Introduction

Polyunsaturated fatty acids (PUFAs), innate structural components of cell membranes, play an important role in membrane physical properties1. Arachidonic acid (ARA), the most abundant C20 PUFA in humans, can act as precursors of endogenous cannabinoid or eicosanoids involved in a variety of cellular processes, such as inflammation, mood, cognitive functions improvement and central nervous system development2. Accordingly, ARA has been widely used in pharmaceuticals, nutrition, cosmetics, breeding feed and other fields3. Lipids of microbial sources have been suggested as the main alternative source of ARA4. Submerged fermentation (SmF) of Mortierella alpina (M. alpina) is the main way to produce ARA5.

In recent years, solid-state fermentation (SSF) has emerged as a promising alternative to SmF due to its advantages, including reduced wastewater production, lower energy consumption, and higher productivity. SSF is a biotechnological process in which microorganism are cultivated in environments with little or no free water, which mimics the natural habitat of fungi, particularly filamentous fungi, thereby enhancing product yield. Enzyme production is one of the most successful applications of microbial SSF, with the production of various enzymes, including cellulase, amylase, protease, and xylanase, successfully achieved through SSF of filamentous fungi6. Furthermore, the high productivity of SSF has facilitated its application in biofuel production, such as the synthesis of lipase by Rhizopus oligosporus7 for biodiesel catalysis, and the ability of fungi like Mucor indicus and Rhizopus oryzae to produce ethanol, with a yield of up to 74% of the theoretical maximum8. Additionally, SSF is frequently employed in the production of bioactive compounds9, such as phenolic compounds, and in the enhancement of plant-derived proteins10. Thus, SSF has established credibility in the production of a broad range of products, including biodiesel, feed bioproducts11, enzyme, bioactive compounds and polyunsaturated fatty acids (PUFAs).

Zygomycetes possess the ability to synthesize lipids that contain a wide range of polyunsaturated fatty acids, which are commonly used in PUFA12 production and have been successfully applied in solid-state fermentations (SSFs). Mucor circinelloides, a strain known for producing gamma-linolenic acid (GLA), generated 42.43 mg of lipids per gram of dry substrate, with a high proportion of unsaturated fatty acids, when utilizing mulberry branches as a solid substrate13. Similarly, Mortierella isabellina can produce 47.9 mg lipid per gram of soybean hull, representing a 3.3-fold increase in lipid content compared to the initial lipid content of the soybean hulls14. Currently, the SSF of Zygomycetes fungus M. alpina primarily focuses on the utilization of natural substrates such as agro-industrial and food by-products including oat bran15, oilseed cakes16, grains17, animal fat by-products18, and mushrooms19, yielding lipids with 30–50% ARA and 15–40 mg of ARA per gram of substrate. Although natural substrate is a cost-effective solid matrix, the ARA content and lipid yield remain insufficient to meet the quality and yield for ARA-related products. Furthermore, separating mycelium from the natural biomass matrix is challenging, which hinders the analysis of the mycelium’s characteristics, lipid and ARA production capacity, and physiological status20. This complicates the understanding of the mechanisms involved in the SSF of M. alpina. Agar-supported solid-state fermentation (AgSF) is commonly used for the isolation, preservation, and screening of strains. Few studies have explored it as a variant of traditional SSF or developed it as a model for studying the physiology of filamentous fungi under SSF conditions21. However, the potential of AgSF as a fermentation strategy was rarely evaluated. Additionally, the development of AgSF as a model for physiological studies in SSF, as well as its application as a fermentation method in M. alpina, has not been investigated to date.

To gain a comprehensive understanding of the production potential, lipid synthesis capacity, mycelium characteristics and physiological status of AgSF in M. alpina, a systematic evaluation of AgSF was conducted using M. alpina M0223. First, differences in mycelial characteristics and lipid profiles between AgSF-M0223 and SmF-M0223 were analyzed. Next, the physiological status of mycelium was assessed through comparative transcriptome analysis. Finally, based on these findings, the fermentation process of AgSF-M0223 was optimized. This study aims to provide a new theoretical foundation for the fermentation of M. alpina, introduce a novel fermentation strategy for producing ARA-rich lipids, and offer new insights into lipid synthesis by oleaginous microorganisms.

Results

Mycelial characteristic and lipid profile of AgSF- and SmF-cultured M. alpina M0223

As shown in Fig. 1a, although the biomass of AgSF-M0223 was slightly lower than that of SmF-M0223, its lipid content was significantly higher, with a 74.26% increase over SmF-M0223 (Fig. 1a). Moreover, ARA comprised 72.53% of total fatty acids (TFAs) in AgSF-M0223, representing a 72.21% increase compared to SmF-M0223, and the ARA content per unit of mycelium was 183.66% higher than in SmF-M0223 (Fig. 1b, c). Additionally, levels of ARA synthesis precursors, including palmitic acid (PA), stearic acid (SA), oleic acid (OA), and linoleic acid (LA), were lower in AgSF-M0223, implying an enhanced conversion efficiency to ARA in AgSF-M0223 (Fig. 1b, c). Overall, the yields of ARA and lipid containing 72.53% ARA in AgSF-M0223 were 4.38 g/L and 7.28 g/L, respectively, representing increases of 141.63% and 48.25% over SmF-M0223 (Fig. 1a).

Fig. 1
figure 1

The comparison of SmF-M0223 and AgSF-M0223. (a) Distincts of biomass, lipid and ARA yield, as well as fatty acid composition (b) and content (c) in dried cell weight (DCW) between SmF-M0223 and AgSF-M0223 (n = 3). (d) Biomass, total fatty acid content in DCW and ARA yield differences, as well as fatty acid composition (e) and content (f) in dried cell weight (DCW) of AH and SH in AgSF-M0223 (n = 3). Asterisk at the top of the error bars indicated significant differences between the two groups: * represents p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, and “ns” denotes no significant difference. (g) Macro-morphological (Macro), optical morphological (Optical) and SEM-based surface morphological (Surface) distinctions between VH produced by SmF-M0223, and AH, SH and PH produced by AgSF-M0223. PA, palmitic acid. SA, stearic acid. OA, oleic acid. LA, linoleic acid. GLA, γ-linolenic acid. DGLA, dihomo-γ-linolenic acid. ARA, arachidonic acid.

Full size image

Morphologically, AgSF-M0223 formed a typical SSF-mycelium network of AH, SH, and penetrative hyphae (PH) embedded in the medium (Fig. 1g). No significant morphological differences were observed between the collected AH and SH. However, the SH was dense, pancake-shaped, and exhibited a higher density than the cotton-like AH, resulting in reduced light transmission. Consistent with this, the biomass of SH was substantially higher than that of AH, while both displayed high ARA content (Fig. 1d-f). Additionally, the morphology of the VH in SmF-M0223 also resembled that of SH, and exhibited a lower density (Fig. 1g). These findings suggested that the high-density SH played a critical role in promoting ARA-rich lipid production by AgSF-M0223.

The lipid and ARA synthesis capacities of AH, SH, and VH were further compared by the changes in lipids, ARA and the precursor fatty acids of ARA over time (Fig. 2a). SH is harvestable after 96 h of cultivation, and it displayed higher levels of lipids, ARA, and major fatty acid precursors compared to VH (Fig. 2b-i). Lipid accumulation in SH was fast from 96 to 120 h (Fig. 2b), accompanied by a rapid increase in ARA (Fig. 2c) and precursor fatty acids (Fig. 2d-i). Following this phase, lipid and ARA accumulation rates slowed, while precursor fatty acid content slowly reduced. VH exhibited a similar accumulation trend, with a prolonged rapid accumulation phase but lower lipid and ARA synthesis rates. AH, which could be harvested after 144 h of cultivation, showed comparable initial lipid and ARA content and consistent temporal change patterns to basal SH, albeit at a slower rate. These results indicated that SH and AH contained higher initial lipid and ARA levels, with SH exhibiting a greater capacity for lipid and ARA accumulation compared to VH. These characteristics contributed to the high ARA-rich lipids of AgSF-M0223.

Fig. 2
figure 2

Time courses of changes in fatty acids content of AH, SH and VH. As shown in (a), the synthesis of ARA undergoes a series of elongation and desaturation process from palmitic acid (C16:0, PA), leading to the sequential production of stearic acid (C18:0, SA), oleic acid (C18:1, OA), linoleic acid (C18:2, LA), γ-linolenic acid (C18:3, GLA), dihomo-γ-linolenic acid (C20:3, DGLA) and arachidonic acid (C20:4, ARA). (b) The changes of lipid content and major fatty acids content (c-i) in dried cell weight (n = 3).

Full size image

Transcriptome analysis among AH, SH, and VH in the period of rapid lipid synthesis

RNA-sequencing of AH, SH, and VH

To clarify why AgSF-M0223 produced high levels of ARA-rich lipid, RNA sequencing and analysis of AH, SH, and VH during the rapid lipid synthesis phase was conducted. A total of 68.59 Gb of clean bases was obtained. Over 97.7% of the reads aligned to the M. alpina M0223 genome (Table S1). With PCA and Pearson correlation analysis, samples of AH, SH, and VH can be clearly distinguished and have good reproducibility. Moreover, several unique genes expressed in AgSF, especially in AH, suggested a distinct transcriptional profile in AgSF (Fig. S1).

Gene set enrichment analysis between AH, SH, and VH

To further investigate transcriptional differences among AH, SH, and VH, gene set enrichment analysis (GSEA) was performed on all genes (Table S2) between AH vs. VH, SH vs. VH, and SH vs. AH, using genes with Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) annotations as a priori gene sets, respectively (Fig. 3). According to the GSEA results between AH vs. VH (Fig. 3a), the significant down-regulation of energy, carbohydrate, amino acid and cofactor metabolism in KEGG, as well as the biosynthesis and metabolism of nucleotides, cofactors, carbohydrate derivatives and proteins in GO, revealed attenuated primary metabolism in AH. Additionally, multiple GO terms related to ribosomal function (e.g., “ribosome” and “ribonucleoprotein complex”) and KEGG pathways for “ribosome” and “aminoacyl-tRNA biosynthesis” were also significantly downregulated in AH, further indicating a weaker protein synthesis in AH. Besides, AH probably underwent less DNA damage based on the downregulation of multiple pathways related to “replication and repair”, such as “base excision repair”, “DNA replication”, and “nucleotide excision repair”, etc. Coupled with the downregulation of “glutathione metabolism”, a key antioxidant22, AH might experience lower oxidative stress than VH, potentially mitigating the oxidation of unsaturated fatty acids.

Fig. 3
figure 3

Gene set enrichment analysis of AH, SH and VH. Significant pathways and GO terms from the GSEA of genes between AH vs. VH (a), SH vs. VH (b) and SH vs. AH (c). p < 0.05 and padj < 0.25 were considered significantly enriched. The color gradient from tawny to green represents high to low significance, and the bubble size represents the number of genes contributing to that pathway or GO term.

Full size image

In SH, significantly downregulated pathways related to “replication and repair” and “glutathione metabolism” showed a similar lower oxidative stress to AH (Fig. 3b). However, GSEA using both GO and KEGG gene sets showed upregulation of ribosomal functions in SH. Specifically, terms such as “ribosome” and “non-membrane-bound organelle” in GO, along with “ribosome biogenesis”, “ribosome”, and “RNA polymerase” in KEGG, were enriched, suggesting enhanced protein synthesis in SH. These physiological statuses were consistent with that of the SSF in the increased enzyme productivity23. Additionally, glycerophospholipid and glycerolipid metabolism, which are typically involved in lipid assembly and fatty acids desaturation, were upregulated in SH, prompting an increased capacity for lipid and ARA accumulation in SH. Compared to SH (Fig. 3c), AH exhibited reduced transcriptional and translational activity, including downregulation of ribosomal functions, RNA polymerase activity, regulation, etc. Metabolic pathways such as “glycolysis”, “pyruvate metabolism”, and “oxidative phosphorylation” were also downregulated, suggesting lower mycelial viability than SH. This further reflected a response in AH to nutrient limitations.

Transcriptional analysis of the lipid metabolism among AH, SH, and VH

Differentially expressed genes (DEGs) associated with lipid metabolism were analyzed. Genes with p < 0.05 among AH, SH, and VH were considered DEGs (Table S3). Glycolysis, the starting point for lipid synthesis, was significantly weaker in the AH than in SH and VH, further illustrating AH-restricted nutrition and its limitations on lipid synthesis. Further, according to the GSEA of “glycolysis”, SH and VH showed similar glycolytic capacity (Fig. 4a1, a2). However, the citrate cycle was weaker in SH than in VH, leading to more glucose being fluxed towards fatty acid synthesis in SH and more being fully oxidized to CO2 in VH (Fig. 4b). Furthermore, the genes encoding glucose-6-phosphate dehydrogenase and malic enzyme, critical for NADPH supply, were significantly upregulated24,25. This suggested an enhanced NADPH supply in SH (Fig. S2, S3). These physiological features aided in the synthesis of fatty acids. Besides, the transcript levels of desaturase and elongase in SH were lower than in VH but remained abundant (Fig. S4). A similar phenomenon was observed in the study of Yu et al.26. These suggested that the abundance of desaturase in M. alpina was enough to support the desaturation process. Enhanced triacylglycerol synthesis (Fig. 4d1, d2) and glycerophospholipid metabolism (Fig. 4e1, e2) in SH likely play a more vital role in accelerating fatty acid conversion to ARA-rich lipids. Additionally, the attenuated fatty acid degradation in SH reduced lipid consumption (Fig. 4c). These differences in physiological features collectively contributed to SH’s high lipid and ARA accumulation capacity.

Fig. 4
figure 4

An overview of transcriptional differences in lipid metabolism pathway among SH, AH and VH. In M. alpina M0223 cultured with glucose as carbon source, pyruvate produced from glucose via the EMP pathway undergoes the TCA cycle in the mitochondria to synthesize citrate. When the produced citrate overflows from the TCA cycle, the carbon flux shifts into fatty acid synthesis by fatty acid synthetase (FAS) in the cytoplasm. The synthesized fatty acids subsequently form triglycerides via the Kennedy pathway or enter glycerophospholipid metabolism and are desaturated or elongated in the form of phosphatidylcholine or acyl CoA via the fatty acid desaturase/elongase system. Besides, the synthesized fatty acids can be degraded by mitochondrial and peroxisomal fatty acid β-oxidation (f). Heatmap of genes with significant differences among AH, SH and VH in EMP (a1), TCA (b), FAO (c), triacylglycerol biosynthesis (d1) and glycerophospholipid metabolism (e1). GSEA results of glycolysis (a2), triacylglycerol biosynthesis (d2) and glycerophospholipid metabolism (e2) between SH and VH.

Full size image

Based on the transcriptomic analysis of lipid metabolism pathways (Fig. 4f), the citrate cycle in SH was weakened, while the synthesis of triglycerides and glycerophospholipid metabolism were enhanced. These changes, coupled with glycolytic capacity comparable to that of VH, directed more glucose toward lipid synthesis. Additionally, fatty acid β-oxidation (FAO) was reduced in SH. Together, these factors contributed to the enhanced lipid synthesis capacity observed in SH. In the study of Krishnan, J. et al.27, hypoxia was shown to inhibit the citrate cycle and enhance triglyceride synthesis, similar to the effects observed in SH, implying a possible hypoxic microenvironment in SH. To investigate this, we measured the oxygen content in SH and VH during the rapid lipid synthesis phase. The oxygen content in the microenvironment of the SH produced by AgSF-M0223 was 0.048 ± 0.045 mg/L, significantly lower than that in the liquid environment of SmF-M0223, which was 4 ± 0.47 mg/L (Fig. S5). This suggests that the SH of AgSF-M0223 is in a hypoxic microenvironment during the rapid lipid synthesis phase. Furthermore, a liquid superficial-static culture (LSSC) method, a static liquid culture without shaking28, was employed for M. alpina M0223. A fungal mat gradually formed at LSSC-M0223 (Fig. S6), which exhibited biomass comparable to that of AgSF-M0223. Although the lipid and ARA yields were lower than those from AgSF-M0223, they were still significantly higher than those from SmF-M0223, increasing by 48.63% and 108.42%, respectively (Fig. S6). These results further underscore the critical role of SH in lipid accumulation under specific physiological conditions.

Enhancing lipid accumulation and growth in AH via improved nutrient availability

According to transcriptomic and lipid profile analyses, lipid synthesis in the AH could be limited by the scarcity of glucose, the starting point for lipid synthesis. A medium spraying strategy was used to provide direct nutrients to the AH. Following this spraying, the expression of genes encoding key enzymes in glucose transport (glut-1), glycolysis (hxk-2, hxk-5, and pfk), and the citrate cycle (acnA, icd, and sucA) was upregulated, (Fig. 5a, b). After 72 h of spraying, the increases in total fatty acids and ARA in AH exceeded those observed in the control group. These indicated enhanced glucose utilization promoted the ability to synthesize fatty acids and ARA in AH (Fig. 5e, f). Similarly, the expression of glut-3, glut-4 and glycolytic enzymes in SH was also upregulated, albeit to a lesser extent than in AH (Fig. 5c). But unlike in AH, the expression of key enzymes in the citrate cycle remained unchanged (Fig. 5d). After spraying, SH also showed significant increases in total fatty acids and ARA levels (Fig. 5e, f), suggesting that the optimization of nutrients would further promote fatty acid and ARA synthesis in SH.

Fig. 5
figure 5

The effect of spraying medium on AH and SH. The expression level of genes encoding key enzymes in EMP (a) and TCA (b) of AH and SH (c, d) after spraying for 3 h (n = 6). (e) Differences in increments of TFA and ARA (f) in AH and SH after 72 h of spraying medium (n = 3).

Full size image

Effect of various nutrient component concentrations on AgSF-M0223

Nutrient levels significantly influence carbon flow toward lipid accumulation or growth within the mycelium. As shown in Fig. 6a1, biomass and lipid accumulation in AgSF-M0223 advanced gradually with increasing glucose concentration, but excessive glucose concentration inhibits mycelial growth (Fig. 6a2). A broader glucose concentration of 40–80 g/L is capable of producing lipids with high ARA content, and 80 g/L is the highest concentration for growth and lipid accumulation.

Nitrogen is essential in the synthesis of cellular components such as amino acids and nucleotides, whereas nitrogen deprivation promotes lipid accumulation29. In AgSF-M0223, nitrogen concentration demonstrated a positive correlation with growth (Fig. 6b1), and excess nitrogen resulted in thicker SH and less AH (Fig. 6b2). Lipid content decreased rapidly at yeast extract concentrations above 9 g/L. Whereas, below 9 g/L, the content and yield of ARA demonstrated a positive correlation with nitrogen availability (Fig. 6b1). 9 g/L is suitable for producing ARA-rich lipid. Besides, similar to the high OA content of 100 g/ L and the high PA content of 20 g/ L, excessively high or low nitrogen concentrations resulted in distinct fatty acid compositions, implying that extreme nutrient concentrations could lead to unusual mycelial status (Fig. 6b1).

The effect of inorganic salt was evaluated by adjusting its concentration while keeping the ratio constant. Inorganic salt concentration mainly affected biomass rather than lipid content and fatty acid composition (Fig. 6c1). Mycelial growth was inhibited at concentrations higher than 5× (Fig. 6c2). Both too-high (> 4×) and low (< 3×) concentrations resulted in reduced biomass and lower yields of ARA and lipids (Fig. 6c1). Therefore, a proper concentration (3 ×) was determined.

Fig. 6
figure 6

The effect of nutrients concentration on AgSF-M0223. (a1) The distinct of biomass, lipid, ARA and fatty acid composition of AgSF-M0223 under different glucose concentration of 20, 40, 60, 80, 100, 120, 160, 200 g/L, (b1) yeast extract concentration of 3, 6, 9, 12, 15, 18, 21 g/L and (c1) inorganic concentration of 1×, 2×, 3×, 4×, 5×, 8×, 10×, 12 × (0.6 g/L of NaNO3: 0.6 g/L KH2PO4: 0.1 g/L MgSO4·7H2O ) (n = 3), as well as their macroscopic morphology (a2, b2, c2). Numbers from low to high represent the macroscopic morphology of AgSF-M0223 from low to high concentrations. Groups that do not share the same letter are significantly different from each other (p ≤ 0.05), while ‘ns’ indicates no significant difference overall, according to the Bonferroni method.

Full size image

After integrating the optimal concentrations of each nutrient component, the yields of ARA and lipid reached 5.61 g/L and 11.12 g/L, respectively, without changing the fatty acid composition, which was 25.22% and 43.48% higher than the control. Further, the yield of lipids and ARA increased to 12.64 g/L and 7.28 g/L after the coupled medium spraying strategy (48 h interval spraying implemented after 8 days of incubation), which was 13.67% and 29.77% higher than that of the unsprayed group, and the fatty acid composition was not affected (Table S4). These indicated that AgSF enhanced both ARA and lipid yields, making it suitable for ARA-rich lipid production.

Discussion

SSF is an eco-friendly technology for cultivating diverse saprophytic fungi to produce high-yield bioactive compounds, enzymes, and other valuable products30,31,32. In this study, we suggested that AgSF, a special SSF, was a promising eco-friendly strategy for ARA-rich lipid production in M. alpina. Further, new insights into the association between ARA-rich lipid accumulation and the physiological status of M. alpina in AgSF were clarified. This deepened the understanding of fungal physiology in SSF and advanced applications for ARA-rich lipid production.

Fungi are well adapted to grow on surfaces33. SSF-cultured filamentous fungi form a three-dimensional mycelial network containing SH and, in some filamentous fungi (e.g., Aspergillus), biofilms21,34. Unlike SmF, where nutrients and oxygen are uniformly distributed, there are inevitable substrate, enzyme, and oxygen concentration gradients in the substrate and fungal mat of SSF, contributing to the special microenvironmental and physiological status of the biofilm35,36 In Aspergillus fumigatus, oxygen content within the biofilm decreased rapidly with increasing density, and there was a steep decline in oxygen levels with increasing depth from the air interface, resulting in a hypoxic microenvironment37,38. Consistent with this, SH produced by AgSF during the rapid lipid synthesis phase is also in a hypoxic microenvironment. Hypoxia has important regulatory roles in lipid metabolism. Up-regulated hypoxia-inducible factor (HIF) 1α inhibits the citrate cycle and activates glycolytic genes and PPARγ, whose products, in turn, activate glycerolipid biosynthesis genes, thereby facilitating carbon flux toward lipid synthesis27,39. The attenuated citrate cycle and enhanced triacylglycerol synthesis in SH hinted at a possible role of the hypoxic microenvironment in M. alpina in efficiently directing carbon flux toward lipid synthesis.

Based on the fatty acid desaturase/elongase system, arachidonic acid (ARA) synthesis involves a series of elongation and desaturation processes. These processes are catalyzed by elongase, Δ9 fatty acid desaturase, Δ12 fatty acid desaturase, Δ6 fatty acid desaturase, Δ6 elongase, and Δ5 fatty acid desaturase, starting from palmitic acid (C16:0, PA) and leading to the sequential production of stearic acid (C18:0, SA), oleic acid (C18:1, OA), linoleic acid (C18:2, LA), γ-linolenic acid (C18:3, GLA), dihomo-γ-linolenic acid (C20:3, DGLA), and finally, ARA (C20:4). Fatty acid desaturases are key enzymes in the ARA synthesis pathway, but the transcription levels were down-regulated in SH compared to VH. Microorganisms can finely regulate unsaturated fatty acid (UFA) synthesis in response to environmental factors. Previous studies have shown that fatty acids and oxygen regulate the transcript levels and mRNA stability of the gene encoding Δ9 fatty acid desaturase (ole1) via the fatty acid response element and hypoxia response element in the ole1 promoter. However, the transcriptional activation of both elements is suppressed by unsaturated fatty acids40. Moreover, environmental factors such as temperature and osmotic pressure can influence desaturase gene expression. For instance, low temperatures up-regulate mRNA levels of desaturase in Saccharomyces cerevisiae41, while high salt concentrations decrease gene expression in Hortaea werneckii42. The regulatory mechanisms behind the decreased transcript levels of fatty acid desaturases in SH produced by AgSF-M0223 remain unclear. However, it is possible that the inherently high transcript and protein abundance of fatty acid desaturases in M. alpina means that their down-regulation has a minimal effect on ARA synthesis, as observed in this study and in a previous study by Yu et al.26. In addition to fatty acid desaturases, glycerophospholipid metabolism plays a crucial role in fatty acid desaturation. In eukaryotes with a fatty acid desaturase/elongase enzyme system, such as Mucor circinelloides and certain plants, oleoyl-CoA must bind to phosphatidylcholine before being used as a substrate for desaturase to synthesize C18:2 or C18:343. Hypoxia has been shown to enhance phospholipid metabolism, thereby increasing the content of unsaturated fatty acids in plants45,46. Moreover, C18 polyunsaturated fatty acid (PUFA) levels were increased after enhancing the channeling of PC-modified fatty acids into the DAG pool by heterologous expression of PDCT47. Enhanced glycerophospholipid metabolism under SH should also play a key role in ARA synthesis. These suggested a potential role for hypoxic environments of oleaginous microorganisms in the production of ARA-rich lipid synthesis.

Furthermore, consistent with the high protein synthesis capacity demonstrated by eukaryotic SSF, ribosomal function was significantly enhanced, and amino acid degradation (particularly, leucine) was weaker in SH. Ribosomes are complexes that synthesize proteins from amino acids and are tightly regulated by nutritional availability in the environment (e.g. leucine, arginine and glutamine) by the target of rapamycin (TOR) signalling48. During nutritional stress, translation of ribosomal proteins is strongly inhibited and degraded49. Evidently, ribosomal functions such as r-protein synthesis and nucleolus were significantly down-regulated in nutritionally restricted AH. It is implied that nutritional availability in the environment could be the reason why eukaryotic SSFs exhibit high protein synthesis capacity.

The special physiological status of AgSF-M0223 conferred a higher yield of ARA-rich lipid than SmF-M0223. Meanwhile, AgSF, a special SSF, did not involve the stirring required for SmF, as well as the discharge and treatment of wastewater6. Moreover, it could avoid the heat and mass transfer issues and substrate heterogeneity associated with traditional SSF, which often leads to unstable product quality and contamination20. Further, LSSC results showed that the formation of biofilm was crucial to the high lipid-accumulation capacity of AgSF. Therefore, some gelling agents have the potential to be used as alternatives to agar to further reduce costs, such as carboxymethyl cellulose, xanthan gum, and polyvinylpyrrolidone50,51,52. Also, even without the use of gelling agents, direct superficial-static culture was feasible. What’s more, an automated scale-up system for AgSF has been developed, integrating sterilization and cultivation processes53. Based on the superior performance of SH accumulating ARA-rich lipid, some biofilm reactors possibly could be available as well54. Advances in mechanical automation would lead to the further development of new gel fermentation equipment. Therefore, the advantages of AgSF-M0223’s high yield of ARA-rich lipid and eco-friendly could be magnified in the booming landscape of artificial intelligence and automation.

Conclusions

In this study, AgSF was developed for the high yield of ARA-rich lipid in M. alpina. Enhanced lipid synthesis in SH and higher starting lipid and ARA content in AH and SH compared to VH together contributed to the superiority of AgSF-M0223. The similar glycolysis, attenuated citrate cycle, and enhanced triacylglycerol synthesis and glycerophospholipid metabolism of SH over VH enhanced glucose flux towards lipid synthesis. Concurrently, down-regulation of fatty acid degradation reduced lipid depletion in SH. AH, on the other hand, exhibited weaker primary metabolism due to restricted nutrients, limiting lipid accumulation. Building on these insights, optimization of AgSF with a spraying strategy further enhanced the ARA-rich lipid production capacity of AgSF-M0223, achieving 12.64 g/L of lipids with 70.41% ARA. This study provided new insights into ARA-rich lipid production mechanisms on SSF and a promising eco-friendly fermentation strategy.

Methods

Microorganism and medium

M. alpina M0223, preserved in our laboratory, was kept on potato dextrose agar (PDA) to ensure strain conservation and obtain spores for fermentation. The fermentation medium contained 80 g/L glucose, 12 g/L yeast extract, 3 g/L NaNO₃, 3 g/L KH₂PO₃, and 0.5 g/L MgSO₄·7 H₂O, with 15 g/L agar for AgSF and no agar for SmF and LSSC. To analyze the effects of nutrient concentration, glucose, yeast extract, and inorganic salt levels were varied individually while keeping the concentrations of other components constant. The spraying medium composition was identical to that of AgSF.

Culture method

The culture method was performed as previously described55. Briefly, M. alpina M0223 was cultured on PDA plates for 8 days, and the spore suspension was then inoculated into the fermentation medium at 10% of the medium volume. The cultures were incubated in (80 ml/250 ml) triangular flasks at 28 °C, 180 rpm for 12 days for SmF. For AgSF, 20 mL of sterilized fermentation medium containing agar was spread evenly on a 90 mm diameter plate to a height of approximately 0.36 cm. The plate was then inoculated with a 10% (v/v) spore suspension and incubated at 28 °C for 12 days. Medium spraying was performed on the third day after the fungal mat had formed (on the eighth day of incubation) using a sterile spray bottle, which contained 1.5 mL of the spray medium. Each plate received 1.5 mL of the medium per spray. The spraying intervals varied in the optimization experiments and included 12, 24, 36, 48, 60, and 72 h between each spray. For LSSC, the spore suspension (10% inoculum) was added to fermentation medium without agar, and 20 mL of the mixture was spread evenly onto a 90 mm plate, followed by incubation at 28 °C for 12 days.

Determination of dry cell weight, total lipid, and fatty acid profiles

To determine the dry cell weight, the vegetative hyphae (VH) in SmF were harvested by filtration, while the mycelium in AgSF was collected by scraping from the surface of agar plates, comprising the aerial hyphae (AH) and surface hyphae (SH). Due to that SH forms a tightly wound mycelial mat while AH consists of a fluffy mycelium growing on top of SH, they can be easily separated. The SH, after the removal of AH, is shown in Fig. S7. The collected mycelium was dried at 60 °C to constant weight. The determination of total lipid was performed as described in reported studies using n-hexane as an extractive agent56. Fatty acid profiles were detected with a simple modification of the method of Chen et al.57 and Yin et al.58. In brief, methylation of fatty acids was performed by boron trifluoride (BF3) using pentadecanoic acid as a standard. Further, the obtained fatty acid methyl esters (FAMEs) were analyzed by a Gas Chromatography system (CLARUS 680, PerkinElmer) equipped with Flame Ionization Detector (FID), taking nitrogen as carrier gas. A capillary column (HP-FFAP, 30 m× 0.32 mm ID × 0.25 μm film) was used and the column temperature was increased from the initial 150 °C to 220 °C at a rate of 10 °C/min and maintained at 220 °C for 10 min. The temperatures of the inlet and detector ports were both set at 250 °C. The injection volume was 1 µL. After collecting the peak area data using appropriate software, the percentage of the target FAME (ARA) in total fatty acids was calculated by dividing the area of the ARA peak by the total area of all peaks, then multiplying this result by 100. The content of ARA in dry cell weight (DCW) was determined by dividing the ARA peak area by the pentadecanoic acid peak area, and then multiplying the result by the amount of pentadecanoic acid added59. Furthermore, for a better comparison between AgSF-M0223 and SmF-M0223, the yield of ARA, lipids, and biomass in AgSF-M0223 is expressed as the amount of ARA and lipids, as well as dry cell weight harvested from 1 L of agar-containing fermentation medium. The plate specifications used in this study were 90 mm in diameter (with an agar surface area of approximately 55 cm2, allowing for the conversion of volume to surface area (1 L of medium provides approximately 0.277 m² of agar plate). This enables the calculation of yield in g/m² surface area, facilitating comparison with other SSF systems. Additionally, the height of the culture medium (approximately 0.36 cm) was determined based on the 20 mL volume used, and the container size can be adjusted accordingly.

Morphology observation by optical microscopy and scanning electron microscope

The optical morphology of M. alpina M0223 was observed using a biological microscope (BX53, Olympus). Before scanning electron microscope analysis (SEM), samples were initially fixed overnight in 2.5% glutaraldehyde, and then underwent gradient dehydration with ethanol concentrations from 30 to 100%, with two washes at 100% ethanol. Subsequently, ethanol was replaced with isoamyl acetate, and samples were lyophilized overnight. The dried samples were sputter-coated with gold for 300 s. SEM images were collected with a Nova NanoSEM 450 (FEI, Holland).

Transcriptome analysis

SH, AH, and VH in the rapid lipid and ARA synthesis period were collected and snap-frozen by liquid nitrogen. Then, RNA was extracted by the RNA Isolation Kit (Beibei Biotechnology) and RNA integrity and quality were assessed by Nanodrop and agarose gel electrophoresis. RNA sequencing was performed by the DNBSEQ-T7 platform. Transcript expression levels were quantitatively analyzed using TPM (transcripts per million reads) as the quantitative index with RSEM60. Gene Ontology (GO)61,62 and the Kyoto Encyclopedia of Genes and Genomes (KEGG)63,64 databases were utilized for the functional annotation of genes and the enrichment analysis of GO and KEGG were performed by Gene Set Enrichment Analysis (GSEA)65. The principal component analysis (PCA), correlation analysis, Venn analysis, and GSEA were performed on the online platform of Major Cloud Platform (https://cloud.majorbio.com/). Displays of GSEA results were done through the R package (ggplot2)66.

RNA extract and quantitative polymerase chain reaction (RT-qPCR) analysis

Collected mycelium was snap-frozen in liquid nitrogen and then ground into powder with liquid nitrogen. RNA was extracted by RNA Isolation Kit (Beibei Biotechnology). After integrity and quality detection by agarose gel electrophoresis and Nanodrop, cDNA was obtained by HiScript III 1st Strand cDNA Synthesis Kit (+ gDNA wiper, Vazyme). Quantitative Real-time PCR (qRT-PCR) of key enzymes in glycolysis and citrate cycle was performed by StepOne™ System with SYBR Green Fast qPCR Mix (ABclonal). The relative expression levels of the genes were determined by 2−ΔΔCt, taking 18 S rRNA as internal control. The primers of qRT-PCR were listed in Table S5.

Detection of oxygen content

After AgSF-M0223 and SmF-M0223 were cultured to the rapid lipid synthesis phase, the oxygen content was measured. The oxygen content in the microenvironment of SH produced by AgSF-M0223 was determined using a microelectrode system connected to an oxygen microsensor (Unisense). In contrast, the oxygen content in the liquid environment of SmF-M0223 was measured using a portable dissolved oxygen meter (HACH).

Statistics analysis

Each experiment used three biological replicates. The analysis between two groups of data was achieved by one-way analysis of variance (ANOVO) in IBM SPSS. The analysis among multiple groups of data was obtained by Bonferroni. P < 0.05 was considered to significant difference.