Growth Potential of Selected Yeast Strains Cultivated on Xylose-Based Media Mimicking Lignocellulosic Wastewater Streams: High Production of Microbial Lipids by Rhodosporidium toruloides

: The potential of Rhodosporidium toruloides , Candida oleophila , Metschnikowia pulcherima, and Cryptococcus curvatus species to produce single-cell-oil (SCO) and other valuable metabolites on low-cost media, based on commercial-type xylose, was investigated. Rhodosporidium strains were further evaluated in shake-ﬂasks using different lignosulphonate (LS) concentrations, in media mimicking waste streams derived from the paper and pulp industry. Increasing the LS concentration up to 40 g/L resulted in enhanced dry cell weight (DCW) while SCO production increased up to ~5.0 g/L when R. toruloides NRRL Y-27012 and DSM 4444 were employed. The intra-cellular polysaccharide production ranged from 0.9 to 2.3 g/L in all fermentations. Subsequent fed-batch bioreactor experiments with R. toruloides NRRL Y-27012 using 20 g/L of LS and xylose, led to SCO production of 17.0 g/L with maximum lipids in DCW (Y L/X ) = 57.0% w / w . The fatty acid (FA) proﬁle in cellular lipids showed that oleic (50.3–63.4% w / w ) and palmitic acid (23.9–31.0%) were the major FAs. Only SCO from batch trials of R. toruloides strains contained α -linolenic acid. Media that was supplemented with various LS concentrations enhanced the unsaturation proﬁle of SCO from R. toruloides NRRL Y-27012. SCO from R. toruloides strains could replace plant-based commodity oils in oleochemical-operations and/or it could be micro-and nano-encapsulated into novel food-based formulas offering healthier food-products.


Introduction
Renewable lignocellulosic feedstock is very challenging from the perspective of a low carbon economy as it is abundant (≈200 billion t. annually), inexpensive, and rich in fermentable sugars [1]. It mainly consists of lignin (15-20%), cellulose (40-50%), and hemicellulose (25-30%) [2]. Hemicellulose in hardwoods i.e., Eycalyptus globulus accounts for around 26% of the dry mass and it contains C6 and C5 sugars with xylose being the predominant one [3]. Commercial-type xylose, with a purity of around 95%, is produced after acid-assisted treatment of various lignocellulosic-based resources including corn cobs and sugarcane bagasse amongst others, while the rich in xylose stream is sequentially condensed and crystallized. This commercial-type xylose is of low-cost while fermentation media containing an initial xylose concentration of 25-110 g/L could simulate the xylose content in spent sulphite liquor (SSL), or other lignocellulosic-derived residues such as the kraft bleach plant effluent (KBPE), the xylose mother liquor, etc. [4,5]. production of specialty products for chemical, cosmetic, and pharmaceutical applications, i.e., value-added fatty acid esters [7]. Last but not least, SCO-derived biolubricants have also been proposed as potent substitutes of their petrochemical-based counterparts [19].
In the present study, several yeast strains belonging to the species Rhodosporidium toruloides, Candida oleophila, Metschnikowia pulcherima, and Cryptococcus curvatus were evaluated for their ability to grow and to produce SCO in non-purified xylose (the so-called commercial-type xylose), deriving from lignocellulosic biomass. The most promising among the previously screened yeasts including two R. toruloides strains were additionally tested under batch shake-flask fermentations in order to further enhance the production of SCO on xylose-enriched fermentation media. The inhibitory effect of applying different LS concentrations on SCO production was also investigated. Eventually the best performing Rhodosporidium toruloides strain (namely the strain NRRL Y-27012) was tested in scaled-up trials under fed-batch mode using a bench top bioreactor. The fatty acid profile of SCO was considered under all the performed fermentation conditions.

Raw Materials and Microorganisms
The yeast strains Rhodosporidium toruloides DSM 4444, R. toruloides NRRL Y-27012, Candida oleophila ATCC-20177, Metschnikowia pulcherima FMCC Y-2, Cryptococcus curvatus NRRL Y-1511, and Cryptococcus curvatus ATCC 20509 were used to produce SCO. The strain with the code FMCC Y is an indigenous yeast, isolated from foodstuffs; the strains with the code NRRL Y were provided by the NRRL culture collection (Peoria, IL, USA); the strains with the code ATCC were purchased from the American Type Culture Collection (Manassas, VA, USA); while the strain with the code DSM was provided by the German Collection of Microorganisms and Cell Cultures (Leibniz, Germany) respectively. All the strains were maintained at T = 4 • C on glass slopes containing 10 g/L glucose, 10 g/L yeast extract, 10 g/L peptone, and 20 g/L agar. A liquid pre-culture with the same composition was prepared as inoculum and it was incubated at T = 28 • C in an orbital shaker (Zhicheng ZHWY 211C; Shanghai, China), at an agitation rate of 180 rpm for 24 h. LS (sodium-derived) were provided by the company LignoTech Iberica (Torrelavega, Spain).

Fermentations for SCO Production
Initial batch fermentations were carried out to screen the aforementioned microbial strains for biomass proliferation and lipid accumulation. Commercial-type xylose (purity of ca. 95%, w/w) was used as the sole carbon source at an initial total concentration of (S 0 ) ≈ 50 g/L. The experiments were conducted in 250 mL Erlenmeyer flasks with a working volume of 50 mL, implicating full aerobic conditions (viz. dissolved oxygen tension-DOT ≥ 20% v/v) throughout fermentations [20]. DOT in the flask cultures was performed by off-line measurement with a selective electrode (HI 9146, Hanna Instruments, Woonsocket, RI, USA) as previously indicated in Filippousi et al. [21]. In all the experiments, for all the strains and irrespective of the culture conditions, trials were performed under full aerobic conditions, with dissolved oxygen tension (DOT) being always ≥20% v/v. A mineral solution containing (in g/L) KH 2 PO 4 , 7.0; Na 2 HPO 4 , 2.5; MgSO 4 ·7H 2 O, 1.5; FeCl 3 , 0.09; ZnSO 4 ·7H 2 O, 0.02; MnSO 4 ·H 2 O, 0.06; CaCl 2 ·2H 2 O, 0.15; and organic nitrogen sources, including 1.0 g/L yeast extract and 2.0 g/L peptone, were added in all fermentations [22]. After sterilization at T = 121 • C for 20 min, the flasks were inoculated with 2% (v/v) yeast pre-culture and incubated at T = 28 • C in an orbital shaker (Zhicheng ZHWY 211C, Shanghai, China) at 180 rpm. The pH was manually adjusted to 6.0 during fermentation using sterile 5 M NaOH. Sequential shake-flask experiments (employing the same conditions of the screening stage) were implemented with the most promising strains of the previously screened microorganisms (R. toruloides strains). Fermentation efficiency was investigated when various LS concentrations (10-40 g/L) were supplemented into the fermentation media.
Fed-batch fermentation was carried out in a 3 L bioreactor (INFORS HT LABFORS), with a working volume of 2 L and incubation temperature of T = 28 • C. A cascade agitation (250-350 rpm) was employed in order to maintain the DOT at 20% of saturation with an aeration rate of 1.5 vvm. The broth media contained ≈50 g/L xylose and 20 g/L LS as well the mineral solution and nitrogen that were reported for the screening stage. The pH was automatically regulated at 6.0 using sterile 5 M NaOH. Feeding supply (350 g/L of commercial xylose solution) was conducted in pulses and it was initiated when the xylose concentration reached around 10 g/L. Inoculation was performed with 10% (v/v) of a 24 h yeast pre-culture. The samples were taken periodically throughout fermentation for sugars, free amino nitrogen (FAN), dry cell weight (DCW), and microbial lipid determination.

Analytical Methods
For the quantification of the yeast total biomass (X, g/L), DCW determination was performed. The whole content of the 250 mL flasks (viz. 50 mL) or sample from the bioreactor (≈20 mL) was collected and cells were harvested by centrifugation (10,000× g, T = 4 • C, 15 min; centrifugation performed in a Hettich Universal 320R-Germany centrifuge), yeast biomass was repeatedly washed with deionized water and re-centrifuged, dried at T = 85 • C for 24 h, and subsequently cooled down in a desiccator. After mechanical disruption of the dried biomass, Folch solution (chloroform:methanol at 2:1, v/v) was added for total lipid extraction and quantification [23]. Specifically, yeast dried biomass (up to 300 mg) was put in a McCartney vial and was covered with a chloroform/methanol (C/M) 2:1 (v/v) blend (up to 25 mL). The whole unit was closed with an aluminum screw cap and was left for at least 6 days in the darkness. Occasionally, the content of the vial was gently mixed with the aid of a glass stick. Then, cell debris was removed through filtration (Whatman ® filter n • 3), (Whatman 1003-185, Whatman Article No. 28413990, US reference) (Merck) the solvent mixture was collected in a pre-weighted evaporator flask and was completely evaporated in a rotary evaporator (R-144, Büchi Labortechnik, Flawil, Switzerland) via vacuum evaporation, and lipids were determined gravimetrically in pre-weighed round bottom flasks [24]. Lipid production (L, g/L) was expressed as g of total lipid that was produced per liter of fermentation broth while intracellular lipid content (Y L/X ) was expressed as g lipid that was produced per 100 g of DCW (%, w/w). The total intra-cellular polysaccharides (IPs, expressed as g per liter of medium and as % of DCW − Y IPs/X ) were measured based on Argyropoulos et al. [25] with slight modifications. Briefly, 50 mg of total biomass was acidified with 10 mL 2 M HCl and subsequently hydrolyzed (T = 100 • C for 30 min) and neutralized to a pH value of 7.0, using 10 mL 2 M NaOH. The solutions were filtered through Whatman filter paper (Whatman ® filter n • 3) and, finally, the content of the sugars that were released (reducing sugars) was determined using the 3,5-dinitrosalicylic acid (DNS) method [26]. The same protocol was applied to determine the xylose concentration.
The fatty acid composition was determined via transformation into fatty acid methyl esters (FAMEs) using sequential treatment with sodium methoxide followed by methanol with HCl as a catalyst. The determination of FAMEs was carried out by gas chromatography (GC) Fisons 8060 equipped with a chrompack column (60 m × 0.32 mm, Chrompack) (CP-Wax 52 CB GC column-Agilent) and a flame ionization detector (FID) using helium as the carrier gas (2 mL/min). The oven program was initiated at 50 • C, heated to 200 • C with a ratio of 25 • C/min (1 min), then increased with ratio of 3 • C/min up to 240 • C, and finally increased to 250 • C with a ratio of 25 • C/min and maintained for 3 min. The detector temperature was set at 250 • C. FAMEs were identified by reference to a standard (Supelco ® 37 Component FAME Mix, 10 mg/mL in CH 2 Cl 2 , 47885-U, Merck, Rahway, NJ, USA) [7]. Analyses were performed in duplicates and the presented data correspond to average values.

Screening of Microorganisms for SCO Production
At an initial stage, all the yeast strains were grown on commercial-type xylose at an initial xylose concentration (S 0 ) adjusted to ca. 50 g/L. SCO production was triggered by the exhaustion of nitrogen in the growth medium, allowing the conversion of xylose to storage lipids [12,27]. The fermentation efficiencies are presented in Table 1. Metschnikowia pulcherrima showed the lowest DCW production (=8.1 g/L), followed by Rhodosporidium strains (11.4-11.6 g/L), Candida oleophila (13.1 g/L), and Cryptococcus strains (16.2-17.1 g/L). The intracellular content of the lipids varied among all the tested microorganisms. The maximum SCO accumulation was achieved with R. toruloides NRRL Y-27012 (Y L/X = 21.9% w/w) and R. toruloides DSM 4444 (Y L/X = 31.6% w/w). SCO production was a quite restricted process in the case of Candida oleophila and M. pulcherima with Y L/X < 10% w/w, while Y L/X was equal to 21.1% w/w when Cryptococcus strains were employed. On the other hand, it appeared that for the strains Candida oleophila and M. pulcherima, cellular metabolism was mainly shifted towards the synthesis of IPs (quantities of polysaccharides per DCW; Y IPs/X in all cases ≥35% w/w). Non-negligible quantities of polysaccharides were also produced by the strain Cryptococcus curvatus NRRL Y-1511 (Y IPs/X = 28% w/w), while the strains that presented quantities of lipid in DCW >20% w/w (the threshold characterizing the microorganism as "oleaginous"; see Papanikolaou and Aggelis [12]) were the two "red" strains (R. toruloides) and the two Cryptococcus curvatus strains, which was in accordance with previous investigations of all these four strains cultured on media enabling the de novo lipid production process [4,5,[28][29][30][31]. Indeed, higher SCO quantities (up to 13.5 g/L) have been reported for R. toruloides NRRL Y-27012 combined with substantial Y L/X values (40-54% w/w) on fermentation substrate containing crude glycerol, obtained as a byproduct stream from the biodiesel industry [28,31]. This is an indication that the de novo lipid synthesis is highly dependent on the carbon sources that are employed. Glycerol is an effective and very competitive carbon source for SCO production compared to xylose [4]. Xylose is typically metabolized through the pentose-phosphate pathway that presents lower efficiency in terms of biomass and metabolite formation, compared to the EMP glycolysis that is used for the catabolism of glycerol or C6 sugars [12,27,32]. In the case of R. toruloides DSM 4444, several studies have demonstrated its enhanced potential to produce SCO in various fermentation substrates. Diamantopoulou et al. [5] reported that R. toruloides DSM 4444 cultivated on high glycerol concentrations showed high lipid accumulating capacities with a maximum SCO concentration of 12.5 g/L and lipid content of Y L/X = 43.0-46.0% w/w. When crude glycerol was replaced with xylose, SCO production decreased. The same yeast strain resulted in remarkable SCO production under batch or fed batch experiments (DCW = 61.2 g/L, Y L/X = 61.8% w/w) when it was cultivated in flour-rich waste streams that were rich in glucose [33]. Studies dealing with Cryptococcus curvatus NRRL Y-1511 have demonstrated a satisfying performance in lactosebased media (SCO = 4.3 g/L, Y L/X = 30%) [29] and xylose-based shake-flask fermentations (SCO = 5.1 g/L, Y L/X = 38%) [34]. Table 1. Quantitative data of yeast strains that were cultivated on commercial-type xylose (S) under nitrogen-limited conditions in shake-flask cultures.

Strain Time (h) S cons (g/L) X (g/L) L (g/L) IPs (g/L)
Cryptococcus The initial xylose concentration (S 0 ) was adjusted to ca. 50 g/L, while the initial concentration of nitrogenous compounds, yeast extract, and peptone was 1.0 and 2.0 g/L, respectively. There are three different points in the fermentations that are presented: (a) when the maximum quantity of DCW (X, g/L) was observed; (b) when the maximum quantity of total lipid (L, g/L) was observed; and (c) when the maximum quantity of endopolysaccharides (IPs, g/L) was achieved. The fermentation time (h), X, L, IPs, and xylose consumed (S cons , g/L) are also depicted for all the above-mentioned fermentation points. Culture conditions: growth on 250 mL conical flasks filled with the 1/5 of their volume agitated at 180 rpm, initial pH of 6.0, pH ranging between 5.1 and 6.1, incubation temperature T = 28 • C. Each experimental point is the mean value of two determinations.

Effect of LS Supplementation on Fermentation Efficiency
In the next step, the effect of various LS concentrations on microbial growth and lipid production was assessed in shake-flask fermentations using the yeast strains that showed good capacity to accumulate intracellular lipids (based on the aforementioned results) and more specifically R. toruloides strains that presented the highest lipid in DCW quantities during growth on xylose-based media. The rationale of this approach was based on the fact that LS-type compounds (together with other xenobiotic substances) are found in various types of lignocellulose-type wastewaters (i.e., SSF, KBPE, xylose mother liquor), while previous investigations have shown that, in several cases, it is possible to modify the cellular metabolism and (in some cases to enhance) the intra-cellular lipid content in R. toruloides strains, when additions of natural substances (i.e., NaCl, hydroglycerolic extract of onion peels, sodium lignosulphonates, etc.) are performed [28,35]. The obtained results related to the addition of LS-type compounds upon the cellular metabolism of R. toruloides are shown in Table 2, while, as mentioned, the bioprocess that is implemented in this stage simulates to a great extent the composition of a variety of xylose that is abundant lignocellulosic-type wastewaters, i.e., SSL. Interestingly, increasing the LS concentration up to 40 g/L, resulted in enhanced biomass formation, in both cases (14.6-14.7 g/L) with Y X/S ranging between 0.30 and 0.36 g/g ( Table 2). SCO production increased up to 5.0 g/L (Y L/X = 37% w/w) and 4.8 g/L (Y L/X = 38% w/w) when R. toruloides Y 27012 and R. toruloides DSM 4444 were, respectively, cultivated on commercial-type xylose that was supplemented with 20 g/L LS. The incorporation of 40 g/L LS into the fermentation medium in both Rhodosporidium cultures drastically shifted the intra-cellular carbon flow towards the bioformation of xylose into lipid-free biomass while simultaneously decreasing the intra-cellular lipid accumulation process (Table 2). More specifically, a sharp decrease (compared to data that were obtained from the experiments with 20 g/L LS) in lipids production, equal to 46% and 50% was, respectively, observed for R. toruloides Y 27012 and R. toruloides DSM 4444 while the lipid-free biomass (viz. total DCW value-total lipids value) was 10.3 g/L and 11.3 g/L, respectively. Based on the consumed xylose calculations (%, w/w of xylose consumed), it is evident that both yeasts were able to catabolize xylose at very satisfying levels (viz. 81-86% w/w of xylose consumed in all cases, despite the presence of LS in the medium). Finally, the addition of LS-type compounds into the medium seemed to be negatively correlated with the presence of IPs (in terms of polysaccharides per DCW) in all of the performed experiments (Table 2). Table 2. Quantitative data of R. toruloides DSM 4444 and NRRL Y-27012 cultivated on commercial-type xylose (S), during batch experiments using various concentrations of lignosulphonates (0-40 g/L).

LS (g/L)
Time (h) X (g/L) L (g/L) IPs (g/L) Y L/X (g/g) Y X/S (g/g) Xylose Consumed (%, w/w) There are three different points in the fermentations are presented: (a) when the maximum quantity of DCW (X, g/L) was observed; (b) when the maximum quantity of total lipid (L, g/L) was observed; and (c) when the maximum quantity of endopolysaccharides (IPs, g/L) was achieved. Each experimental point is the mean value of two determinations. The culture conditions were as in Table 1.
An interesting observation is the fact that in some cases the highest DCW and SCO production did not occur at the same fermentation time (Table 2). It was also observed that the maximum quantities of lipid that were obtained (in g/L) did not coincide with the maximum accumulation of lipids (Y L/X , in % w/w) for both R. toruloides strains (see Figures 1 and 2). Obviously, the carbon flow during fermentation after nitrogen depletion (at t ≥ 48 h, data not shown) resulted in enhanced SCO production but eventually, lipids underwent a partial degradation process, although xylose amounts were non-negligible in the medium (see Figures 1 and 2). At the latter growth stages, the somehow low xylose concentration could not fulfill the metabolic requirements of the strains leading to further consumption of the intra-cellular lipids, favoring the lipid-free material formation (for review see Papanikolaou and Aggelis [12,27]). The degradation of previously produced cellular lipids has been already reported for many species of oleaginous microorganisms including but not limited to Yarrowia lipolytica [5,36], Cryptococcus curvatus [20], Mortierella isabellina [37], and Trichoderma viride [38]. On the other hand, interplay between the biosynthesis of IPs and lipids (viz. initial production of polysaccharides and subsequent decrease in the value of polysaccharides per DCW accompanied by an increase in the value of lipids per DCW, in accordance the current study; see: Figures 1b and 2b) has already been reported by strains belonging to Cryptococcus curvatus, R. toruloides, and M. isabellina [4,5,20,24,39].
The valorization of waste streams that contain xenobiotic compounds can substantially alter the profile of the final biometabolites that are produced by various types of yeasts compared to the control trials [40][41][42]. From all the obtained results in the current investigation, the strains' resistance towards high LS amounts was demonstrated while maintaining their ability to produce considerable amounts of DCW and SCO. In some cases, and up to a certain added LS quantity, lipid accumulation was stimulated compared to the control experiment (viz. no LS added). This is in accordance with previous studies that report these species' suitability to produce biomass, when cultivated in similar fermentation configurations [20,43]. There is limited information that can correlate the yeasts tolerance to phenolic compounds (contained in LS) with metabolomics and proteomics patterns [35]. It is assumed that phenolic compounds that can be found in several types of industrial wastewaters (i.e., aldehydes, phenolic derivatives, and other compounds that are found in various residues such as SSL, olive-mill wastewaters, etc.) are responsible for the low viability of oleaginous yeasts, due to the weak bioconversion capacity of the latter to the former, thus this fact has not been thoroughly investigated or verified [44]. In a recent study, the metabolism of Trichosporon cutaneum, R. toruloides, Rhodotorula glutinis, and Y. lipolytica was investigated when cultivated in wheat straw hydrolysate. Only T. cutaneum, showed satisfactory cell growth and lipid production in the presence of residual phenolic aldehydes [44]. Several other non-conventional yeast strains, i.e., Y. lipolytica [45], Candida cylindracea [40], and C. tropicalis [43] have shown remarkable resistance to olive mill wastewater (OMW)-based fermentation media containing phenolic compounds for the production of value-added metabolites. Additionally, zygomycetes, i.e., Mortierella sp., Cunninghamella echinulata, Mucor sp., Thamnidium elegans, and Zygorhynchus moelleri have demonstrated enhanced capacities to remove phenols from OMW streams and produce significant amounts of SCO (Y L/X = 60% w/w). Despite the fact that zygomycetes strains do not possess ligninolytic enzymes, i.e., laccases and other oxidases, phenolic compound removal could be attributed to phenol absorption rather than oxidative processes [46]. The valorization of waste streams that contain xenobiotic compounds can substantially alter the profile of the final biometabolites that are produced by various types of yeasts compared to the control trials [40][41][42]. From all the obtained results in the current investigation, the strains' resistance towards high LS amounts was demonstrated while  The valorization of waste streams that contain xenobiotic compounds can substantially alter the profile of the final biometabolites that are produced by various types of yeasts compared to the control trials [40][41][42]. From all the obtained results in the current investigation, the strains' resistance towards high LS amounts was demonstrated while

Fed-Batch Fermentation
Taking into consideration the previously reported results, R. toruloides NRRL Y-27012 that presented noticeable growth and production of lipids during growth on xylose-based nitrogen-limited cultures that were supplemented with LS compounds (DCW = 13.6 g/L and L = 5.0 g/L; see Table 2) was further evaluated, and fermentations were scaled up in fed-batch bioreactor trials (active volume = 2.0 L). As performed previously, commercialtype xylose (S 0 ≈ 50 g/L) and nitrogenous compounds (1.0 g/L yeast extract and 2.0 g/L peptone) combined with 20 g/L LS were supplemented into the fermentation medium. The kinetics of DCW, lipids, and IPs production as well as xylose consumption and conversion yield of microbial lipids that were produced per unit of xylose consumed are presented in Figure 3a-c. The yeast stain seemed to perform very well even under this relatively high LS concentration. As it can be observed in Figure 3a, SCO gradually increased up to the later stages of fermentation. DCW increased with a slow rate up to 89 h of fermentation while the production rate was much higher thereafter, until the later stages of the experiment. The xylose consumption rate was significant irrespective of the presence of assimilable nitrogen (up to t ≈ 40 h; kinetics of FAN not presented) into the medium, or thereafter, and specifically when the pulse of concentrated xylose into the medium occurred ( Figure 3b). Furthermore, as in the previous experiments that were performed in shake-flask experiments, polysaccharides in DCW values (Y IPs/X , % w/w) were relatively elevated at the early growth steps (i.e., Y IPs/X > 20% w/w) clearly decreasing as the fermentation proceeded, while exactly the opposite trend was observed for the values of lipids that were produced per DCW (Y L/X , % w/w), indicating the previously mentioned interplay between the biosynthesis and accumulation/degradation of storage polysaccharides and lipids (in accordance with the results that were reported for other R. toruloides strains [5,24,28]).
Compared to the respective trial that was carried out in shake-flask mode (see Table 2, Figure 2), SCO was maximized at 17.0 g/L combined with maximum DCW = 29.7 g/L and intra-cellular lipid content of ≈57.0% w/w in the performed fed-batch experiment after 312 h (see Table 3). It is well documented that fed-batch cultures can lead to higher cell densities than traditional batch operations and simultaneously alleviate the inhibitory effect that is caused by high nutrient concentrations, regulating the flow rate of the feeding medium. Fed-batch strategies can avoid the phenomenon of catabolite repression by keeping the carbon concentration at relatively low levels [47]. Table 3. Literature-cited publications focusing on R. toruloides strains that were cultivated on xyloserich media, synthetic or derived from renewable lignocellulosic resources.

Strain
Substrate Mode X (g/L) SCO (g/L) Y L/X (% w/w) Reference    The representation of SCO that was produced f (consumed xylose) and the subsequent linear regression for the whole set of data (see Figure 3c) represents the global conversion yield of SCO that is produced per unit of xylose consumed (Y L/S ). In the present culture conditions, this received the quite impressive value of 0.19 g/g (Figure 3c), a value that is amongst the most promising ones for wild-type oleaginous microbial strains cultivated on sugar-based nitrogen-limited media [12,14,27,39]. The global conversion yield increased by 59.1% compared to shake-flask fermentation that was performed with 20 g/L LS with the same strain. The R value of the regression was quite satisfactory for the linear regression (R = 0.98).
In any case, the production of DCW and microbial oil for R. toruloides NRRL Y-27012 cultivated on commercial-type xylose that was supplemented with high LS concentrations (20 g/L) is very promising and amongst the highest ones that are reported in literaturecited publications that deal with similar fermentation media and culture configurations. Table 3 summarizes publications focusing on R. toruloides strains that were cultivated on xylose-rich media, synthetic or derived from renewable lignocellulosic resources. The strain NCYC 1576 was cultivated on the hemicellulose liquid fraction from hot water extraction of Betula pendula. Even after biodetoxification of this stream using Bacillus sp., followed by the removal of other lignocellulose-derived inhibitors, i.e., furfural and phenols, the SCO production was rather low (1.8-2.8 g/L) with DCW production ranging between 7.1 and 7.5 g/L [16]. In another study, the adapted strain CCT 0783 resulted in a satisfying SCO concentration (3.7-4.5 g/L) with an intracellular content of Y L/X = 34.6-52.8% w/w when it was grown on undetoxified sugarcane bagasse hemicellulosic hydrolysate as a low-cost substrate [49]. Acid-pretreated switch grass hydrolysates were reported as very promising feedstock for SCO production using the strain NRRL Y-1091. A two stage batch strategy that was optimized in terms of N and C loading, led to substantially high lipid concentrations (26.2 g/L) and DCW formation (42.6 g/L) [50]. In another study, the same strain was cultivated in detoxified wheat straw hemicellulosic hydrolysates. After the removal of 77.2% of inhibitors with activated charcoal and two-fold concentration, the highest concentration of SCO was 1.0 g/L (Y L/X = 9.4%) [15]. Finally, better fermentation efficiency (DCW = 9.9 g/L, SCO = 2.4 g/L, Y L/X = 24.6% w/w) has been reported for R. toruloides ATCC 10788 when detoxified wheat straw hydrolysates were employed [51] ( Table 3).

Total Fatty Acid Profile of Cellular Lipids
The fatty acid (FA) profile of SCO that was derived from R. toruloides yeast strains that were cultivated on commercial-type xylose containing on various concentrations of LS under batch and fed-batch fermentations is presented in Figure 4. Data on FA methyl esters of SCO that was produced during different fermentation times (from the lag to the stationary phase), showed slight variations with respect to oleic acid (C18:1), palmitic acid (C16:0), stearic acid (C18:0), linoleic acid (C18:2), and alpha-linolenic acid (C18:3). The predominant fatty acid in all the fermentations that were performed under batch mode was C18:1 (50.3-58.2%), followed by C16:0 (24.9-31.0%) while the other FAs were found in percentages < 10%. In the case of fed-batch trials, C18:1 reached up to 63.4% with C16:0 varying between 23.9 and 28.9%. C18:2 and C18:0 were detected in amounts lower that 11.0%. This is a strong indication that the FAs profile in cellular lipids is a primarily strain-specific process. An average FA composition of lipids that were obtained from R. toruloides strains was 44.5-66.6% of C18:1, 18.0-35.3% of C16:0, 0.4-11.2% of C16:1, 4.5-16.3% of C18:0, 4.4-17.7% of C18:2, and 0.2-3.8% of C18:3 [35]. Added to strain-specific fatty acid synthesis as the major determinant, the FA compositions of microbial oil can show extensive or low variations that can be attributed to the employed microbial strain, type of fermentation media (i.e., sugars or hydrophilic substrates), the microbial growth stage, and the conditions of the micro-and/or macro-environment [5,29,31,34,52,53].   Overall, in this study, novel wild-type and non-conventional yeast strains were screened under nitrogen-limited conditions on xylose-based media, demonstrating inter- It is interesting to indicate that the FA C18:3, which belongs to the ω-3 polyunsaturated FA (PUFAs) family, was detected in all the batch trials with R. toruloides NRRL Y-27012 even in relatively low quantities (up to 3.0%) (Figure 4b, 4c). Very long chain and rarely found in the plants or fish PUFAs (e.g., EPA, DHA, etc.) were not at all detected, since these compounds are the principal storage lipophilic compounds in oleaginous fungi and algae [54,55]. These rarely found FAs can be produced in significant quantities inside the yeast cells only after genetic modifications have been performed [54,55]. On the other hand, some strains of R. toruloides, R. diobovatum, and R. fluviale have been reported for their ability to produce low amounts of C18:3 and long chain fatty acids (C20-C24), the total content of which is <6% [35]. R. kratochvilovae HIMPA1 produced almost 30% of C20-C27 FAs, which is rather unconventional with respect to other Rhodosporidium species [56]. In contrast, in the fed-batch fermentation, the FA C18:3 was not detected at all, probably due to the different regime of aeration and agitation compared to the shake-flask experiments. This phenomenon has been previously reported for Cunninghamella bainieri, indicating lower amounts of gamma-linolenic acid when different aeration and agitation regimes were applied in the bioreactor trials related to the respective flask experiments [57].
Considering the various LS concentrations that were applied, it was observed that increasing LS amounts led to SCO with a slightly higher percentage of C16:0 in the case of R. toruloides DSM 4444 (Figure 4a). In the case of R. toruloides NRRL Y-27012, the supplementation of media with various LS concentrations seemed to enhance the unsaturation profile of the produced SCO (64.9-68%) while the unsaturated FAs in the control medium accounted for 62.4-62.7% (Figure 4b). As also reported in the literature, it can be suggested that the addition of xenobiotic compounds (i.e., phenolic substances found in various wastewaters, essential oils, etc.) has not any systematic effect upon the FA composition of the cellular lipids that were produced in various types of yeast species that were cultivated on sugar-or glycerol-based substrates [42,45].
Overall, in this study, novel wild-type and non-conventional yeast strains were screened under nitrogen-limited conditions on xylose-based media, demonstrating interesting DCW production, variable SCO, and polysaccharide synthesis and significant xylose assimilation. Thereafter, two of the previously screened strains belonging to R. toruloides were studied with regards to their SCO-producing capacities in xylose-based and LS-added media, mimicking industrial xylose-containing wastewaters. Lipids from both Rhodosporidium strains are of high importance due to their interesting lipid profile with respect to simultaneous PUFAs and carotenoids production. PUFAs are essential FAs that cannot be synthesized in humans and must be obtained from external sources such as vegetable oils and fish lipids. There is a critical role of EFAs and their metabolic products for the maintenance of the structural and functional integrity of the central nervous system and retina [58]. PUFA production from plant seeds is influenced by region, climate, and seasons, resulting in variable quantities and qualities of the oils. Increasing market demand for PUFA-containing commodity oils, inadequate supply of PUFAs from agricultural and animal sources, and recently, the latter developments and economic events (i.e., the economic crisis of the last years accompanied by the war in the Eastern Europe, and so on) resulted in a very important increase in the cost of several types of agro-industrial/agricultural products and edible commodities such as the plant-origin lipids (i.e., sunflower oil, soybean oil, etc.). Due to their resemblance in the fatty acid composition with the various types of plant-based commodity oils (i.e., soybean oil, rapeseed oil, olive pomace oil, etc.), and their potential of being produced through waste and residues valorization, SCOs do not have competition with food products and can be successfully implicated as non-conventional replacements of these mentioned plant-based commodity oils [59]. Finally, in our study, both R. toruloides strains that were tested produced carotenoids (data not shown) while visual observation indicated their interesting concentrations. Carotenoids have been widely known for their antioxidant, anticancer, and antimicrobial properties [60]. They could be incorporated into novel food-based formulas via micro-and nano-encapsulation, offering healthier food products with prolonged shelf life [60][61][62].
Funding: This research received no external funding.