Figure 4 : Sensitivity analyses of oleochemical pathways. We
used a global sensitivity analysis to examine the influence of enzyme
concentrations on three distinct biochemical objectives: total
production, unsaturated fraction, and average length of oleochemical
products. In our models, total production is most strongly influenced by
the concentrations of either (i) the thioesterase and a single
product-specific enzyme or (ii) two product-specific enzymes, while
unsaturated fraction and average chain length are most sensitive to the
concentrations of core FAS enzymes.
profiles. For the methyl ketone pathway, our results indicate that
thioesterase concentration has an outsized influence on both total
production and chain length, but also suggest that FadD and
β-ketoacyl-CoA thioesterase (FadM) can affect both objectives. This
finding is consistent with in vivo data showing that the
thioesterase, FadD, and FadM need overlapping substrate specificities to
maximize the production of specific chain lengths. Importantly, severalin vivo studies have showed that overexpression of acyl-coenzyme
A dehydrogenase (FadE) can improve titers of methyl ketones, but our
sensitivity analysis does not reproduce this influence; this discrepancy
suggests that FadE may be overactive in our model (Yan et al., 2020). We
conclude our brief comparison of general effects by noting that in both
model and experiment, the concentrations of final enzymes in the FAME
and FAEE pathways (i.e., the O-methyltransferase (MT) and the wax
synthase (WS), respectively) strongly influence total production,
perhaps a result of their low activities, relative to other enzymes
(Sherkhanov et al., 2016; Steen et al., 2010).
We used our models to investigate several intriguing effects fromin vivo studies. An experimental study of the alkane pathway
examined different expression levels of acyl-ACP reductase (AAR) and
(ii) an aldehyde deformylating oxygenase (ADO) by using alternative
plasmids, promoters, and pseudo-operon configurations (Song et al.,
2016). Shifts in the expression of both enzymes had a prominent
influence on total production, but not product profile. We recreated
this effect by changing the AAR:ADO ratio in our model; production
peaked at an intermediate ratio, while average chain length stayed
constant (Figs. 5A-5B). This behavior suggests that AAR and ADO catalyze
rate-limiting steps and must be overexpressed at similar levels to
improve titer. Their inability to affect product profile, in turn,
probably reflects their narrow substrate specificities (i.e.,
C16 and C18). Intriguingly, the same
experimental study also showed that the overexpression of FabH and FabB
can reduce alkane production. Our model