Introduction

The development and implementation of a circular bio-based economy have been extensively promoted in recent years as an alternative to the use of fossil resources. This transformative process will definitely contribute to the UN’s Sustainable Development Goals, involving a wide range of economic sectors and industries (e.g., food, forestry, pharmaceuticals, chemistry and textiles) [1]. Among the diverse bioproducts available, biofuels have a central role in develo** such a bioeconomy, and extensive research efforts have been targeted at placing these compounds on the market. Independently of the raw material from which biofuels are produced, these bio-based products can enable energy independence, reduce greenhouse gas emissions, and enhance sustainable economic development [2]. In addition, costs can be decreased by the use of residual products. Among the different residues, low-input farming have the smallest carbon and nitrogen footprints and are usually cheaper feedstocks.

Biodiesel is the second most-produced liquid biofuel after bioethanol worldwide, reaching 48 billion litres in 2021 [2]. Current biodiesel production is based on vegetable oils and animal fats. These raw materials are usually associated with limited availability, competition with arable land and high costs. In contrast, single-cell oils (SCOs, also known as microbial lipids) obtained from oleaginous microorganisms have been positioned as a promising alternative for intermediate hydrocarbon biofuel production. In addition to having a similar composition to vegetable oils, these feedstocks do not depend on geographic location, seasonal changes, harvest schedules or transportation, and they usually exhibit fast production rates and are easy to scale up for industrial processing [3, 4]. However, large-scale biodiesel production from microbial oils still requires overcoming some obstacles, including harsh microbial culture conditions and difficulties during microbial harvesting and oil extraction [5].

Oleaginous microorganisms are bacteria, yeasts, filamentous fungi, and microalgae that may accumulate lipids intracellularly with concentration of over 20% (w/w) of their dry weight and, under special condition, up to 80% (w/w) [5,6,13].

Productivity in terms of high concentrations generated in relatively short periods of time is one of the most critical considerations in microbial oil production. Furthermore, an oil content of 40% has been set as the threshold-accumulated amount to reduce waste production during downstream processing [14]. Optimising process parameters and designing effective fermentation strategies have been the main targets for improving the conversion of raw materials into SCO [15].

Yeast cells trigger lipid production mainly under nutrient-limiting conditions and in the presence of high carbon/nitrogen (C/N) ratios [5, 16, 17]. Nitrogen, sulphur and phosphorus are the main nutrient-limiting factors used to activate lipogenesis. However, low concentrations of a specific nutrient usually result in low cell biomass and may even cause cell growth inhibition [18]. André et al. reported a maximum cell biomass concentration of 4.2–8.2 g/L dry cell weight (DCW) when the yeast Yarrowia lipolytica was grown on a medium having crude glycerol (30 g/L) and ammonium sulphate and yeast extract (0.5 g/L each) as major carbon and nitrogen sources. Although the lipid accumulation found was low (maximum 0.20 g lipids/g DCW), non-negligible amounts of citric acid and mannitol were found in the growth medium. [19]. Similarly, Rhodotorula glutinis reached a cell biomass of 5.3 g/L when grown in pure glycerol at about 700 C/N ratio [20]. Glucose-based media have also yielded low cell biomass concentration (4.5 g DCW /L) when using Y. lipolytica at a C/N ratio of about 75 [21]. High lipid accumulation (47.5 w/w) was obtained using Y. lipolytica in a glucose-based media limited in nitrogen and magnesium [22]. Subsequently, by applying an adaptive laboratory evolution strategy of the same strain used in the work, as mentioned earlier, it was possible to increase the lipid content by 30%. The evolved strain was obtained after 77 generations in lack of nitrogen and magnesium while using glucose as a carbon source [23]. Final cell biomass concentrations are directly linked to the final microbial oil concentrations. Hence, final cell biomass concentrations of about 4.5–6.5 g/L DCW can result in up to 2 g/L of lipids, with oil content ranging from 30 to 45% [21,22,23]. Y. lipolytica, has also shown the capacity to accumulate endo-polysaccharides in the early stages of culture despite the presence of nitrogen in the medium [24]. This interesting trait has also been reported in other yeast such as Rhodosporidium toruloides, Cryptococcus curvatus and Lipomyces starkeyi. These endo-polysaccharides can be therefor considered as advantageous co-products increasing the economic viability and sustainability of the SCO bioprocess [25,26,27].

In contrast to the batch operational mode, using fed-batch and continuous cultivation modes can improve substrate utilisation, thus contributing to higher conversion yields [28, 29]. These alternatives favour greater productivity and prevent the systems from inhibiting cell growth. Karamerou and Webb have reviewed how these operational modes influence the process to provide the necessary nutrients during each growth phase and support high lipid production titers [13].

In fed-batch culture, the substrate is fed to the system at different stages during the cultivations phase, while the product remains in the bioreactor until the end of the run. Two common fed-batch strategies are used for SCO production: (1) fed-batch cultivations with pulse-feeding medium addition and (2) fed-batch cultivations with continuous medium supply. Regardless of the strategy use, fed-batch cultivation effectively increases final cell biomass and lipid concentrations [30]. For instance, a fed-batch strategy using a glucose medium increased the cell biomass concentration of Candida viswanathii from 17 g/L DCW with the batch strategy to 21 g/L DCW [31]. In addition to the higher final cell biomass concentration, these authors reported an increase in the oil content from 33 to 50%, doubling the final lipid concentration (from 5.6 g/L to 10.5 g/L). Raimondi et al. also increased the final oil concentration obtained with the yeast Candida freyschussii using a fed-batch strategy with pure glycerol media [32]. In that study, 2-pulse, 4-pulse and continuous fed-batches increased SCO content 2.0-, 4.4- and 6.1-fold, respectively, for final lipid concentrations of 9.1 g/L, 20 g/L and 28 g/L compared to the 4.6 g/L concentration obtained with the batch strategy. In contrast to the study with C. viswanathii which increased its intracellular oil content by 1.5 times [31], C. freyschussii always produced an oil content in the range of 30–35% independent of the operation mode and the feeding strategy [32].

The fed-batch operation mode with a pulsed medium addition strategy can follow different feeding patterns based on parameters such as a specific nutrient concentration (commonly the carbon source), the dissolved oxygen levels and the fermentation time [20, 33,34,35]. One of the main approaches used for the further addition of the carbon source is when this nutrient is at low levels. Maina et al. fed the system with a glucose solution when the concentration of this component decreased to around 5 g/L [33]. A similar strategy was followed by Thiru et al., where the system was fed with a concentrated glycerol solution when the concentration was below 3 g/L [34]. This feeding strategy contributes to increased final cell biomass concentrations. For instance, a final cell biomass concentration as high as 110 g/L DCW was obtained by Tsakona et al. when growing Lipomyces starkeyi in a flour-rich hydrolysate using pulse addition when glucose content dropped below 20 g/L [36]. The relative dissolved oxygen level is another parameter that can be used to design the feeding strategy. Dissolved oxygen content increases when the metabolic activity of the fermentative microorganism decreases, mainly due to carbon source depletion. When observing such a shift in the system, the carbon source is then fed accordingly. This strategy was followed by Meesters et al. using Cryptococcus curvatus as the fermentative microorganism and 87% pure glycerol as the carbon source [37]. Although very high final cell biomass concentrations of up to 118 g/L were obtained, the authors reported a maximum oil content of only 25%. In addition to monitoring the carbon concentration and oxygen levels, the feeding strategy can be done by simply considering a specific time interval. R. glutinis was cultured in a glycerol-based medium following a fed-batch strategy where the fresh substrate was added to the system every 24 h, almost doubling the final cell biomass concentration (9.4 g/L vs. 5.3 g/L) and the lipid content (2.6 g/L vs. 1.7 g/L) [20]. Uçkun Kiran et al. also followed this process scheme with Rhodosporidium toruloides using crude glycerol as the carbon source [38]. After a batch phase of 72 h, crude glycerol was added to the system every 24 h to achieve final concentrations of 70–90 g/L. This approach produced a final cell biomass concentration of 23.1 g/L and a lipid content of 9.4 g/L, compared to 12.1 g/L and 6.1 g/L, respectively, obtained in the batch mode.

As an alternative to the pulse-feeding strategy, the substrate can be fed to the system continuously. With the continuous fed-batch strategy, finding the optimal feeding rate is of utmost importance for effective lipid production, although only a few studies have addressed this point [13, 31, 32, 39]. For instance, Zhao et al. designed a continuous feeding strategy to keep the glucose concentration at low levels (< 5 g/L) during the feeding phase [39]. This process scheme increased the cell biomass concentration and oil content from 89 g/L DCW and 52.2% to 127.5 g/L DCW and 61.8% compared to a pulsed-feeding strategy. In contrast, Raimondi et al. followed a continuous feeding strategy using glycerol as the carbon source, which resulted in glycerol accumulation in the medium at the initial stages of substrate addition [32]. However, it also produced a higher cell biomass concentration when compared to the pulsed-feeding strategies. Other alternatives to the constant feeding of the substrate for SCO production are exponential and variable feeding, where the feeding rate is adjusted according to specific equations or to maintain a particular parameter (e.g. C/N ratio) [40, 41]. These strategies have also resulted in high cell biomass concentration with a high accumulation percentage. For example, fed-batch cultivation of R. glutinis with exponential feeding resulted in a final cell biomass concentration of about 40 g/L with an oil content of 43% [41]. On the other hand, a final cell biomass concentration as high as 132 g/L DCW with an oil content of about 55% was observed when culturing Trichosporon oleaginosus with a feeding strategy based on maintaining a constant C/N ratio [40].

During fed-batch fermentation, it is essential to consider the dilution effect to avoid reducing cell density and to keep nutrient concentrations at adequate levels [13]. In this context, using concentrated substrate solutions containing all required nutrients during the feeding phase is essential [32].

Continuous cultivation in SCO production is also of interest because it increases productivity compared to the batch fermentation mode [18]. This system continuously provides carbon and nitrogen sources while allowing continuous collection of products and cells. A constant C/N ratio is preserved when reaching the steady state phase at a specific dilution rate (D). Under these conditions, cell concentration remains constant since growth and substrate uptake rates do not vary, while lipid production is promoted by controlling the nitrogen concentration. The dilution rate is the main parameter affecting the final cell biomass concentration and lipid content. In this regard, Papanikolaou and Aggelis observed a reduction in the lipid content from 3.5 to 0.3 g/L and in the cell biomass concentration from 8.1 to 3.8 g/L after increasing the dilution rate from 0.03 h-1 to 0.13 h-1 during the continuous cultivation of Y. lipolytica LGAM S(7)1 in crude glycerol [42]. Similar results have also been found when using other carbon sources and microbial strains. For instance, continuous fermentation of R. toruloides AS 2.1389 in glucose media at D ranging 0.02–0.20 h-1 resulted in cell biomass concentrations of 1.63–8.67 g/L DCW and lipids content of 0.21–5.36 g/L, the lower values being obtained with the higher dilution rates [43]. Similarly, cell biomass concentration and lipid content in C. curvatus decreased from 5.1 to 0.8 g DCW/L and 3.4 to 0.12 g lipid/L after increasing D values from 0.01 to 0.11 h-1 in an acetate-based medium [44].

In addition to affecting lipid production, the feeding rate in continuous cultivation is critical in directing carbon flux and cellular metabolism to energy, maintenance and product synthesis [42, 45]. This system benefits from continuously harvesting cells that can be processed directly after collection. Nevertheless, long-term fermentation can suffer from media sedimentation and biofilm formation. Furthermore, there is a high risk of contamination, and the yield can fluctuate depending on cellular changes [15].

Other advanced cultivation systems have been developed to optimise the process of SCO production. One such system is two-stage fermentation, which first targets producing cell biomass in nutrient-rich conditions and then on lipid accumulation under nutrient limitations and in an excess of carbon [46]. Therefore, the two-stage lipid production process focuses seeks to improve lipid production by optimising both cell proliferation and lipid accumulation stages. The accumulation stage requires optimisation when using this cultivation system, since cell growth only requires nutrient-rich media. In this context, these cultivation systems can follow a two-stage batch or two-stage strategy with a feeding addition [45, 47].

Lignocellulosic biomass and industrial wastes as raw material

During SCO production, the cost of raw materials can account for 40–70% of the total overall costs, with highly refined polysaccharides such as industrial-grade glucose and sucrose accounting for about 80% of such costs [47,48,49,50,51]. Using low-cost substrates, including lignocellulosic biomass and industrial waste, are promising alternatives for microbial lipid production [52,53,54].

Residual lignocellulosic biomass and waste-derived substrates are excellent candidates as carbon sources to make this process both cost-effective and environmentally friendly. In this context, sugarcane bagasse residues, cheese whey, corn stover, potato wastewater or orange peel extracts have been previously investigated for SCO production [55,56,57,58,59]. Table 1 summarises the most recent literature on using biomass residues for SCO production, mainly focusing on lignocellulosic sources.

Oleaginous yeasts can also use hydrophilic and hydrophobic substrates to accumulate lipids via two pathways: de novo and ex novo lipid synthesis [60]. Notably, both pathways offer distinct advantages: de novo lipid fermentation can generate a high quantity of lipids. In contrast, ex novo lipid fermentation can modify the lipid compositions to satisfy the requirements of the chemical or food industries. Thus, improving and upgrading fatty materials utilized as substrates can produce “tailor-made” lipids of high-added value [61]. The combined production of de novo and ex novo lipids has been studied, for instance, by using the yeast Trichosporon dermatis and a mixed medium combining an acid hydrolysate of corn cob and a soybean. That study confirms the potential of using both substrates (hydrophilic and hydrophobic) for the production of lipids with application in the production of biodiesel or lipid-based chemical compounds. [

Fig. 2
figure 2

Important aspects requiring further investigation towards reaching cost-effective microbial oil production

Full size image

Process optimisation is also an important element to consider during microbial oil production. In this sense, it is important to improve further the cultivation strategy by selecting the best bioreactor type (e.g., stirred tank, airlift/bubble columns), working with high cell densities, finding the appropriate substrate feeding system during the accumulation phase (e.g., pulse-feeding, two-batch and continuous feeding strategy), and evaluating effective co-cultivation methods allow increased total lipid production and maximising co-product recovery. Optimisation of process parameters, including media composition, pH, temperature and aeration rate, is also crucial. The use of new technology using big data analytics and/or machine learning algorithms might contribute towards process optimisation. Finally, downstream processing also demands further research efforts in order to develop and implement novel methods for product recovery, including the use of solvent-free or green solvent methodologies to improve the sustainability of the overall process scheme. These downstream processing technologies must also be targeted at develo** new processes allowing the continuous treatment of cell biomass and harvesting the product of interest.