Relationship of microbial communities and suppressiveness of Trichoderma fortified composts for pepper seedlings infected by Phytophthora nicotianae

The understanding of the dynamic of soil-borne diseases is related to the microbial composition of the rhizosphere which is the key to progress in the field of biological control. Trichoderma spp. is commonly used as a biological control agent. The use of next generation sequencing approaches and quantitative PCR are two successful approaches to assess the effect of using compost as substrate fortified with two Trichoderma strains (Trichoderma harzianum or Trichoderma asperellum) on bacterial and fungal communities in pepper rhizosphere infected with Phytophthora nicotianae. The results showed changes in the bacterial rhizosphere community not attributed to the Trichoderma strain, but to the pathogen infection, while, fungi were not affected by pathogen infection and depended on the type of substrate. The Trichoderma asperellum fortified compost was the most effective combination against the pathogen. This could indicate that the effect of fortified composts is greater than compost itself and the biocontrol effect should be attributed to the Trichoderma strains rather than the compost microbiota, although some microorganisms could help with the biocontrol effect.


Introduction
Sweet pepper (Capsicum annum L.) is one of the main horticultural crops in Murcia (Southeast Spain), with over 1334 ha of total cultivated area under greenhouses. In this area, Phytophthora nicotianae (P. nicotianae) has been reported as the main causal agent of Phytophthora root rot in pepper plants [1]. The management of this disease is based on phenylamide fungicides but fungicide-tolerant strains have been detected [2].
Seedlings with two true leaves (14 days after sowing) were inoculated with the pathogen (P. nicotianae) (n = 30). The inoculation dose for all treatments was 10 5 cfu g -1 substrate (P_CPTh; P_CPTa; P_CP; P_P). Substrate samples were collected when the experiment was set up and when seedlings were harvested four weeks after sowing (rhizosphere) and stored at -80˚C for molecular analysis.
Seedling infection by P. nicotianae was recorded every day and the cumulative number of infected plants was also recorded from the day after pathogen inoculation. Percentage of dead seedling was calculated as the percentage of diseased plants out of total number of growing plants.

DNA extraction
Extractions of total DNA from substrates of each treatment inoculated and not inoculated with pathogen at the end of the experiment (0.5 g) were carried out in triplicate. The DNA extraction was carried out using the FastDNA 1 Spin Kit for soil (Q-Biogene, Carlsbad, CA, USA) following the manufacturer´s instructions. DNA concentrations of samples were determined using a NanoDrop 1 ND-1000 Spectrophotometer (Thermo Fisher Scientific Inc., USA) and stored at -20˚C until required.

Next Generation Sequence (NGS)
For molecular analysis of bacterial communities, variable regions V1, V3, V6 and V1-V2 of the 16S rRNA gene were amplified using the four primers pairs 8F/120R, F388/R534, F968/R1073 and 8F/R361 [20,21] and for fungi ITS1 region was amplified with a pair of primers ITS2/ ITS5 [22]. Each replicate sample was amplified, and amplicons for the triplicate samples were purified using the kit QIAquick PCR Purification Kit (Qiagen, Hilden, Germany), and composited together into equimolar concentration prior to sequencing. For bacterial PCR amplification, each 25 μl PCR mix contained the following regents: 1X KAPA2G Fast HotStart ReadyMix2 (2X) (Kapa Biosystems, Boston, MA, USA), 1.5 mM MgCl 2 , 0.5 μM of each primer, and 5 μl of DNA. The thermal cycler conditions were firstly 15 cycles of denaturation at 90˚C for 30s, amplification with a temperature gradient of 70˚C-50˚C for 30 s and a final extension of 72˚C for 30s. Secondly, samples had 30 cycles of denaturation at 94˚C for 45 s, amplification at 50˚C for 45 s and a final extension of 72˚C for 45 s.
For fungi amplification, each 25 μl PCR mix contained the following regents: 1X PCR buffer (Biotools, Madrid, Spain), 1.5 mM MgCl 2 , 0.4μM of each primer, 0.5 mg mL -1 BSA, 200μM dNTPs mix, 2. 5 U TaqPolymerase, and 2 μl of DNA. The thermal cycler conditions were firstly 1 cycle of denaturation at 95˚C for 5 min, followed by 35 cycles of denaturation at 95˚C for 1 min, amplification at 56˚C for 40 s, extension of 72˚C for 1 min, and a final extension of 72˚C for 7 min.
A library was created using Ion Plus Fragment Library Kit, and barcodes were added by Ion Xpress ™ Barcode Adapters 1-96 Kit (Life Technologies, Carlsbad, CA, USA). The template preparation was performed by Ion OneTouch™ 2 System and the Ion PGM™ Template Kit OT2 400 (Life Technologies, Carlsbad, CA, USA). Finally, the platform sequenced the samples using Ion Torrent PGM (Life Technologies, Carlsbad, CA, USA) with the kit Sequencing Kit Ion PGM 400 in chips Ion 318 Chip kit and Ion 314 Chip kit.
Briefly, before sequences processing, it was carried out a quality-filtering step with QIIME, where sequences smaller than 60bp or with a mean quality score below 25 were removed. After that, primers and barcodes were removed and a chimera filter (QIIME) was used. De-replication and singletons discarding steps were performed using USEARCH software. The remaining high quality sequences were grouped in operational taxonomic units (OTUs) following open-reference OTU picking protocol for QIIME, where sequences were clustered against the Green Genes v13_8, using the uclust ref algorithm at 97% similarity. Sequences not matching the database were subsequently clustered novo. A representative set of OTUs was generated and then the taxonomy of each OTUs was assigned using the same database.
For fungi we used the software packages QIIME v1.8.0. [23], USEARCH v7.0.1090 [24,25] and ITSx extractor [26] following ITS Profiling Data Analysis Pipeline protocol recommend by Brazilian Microbiome Project (http://brmicrobiome.org). As in the previous protocol, during the quality-filtering step, sequences smaller than 60bp or with a mean quality score below 25 were removed. Primers were removed and a chimera filter (QIIME) was used. De-replication step was carried out, and Singletons were eliminated, both with USEARCH software. The variable regions of the remaining sequences were extracted with the ITSx extractor. These sequences were grouped in OTUs with the Usearch with Uparse method. The remaining high quality sequences were grouped in operational taxonomic units (OTUs) following open-reference OTU picking protocol for QIIME, where sequences were clustered against the UNITE database using the uclust ref algorithm at 97% similarity. For P. nicotianae and T. harzianum primers and probes were described in Blaya et al. [1] and Lopez-Mondejar et al. [27] respectively. For T. Asperellum, we designed a new primers and probe on the transcription enhancer factor-1 (TEF-1) DNA binding site: TaspFw: 5'-GGCAGCAACCCCGCTAT-3'; TaspRv: 5'-ACGACGCGATTGAGCAAATA-3' and Tasp-Pr: 5'-6FAM-CCACTGCACCTCTTCCATCACCCA-ZEN/IBFQ-3').The efficiency was 95% and sensibility 10 copies. The specificity was shown on S1 Table  The thermocycling conditions for both Trichoderma strains and the pathogen P. nicotianae were 95˚C for 10 min, followed by 40 cycles of 95˚C for 10 s and 60˚C for 40 s. Each sample was analyzed in triplicate and to control the potential presence of PCR inhibitions and internal positive control (IPC) (TaqMan 1 Exogenous Internal Positive Control Reagents, Applied Biosystems 1 ) were included in all reactions according to the manufacturer´s recommendations. Each run contained one negative (bdW) and a DNA positive control.

Statistical analysis
Results from the suppressive bioassay and quantification of BCAs and P. nicotianae were subjected to the non-parametric Kruskal-Wallis test and median P-values 0.05 were considered significant. Statistical analyses were performed using SPSS 23.0 software (SPSS Inc., Chicago, IL, USA). Principal Component Analysis of bacterial and fungal communities was performed with Canoco for Windows 4.5.

Suppressive effect and P. nicotianae abundance in the different substrates
The suppressive effect of the compost fortified with T. asperellum (P_CPTa) was significantly (p<0.05) more effective than the one fortified with T. harzianum (P_CPTa) (Fig 1A). The treatment P_CPTa showed the least percentage of dead pepper seedlings (p<0.05) while, the non-fortified compost and peat (P_CP and P_P) showed the highest dead seedling (Fig 1A).
qPCR data showed significant differences (p<0.05) of ITS copies g -1 of P. nicotianae after 28 days of sowing between treatments (Fig 1B). The lowest number of P. nicotianae was observed in treatment (P_CPTa) fortified with T. asperellum (p<0.05), while the treatments with compost treatment (P_CP) showed the highest values (p<0.05) compared with the rest of treatments ( Fig 1B).
Abundance of BCAs in the rhizosphere of different substrates qPCR showed that the amount of T. asperellum did not change significantly (p<0.05) with the pathogen after 28 days of inoculation (P_CPTa) ( Table 1), while in the absence of the pathogen (P. nicotianae) (CPTa) the amount of T. asperellum decreased significantly (p<0.05) ( Table 1).
On the other hand, T. harzianum did not change with or without pathogen inoculation after 28 days of inoculation (P_CPTh, CPTh) ( Table 1). Values were 4 log times higher for T. harzianum compared with T. asperellum. Peat (P) also showed natural T. harzianum but it was significantly lower than the amount of T. harzianum in compost and did not change along the bioassay with or without pathogen inoculation (P_CPTh, CPTh).

Bacterial composition of different treatments by NGS analysis
After filtering the readings based on quality control, 868.608 sequences with an average length of 181 bases were obtained from 8 samples and assigned to 65.643 OTUS. All of the sequences obtained were classified into 38 phyla and the remaining sequences were classified as unassigned and other bacteria. The relative abundance (>0.5%) for all samples showed Proteobacteria as the dominant phylum with Alphaproteobacteria (41.93%) and Gammaprotobacteria (6.76%) being the most dominant class within that phylum. Actinobacteria, Bacteroidetes and Chloroflexi were also relatively abundant (19.53%, 10.46% and 5.78% respectively) (all accounting of the OTUs in all samples) (Fig 2). The relative abundance at the phylum level varied across the different samples (Fig 2). Some differences were found between P. nicotianae non-infected compost treatments (CPTh, CPTa and CP) and peat treatment (P). Peat treatment showed higher relative abundance of Gammaproteobacteria, TM7 and Chlorobi and lower relative abundance of Gemmatimonadetes and Chloroflexi. No differences were shown between fortified composts (CPTh and CPTa) and compost treatment (CP).
At this level three clear groupings of the samples could be observed on axis 1 (69.2%) of the PCA (Fig 3A), one group that grouped P. nicotianae infected fortified compost treatment with both Trichoderma strains (P_CPTh, P_CPTa); a second group that grouped peat treatments independently of pathogen inoculation (P, P_P); and the third group that grouped the rest of treatments (CP, P_CP, CPTh and CPTa) (Fig 3A).
PCA of the 22 most abundant genera obtained (S2 Table) with a relative abundance (>0.5%) resulted in a similar separation to phylum level on axis 1 (86%) (Fig 3B). A first group of P_CPTh and P_CPTa; a second group of the peat treatments (P and P_P) and the third group containing the other treatments.

Fungi composition of different treatments by NGS analysis
After filtering the readings based on quality control, 1173993 sequences with an average length of 211 bases were obtained from 8 samples assigned to 216457 OTUS.
Taxonomic assignment of fungal OTUs revealed that at the phylum level, the Ascomycota were the most abundant in all treatments (71.39%) followed by Basidiomycota (25.28%) and Zygomycota (1%) (all accounting of the OTUs in all samples). The percentages of sequences that were classified as unidentified fungi were lower of 2.3% of the OTUs in all samples. The relative abundance at the phylum level varied across the different treatments (Fig 4). Independently of P. nicotianae infection, compost treatments (CPTh, CPTa, CP and P_CPTh, P_CPTa and P_CP) showed higher Basidiomycota and lower Ascomycota than peat treatments (P and P_P). The relative abundance of Zygomycota in the P. nicotianae infected treatments decreased in all compost treatments (P_CPTh, P_CPTa and P_CP), but not in peat (P_P).

T. asperellum (Log copies TEF
At the genus level, the PCA of the top 22 classified fungi (>0.5%) (S3 Table revealed that the microbial community varied across the different treatments independently of pathogen. Three groups were observed across Axis 1 (78.3%). One composed of CPTh and P_CPTh; the second formed by the other compost treatments (CPTa, P_CPTa, CP and P_CP); and the third one composed of the peat treatments (P_P and P) (Fig 5).

Discussion
Compost has been widely used to control plant disease caused by different pathogens such as Phytophthora spp. [19], although it is well known that not all composts are suppressive and depend on the extant antagonist microorganisms, plant host, pathogen species involved and the characteristics of compost [28]. The addition of a specific biological control agent to compost has been reported as leading to a substrate with a broader-ranging suppressive effect [29,30].
Our experiment showed that composts fortified with the two different Trichoderma strains showed a lower percentage of dead seedlings, being in both cases significantly more effective against P. nicotianae than non-fortified compost and peat. The combination of organic substrates and biocontrol agents suppress plant pathogens through different mechanisms which can be summarized as direct interactions such as nutrient and space competence, antibiotic compound production and mycoparasitism; and indirect interaction through plants, such as systemic and acquired resistance (ISR and SAR) [31,5,32].
P. nicotianae abundance measured on T. harzianum fortified compost did not decreased assuming that no direct pathogen-BCA interaction occurs, and that plant defense induction could be involved [33]. On the other hand, T. asperellum fortified compost decreased P. nicotianae abundance and it could be due to a direct pathogen-BCA interaction, probably, producing a mycoparasitism process by T. asperellum hydrolytic enzyme secretion, causing the hydrolysis of the pathogen cell wall or antibiotic compound production [33]. Furthermore, the population of T. asperellum fortified compost decreased significantly (p<0.05) after 28 days, while the population of T. harzianum did not change during the bioassay. This could be due to the different suppressive mechanisms and could be influenced by the composition of the root exudates and the microbes in the rhizosphere where T. harzianum could be better adapted.
The study of bacteria and fungi communities through 16S and ITS sequencing analysis showed some differences as other authors pointed out [18,34,35,36] which would help to understand the different suppresiveness effects observed.
The bacterial phyla abundance revealed that the most abundant phyla for all treatments were Alphaproteobacteria and Gammaprotobacteria, documented as typical compost bacteria [37] and Actinobacteria and Bacteroidetes. Similar results were also observed for different types of composts and peat [34,35,36,38]. Bacterial community from the fortification of compost with the two different Trichoderma strains did not show any important differences in the relative abundance of different phyla compared to non-fortified compost. Nevertheless, peat treatments were enriched by Gammaproteobacterias [39] and Candidate division TM7, as first identified in a German peat bog [40], and as has been shown to be present in soil, sediments, wastewater, animals and a host of clinical environments [41,42].
PCA of bacterial community showed that the variation at phylum level is more due to the infection with the pathogen than the introduction of Trichoderma strains. The interaction between P. nicotianae-Trichode rma strains and rhizosphere of fortified compost treatment showed Bacteroidetes enrichment and in particular the more relative abundance of Pedomicrobium, Hyphomicrobium, Bacillus, Bdellovibrio and Gammaproteobacteria compared to nonfortified compost treatment, indicating that they may be involved in disease suppression of P. nicotianae. Similar results were observed by Kyselkova et al. [43] in soil suppressive to tobacco black rot caused by Thielaviopsis basicola, or against R. solani infection especially controlled by Pseudomonadaceae [44]. Blaya et al. [38] also demonstrated that these genera were found in pepper rhizosphere when suppressive composts were used against P. nicotianae, but not in the ones that did not show suppressive effect.
On the other hand, P. nicotianae infection showed a relative increase of Actinobacteria but not only in the fortified compost treatment. This could indicate that in this case these microbial communities have a bigger role in decomposition of organic materials particularly for degradation of macromolecules such as cellulose, hemicellulose, etc. [45] than in a positive impact on plant disease suppression due to a strong ability to produce antibiotic-like compounds [46,47]. In this sense, Bonanomi et al. [48] concluded that disease suppression was only correlated with Actinobacteria in a limited number of experimental cases.
Between the top 22 abundant genera, Bacillus was relatively more abundant in compost treatments than in peat. It is characterized as being able to form stable and extensive biofilm [49] composed by secreted antifungal compounds, such as surfactin, bacillomycin and microlactin that protect plants against attack by soil-borne pathogens [49,50]. Other genera of bacteria that have been found in compost treatment higher than in peat described as biological control were Streptomyces [51] and Microbacterium [52].
In the fungal community study, Ascomycota and Basidiomycota were the most abundant phyla, in agreement with other studies with similar organic materials [53,38]. Ascomycota has been observed during different composting processes [34,36], where most of the microorganisms were saprophytic and lived on dead organic matter that they help to decompose [54]. The variation at phylum level is due principally to the type of substrate and not to the pathogen infection. T. harzianum fortified compost showed a different fungal diversity, by increasing Ascomycota and by decreasing Basidiomycota, to the other two compost treatments.
It can be considered that different genera as Trichoderma, Fusarium and Myriococcum were the most abundant of compost treatment compare to peat independently of pathogen infection that have been related with suppressiveness. Trichoderma species and Fusarium species has been considered as biological control agent [27,55] or even species from genera Myriococcum has been described to produce antifungal antibiotics [56].

Conclusion
The use of different biological control agents such as T. harzianum and T. asperellum in fortified compost can be more effective to reduce P. nicotianae symptoms in pepper seedlings than only the compost. In addition, the results showed that T. asperellum fortified compost turned out to be the most effective combination against the pathogen. Changes in the bacterial rhizosphere community were not attributed to the Trichoderma strain but to the pathogen infection. On the other hand, the fungal rhizosphere community depended on the substrate but was not affected by plant infection.
These results suggest that the use of next generation sequencing approaches represents a useful method for studying microbial interactions in the rhizosphere, and it is essential to know the effect of biological control agents such as Trichoderma spp. in the plant substrate.
Supporting information S1