Vanillic acid changed cucumber (Cucumis sativus L.) seedling rhizosphere total bacterial, Pseudomonas and Bacillus spp. communities

Soil microorganisms are key drivers of plant productivity in terrestrial ecosystems, yet controls on their diversities and abundances are not fully elucidated. Phenolic acids, released through plant root exudation and residues decomposition, are usually referred as autotoxins of several crops, including cucumber. In this study, effects of vanillic acid (VA) on cucumber rhizosphere microbial communities were investigated by treating cucumber seedlings with VA every two days for five times. Amplicon sequencing, PCR-denaturing gradient gel electrophoresis and quantitative PCR were used to analyzed the 16S rRNA genes of total bacterial, Pseudomonas and Bacillus spp. communities. Results showed that VA at 0.05 μmol g−1 soil changed total bacterial community diversity and composition. In particular, VA inhibited the relative abundances of genera with plant-beneficial potentials, such as Bacillus and Lysobacter spp. Moreover, VA changed Pseudomonas and Bacillus spp. community compositions by altering the number and/or relative abundances of their OTUs; and decreased Bacillus spp. community abundance at 0.02 to 0.2 μmol g−1 soil and Pseudomonas spp. community abundance at 0.2 μmol g−1 soil. Overall, VA changed cucumber seedling rhizosphere total bacterial, Pseudomonas and Bacillus spp. communities, which maybe be associated with the adverse effects of VA on cucumber growth under soil conditions.

. Relative abundances of main bacterial phyla (a) and classes (b) in cucumber rhizosphere soils treated with water (W) or vanillic acid at 0.05 μmol g −1 soil (VA). Bacterial phyla and classes with average relative abundances >0.5% in at least one treatment were shown. Data are represented as the means of three independent replicates.
Bacterial Community Diversity and Structure. Rhizosphere soils treated with vanillic acid had higher Shannon and inverse Simpson indices than rhizosphere soils treated with water (P < 0.05) (Fig. 4a). However, number of OTUs, ACE and Chao indices were similar in rhizosphere soils treated with water and vanillic acid (Fig. 4a).
Principal coordinates analysis at the OTU level, based on both Bray-Curtis and UniFrac distance dissimilarities, revealed a clear separation between rhizosphere soils treated with water and vanillic acid (Fig. 4b,c).

Bacillus and Pseudomonas spp. Community Compositions. Eight and five OTUs aligned to Bacillus
and Pseudomonas spp., respectively, were detected in all treatments as determined by Illumina Miseq sequencing (Table S1). OTU275, OTU423 and OTU1458 were classified as B. luciferensis, P. flexibilis and P. resinovorans, respectively; while most of these OTUs could be aligned at the species level. For Bacillus spp., six OTUs were detected in rhizosphere soils treated with vanillic acid and eight OTUs in rhizosphere soils treated with water; the relative abundances of OTU275, OTU302 and OTU1222 were lower in rhizosphere soils treated with vanillic acid than in rhizosphere soils treated with water (P < 0.05). For Pseudomonas spp., the relative abundance of OTU1587 were higher in rhizosphere soils treated with vanillic acid than in rhizosphere soils treated with water (P < 0.05).

Figure 2.
Cladograms, generated from LEfSe analysis, represent the polygenetic distribution of cucumber rhizosphere soil bacterial taxa. LEfSe analysis was based on the data of three independent replicates of each treatment. Bacterial taxa that are significantly enriched in each treatment with LDA scores larger than 2.0 are shown. Significantly discriminant taxon nodes are colored: red for cucumber rhizosphere soils treated with water (W), green for vanillic acid at 0.05 μmol g −1 soil (VA). Yellow circles represent non-significant differences in abundance between treatments for that particular taxon. Each circle's diameter is proportional to the taxon's abundance. Labels are shown of the phylum, class and order levels. The LDA scores of each significantly discriminant taxon from the phylum to genus levels are shown in Figure S2.  Figure S3a). Principal component analysis also could not clearly separate the five treatments from each other (Fig. 5a). Number of bands, Shannon-Wiener index and evenness index were not affected by exogenously applied vanillic acid ( Figure S4). For Pseudomonas spp. community, visual inspection of the DGGE profiles revealed different banding patterns among treatments ( Figure S3b). Principal component analysis clearly distinguished the five treatments from each other (Fig. 5b). Compared with rhizosphere soils treated with water, rhizosphere soils treated with 0.2 μmol g −1 soil vanillic acid had lower number of bands, and rhizosphere soils treated with 0.02 μmol g −1 soil vanillic acid had higher Shannon-Wiener index and evenness index ( Figure S4).
Bacillus and Pseudomonas spp. Community Abundances. Quantitative PCR revealed that all concentrations of vanillic acid (0.02, 0.05, 0.1, 0.2 μmol g −1 soil) significantly decreased rhizosphere Bacillus spp. community abundance (P < 0.05) (Fig. 6a). The lowest Bacillus spp. community abundance was observed in rhizosphere soils treated with 0.2 μmol g −1 soil vanillic acid. Vanillic acid at 0.02, 0.05 and 0.1 μmol g −1 soil did not significantly affect cucumber rhizosphere Pseudomonas spp. community abundance (Fig. 6b). Vanillic acid at 0.2 μmol g −1 soil significantly decreased rhizosphere Pseudomonas spp. community abundance (P < 0.05). Bacillus and Pseudomonas spp. community abundances in rhizosphere soils treated with water was 19 and 1.76 times of these in rhizosphere soils treated with 0.2 μmol g −1 soil vanillic acid, respectively.  Table 1. The most abundant bacterial OTUs in cucumber rhizosphere soils treated with water (W) or vanillic acid at 0.05 μmol g −1 soil (VA). OTUs were delineated at 97% sequence similarity. Only OTUs with average relative abundances >0.5% in at least one treatment were presented. Values were expressed as mean ± standard error (n = 3). OTU ID in bold indicates its relative abundance was significant different between treatments according to Welch's t test (P < 0.05, Bonferroni corrected).

Discussion
Plant-derived carbon metabolites, released through root exudation and residue composition, can be assimilated by soil microorganisms, the growths of which are usually limited by carbon resources 25 . It was observed that phenolic acids were rapidly decomposed by soil microorganisms after entering the soil 26,27 . Therefore, vanillic acid was applied into the soil periodically as described before 26 . Generally, the bioactivity of toxic compounds on plants and microorganisms was concentration dependent 28 . Previously, we found that the concentration of vanillic acid in cucumber rhizosphere was about 0.05 μmol g −1 soil in a continuous monocropping system 22 . In both natural and agricultural ecosystems, the concentration of soil phenolic acids was shown to range from 0.01 to 0.5 μmol g −1 soil [29][30][31] . Therefore, concentrations of vanillic acid used in this study (0.02 to 0.2 μmol g −1 soil) were within the realistic range of concentrations in the soil reported before. Blum et al. 32 found that phenolic acids inhibited cucumber seedling growth but stimulated the rhizosphere phenolic acid-utilizing bacteria. We also demonstrated that vanillic acid at concentration ≥0.05 μmol g −1 soil inhibited cucumber seedling growth and increased rhizosphere bacterial community abundance 23 . In this study, high-throughput amplicon sequencing of bacterial 16S rRNA gene was used to further illustrate the taxonomies of these responsive bacteria. Results showed that relative abundances of some bacteria such as Rhodanobacter, Steroidobacter and Ohtaekwangia spp., which have been reported to be involved in phenolic compounds degradation [33][34][35] , were enriched in cucumber rhizosphere soils treated with vanillic acid. Future in vitro studies should focus on validating the capabilities of these microorganisms to metabolize vanillic acid.
In continuous monocropping systems, plant usually showed reduced growth and vigor, and increased disease index 5,36 For example, continuous monocropping of cucumber stimulated the abundance of Fusarium oxysporum f.sp. cucumerinum, the causing agent of cucumber Fusarium wilt disease 4 . Illumina MiSeq sequencing showed that, compared with rhizosphere soils treated with water, rhizosphere soils treated with vanillic acid (0.05 μmol g −1 soil) had higher relative abundances of Bacillus, Archangium and Lysobacter spp., which contained species to promote plant growth and inhibit plant pathogens 19,37,38 . Quantitative PCR also found that all concentrations of vanillic acid (0.02 to 0.2 μmol g −1 soil) decreased cucumber rhizosphere Bacillus spp. abundance. Some species in Bacillus spp. are able to inhibit plant soil-borne pathogens, including Fusarium oxysporum f.sp. cucumerinum, and promote cucumber growth 19 . Therefore, decreases in cucumber rhizosphere plant-beneficial microorganisms induced by vanillic acid may contribute to increased soil-borne diseases in the continuous monocropping system.
In this study, vanillic acid increased the relative abundances of Arenimonas 39 , Gemmatimonas 40 , Haliangium 41 , Opitutus 42 , Pseudolabrys 43 , Steroidobacter 44 and Rhodanobacter spp. 45 , which contained taxa with denitrification capabilities, dissimilatory reduction of nitrate to nitrous oxide and N 2 . Meanwhile, vanillic acid decreased the relative abundance of Nitrospira spp. 46 , which had nitrification capabilities, oxidation of ammonia to nitrite. These results validated previous studies showing that phenolic acids were able to inhibit nitrification 47 and some denitrifiers were able to use phenolic acids as carbon sources 48 . Nitrogen is one of the major limiting elements in agricultural ecosystems 49 . Our results indicated that there was possibility that phenolic acids could influence plant growth through regulating rhizosphere nitrogen transformations, such as inhibiting nitrification and promoting denitrification.
Illumina MiSeq sequencing and PCR-DGGE analyses showed that vanillic acid changed the structure and composition of cucumber rhizosphere Pseudomonas spp. community. Illumina MiSeq sequencing revealed that vanillic acid also changed the composition of cucumber rhizosphere Bacillus spp. community. Moreover, quantitative PCR analysis showed that vanillic acid decreased Bacillus spp. community abundance at 0.02 to 0.2 μmol g −1 soil, and decreased Pseudomonas spp. community abundance at 0.2 μmol g −1 soil. It has been shown that agricultural intensification could negatively influence soil Pseudomonas and Bacillus spp. communities 16,20,21,50 . For example, continuous monocropping of Radix pseudostellariae changed the community structure and decreased the abundance of soil Pseudomonas spp. 20 . Continuous monocropping of Helianthus tuberosus changed soil Pseudomonas and Bacillus spp. community structures 51 . It is well known that continuous monocropping can lead to accumulation of autotoxins, including phenolic acids, in the soil 6,22,48 . Thus, accumulation of phenolic autotoxins may be linked the effects of continuous monocropping on Pseudomonas and Bacillus spp. communities.
Plants can influence the diversity and composition of rhizosphere microbial communities through releasing root exudates 10 . Generally, rhizosphere microbial communities have higher abundances but lower diversities than those of the bulk soil 8 . It has been observed that artificially applied phenolic acids can alter the composition and diversity of soil microbial communities in absence of host plants 10,15 . Phenolic autotoxins can damage plant root and enhance ion leakage 52 and these changes may also affect rhizosphere microbial communities. Therefore, besides its direct effects, vanillic acid may also indirectly changed cucumber rhizosphere microbial communities through its effects on physiological status of cucumber. These also suggest that rhizosphere and bulk microbial communities may respond differently to phenolic acids, which need to be further elucidated.
Phenolic acids are ubiquitous secondary metabolites in plants 2,6 . However, the composition of phenolic acids in root exudates or rhizosphere soils differed among crop species 2,7,22,31 . Liu et al. 15 found that benzoic acid, a phenolic compound found in peanut (Arachis hypogaea) root exudates, increased the relative abundance of Burkholderia spp. in soil. However, vanillic acid did not affect the relative abundance of Burkholderia spp. in this study. These indicated that different phenolic acids may have different effects on soil microbial communities. Previous studies have also revealed the structure-function relationships of the phytotoxic and antimicrobial activities of phenolic acids 6,53 . For example, cinnamic acid derivatives had higher inhibitory effects than their corresponding benzoic acid derivatives on cucumber seedlings 54 . Phenolic acids with the hydroxyl group in the position para to the carboxyl side chain, such as benzoic acid, can be easily metabolized by the maize pathogen Cochliobolus heterostrophus; while phenolic acids lacking the para-hydroxy group or their para-methoxy substituted derivatives, such as ferulic acid, have higher antifungal activity 53 . These evidences may help to explain phenolic acids act as autotoxins of several crops though all plants releases phenolic acids.

Conclusion
Overall, our results revealed that vanillic acid changed cucumber rhizosphere total bacterial, Pseudomonas and Bacillus spp. community compositions, and Pseudomonas and Bacillus spp. community abundances. In particular, Illumina MiSeq sequencing showed that vanillic acid at 0.05 μmol g −1 soil inhibited the relative abundances rhizosphere microorganisms with plant-beneficial potentials. Quantitative PCR analysis showed that vanillic acid decreased Bacillus spp. community abundance at 0.02 to 0.2 μmol g −1 soil, and decreased Pseudomonas spp. community at 0.2 μmol g −1 soil. It is clear that phenolic acids are toxic to plants 2,6,7,52 . Results obtained from the present study suggested that phenolic acids may also inhibit plant growth through changing rhizosphere microbial communities, which need to be further explored by evaluating the effects of phenolic acids on the functions of rhizosphere microbial communities and their relationships with plant performance.

Materials and Methods
Pot Experiment. The soil used in this experiment was collected from the upper soil layer (0-15 cm) of an open field in the experimental station of Northeast Agricultural University, Harbin, China (45°41′N, 126°37′E), which was covered with grasses and undisturbed for more than 15 years. The soil has a sandy loam texture, contained organic matter, 3.67%; available N, 89.02 mg kg −1 ; Olsen P, 63.36 mg kg −1 ; available K, 119.15 mg kg −1 ; EC (1:2.5, w/v), 0.33 mS cm −1 ; and pH (1:2.5, w/v), 7.78. Cucumber seedlings (cv. Jinlv 3) with two cotyledons were transplanted into pots contained 150 g soil and maintained in a greenhouse (32 °C day/22 °C night, relative humidity of 60-80%, 16 h light/8 h dark). There was one cucumber seedling per pot. No fertilizer was added to the soil.
Cucumber seedlings at the one-leaf stage were treated with different concentrations of vanillic acid (0.02, 0.05, 0.1, 0.2 μmol g −1 soil) every two days for five times as recommended before 26 . The solution pH was adjusted to 7.0 with 0.1 M NaOH solution, because the soil pH is widely accepted as a dominant factor that regulates soil microbial communities55. Cucumber seedlings treated with distilled water were served as the control. Soil water content was adjusted every two days with distilled water to maintain a constant weight of pots. There were five treatments (four concentrations of vanillic acid and one control) in total. Each treatment had five seedlings and was replicated three times.
Rhizosphere Soil Sampling and DNA Extraction. Ten days after the first application of vanillic acid, cucumber rhizosphere soil samples were collected as described before 5 . Briefly, cucumber roots were gently removed from the pot, and soils loosely attached to cucumber roots were carefully removed by manual shaking. Then, soils tightly adhering to roots were removed from the root surface by a sterile brush and considered as rhizosphere soils. After sieving (2 mm), rhizosphere soil samples were stored at −70 °C. Samples from five plants in each replicate were combined to make a composite sample. There were three composite rhizosphere samples for each treatment.
Total soil DNA was extracted with the PowerSoil DNA Isolation Kit (MO BIO Laboratories, Carlsbad, USA) as per the manufacturer's instructions. Each composite soil sample was extracted in triplicate and the extracted DNA solutions were pooled.
High-throughput Amplicon Sequencing and Data Processing. Total rhizosphere soil bacterial community compositions were estimated with high-throughput sequencing on an Illumina MiSeq platform. Primer set of F338/R806 was used to amplify V3-V4 regions of the bacterial 16S rRNA gene as described before 4,56 . Both the forward and reverse primers also had a six-bp barcode unique to each soil sample. The PCR protocol was: 95 °C for 3 min; followed by 27  reactions were pooled and purified using an Agarose Gel DNA purification kit (TaKaRa, China). Then, purified amplicons were quantified by a TBS-380 micro fluorometer with Picogreen reagent (Invitrogen, USA), and mixed accordingly to achieve the equal concentration in the final mixture. The mixture was then paired-end sequenced (2 × 300) on an Illumina Miseq platform at Majorbio Bio-Pharm Technology Co., Ltd., Shanghai, China.
Raw sequence reads were de-multiplexed, quality-filtered, and processed using FLASH 57 as described before 4 . Operational taxonomic units (OTUs) were delineated at 97% sequence similarity with UPARSE using an agglomerative clustering algorithm 58 . Then, a representative sequence of each OTU was taxonomically classified through BLAST in Ribosomal Database Project (RDP) database 59 . Chimeric sequences were identified and removed using USEARCH 6.1 in QIIME 60 . To avoid potential bias caused by sequencing depth, a random subsampling effort of 24,245 16S rRNA gene sequences per sample was performed for further analysis. The data set was deposited in the NCBI-Sequence Read Archive with the submission Accession Number SRP119631.
PCR-DGGE Analysis. Semi-nested PCR protocols were used to amplify Pseudomonas and Bacillus spp. 16S rDNA fragments. Primer sets of PsF/PsR and GC-338F/518 R were used for the first and second round of PCR amplification of Pseudomonas spp. 16 , respectively; while BacF/BacR and GC-338F/518 R were used for Bacillus spp. 21  Standard curves were made with a 10-fold dilution series (10 2 -10 8 ) of plasmids containing 16S rRNA genes of Pseudomonas and Bacillus spp. from soil samples. Sterile water was used as a negative control to replace the template. All amplifications were done in triplicate. The specificity of the products was confirmed by melting curve analysis and agarose gel electrophoresis. The threshold cycle (Ct) values obtained for each sample were compared with the standard curve to determine the initial copy number of the target gene.
Statistical Analysis. For Illumina Miseq sequencing data, the defined OTUs were used to calculate taxon accumulation curves with the 'vegan' package in 'R' 62 . Alpha diversity indices, Chao, ACE, Shannon index and inverse Simpson index were calculated using QIIME 60 . For beta diversity analysis, weighted UniFrac distances and Bray-Curtis distances were calculated using QIIME 60 and 'vegan' package in 'R' 62 , respectively. Principal coordinates analysis was conducted to visualize the community similarity with the 'vegan' package in 'R' 62 . Linear discriminant effect size (LEfSe) analysis was used to identify microbial taxa that were significantly associated with each treatment with an alpha value of 0.05 for the Kruskal-Wallis test and a threshold of 2.0 for logarithmic linear discriminant analysis (LDA) scores 63 . Differences in relative abundances of microbial taxa between treatments were analyzed using Welch's t test with Bonferroni correction in 'STAMP' 64 .
The DGGE profiles banding patterns were analyzed with Quantity One V4.5 (Bio-Rad Lab, LA, USA). Principal component analysis was used to compare the band patterns between samples with Canoco for Windows 4.5 software (Plant Research International, Wageningen, the Netherlands). The microbial community diversity indices, including number of bands, Shannon-Wiener index and evenness index, were calculated as described before 5 .
Data were analyzed by analysis of variance (ANOVA). For alpha diversity indices from Illumina Miseq sequencing, mean comparison between treatments was performed based on the Welch's t test at the 0.05 probability level. For diversity indices from the PCR-DGGE analysis, mean comparison between treatments was performed based on the Tukey's honestly significant difference (HSD) test at 0.05 probability level.