathway were identified. I could then use these genes as bait for bioinformatics analysis to identify other co-expressed candidate pathway genes that may encode enzymes with potential roles in the formation of the acyl chain. As the early pathway were known to be highly expressed in the primordial tissue, I also prioritised candidate genes with strong expression in that tissue.  We had also mined the Q. saponaria genome for predicted biosynthetic gene clusters, since the genes for some plant natural product pathways are known to be physically clustered in plant genomes.  Indeed some of the early QS-21 pathway genes were clustered in the genome.  We therefore kept our eyes open for any additional candidate genes that were located in these gene clusters, following a guilt by association strategy.  To avoid missing anything, I casted a wide net and cloned dozens of candidate genes with the aim of testing their activities in heterologous host systems, namely yeast and N. benthamiana. The latter being particularly central to this work as expressing genes in a similar system circumvent many shortcomings.

I first identified a carboxy-CoA ligase that activates 2-methylbutyric acid, a degradation product of isoleucine, and six enzymes from the same family that catalyse the next step. But then things got complicated. The order of the following enzymatic reactions was unknown, especially when and how the acyl chain was attached to the glycosylated triterpene scaffold. So I gathered all the Agrobacterium strains containing my candidate genes and the genes necessary to biosynthesise the scaffold and the acyl chain precursor and I coinfiltrated them (around 80 strains) in a single leaf of Nicotiana benthamiana. This was pushing the limits to what had been attempted in the past. That many strains mixed together into a single solution added up to an absorbance of 17, which was the thickest I have ever infiltrated in a N. benthamiana leaf! After the deed was done, I waited four days for the genes to be transferred into N. benthamiana nuclei and to be expressed by the N. benthamiana machinery.  Incredibly, I detected a tiny amount of QS-21, indicating that all the essential genes were among the pool of candidates. It was then a matter of repeating infiltrations with narrowing subsets of candidates to finally find the five remaining enzymes necessary to complete the biosynthetic pathway of QS-21. To prove beyond a doubt that we did produce QS-21 by heterologously expressing Quillaja saponaria genes in N. benthamiana, we wanted to produce enough of it for purification and subsequent NMR analysis. The problem was that the yield was low, so pathway optimization was necessary. We did that by increasing the substrate levels of CCL1, either by supplying the agrobacterium strain mix with commercially available substrate, either by co-infiltrating a feedback-insensitive upstream enzyme. We could then make enough of QS-21 to generate unambiguous NMR data, opening the door to free-from-tree production of QS-21.

Our paper “Complete biosynthesis of the potent vaccine adjuvant QS-21”² ends the quest of unlocking the biosynthetic route that Quillaja saponaria employs to produce QS-21. This new knowledge not only opens the door to heterologous production of QS-21, promising a future where its supply isn’t constrained by natural sources, but also will enable precise structure/function studies to better understand how QS-21 triggers such a powerful immune response when included in a vaccine formulation.

¹ Reed J. et al. 2023. Elucidation of the pathway for biosynthesis of saponin adjuvants from the soapbark tree. Science. 379(6638) pp1252-1264

²Martin, L.B.B., Kikuchi, S., Rejzek, M. et al. Complete biosynthesis of the potent vaccine adjuvant QS-21. Nat Chem Biol (2024). https://doi.org/10.1038/s41589-023-01538-5