Auxin-mediated protein depletion for metabolic engineering in terpene-producing yeast

In metabolic engineering, loss-of-function experiments are used to understand and optimise metabolism. A conditional gene inactivation tool is required when gene deletion is lethal or detrimental to growth. Here, we exploit auxin-inducible protein degradation as a metabolic engineering approach in yeast. We demonstrate its effectiveness using terpenoid production. First, we target an essential prenyl-pyrophosphate metabolism protein, farnesyl pyrophosphate synthase (Erg20p). Degradation successfully redirects metabolic flux toward monoterpene (C10) production. Second, depleting hexokinase-2, a key protein in glucose signalling transduction, lifts glucose repression and boosts production of sesquiterpene (C15) nerolidol to 3.5 g L−1 in flask cultivation. Third, depleting acetyl-CoA carboxylase (Acc1p), another essential protein, delivers growth arrest without diminishing production capacity in nerolidol-producing yeast, providing a strategy to decouple growth and production. These studies demonstrate auxin-mediated protein degradation as an advanced tool for metabolic engineering. It also has potential for broader metabolic perturbation studies to better understand metabolism.


Supplementary result 2: Variation of Hxk2p-engineered yeast
To test the effect of Hxk2p depletion on nerolidol production, we constructed the strain NLD128 expressing TIR1 and with AID*-CUP1 fused to the C-terminal of Hxk2p. Consistent with the reference NLD401, the genes from the mevalonate pathway, farnesyl pyrophosphate synthase gene ERG20, and Y-FAST-2A-AcNES1 were controlled by GAL promoters in NLD128. We firstly characterised four NLD128 clones ( Supplementary Figure 2b-d). On average, ~2.4 g L -1 nerolidol was produced in flask cultivation. However, a dramatic variation was observed among these four clones. The first clone, NLD128-1, produced ~3.6 g L -1 nerolidol in flask cultivation with NAA added at 0 hour and in precultures, whereas other clones produced much less amount. The Y-FAST fluorescence levels in the low-productivity clones were much lower, compared to that in NLD128-1 and induced Y-FAST fluorescence levels in the reference strain NLD401. Further analyses are necessary to elucidate the genetic basis responsible for this variation.
To validate the repeatability of the experiments, we re-transformed the strain o128R to generate the strain NLD128 (Supplementary Table 5). Seven clones were replicated on YNB-glucose agar supplemented with HMBR. We examined the fluorescence from these seven colonies using a Safe hour and in precultures, and in NLD128 L, ~2.5 g L -1 nerolidol was produced. Y-FAST fluorescence in NLD128H were much higher than that in clones NLD128L. HXK2 modifications in NLD128H and NLD128L was confirmed by DNA sequencing, and the sequence was as being designed. The causes of this variation might be due to the variations in plasmid copy number or other genetic loci.

Supplementary result 3: Induction of GAL promoter in terpene-producing strains
In terpene-producing strains, GAL80 was disrupted to enable auto-induction of GAL promoters, which were used to regulate terpene anabolic pathways, after glucose depleted. Examining the induction of the GAL promoter is important for understanding the transcriptional responses to altered cultivation conditions and genetic modifications in these strains. We employed Y-FAST to examine the induction of the GAL promoter in flask cultivation (Figure 2d, 3f, 4d, & 4h) by analysing the mean value of Y-FAST fluorescence of populations. However, these data did not reflect the induction in individual cells.
Here, we further analysed the data to understand the GAL promoter induction in yeast populations present in flask cultivation.
In limonene-producing strain LIM401 (without ERG20 modified) and LIM141 (with engineered depletion of Erg20p), Y-FAST was controlled by the GAL10 promoter. In the exponential growth phase when glucose was used as the carbon source, Y-FAST expression was repressed in both and then Similar to the GAL10 promoter, the GAL2 promoter showed a binary induction in reference nerolidolproducing strain NLD401, when glucose was used as the carbon source (Supplementary Figure 3d).
When ethanol was used as the carbon source, the GAL promoter is supposed to be fully derepressed.
However, a binary induction of the GAL promoter was also observed in strain NLD401 when ethanol was used as the carbon source (Supplementary Figure 3e).
In nerolidol-producing strain NLD128 with engineered depletion of Hxk2p, the GAL promoter was

Supplementary result 4: Viability of terpene-producing strains in flask cultivation
To test whether depleting target proteins affected the strain fitness, we employed a flow cytometrybased propidium iodide exclusion assay to examine the viability of cells in flask cultivation (Supplementary Table 6).
For limonene-producing strain LIM401 (without ERG20 modified) and strain LIM141 (with induced depletion of Erg20p), the viability at 72 hour and/or 96 hour were assayed with glucose used as the carbon source. Strain LIM141 showed the survival rates higher than that in strain LIM401 (Supplementary Table 6#a-e).
For nerolidol-producing reference strain NLD401 (without target gene modified), we examined the influence of auxin on cell viability when glucose was used as the carbon source. Compared to the strain free of auxin, adding auxin had minor effect on cell viability (Supplementary Table 6#f Table 6#g).
Similarly, the nerolidol-producing strain NLD138 with engineered depletion of Acc1p did not showed a remarkable difference on cell viability (Supplementary Table 6-#j-l).
When ethanol used as the carbon source, strain NLD401 and strain NLD138 did not show a significant difference in cell viability (Supplementary Table 6#m-n).
These showed that in our cases, auxin-induced depletion of target proteins did not cause dramatic change on cell viability. However, the terpene-producing strains showed a fitness problem, as shown by a lower viability compared to farnesene-producing yeast previously reported 8 .