Microbial interaction induced changes of the metabolite profile
The co-culture of twenty microorganisms with A. sydowii on bran medium showed different degrees of induction between the cultures, among which B. subtilis could significantly induce A. sydowii to produce metabolites. After 12 days, the color of the hyphae of A. sydowii changed from dark green to light green in co-culture, and red-brown exudate was generated at the junction of B. subtilis and A. sydowii (Fig. 1). Moreover, a significant deadlock model was observed. This phenomenon indicates that during co-culture, A. sydowii and B. subtilis generated compounds due to the stress response at the confrontation zone, inhibiting the growth of the other. In order to study this phenomenon, we collected the confrontation zone and analyzed the metabolites.
The extracts from the bran medium of co-culture or pure culture were compared by LC-MS/MS, and 206 strong signal features whose intensity was over 10% of the highest intensity peaks were detected. The partial least squares discriminant analysis (PLS-DA) analysis of these peaks revealed the intrinsic variation in the data set. In the score plot, the samples from the co-culture are clearly separated from the two pure cultures, indicating the changes of metabolite profile (Fig. 2A). The heatmap generated by hierarchical clustering analysis (HCA) of these 206 features based on the MS data showed that co-culture caused global changes in the metabolomes (Fig. 2D). The heatmap also revealed that 25 features were identified only in co-culture, indicating that about 12.1% of the candidate features were newly biosynthesized during co-culture. In addition, 156 features were recorded in both pure culture and co-culture. Among them, 70 features in the co-culture system were significantly decreased, while 4 features in the co-culture system were up-regulated when compared with the pure culture of B. subtilis. On the contrary, there were only 8 features in the co-culture that were significantly decreased, and 37 features in the co-culture that were up-regulated when compared with the pure culture of A. sydowii (Fig. 3). In the loading plots of PLS-DA, the 25 newly biosynthesized features were mainly deviated from the center and clustered into the lower right zone of the plot (Fig. 2B). These features showed good linear correlation. Only the features that have a large contribution to the classification generated by co-culture are distributed on this line. The features that contribute more to the classification were closer to the lower right, and the features that contribute less to the classification were clustered on the upper left of the line and were closer to the origin (Additional file 1: Fig. S1). In the meantime, the variable importance in projection (VIP) score data indicated new biosynthesized features (N1-N4, N7, N13 and N20) with molecular weights of m/z 168.4234, 266.1459, 282.1436, 282.4537, 353.1765, 402.1640 and 480.3248 were ranked in top features detected by VIP score (Fig. 2C). These data indicated that the newly biosynthesized features in co-culture made important contribution to group classification.
Metabolomics study of newly biosynthesized metabolites in the co-culture
To understand the structure of the new biosynthesized metabolites, the 25 features were identified with the integrated approach. Take the identification of feature N2 as an example. The adduct ions of feature N2 were detected as m/z 265.1459 [M − H]− and m/z 325.2437 [M + CH3COOH − H]−, suggesting that the molecular weight was 266.1459. The PLS-DA analysis indicated that this feature was detected only in the co-culture and contributed greatly to the cluster. The structure of N2 was mainly identified by Level 2 process because there is no structural candidate that could be obtained in MS/MS databases supplied by MS-DIAL in Level 1 process. After comparing in silico spectra and the MS/MS data provided by MS-FINDER, N2 was identified as sydonic acid based on the fact that the mass peaks (m/z 265.1459, 253.4632, 249.1126, 180.0446, 137.0338 and 93.0326) matched well with the MS/MS database (Additional file 1: Fig. S2) with the lowest molecular error of 0.2327 and the highest structure score of 7.55. Similarly, among the other 20 features, 5 features were identified through Level 1 process and 15 features were identified through Level 2 process. The detailed information of the metabolites was summarized in the Table 2. There were 5 major classes of metabolites induced by co-culture, including sesquiterpenes, macrolides, esters, polyketides, and flavonoids. These metabolites of microorganism, are usually not generated in the normal condition of microorganisms, and can only be synthesized under certain stress. These compounds were reported to participate in the defense and communication between microbial cells, promote metabolism, and have a certain bacteriostatic effect [17].
In order to verify the validity of the identification approach, five compounds (N1, N2, N3, N4 and N13) which had higher VIP scores in PLS-DA analysis indicating lager contributions to the cluster, were isolated and purified through silica gel column chromatography, ODS column chromatography and preparative HPLC from the confrontation zone of the co-culture. According to the results, N1 (20 mg), N2 (60 mg), N3 (19 mg), N4 (13 mg) and N13 (21 mg) with purity over 95% were obtained. After analyzing the NMR data, the compounds were identified as Orsellinic acid (N1) [18], Sydonic acid (N2) [19], (7S)-(−)-10-Hydroxysydonic Acid (N3) [17], (R)-(-)-Hydroxy-sydonic acid (N4) [20], and Macrolactin A (N13) [21], which were consistent with those identified by the approach above, suggesting the credibility of the approach. The structures of the 7 compounds were shown in Fig. 4 and the detailed NMR data were provided in Additional file 1: compounds information.
Identification of novel metabolites in the co-culture
There were still 4 features (N6, N7, N9 and N20) that did not match any features in the public MS/MS spectrum library. To elucidate the structures of these possible novel metabolites generated through co-culture, their MS/MS data were analyzed with Level 3 process which was assisted by GNPS platform and manual dereplication. The GNPS approach can capture similar structures and analog features into the same cluster regardless of retention times in the LC-MS.
The GNPS data indicated that three induced features, including N6 (m/z 350.1610), N7 (m/z 352.1765) and N9 (m/z 368.1713) were clustered, suggesting that these features have very close structural relationships (Fig. 5). As none of the three features was identified within the LC-MS/MS database, compound N7, with the highest content in LC-MS data, was separated and purified. Compound N7 was obtained as white powder. The UV absorption was at 213 nm, 254 nm and 298 nm. The molecular formula C18H27O6N was indicated by the HRESIMS at m/z 354.1908 [M + H]+ (calculated as 354.1872), indicating 6 degrees of unsaturation. The 1H NMR and 13C NMR of N7, N2 and the reference data of N2 [19] were shown in Table 1. The 13C NMR and HMQC spectra indicated the presence of a total of 18 carbon signals attributable to one carboxyl carbon at δc 172.05 (C-10); one ketone carbon δc 166.14 (C-7) ; three methyl groups at δc 28.48 (C-8’), δc 22.57 (C-6’) and δc 22.37 (C-7’); four methylene groups at δc 61.35 (C-9), 41.60 (C-2’), 21.33(C-3’) and 38.78(C-4’); five methines including δc 22.37 (C-7’), three aromatic carbons at δc 126.64 (C-6), δc 117.47 (C-5) and δc 115.10 (C-3), one oxygenated carbon at δc 55.53 (C-8); four quaternary carbons including three aromatics at δc 154.74 (C-2), δc 135.86 (C-1) and δc 133.50 (C-4), one oxygenated atom at δc 75.04 (C-1’).
Table 1
1H and 13C NMR spectral data for compound N7 in DMSO-d6, N2 in CD3OH and the reference data of sydonic acid in CD3OH (500 MHz for 1H NMR and 125 MHz for 13C NMR)
No.
|
N7
|
N2
|
Sydonic acid
|
|
δC
|
δH (J in Hz)
|
δC
|
δH (J in Hz)
|
δC
|
δH (J in Hz)
|
1
|
135.9
|
|
137.9
|
|
138.0
|
|
2
|
154.7
|
|
157.0
|
|
156.9
|
|
3
|
115.1
|
7.24 (1H, d, 1.35)
|
118.7
|
7.37 (1H, d, 1.6)
|
118.6
|
7.36 (1H, d, 1.6)
|
4
|
133.5
|
|
131.6
|
|
131.6
|
|
5
|
117.5
|
7.28 (1H, dd, 1.35, 8.05)
|
121.5
|
7.44 (1H, dd, 1.6, 7.9)
|
121.5
|
7.43 (1H, dd, 1.6, 7.9)
|
6
|
126.6
|
7.35 (1H, d, 8.05)
|
127.7
|
7.25 (1H, d, 7.9)
|
127.7
|
7.25 (1H, d, 7.9)
|
7
|
166.1
|
|
169.9
|
|
169.9
|
|
8
|
55.5
|
4.42 (1H, m)
|
|
|
|
|
9
|
61.4
|
3.77 (2H, m)
|
|
|
|
|
10
|
172.1
|
|
|
|
|
|
1'
|
75.0
|
|
78.0
|
|
77.9
|
|
8’
|
28.5
|
1.51 (3H, s)
|
28.9
|
1.59 (3H, s)
|
29.0
|
1.59 (3H, s)
|
2’
|
41.6
|
1.94 (1H, m)
|
43.7
|
1.94 (1H, m)
|
43.6
|
1.94 (1H, ddd, 4.5, 12.2, 13.7)
|
|
|
1.66 (1H, m)
|
|
1.77 (1H, m)
|
|
1.77 (1H, ddd, 4.8, 11.6, 13.7)
|
3’
|
21.3
|
0.99 (1H, m)
|
22.9
|
1.29 (1H, m)
|
22.9
|
1.18 (1H, m)
|
|
|
1.27 (1H, m)
|
|
1.34 (1H, m)
|
|
1.33 (1H, m)
|
4’
|
38.8
|
1.05 (2H, m)
|
40.4
|
1.13 (2H, m)
|
40.4
|
1.13 (2H, m)
|
5’
|
27.3
|
1.43 (1H, m)
|
28.9
|
1.47 (1H, m)
|
28.8
|
1.47 (1H, m)
|
6’
|
22.6
|
0.77 (6H, d, 6.6)
|
22.8
|
0.81 (6H, d, 6.4)
|
22.9
|
0.81 (6H, d, 6.5)
|
7’
|
22.4
|
23.0
|
23.0
|
N-H
|
|
8.15 (1H, d, 7.6)
|
|
|
|
|
Table 2
List of induced features only in the co-culture of A. sydowii with B. subtilis analyzed by LC-HRMS
No
|
m/z(-)
|
RT (min)
|
Molecular formula
|
Name
|
Classification
|
Identification method
|
N1
|
168.42
|
12.01
|
C8H8O4
|
Orsellinic acid
|
Sesquiterpene
|
Level 2, 4
|
N2
|
266.14
|
30.79
|
C15H22O4
|
Sydonic acid
|
Sesquiterpene
|
Level 2, 4
|
N3
|
282.14
|
13.22
|
C15H22O5
|
(7S)-(−)-10-Hydroxysydonic Acid
|
Sesquiterpene
|
Level 2, 4
|
N4
|
282.45
|
14.01
|
C15H22O5
|
(R)-(-)-Hydroxy-Sydonic acid
|
Sesquiterpene
|
Level 2, 4
|
N5
|
337.37
|
25.88
|
C18H27NO5
|
Neostemodiol
|
Alkaloid
|
Level 2, 3
|
N6
|
351.16
|
12.73
|
C18H25NO6
|
Hydroxy serine Sydonate
|
Sesquiterpene
|
Level 3
|
N7
|
353.19
|
20.82
|
C18H27NO6
|
Serine sydonate
|
Sesquiterpene
|
Level 4
|
N8
|
367.4
|
22.65
|
C19H29NO6
|
Piperidinyl acetic acid ethyl ester
|
Ester
|
Level 1
|
N9
|
369.17
|
31.44
|
C18H27NO7
|
3’-Alkene serine Sydonat
|
Sesquiterpene
|
Level 3
|
N10
|
381.17
|
31.44
|
C19H27NO7
|
Bruceolline
|
Alkaloid
|
Level 2, 3
|
N11
|
395.19
|
22.26
|
C20H29NO7
|
Ruwenine
|
Macrolide
|
Level 1
|
N12
|
398.09
|
11.57
|
C21H18O8
|
Auramycinone
|
Antibiotic
|
Level 2
|
N13
|
402.16
|
19.82
|
C24H34O5
|
Macrolactin A
|
Macrolide
|
Level 2, 4
|
N14
|
424.11
|
20.42
|
C23H20O8
|
Dehydrovillosin
|
Polyketide
|
Level 1
|
N15
|
424.13
|
11.86
|
C20H24O10
|
Rutarin
|
Glucoside
|
Level 2
|
N16
|
426.09
|
14.44
|
C22H18O9
|
Maggiemycin
|
Polyketide
|
Level 2
|
N17
|
436.13
|
14.3
|
C21H24O10
|
Phlorizin
|
Antioxidant
|
Level 1
|
N18
|
444.19
|
20.07
|
C21H32O10
|
Penstemide
|
Ester
|
Level 2
|
N19
|
452.12
|
14.43
|
C21H24O11
|
Glucuronide
|
Ester
|
Level 1
|
N20
|
480.32
|
39.51
|
C31H44O4
|
Macrolactin U’
|
Macrolide
|
Level 4
|
N21
|
502.255
|
28.94
|
C28H38O8
|
Cavipetin
|
Ester
|
Level 2
|
N22
|
514.24
|
18.41
|
C31H34N2O5
|
Telmisartan
|
Aromatic
|
Level 2
|
N23
|
558.26
|
15.84
|
C27H42O12
|
Valeriotriate B
|
Ester
|
Level 2
|
N24
|
601.32
|
38.12
|
C34H43N5O5
|
Methylurea
|
Ester
|
Level 2
|
N25
|
706.36
|
20.78
|
C42H50N4O6
|
Tetrastachynine
|
Peptide
|
Level 2
|
The inspection of the 1H NMR spectrum indicated the presence of δH 7.35 (d, J = 8.05 Hz, H-6), 7.28 (dd, J = 8.05,1.35 Hz, H-5) and 7.24 (d, J = 1.35 Hz, H-3), which was a typical spectrum of 1,3,4-trisubstituted benzene ring. This sub-structure was confirmed by the relationship between H-5 (δH 7.28) to H-6 (δH 7.35) in 1H-1H COSY, and H-3 (δH 7.24) to C-1 (δc 135.86), H-5 (δH 7.28) to C-1, and H-6 (δH 7.35) to C-4 (δc 133.5) in HMBC. The correlations from H-7’ (δH 0.77) to H-2’ (δH 1.94,1.66) in 1H-1H COSY indicated the existence of a hexane substructure. The HMBC relationship between H-5 (δH 7.28) with C-7 (δc 166.14) and H-6 (δH 7.35) with C-1’ (δc 75.04) implied the attachment of C-4 to the C-1 position. The position of the amine bond was determined by the relationship of H-8 to C-7 in HMBC (Fig. 6; Additional file 1: Fig. S3-S9).
The absolute configuration of N7 was deduced based on the comparison of experimental data and calculated ECD curves by Gaussian 09. The conformers were optimized using DFT at the B3LYP/6-31G (d) level in methanol. The energies were calculated using the TDDFT methodology at the B3LYP/6-31G (d, p) level in MeOH with PCM model (Additional file 1: Fig. S10-S13; Table S1-S8). The calculated CD spectrum of N7 (1’R, 8S) agreed well with the experimental CD curve (Fig. 7), indicating that absolute configuration of N7 was 1’R, 8S, and was named as Serine sydonate.
The structures of compounds N6 and N9 were mainly determined by the LC-MS/MS data from the negative-ion mode and compared with N7. Comparison of the fragment ions of the compounds showed some common fragments of m/z 224.0562, m/z 194.0458 and m/z 150.0563, indicating that these features had the same backbone structure and belonged to a series of structural derivatives (Additional file 1: Fig. S14). The structures of compounds N6 and N9 were determined by comparing the negative-ion mode LC-MS/MS data with N7 because these three compounds shared similar MS/MS pattern and the content of N6 and N9 was low in broth. For the LC-MS/MS data of N7, the abundant fragments were m/z 334.1656 and 322.1660, which arose from molecular ion m/z 352.1765 by the loss of H2O (18 Da) and two methyl groups (30 Da), respectively. The fragments of m/z 304.1557 and 290.1758 were generated by further facile loss of H2O (18 Da) and CH3OH (32 Da) from the fragment of m/z 322.1654, respectively. The fragments of m/z 224.0562, m/z 194.0458 and m/z 150.0563 indicated the substructure of serine substituted hydroxy-benzoic acid. The molecular weight of N6 was 351.1610, which was common neutral loss of 2 Da of N7, indicating that the compound N6 was most likely the dehydrogen product of N7. The fragments of m/z 224.0562, m/z 194.0458 and m/z 150.0563 suggested the existence of the substructure of serine substituted hydroxy-benzoic acid in N6. The stable m/z 302.1394 fragment, which represented the conjugated olefin structure formed by the dehydroxylation of hydroxyl methylheptane, suggested that the double bond was located at the 4' position. Thus, compound N6 was determined as 4’-alkenyl serine sydonate. For compound N9, the fragments of m/z 224.0562, m/z 194.0458 and m/z 150.0563 were also the characteristics of the substructure of serine substituted hydroxy-benzoic acid. The residue mass was 18 Da higher than that of N6, suggesting that N6 was the dehydration product of N9. The stable m/z of 302.1394 in N6 and m/z of 320.1497 in N9 also indicated that dehydration occurred in the substructure of hydroxyl methylheptan. Thus, the structure of N9 was determined as hydroxyl serine sydonate. However, whether this hydroxyl group was located at 4’ or 5’ position could not be determined by LC-MS/MS data alone (Fig. 8). N6 and N9 (4’-hydroxyl or 5’-hydroxyl serine sydonate) were found to be novel compounds after database searching.
As compound N20 cannot be connected with the other metabolites in Level 3, it was forwarded to Level 4 for direct isolation and purification. This compound was isolated as pale-yellow creamy solid. The molecular formula C31H44O4 was indicated by the HRESIMS at m/z 503.3134 [M + Na]+ (calculated for 480.3240), indicating 10 degrees of unsaturation. The UV absorption was at 212 nm and 273 nm. The 1H NMR (DMSO-d6, 500 MHz) and 13C NMR (DMSO-d6, 125 MHz) data were provided in the Additional file 1: compounds information. The 1D NMR data were consistent with the data of the known compound Macrolactin U identified by Xue et al. [22], and its relative configuration was 4S, 5S. However, the methyl group H3-29 (δ 1.10, d) and H2-6 (δ 2.56, m) were defined as trans due to the NOESY relationship between H3-29 and H-5 (δ 4.25, m) in compound N20 (Additional file 1: Fig. S15-S22). Thus, the relative configuration of C-4 and C-5 were S and R, (Fig. 9), which was different from S and S in Macrolactin U, respectively. Therefore, compound N20 was identified as the isomer of Macrolactin U and named as Macrolactin U’, which is also a novel compound. Unfortunately, the absolute configuration of N20 was not determined yet as no obvious difference was identified between ECD and crystal of N20, which could not be obtained due to the limited amount of the compound.
Thus, a total of 25 features induced only in the co-culture were identified by the combination of the computational approach (MS-DIAL), the web-based tools (GNPS and MetaboAnalyst) with chemical isolation and purification (Table 2, Additional file 1: Fig S23). Four of the features were novel metabolites, including two compounds confirmed by NMR. Five known compounds were also purified to verify the validity of the approach.
Biological Activity Assay
The isolated compounds N1-N4, N7, N13 and N20 were evaluated for their anti-nematode activity and antidiabetic activity. Among the compounds, compound N3 showed a certain degree of anti-nematode activity with an IC50 of 50 µM. Furthermore, compounds N2-N4, N7 and N13 exhibited potent activity against SHP1 and PTP1b, both of which are targets for the development of diabetes (Table 3). In addition, compounds N7 and N13 displayed inhibition activities against protein tyrosine phosphatase (CD45) with IC50 values of 16.0 µM and 17.9 µM.
Table 3
|
IC50 (µM)
|
PTP1b
|
SHP1
|
CD45
|
N1
|
>20
|
>20
|
>20
|
N2
|
7.97±0.24
|
8.35±0.35
|
>20
|
N3
|
15.88±0.13
|
>20
|
>20
|
N4
|
>20
|
15.72±0.11
|
>20
|
N7
|
14.18±0.21
|
11.68±0.08
|
16.03±0.38
|
N13
|
>20
|
14.61±0.39
|
17.89±0.92
|