The overall goal of our studies is to enable the use of carbon sources present in hydrolysates of lignocellulosic biomass for triacylglycerides (TAG) production. Here we have focused on understanding the individual and synergetic effect of inhibitors found in lignocellulosic hydrolysates on the growth.
Construction and characterization of a xylose consuming strain (SZYL004).
The efficient utilization of lignocellulosic hydrolysate requires that xylose can be used as a carbon source. Therefore, we genetically engineered Y. lipolytica to improve xylose catabolism. The genome of Y. lipolytica contains open reading frames for all three proteins required to catabolize xylose: xylose reductase (XR; YALI0D07634/YALI1_D09870g), xylitol dehydrogenase (XDH; YALI0E12463/YALI1_E15452g), and xylulose kinase (XK; YALI0F10923/YALI1_F14583g) [22]. These genes are homologous to those of yeast species able to metabolize xylose by a pathway in which, after uptake, D-xylose is first reduced to xylitol by XR and further oxidized to D-xylulose mediated by XDH. In a final step, D-xylulose is then phosphorylated by XK to yield D-xylulose-5-P which can enter the pentose phosphate pathway [7]. However, many strains of Y. lipolytica do not grow on xylose, or their growth is highly depended on culturing conditions and pre-adaptation [7].
We constructed a xylose consuming strain (SZYL004) by overexpressing the three native xylose genes XR, XK, and XDH, under the control of pPYK1, pTEF1, and pGAPDH promoters respectively. An overexpression of only XK and XDH resulted in only minor growth on xylose (Figure S1), suggesting that the native expression level of XR is not sufficient which is in contrast to previous reports [23, 24]. The xylose consuming strain (SZYL004) showed an increased lag phase when cultivated in media containing only xylose as a carbon source (Figure S2), likely because of an insufficient transport of xylose into the cell, which could be further improved by overexpressing xylose transporter proteins [25]. This lag phase could be prevented using media containing both glucose (2%) and xylose (8%). In this media, SZYL004 did show an increased lipid production (up to 60%) and increased biomass production compared to its parental strain OKYL049 (Fig. 2).
The tolerance of Y. lipolytica to individual inhibitors
To determine which inhibitor classes pose a problem for Y. lipolytica, we decided to evaluate the effect of a set of individual representative inhibitory compounds (chosen from [26]). First, we tested the effect of six single inhibitors in different carbon sources to assess the tolerance range of Y. lipolytica. Then the inhibitors were combined to mimic the composition of a real hydrolysate [26].
Effect of the overexpression of the xylose genes on the inhibitor tolerance
Inhibitor tolerance in the xylose engineered strain (SZYL004) and its parental strain (OKYL049) were evaluated to exclude that the overexpression of the three native xylose genes (XR XK, and XDH) affect inhibitor tolerance, e.g. by affecting the central metabolism. This was investigated by comparing the inhibitor effects on the xylose engineered strain SZYL004 and its parental strain OKYL049 during growth on glucose. Both strains showed a very similar growth behavior in all tested inhibitors (Fig. 3), indicating that the overexpression of the three native xylose genes does not significantly affect inhibitor tolerance of Y. lipolytica.
Effect of the carbon source on the inhibitor tolerance
Xylose catabolism affects tolerance to furfural, coniferyl aldehyde and acetic acid, but not to HMF, formic acid, and cinnamic acid (Fig. 3). It is possible that differences in metabolism caused by the choice of carbon source could affect inhibitor tolerance. The strain growing on xylose showed a lower growth rate and a longer lag phase than the strain growing on glucose. However, no major carbon source dependent differences in inhibitor tolerance were observed in the presence of either formic acid, cinnamic acid and HMF (Fig. 3).
However, when SZYL004 was growing in the presence of furfural with glucose or xylose as carbon and energy source a clear difference was observed (Fig. 3). At the same furfural concentration, SZYL004 growing on xylose displays a longer lag phase than during growth on glucose. For example, at 30 mM furfural SZYL004 on glucose grows after 70 h, while on xylose only grows after 115 h. It is likely that the introduction of the two genes XR and XD lead to cofactor imbalance compared to growth on glucose, since XR uses NADPH and XDH uses NAD+ as a cofactor [24]. Moreover, Y. lipolytica cannot convert cytosolic excess of NADH into NADPH [27]. Additionally, furfural detoxification causes a decrease in the intracellular levels of NADPH in S. cerevisiae [10]. The combination of these two factors increases the burden on the cell while metabolizing xylose, which could explain the higher resistance to furfural of SZYL004 on glucose than on xylose.
For coniferyl aldehyde, we saw a tendency to less sensitivity during growth on xylose compared to glucose. In glucose, there was a small increase in lag phase up to concentrations of 1 mM, which was not observed during growth on xylose (Fig. 3).
A lower sensitivity of xylose grown cells was also observed for acetic acid. While in glucose a concentration of 37.5 mM acetic acid increased the lag phase, this was not observed on xylose (Fig. 3).
Effect of weak acids on the growth of Y. lipolytica
Acetic acid was well tolerated by Y. lipolytica up to 18.75 mM. At concentrations between 18.75 mM and 56.25 mM, both the lag phase and growth rate were affected. Acetic acid did not allow growth at concentrations of 75 mM (Fig. 3).
Formic acid showed similar effects as acetic acid, but the tolerance range was narrower. At concentrations up to 18.75 mM formic acid did not affect the growth of Y. lipolytica. Formic acid did not allow growth at concentrations of 37.5 mM and higher (Fig. 3). To further characterise the formic acid sensitivity, intermediate concentrations were tested in xylose containing media. The results show that growth was possible up to a concentration of 30 mM formic acid. However, there was a major increase in the lag phase (25 mM: 18 h; 30 mM: 72 h) (Fig. 4 a). This shows that there is a threshold for formic acid tolerance in our Y. lipolytica strain of approximately 25 mM.
Acetic acid and formic acid are toxic compounds because they lead to intracellular anion accumulation. The undissociated form of the acid can diffuse throught the plasma membrane and dissociate inside the cell, leading to anion accumulation. Formic acid shows higher toxicity than acetic acid likely due to a lower pKa value (3.75 at 25 °C) than acetic acid (4.75 at 25 °C). Therefore, at the same molarity, the intracellular pH will be more affected by formic acid than by acetic acid [12]. Formic acid also has a smaller molecular size than acetic acid and thus can diffuse more easily through the plasma membrane [12]. Formic acid has been shown to induce ROS production and to induce apoptosis in S. cerevisiae [28]; a similar effect might contribute to the formic acid toxicity in Y. lipolytica. Furthermore, acetic acid could be less toxic than formic acid because the dissociated form (acetate) can be activated to acetyl-CoA in the cytoplasm by acetyl-CoA synthetase and then be incorporated into lipids or transported into the mitochondria to enter the tricarboxylic acid (TCA) cycle [29].
The tolerance of another oleaginous yeast, Rhodosporidium fluviale, was previously tested and it was found that its growth was strongly affected at a concentration of 10.9 mM formic acid and 16.7 mM of acetic acid [14]. Thus, Y. lipolytica seems to display an advantage over both S. cerevisiae and R. fluviale in this respect.
Effect of aromatic compounds on the growth of Y. lipolytica
Cinnamic acid shows strong effects on the growth at concentrations above 2.5 mM, while coniferyl aldehyde has strong effects at 2.5 mM.
The inhibitory effects of cinnamic acid on Y. lipolytica was initially tested for concentrations up to 1 mM (148 mg/L) (Fig. 3). In that range, cinnamic acid did not show any effect on the growth parameters. As a follow up we also tested cinnamic acid concentration of 2.5 mM and 5 mM to find the upper limit of cinnamic acid tolerance for Y. lipolytica (Fig. 4 b). Y. lipolytica displayed only a slight reduction in growth rate and yield in 2.5 mM cinnamic acid (370 mg/L). However, at concentrations of 5 mM, the overall growth was strongly affected.
The high tolerance towards cinnamic acid is not only interesting in the context of hydrolysate utilization, but also for other metabolic engineering applications. For instance, cinnamic acid can be converted to p-coumaric acid, which is an intermediate in the production of flavonoids, e.g. kaempferol or naringenin [30]. Previous studies on S. cerevisiae suggest that increasing concentrations of cinnamic acid up to 35 mg/L led to an increase of the lag phase and a gradual reduction of the growth rate [31]. Even though with engineering efforts S. cerevisiae tolerance to cinnamic acid can be improved [18], the high cinnamic acid tolerance of Y. lipolytica could allow high production of flavonoids in the future with less engineering efforts.
The inhibitory effects of coniferyl aldehyde on Y. lipolytica were at first also tested for concentrations only up to 1 mM (178 mg/L). These concentrations had a minor impact on the growth on glucose, but not on xylose-based media (Fig. 3). To explore the limit of coniferyl aldehyde tolerance, concentrations of 2.5 mM (445 mg/L) and 5 mM (891 mg/L) were tested. In 2.5 mM Y. lipolytica showed a prolonged lag phase, but a similar growth rate and maximum OD600 compared to that of the control (Fig. 4 c). Growth was not detectable at coniferyl aldehyde concentrations of 5 mM even after 120 h of cultivation. The mechanism of coniferyl aldehyde toxicity has not been fully elucidated, but a correlation with ROS accumulation has been observed. A transcriptomic analysis of a strain of S. cerevisiae evolved for tolerance to CA showed an upregulation of 11 genes with a role in oxidative stress response [32].
Effect of furanic aldehydes on the growth of Y. lipolytica
Furfural increases the lag phase in a linear matter, while HMF reduces the growth rate. To investigate the effect of furfural, our strains were cultivated in media containing different concentrations of furfural (Fig. 3) and we found that every 1 mM of furfural increased the lag phase of SZYL004 by 2.4 h and 3.7 h with glucose or xylose as a carbon source, respectively (Fig. 5 a). Y. lipolytica cannot metabolize furfural or HMF to enter cellular metabolism, but has non-specific enzymes such as reductases and dehydrogenases that can reduce or oxidase these compounds to less inhibitory alcohols (furfuryl alcohol and HMF alcohol) or acids (furoic acid and HMF acid) [33, 34]. After inhibitor degradation, cells show normal growth. Increasing the inoculum density yielded faster detoxification and a shorter lag phase (Fig. 5 b and c) and can be considered one way to handle the inhibitory effects of furfural in hydrolysates, as previously suggested [35].
The addition of HMF led to a complex phenotype regarding the growth parameters (Fig. 3). With increasing HMF concentration we observe two different growth phases. While an increase of lag phase caused by furfural corresponded to a lack of all growth, in HMF we observe a low growth rate until an OD600 of approximately 0.3, at which point the growth rate increased up to a rate similar to the control. The most straightforward explanation for this behavior is that HMF only partially inhibits the growth until the HMF is detoxified, after which normal growth resumes.
Previous studies testing the tolerance of Y. lipolytica found no tolerance to furfural when testing its growth on 0.5 g/L furfural (5 mL media in test tubes at 23 °C) [36]. Our strain showed an extended lag phase but grew in furfural concentrations up to 2.9 g/L. These results indicate that there are differences in tolerance between different strain isolates.
Effect of an inhibitor mixture on the growth of Y. lipolytica
During the process of hydrolysate treatment, multiple inhibitors are formed, which affect the growth performance of microorganisms in a synergistic manner [8]. Y. lipolytica SZYL004 was tested during growth on xylose based media, supplemented with an inhibitor mixture mimicking the composition of a real hydrolysate [26](Fig. 6). The concentrations of the inhibitors correspond to 20 percentage of the maximum concentrations that were initially tested in previous experiments. The strain did not grow in the full inhibitor mixture. We therefore decided to evaluate the contribution of the individual inhibitory compounds in causing the growth phenotype. To this end, we evaluated growth in mixtures containing only five of the six inhibitors, leaving out one inhibitor at a time. We observed that omission of any of the six different inhibitors could partially rescue the growth (Fig. 6). The omission of furfural showed the strongest impact on the growth, followed by coniferyl aldehyde, and formic acid. These results are not surprising for furfural and formic acid since in previous experiments with single inhibitory compounds both of them showed a strong impact on cell growth (Fig. 3). However, that coniferyl aldehyde omission resulted in such a strong recovery of growth was highly surprising, since we observed high tolerance (up to 2.5 mM) in previous experiments with single inhibitors (Fig. 4). This implies that inhibitor toxicity is highly dependent on synergistic effects. Phenolic compounds (such as coniferyl aldehyde) [20], furfural [9], and formic acid [28] have all been reported to induce ROS formation. Therefore, it is likely that the synergistic effects we observe could be linked to ROS-related effects. While the amount of ROS generated by each inhibitor might be tolerated, combining multiple inhibitors could put a burden on the cell that leads to growth arrest.