B. trispora ATCC 14271(+) and 14272(-) were obtained from the American Type Culture Collection (Manassas, VA, USA) and cultivated as previously described (Wang et al., 2016a). Cells were divided into two groups, the control group and TPA-treated group, with five biological duplicates. For the control group, B. trispora ATCC 14271(+) and 14272(-) were fermented for 4 days, with three repeats for each sample. For the TPA-treated group, after the cultures had fermented for 2 days, TPA (1.8 g/L) was administered for 2 days (Wang et al., 2016a). The BL21 strain (Escherichia coli DE3) was cultivated as our previously method (Wang et al., 2016b).

B. trispora cell cultures of control and TPA-treated groups were harvested after fermenting for 4 days and re-suspended twice using ion-free ultrapure water. Then, the cultures were dissolved in 200 μl of ion-free ultrapure water per gram of wet weight. The cultures were then crushed and homogenized by liquid nitrogen grinding and tissue grinder, respectively. Then, methanol/acetonitrile (1:1, v/v) was added, and the cells were broken up in two rounds by using a Scientz-IID sonifier (Ningbo Scientz Biotechnology Co., China) for 30 min each round. The cells were then incubated at −20°C for 1 h to precipitate protein. The samples were centrifuged at 13,000 rpm for 15 min, and the supernatants were used as the metabolite samples and stored at −80°C after freeze-drying.

The metabolite samples were separated by UHPLC-Q-TOF-MS/MS using an Agilent 1290 Infinity Series UHPLC (Agilent Technologies, Santa Clara, CA, United States). UHPLC was performed at 25°C, with a linear gradient as follows: 0–1 min, 5% phase A (ddH₂O, 25 mM ammonium acetate and 25 mM ammonia), 95% phase B (acetonitrile); 1–14 min, 35% phase A, 65% phase B; 14–18 min, 60% phase A, 40% phase B; and 18–23 min, 5% phase A, 95% phase B, with a flow rate of 0.3 ml/min. Time-of-flight mass spectrometry was performed using a hybrid triple quadrupole (Triple TOF 5600 system, AB SCIEX, Framingham, United States) with electrospray ionization operated in both positive and negative modes. The TOF MS scanning m/z range was 60–1,000 Da, and the product ion scanning m/z range was 25–1,000 Da. The MS spectrum was obtained using information-dependent acquisition with high sensitivity.

The raw MS data were converted to mzXML files using ProteoWizard MSConvert and processed using XCMS software (Waters) for feature detection, retention time correction, and alignment (Benton et al., 2015). The metabolites were identified by accurate masses (<25 ppm) and matched with the standards database built by Shanghai Applied Protein Technology (Shanghai, China). After pareto-scaling, multidimensional statistical analyses [including principal component analysis (PCA), partial least squares discriminant analysis (PLS-DA), and orthogonal partial least squares discriminant analysis (OPLS-DA)] were implemented in SIMCA-P 14.1 (Umetrics, Umea, Sweden). Univariate statistical analyses [including Student’s t-test and fold change (FC) analysis] were implemented in R programming language. For multidimensional statistical analyses, variable importance for the projection (VIP) was used to measure the degree of influence of each metabolite. The values VIP > 1 and P < 0.1 were set to determine differential metabolites, and VIP > 1, P < 0.05, and FC > 1.5 were set to determine significantly differential metabolites.

Hierarchical clustering and correlation analysis of differential metabolites were performed to accurately screen for ionic metabolites and measure the correlation of significantly differential metabolites. Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment was performed using the KEGG database¹.

B. trispora ATCC 14271(+) and 14272(-) were fermented for 2 days, and TPA (1.8 g/L) was added to the cultures. After the samples were shaken at 28°C for 0, 1, 4, 8, 16, 32, and 64 h, the cells were harvested and lysed by using Scientz-IID sonifier. Then, the pH of the cells and fermentation was measured with a pH meter. To detect the vesicles structure in cells, hyphae were stained with 10 μM Mito-tracker Green (MG) (maximum excitation wavelength 490 nm; maximum emission wavelength 516 nm) (Beyotime Institute of Biotechnology, Shanghai, China) (Swift, 2003) and kept in the dark for 40 min at 33°C. Mycelial morphology was then detected with an upright fluorescence microscope (Ni-U; Nikon Co., Tokyo, Japan) after treatment with TPA for 2 days. Bacteriostatic ability were analyzed using E. coli; 10⁷ cfu/ml E. coli were added before TPA was added, and after TPA (1.8 g/L) was administered for 2 days, the cell density of E. coli was measured by a blood-cell-counting plate after shaking at 180 rpm, 28°C for 0, 1, 4, 8, 16, 32, and 64 h.

All experiments concerning data comparisons were performed three times. Statistical significance was determined using IBM SPSS Statistics 11.0 (SPSS Inc., Chicago, IL, United States), with the threshold set at P < 0.05.

After UHPLC-Q-TOF/MS analysis, 5,744 and 6,030 ionic peaks were extracted in positive and negative modes, respectively, and the total ion chromatograms of the control and TPA groups were obtained (Supplementary Figure 1). PCA (R2X > 0.5) and PLS-DA (Q2 > 0.5) analyses indicated that the models were stable and reliable. The OPLS-DA model of the control and TPA groups was established based on the modified PLS-DA. The score charts of PCA and OPLS-DA models (Supplementary Figure 2) showed that the TPA group evidently diverged from the control group, indicating that there were significant differences between and small variations within the two groups.

After VIP calculation, 119 and 226 differential metabolites were identified (VIP > 1 and P < 0.1, Supplementary Table 1) in positive and negative modes, respectively. Of these, 22 and 23 metabolites, respectively, were described (Supplementary Table 2). The hierarchical clustering of differential metabolites in positive and negative modes showed that samples of the same groups were clustered together (Figure 1). Furthermore, metabolites clustered together had similar expression patterns, suggesting that these metabolites may participate in closer reaction steps in the metabolic process.

The correlation of differential metabolites in positive and negative modes was evaluated (Figure 2). The positive mode indicated that PPE was significantly positively correlated with unsaturated fatty acids [including 4-hydroxybenzoate (4-HBA), 2E-eicosenoic acid, linoleic acid, and 16-hydroxypalmitic acid] and negatively correlated with 2-(4-hydroxyphenyl)ethanol, methionine, homogentisic acid, uracil, and thymidine. The negative mode indicated that cyclic dipeptides (His-Pro, His-Ile, and His-Phe) had significant positive correlations with thioetheramide-PC (TEA-PC), PC(16:0/16:0) (lecithin), 3,3-dimethylacrylic acid, and deoxyadenosine and negative correlations with glutathione disulfide (GSSG), betaine, and PPE.

To elucidate the mechanism of TPA-enhanced lycopene production in B. trispora, the significantly differential metabolites were screened with VIP > 1, P < 0.05, and FC > 1.5 as the standards, and 15 metabolites were identified (Table 1). Of these, 10 were down-regulated, and 5 were up-regulated. The down-regulated metabolites were mainly cyclic dipeptides (His-Pro, His-Ile, His-Phe, and Ile-Asn), organic acid [3,3-dimethylacrylic acid, phenyllactic acid, and isovalerylglycine (IVG)], lipids (TEA-PC and lecithin), and deoxyadenosine. The up-regulated metabolites were unsaturated fatty acids [16-hydroxypalmitic acid, linoleic acid, 2E-eicosenoic acid, and 11(Z),14(Z)-eicosadienoic acid (ω-3 PUFA)] and 4-HBA.

QS is a complex regulatory network with global control over diverse target functions, including nutrient acquisition, biofilm formation, and motility (Liu et al., 2010; Hallan et al., 2020). Cyclic dipeptides are QS autoinducers and can freely penetrate the cell membrane and activate the QS system (which includes regulation of cell membranes, cell growth, secondary metabolite production, and anti-oxidation) between organisms of the same species (Holden et al., 1999). In specific environments, the higher the number of microorganisms, the higher the concentration of cyclic dipeptide secreted (Wang et al., 2010). However, the stability of cyclic dipeptides is greatly reduced under high pH conditions (Gao et al., 2006). In this study, the extracellular and intracellular pH were both high after treatment with TPA (Figure 3), indicating that the down-regulation of cyclic dipeptides was caused by the increase in pH. Additionally, the up-regulation of His further increased intracellular pH. Thus, we speculated that TPA could enhance lycopene accumulation not only by the inhibition of the β-carotene cyclization, but also by the down-regulation of cyclic dipeptides, which induced the increase of intracellular and extracellular pH, and thereby starting QS.

The cell membrane is mainly composed of phospholipids and proteins. Approximately, 70–80% of the phospholipids are phosphatidylethanolamine, and an increase in its content will increase the volume of the cell membrane. In this study, TEA-PC and lecithin were down-regulated and unsaturated fatty acids were up-regulated after treatment with TPA. This effect may have been caused by the hydrolysis of lecithin by phospholipase. TEA-PC is an inhibitor of the competitive reversible secretory phospholipase A2, which can catalyze the hydrolysis of ester bonds at the sn-2 position of lecithin, producing free fatty acids and lysolecithin (Kita et al., 2019). Unsaturated fatty acids have antioxidant effects, and the solubility of lycopene in unsaturated fatty acids is high (Failla et al., 2014). Furthermore, linoleic acid can enhance the production of red pigments in Monascus ruber by accelerating the cAMP-PkA pathway (Huang et al., 2017). This indicates that TPA can increase the solubility of lycopene in cells and promote lycopene production through the induction of linoleic acid. However, in this study, lysolecithin did not show differential levels, and its derivative PPE had the greatest up-regulation in differential metabolites.

Furthermore, tertiary amines will first act on the lipid region of the cell membrane (Zhang et al., 2007; Hermetter et al., 2013). Thus, we speculated that the up-regulation of PPE was caused by the direct action of TPA on lecithin. Moreover, PPE can shuttle through the cell membrane, relax the local phospholipid bilayer, and accumulate β-carotene in the cell membrane, thereby increasing its production (Wu et al., 2017). These effects indicate that, in this study, up-regulated PPE could increase the amount of the cell membranes and penetrate the phospholipid bilayer to relax the cell membrane.

Among the differential metabolites, the down-regulation of choline and up-regulation of downstream betaine suggest that choline levels increase to produce betaine in glycine, serine, and threonine metabolism (Supplementary Figure 3). Betaine is a quaternary amine alkaloid, which can donate methyl with high activity, promote lipid metabolism, and thereby promote cell growth. Furthermore, choline is a component of cell membrane, and changes in its content are closely related to the morphology of filamentous fungi (Markham et al., 1993). When the concentration of choline decreases, the hyphae branch and expand, and many cytoplasmic vesicles accumulate locally (Markham et al., 1993). In this study, choline was down-regulated, the hyphae swelled, and many localized small vesicles were observed (Figure 4). We speculated that this morphological change in hyphae is beneficial for the accumulation of lycopene in the cell membrane and intracellular vesicles.

Interestingly, among the significantly differential metabolites, bacteriostatic phenyllactic acid, IVG, 3,3-dimethylacrylic acid, and deoxyadenosine were all down-regulated. Phenyllactic acid is a recently discovered type of natural bacteriostatic metabolite that can inhibit the growth of bacteria such as E. coli and Listeria monocytogenes (Feasley et al., 2006). It is also a substrate for the synthesis of betaine and 4-HBA, which were up-regulated in this study. Moreover, both betaine and 4-HBA affect fungal morphogenesis and promote mycelial growth (Chong et al., 2009; Zou et al., 2016). Our findings indicate that the metabolic flow to lipid metabolism and mycelial growth were increased, and the bacteriostatic metabolic flow was reduced.

The differences in E. coli densities between the TPA and control groups, shown in Figure 3A, may be because TPA itself has certain bacteriostatic abilities and it could transport into cell for binding cyclase (Wang et al., 2016a) and lecithin (Hermetter et al., 2013). The detected data on bacteriostatic ability showed that the antibacterial ability of TPA-treated B. trispora was indeed weakened. This may be a manifestation of B. trispora maintaining its own metabolic balance, which also reminds us to pay attention to the aseptic operation when using TPA to increase lycopene production.

Furthermore, most unsaturated fatty acids were up-regulated, and the differential metabolite GSSG was up-regulated 1.24-fold (Supplementary Table 2). GSSG is produced by the oxidation of GSH in peroxisomes, where the β-oxidation of fatty acids is also carried out, suggesting that peroxisomes play an important role in the action of TPA. Furthermore, peroxisomes can relieve the toxicity of amines through transamination (Nazarko et al., 2007), which suggests that the up-regulated PPE in this study may be produced by transamination between TPA and lysolecithin in peroxisomes.

The correlation analysis of differential metabolites showed that cyclic dipeptides have significant positive correlations with cell membrane-related TEA-PC and lecithin and bacteriostatic ability related to 3,3-Dimethylacrylic acid and deoxyadenosine; they also have a significant negative correlation with peroxisome-related GSSG, cell membrane-related PPE, and choline (Figure 2). Our data suggest that TPA can enhance lycopene accumulation not only by inhibiting the cyclization of β-carotene but also by down-regulating cyclic dipeptides to initiate QS. Other effects of TPA include the up-regulation of unsaturated fatty acids, PPE, and 4-HBA and down-regulation of choline and bacteriostatic metabolites for metabolic flux redistribution, locally swelling mycelia with vacuoles, and high lycopene capacity.