Oleaginous microorganisms can accumulate considerable amounts of lipids under various nutrient limitation conditions, among which nitrogen limitation is recognized as the most successful induction strategy, and is most widely used. Large amounts of omics data had shown some clues of the regulation of carbon flux redistribution to lipids. For example, proteomic analysis demonstrated that nitrogen depletion can upregulate the glycolytic pathway, while the activity of TCA cycle was retarded, thus, leading more carbon flux to fatty acid biosynthesis in Mucor circinelloides (Tang et al., 2016). Other types of nutrient starvation that can enhance lipid accumulation include phosphate, silicon, and sulfate limitation (Wu et al., 2011; Ren et al., 2013; Bellou et al., 2016b; Ota et al., 2016).

Generally, two different pathways are involved in lipid synthesis: de novo lipid synthesis and ex novo lipid synthesis. Fundamental differences on the biochemical level exist between de novo lipid synthesis from hydrophilic substrates and ex novo lipid synthesis from hydrophobic substrates (Donot et al., 2014). Some studies have reported that nutrient limitation is only effective in the induction of de novo lipid synthesis, when growth is carried out on various hydrophilic substances (Papanikolaou and Aggelis, 2011a,b; Donot et al., 2014). Furthermore, nutrient limitation will induce lipid biosynthesis while concomitantly causing growth inhibition (Sajbidor et al., 1990; Nie et al., 2014; Zienkiewicz et al., 2016). However, it is a significant goal to obtain both a great quantity of biomass and a high lipid content for improved lipid productivity. Thus, in practice, oleaginous microorganisms are first cultured in nutrient-rich media for rapid propagation in early process stages, while nutrient limitation is introduced in later stages to induce the overproduction of lipids (Lin et al., 2011; Zhu et al., 2016). For instance, Nie et al. (2014) reported a three-stage fermentation strategy that changes the nutrient components of the medium, which was designed for the efficient production of oil rich in arachidonic acid (ARA) using Mortierella alpina.

The composition and quantity of microbial lipids is species-dependent and can be affected by external environmental factors, such as light intensity, temperature, salinity, dissolved oxygen, etc. (Vitova et al., 2015). Furthermore, in autotrophic microalgae, light capture for photosynthesis is crucial for growth and accumulation of lipid reserves. Thus, adequate light intensity favors the overproduction of microalgal lipids. For example, high light conditions trigger the formation of neutral lipids in the microalgae Scenedesmus abundans and Nannochloropsis sp., as well as Botryococcus braunii and a further Botryococcus species (Kojima and Zhang, 1999; Pal et al., 2011; Yeesang and Cheirsilp, 2011; Mandotra et al., 2016).

The incubation temperature has been reported as an important abiotic factor affecting growth and microbial oil accumulation (Laoteng et al., 2011). In Nannochloropsis oculata, heat stress rapidly stimulates neutral lipid formation, while raising the temperature up to 25°C causes an elevation of total lipids (Converti et al., 2009). On the other hand, the incubation temperature of oleaginous microorganisms influences their fatty acid composition by changing the desaturase enzyme activity (Amaretti et al., 2010; Xu et al., 2012). James et al. (2013) found that a switch to temperatures lower than 25°C decreased the content of saturated fatty acids in the microalga Chlamydomonas reinhardtii. Likewise, the concentration of ARA in M. alpina changed from 7.3 to 9.2 g/L, when the temperature changed from 25 to 20°C (Peng et al., 2010).

In some species of microalgae, an increase of intracellular lipid content was observed as a response to osmotic pressure due to salinity stress (Takagi et al., 2006; Yang H. et al., 2014). For example, Duan et al. (2012) reported that Chlorella vulgaris experienced a 21.1% increase of lipid yield when exposed to salt pressure. Biosynthesis of lipids is also induced by other physical environmental stresses, such as low oxygen in the microalga Aurantiochytrium sp., and dehydration in the green alga Chlorella kessleri (Jakobsen et al., 2008; Shiratake et al., 2013).

Reactive oxygen species, such as superoxide (O₂^-), hydroxyl radical (OH^•), and hydrogen peroxide (H₂O₂) are formed by the partial reduction of oxygen, which is an inevitable aspect of life under aerobic conditions (Montibus et al., 2015). Cellular ROS metabolism is tightly regulated by a battery of biological redox mechanisms, including both antioxidant enzymes and non-enzymatic antioxidant molecules. In a normal physiological state, the level of cellular ROS is in a stable dynamic equilibrium (Zhang J. et al., 2016). However, under specific adverse environmental stimuli, the balance between cellular ROS production and elimination can be disturbed, leading to a locally increased concentration of ROS – i.e., “oxidative stress” (Lushchak, 2011).

Similar to other forms of environmental stress, oxidative stress results in damage of nucleic acids, proteins, and lipids. However, indeed in many eukaryotes, there are well-described mechanisms in which ROS play a key role in the response and adaptation to environmental changes (Bhattacharjee, 2012; Hideg et al., 2013). According to recent reports, ROS may act as decisive signaling molecules of the cellular responses to environmental stresses in oleaginous microorganisms. The relationship between ROS levels and a number of different kinds of stress has been reported, including adverse light conditions, nitrogen depletion, unfavorable salinity, as well as high or low temperature (Shih et al., 2014; Chokshi et al., 2015; Garcia-Rios et al., 2016). In C. vulgaris, the specific intracellular ROS level can serve as a general quantitative marker for stress, irrespective of the type of stress induced (Menon et al., 2013).

Eukaryotic microorganisms (e.g., yeasts, molds, and microalgae) possess highly conserved molecular mechanisms that enable them to survive in adverse environments. The machinery encompassing stress sensors, signal transduction and response elements is common to these mechanisms (Lushchak, 2011). In Figure 1, we summarized what is known about stress sensing and the putative concomitant ROS generation.

Nicotinamide adenine dinucleotide phosphate oxidases are highly conserved plasma- and endo-membrane enzymes present in animals, plants (including microalgae), and fungi. Some researchers also summarized the functions of certain macromolecules as possible sensors for environmental changes in plants and mammals (Hernandez-Onate and Herrera-Estrella, 2015; Zhu, 2016). Specific signals (e.g., Ca²⁺) that are generated by these stress sensors can lead to the phosphorylation and activation of NADPH oxidases, and activated NADPH oxidases can in turn utilize cytosolic NADPH as the electron donor to reduce extracellular O₂ to O₂^-, causing the subsequent formation of H₂O₂ (Herve et al., 2006; Mignolet-Spruyt et al., 2016).

The electron transfer chains (ETC), which exist in mitochondria, endoplasmic reticulum, and peroxisomes, are regarded as the main sources of intracellular ROS (Starkov, 2008; Sandalio et al., 2013). Under certain stimuli, growing ROS concentrations can overwhelm the cellular antioxidant defense systems, causing ROS to be released into the matrix via osmosis (Manoli et al., 2007; Mignolet-Spruyt et al., 2016). Using TEM, Yu et al. (2016) observed mitochondria in M. alpina that seemed to be enlarged and less compact than normal with carbon limitation, which in turn may generate ROS. Furthermore, abiotic stresses such as UV and heat instantaneously lead to modifications of the membranes, such as membrane-lipid peroxidation, which are potential sources of ROS. Unlike NADPH oxidase-generated ROS, these membrane-lipid peroxidation products induce non-specific responses to various kinds of environmental stress (Bhattacharjee, 2012).

Compared with other types of ROS, such as O₂^- and OH^•, H₂O₂ is much more stable and can pass membranes through aquaporins, which is an advantage with regard to its signaling capacity (Reczek and Chandel, 2015). The transduction of H₂O₂-based signals is mainly centered on sulfur chemistry, with the main player being the reversible oxidative modification of cellular sulfur-containing groups (e.g., cysteine residues and thioredoxin), which in turn results in disturbances of cellular metabolism and signaling pathways in eukaryotic microorganisms (Veal et al., 2007; Scott and Eaton, 2008).

As shown in Figure 1, the key part of the adaptive response to H₂O₂ is the transcriptional reprogramming of gene expression. The AP-1 family transcription factors are highly conserved among eukaryotes. They are involved in genetic responses to H₂O₂ signals associated with the Skn7-type response regulators (Carmel-Harel et al., 2001; Hagiwara et al., 2016). H₂O₂ can also oxidize critical cysteine thiol groups of phosphatases in mitogen-activated protein kinase (MAPK) pathways, which appears to be a conserved signaling mechanism for diverse environmental stimuli (Ikner and Shiozaki, 2005; Lushchak, 2011). For instance, the unicellular yeast Schizosaccharomyces pombe utilizes a two-component signaling system coupled to a MAPK pathway to perceive and respond to local ROS accumulation (Nguyen et al., 2000; Lara-Rojas et al., 2011). Finally, a number of recent studies have shown that post-translational changes induced by ROS also play an important role in the rapid response to oxidative stress, in addition to the altered global gene expression patterns (Ralser et al., 2009; Morano et al., 2012).

Oleaginous microorganisms mainly accumulate neutral lipids, which account for nearly 90% of their lipid storage (Juneja et al., 2013; Donot et al., 2014). In addition to their traditional and most-studied function in carbon and energy storage, these lipids, and especially PUFAs, may act as antioxidants or otherwise protective defense molecules in the stress response (Hu et al., 2008). For example, the increase of fatty acid unsaturation under low temperature stress has been generally linked to an adaptation mechanism which the lipid-producing fungi use to maintain membrane fluidity (Certik and Shimizu, 1999; Ji et al., 2014). The overproduction of lipids presents an indispensable buffer against stress conditions.

As shown in Table 1, improving experimental evidence seems to point in the direction that intracellular ROS may in fact be mediators of lipid accumulation in oleaginous microorganisms (Yilancioglu et al., 2014; Chokshi et al., 2015). A number of studies showed that stress-induced lipid accumulation always goes along with increasing antioxidant defenses (e.g., oxidative-stress response proteins) or increasing intracellular ROS levels (Yang et al., 2013; Yilancioglu et al., 2014). As stress markers, the specific intracellular ROS levels have been found to be linked to the specific intracellular neutral lipid levels in an inverse and direct power law fashion in C. vulgaris (Menon et al., 2013). In fact, the proper exposure to exogenous ROS, such as H₂O₂, but also certain nanomaterials, can trigger neutral lipid formation (Yu et al., 2015).

However, only a few possible mechanisms of ROS-mediated lipid accumulation have been illuminated, and direct experimental evidence is still absent. Some hypotheses surrounding the mechanisms of ROS-dependent lipid accumulation are shown in Figure 1. As lipids are highly reduced molecular species, neutral lipid overproduction requires large quantities of NADPH, whose primary sources are the oxidative pentose phosphate pathway and malic enzyme (Wasylenko et al., 2015). In what appears to be a common mechanism, within seconds of oxidative stress the carbon metabolic flux changes from glycolysis to the oxidative pentose phosphate pathway by post-translational modification of glycolytic enzymes (Ralser et al., 2009). On the other hand, the stored neutral lipids in oleaginous microorganisms are deposited within lipid droplets, and there is evidence that endoplasmic reticulum stress is an activator of lipid droplets formation (Jacquier et al., 2011). Since ROS are a well-known trigger of endoplasmic reticulum stress, this is another possible mechanism by which ROS stress enhances the formation of lipid droplets (Yu et al., 2015). Moreover, previous studies have demonstrated that some transcription factors, such as AP-1, which are associated with the oxidative-stress response, also participate in a regulatory network that induces secondary metabolism (Montibus et al., 2015). Considering the crosstalk between lipid biosynthesis and secondary metabolism, this appears to be another potential link between intracellular ROS signaling and lipid metabolism (Reyes et al., 2014; Zhang Z. et al., 2016).

Furthermore, the overall level of intracellular ROS cannot be ignored. The generation of low levels of ROS by environmental stresses initiates adaptive responses, including lipid biosynthesis and accumulation. By contrast, high levels of ROS may cause cellular damage, leading to the consumption of stored lipids as energy source in order to maintain cellular homeostasis. It has also been reported that the external addition of antioxidants, such as sesamol, ginsenosides, and ascorbic acid, can scavenge intracellular ROS and enhance PUFA content in oleaginous microorganisms (Liu et al., 2015; Ren et al., 2017).