Genotype May Influence Bacterial Diversity in Bark and Bud of Vitis vinifera Cultivars Grown under the Same Environment

Viticulture is globally an important economic activity, and grapevine microbiomes hold a significant role in influencing yield and quality. Earlier studies showed that cultivar and agronomic management affect grapevine microbiome structure and, potentially, the quality of the end product. While microbial dynamics and ecology were established on some grapevine tissues, i.e., leaves and grapes, there is less knowledge deciphering microbiomes on other tissues, i.e., barks and buds. Moreover, although the impact on the microbiome of the so-called “vitivinicultural terroir” is well established, there are limited data considering microbiomes of genetically diverse cultivars within the same environment. Our study aims to explore microbiome diversity on bud and bark tissues of 37 different grapevine cultivars under the same environment and agronomic management. We targeted the V2-9 regions of the 16S rRNA gene of the microbiomes in bark and buds at the onset of new vegetation and bud expansion using Ion Torrent sequencing technology. Our results show that these tissues display high bacterial diversity regardless of cultivars’ use. Proteobacteria, Bacteroidetes, and Actinobacteria were the most prevalent among 11 detected phyla. The genotype of the cultivar seems to affect bacterial diversity and structure (p < 0.001) within the same environment. Our approach highlights the efficiency of high-throughput sequencing to unfold microbiomes of several grapevine parts that could be an important source of microbial inoculation and an important molecular fingerprint of the wine and grape end products.


Introduction
Viticulture is an economically important crop spanning an area of 7.5 mha with a total production exceeding 78 × 10 6 tons and a revenue of more than 30 billion Euros of wine [1].
Therefore understating the dynamics of product quality are of great economic importance [1]. Due to their abundance and activity, Vitis vinifera associated microbes are key in regulating growth, plant health, and grape and wine quality [2][3][4]. Even beneficial microbes in wine production, i.e., lactic acid bacteria, can cause undesirable changes in wine flavor under certain conditions, rendering the wine undrinkable. Similarly, several associated bacteria can cause some serious wine spoilage [5]. Despite the great progress in describing grapevine microbiomes and their implications on their growth, yield, and product quality, little is known about how different cultivars (as microbial hosts) can shape associated microbial assemblage.

DNA Extraction
Frozen plant samples were mixed vigorously (10 min at max speed) with 25 mL of TENP buffer. The supernatant (20 mL) was collected by centrifugation (3000× g; 5 min) in a new 50 mL tube. The supernatant was centrifuged at 14,500 rpm for 10 min and the pellet was collected and stored at −80 • C for DNA extraction [11]. DNA was extracted using a NucleoSpin ® Plant II kit (Macherey-Nagel GmbH, Dylan, Germany) according to the manufacturer's instructions. All DNA samples were kept at −20 • C for further use. Samples DNA quantity and quality were assessed by NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Inc., Walsham, MA, USA) and visualized in 1.0% agarose gel electrophoresis.

High-Throughput Sequencing and Statistical Analysis
Samples were analyzed for bacterial taxonomy using the Ion 16S™ Metagenomics Kit according to the manufacturer's instructions (#A26216, Thermo Fisher Scientific, Inc., Walsham, MA, USA). Briefly, the 16S rDNA metagenomic library consisted of pooled and tagged (Ion Xpress Barcode Adapters, #4471250, #4474009, #4474518, Thermo Fisher Scientific, Inc., Walsham, MA, USA) amplicons targeting V2-4-8 and V3-6, 7-9 sets; the combination of the two primer pools allows for sequence-based identification of broad range of bacteria within a mixed population [27]. Briefly, 3 ng of high-quality DNA was amplified using 2X Environmental Master Mix with the provided 10X 16S Primer Sets (Thermo Fisher Scientific, Inc., Walsham, MA, USA). PCR products were purified with AMPure ® XP beads solution (#A63881, Beckman Coulter, Breia, CA, USA), quantified, pooled, and end-repaired using Ion Plus Fragment Library Kit (#4471252, Thermo Fisher Scientific, Inc., Walsham, MA, USA). Then, amplicons were adapter-ligated, nick repaired, quantified using an Ion Universal Library Quantitation Kit (#A26217, Thermo Fisher Scientific, Inc., Walsham, MA, USA), and the library was pooled and diluted to the required concentration. The library template was prepared using Ion 520™ & Ion 530™ Kit-OT2 (#A27751, Thermo Fisher Scientific, Inc., Walsham, MA, USA) and was loaded on a Ion 520™ Chip (#A27762, Thermo Fisher Scientific, Inc., Walsham, MA, USA). Sequencing was done at the Plant Genetics and Breeding Lab of AUTH on a GeneStudio S5 platform using Ion Torrent™ sequencing technology (Thermo Fisher Scientific, Inc., Walsham, MA, USA). Sequencing progress and raw sequences were processed through the Ion Torrent Suite (ver. 5.12.2, Thermo Fisher Scientific, Inc., Walsham, MA, USA), then BAM files were uploaded to the Ion Reporter Platform to obtain Ion 16S Metagenomics report through Ion 16S Metagenomics Workflow (Thermo Fisher Scientific, Inc., Walsham, MA, USA). Ion 16S Metagenomics Workflow uses a premium curated Applied Biosystems™ MicroSEQ™ ID 16S rRNA database and a curated Greengenes database, that are not user customizable (i.e., thresholds and cut-off values). Therefore, Fastq files (QC-, adaptor-, and barcode trimmed) were exported using the FileExporter plugin through Ion Torrent Suite and uploaded on the MG-RAST server [28] under the project number mgp91606. After MG-RAST QC (dynamic error removal (DRISEE), dynamic trimming (DYNAMICTRIM), denoising and normalization (FASTQ-MCF); see details on MG-RAST under Processing Information per submitted sample) taxonomy was assigned using Greengenes database (v.13_8) at 97% identity (e value 10 −5 ). The resulting taxonomy matrix (OTUs Table; not shown) was used for all subsequent analysis. Descriptive diversity indices (α-diversity, rarefaction curves) and statistics (PCoA, Clustering based on Bray-Curtis and UPGMA, PERMANOVA, and ANOSIM) were calculated using PAST3 software [29]. Venn diagrams were carried out according to Heberle et al. [30].

Sequence Analysis
High-throughput sequencing was performed using an Ion Torrent platform GENE STUDIO S5 at the Laboratory of Plant Genetics and Breeding, School of Agriculture, Aristotle University of Thessaloniki, Greece. Seven hypervariable regions (V2, V3, V4, V6, V7, V8, and V9) of the 16S rRNA gene were amplified and sequenced. Sequencing yielded 3,836,209 reads in total, with an average of 15,775 classified sequences per cultivar with an average read length of 234 bp. Accordingly, V3 provided the greatest sequencing depth that is the number of reads generated per sample (33% of total reads), followed by V6 and 7 (20%), V4 (18%), V8 (18%), V2 (10%), and, V9 (>0.1%) for profiling the bacterial diversity. The proportion of the unclassified reads ranged from the lowest of Limniona (60%) to the highest of Mavro (97%) with an average of 84% and were excluded from further analyses. Though the rarefaction curve was not parallel with the x-axis, the Good's coverage of bacterial diversity reached 99.9%, with the majority of microbial diversity being captured ( Figure S1

Microbial Diversity among Cultivars
Among the 11 phyla present in 37 cultivars, Proteobacteria, Bacteroidetes, and Actinobacteria were the most dominant, comprising 99% of the total OTUs ( Figure 2). The proportion of the aforementioned phyla varied among the cultivars, ranging from 94% for Mavro to 100% for Karabraimis, with an average of 99% ( Figure 2).  Microbial diversity indices varied among cultivars. Richness (Chao1) scores ranged among the cultivars from 183 to 544 (Table 2). Specifically, Cabernet Sauvignon held the lowest and Korinthiaki the highest Chao1 score. Regarding OTUs, the highest value was obtained by Limniona (266 and 164), both including and excluding singletons, respectively. The lowest values were shown by Cabernet Sauvignon with singletons (88), and by Grenache (37) excluding singletons ( Table 2). Shannon diversity ranged from 1.2 (for Syrah) to 3.1 (for Mavrokorakas). Simpson's evenness index ranged from 0.45 to 0.92 for Savvatiano and Mavrokorakas, respectively.      The first two PCoA axes (1 and 2) explain 35% of the total variation ( Figure 3). Six groups were separated by hierarchical clustering (Figure 4). Significant differences in the microbial community structure were confirmed by PERMANOVA (p < 0.001 and F = 6.617) and ANOSIM (p < 0.001 and R = 0.6964) statistical analyses. Based on Venn diagram analyses (data not shown), 26 of the 37 cultivars showed the presence of cultivar-specific bacterial genera. The abundance of cultivar-specific genera ranged from 1-13 (Table 3), forming six distinct groups of cultivars. These six groups comprise 64 bacterial genera; interestingly, each cultivar had a unique bacterial fingerprint. While the remaining 11 cultivars from one group share 140 bacterial genera (Table 3).       Of the total 204 bacterial genera identified, the following eleven: Curtobacterium, Fischerella, Frigoribacterium, Hymenobacter, Kineococcus, Massilia, Pantoea, Pseudomonas, Pedobacter, Rathayibacter, and Sphingomonas were present in all cultivars, except the Riesling. In the latter, the Pantoea genus was absent. The above-mentioned 11 genera are considered as a core microbiome. The relative abundance of this core microbiome ranged from 68-99%, with Limniona harboring the lowest and Robola the highest, respectively ( Figure 5). Appl. Sci. 2020, 10, x FOR PEER REVIEW 3 of 3

Assessment of Vine Bacteria of Economic Importance
The study revealed well-known bacteria with economic implications for viticulture and winemaking. Of all the studied cultivars, only seven cultivars presented such bacteria. Agrobacterium vitis and A. tumefaciens, which are responsible for crown and cane gall disease, were detected in Tsaousi and Riesling cultivars. Riesling cultivar harbored both A. vitis and A. tumefaciens, with the former presented in a 12-fold species abundance compared to Tsaousi, while Tsaousi contained only A. tumefaciens. Lactic acid bacteria (LAB), including Lactobacillus and Pediococcus, and acetic acid bacteria (AAC), such as Gluconobacter associated with wine production, were detected in low abundance. Specifically, Lactobacillus sp. was observed in Vidiano, Limniona, Voidomatis, and Kakotrygis, and Pediococcus spp. were only discovered in Mavro, while Gluconobacter was observed only in Serifiotiko cultivar.

Discussion
The aim of this study was to investigate the roles that Vitis vinifera cultivars might play, as microbial hosts, in shaping the differentiation of associated microbiomes. The hypothesis in the study was that if there is any influence of the genotype in the bacterial assemblage that overwinter on the vine canes, this should be evident as differences in the bacterial populations detected early in the spring on vines of different genotypes in the same vineyard, same soil and environment, and cultivated with the same agricultural practices. Here, we specifically focused on one-season-old canes' barks and buds, which to our knowledge, no other study compared their associated microbiomes yet. High-throughput sequencing revealed rich microbiomes on these parts from several cultivars, supporting that trunk bark is an important microbial host (microbial terroir [7]). Interestingly, cane samples hosted several taxa that are typically identified in many studies across the world. On the other hand, a great portion of our sequencing data revealed unclassified microbial sequences. However, it is not uncommon for sequencing studies to result in high numbers of valid but unclassified reads. It could be due to experimental noise and richer microbiomes with some organisms less well described in the different databases [31]. Also, the differences in the microbiomes of the studied cultivars partly support the idea that vineyards consisting of different grape cultivars appear to harbor more diverse microbiomes. A possible selection pressure imposed on the microbiome by grapevine genotype might be one explanation, as indicated in other studies [4].
Patterns of microbiome divergence among cultivars became evident, and the majority of Vitis microbiomes grouped close to the lower part of the plot (Figure 3). Interestingly, Vertzami, Cardinal, Soultanina, and Syrah cultivars formed a distinct group. Similar patterns of microbe divergence were also detected in other studies with different cultivars [4,62]. Moreover, the results of Venn diagrams revealed that Limniona hosted unique bacterial genera ranging from 40 to 72% of the total taxa when compared to Victoria and Grenache, respectively. Considering free-living microbes, it is well documented how geography and environment shape microbial community structure. In vineyard ecosystems it was recently demonstrated that microbial variability correlates with the geographic location of the cultivars, environmental characteristics and agronomic systems [63,64]. In a recent biogeographic investigation, geographic distance and environment appear to be important factors shaping Vitis associated Saccharomyces cerevisiae diversity [65]. In our study, geographic distance was not a factor controlling microbial diversity, because all cultivars were from the same vineyard. Interestingly, the genotype of the cultivars appeared to have a drastic impact on the bacterial community, influencing the fitness of certain V. vinifera associated microbes.
The detected genera (Hymenobacter, Massilia, Frigoribacterium, Sphingomonas, Pseudomonas, Pedobacter, Pantoea, etc.) in the vineyard are known to have distinct beneficial and pathogenic roles. For instance, among Pseudomonas spp., Ps. putida had a presence of 11% among the other species of this genus and is known for its growth-promoting role [66,67]. Moreover, Ps. fluorescens (12%) and Ps. aeruginosa (0.02%) suppress different pathogens [68,69]. However, other Pseudomonas spp. such as Ps. viridiflava (15%) and Ps. syringae (0.4%) [68,69] are known for their pathogenicity. Only Massilia timonae was detected in this study, which is known for its plant-growth-promoting and protection in the rhizosphere and seed surface [70]. Massilia taxa have been reported in the phyllosphere of different plants, including lettuce and apple [71][72][73][74][75]. Moreover, the genus Sphingomonas generally acts as a plant-protective by suppressing disease symptoms and decreasing pathogen growth [53]. Researchers have demonstrated that the leaf bacterium Sphingomonas spp. can protect plants against the leaf-pathogenic Ps. syringae through substrate competition [74]. Also, Sphingomonas spp. are known for their plant growth-promoting through the production of plant growth-stimulating factors [75]. However, S. melonis was described as the causative agent of the brown-spot disease of yellow Spanish melons (Cucumis melo) [76] and was also detected with a presence of 97% among the other species of this genus. Furthermore, members of Pantoea act as biocontrol agents or plant-growth promoters [75], including Pantoea agglomerans (95%). This species works against different pathogens like Erwinia amylovora, causing fire blight disease of pear and apple fruit trees [77]. Some economically important bacterial genera such as Agrobacterium, Acetobacter, Gluconobacter, Pediococcus, Lactobacillus spp. were observed in trivial abundance. Although no observable effect is anticipated during this stage of the grapevine lifecycle, they might later be inoculants for the grapevine itself or its products, causing economic losses if the conditions are suitable. From our study, it is evident that cane tissues host both beneficial and pathogenic taxa, with important economic implications in viticulture. Thus, the dynamics of beneficial and pathogenic taxa through seasonal grapevine growth remain unclear. Future studies should investigate V. vinifera microbiomes at a greater temporal resolution to better understand the control and factors affecting the trade-off balance between beneficial and pathogenic taxa in vineyards.
Results of this study may be helpful to investigate the origin of the grapevine associated microbiomes in other tissues. Previous studies indicated that grapevine microbiomes on leaves and grapes originated primarily from the adjacent soil microbiome [6,7,37,39,47,78]. This could be due to the direct dissemination of soil particles or any other form of passive transportation, e.g., displacements of plant surface. Another concept of microbial inoculation considers direct inoculation of various vine parts through buds. Trunk bark was also considered as a permanent habitat and source of inoculation in a vineyard [6,7,79]. Hence, the buds, which are responsible for the annual growth cycle of grapevines [80], together with the barks of the one-season-old canes, could be the direct source of inoculants because they are in closer contact with leaves, inflorescences, and grapes than any other sources. The origin and mode of inoculation require further investigation and could potentially help us understand the importance of the local environment and cultivar selection in controlling microbial dynamics on V. vinifera cultivars.

Conclusions
Metagenomics mining revealed that a core microbiome is hosted by all studied V. vinifera cultivars conserved in the same environment. On the other hand, unique metagenomics fingerprints were unraveled for each cultivar, suggesting a role of cultivars' genotype in shaping the associated microbial community. The vine microbiome's complexity and diversity were highlighted, although a percentage of sequence data still remains unassigned. Thus, further microbial metagenomics research is vital for complete coverage of the microbial genome available in vineyards. The host-associated microbes require further investigation in greater temporal and spatial patterns to assess the importance of grapevine microbiomes to delineate the wines regional character and provide further tools for sustainable agriculture.