Lipid and Carotenoid Production by a Rhodosporidium toruloides and Tetradesmus obliquus Mixed Culture Using Primary Brewery Wastewater Supplemented with Sugarcane Molasses and Urea

In this study, Rhodosporidium toruloides and Tetradesmus obliquus were used for lipid and carotenoid production in mixed cultures using primary brewery wastewater (PBWW) as a culture medium, supplemented with sugarcane molasses (SCM) as a carbon source and urea as a nitrogen source. To improve biomass, lipid, and carotenoid production by R. toruloides and T. obliquus mixed cultures, initial SCM concentrations ranging from 10 to 280 g L−1 were tested. The medium that allowed higher lipid content (26.2% w/w dry cell weight (DCW)) and higher carotenoid productivity (10.47 µg L−1 h−1) was the PBWW medium supplemented with 100 g L−1 of SCM and 2 g L−1 of urea, which was further used in the fed-batch mixed cultivation performed in a 7-L bioreactor. A maximum biomass concentration of 58.6 g L−1 and maximum lipid content of 31.2% w/w DCW were obtained in the fed-batch cultivation. PBWW supplemented with SCM was successfully used as a low-cost medium to produce lipids and carotenoids in a R. toruloides and T. obliquus mixed culture, with higher productivities than in pure cultures, which can significantly reduce the cost of the biofuels obtained.

coupled with fluorescent dyes. This technique allows, using a simple and easy method, discriminating the evolution of each microorganism throughout the mixed culture development, as well as monitoring the individual stress response of each microbial population [13].
The yeast Rhodosporidium toruloides and the microalga Tetradesmus obliquus have been widely reported as oleaginous microorganisms [1,[14][15][16]. Moreover, both species are able to produce carotenoids with a commercial interest: R. toruloides produces β-carotene, torulene, and torularhodin, which have a high interest in the food industry [1,9] and T. obliquus produces significant amounts of different carotenoids such as β-carotene or lutein [17,18]. If during the bioprocess microorganisms can produce, beyond the production of lipids, other high-value-added biocompounds, such as carotenoids, the biodiesel overall process cost could be reduced.
In this work, PBWW collected from a brewery company was used as feedstock for R. toruloides and T. obliquus mixed cultures. Due to a low carbon and nitrogen load in the effluent as observed by Dias et al. [9], PBWW was supplemented with SCM as a carbon source and urea as a nitrogen source. To improve the efficiency of the lipid and carotenoid production by R. toruloides and T. obliquus mixed cultures, different SCM concentrations ranging from 10 to 280 g L −1 were tested. The SCM concentration, which allowed higher biomass, lipid, and carotenoid production, was used in the medium composition of fedbatch mixed cultivation. The fed-batch cultivation was developed according to Dias et al. [1,9], which allowed extending the microorganisms' exponential and stationary growth phases and triggering the activation of secondary metabolic pathways. However, in this work, no dual-stage pH control strategy was used to stimulate the microorganisms' growth and metabolite production since the two microorganisms used in the mixed culture have distinct optimum pH values. Flow cytometry (FC) was used in all cultivations to monitor the microorganisms' proportions throughout the cultivations.

Primary Brewery Wastewater and Sugarcane Molasses
The PBWW used in this work was the same used by Dias et al. [12]. The characteristics of the PBWW used, in terms of waste quality parameters, are described in Table 1. As in the work of Dias et al. [12], PBWW was supplemented with SCM, which was provided by the Sidul Company at Alhandra, Lisbon, with a total sugar concentration of 1138.95 ± 5.02 g L −1 with 871.54 ± 3.37, 155.97 ± 3.48 and 111.44 ± 0.82 g L −1 of sucrose, glucose, and fructose, respectively.

Microorganisms
The microorganisms used in this work were as follows: the yeast Rhodosporidium toruloides NCYC 921, acquired from the National Collection of Yeast Cultures (Norwich, UK), which was stored at 4 °C on slants of malt extract agar; the microalga Tetradesmus obliquus (ACOI 204/07), purchased from the Coimbra University Culture Collection, Portugal. T. obliquus was maintained as described by Dias et al. [13].

Growth Conditions
Inoculum R. toruloides and T. obliquus inocula were performed according to Dias et al. [12].

Mixed Cultivations-Optimization of the SCM Concentration for Maximum Biomass, Lipid, and Carotenoid Production Using Shake Flasks
To optimize the SCM concentration to obtain maximum biomass, lipid, and carotenoid production by R. toruloides and T. obliquus mixed cultures, six batch experiments were performed using PBWW as the culture medium, supplemented with SCM as carbon source, at concentrations ranging from 70 to 280 g L −1 , and 2 g L −1 of urea as nitrogen source [9], in addition to the ones already presented by Dias et al. [12] with SCM concentrations of 10, 20, 40, and 100 g L −1 . The mixed culture experiments were performed according to Dias et al. [12] in 1-L baffled Erlenmeyer flasks containing 180 mL of the respective sterile culture medium ( Table 2). The initial medium pH of each mixed culture was adjusted to 6.0 by NaOH or HCl solution addition after sterilization, and the initial cell density of each microorganism in the mixed culture was 1.38 × 10 6 cells mL −1 (total of 2.75 × 10 6 cells mL −1 ) according to Dias et al. [12]. Experiments were conducted in duplicate, at 30 °C and 130 rpm in an orbital shaker (Unitrom Infors, Switzerland) under the continuous artificial light of two led strips with an average light intensity of 3.7 μmol photons m −2 s −1 (measured by Phywe Lux-meter). Mixed culture batch cultivations were monitored until the late stationary phase was reached.

Fed-Batch Mixed Cultivation
After selecting the best SCM concentration for maximum biomass, lipid, and carotenoid production by R. toruloides and T. obliquus mixed cultures, both microorganisms were grown under a fed-batch regime in a bench-scale 7-L bioreactor (FerMac 360 bioreactor, Electrolab Biotech, UK) using PWBB supplemented with 100 g L −1 of SCM and 2 g L −1 of urea (PWBB + 100SCM + U) as medium culture. A volume of 0.15 L of exponential growing yeast and microalgae cells in a proportion of 1:1 was used to inoculate the bioreactor containing an initial volume of 1.45 L of PBWW + 100SCM + U. The feeding strategy followed the same protocol described by Dias et al. [9]. After the batch cultivation period, the culture was fed with a concentrated solution containing PBWW, SCM, and urea (carbon (C) + nitrogen (N) solution, containing 4.67 g L −1 of urea and 262.8 ± 0.5 g L −1 of total sugars, in PBWW) using a peristaltic pump in order to extend the yeast exponential phase (active growth phase). As the culture entered the stationary phase, a concentrated SCM solution (carbon (C) solution, containing 295.5 ± 0.6 g L −1 of total sugars, in PBWW) was fed to the culture, using a peristaltic pump, in order to induce carbon excess conditions

Microorganisms' Cell Dry Weight Quantification
The microorganisms' DCW in the mixed cultures was quantified gravimetrically by centrifugation (Biofuge 15, Portugal), in triplicate, as described by Dias et al. [12].

Sugar Consumption Evaluation
The concentrations of sucrose, fructose, and glucose during the fed-batch cultivation were determined using HPLC with an Aminex HPX-87P column (Bio-Rad, USA) in an Agilent Chromatographer, equipped with a diode array detector (DAD) and a refractive index detector (RI). All samples were filtered through 0.22 µm membrane filters before analysis. The data obtained were analyzed with the Agilent ChemStation software.

Chemical Oxygen Demand, Kjeldahl Nitrogen, Ammonia Nitrogen, and Phosphorus Concentrations
Chemical oxygen demand (COD) was determined according to the method of Horwitz & Latimer [19], based on the oxidation of the organic matter by potassium dichromate in the presence of sulfuric acid and silver ions as a catalyst, and performed as described by Dias et al. [9]. The Kjeldahl nitrogen was determined according to the modified Kjeldahl method adapted from the standard method 4500-Norg B [20].
The ammonia nitrogen was analyzed using an ammonium selective electrode, CRISON-Multimeter MM 41.

Fatty Acid Analysis
Since the yeast saponifiable lipidic fraction is the one that is used for biodiesel purposes, the yeast lipids were converted into fatty acid methyl esters through a transesterification reaction and analyzed by gas-liquid chromatography using a SCION GC 436 chromatographer (Bruker, Germany), equipped with a flame ionization detector (FID), according to Freitas et al. [21].

Carotenoid Quantification
The carotenoid quantification at the end of the different batch and fed-batch cultivations was performed according to Dias et al. [9]. Total carotenoids were quantified by spectrophotometry (Hitachi-2000, Shanghai, PRC). Spectra were run between 380 and 700 nm, and the concentration of total carotenoids was calculated (in equivalents of β-carotene) using the Beer-Lambert equation at the maximum absorbance (450 nm). A value of 262 L g −1 cm −1 was used for the specific optical extinction coefficient (E 1% 1 cm ) [21]. HPLC was used to identify and quantify the major extracted carotenoids. The HPLC system used consisted of a liquid chromatograph Hewlett Packard HP-1100 series (Hewlett Packard, Waldrom, Germany), with a UV/Vis detector adjusted to the wavelength relative to the maximum of absorbance found for the main pigments (457 nm for beta-carotene and lutein, 489 nm for torulene, and 600 nm for chlorophyll a). A C18 reversed-phase column, Vydac 201 TP54, was used with a mobile phase of methanol:acetonitrile (90:10 v/v) at an isocratic flow rate of 1 mL min −1 . The identification of β-carotene and lutein was carried out by comparing the retention times of these carotenoids with those of the standard compounds (β-carotene: 97%, CalBiochem; lutein: 90%, Sigma). Moreover, the identification of chlorophylls and torulene was carried out by the elution of extracts from T. obliquus and R. toruloides, respectively, and a comparison of the obtained chromatogram peaks with data from literature [22][23][24]. Since lutein and chlorophylls are only produced by the microalgae and torulene is only obtained from the yeast, the presence of these pigments could be attributed solely to the existing metabolism of the respective microorganism.

Flow Cytometry
According to Dias et al. [13], flow cytometry (FC) was used to monitor R. toruloides and T. obliquus mixed cultivations, while carboxyfluorescein diacetate (CFDA) was used to perform the enzymatic activity studies to detect cell stress.

Microscopic Observations and Nile Red Fluorescence Microscopy
R. toruloides and T. obliquus mixed culture cells at the end of the fed-batch cultivation (142 h) were observed under the optical and fluorescence microscope, after staining with Nile Red dye. Five hundred microliters of a sample taken from the mixed culture was diluted in PBS buffer and mixed with 5 μL of Nile Red at 0.2 mg mL −1 and incubated in the dark for 30 min. After the incubation period, the cells were observed using an Olympus BX60 optical and fluorescence microscope equipped with a fluorescence illuminator and a frame camera with 100 × magnification (oil immersion lens).

Optimization of the SCM Concentration for Maximum Biomass, Lipid, and Carotenoid Production
In order to optimize the SCM concentration in the PBWW medium described by Dias et al. [12] to obtain maximum biomass, lipid, and carotenoid production by R. toruloides and T. obliquus mixed cultures, ten batch experiments were conducted in 1 L Erlenmeyer flask with initial SCM concentrations ranging from 10 to 280 g L −1 . The kinetic parameters obtained for these experiments are presented in Table 3. Maximum biomass concentration increased as the SCM concentration increased in the medium, reaching a plateau of around 30 g L −1 of maximum biomass concentration for SCM concentrations equal to or higher than 100 g L −1 (Table 3). However, at the SCM concentration of 280 g L −1 , the maximum biomass concentration obtained was 2.13 g L −1 ( Table 3). These results indicate that there was substrate saturation for SCM concentrations equal to or higher than 100 g L −1 , but for the SCM concentration of 280 g L −1 , substrate inhibition was observed. In terms of biomass productivity, the highest biomass productivity was reached in the PBWW mixed cultivation performed with 40 g L −1 of SCM (428.8 mg L −1 h −1 , Table 3) [12].
The specific growth rate (µ) increased as the SCM concentration increased in the medium, reaching its maximum of 0.24 h −1 in the essay with 100 g L −1 of SCM (Table 3). However, for assays with SCM concentrations higher than 100 g L −1 , the µ decreased significantly, reaching 0.01 h −1 for the cultivation in which 280 g L −1 of SCM was used (Table 3). These results demonstrate that the mixed culture growth was inhibited for initial SCM concentrations above 100 g L −1 .
Relatively to the lipid content, a maximum of 26.2% w/w of total fatty acid (TFA) was obtained for the cultivation using 100 g L −1 of SCM (Table 3) [12]. Nevertheless, the highest lipid productivity of 74.95 mg L −1 h −1 was obtained for the mixed cultivation using 70 g L −1 of SCM (Table 3). For lipid accumulation by oleaginous microorganisms such as R. toruloides and T. obliquus, the C/N ratio is an important factor [9] since it requires excess carbon over nitrogen to stimulate lipid production [9]. The media tested in the batch cultivations correspond to C/N ratios between 5 and 144 ( Table 2). In the conditions tested, the C/N ratio that allowed for higher lipid content was 52.
Considering the total carotenoid production at the end of the different batch experiments, although the highest maximum carotenoid content was obtained for the mixed culture performed with 10 g L −1 of SCM (63.6 µg g −1 ; Table 3), the carotenoid productivity for this cultivation was the lowest (Table 3). Indeed, the highest carotenoid productivity of 10.47 µg L −1 h −1 was obtained for the mixed culture performed with 100 g L −1 of SCM (Table 3). Table 4 presents the results obtained after the HPLC analysis of the carotenoid extracts at the end of the batch experiments. It was possible to identify the presence of the following main pigments: lutein, β-carotene, chlorophylls, and torulene. Lutein and chlorophylls are exclusively produced by T. obliquus, torulene is only produced by R. toruloides, while β-carotene is produced by the two microorganisms. Besides these pigments, it was also possible to observe the presence of other carotenoids that, although were not identified, could be related to those produced by both or each microorganism, through comparison with the HPLC chromatograms of the extracts from R. toruloides and T. obliquus. These unidentified pigments included a few more polar compounds (probably oxygenated carotenoids) produced by both microorganisms and other less polar carotenoids, e.g., carotenes, produced mainly by R. toruloides. Therefore, besides the four main pigments detected, in order to analyze and compare the pigments detected in the samples collected at the end of all cultivations, the other non-identified pigments present in the extracts were thought to be other carotenoids, mainly from R. toruloides, and carotenoids from both microorganisms. As it is possible to observe in Table 4, in the batch experiments with lower SCM concentrations (10 and 20 g L −1 ), the main pigments present in the extracts were lutein and chlorophylls, which are exclusively produced by T. obliquus (in the 10 and 20 g L −1 cultivations, 52.0% and 62.1% of the total carotenoids was lutein and 14.0 and 12.8 were chlorophylls, respectively, Table 4). This is in accordance with the FC results, as these two batch cultivations were the ones showing higher microalga proportions at the end of the cultivations (2.4% and 1.7% at the 10 and 20 g L −1 cultivations, respectively, Table 5). These results corroborate the conclusion that, although T. obliquus cells were present at low proportions in the mixed cultures with 10 and 20 g L −1 of SCM, the microalga cells were able to produce lutein and chlorophyll, the latter pigment related to the microalgal metabolic activity. On the other hand, the carotenoids attributed to R. toruloides increase their contribution to the pigment profile with the increase in the SCM concentration, reaching a value of almost 77% of the total carotenoids in the cultivation with initial SCM concentrations of 120 g L −1 . Conversely, the main pigments related to T. obliquus metabolism, lutein, and chlorophylls, decreased their contribution with the increase in SCM initial concentration from 120 to 200 g L −1 . Moreover, for the cultivations with the highest SCM initial concentration (240 g L −1 ), these T. obliquus main pigments were not detected ( Table 4). As previously referred to, T. obliquus growth was probably limited due to insufficient light penetration in the medium, which seriously affected the microalga growth and the carotenoid production. Table 5 presents the proportions of R. toruloides and T. obliquus cells obtained by FC in the batch mixed cultivations performed with different initial SCM concentrations. In all cultivations, the yeast cells dominated the cultivations, representing, at the end of the cultivations, between 97.6% of all cells in the cultivation with 10 g L −1 of SCM and 99.9% in the cultivations with 200 and 240 g L −1 of SCM (Table 5). Since R. toruloides is a heterotrophic organism, it will use the carbon present in the medium to produce biomass and other valuable biocompounds such as lipids and carotenoids. On the contrary, T. obliquus can grow under photoautotrophic, heterotrophic, or mixotrophic conditions, depending on the environment and nutrient availability. However, as the yeast grows faster than the microalgae, the microalgae growth is limited. In addition, as the yeast biomass concentration attained 30 g L −1 in some cultivations, or even more, the light penetration into the medium could have been insufficient for the microalga's autotrophic metabolism, which might have been hindered by self-shading and light absorption. In fact, the batch cultivation where lower maximum biomass was obtained was the assay with 10 g L −1 of SCM, which attained 3.73 g L −1 of maximum biomass (Table 3) and was also the one with a higher percentage of T. obliquus at the end of the cultivation (2.4% of microalga cells, Table 4). Martins et al. [25] studied the effect of high glucose concentrations (10-200 g L −1 ) on the growth and lipid production of R. toruloides NCYC 921, grown on lignocellulosic hydrolysate (obtained from Miscanthus biomass). The authors reported maximum biomass concentrations between 2.24 and 2.92 g L −1 for the different yeast cultivations developed on synthetic media with glucose, which are considerably lower values than the ones obtained in this work using PBWW media supplemented with SCM and urea. Also, the maximum lipid content obtained by the authors was 11.1% w/w of TFA for the 90 g L −1 glucose concentration assay [25], a lower value than the one obtained in this work by the mixed culture with 100 g L −1 of SCM (26.3% w/w of TFA). These results may suggest Table 3 Rhodosporidium toruloides and Tetradesmus obliquus kinetic parameters obtained in the batch mixed cultivations grown on primary brewery wastewater media supplemented with initial sugarcane molasses (SCM) ranging from 10 to 280 and 2 g L −1 of urea a Biomass productivity (mg L −1 h −1 ) for the maximum biomass concentration was calculated as follows:  the advantage of using mixed yeast and microalgae cultures grown on PBWW and SBWW over the yeast pure culture grown on synthetic media with glucose. This step aimed to select the SCM concentration that induced higher lipid and carotenoid production for a R. toruloides and T. obliquus mixed culture grown in a PBWW medium. To understand if these results were statistically relevant, ANOVA analysis was performed on the lipid and carotenoid data displayed in Table 3. All the results confirmed that the data obtained were statistically significant (lipid content: F-ratio = 11.51; p = 0.004; lipid productivity: F-ratio = 7.14; p = 0.017; carotenoid content: F-ratio = 6.46; p = 0.022; Table 4 Tetradesmus obliquus and Rhodosporidium toruloides pigment profile obtained at the end of the different batch mixed cultivations grown on primary brewery wastewater media supplemented with initial sugarcane molasses (SCM) ranging from 10 to 280 and 2 g L −1 of urea   carotenoid productivity: F-ratio = 13.99; p = 0.002). Thereby, the medium that allowed higher lipid content (26.2% w/w, Table 3) and higher carotenoid productivity (10.47 µg L −1 h −1 , Table 3) was the PBWW medium supplemented with 100 g L −1 of SCM and 2 g L −1 of urea. Despite the assay containing 70 g L −1 SCM achieving higher lipid productivity, since carotenoids have a higher market value than biodiesel, the medium formulation that attained higher carotenoid productivity was the one containing 100 g L −1 of SCM, which was selected for the medium composition used in the fed-batch mixed cultivation.

Flow Cytometry Controls
FC is a useful tool to identify and analyze cells in suspension. It allows for the identification of cells from debris/particles, based on their light scatter properties. However, when the media used for cultivation is full of waste particles, such as the medium used in this work in the fed-batch cultivation (PBWW supplemented with 100 g L −1 of SCM and 2 g L −1 of urea), it is important to assure the microbial cell differentiation from the background. Figure 1a, b presents the FSC/SSC and the PC7-A/SSC dot plots showing the noise generated by the PBWW medium supplemented with 100 g L −1 of SCM and 2 g L −1 of urea. After inoculation, it is possible to identify two new populations, concerning the yeasts and the microalgae populations (Fig. 1c, d).
Differentiation between R. toruloides and T. obliquus cells was performed based on their different sizes and internal complexities detected in the FSC/SSC dot plot (Fig. 1c) and by back-gating the population displayed in the PC7-A/SSC dot plot, which shows chlorophyll fluorescence emitted by the microalgae cells (Fig. 1d). As it is possible to observe in Fig. 1a, b, the yeast and microalgae gates do not overlap the medium particles, which prove the effectiveness of using FC to differentiate two microbial populations during the fed-batch mixed cultivation.

Fed-Batch Cultivation
In order to achieve higher biomass, lipid, and carotenoid production than those obtained in the previous batch cultivations, fed-batch mixed cultivation was performed in a 7-L bioreactor, and the results obtained are presented in Fig. 2. The batch phase lasted 23 h, attaining the mixed culture, at this time, a biomass concentration of 16.1 g L −1 (Fig. 2a, b). Although initially, at inoculation time, the proportion of cells of each of the microorganisms was close (41.63% and 58.37% of yeast and microalgae cells in the mixed culture, respectively, Fig. 3a), at 23 h of cultivation, the proportion of yeast cells had risen to 99.58% (Fig. 3b), compared to the proportion of microalga cells that was only 0.42% (Fig. 3b).
Even though at this time the sugar concentration was not close to zero (Fig. 2c), the levels of total nitrogen decreased sharply (Fig. 2d), as well as the dissolved oxygen percentage (DO) which was 0% at this time (Fig. 2e).
At 23 h, a fed-batch phase was started by the addition of a PBWW + SCM + U solution (C + N solution) and the speed rate was increased to 600 rpm (Fig. 2f) in order to enhance the oxygen availability in the medium, which was, afterward, always higher than 50% until the end of the experiment (Fig. 2e). At 96 h, the microorganisms' biomass concentration stopped increasing (Fig. 2a; Table 6). At this time, the feeding solution was changed to a PBWW + SCM solution (C solution) in order to induce lipid and carotenoid production 1 3 (Fig. 2). A maximum biomass concentration of 58.6 g L −1 was achieved at 96 h and the maximum biomass productivity was obtained at 32 h (0.79 g L −1 h −1 ; Fig. 2a; Table 6).
Dias et al. [9] used the same fed-batch strategy to produce lipids and carotenoids from a R. toruloides pure culture using secondary brewery wastewater (SBWW) supplemented with 10 g L −1 of SCM and 2 g L −1 of urea. The authors reported lower maximum biomass concentration (42.5 g L −1 at 126.5 h) and lower maximum productivity (0.55 g L −1 h −1 , at 48.25 h) (Table 7) as expected, since this present work, initially, used 100 g L −1 of SCM, and some synergistic (symbiotically) behavior of the mixed culture was highlighted. Additionally, none of the other yeasts and microalgae pure and mixed cultures developed in brewery effluents presented in Table 7 attained biomass concentrations close to the ones obtained in this work. However, in most of the previous works presented in Table 7, lower carbon concentrations were used by the authors, which can explain the lower biomass concentrations. For instance, in Dias et al. [12] and Schneider et al. [10], 10 g L −1 and 9.38 g L −1 of initial carbon concentrations were used, respectively, which resulted, at the end of the batch and fed-batch cultivations, in 42.5 and 7.38 g L −1 of biomass, respectively (Table 7). In addition, these authors used shake flasks to develop the cultures, which  Fig. 1 a,  contributed to the lower final biomass concentrations since mass transfer limitations (particularly oxygen) often occur in shake flask cultivations due to inefficient agitation and aeration, and no pH control was performed. When Tetradesmus (formerly Scenedesmus) obliquus (SAG 276-3d) was developed on pure brewery cultivations without carbon supplementation, lower biomass productions (0.90 g L −1 ) were reported [26]. Marchão et al. [11] obtained 0.95 g L −1 of microalgae biomass when Scenedesmus obliquus (ACOI 204/07) was cultivated in pure BWW (Table 7). Relatively to the mixed cultures presented in Table 7, Dias et al. [27] performed brewery wastewater cultivations supplemented with 10 g L −1 of SCM, without any nitrogen supplementation ( Table 7). The maximum biomass production obtained by the authors was only 2.17 g L −1 . However, when the PBWW was supplemented with 10 g L −1 and 2 g L −1 of urea, the biomass production attained 3.73 g L −1 [12]. With the increase of the SCM concentration to 100 g L −1 , the biomass production by the mixed culture grown in shake flasks increased to 30.60 g L −1 (Table 7), a lower value than the one obtained in this work in the fed-batch cultivation (58.6 g L −1 , Table 7), as expected since they grow in bioreactors, which allow better mass transfer conditions than in shake flasks and pH control is carried out, keeping this parameter longer in the comfort zone. Indeed, the specific growth rate (µ) of 0.14 h −1 ( Table 6) observed for the fed-batch mixed cultivation presented in this work was considerably higher than the one reported by Dias et al. [9] (0.0453 h −1 ), since the carbon concentration was tenfold higher in the present work and the former cultivation was conducted in a bioreactor.
As previously observed for the batch mixed cultivations, the yeast dominated the fedbatch mixed cultivations (Fig. 3b-e). T. obliquus cells were observed at low proportions throughout all the mixed cultures by FC (Fig. 3) and by microscopic observations (Fig. 4a). Figure 4a, b presents the microscopic observation of R. toruloides and T. obliquus cells at the end of the fed-batch mixed cultivation (142 h). As it is possible to observe, most of the cells in the mixed culture were yeast, as confirmed by FC (Fig. 3b-e and Fig. 4).
Concerning the lipid production, as it is possible to observe in Fig. 2g, there was an increase throughout the cultivation, likely due to the contribution of growth-associated lipids [21]. Maximum lipid productivity of 0.17 g L −1 h −1 was obtained at 74 h ( Fig. 2g; Table 6) and maximum lipid content of 31.2% w/w of total fatty acid (TFA) was obtained at 104 h ( Fig. 2g; Table 6), which is a higher value than all of the lipid contents reported so far for yeasts and microalgae pure cultures grown on brewery effluents, as shown in Table 7. Yen et al. [28] reported a lipid content of 38% (w/w) for a R. glutinis and S. obliquus mixed culture using glucose as a carbon source, a higher value than the one obtained in this work. However, since glucose is an expensive carbon source and its use increases the process costs, especially at a large scale, alternative and cheaper carbon sources must be used. Relatively to the ratio of saturated fatty acids (SFA), monounsaturated fatty acids (MUFA), and polyunsaturated acids (PUFA) during the fed-batch cultivation, although initially in the batch phase, the proportion of SFA, MUFA, and PUFA were very close (~ 30% w/w of TFA, Fig. 2h), at the end of the batch phase, SFA represented 28.2% w/w of TFA, MUFA represented 63.1% w/w of TFA, and PUFA represented 8.7% w/w of TFA (Fig. 2h). These proportions were kept practically stable during the fed-batch phase (Fig. 2h). Except in the batch phase, during which linolenic methyl ester (18:3) was produced by the mixed culture but at low proportions, during the fed-batch phase, there was no production of C18:3 by the mixed culture, which is an advantage for biodiesel production, as C18:3 has a high rate of oxidation and reduces the biodiesel quality.
At the end of the fed-batch mixed cultivation, 0.073 mg g −1 of carotenoid content was obtained, which represents carotenoid productivity of 0.022 mg L −1 h −1 ( Table 6). Dias et al. [9] obtained a maximum carotenoid content of 0.23 mg g −1 and maximum carotenoid productivity of 0.069 mg L −1 h −1 (Table 7), which are higher values than the ones obtained by the mixed culture reported for this work. Schneider et al. [10] also investigated the production of microbial lipids for biodiesel production and high-value carotenoids by a Rhodotorula glutinis pure culture using, as a culture medium, industrial brewery effluent supplemented with glucose as a carbon source. The authors reported a carotenoid content of 0.16 mg g −1 , which is also a higher value than the value obtained in this work for the mixed culture, although the lipid content was lower than the one here reported (Table 7).
On contrary to what was observed for the biomass and lipid production, the carotenoid production was not improved during the R. toruloides and T. obliquus mixed cultivation. Lipids are growth-associated compounds [21] that follow the same trend as biomass concentration, being produced during the microorganisms' active growth phase. However, according to Freitas et al. [21], carotenoids are mixed products as they are associated with yeast growth and are also produced during the stationary phase; but, the medium pH strongly influences the microorganisms' growth and metabolite production. Dias et al. [14] studied the effect of the medium pH on Rhodosporidium toruloides NCYC 921 carotenoid and lipid production evaluated by flow cytometry. The authors concluded that the yeast biomass concentration was maximal at pH 4.0 as well as the lipid production, and the maximum carotenoid content was obtained for pH 5.0. More precisely, the authors observed an almost twofold higher the carotenoid content at pH 5.0 compared to the results obtained for pH 6.0 (63.37 µg g −1 and 36.15 µg g −1 for pH 5.0 and 6.0, respectively) [14]. This can explain the lower carotenoid content obtained by the mixed culture presented in this work. Dias et al. [1,9] performed the yeast pure fed-batch cultivation using a dual-stage pH control to improve the lipid and carotenoid production by the yeast: the active yeast growth was performed using pH 4, and the lipid accumulation phase was 429 performed using pH 5. However, the PBWW fed-batch mixed cultivation described in the present work was totally performed at pH 6 since the dual-stage pH control strategy performed by Dias et al. [9] was not possible to perform in this case, as pH 4 and 5 are not adequate pH values for the growth of T. obliquus.
In terms of sugar consumption, at inoculation time, 83.4% of the total sugars were sucrose (Fig. 2c) and only 9.9 and 6.7% were glucose and fructose, respectively (Fig. 2c). At 23 h, all the sucrose had been converted to glucose and fructose (Fig. 2c). The levels of glucose were always lower than the fructose ones, reaching 0 g L −1 at 96 h, despite the presence of 47.2 g L −1 of fructose, at that time, in the cultivation broth (Fig. 2c). These results show the clear preference of the microorganisms for sucrose and glucose over fructose during the mixed culture evolution. Fig. 2 Different parameters obtained during Rhodosporidium toruloides and Tetradesmus obliquus fedbatch mixed cultivation using primary brewery wastewater supplemented with 100 g L −1 of sugarcane molasses and 2 g L −1 of urea. Biomass values represent the average value of two independent replicates with a standard deviation lower than 10% (n = 2). The fatty acid values were obtained from two independent replicates, with a standard deviation not exceeding 10%. Biomass productivity (g L −1 h −1 ) was calculated as follows: (X t − X 0 ) / (t t − t 0 ), where X t is the biomass concentration at the instant t, X 0 is the biomass concentration at t = 0. Total TFA productivity (g L −1 h − . 1 ) was calculated as follows: (TFA t − TFA 0 ) / (t t − t 0 ), where TFA t is the TFA concentration at the instant t, TFA 0 is the TFA at t = 0. Values of Kjeldahl nitrogen, ammonium nitrogen, COD, and sugar concentration represent the average of the values obtained from two independent replicates with a standard deviation lower than 10% (n = 2) for the different parameters Relatively to the total nitrogen levels throughout the cultivation, a sharp decrease was observed during the batch phase as above mentioned (Fig. 2d), which represented 67.6% of total nitrogen removal during this phase. During the fed-batch phase, there was an accumulation of total nitrogen in the medium. The same pattern was observed for the ammonia nitrogen: there was a slight consumption during the batch phase (30% consumption), followed by accumulation in the C + N solution fed-batch phase (Fig. 2d). When the feeding solution was changed to the C solution, the levels of ammonia nitrogen started to decrease as urea was no longer added to the culture medium (Fig. 2d). However, even though urea was no longer added to the culture broth, the levels of total N increased until 120 h (Fig. 2d). The reason for this, as referred by Dias et al. [9], is related to the SCM that contains nitrogenous organic compounds that increase the level of N in the medium.
In terms of COD removal, in the batch phase, 38.8% of the COD was removed (Fig. 2i). Afterward, with the addition of the feeding solutions (C + N and C), an accumulation of COD in the culture broth was recorded. Neither of the wastewater parameters studied was in accordance with the emission value limits (ELV) of discharge, as the supplementation of the brewery effluent with SCM constrains the effluent treatmen, in terms of organic carbon and nitrogen removal [9]. However, the main goal of this work was to obtain a low-cost primary brewery effluent medium to produce lipids and carotenoids by a R. toruloides and T. obliquus mixed culture, and this goal was achieved. PBWW supplemented with SCM and urea has a residual cost when compared to traditional minimal media supplemented with glucose: brewery wastewater is a waste from the brewery industry and has no residual cost, being, inclusively, a burden for the companies that have to support the effluent treatment cost before it can be discharged into the environment; SCM is a sub-product of the sugar industry and has a residual cost of about 50€/ton (www. aliba ba. com) and urea costs, approximately, 250€/ton (www. aliba ba. com). However, traditional media supplemented with glucose as a carbon source are considerably more expensive than the medium used in this work, as a ton of glucose can cost 500€ or more (www. aliba ba. com). The approach here reported can be successfully used as a low-cost medium to produce lipids and carotenoids by a R. toruloides and T. obliquus mixed culture, which can reduce significantly the cost of the biofuels obtained.
In order to visualize the intracellular lipid droplets, the mixed culture cells were stained with Nile Red (Fig. 5). The cells were effectively stained with Nile Red, which confirms the presence of the lipid droplets inside the cells. Two types of Nile Red fluorescence can be observed in Fig. 5: the yellow Nile Red fluorescence represents neutral lipids and the red Nile Red fluorescence represents polar lipids. Figure 6 Fig. 3 a-e FSC versus SSC dot plots concerning the fed-batch mixed cultivation of Rhosdosporidium toruloides and Tetradesmus obliquus using primary brewery wastewater supplemented with 100 g L −1 of sugarcane molasses and 2 g L. −1 of urea this cultivation, CFDA results for the microalga are not presented. During all the cultivation, R. toruloides presented more than 90% of the cells with enzymatic activity, indicating that R. toruloides cells were adapted to the growth conditions and maintained their enzymatic systems active throughout all the cultivation (Fig. 6a-c).

Conclusions
Primary brewery wastewater (PBWW) is a promising low-cost substrate for microbial lipid production. In this work, PBWW supplemented with sugarcane molasses (SCM) and urea was used as the culture medium for Rhodosporidium toruloides and Tetradesmus obliquus mixed cultures. It was demonstrated that PBWW supplemented with SCM provides a lowcost medium for lipid and carotenoid production by a mixed culture of R. toruloides and T. obliquus. The strategy here presented can be a way to reduce the cost of biodiesel production as industrial effluents and low-cost substrates (SCM) are used as culture medium while producing, at the same time, carotenoids with high market value.