The wide‐ranging phenotypes of ergosterol biosynthesis mutants, and implications for microbial cell factories

Yeast strains have been used extensively as robust microbial cell factories for the production of bulk and fine chemicals, including biofuels (bioethanol), complex pharmaceuticals (antimalarial drug artemisinin and opioid pain killers), flavours, and fragrances (vanillin, nootkatone, and resveratrol). In many cases, it is of benefit to suppress or modify ergosterol biosynthesis during strain engineering, for example, to increase thermotolerance or to increase metabolic flux through an alternate pathway. However, the impact of modifying ergosterol biosynthesis on engineered strains is discussed sparsely in literature, and little attention has been paid to the implications of these modifications on the general health and well‐being of yeast. Importantly, yeast with modified sterol content exhibit a wide range of phenotypes, including altered organization and dynamics of plasma membrane, altered susceptibility to chemical treatment, increased tolerance to high temperatures, and reduced tolerance to other stresses such as high ethanol, salt, and solute concentrations. Here, we review the wide‐ranging phenotypes of viable Saccharomyces cerevisiae strains with altered sterol content and discuss the implications of these for yeast as microbial cell factories.

Strains lacking or producing alternative sterols exhibit a range of phenotypes. In recent years, yeast strains have been engineered as microbial cell factories for a wide variety of heterologous products, and a common approach to increase strain productivity is to suppress native pathways that compete for substrate. In many cases, particularly for terpene engineering, ergosterol biosynthesis has been suppressed in production strains (Moses et al., 2014;Paddon et al., 2013). It is therefore industrially relevant to consider the impact of altered sterol compositions on yeast health, in order to optimize production strains and processes.
This article provides an overview of ergosterol biosynthesis in the budding yeast S. cerevisiae, reviews the phenotypes of viable strains

| ERGOSTEROL BIOSYNTHESIS IN SACCHAROMYCES CEREVISIAE
The biosynthesis of ergosterol from acetyl-CoA is often grouped into three modules: (a) the mevalonate biosynthesis module, (b) the farnesyl pyrophosphate (FPP) biosynthesis module, and (c) the ergosterol biosynthesis module (Figure 1; reviewed in Hu et al., 2017). The ergosterol biosynthetic enzymes are localized in different subcellular compartments of the yeast cell, with Erg proteins of the first, second, and third modules predominantly located in the mitochondria (Isamu, Jun, Hiroshi, & Hirohiko, 1973), cytoplasm and endoplasmic reticulum (ER; Nishino, Hata, Taketani, Yabusaki, & Katsuki, 1981), respectively.
The mevalonate biosynthesis module consists of three steps and begins with the condensation of two acetyl-CoA molecules to acetoacetyl-CoA, catalysed by acetyl-CoA acetyltransferase (Erg10).
The hydroxymethylglutaryl-CoA synthase enzyme (Erg13) then F I G U R E 1 Ergosterol biosynthesis in Saccharomyces cerevisiae Reactions of the mevalonate biosynthesis module (red), farnesyl pyrophosphate (FPP) biosynthesis module (yellow) and ergosterol biosynthesis module (green). The enzyme names are given in red. Enzymes which are non-essential for growth in aerobic conditions are emboldened catalyses the condensation of a third molecule of acetyl-CoA to form 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA), which is reduced by HMG-CoA reductase (Hmg1 and Hmg2) to mevalonic acid.
The second module involves a series of phosphorylation reactions and begins with the phosphorylation of mevalonate by mevalonate kinase (Erg12) to phosphomevalonate, which is further phosphorylated to mevalonate-5-pyrophosphate by phosphomevalonate kinase (Erg8). Mevalonate pyrophosphate decarboxylase (Mvd1/Erg19) subsequently catalyses the decarboxylation of mevalonate-5-pyrophosphate to isopentenyl pyrophosphate (IPP). The 5-carbon prenyl phosphate, IPP is then isomerised to dimethylallyl pyrophosphate by IPP isomerase (Idi1). The condensation of the isomers IPP and dimethylallyl pyrophosphate result in the formation of geranyl pyrophosphate, which condenses with another molecule of IPP to form FPP. The above two condensation reactions between the pyrophosphates are catalysed by FPP synthase (Erg20).
The ergosterol biosynthesis pathway is energy intensive, with at least 24 molecules of ATP and 16 molecules of NADPH estimated to be required for the de novo synthesis of one molecule of ergosterol (Hu et al., 2017). In addition, oxygen is an essential cofactor for several enzymes, which makes the ergosterol biosynthesis pathway dependent on the availability of environmental oxygen (Galea & Brown, 2009). However, in aerobic conditions, yeast cells can overproduce ergosterol, an excess of which is cytotoxic to yeast cells in the free form (Liu, Xia, Nie, Wang, & Deng, 2019). Therefore, stringent regulation and maintenance of cellular sterol homeostasis are critical.
To attenuate the excessive sterol pool, yeast cells either secrete sterol acetates into the extracellular matrix (Hu et al., 2017) or store steryl esters in lipid particles (Taylor & Parks, 1978). Under anaerobic conditions, sterol uptake becomes essential for growth and is obtained either from esterified sterol reserves that can be readily interconverted to free sterols, or from the environment using an import process mediated by ATP-binding sterol transporters Aus1 and Prd11 that are repressed in the presence of oxygen (reviewed in Hu et al., 2017;Liu et al., 2019). The regulation of ergosterol biosynthesis is tightly controlled at several check points and accomplished by feedback mechanisms at the level of transcription, translation, and posttranslation, including the inhibition of key enzymes by ergosterol pathway intermediates and gene activation by transcription factors Upc2, Ecm22, Rox1, and Mot3 (reviewed in Liu et al., 2019).

| STEROL PROFILES OF VIABLE ERGOSTEROL BIOSYNTHESIS MUTANTS
Most of the studies in this review report the phenotypes of viable ergosterol biosynthesis mutants, which vary significantly in sterol composition. Yeast strains deficient in Erg6, Erg2, Erg3, Erg5, or Erg4 are able to grow under routine laboratory conditions (Palermo, Leak, Tove, & Parks, 1997). The remaining ERG genes are classified as essential, although erg24Δ strains are able to grow under specific conditions, such as in defined synthetic media but not yeast extract peptone-based media under aerobic conditions (Crowley, Smith, Leak, & Parks, 1996;Lorenz & Parks, 1992;Palermo et al., 1997).
At the PM of wild type yeast, ergosterol is the predominant sterol, with minor amounts of zymosterol (Zinser et al., 1993). Yeast strains with deletions of genes in the late steps of ergosterol biosynthesis accumulate sterols that differ from ergosterol in the number and position of double bonds in the B-ring and the side chain of the sterol molecule. Sterol profiling suggests that Erg enzymes catalysing the last five steps of ergosterol biosynthesis have reduced substrate specificity and can accept a broad range of similar sterol structures.
As a result, mutants of late ERG genes accumulate mixtures of sterols including the precursors of the enzymes they lack. The most abundant sterols in these yeast mutants are depicted in Figure 2, and the potential catalytic activities that lead to their synthesis are discussed here.
Small changes in sterol methyl groups and positions of unsaturation in the ring skeleton or aliphatic tail, significantly impact the dynamics of lipid bilayers (Ranadive & Lala, 1987;Shahedi, Orädd, & Lindblom, 2006). Sterol structure impacts van der Waals interactions with other lipid bilayer components and influences the packing effects of sterols when included in model membranes. Notably, small variations in saturation within the stiff sterol ring structure can change the orientation of sterol rings, rendering a sterol flat (for example, ergosterol) or bended (for example, fecosterol and episterol) (Shahedi et al., 2006). These differences significantly alter the biophysical properties of the membrane (Ranadive & Lala, 1987 (Figure 3; Arora, Raghuraman, & Chattopadhyay, 2004;Dufourc, 2008;Hsueh et al., 2007;Hsueh, Gilbert, Trandum, Zuckermann, & Thewalt, 2005;Low, Rodriguez, & Parks, 1985;Shrivastava & Chattopadhyay, 2007;Soubias, Jolibois, Massou, Milon, & Réat, 2005). In particular, sterols are considered to interact strongly with the long and highly saturated acyl chains of sphingolipids (Ahmed, Brown, & London, 1997;Wattenberg & Silbert, 1983). In the "lipid raft" model for PM structure, microdomains rich in sterol and sphingolipid are considered to form detergent-insoluble rafts of liquid ordered phase, surrounded by bulk liquid disordered phase. The reliability of using model membranes and detergent solubility studies to define PM structure has been controversial (reviewed in Hancock, 2006;Lichtenberg, Goñi, & Heerklotz, 2005;Munro, 2003); however in yeast, studies using green fluorescent protein which anchors in the outer leaflet of the PM. In a study by Abe and Hiraki (2009), with the exception of erg5Δ, viable ergΔ strains exhibited accelerated rotation of TMA-DPH in the PM; the rotational diffusion coefficients were significantly higher than wild type in the order erg2Δ > erg6Δ > erg3Δ > erg4Δ. The erg2Δ, erg6Δ, and erg4Δ strains also exhibited a decrease in order compared to wild type, with erg2Δ exhibiting the strongest deviation from the wild type. Overall, this study indicated decreased rigidity of the PM in ergΔ strains, with an increase in the occurrence of voids, particularly for the erg2Δ strain. The erg2Δ and erg6Δ strains accumulate sterols with unsaturation in carbon position 8(9), which differs considerably from the carbon 7(8) unsaturation in ergosterol ( Figure 2). These altered sterols could account for the striking differences in PM dynamics.
A number of studies have indicated that sterols have a role in maintaining the compartmentation of PM proteins. For example, in a genome-wide screen for genes affecting the colocalisation of MCC domain proteins, Grossmann et al. (2008) identified that loss of Erg2, Erg24, or Erg6 strongly disrupted MCC formation. Staining of F I G U R E 3 Membrane phases (a) Simplified schematic of a lipid bilayer in 'liquid disordered' phase. In this highly fluid state, lipids exhibit irregular packing with increased disorder and lateral mobility.
(b) Below melting temperature, lipid bilayers pack tightly in a solid-like phase known as 'gel phase'. This impedes the lateral movement of lipids. (c) When sterols are included in model membranes, bilayers form a 'liquid ordered phase' in which acyl chains are packed more closely than in liquid disordered phase, but retain lateral mobility S. cerevisiae cells with filipin (a fluorescent polyene that interacts with 3 0 -β-sterols) has indicated that MCC domains are rich in ergosterol (Grossmann, Opekarová, Malinsky, Weig-Meckl, & Tanner, 2007), and the filipin staining of MCC domains was not as distinct as wild type in erg2Δ, erg24Δ, or erg6Δ strains, indicating reduced compartmentation of sterols (Grossmann et al., 2008). Notably, Jin, McCaffery, and Grote (2008) suggest that filipin might only bind free ergosterol, as a strain with approximately 50% reduction in sphingolipid and a modest reduction in ergosterol content, exhibits bright uniform staining around the PM. However, it is also possible that sterol distribution is impacted by reduced sphingolipid content.
A number of hypothesised mechanisms for PM compartmentation have been investigated. Spira et al. (2012) found actin depolymerisation to have a minor impact on protein distribution, whereas enzymatic degradation of cell wall induced aggregation of membrane proteins into large patches. Reduced lipid complexity (following modification of either phospholipid, sphingolipid, or sterol content) was concluded to be strongly associated with reduced protein segregation. Membrane potential also appears to influence the lateral distribution of both sterols and proteins. Grossmann et al. (2007) applied a proton gradient uncoupler to yeast strains and found depolarisation of PM to result in dispersion of ergosterol and proteins from MCC patches. Notably, when cells were first treated with the sterol dye filipin and then with the proton gradient uncoupler, ergosterol did not disperse, and Can1 was no longer released from MCCs, suggesting that ergosterol has a structural role in maintaining MCC domains.
Overall, the work of Spira et al. (2012) and others indicate that ergosterol has a role in maintaining lateral compartmentation of the PM. Heterologous PM proteins produced in yeast may not localise to PM microdomains as anticipated in cells with altered sterol content, without further engineering. For example, Grossmann et al. (2008) report reduced localization of Hup1 (a hexose/H + symporter from Chlorella kessleri) at MCC domains of S. cerevisiae erg2Δ, erg24Δ, and erg6Δ strains. Further phenotypes of ergosterol biosynthesis mutants are likely to be explained in part by the altered activities of PM processes, which also have implications for yeast as microbial cell factories as discussed below.  -Peck et al., 2002;Munn et al., 1999).
The TORC1 complex functions as a master regulator of cell growth.
Under conditions favourable for growth, TORC1 has kinase activity and promotes processes required for cell growth and division, such as ribosome biogenesis and translation initiation. Activation of TORC1 has been shown to promote the endocytosis of some transporters, such as found an erg3Δ strain to be more sensitive to treatment with the TORC1 inhibitor rapamycin, as determined by reduced growth in the presence of rapamycin and rapid dephosphorylation of the TORC1 target Sch9 when treated with rapamycin. These results suggest reduced TORC1 activity in erg3Δ under normal growth conditions; however, the results could potentially also be due to increased intracellular accumulation of rapamycin, if efflux of rapamycin is reduced in erg3Δ (see Section 4.4). The TORC1 complex is sited at vacuolar membranes, and vacuoles are highly fragmented in many ergΔ strains (Heese-Peck et al., 2002;Munn et al., 1999). It is therefore plausible that the regulation and activity of TORC1 could be perturbed in ergΔ strains. It is therefore plausible that the regulation and activity of TORC1 could be perturbed in ergΔ strains. Alternatively, TORC1 activity could be reduced in ergΔ strains in response to other physiological conditions that are unsupportive for growth; multiple signals including nutrient availability, salt stress, redox stress, and temperature stress regulate TORC1 activity; however, the mechanisms by which inputs are sensed and integrated remain to be elucidated (Loewith & Hall, 2011).
Yeast endocytosis is commonly investigated by studying uptake In wild type cells, the transmembrane receptor of α-factor, Ste2, undergoes a basal level of endocytosis in the absence of mating pheromone.
Upon binding α-factor however, Ste2 is hyperphosphorylated, ubiquitinated, and the receptor-ligand complex is rapidly internalised (Toshima, Nakanishi, Mizuno, Toshima, & Drubin, 2009). Internalisation of α-factor has been found to be similar to wild type in erg3Δ and erg4Δerg5Δ strains, delayed in an erg6Δ strain, significantly reduced in erg2Δ and erg2Δerg3Δ strains, and strongly reduced in an erg3Δerg6Δ strain (Heese-Peck et al., 2002;Munn et al., 1999). In Notably, ergosterol has been found to interact with the kinase Yck2 (Li, Gianoulis, Yip, Gerstein, & Snyder, 2010), which has a role in Ste2 phosphorylation (Feng & Davis, 2000). It is possible that this kinase does not function efficiently in strains with modified sterol composition.
In strains where sufficient α-factor is internalised, degradation of the pheromone has been detected, indicating successful delivery to the vacuole (Munn et al., 1999). However, vacuolar accumulation of the water-soluble dye LY is significantly reduced in most ergΔ strains (Heese-Peck et al., 2002;Munn et al., 1999). Requirements for the internalisation of LY in fluid phase and the lipophilic dye FM4-64 (chemical structures shown in Figure 4), which binds membranes, may be different, or basal endocytosis levels may be reduced in ergΔ strains owing to altered Ypk1 activity. Ergosterol not only copurifies with the Ypk1 kinase from wild type S. cerevisiae cells but also augments Ypk1 kinase activity in vitro (Li et al., 2010). Additionally, Ypk1 isolated from erg4Δ cells exhibits five-fold less kinase activity than  (Kato & Wickner, 2001). It is notable that in the studies by Munn et al. (1999) and Heese-Peck et al. (2002), erg6Δ was the only ergΔ strain that exhibited normal vacuole morphology and was also the only strain to accumulate LY dye similarly to wild type.
With regards to microbial cell factories, these studies indicate that in strains with altered sterol content, compounds are internalised by endocytosis at different rates. This is of relevance to microbial cell factory cultures in which chemicals are added to induce or suppress specific cellular processes (discussed further in Section 4.4). Meanwhile, the reduced accumulation of Tat2 (a high affinity tryptophan and tyrosine permease) at PM means that ergΔ strains may have higher requirements for tryptophan and tyrosine in growth media (Umebayashi & Nakano, 2003), and indicates that when strains are engineered to produce heterologous proteins, they may not be traf-

| Mating
The yeast mating response (reviewed in Bardwell, 2005)  for Ste20 (Tiedje, Holland, Just, & Höfken, 2007). In the study by Aguilar et al. (2010), schmoo formation but not cell fusion was rescued by exogenously supplying ergosterol. Intriguingly, deletion of ERG5 in erg4Δ cells restored both schmoo formation and membrane fusion between mating pairs. These results suggest that accumulation of the highly unsaturated ergosta-5,7,22,24(28)-tetraenol, which accumulates in the erg4Δ strain, restricts the ability of mating pairs to fuse.
With regards to the development of yeast microbial cell factories, these studies highlight that mating occurs with reduced efficiency in strains accumulating altered sterols. Therefore, mating as an engineering strategy should generally be avoided if sterol content has been altered. Additonally, biosensors that utilise pheromone response path-

| Tolerance to mycotoxic compounds
Examples of compounds that have been applied to S. cerevisiae ergΔ mutants are included in Figure 4, and the tolerances of ergΔ strains to various exogenous compounds are reported in Table 2.
As discussed in Section 4.2, the reduced vacuolar accumulation of LY dye in many ergΔ strains could be due to reduced rates of fluid phase endocytosis; however, in the development of antifungal pharmaceuticals, blocking ergosterol biosynthesis has been found to decrease yeast tolerance to a wide range of drugs. Sykes (1979) (Graham, Scott, & Emr, 1993;Vogel, Lee, Kirsch, Rose, & Sztul, 1993).
It has been hypothesised that suppressing ergosterol biosynthesis results in increased intracellular accumulation of exogenously applied chemicals, which reduces the strain's tolerance to mycotoxic compounds. As discussed below, this has been attributed to hyperpolarisation of the PM, increased membrane permeability, and decreased activity of efflux pumps such as Pdr5 (an ATP-binding cassette transporter, overexpression of which confers resistance to a broad range of compounds; Kolaczkowski et al., 1996).
The increased susceptibility of ergΔ mutants to cationic compounds, such as hygromycin B and trimethylammonium, has been partially attributed to an increase in membrane hyperpolarisation, as indicated by uptake of potentiometric dyes (Barreto et al., 2011;Kodedová & Sychrová, 2015), although the efflux of such dyes could be reduced in ergΔ strains (Kodedová & Sychrová, 2015). Meanwhile, Abe and Hiraki (2009) highlight a correlation between increased cycloheximide susceptibility, and the reduced membrane rigidity of different ergΔ strains, as indicated by anisotropy of the fluorescent dye TMA-DPH. Kaur and Bachhawat (1999) also report increased cycloheximide susceptibility for erg6Δ, erg2Δ, erg3Δ, and erg4Δ strains and investigated the efflux activity of Pdr5, for which cycloheximide is a substrate. Kaur and Bachhawat (1999) found the energy-dependent efflux of the fluorescent dye Rhodamine 6G (another known substrate for Pdr5), to be much slower than wild type in erg4Δ, erg2Δ, and erg6Δ strains. Efflux in the erg3Δ strain was also delayed but to a lesser extent than the other strains. Overexpression of PDR5 increased cycloheximide tolerance in erg4Δ, erg3Δ, and to a lesser extent in erg2Δ. Further evidence for sterol composition impacting the activity of efflux pumps came from a study by Kontoyiannis (2000), in which the fluconazole resistance imparted by an activated allele of transcription factor gene PDR1 (termed PDR1-100), resulting in increased PDR5 transcription, was found to be diminished in the absence of Erg3.
Nystatin is a polyene that has been shown to bind ergosterol in synthetic liposomes and form pores in membranes at high concentrations (Coutinho, Silva, Fedorov, & Prieto, 2004). Amphotericin B (AmB) is another polyene with mycotoxic effect. Recent studies utilizing chemical derivatives of AmB, membrane permeabilisation assays, ergosterol-binding assays, and nuclear magnetic resonance spectroscopy, have indicated that AmB mycotoxicity is relayed primarily through sequestration of ergosterol into cell surface aggregates (Anderson et al., 2014;Gray et al., 2012;Palacios, Dailey, Siebert, Wilcock, & Burke, 2011). Strains deficient in Erg5 (and to a much lesser extent Erg2, Erg4, or Erg6) have increased tolerance to AmB (Gazdag et al., 2014). It is plausible that in the strains with enhanced polyene tolerance, the drug has reduced affinity for the sterols present.
Steroidal glycoalkaloids have also been reported to permeabilise membranes via interaction with sterols (Keukens et al., 1992). In a study by Simons et al. (2006), the steroidal glycoalkaloid α-tomatine was found to induce electrolyte leakage when applied to S. cerevisiae, whereas the aglycone tomatidine did not, and both compounds displayed mycotoxic effect. When the tolerances of erg2Δ, erg3Δ, and erg6Δ strains were investigated, erg6Δ was more tolerant to α-tomatine, whereas erg2Δ and erg3Δ were less tolerant. Conversely, the erg3Δ strain had greatly enhanced tolerance to the aglycone tomatidine. Again, differing affinities of these compounds for different sterols would explain the phenotypes observed.
During the fermentation of yeast cell factories, chemicals of varying biochemical properties may be added to cultures to induce, suppress, or augment particular cellular processes. For example, the auxin-inducible degron system from plants has been used to degrade target proteins in yeast (Nishimura, Fukagawa, Takisawa, Kakimoto, & Kanemaki, 2009); tetracycline and related compounds have been used to repress or activate genes (Bellí, Garí, Piedrafita, Aldea, & Herrero, 1998;Garí, Piedrafita, Aldea, & Herrero, 1997); and the hormone β-estradiol has been used for precise control of gene expression (Louvion, Havaux-Copf, & Picard, 1993;Ottoz, Rudolf, & Stelling, 2014). Chemicals may also be added to inhibit specific metabolic pathways, to increase the intracellular concentration of substrates for fine chemical biosynthesis. For example, lovastatin inhibits HMG-CoA reductase activity (Gardner, Shan, Matsuda, & Hampton, 2001), and terbinafine inhibits squalene epoxidase activity (Ryder, 1992), both of which result in accumulation of substrates for the synthesis of highvalue terpenes. Lower concentrations of these expensive chemicals may be required for yeast strains with altered sterol composition, which could drastically reduce production costs. For example, Cdr1 (the C. albicans homologue of Pdr5) has been found to export β-estradiol (Krishnamurthy, Gupta, Snehlata, & Prasad, 1998), therefore ergΔ strains with reduced Pdr5 activity may accumulate β-estradiol (which currently costs approximately £1,700 per 100 g) at a faster rate than wild type. On the other hand, some feedstocks contain inhibitors to yeast growth, such as furfural and coniferyl alcohol (Deparis, Claes, Foulquié-Moreno, & Thevelein, 2017), which may have an increased impact on strains with altered sterol content.

| Ion homeostasis and osmoregulation
Potassium is necessary for multiple aspects of yeast growth and survival, while sodium accumulation has toxic effects, through potassium Pyranine and pHlourin fluorescence measurements indicate that intracellular pH is lower in strains with reduced ergosterol levels (Calahorra, Lozano, Sánchez, & Peña, 2011;Kodedová & Sychrová, 2015), and ergΔ strains have a reduced ability to control intracellular pH upon treatment with sodium chloride (Kodedová & Sychrová, 2015). The pH of vacuolar compartments in erg2Δ, erg3Δ, erg6Δ, and erg24Δ strains was reported to be higher than wild type by Zhang et al. (2010). In this study, vacuolar fragments isolated from an erg24Δ strain exhibited reduced V-ATPase activity. Fragmented vacuoles and perturbed activities at vacuolar membrane have also been reported in many ergΔ strains (Heese-Peck et al., 2002;Kato & Wickner, 2001).
A number of studies using the fluorescent potentiometric dye 3,3 0 -dipropylthiacarbocyanine iodide have indicated that reduced ergosterol content results in hyperpolarisation of the PM (Calahorra et al., 2011;Kodedová & Sychrová, 2015), although the enhanced accumulation of this dye could also result from reduced rates of efflux through multidrug resistant pumps (Kodedová & Sychrová, 2015). The increased susceptibility of ergΔ mutants to cations has been highlighted as another indicator of membrane hyperpolarisation. Welihinda et al. (1994) report S. cerevisiae erg6Δ to be less tolerant to lithium and sodium cations, associated with three-to four-fold higher rates of influx, while efflux rates remained similar to wild type. Kodedová and Sychrová (2015) have since reported erg2Δ, erg3Δ, erg4Δ, and erg6Δ strains to have reduced tolerance to high levels of lithium ions and sodium chloride, although an erg5Δ strain had similar tolerances to wild type in this study.
Meanwhile, erg2Δ and erg6Δ strains have been found to exhibit reduced growth in low potassium conditions, with erg3Δ, erg4Δ, and erg5Δ strains impacted to a lesser extent (Barreto et al., 2011). Confocal microscopy of Trk1-GFP has indicated reduced PM localisation of this high affinity potassium transporter in erg6Δ strains (Barreto et al., 2011). Therefore, the reduced growth in low potassium reported for certain ergΔ mutants could be due to reduced presence of potassium transporters at the PM.
Some ergΔ strains have also been found to be less tolerant to treatment with high concentrations of glucose or sorbitol, which would induce osmotic pressure (Kodedová & Sychrová, 2015). In addition to potentially being more susceptible to osmotic stress, due to disrupted ion homeostasis, ergΔ strains may exhibit defects in their response to osmotic stress. The kinase Ssk22 is involved in relaying the osmotic stress response in yeast, as part of a MAP kinase cascade that activates Hog1 (Posas et al., 1996), and studies by Li et al. (2010) indicate a role for ergosterol in maintaining Ssk22 protein levels. Conversely, the hyperosmotic stress response in yeast involves rapid repression of ERG2 and ERG11 (mediated by the MAP kinase Hog1), and overexpression of the ergosterol pathway confers increased susceptibility to salt stress (Bhattacharya et al., 2018;Montañés, Pascual-Ahuir, & Proft, 2011). These cellular responses are coherent when considering the wider context of the hyperosmotic stress response, which involves cell cycle arrest (Clotet et al., 2006;Escoté, Zapater, Clotet, & Posas, 2004) and diversion of metabolic resources towards glycerol biosynthesis (reviewed in Hohmann, 2002).
These studies suggest that careful design of growth media is required in fermentations using strains with altered sterol content.
For example, such strains may be less tolerant to high sugar and salt concentrations and may also require higher concentrations of potassium. Alternatively, strains may be further engineered to increase ion homeostasis (Deparis et al., 2017). The latter may be of preference; in first generation bioethanol production (utilising food crop biomass), streams of 35% sugar are used in order to attain ethanol titres of 16-18% (Deparis et al., 2017). In second generation bioethanol production (utilising nonfood crops as feedstock), fermentations are usually limited to 12% sugar; however, high concentrations of sodium salt may be present, due to feedstock pretreatment conditions and/or accumulation of sodium salt in pipelines and fermenters (Deparis et al., 2017). Certain feedstocks also contain metal cations, to which ergΔ strains may be more susceptible. For example, sugarcane bagasse usually contains Mg 2+ , Fe 2+ , Mn 2+ , Cu 2+ , and Zn 2+ (Deparis et al., 2017).

| Ethanol tolerance
Many yeast including S. cerevisiae rapidly convert sugars to ethanol, as a means to outcompete competitors in sugar-rich environments (Dashko, Zhou, Compagno, & Piškur, 2014 Tierney, Block, & Longo, 2005;Vanegas, Contreras, Faller, & Longo, 2012;Vanegas, Faller, & Longo, 2010). Elevated levels of unsaturated fatty acids and ergosterol have been associated with increased ethanol tolerance in a number of studies (reviewed in Henderson & Block, 2014). For example, Aguilera, Peinado, Millán, Ortega, and Mauricio (2006) compared the ethanol tolerance and lipid composition of five S. cerevisiae strains and found increased ergosterol, oleic acid, and palmitoleic acid content to be highly correlated with ethanol tolerance. The hypothesised role of ergosterol in ethanol tolerance is supported by the decreased ethanol tolerance of certain ergΔ strains. In a study by Liu et al. (2017), an erg3Δerg5Δ strain was found to grow at 35% the rate of a wild type strain in the pres- (2017) subsequently compared the thermotolerance of erg2Δ, erg3Δ, erg4Δ, erg5Δ, erg3Δerg4Δ, erg3Δerg5Δ, and erg4Δerg5Δ strains and found them to exhibit a higher growth rate than wild type at 39.5 C.
Of the individual mutants, erg3Δ and erg5Δ had the highest thermotolerance, and these mutations had an additive effect, with the erg3Δerg5Δ strain showing 2.24-fold increase in growth rate relative to wild type at 39.5 C .
As explored by Abdel-Banat et al. (2010), there are many potential cost savings associated with increasing the thermotolerance of microbial cell factories. In the production of ethanol from starch for example, starch is liquefied at 90 C with a thermostable α-amylase, cooled for either separate or simultaneous saccharification and fermentation (32-35 C), and then heated again for ethanol extraction.
During the fermentation phase, cooling is also required, as fermentations generate heat (Deparis et al., 2017 To test the translatability of their findings to other fungal species, Liu et al. (2017) removed the ortholog of S. cerevisiae Erg5 from the filamentous fungus Penicillium oxalicum, and found the mutant strain to exhibit significantly improved growth at 37 C relative to wild type.
This is of industrial relevance as P. oxalicum grows optimally at 30 C but secretes cellulases that function optimally at 50 C. Hence, modification of sterol content could be a promising approach to develop a range of fungal strains for high-temperature fermentation.

| CONCLUSIONS
The viable mutants of the ergosterol biosynthesis pathway exhibit a range of phenotypes, including altered localisation of specific PM proteins, fragmentation of vacuoles, reduced completion of mating, altered susceptibility to chemical treatment, PM hyperpolarisation, increased tolerance to high temperatures, and reduced tolerance to other stresses such as high ethanol, salt and solute concentrations.
Notably there are differences in the severity of phenotype between different ergΔ mutants, reflecting the impact of different sterol compositions. The removal of methyl groups from the carbon 4 and carbon 14 of lanosterol would appear to be critical for sterol function, as strains deficient in Erg11, Erg25, Erg26, Erg27 or Erg28 are non-viable, although interestingly the erg24Δ strain can grow under specific conditions (Crowley et al., 1996).
Of the viable ergΔ mutants, strains lacking Erg2 often have the strongest phenotype. Time-resolved fluorescence anisotropy measurements indicate that there is greatly elevated freedom of rotational movement within the PM of this strain, and increased disorder ). The erg2Δ strain accumulates sterols with unsaturation in carbon position 8(9), which differs considerably from the carbon 7 (8) unsaturation in ergosterol (Figure 2). These altered sterols could account for the striking differences in PM dynamics. Meanwhile, the lack of a carbon 24 methyl group in erg6Δ strains, is likely to alter interactions of the sterol side chain with fatty acid tails within the hydrophobic core of lipid bilayers.
In some instances, the accumulation of a specific sterol structure is hypothesised to impact a specific process. For example, the accumulation of ergosta-5,7,22,24(28)-tetraenol in erg4Δ strains is postulated to impede membrane fusion between mating cells (Aguilar et al., 2010). As discussed in section 4.3, the aliphatic tail of this sterol is particularly rigid due to the presence of two double-bonds.
Many of the phenotypes observed are due to altered processes at cell membranes. For example, reduced activity of efflux pumps and reduced V-ATPase activity. Multiple studies indicate that signalling processes are disrupted or respond differently in strains with altered sterol content. For example, the reduced phosphorylation of Ste2 in the presence of mating pheromone (Heese-Peck et al., 2002), and the trafficking of Tat2 to vacuoles as opposed to the PM (Estrada et al., 2015). Notably, in the analysis of metabolites that co-purify with kinases, ergosterol was found to co-purify with 15 different kinases (Li et al., 2010), including Ypk1 which regulates rates of endocytosis (deHart et al., 2002), Yck2 which is involved in Ste3 phosphorylation (Feng & Davis, 2000), and Ssk2 which is involved in responding to osmotic stress (Posas et al., 1996). Furthermore, ergosterol was found to augment Ypk1 kinase activity in vitro (Li et al., 2010), suggesting that yeast sterols could have a physiological role in facilitating the activity of some kinases.
The implications of ergΔ phenotypes in the development of yeast as microbial cell factories are many. In some cases, removing specific ergosterol biosynthesis genes may be beneficial. For example, when increased susceptibility to an expensive drug is desirable, or if it is optimal for fermentations to take place at higher temperatures. It may also be beneficial to suppress ergosterol biosynthesis in order to maximize flux towards an alternate pathway. Notably, suppression of specific enzymes in ergosterol biosynthesis has been found to increase expression of most genes in the ergosterol pathway (Caspeta et al., 2014), potentially increasing the availability of precursors of interest.
However, there are important considerations to factor into the design of growth media when sterol profiles are altered; namely, the reduced tolerances of strains with altered sterol content to osmotic stress, high ethanol content, and low concentrations of specific amino acids and ions.
In conclusion, there are many potential benefits to modifying sterol composition in yeast cell factories, however the phenotypes of strains with altered sterol content are wide-ranging and should be considered when engineering strain and culture conditions in order to maximize productivity.