The transcriptionally active bacterial communities of grapevine rhizosphere in dependence on rootstock and scion variety

The rhizosphere is where crucial processes for the productivity of viticultural systems occur. The composition of the bacterial communities associated with the rhizosphere of grapevines is known to depend on plant genotype. However, the genotype of grafted grapevines differs between scion and rootstock; the role of each genotype is unclear. To untangle the effect of scion and rootstock, the rRNA (V4–V5 region of 16S rRNA) extracted from the rhizosphere of the grape varieties Riesling and Mueller-Thurgau ungrafted vs grafted on different rootstocks was sequenced. The bioinformatic analysis with tools designed to be robust for compositional data showed that the investigated rootstocks or scions or combinations, respectively, recruited bacterial communities with distinguishable traits. Statistical differences were revealed between ungrafted Riesling vs Mueller-Thurgau, between grafted Riesling vs ungrafted Riesling, and between ungrafted Mueller-Thurgau vs grafted Mueller-Thurgau. Thus, confirming the role of scion and rootstock genotype as a driver of the structure and composition of bacterial communities in the rhizosphere of grapevines.


INTRODUCTION
Plants are colonised below and above ground by a variety of microbes that serve their mutualistic benefits. The rhizosphere microbiome is often described as a positive interaction between plants and microorganisms (Ryan et al., 2009;Taye et al., 2019). In particular, the microbiota associated with the plant rhizosphere is involved in important processes such as growth modulation, defence responses, and nutrient uptake (Berendsen et al., 2012;Durán et al., 2018;Hu et al., 2018;Li et al., 2020;Schlaeppi and Bulgarelli, 2015;Zhang et al., 2019). To date, the majority of rhizobiome studies has focused on model plants such as Arabidopsis thaliana (Alegria Terrazas et al., 2016) or annual crop plants such as barley (Bulgarelli et al., 2013), canola, wheat, pea, and lentil (Cordero et al., 2020), oilseed rape (Etesami and Alikhani, 2016) and maize (Peiffer et al., 2013). However, it is also remarkably important to consider the rhizosphere microorganisms of perennial plants, such as grapevines (Marasco et al., 2018), for supporting crop growth, especially under difficult conditions (Timmusk et al., 2014).
Microbial communities associated with grapevines and wine, respectively, have already been extensively studied (Burns et al., 2016;Hendgen et al., 2018;Holland et al., 2014). In vineyard soil, the microbial communities have been described, for example, as a function of spatial distribution or management practices (Bokulich et al., 2014;Hendgen et al., 2018;Holland et al., 2014;Vega-Avila et al., 2015). Most of these studies were based on soil samples taken near the grapevines, not on veritable rhizosphere samples. Nonetheless, the rhizosphere is the location where crucial processes for the productivity of agricultural systems take place, mediated by microorganisms. Therefore, the processes and microorganisms in the rhizosphere need to be considered to study the direct interaction between the grapevine root and the surrounding soil.
The choice of grapevine rootstock variety impacts the microorganisms in the grapevine rhizosphere (Berlanas et al., 2019;Marasco et al., 2018). Since the 19th century, Vitis vinifera cultivars have been grown as scions grafted onto Phylloxera-tolerant Vitis sp. rootstocks. However, understanding the mechanisms underlying grafted vines and the interactions between scion and rootstock is still beginning (as reviewed by Gautier et al., 2019). For future research, the selection of the rootstock is, thus, inevitable, also with regard to its microbial community (Zarraonaindia et al., 2015). Vink et al. (2021) investigated differences in the microbial communities in the rhizosphere of an 11-year-old grapevine of four scion cultivars and four rootstock types regarding alpha and beta diversity indices, concluding that bacterial diversity is affected by both scion and rootstock variety. However, this effect depends on the diversity measures and the specific rootstock-scion combinations considered (Vink et al., 2021). Further studies on this specific topic are needed (Berlanas et al., 2019). Therefore, a study was conducted under controlled conditions to reduce the variability resulting from all factors except the genotype of the rootstock and scion with potted grapevines. In three different experimental designs, the rhizobiome of two ungrafted Vitis vinifera varieties (Riesling and Mueller-Thurgau) were compared with varieties of Riesling and Mueller-Thurgau grapevines grafted on different rootstocks. A metabarcode analysis was performed on the extracted RNA (Carvalhais et al., 2012;Garoutte et al., 2016;Turner et al., 2013) to get a better insight into the active microbiome of the grapevine rhizosphere. To the best of our knowledge, no compositional data analyses of 16S rRNA data of the grapevine rhizosphere microbiome with different scions and rootstock cultivars have been carried out before. Data derived from high-throughput sequencing of biological samples must be considered as compositions rather than counts, as ratio-based analyses can lead to qualitatively incorrect conclusions (Fernandes et al., 2014;McLaren et al., 2019;Quinn et al., 2019). The effect of the grape variety and rootstock, respectively, on the bacterial communities in the rhizosphere could be investigated due to the controlled experimental conditions.

Plant material
The independent experiments were located at Hochschule Geisenheim, University, Germany. The grapevines were planted in pots (15 cm

Rhizosphere sampling
Rhizosphere sampling was performed at three different time points for the three experimental setups: May 2019, October 2019, and October 2020. A total of four Riesling, four Mueller-Thurgau, and four controls were sampled in May 2019. Five samples per grapevine rootstock were collected in October 2019. Additionally, nine controls, nine ungrafted Riesling and Mueller-Thurgau, and three samples per grapevine rootstock were collected in October 2020. Only soil attaching to the roots was considered as rhizosphere soil for sampling and microbiome analyses. For controls ("No plant"), bulk soil was sampled in the same horizons of the pot as where the roots of the vines grow in the pots with a plant.

RNA extraction, Reverse-Transcriptase-PCR, and Ion Torrent Sequencing
RNA extraction from the rhizosphere soil and cDNA amplification were done according to Rosado-Porto et al. (2021). First Ion Torrent PCR was performed with a KAPA HiFi Polymerase kit (VWR International GmbH, Darmstadt, Germany), amplifying the partial sequence of the hypervariable regions (V4 and V5) of the 16S rRNA gene with the primer 520F (5'-AYTGGGYDTAAAGNG-3', (Claesson et al., 2009)) and 926R (5′-CCGYCAATTYMTTTRAGTTT-3', (Engelbrektson et al., 2010)). Amplification parameters were 3 min at 95 °C followed by 35 cycles at 98 °C for 20 s, 55 °C for 30 s, 72 °C for 30 s, and finally, 72 °C for 5 min. Second Ion Torrent PCR with primers, including barcodes and Ion Torrent sequencing adapters, was conducted with the PCR product from the first Ion Torrent PCR, as suggested by Berry et al. (2011). The PCR was performed with the following amplification parameters: 3 min at 95 °C followed by 10 cycles at 98 °C for 20 s, 55 °C for 30 s, 72 °C for 30 s, and finally, 72 °C for 7 min. The PCR products were applied to a 1 % agarose gel and subsequently purified using NucleoSpin® Gel and PCR Clean-up (MACHEREY-NAGEL GmbH & Co. KG, Düren, Germany). Further, this product was purified with DNA purification beads NucleoMagVR NGS clean-up kit (MACHEREY-NAGEL GmbH & Co. KG, Düren, Germany). The Ion Torrent Sequencing was done according to Kaplan et al. (2019).

Bioinformatic analysis of the sequencing data
Bioinformatic analysis was performed with QIIME 2 2020.11 (Bolyen et al., 2019). The obtained raw sequences were demultiplexed using cutadapt (Martin, 2011) with no errors allowed in the barcode sequences. Quality control, sequence denoising, clustering to amplicon sequence variants (ASVs), dereplication, and removal of chimaera sequences were conducted with DADA2 (Callahan et al., 2016) (via q2-dada2). Thereby, the first 15 base pairs (bp) were removed, and the sequences were cut to a length of 312 bp (May 2019), 317 bp (October 2019), and 312 bp (October 2020), respectively. Sequences that did not pass the quality control were removed. Therefore, two samples of 125AA (Timepoint October 2019) and one sample of SO4 (Timepoint October 2019) were not considered in the following analysis. Afterwards, the taxonomy was assigned as described previously by Dries et al. (2021a), and all ASVs belonging to chloroplasts and mitochondria were removed. All ASVs were aligned with time point grape variety (n) rootstock

Riesling (4) ungrafted
No plant (4) Control October 2019 Riesling (5)  SO4 Riesling (5)  125AA Riesling (5)  Boerner Riesling (5)  8B October 2020 Riesling (9)  ungrafted Riesling (3)  SO4 Riesling (3)  125AA Riesling (3)  Boerner Riesling (3)  8B Mueller-Thurgau (9) (Anderson, 2017), respectively. Beta diversity was measured using DEICODE (Martino et al., 2019) with a filtered beta diversity ordination file. For this purpose, all feature counts below 10 and all sample counts below 500 were removed. Beta diversity for all time points was visualised within a compositional biplot, displaying the eight most important features. From the ASVs shown as arrows in the DEICODE graphs, a more accurate taxonomic affiliation was done with a pairwise alignment at the online database EzBioCloud (Yoon et al., 2017). The distance matrices were also analysed by the PERMANOVA test, using 999 permutations. The significance of differential abundance was determined using ALDEx2 (Fernandes et al., 2013). Therefore, all feature frequency counts below 10 and all sample counts below 2 were removed.

Bacterial alpha diversity across the different experimental set-ups
The 20 dominating bacterial families for the three-time points are shown in Figures 1 and 2. Control soil showed clear differences in the 20 dominating bacterial families compared to the ungrafted Mueller-Thurgau and Riesling ( Figure 1A). In the rhizosphere of Riesling grapevine grafted onto four different rootstocks (SO4, 125AA, Boerner, 8B), the dominating bacterial families in the rhizosphere showed no clear differences of the 20 dominating families comparing the four rootstocks ( Figure 1B). For time point October 2020 (Figure 2), some differences were found between Riesling and Mueller-Thurgau ungrafted compared to the grafted varieties and the control of the 20 dominating bacterial families.
For ungrafted grape varieties Mueller-Thurgau and Riesling, significant effects of the grape variety were found in alpha diversity indices Shannon and observed features ( by rare species, the Simpson index gives more weight to evenness and common species. For the grape variety Riesling with four different rootstocks, examined in October 2019, and for time point October 2020, no statistical differences were found in the alpha diversity (Shannon, Simpson, and observed features).

Bacterial beta diversity across the different experimental set-ups
The compositional beta diversity metric was calculated using a robust Aitchison PCA via DEICODE. Aitchison distance is a Euclidean distance between samples after centre log ratio (clr) transformation. For all experimental set-ups and time points,   significant differences in the beta diversity of the bacterial communities were detected (Figure 3). The determinant of the bacterial communities in the rhizosphere is the grape variety and rootstock, or grape variety and rootstock combination, respectively. PERMANOVA pairwise results reveal statistical differences (  Table 3). Statistical differences were also detected between Mueller-Thurgau grafted vs Riesling ungrafted, Mueller-Thurgau grafted vs the control, Riesling ungrafted vs Riesling grafted onto different rootstocks, Riesling ungrafted vs the control, and Riesling grafted vs the control (p < 0.5, Table 3). However, no statistical differences in beta diversity were detected between Mueller-Thurgau grafted and Riesling grafted (p = 0.21, Table 3).  This bacterial genus is also displayed in Figure 3A) as an  Figure 3C), Acidobacteriaceae, "Solibacteraceae", Comamonadaceae, Caulobacteraceae (Phenylobacterium deserti, also shown in Figure 3C)

DISCUSSION
The experimental design of this study aimed at minimizing the variability coming from all factors except grape variety and rootstock genotype using grapevines in pots under the same environmental conditions. The differences in the bacterial communities between the two grape varieties, Riesling and Mueller-Thurgau, ungrafted and grafted onto different rootstocks, were investigated using compositional data analyses. The observed ASVs revealed that the rootstock and scion rhizosphere, respectively, recruited complex bacterial communities mainly composed of Proteobacteria, Verrucomicrobiota, Chloroflexi, Myxococcota, Acidobacteria, and Gemmatimonadota. A similar study by Dries et al. (2021a) with ungrafted grapevine rootstocks also showed the phyla Proteobacteria, Acidobacteria, and Gemmatimonadota as some of the predominant bacterial communities in the rhizosphere. Other studies have come to comparable conclusions, independent of factors such as grape variety and rootstock (Berlanas et al., 2019;Coller et al., 2019;Marasco et al., 2018;Novello et al., 2017;Torres et al., 2021;Vink et al., 2021;Zarraonaindia et al., 2015). Gobbi et al. (2022) showed in their study that Proteobacteria occurred with the highest relative abundances, and Actinobacteria and Acidobacteria were the second-most or third-most abundant bacterial phylum. In this study, we also detected Actinobacteria in the samples (Supplementary Material), but not as one of the most abundant bacterial phyla, which could be due to the commercial soil used for the experiments. The dominating families give a higher resolution on the taxonomic level which is more suitable for comparison than the phyla level. During all experimental time points, Opitutaceae, Reyranellaceae, "Solibacteraceae", and Solimonadaceae were found as dominating families, among others (Figures 1 and 2). This is also consistent with a study conducted by Marasco et al. (2018). They revealed "Solibacteraceae" as one of the families shaping the topology of the bacterial network in the grafted root system (Marasco et al., 2018). Comamonadaceae (Figures 1A and  2) were also found in a study by D'Amico et al. (2018) in the rhizosphere of grapevine rootstocks 5BB and Paulsen 1103. As a wider variety of bacteria is found in soil, it can be assumed that soil serves as a primary reservoir for potential plant-associated bacteria (Zarraonaindia et al., 2015). However, soil, as well as the rhizosphere, can also form a path of infection for soil-borne pathogens (Berlanas et al., 2019). For grapevines, this includes pathogens such as Fusarium oxysporum (Freire Cruz and Carvalho Pires, 2014) or Sorosphaera viticola (Neuhauser, 2009).
The alpha diversity indices in this study revealed higher observed features for ungrafted Riesling grape varieties in May 2019 (Table 2). This may indicate that ungrafted Riesling grapevines could recruit a higher number of bacteria in their rhizosphere than Mueller-Thurgau grapevines. The underlying reason for this varying colonisation could originate from the root exudates. Indeed, root exudates are strongly influenced by the cultivar, and they are considered among the most important factors in the recruitment of the microbiome (Kusstatscher et al., 2021;Marasco et al., 2018;Wei et al., 2019). As a key role in the rhizosphere ecosystem, it is essential to understand the root exudation patterns to unravel the subsequent effects on the surrounding soil and microbial communities (Yee et al., 2021). Berlanas et al. (2019) investigated in their study the grapevine rootstock variety as one of the factors shaping the vineyard microbiome. In a study conducted by Vink et al. (2021), the authors observed highly specific cultivar-rootstock interaction effects on the microbiome only occurring for a few specific rootstockcultivar combinations. In a recent study by Marasco et al. (2022), the authors showed that the interaction of rootstock and scion resulted to be more important in shaping the root system microbiome than the rootstock and scion considered separately. They investigated four different rootstocks and three scions in seven combinations (Marasco et al., 2022).
The alpha diversity indices for Riesling and Mueller-Thurgau grafted onto different grapevine rootstocks revealed no statistical differences (Table 2). However, three out of four rootstocks used for this experiment have emerged from the same breeding: Vitis berlandieri Planch. × Vitis riparia Michx., which may be an explanation of the statistical same bacterial communities in the rhizosphere. The fact that most cultivated grapevines are genetic chimaeras with two different genotypes (Marín et al., 2021) complicates the separation of the genotypes from each other regarding their related microbial communities. Hence, all the studies indicate that certain taxa are always apparently present at all times in the rhizosphere of the grapevine. However, no statistical differences regarding alpha diversity indices were found between ungrafted Mueller-Thurgau and ungrafted Riesling for the time point October 2020, which could be due to differences between May 2019 and October 2020 regarding the time of growth in the greenhouse under field conditions. Since grapevines for May 2019 were grown only four months instead of six months, alpha diversity indices could have been changed.
The beta diversity revealed statistical differences in the three experimental setups. Pairwise PERMANOVA results showed statistical differences between ungrafted Mueller-Thurgau and ungrafted Riesling grape varieties but no statistical differences between different rootstocks. The results indicate that the grape variety may be a driving factor of the bacterial communities in the rhizosphere, while the combination of grafted vine and rootstock may have a minor effect. This is also in accordance with a former study, showing the ungrafted grapevine rootstocks as a driver of the bacterial communities (Dries et al., 2021a). Thus, it might be concluded that grafting grapevines onto rootstocks implies a change in bacterial communities. Moreover, Vink et al. (2021) revealed in their study that the main determinant of the bacterial communities was scion variety for the alpha diversity and a significant interaction between scion and rootstocks regarding the beta diversity. However, the authors did not refer to ungrafted grape varieties. In another study conducted by Berlanas et al. (2019), they described the rootstock genotype as the most important factor in shaping the microbiome. Wright et al. (2022) revealed that the rootstock was a significant factor driving the root microbiome with the grape variety New York Muscat ungrafted and grafted onto two different rootstocks.
Comparing the ASVs strongly influencing the principal component axis from the PCoA biplot ( Figure 3) with the results of the compositional differential abundance analyses (ALDEx2, Tables 1-3, Supplementary Material) for time point May 2019 revealed ASVs next relative to Nevskia terrae, uncultured Candidatus Solibacter, and uncultured Deltaproteobacteria as those taxa appearing in both analyses. Nevskia terrae was already described as a bacterium isolated from soil in Korea (Kim et al., 2011), belonging to the Xanthomonadaceae. Some members of this family are already described as plant-growth-promoting (Cutiño-Jiménez et al., 2020). Candidatus Solibacter was described previously as a bacteria inhabiting the rhizosphere of walnut trees (Bai et al., 2020) and Deltaproteobacteria in rice . For time point October 2020, all eight displayed ASVs strongly influencing the principal component axis from the DEICODE beta diversity also appeared in the compositional abundance analyses (ALDEx2, Tables 4-9, Supplementary Material). The next relative Phenylobacterium deserti was first isolated from desert soil (Khan et al., 2017), and other Phenylobacterium were also isolated from different soil samples (Khan et al., 2018;Li et al., 2019). The next relative of uncultured "Acidibacter" belongs to Gammaproteobacteria; members of this class are known for plant growth-promoting traits, inhabiting the rhizosphere (Madhaiyan et al., 2017), root nodules (Ibáñez et al., 2009), or plant tissue (Madhaiyan et al., 2020).
However, further research on this topic is needed to reveal all effects of rootstock and grapevine scion combinations on the bacterial communities in the rhizosphere. Here, it should be emphasised again that in practice, Vitis vinifera cultivars are usually grown as scions grafted onto Phylloxera-tolerant Vitis sp. Darriaut et al. (2022) indicate that grapevines can select different but potentially beneficial microorganisms via rootstock and scion genotypes. The effects of rootstock-scion interactions should, therefore, be considered primarily from a practical perspective. Nevertheless, it is remarkable that the bacterial communities change with grafting. Therefore, the next approach should be to find out whether this change brings advantages or even disadvantages. Moreover, it is important to understand the complexity of the grapevine holobiont as a crucial issue for the future of the wine industry (as reviewed by Bettenfeld et al., 2021). In addition, the effects on the growth and health of the vine must be revealed from a practical perspective. A thorough understanding of microorganisms in vineyard soil and the complex relationships between microbial communities, soil properties, and plants are crucial for enhancing plant productivity, grape production, biogeochemical processes, and vineyard management practices (Di Liu et al., 2019;Dries et al., 2021b;Holland et al., 2014;Liang et al., 2019;Yee et al., 2021). Inferring, the colonisation of the microorganisms in the rhizosphere and the root exudation patterns of grapevine must be investigated. Future research in this field is inevitably required to provide a better understanding of the rhizospheric grapevine microbiome in the context of root exudates, grape variety, and rootstock, as well as different environmental conditions.

CONCLUSION
The results from these experimental designs reveal differences in the bacterial communities in the rhizosphere of grafted or ungrafted grapevine varieties, respectively. The bacteria in the rhizosphere of grapevine are affected by both grapevine variety and scion-rootstock combination. While differences were observed between ungrafted vs grafted grape varieties, especially in terms of beta diversity, no differences were observed between the different rootstocks. Thus, the grapevine cultivar appears to have a predominant role compared with the rootstock in shaping the rhizosphere microbiota. Further research is needed to provide a better understanding of the different microorganisms in the grapevine rhizosphere regarding the scion-rootstock combinations. Moreover, the effects on the grapevine growth and health, and also on the wine quality, have to be revealed.