Tracing the mass flow from glucose and phenylalanine to pinoresinol and its glycosides in Phomopsis sp. XP-8 using stable isotope assisted TOF-MS

Phomopsis sp. XP-8, an endophytic fungus from the bark of Tu-Chung (Eucommia ulmoides Oliv) showed capability to biosynthesize pinoresinol (Pin) and pinoresinol diglucoside (PDG) from glucose (glu) and phenylalanine (Phe). To verify the mass flow in the biosynthesis pathway, [13C6]-labeled glu and [13C6]-labeled Phe were separately fed to the strain as sole substrates and [13C6]-labeled products were detected by ultra-high-performance liquid chromatography-quadrupole time of flight mass spectrometry. As results, [13C6]-labeled Phe was incorporated into [13C6]-cinnamylic acid (Ca) and p-coumaric acid (p-Co), and [13C12]-labeled Pin, which revealed that the Pin benzene ring came from Phe via the phenylpropane pathway. [13C6]-Labeled Ca and p-Co, [13C12]-labeled Pin, [13C18]-labeled pinoresinol monoglucoside (PMG), and [13C18]-labeled PDG products were found when [13C6]-labeled glu was used, demonstrating that the benzene ring and glucoside of PDG originated from glu. It was also determined that PMG was not the direct precursor of PDG in the biosynthetic pathway. The study identified the occurrence of phenylalanine- lignan biosynthesis pathway in fungi at the level of mass flow.

Precursor feeding and enzymatic activity measurements indicate that this strain synthesizes PDG via many steps, such as during mass flow of the phenylpropanoid pathway 14 . Genomic annotation indicates that the phenylpropane pathway exists in this strain 15 and some other microorganisms 16 . However, the functions of the denoted genes have not been verified until now. Therefore, it is necessary to verify the entire PDG biosynthetic pathway in Phomopsis sp. XP-8.
Using stable or radioactive isotope-labeled compounds is an efficient and reliable strategy to verify the mass flow of unknown biosynthetic pathways by tracing the isotope-labeled compounds from substrates to products 17 . 13 C-labeled substrates have been used to shed light on the biodegradation pathways of organic pollutants 18 . Isotope labeling combined with high-resolution mass spectrometry have also been used to track the abiotic transformation of pollutants in aqueous mixtures 19 . In recent years, liquid chromatography-mass spectrometry (LC-MS) and ultra-high-performance liquid chromatography (UPLC) systems have been developed to facilitate the analysis of many substances at the same time with high sensitivity and selectivity 20 . Stable isotope-labeled compounds have also been employed in several areas of biomedical research 21 . The combination of stable isotope-labeling techniques with MS has allowed rapid acquisition and interpretation of data and has been used in many fields, including distribution, metabolism, food, and excretion studies [22][23][24] . The biochemical pathway of the aromatic compounds in tea has been also been revealed using the stable isotope labeling method 25 .
In this study, we applied stable isotopes labeling and MS to trace the PDG biosynthetic pathway. Stable isotope-labeled 13 C 6 glu and 13 C 6 Phe were used as the substrates and ultra-high-performance liquid chromatography-quadrupole time of flight mass spectrometry (UPLC-Q-TOF-MS/MS) was used to identify the products.

Materials and Methods
Microorganism and chemicals. Phomopsis sp. XP-8 previously isolated from the bark of Tu-Chung and stored at the China Center for Type Culture Collection (Wuhan, China) (code: Phomopsis sp. CCTCC M 209291) was used in this study.
Phe (purity ≥98%, Sigma, St. Louis, MO, USA), Ca and p-Co (purity ≥98%; Aladdin, Shanghai, China), PDG, PMG, and Pin (purity ≥99%; National Institutes for Food and Drug Control, Beijing, China) were used as the standards (dissolved in methanol) for the structural analysis and product identification. [ 13 C 6 ]-Labeled Phe and glu were purchased from the Qingdao TrachinoidCo (≥99%; Qingdao, China). The purity of the [ 13 C 6 ]-labeled Phe and glu was 99%. Methanol (HPLC grade) was purchased from Fisher Scientific (Fairlawn, NJ, USA). The water used in the experiment was purified using a Milli-Q water purification system (18.5 M) (Millipore Corp., Bedford, MA, USA). Other reagents and chemicals were of analytical grade. shown as solid arrow: synthesis of pinoresinol (Pin) occurs via oxidative coupling of coniferyl alcohol; coniferyl alcohol can be synthesized from glu through the phenylpropanoid pathway 10 ; PMG and PDG are converted from Pin by UDP-glucose-dependent glucosyltransferase 8 . while the biosynthesis of PDG from Pin has not been detected in plants, so they are connected with dashed arrow. The substrates which have been found in the fermentation medium of Phomopsis sp. XP-8 in our preliminary experiment were labeled as red.  a ratio of 10 g cells (wet weight) per 100 mL medium. To track the mass flow from glu to PDG, glu was changed to 5 g/L [ 13 C 6 ]-labeled glu in the above medium and the same conditions were used for bioconversion.
Bioconversion with Phe as the sole substrate was carried out in medium without glu, 7 mM [ 13 C 6 ]-labeled phenylalanine, and the prepared Phomopsis sp. XP-8 cells at a ratio of 10 g wet cells per 100 mL medium.
All bioconversions were carried out for 48 h at 28 °C and 180 rpm. At the end of bioconversion, the broth was collected and filtered through an intermediate speed qualitative filter paper before the products were detected.
Identification of the accumulated products during bioconversion. The products were extracted from the vacuum-evaporated (0.09 MPa, 50 °C) bioconversion broth with methanol and adjusted to 4 mL for the UPLC measurements after filtration through a membrane (0.45 µm, 13 mm diameter; Millipore, Billerica, MA, USA). The UPLC analysis was performed on a Waters Acquity UPLC system (Waters Corp., Milford, MA, USA), equipped with a binary pump, a thermostatically controlled column compartment, and a UV detector. Gradient elution was performed on an Acquity UPLCTM BEH C18 column (50 mm × 2.1 mm I.D., 1.7 m; Waters) and the column temperature was maintained at 30 °C, while sample temperature was 10 °C 13 .
The MS analysis of the products was carried out on a Q-TOF Premier TM with an ESI source (Waters Corp.) at the optimized parameters of: capillary voltage, 2.8 kV; sampling cone voltage, 20 V; extractor voltage, 4 V; source temperature, 100 °C; desolvation temperature, 250 °C, and flow rate of the desolvation gas (N 2 ), 400 L/h. The collision cell parameters for the Q-TOF-MS/MS analysis were: collision gas (Argon) flow rate, 0.45 L/h; collision energy, 15-35 eV. The mass spectra were recorded using full scan mode over a mass range of m/z 100-800 in negative ion mode. The MS acquisition rate was set to 1.0 s, with a 0.02 s interscan delay. The Q-TOF-MS/MS experiments were carried out by setting the quadrupole to allow ions of interest to pass prior to fragmentation in the collision cell.
Accurate mass measurements were obtained by means of a lock mass that introduces a low flow rate (3 L/min) of a chrysophanol (253.0499) calibrating solution in the ESI-Q-TOF-MS and ESI-Q-TOF-MS/MS. All operations and acquisition and data analyses were controlled by Masslynx V4.1 software (Waters Corp.). Data processing. Peak detection, alignment, and identification of the detected compounds were performed using Masslynx V4.1 software (Waters Corp.). The MS/MS fragmentation patterns were used for informative non-targeted metabolic profiling of the LC-MS data, and the acquired LC-MS/MS spectrum was identified after comparison with spectra proposed by the Mass bank database (www.massbank.jp), the KEGG database, and related reports.

Results
Detection of products converted from unlabeled glu. Production of PDG, PMG, Pin, Phe, p-Co, and Ca were detected in bioconversion systems using glu as the sole substrate. Data in Figs. 2-5 show the mass spectra of these compounds accumulated in the bioconversion systems and the corresponding standards. www.nature.com/scientificreports www.nature.com/scientificreports/ Production of Phe was detected as m/z = 164.08, and m/z = 147.06 ( Fig. 2A-4), which was consistent with the data obtained from the corresponding standards ( Fig. 2A-2). Similarly, production of PDG, PMG, Pin, p-Co, and Ca was also detected in the bioconversion system, indicating that glu was converted to these products by Phomopsis sp. XP-8, as only glu was provided in the bioconversion system. www.nature.com/scientificreports www.nature.com/scientificreports/ Identification of products converted from [ 13 c 6 ]-labeled Phe. The phenylpropanoid pathway in plants starts with Phe and ends with p-Co. The same mass flow was previously detected during PDG biofrom glu by Phomopsis sp. XP-8 13 . To verify this finding and the role of the Phe pathway in the biosynthesis of PDG, PMG, and Pin, [ 13 C 6 ]-labeled Phe was used as the sole substrate in the bioconversion system without glu (mainly used as the glucoside donor). As results, 13 C labeled Pin, Phe, p-Co, and Ca were successfully detected (Fig. 6). The products were successfully detected at the same RT of their corresponding unlabeled standard substrates. All 13 C-labeled product data and their corresponding standard substrates are summarized in Table S1 (Supporting information). www.nature.com/scientificreports www.nature.com/scientificreports/ As shown in Fig. 6B, 13 C-labeled Ca was detected as m/z = 153.07, indicating that six 13 C from [ 13 C 6 ]-labeled Phe were incorporated into Ca. A daughter ion of 13 C-labeled Ca was obtained at m/z = 109.08, indicating six 13 C referring to the standard Ca (m/z = 103.06). The structure of 13 C-labeled Ca without -COO − was observed at m/z = 109.08. Therefore, it was deduced that the six 13 C were incorporated into the benzene ring of Ca not into -COO − . P-Co produced in the conversion system was detected as m/z = 169.05 and revealed six 13 C by consulting the p-Co standard (Fig. 6C). A daughter ion of 13 C-labeled p-Co was obtained at m/z = 125.07, indicating 6 Da mass www.nature.com/scientificreports www.nature.com/scientificreports/ shift than p-Co standard (m/z = 119.06). The structure of 13 C-labeled p-Co without -COO − (44 Da lost) was observed at m/z = 125.07. Therefore, it was deduced that the six 13 C might be distributed in the benzene ring. 13 C-labeled Pin was detected (Fig. 6D-1) and compared with the mass spectra of the Pin standard (C 20 H 22 O 6 , RT = 9.736 min, detected as m/z = 357.13 and m/z = 151.04 respectively) (Table S2, Supporting information). 13 C-labeled Pin was detected as m/z = 369.05, indicating 12 Da mass shift than Pin standard (m/z = 357.13). A daughter ion of 13 C-labeled Pin was observed at m/z = 157.06, which showed a mass increase of 6 Da than Pin standard (m/z = 151.04). The structure of 13 C-labeled Pin with loss of a benzene ring was identified as the major daughter ion of m/z = 157.06 (Fig. 6D-1). This result confirmed that the six 13 C were distributed in a benzene ring,   13 C stable isotopes were signed by coloured dots. Among the dots, red dots mean that the 13 C stable isotopes were converted from 13 C isotopes labeled glucose through Phosphoenolpyruvate (PEP); blue dots mean that the 13 C isotopes were converted from 13 C isotopes labeled glucose through Enthrose 4-phosphate; green dots means that the 13 C stable isotopes were converted from another 13  The isotopic patterns observed in the MS and MS/MS spectra suggest the 13 C from 13 C 6 -labeled glu were incorporated into the products of Phe (Fig. 7A), Ca (Fig. 7B) p-Co (Fig. 7C), PDG (Fig. 7D), PMG (Fig. 8A), Pin (Fig. 8B), respectively. The observed mass shifts, indicating the number of incorporated 13 C, were shown in the spectra. The detailed information on the products and possible positions of 13 C in the products are summarized in Table S2 (Supporting information).
Interestingly, the analysis revealed that 13 C 6 -labeled glu were incorporated into the core structure of PDG and PMG, and their glycosides. Additionally, the maximum of 16 13  Possible pathways for biosynthesis of PDG and PMG. The evidences for the possible biosynthetic pathways of PDG, PMG, and Pin are summarized in Figs. 9 and 10. The pathway from Phe to Pin, glu to Phe, Pin, PMG and PDG was verified (Fig. 9A,B, Supporting information Table S1 and Table S1). In addition, the bioconversion between PDG and PMG in Phomopsis sp. XP-8 was reported for the first time, and the analysis was showed in Fig. 10.
As shown in Fig. 10, two structures of PMG were detected: one was [ 13 C 12 ]-PMG with two benzene rings converted from 13 C-labeled glu and an unlabeled glycoside (PMG m/z 531.29), and the other was [ 13 C 18 ]-PMG with both benzene ring structures converted and a glucoside from 13 C-labeled glu (PMG m/z 537.33). Similarly, two PDG structures were detected: one was [ 13 C 18 ]-PDG with a two benzene ring structure and one glycoside converted from 13 C-labeled glu (PDG m/z 699.27); the other one was [ 13 C 24 ]-PDG with two benzene rings and two glycosides from 13 C-labeled glu (PDG m/z 705.26). If PMG was the direct precursor of PDG, PMG m/z 531.29 would be converted to PDG m/z 699.27 by bonding one [ 13 C 6 ]-labeled glycoside through glycosylation; PMG m/z 537.33 could also be converted to PDG m/z 699.27 by bonding one unlabeled glycoside through glycosylation and to PDG m/z 705.26 by bonding one [ 13 C 6 ]-labeled glycoside. If this is true, PDG m/z 699.27 would have two glucoside sources, whereas PDG m/z 705.26 would have only one glucoside source. Therefore, the concentration of PDG m/z 705.26 should be lower than PDG m/z 699.27. However, the data show that the relative abundance of PDG m/z 705.26 was much higher than that of PDG m/z 699.27 (Fig. 7D-3). Therefore, PMG was not the precursor of PDG.
In contrast, if PDG was the direct precursor of PMG, PDG m/z 699.27 would be converted to PMG m/z 531.29 by hydrolyzation of one [ 13 C 6 ]-labeled glycoside and to PMG m/z 537.33 by hydrolyzation of one unlabeled glycoside; PDG m/z 705.26 would be converted to PMG m/z 537.33 by hydrolyzation of one [ 13 C 6 ]-labeled glycoside. If this is true, PMG m/z 537.33 would have two glycoside sources, whereas PMG m/z 531.29 would have only one source. The concentration of PMG m/z 531.29 should be lower than PMG m/z 537.33. However, the data show that relative abundance of m/z = 531.29 was higher than that of m/z = 537.33 (Fig. 8A-3). Therefore, PDG was not the precursor of PMG.

Discussion
The 13 C stable isotope labeling method was successfully used in this study to verify the phenylpropanoid-pinoresinol and biosynthetic pathway of its glycosides in Phomopsis sp. XP-8 during mass flow. This lignan biosynthetic pathway was only reported in plants until now 8,9 , so it was very significant to verify the occurrence of this pathway in microorganisms. Stable Isotope-assisted metabolomics is an efficient way to trace and identify bio-transformed products and the metabolic pathways involved in their formation, such as understanding the fate of organic pollutants in environmental samples 17 . It was the first time to use this method to verify the Phenylpropanoid-pinoresinol in a microorganism. In our previous studies, many methods such as precursor feeding 13 , detection of enzyme activity 14 , and genomic annotation 15 have been used to analyze the Phenylpropanoid-pinoresinol biosynthetic pathway in Phomopsis sp. XP-8. Through these studies, the precursors, enzymes activity and genes of PDG biosynthetic pathway have been found. The 13 C stable isotope labeling method gave further verification on the occurrence of lignan biosynthetic pathway in microorganisms by now. In addition to this, it is the first time that differences between the PDG and PMG biosynthetic pathways have been verified.
The results obtained in this study verify the existence of the phenylpropanoid-lignan metabolic pathway in Phomopsis sp. XP-8. Genomic annotation is an efficient way to discover the pathways that are normally difficult to reveal by metabolic and enzymatic evidence due to low intermediate accumulation, low end-product, and silent gene expression under normal conditions. This method has been successfully used to identify the existence of a phenylpropanoid metabolic pathway in Aspergillus oryzae 26 , and the molecular genetics of naringenin biosynthesis, a typical plant secondary metabolite in Streptomyces clavuligerus 27 , and the occurrence of the phenylpropanoid-lignan pathway in Phomopsis sp. XP-8 15 . This study reports the existence of the phenylpropanoid-lignan pathway in Phomopsis sp. XP-8 during mass flow and identified the metabolites.
Additional studies should illustrate the origin of the genes in the phenylpropanoid-lignan pathway of Phomopsis sp. XP-8. Horizontal gene transfer (HGT) has long been recognized as an important force in the evolution of organisms 28 . HGT occurs among different bacteria and plays important roles in the adaptation of microorganisms to different hosts or environmental conditions 29 . More and more evidence for gene transfer between distantly related eukaryotic groups has been presented 28 .Therefore, we cannot exclude the possibility that XP-8 may have acquired the genes related to the lignan biosynthetic pathway from its host plant by HGT during long-term symbiosis and evolution. However, further evidence is still needed to verify this proposed process.
The results obtained in this study provide useful information on the biosynthesis of lignans and their glycosides via microbial fermentation. Biosynthesis of lignans is of great interest to organic chemists as it provides a model for biomimetic chemistry and has extensive applications 30 . Improvement has been made in the techniques to biosynthesize lignan products by regulating the lignan biosynthetic pathway in trees through genetic modifications 31 . However, the lignan biosynthetic pathway has rarely been reported. More importantly, the bioconversion sequence from Pin to PDG and the direct precursor of PDG have remained unclear until now. In previous studies on Phomopsis sp. XP-8, the highest production of PDG and PMG did not occur simultaneously 12 and PMG was not the precursor of PDG because PDG production decreased and/or disappeared when PMG yield increased 13 . The present study demonstrated that PMG was not the precursor of PDG, and PDG was not the precursor of PMG, indicating that Pin might be converted to PMG and PDG via two different pathways in Phomopsis sp. XP-8, which has not been revealed in plants.
Furthermore, this study revealed that the bioconversion of Pin, PMG, and PDG from glu occurred simultaneously as that from Phe. We found that the benzene ring structure of Phe did not open throughout the entire Pin bioconversion process in Phomopsis sp. XP-8 when Phe was used as the sole substrate, indicating that the Pin benzene ring originated from Phe. Glu was converted to Phe and was the sole glycoside donor for PDG biosynthesis. Therefore, glu not only participated in the formation of glycosides in PDG, but also provided the PDG benzene ring structure. This is different from that found in plants, indicating there might be some other different pathways to produce these products in Phomopsis sp. XP-8.
Not all intermediates in the KEGG-identified plant-lignan biosynthetic pathway related to Pin, PMG, and PDG formation were found in Phomopsis sp. XP-8, such as caffeic acid, ferulic acid, and coniferyl alcohol (Fig. 1). This may be because the pathways after p-Co are different in XP-8 from those in plants, or the accumulation of these intermediates was too low to be detected. Further studies are needed to verify this hypothesis.
In conclusion, the capability of Phomopsis sp. XP-8 to biosynthesize Pin, PMG and PDG from [ 13 C 6 ]-Phe and [ 13 C 6 ]-glu was verified. The study illustrated the phenylpropanoid-pinoresinol biosynthetic pathway in microorganism by using stable isotope assisted UPLC-Q-TOF-MS/MS, thus, demonstrating a completely new way to