3.1. Responses of soil physical-chemical properties and soil enzyme to the invasion of Fom
In this study, modifications in physical-chemical properties and enzyme of the soil in which it was infected by Fom were described in Supplementary Table S6. The soil organic matter content (SOC), pH, electrical conductivity (EC), available nitrogen content (AN), available phosphorus content (AP) and available iron content (AFe) increased in S-FOM treatment. The significant increase in the value of SOC (57.7%), pH (2.0%), and AFe (27.7%) were obtained in S-FOM soil. The control group had the higher values of soil fluorescein diacetate (FDA) hydrolase and polyphenol oxidase (PPO). Although the inhibitory effects of different enzymes were different, the invasion of Fom significantly inhibited these enzymes. For example, the obvious decreases in the activities of FDA hydrolase (47.5%) and PPO (17.4%) were obtained in S-FOM soil. Soil FDA hydrolase and PPO activity were negatively correlated with the relative abundance of F. oxysporum, pH, organic matter content, and available Fe. This result indicated that the inhibition of F. oxysporum on enzyme activities of the soil might play critical roles in the conversion of organic matter and the change of pH.
3.2. Responses of community structure and diversity of soil to the invasion of Fom
The data set including 10 samples comprised a total of 22,492,589 contigs with 2,005,895 to 2,511,389 sequences per sample (mean 2,161,810.6 (Control) and 2,336,707.2 (S-FOM)) and 21,972,887 scaffolds with 1,931,781 to 2,469,478 sequences per sample (mean 2,106,715 (Control) and 2,287,862.4 (S-FOM)). A total of 6,923 microbes were assembled. The higher observed species and Chao1 index were detected for S-FOM, whereas the difference was not significant. The Fom infected samples showed a significantly higher Simpson index and Shannon index of bacteria than Control, however, the Fom infected samples showed lower Simpson index and Shannon index of fungi than Control (Fig. 1A). When the beta diversity was assessed, the Control and S-FOM was significantly different (R2 = 0.43,P = 0.013) (Supplementary Fig. S1A). Redundancy analysis (RDA) revealed that the environmental factor variable possesses significant correlation with community structure. The variable of pH (R2 = 0.65, P = 0.017) produced significant positive correlations with S-FOM microbes communities, the variable of FDA hydrolase (R2 = 0.84, P = 0.002) and PPO (R2 = 0.58, P = 0.043) produced significant negative correlations with S-FOM microbes communities (Supplementary Fig. S1B).
The soil was predominantly colonized by Proteobacteria, Actinobacteria, Acidobacteria, Gemmatimonadetes, Planctomycetes, Firmicutes, Nitrospirae, and Bacteroidetes, accounting for 63.0% of the total scaffolds, when the microbe was assessed at the phylum level (Fig. 1B). Ascomycota and Basidiomycota were the largest fungal taxonomic group, but accounted for a substantially lower proportion of microbe-assigned reads (Fig. 1D). The higher abundance of increased Proteobacteria, Firmicutes, and Nitrospirae were detected in the S-FOM samples that were subjected to invasion of F. oxysporum. Meanwhile, the obvious decreases in the relative abundance of Acidobacteria, Gemmatimonadetes, Planctomycetes, and Basidiomycota were obtained in S-FOM soil (Fig. 1B and D). The modifications in the composition of the core microbe at the genus level of the S-FOM samples were described in Fig. 1C, Fig. 1E, and Supplementary Fig. S1C. The invasion of Fom significantly influenced these microbes, although the influence differed in different microbes. For example, the obvious decreases in the relative abundance of Aspergillus, Pseudomonas, Sphingobium, Lysobacter, Variovorax, Sporisorium, Methylobacterium, Mycobacterium, Micromonospora, Actinoplanes, Mycolicibacterium, Amycolatopsis, Frankia, Pseudonocardia, and Ilumatobacter were obtained in S-FOM samples. Simultaneously, the relative abundance of Fusarium, Agromyces, Ramlibacter, Streptomyces, Conexibacter and Kribbella were significantly increased in S-FOM samples. Sphingomonas, Trichoderma, and Bacillus had no difference between the control sample and the S-FOM samples. At the species level, the relative abundance of Trichoderma asperellum, T. atroviride, Bacillus anthracis, B. kochii, Talaromyces stipitatus and so on showed an increased tendency in response to the invasion of F. oxysporum (Supplementary Fig. S2). Meanwhile, Fusarium sp. was composed of F. incarnatum, F. culmorum, F. delphinoides, F. dimerum, F. equiseti, F. fujikuroi, F. graminearum, F. langsethiae, F. oxysporum, F. pseudograminearum, F. solani, F. venenatum, F. verticillioides, and Unclassified Fusarium. The F. oxysporum, F. solani, and F. fujikuroi showed significantly higher proportions than others. A significant increase of the relative abundance of F. oxysporum (51.7%) was observed in S-FOM samples, while there was no difference in other species.
3.3. Co-occurrence networks analysis of microbial communities
To study the effect of the invasion of Fom on microbial interactions, the networks were constructed (Supplementary Fig. S3). The co-occurrence network was constructed from 10 samples (Control and S-FOM) based on the strong and significant (P < 0.05) correlations of microbial species. In total, 5035 nodes (5025 microbial nodes and 10 environmental nodes) and 17912 edges were shared among co-occurrence networks, and the co-occurrence networks contains 89.54%-92.13% microbial biomass (Supplementary Table S7). The degrees of the networks followed power-law distributions, the average path length was significantly larger than the corresponding random networks, which indicated that co-occurrence networks was a small world (Supplementary Fig. S2 and Table S7). Meanwhile, modularity values were higher than the random networks and greater than 0.34. This indicated that the co-occurrence networks had modular structures. For the networks, approximately 82.75% of the nodes belonged to bacteria, 7.98% of the nodes belonged to eukaryota, 7.36% of the nodes belonged to archaea, 1.91% of the nodes belonged to viruses. Proteobacteria (33.29%), Actinobacteria (18.23%), and Firmicutes (7.02%) accounted for more links in the network. There are 353 modules among the networks, of which 15 modules had more than 50 nodes, such as module 1141, module 676, module 1458, module 1465, module 168, module 1460, module 1229, module 1163, module 1124, module 1281, module 1250, module 1238, module 1225, module 1286 and module 1283, the nodes account for 37.87% of the total number of nodes and the relative abundance of microbe the nodes account for 70.39–75.23% of the total number of microbe, including Proteobacteria (35.26%), Actinobacteria (20.86%), Firmicutes (6.25%), Euryarchaeota (2.94%), Thaumarchaeota (2.73%), Ascomycota (2.47%), and Bacteroidetes (2.15%) (Supplementary Fig. S3 C and D). We investigated the distribution of soil microbial species by Zi-Pi plot (Supplementary Fig. S4). In the network, 94.22% of the nodes belong to the peripherals, 4.39% of the nodes belong to the connectors, 0.71% of the nodes belong to the module hubs, 0.08% of the nodes belong to the network hubs, among them, the connectors, module hubs and the network hubs were considered as key microorganisms. Among the 261 key microbial species, 159 microbial species belong to Proteobacteria, 27 microbial species belong to the Actinobacteria, and the relative abundance of 13 microbial species were increased in S-FOM samples (Table 1). The relative abundance of sp1571 (Kribbella flavida), sp1615 (Microlunatus soli), sp2002 (Streptomyces sp. 4F), sp4492 (Devosia sp.), sp5385 (Burkholderia sp.), sp5454 (Paraburkholderia sprentiae), sp5496 (Acidovorax sp.), sp5501 (Alicycliphilus denitrificans) and sp5675 (Chromobacterium vaccinii) were significantly positively correlated with the relative abundance of F. oxysporum. The relative abundance of sp5951 (Haliangium ochraceum), sp744 (Acidimicrobium ferrooxidans) and sp4354 (Bradyrhizobium erythrophlei) were negatively correlated with the relative abundance of F. oxysporum.
Table 1 The topological features of key microorganisms in co-occurrence networks and the correlation coefficient with F. oxysporum and Talaromyces purpureogenus
OTU ID
|
Phylum
|
Genus
|
Relative abundance
|
Degree
|
Betw centrality
|
Zi
|
Pi
|
Correlation coefficient
|
Control
|
S-FOM
|
Fom
|
Strain Q2
|
sp1227#
|
Actinobacteria
|
Arsenicicoccus
|
1578.3±67.55
|
1670.77±153.65
|
6
|
3156.53
|
-1.10
|
0.67
|
0.479
|
0.261
|
sp1571#
|
Actinobacteria
|
Kribbella
|
8311.72±732.74
|
16815.91±4769.13**
|
10
|
5398.30
|
1.23
|
0.64
|
0.806**
|
0.697*
|
sp1615#
|
Actinobacteria
|
Microlunatus
|
1270.57±39.9
|
1610.00±195.1**
|
8
|
638.42
|
0.37
|
0.66
|
0.794**
|
0.576
|
sp2002#
|
Actinobacteria
|
Streptomyces
|
1353.06±56.39
|
1790.82±285.96*
|
7
|
1206.15
|
0.37
|
0.69
|
0.733*
|
0.709*
|
sp4492#
|
Proteobacteria
|
Devosia
|
3952.31±519.85
|
5870.62±674.90**
|
5
|
1503.80
|
-0.49
|
0.72
|
0.842**
|
0.794**
|
sp4982#
|
Proteobacteria
|
Azospirillum
|
3357.51±94.17
|
3516.69±201.64
|
8
|
412.30
|
0.94
|
0.66
|
0.503
|
0.273
|
sp5385#
|
Proteobacteria
|
Burkholderia
|
7092.2±295.86
|
7920.63±713.51*
|
4
|
390.17
|
-1.08
|
0.63
|
0.685*
|
0.552
|
sp5454#
|
Proteobacteria
|
Paraburkholderia
|
1091.48±59.02
|
1208.08±101.81
|
3
|
5808.20
|
-1.25
|
0.67
|
0.697*
|
0.527
|
sp5496#
|
Proteobacteria
|
Acidovorax
|
457.68±32.01
|
617.36±56.87**
|
3
|
337.52
|
-1.57
|
0.67
|
0.939**
|
0.806**
|
sp5501#
|
Proteobacteria
|
Alicycliphilus
|
2216.43±86.58
|
2684.07±222.70**
|
14
|
4476.52
|
-0.41
|
0.71
|
0.939**
|
0.733*
|
sp5675#
|
Proteobacteria
|
Chromobacterium
|
1094.98±83.71
|
1246.80±102.88*
|
5
|
564.79
|
-1.08
|
0.72
|
0.891**
|
0.709*
|
sp5951#
|
Proteobacteria
|
Haliangium
|
22571.04±703.57
|
20814.08±1675.12
|
4
|
711.62
|
-1.03
|
0.63
|
-0.467
|
-0.491
|
sp744#
|
Actinobacteria
|
Acidimicrobium
|
1670.81±76.07**
|
1539.43±91.80
|
13
|
22530.34
|
-0.30
|
0.70
|
-0.600
|
-0.648*
|
sp1175&
|
Actinobacteria
|
Modestobacter
|
4728.08±178.51
|
4992.08±366.15
|
10
|
1243.11
|
2.53
|
0.46
|
0.527
|
0.297
|
sp4354&
|
Proteobacteria
|
Bradyrhizobium
|
16015.93±532.84
|
15386.41±1261.79
|
19
|
7755.65
|
2.62
|
0.28
|
-0.285
|
-0.515
|
sp7444&
|
Unclassified Bacteria
|
278853.56±5378.65
|
292085.12±19196.42
|
18
|
6825.71
|
2.75
|
0
|
0.455
|
0.382
|
Fom
|
Fusarium oxysporum
|
38.80±4.22
|
58.23±11.67*
|
/
|
/
|
/
|
/
|
1
|
0.794**
|
Strain Q2
|
Talaromyces purpureogenus
|
27.09±5.93
|
51.87±13.13*
|
/
|
/
|
/
|
/
|
0.794**
|
1
|
# means connectorssp; & means module hubs. * indicate a significant difference or correlation (P<0.05); ** indicate a significant difference or correlation (P<0.01). 1227: Arsenicicoccus sp.; sp1571: Kribbella flavida; sp1615: Microlunatus soli; sp2002: Streptomyces sp. 4F; sp4492: Devosia sp. A16; sp4982: Azospirillum thiophilum; sp5385: Burkholderia sp.; sp5454: Paraburkholderia sprentiae; sp5496: Acidovorax sp. RAC01; sp5501: Alicycliphilus denitrificans; sp5675: Chromobacterium vaccinii; sp5951: Haliangium ochraceum; sp744: Acidimicrobium ferrooxidans; sp1175: Modestobacter marinus; sp4354: Bradyrhizobium erythrophlei; sp7444: Unclassified Bacteria. the same below
OTU ID
|
Phylum
|
Genus
|
Relative abundance
|
Degree
|
Betw centrality
|
Zi
|
Pi
|
Correlation coefficient
|
Control
|
S-FOM
|
Fom
|
Strain Q2
|
sp17
|
Actinobacteria
|
Actinobacteria
|
219193.85±5453.8
|
229361.99±17870.27
|
48
|
17303.9
|
3.70
|
0.04
|
0.491
|
0.309
|
sp28
|
Proteobacteria
|
Burkholderiales
|
58742.72±4327.09
|
75223.07±5342.81**
|
15
|
2143.72
|
3.18
|
0.34
|
0.939**
|
0.733*
|
sp60
|
Proteobacteria
|
Myxococcales
|
15990.17±517.83
|
16788.61±1396.37
|
34
|
103336.29
|
3.93
|
0.11
|
0.588
|
0.503
|
sp61
|
Proteobacteria
|
Bradyrhizobium
|
16015.94±532.85
|
15386.41±1261.79
|
45
|
61630.75
|
3.96
|
0.13
|
-0.285
|
-0.515
|
sp78
|
Proteobacteria
|
Anaeromyxobacter
|
11863.22±432.14
|
12436.17±999.12
|
27
|
21087.43
|
3.02
|
0.07
|
0.552
|
0.479
|
sp115
|
Proteobacteria
|
Bradyrhizobium
|
7959.14±370.72
|
7464.79±667.44
|
44
|
21355.7
|
3.96
|
0.09
|
-0.297
|
-0.515
|
sp129
|
Proteobacteria
|
Myxococcus
|
6457.37±267.92
|
6955.99±573.05
|
43
|
168055.48
|
4.23
|
0.37
|
0.564
|
0.491
|
sp130
|
Proteobacteria
|
Bradyrhizobium
|
6510.68±226.33
|
6357.98±519.76
|
39
|
40592.81
|
3.37
|
0.10
|
0.006
|
-0.309
|
sp149
|
Proteobacteria
|
Bordetella
|
5217.8±254.93
|
6173.29±578.51*
|
25
|
33010.98
|
2.69
|
0.70
|
0.891**
|
0.782**
|
sp182
|
Acidobacteria
|
Acidobacterium
|
5129.82±157.81
|
4827.77±347.97
|
26
|
16365.47
|
2.71
|
0.14
|
-0.333
|
-0.285
|
sp192
|
Proteobacteria
|
Thiobacillus
|
4371.98±243.68
|
5257.59±502.88*
|
21
|
7835.58
|
2.98
|
0.53
|
0.891**
|
0.782**
|
sp201
|
Proteobacteria
|
Sinorhizobium
|
4248.08±161.02
|
4684.56±385.43
|
43
|
444514.48
|
6.22
|
0.62
|
0.782**
|
0.576
|
sp204
|
Proteobacteria
|
Bradyrhizobium
|
4435.02±177.85
|
4345.54±320.6
|
41
|
14945.07
|
3.60
|
0.09
|
-0.018
|
-0.358
|
sp214
|
Acidobacteria
|
Terriglobus
|
4440.03±75.16
|
4250.3±283.31
|
31
|
58495.42
|
2.86
|
0.32
|
-0.333
|
-0.333
|
sp241
|
Proteobacteria
|
Blastochloris
|
3839.65±134.9
|
3882.32±283.18
|
33
|
12536.46
|
2.90
|
0
|
0.236
|
-0.006
|
sp243
|
Acidobacteria
|
Granulicella
|
4026.25±95.71
|
3803.9±278.89
|
34
|
61955.15
|
3.93
|
0.11
|
-0.273
|
-0.261
|
sp259
|
Proteobacteria
|
Deltaproteobacteria
|
3686.93±112.72
|
3767.67±274.11
|
23
|
12673.7
|
2.56
|
0
|
0.479
|
0.442
|
sp261
|
Proteobacteria
|
Stenotrophomonas
|
2927.55±169.54
|
3928.56±342.72**
|
14
|
6050.23
|
3.18
|
0.24
|
0.964**
|
0.758*
|
sp287
|
Actinobacteria
|
Frankia
|
3361.42±100.35
|
3336.42±271.59
|
34
|
56108.42
|
2.54
|
0.21
|
0.079
|
-0.176
|
sp315
|
Proteobacteria
|
Bradyrhizobium
|
2977.26±114.01
|
2986.82±221.23
|
44
|
28845.71
|
3.60
|
0.20
|
0.115
|
-0.091
|
sp331
|
Firmicutes
|
Limnochorda
|
2874.85±41.89
|
2862.25±205.78
|
26
|
18652.51
|
3.02
|
0
|
0.115
|
-0.03
|
sp370
|
Acidobacteria
|
Granulicella
|
2525.3±114.82
|
2422.1±191.61
|
27
|
18444.13
|
3.17
|
0
|
-0.297
|
-0.2
|
sp372
|
Acidobacteria
|
Acidobacteriaceae
|
2524.3±69.87*
|
2290.09±175.12
|
30
|
34555.83
|
3.32
|
0.13
|
-0.745*
|
-0.697*
|
sp457
|
Proteobacteria
|
Paraburkholderia
|
1709.18±67.22
|
1883.88±175.16
|
60
|
415485.3
|
7.83
|
0.70
|
0.758*
|
0.515
|
sp476
|
Proteobacteria
|
Chelatococcus
|
1682.15±24.3
|
1704.7±132.29
|
41
|
44814.77
|
3.60
|
0.09
|
0.358
|
0.018
|
sp509
|
Proteobacteria
|
Thiodictyon
|
1545.94±100.82
|
1710.59±135.38
|
19
|
74875.8
|
2.82
|
0.79
|
0.794**
|
0.673*
|
sp624
|
Proteobacteria
|
Chromobacterium
|
1171.53±63.5
|
1348.25±103.31*
|
81
|
815503.77
|
3.85
|
0.92
|
0.842**
|
0.685*
|
sp694
|
Actinobacteria
|
Nocardia
|
1022.74±20.87
|
1073.01±69.13
|
39
|
24694.67
|
2.90
|
0
|
0.624
|
0.406
|
sp728
|
unidentified
|
unidentified
|
1028.83±81.3
|
978.54±80.43
|
42
|
26429.36
|
3.20
|
0
|
-0.358
|
-0.467
|
sp852
|
Proteobacteria
|
Ahniella
|
780.82±19.17
|
863.22±73.02
|
17
|
105279.65
|
2.67
|
0.48
|
0.721*
|
0.636*
|
sp870
|
Actinobacteria
|
unidentified
|
793.18±41.23
|
801.08±69.09
|
43
|
17083.01
|
3.30
|
0
|
0.200
|
0.127
|
sp1067
|
unidentified
|
unidentified
|
550.49±39.26
|
554.54±49.22
|
49
|
83041.16
|
3.80
|
0.04
|
0.091
|
-0.127
|
sp1111
|
Proteobacteria
|
Desulfomicrobium
|
544.85±29.15
|
516.7±51.37
|
26
|
15229.88
|
3.02
|
0
|
-0.248
|
-0.212
|
sp1120
|
Proteobacteria
|
Desulfuromonadales
|
528.33±28.83
|
504.03±44.55
|
27
|
53267.22
|
3.17
|
0
|
-0.261
|
-0.297
|
sp1208
|
Gemmatimonadetes
|
unidentified
|
455.52±18.75
|
426.39±39.95
|
29
|
15740.84
|
3.47
|
0
|
-0.527
|
-0.321
|
sp1449
|
Deinococcus-Thermus
|
Deinococcus
|
288.51±6.72
|
283.98±20.24
|
29
|
16332.01
|
3.32
|
0.07
|
-0.079
|
0.103
|
sp1899
|
unidentified
|
unidentified
|
141.89±17.57
|
125.12±19.8
|
39
|
20524.19
|
2.90
|
0
|
-0.600
|
-0.588
|
sp5376
|
Actinobacteria
|
Microbacterium
|
6.95±2.21*
|
3.94±1.35
|
23
|
90227.86
|
2.56
|
0
|
-0.867**
|
-0.673*
|
sp5882
|
Ascomycota
|
Neurospora
|
3.37±1.99
|
4.12±1.03
|
36
|
164.45
|
2.59
|
0
|
-0.067
|
0.261
|
sp7622
|
Crenarchaeota
|
Sulfolobaceae
|
1.22±0.26
|
0.96±0.59
|
54
|
12867
|
2.55
|
0
|
-0.055
|
-0.309
|
Fom
|
Fusarium oxysporum
|
38.80±4.22
|
58.23±11.67*
|
/
|
/
|
/
|
/
|
1
|
0.794**
|
Strain Q2
|
Talaromyces purpureogenus
|
27.09±5.93
|
51.87±13.13*
|
/
|
/
|
/
|
/
|
0.794**
|
1
|
3.4. Isolation and screening of biocontrol agents from pathogenic soil of F. oxysporum
The soil invaded by F. oxysporum was used for the discovery of beneficial microbial species antagonism against Fom. More than 100 microbial isolates were screened and successfully revealed 14 biocontrol agents with strong antagonistic activity to Fom on the plate, including 7 bacterial strains, 2 Penicillium sp. (1 Talaromyces sp.), and 5 Trichoderma sp.. Fom inhibition rate was 52%-83% (Supplementary Table S1). The secondary screening was conducted in pot trials and successfully revealed 4 biocontrol agents that have strong control effects on Fusarium wilt of bitter gourd in the greenhouse, including bacterial strain SK2 and SK6, Talaromyces sp. strain Q2, and Trichoderma sp. strain M2, their control efficacy on Fusarium wilt were 68%-79% (Supplementary Table S8). Meanwhile, the relative abundance of sp1571 (Kribbella flavida), sp2002 (Streptomyces sp. 4F), sp4492 (Devosia sp.), sp5496 (Acidovorax sp.), sp5501 (Alicycliphilus denitrificans), sp5675 (Chromobacterium vaccinii) and F. oxysporum were significantly positively correlated with the relative abundance of strain Q2, and the correlation coefficient between strain Q2 and F. oxysporum reaches 0.79 (Table 1).
3.5. Micromorphology, biocontrol potential, and phylogenetic analysis of strain Q2
Strain Q2 had strong antagonistic activity to Fom on the plate (Fig. 2A ). Colonies of strain Q2 growing moderately fast, reaching 70.46 mm on PDA, 49.44 mm on CA, 76.68 mm on CYA, 76.22 mm on OA, 70.66 mm on CMA, 75.32 mm on MM, 67.06 mm on SSA, 61.00 mm on SDA and 66.56 mm on SMA at 25 ℃ for 7 days (Fig. 2D and Supplementary Table S9 ). Colonies of strain Q2 on PDA moderately deep, plane; margins low, entire; marginal mycelia yellow and white; conidia pile greyish green to dull green; sometime exudates minute reddish droplets; reverse dark ruby (Fig. 2D). Under light microscope, hyphae of strain Q2 were septate, smooth to coarse, and had branched. Conidiphore of strain Q2 produces from hyphal string. Conidiophores biverticillate; stipes smooth walled, top not swollen (Fig. 2B and C), metulae three to five, divergent, 8.0–13 µm × 2.3-3.0 µm; phialides acerose, three to six per metulae, 10–20 µm×1.8-3.0 µm; conidia oval or round, smooth walled, 3–3.5 × 2–2.5 µm. Conidia chain loose, similar cylinder shape. Ascomata not observed. The optimal growth temperature was 30 ℃, maximum 55 ℃; optimal growth pH was 7; optimal carbon source was glucose (Supplementary Fig. S5).
Strain Q2 had great potential for biological control. For example, it was able to produce β-1, 3-glucanase and chitinase (Fig. 2E and F). The β-1, 3-glucanase activity of the crude culture filtrate was 2.51 U·mL− 1 (Q2) and that of the culture filtrate of substrate induced was 3.13 U·mL− 1 (Q2-CW), 4.20 U·mL− 1 (Q2-Lam) and 4.38 U·mL− 1(Q2-CW-Lam), respectively (Fig. 2G). The chitinase activity of the crude culture filtrate was 1.79 U·mL− 1 (Q2) and that of the culture filtrate of substrate induced was 2.22 U·mL− 1 (Q2-CW), 1.92 U·mL− 1 (Q2-cChi) and 2.40 U·mL− 1 (Q2-CW-cChi), respectively (Fig. 2H). Strain Q2 was able to inhibit mycelia growth of eleven pathogenic fungi such as F. oxysporum f. sp. momordicae (Fom), F. graminearum, F. moniliforme, F. oxysporum f. sp. cucumarinum, F. oxysporum f. sp. niveum, Pyricularia oryzae, Trichothecium roseum, Cryphonectria parasitica, Cytospora chrysosperma, Phytophthora parasitica, Rhizoctonia solani (Supplementary Table S10). After co-culture for 5 days, obvious inhibitory zone was occurred between the colony of strain Q2 and Fom, which inhibited the mycelia growth of Fom by 65.86% (Fig. 2A and Supplementary Table S10).
The DNA sequencing was used to identify the strain Q2. The lengths of nucleotide sequence for ITS, beta-tubulin gene (BenA), and calmodulin gene (CaM) was 585 bp, 446 bp and 705 bp, respectively and be submitted to GenBank with the accession number KX432212, KY047419 and KX781300. The maximum likelihood (ML) phylogenetic tree showed that strain Q2 was the same as the type strain Talaromyces purpureogenus CBS 286.36, which indicated that strain Q2 is T. purpureogenus strains (Supplementary Fig. S6A). Together with the morphological identification, we determined the strain Q2 to be T. purpureogenus, anamorphic type Penicillium purpurogenum. The genome of strain Q2 was assembled with PacBio technology, has 57 scaffolds with a length of 27.73 Mb, including 9,600 genes with the average length of 1,633 bp (Supplementary Table S11). Although, 8968 genes of strain Q2 could be annotated in the KOG database, the function of 41.1% was unknown (Supplementary Fig. S6B). Meanwhile, a total of 463 CAZymes (Carbohydrate active enzymes) have been identified, encoding 200 glycoside hydrolases (GHs), 83 glycosyl transferases (GTs), 84 carbohydrate esterases (CEs), 79 auxiliary activities (AAs), 15 carbohydrate-binding modules (CBMs), and 2 polysaccharide lyases (PLs) (Supplementary Fig. S6C). The number of CE1 (Acetyl xylan esterase), CE10 (arylesterase), GH18 (chitinase), GH3 (β-glucosidase), GH16 (xylanase), and GH109 (α-N-acetylgalactosidase) were significantly higher than other CAZymes (Supplementary Fig. S6D).
3.6. The control efficacy of strain Q2 on Fusarium wilt of bitter gourd
The control efficacy of strain Q2 on Fusarium wilt of bitter gourd was 63.4% (greenhouse) and 60.2% (field) (Table 2). The disease incidence and disease index of bitter gourd seedlings were significantly decreased after treatment with strain Q2. Besides, the time of bitter gourd seedling infected by Fom was obviously postponed, which demonstrated that the resistance to Fusarium wilt of bitter gourd seedlings was enhanced after inoculation with strain Q2. Meanwhile, the obvious modifications the composition of the fungal communities at the genus level of the Fom pathogenic soil treated with strain Q2 were discovered in this study (Supplementary Fig. S7). The lower observed species and Chao1 index were detected in TP samples (The pathogenic soil with Fusarium inoculation was treated with antagonistic fungus T. purpureogenus strain Q2), whereas the difference was not significant, the TP samples showed a significantly lower Simpson index and Shannon index of fungi than S-FOM samples, but, the TP samples showed a significantly higher Simpson index and Shannon index of bacteria than S-FOM samples (Supplementary Fig. S7A). The relative abundance of Actinobacteria, Firmicutes, Ascomycota, Chytridiomycota, Streptomyces, Conexibacter, Sphingomonas, Rhodoplanes, Gemmatirosa, Sphingobium, Nocardioides, Mesorhizobium, Lysobacter, Mycobacterium, Micromonospora, Talaromyces, Aspergillus, Thermothelomyces, and Chaetomium were increased in TP samples. The decreases in the relative abundance of Fusarium, Cordyceps, Colletotrichum, Anthracocystis, Trichoderma, Thielavia, Cladophialophora, Gaeumannomyces, Sporisorium, Verticillium, Fonsecaea, Exophiala, Trametes, and Diplodia were observed in TP samples (Supplementary Fig. S7 B-E). Talaromyces became the largest fungal taxonomic group in the TP samples compared with S-FOM samples, the relative abundance of strain Q2 that obtained by comparing the genome of the strain Q2 to the metagenomic sequencing data in the TP samples was 2828.01 reads compared to S-FOM samples with 51.87 reads and control samples with 27.09 reads (Supplementary Fig. S7E and I). A random forest analysis was then applied to identify the major microbes that contribute to the variation in soil microbiota. Ascomycota, Thaumarchaeota, Euglenida, Firmicutes, Chytridiomycota, Actinobacteria, Nitrospirae, Bacteroidetes, Microsporidia and so on were the primary factor that affected features of soil of microbial community (Supplementary Fig. S7 F). Spearman’s correlation network analyses confirmed that Actinobacteria, Ascomycota, Nitrospirae, Tenericutes, Bacteroidetes, Bathyarchaeota, Beckwithbacteria, Campbellbacteria, Saccharibacteria, Tectomicrobia, Woesebacteria, Wolfebacteria, Elusimicrobia, Thaumarchaeota and T. purpureogenus obviously influenced the Simpson index of soil microbiota (Supplementary Table S12). Meanwhile, T. purpureogenus seems to have a strong positive correlation with Actinobacteria, Firmicutes, Thaumarchaeota, Ascomycota, Bacillariophyta, Bacteroidetes, Beckwithbacteria, Campbellbacteria, Saccharibacteria, Woesebacteria, Wolfebacteria, Elusimicrobia, and Tenericutes (Supplementary Table S12). The relative abundance of Fusarium decreased in the TP sample (Supplementary Fig. S7G), but the number of Fusarium in the rhizosphere of bitter gourd had decreased significantly on the Fusarium selective medium (Komada’s medium) (Supplementary Fig. S7H). The density of Fusarium under TP treatment was lower, with only 10,700 cfu·g− 1 soil compared to 17, 600 cfu·g− 1 soil in S-FOM treatment. Meanwhile, strain Q2 could colonize in soil, but its increasing trend in soil was gradually decreasing (Supplementary Fig. S7J).
Table 2 The biocontrol efficacy of strain Q2 on Fusarium wilt of bitter gourd in pot and field
Treatment
|
Pot
|
Field
|
Incidence (%)
|
Disease index
|
Control efficacy (%)
|
Incidence (%)
|
Disease index
|
Control efficacy (%)
|
Control
|
/
|
/
|
/
|
6.94±2.40 c
|
2.31±1.22 c
|
/
|
C-TP
|
/
|
/
|
/
|
2.78±2.41 c
|
0.93±0.80 c
|
/
|
S-FOM
|
95.70±7.45
|
72.07±24.85
|
/
|
45.56±3.85
|
34.44±3.92
|
/
|
TP
|
63.20±25.19
|
26.38±13.87
|
63.40
|
27.78±5.09
|
13.7±1.34
|
60.22
|
Data are mean±SD. Different letters in the same column indicate significant difference at P<0.05 level by Student's t-test.
|
To constructed the co-occurrence network between Control, S-FOM and TP to explain the influence F. oxysporum f. sp. momordicae (Fom) and T. purpureogenus (Tp) to soil microbiota (Fig. 3A and 3B), we compared the differences in network-level topological features between Control, S-FOM and TP. For different treatments, soil pH was significantly lower in TP treatment than in Control and S-FOM. However, electrical conductivity (EC), available phosphorus (AP), available iron (AFe), available manganese (AMn), soil organic matter (SOC), FDA hydrolase, polyphenol oxidase (PPO), urease (UE) and acid phosphatase (ACP) were higher in TP treatment than in Control and S-FOM (Supplementary Table. S6). Similar to soil enzyme, values for the vertex number, edge number, modularity and degree centralization were significantly higher for sub-networks in TP treatment than in Control, according to Tukey HSD tests (Fig. 3C). In contrast, values for the average path length and density were lower for sub-networks in TP treatment than in Control (Fig. 3C). These results suggest that the soil microbiota in TP treatment was more closely associated than in Control. To evaluate the relative contribution of soil chemical properties and soil enzyme activity to network-level topological features of the soil microbial network. In total, compared with soil chemical properties, the soil enzyme activity contributed to more of the variation in the network-level topological features. The contribution of soil FDA hydrolase, urease (UE), acid phosphatase (ACP), soil pH, available iron (AFe) and available manganese (AMn) to the network-level topological features overwhelmed that of other environmental factors with multiple regression analysis (Fig. 3D). Compared with F. oxysporum, T. purpureogenus seems to have a stronger positive correlation with vertex number, edge number, modularity and degree centralization and negative correlation with density based on Spearman’s correlation analysis (Fig. 3E). Meanwhile, F. oxysporum and T. purpureogenus had a strong positive correlation with SOC, AFe, S-UE and S-ACP, and T. purpureogenus also have a strong positive correlation with EC and AP (Fig. 3E). The degrees of the networks followed power-law distributions indicated that co-occurrence networks was a small world (Supplementary Fig. S8A). The co-occurrence network contained 1718 nodes (1717 microbial nodes and 1 environmental nodes), 3445 edges, and 72.73%-74.80% microbial biomass (Fig. 3). In the network, 98.72% of the nodes belong to the peripherals, 1.05% of the nodes belong to the connectors, and 0.17% of the nodes belong to the module hubs, according to the distribution of soil microbial species by Zi-Pi plot (Supplementary Fig. S8B). Among the 1717 key microbial species, 33.93% microbial species belong to Proteobacteria, 20.9% microbial species belong to Actinobacteria, 6.52% microbial species belong to Firmicutes, 3.78% microbial species belong to Thaumarchaeota, 3.08% microbial species belong to Euryarchaeota, and 2.91% microbial species belong to Ascomycota (Supplementary Fig. S8C). The relative abundance of Actinobacteria and Firmicutes were higher in TP treatment than in Control and S-FOM. There are 404 modules among the networks, of which 15 modules had more than 20 nodes, sunch as module 57, module 38, module 12, module 11, module 26, module 155, module 118, module 130, module 117, module 249, module 168, module 162, module 175, module 106 and module 180. The nodes of the 15 modules account for 41.58% of the total number of nodes. The microbial abundance of the 15 modules account for 50.09–55.50% of the total number of microbe, including Proteobacteria (36.36%), Actinobacteria (25.17%), Firmicutes (5.59%), Euryarchaeota (2.94%), Thaumarchaeota (2.94%), Bacteroidetes (1.82%), and Ascomycota (1.68%). The results suggest that Proteobacteria and Actinobacteria in soil were more often located in core positions in the network than other species.
The co-occurring network was built based on S-FOM and TP, in order to more clearly observe the influence of T. purpureogenus on the network interactions among microbes. The co-occurrence network included 4954 nodes (4946 microbial nodes and 8 environmental nodes), 16275 edges, 1752 modules and 87.79%-89.00% microbial biomass (Supplementary Fig. S9A). Proteobacteria, Actinobacteria, Acidobacteria Planctomycetes and Firmicutes were dominant phyla (Supplementary Fig. S9B). 13 modules had more than 50 nodes, sunch as module 897, module 1725, module 1491, module 1441, module 1448, module 1468, module 1368, module 543, module 1529, module 1450, module 1409, module 1492 and module 1505 (Supplementary Fig. S9A). There are 1512 nodes, 33 phyla and 62.77%- 64.63% microbial biomass in 13 larger modules (Supplementary Fig. S9 C). T. purpureogenus participated in the construction of co-occurrence network as environmental factor. It showed significant negative correlations with the Acidobacteria, Firmicutes, Chloroflexi Armatimonadetes, Chromerida, Chytridiomycota and Chlamydiae, obvious positive correlations with Actinobacteria and Dictyoglomi (Supplementary Fig. S9 E). In contrast, F. oxysporum showed positive correlations with the Acidobacteria, Firmicutes, Chloroflexi Armatimonadetes, Chromerida, Chytridiomycota and Chlamydiae, negative correlations with Actinobacteria and Dictyoglomi. Meanwhile, T. purpureogenus tended to co-occur with Actinobacteria and Proteobacteria of module 897. Module 897 was a module with Actinobacteria as the main component (Supplementary Fig. S9 D). The microbial species, such as Kribbella flavida, Frankia casuarinae, Kitasatospora setae, Nocardiopsis alba, Streptomyces niveus and Xylanimonas cellulosilytica, in module 897 were significantly positively correlated with the relative abundance of T. purpureogenus. The results indicated that T. purpureogenus strain Q2 involved in the resistance of soil microbial communities to the invasion of F. oxysporum, and the growth and development of F. oxysporum was inhibited by T. purpureogenus and its “friends”.
3.7 Inhibition of the growth and development of Fom by strain Q2 in co-culture
To understand the mechanisms of inhibition of growth and development of Fom by strain Q2 under natural condition. In this study, the liquid co-cultivation of strain Q2 and Fom was firstly attempted, and the growth of strain Q2 and FOM in the axenic and co-culture conditions in PDB were significantly different at 48 h. It was observed that the culture liquor of FOM appeared beige, thicker, hazy mixture, while the color of culture liquor of strain Q2 was relatively lighter and appeared red white after they were singly-cultured for two days. However, when the inoculation ratio of strain Q2 and FOM was greater than 2:1, after co-cultivation for two days, the culture broth color was red and deepened with time, and the mycelia of strain Q2 were dominant (Supplementary Fig. S10). Fom hyphae from the control group was not changed under the light microscope and hyphae grown and developed normally (Fig. 4A). Strain Q2 hyphae of all the groups were normally grown and developed, and the color of strain Q2 was not changed (Fig. 4A). There were morphological changes that hypha swelling appears round, the appearance of septa digestion, cytoplasma concentration and more than globular material in hypha cells of Fom compared with control group (Fig. 4A). Propidium iodide (PI) staining showed that the Fom conidia with red pigment accumulation lost vitality in T1 to T4 samples (Fig. 4B and C). The number of conidia of Fom in culture liquor was lower in T1, T2, T3, T4 and T5 treatment, with only 8.20 × 107 cfu·mL− 1, 6.73 × 107 cfu·mL− 1, 9.83 × 106 cfu·mL− 1, 4.33 × 106 cfu·mL− 1 and 1.77 × 106 cfu·mL− 1, respectively, compared to 1.20 × 108 cfu·mL− 1 in control treatment at 5th days of cultivation (Fig. 4D). The results showed that the hyphae growth and spores production of Fom was inhibited significantly in the co-culture by strain Q2, and the red pigment produced by Q2 penetrated cell wall and cell membrane of the dead cell of Fom and deposited.
The presence of propidium iodide and the red pigment in Fom cells indicated that the integrity of the cell wall and cell membrane of Fom were damaged. We investigated the effect of strain Q2 on malondialdehyde (MDA) concentration and triglyceride content of Fom cell. The results showed that the MDA content and triglyceride content of Fom cell increased in T1 to T4 samples compared to control (Fig. 4F and G). The MDA content and triglyceride content in T2 treatment were significantly higher than those in the control. At the same time, it is worth noting that the small fragments increased in the culture filtrate of T3 samples (Fig. 4B), and the N-acetylglucosamine (N-acetylglucosamine is a hydrolysate of chitin) content in the culture filtrate of T1 to T3 samples was significantly higher than that in the control (Fig. 4E), and the content of N-acetylglucosamine gradually increases with inoculation concentration of strain Q2 increases. The results indicated that strain Q2 maybe inhibit the growth and development of Fom by destroying the integrity of the cell wall and cell membrane of Fom.
3.8 The influence of strain Q2 on the metabolic function of Fom at transcriptional level
A total of more than 336.47 million high quality clean reads were obtained after removing duplicate reads, trimming adapters and low quality sequences from six libraries. Q30 percentage was over 92% for all the samples. These data showed that the RNA-Seq quality was applicable for further analysis. In FOM samples, 89.20 % to 89.73 % of the total reads were mapped to the genome of F. oxysporum f. sp. cubense race 1; and more than 95 % of the reads in each sample were uniquely mapped to the genome. In Q2-FOM samples, 28.25 % to 47.17 % of the total reads were mapped to the genome of F. oxysporum f. sp. cubense race 1; and 96.86 % to 97.15% of the reads in each sample were uniquely mapped to the genome (Supplementary Table S13).
Differentially expressed genes (DEGs) were identified between FOM and Q2-FOM groups through comparing the expected read counts normalized using the DEseq package. To compare the expression levels of the DEGs between two groups, sequences count based differential expression analysis was performed. It was observed that 4,893 genes expression levels in FOM were identified as differentially expressed in FOM and Q2-FOM groups, including 2,435 up-regulated and 2,458 down-regulated (Supplementary Fig. S11A). The Gene Ontology analysis (GO) and KEGG analysis of DEGs was further conducted (Supplementary Fig. S11B and C). Following GO annotation, the DEGs were classified into different groups belonging to biological process (BP), molecular functions (MF) and cellular component (CC) categories. In the all category, the highly represented groups were ‘intrinsic component of membrane’ (410 up-regulated genes and 748 down-regulated genes), ‘integral component of membrane’ (410 up-regulated genes and 744 down-regulated genes), ‘membrane part’ (416 up-regulated genes and 748 down-regulated genes), ‘membrane’ (428 up-regulated genes and 762 down-regulated genes), ‘oxidoreductase activity’ (355 up-regulated genes and 207 down-regulated genes), ‘transmembrane transport’ (149 up-regulated genes and 256 down-regulated genes) and ‘transmembrane transporter activity’ (65 up-regulated genes and 146 down-regulated genes) groups. In the Q2-FOM groups, the genes associated with membrane and transport appeared to be the most significantly enriched. To further investigate the biochemical pathways of these DEGs in the process of T. purpureogenus inhibiting the growth of Fom, we mapped all DEGs identified in the RNA sequencing to terms in the KEGG database. The results showed the DEGs could be classified into four pathways, including metabolism, genetic information processing, environmental information processing and cellular processes. Top 20 enriched KEGG pathways associated with DEGs affected by T. purpureogenus were showed in Supplementary Fig. S11B, including glyoxylate and dicarboxylate metabolism, glutathione metabolism, valine, leucine and isoleucine degradation, glycerophospholipid metabolism, phenylalanine metabolism, nitrogen metabolism, propanoate metabolism, pyruvate metabolism, ABC transporters, galactose metabolism, tyrosine metabolism and glycerolipid metabolism. These results further indicated that stress of strain Q2 may greatly influence carbohydrate metabolism, amino acid metabolism, lipid metabolism, energy metabolism, component of membrane and membrane transport of F. oxysporum.
Based on the STRING output, the interaction of 371 coding genes was described (Fig. 5A), including cell cycle relating genes, starch and sucrose metabolism relating genes, amino sugar and nucleotide sugar metabolism relating genes, glycan biosynthesis and metabolism relating genes and biosynthesis of secondary metabolites relating genes (Supplementary Table S14). Real-time PCR was used to validate the expression levels of thirteen genes (Fig. 5C), including cell cycle relating genes, starch and sucrose metabolism relating genes, amino sugar and nucleotide sugar metabolism relating genes and glycan biosynthesis and metabolism relating genes. Among these genes, FOC1 g10015395 (CDC5, Cell cycle serine/threonine-protein kinase), FOC1 g10015375 (CG21, G2/mitotic-specific cyclin-B), FOC1 g10014984 (MPG1, Mannose-1-phosphate guanyltransferase), FOC1 g10014983 (HXKG, Glucokinase), FOC1 g10014811(GNPI1, Glucosamine-6-phosphate isomerase 1), FOC1 g10014809 (NAGA, N-acetylglucosamine-6- phosphate deacetylase), FOC1 g10014793 (INV, Invertase), FOC1 g10014440 (GPI1, N-acetylglucosaminyl- phosphatidylinositol biosynthetic protein), FOC1 g10013913 (ARK1, Serine/threonine-protein kinase), FOC1 g10013617 (BGL1B, Beta-glucosidase 1B), FOC1 g10012541 (TREA, Trehalase), FOC1 g10012205 (MALT, Alpha-glucosidase), FOC1 g10012008 (GPI18, GPI mannosyltransferase 2), FOC1 g10010259 (BIR1, Protein), FOC1 g10009667 (GPI15, Phosphatidylinositol N-acetylglucosaminyltransferase subunit), FOC1 g10009617 (WSC1, cell wall integrity and stress response component 1), FOC1 g10009158 (TPSY, Putative alpha, alpha-trehalose-phosphate synthase), FOC1 g10008820 (AGDC, Probable alpha/beta-glucosidase), FOC1 g10008111 (ASE1, Anaphase spindle elongation protein 1), FOC1 g10006897 (GPI2, Phosphatidylinositol N-acetylglucosaminyltransferase subunit), FOC1 g10005945 (TPS2, Trehalose-phosphatase), FOC1 g10005528 (CHS7, Chitin synthase 7), FOC1 g10003572 (AMYG, Glucoamylase) and FOC1 g10003114 (BUB1, Checkpoint serine/threonine-protein kinase) were down-regulated in Q2-FOM. FOC1 g10015825 (GUN, Endoglucanase), FOC1 g10015543 (AGM1, Phosphoacetylglucosamine mutase), FOC1 g10012466 (GAL10, Bifunctional protein) and FOC1 g10011050 (YBB2, Probable mannose-1-phosphate guanyltransferase) were up-regulated in Q2-FOM. Four GPI biosynthesis and one chitin synthases genes were all down-regulated in Q2-FOM. The gene expression results from the real-time PCR were consistent with the transcriptome data (Fig. 5B).
We found a gene with CFEM domains (the characteristic eight cysteine residues at N terminus) and putative GPI-anchored site (FOC1 g10014283) in further analysis of differentially expressed genes. The expression level of FOC1 g10014283 was significantly decreased in Q2-FOM samples compared to FOM samples, although his function is unknown in F. oxysporum (Fig. 6A and Supplementary Table S14). We experimentally isolated and characterized an FOC1 g10014283 orthologue, FomCFEM, in F. oxysporum f. sp. momordicae.
3.9 FomCFEM is important for full virulence and stress tolerance
The SMART-PFAM revealed that FomCFEM consists of a CFEM domain, a signal peptide, and the potential GPI-modification site (the amino acids N291 and A292) (Supplementary Fig. S12A and Supplementary Fig. S13). A phylogenetic analysis of the putative FomCFEM, homologous sequences of Fusarium and other fungal genes with CFEM domain indicated that homologous sequences of FomCFEM were common in Fusarium and the FomCFEM is closely homologous to the gene with CFEM domain from Trichoderma asperellum (GFP54046.1) (Supplementary Fig. S12B). To assess functions of FomCFEM, in F. oxysporum f. sp. momordicae by homologous recombination, FomCFEM deletion transformants harboring a hygromycin-resistance gene instead of the entire FomCFEM were generated. After PCR and Southern blot assays, the predicted FomCFEM separate mutants were obtained (Supplementary Fig. S12C and Fig. S14B). One of the mutants (designated as ∆FomCFEM) was selected for complementation by reintroducing the entire FomCFEM gene under the control of its native promoter into the ∆FomCFEM mutant (Supplementary Fig. S12D).
To determine the role of FomCFEM in pathogenicity, the rhizosphere soil of bitter gourd seedlings were inoculated with conidial suspensions of WT, ∆FomCFEM mutant, and ∆FomCFEM-C strains (the inoculation amount was 1×106 spores/g·soil). The disease incidence and disease index of bitter gourd seedlings were significantly decreased after treatment with ∆FomCFEM mutant (Fig. 6B and C). The ∆FomCFEM mutant resulted in a 58.8% reduction in virulence compared to the WT strain. These results suggest that FomCFEM is essential for full virulence in F. oxysporum f. sp. momordicae. Although conidial production of ∆FomCFEM was decreased compared with WT and ∆FomCFEM-C strains, there was no difference in the conidium morphology and conidial germination of ∆FomCFEM, WT, and ∆FomCFEM-C (Fig. 6F-G). In addition, to investigate whether FomCFEM is involved in external stress tolerance, growth assays of indicated strains on CM media supplemented with salt stress (0.7 mol·L− 1 NaCl and 0.7 mol·L− 1 KCl), osmotic stress (1 mol·L− 1 sorbitol), H2O2 (5 mmol·L− 1) and cell wall stress (0.05% SDS, and 0.2 g·mL− 1 CR) were analyzed, respectively. The results showed that the ∆FomCFEM strain was more hypersensitive to the cell wall stress compared with WT and ∆FomCFEM-C strain (Fig. 6D and E). These results indicated that ∆FomCFEM may play an important role in the integrity and function of the cell wall of F. oxysporum f. sp. momordicae.