Changes of bacterial communities in the rhizosphere of sugarcane under elevated concentration of atmospheric CO2

It is believed that climate change will influence most of interactions that sustain life on Earth. Among these, the recruitment exerted by plants in their roots vicinity can change, leading to differential assemblages of microbiomes in the rhizosphere. We approached this issue analyzing the variations in the composition of bacterial communities in the rhizosphere of sugarcane cultivated under two concentrations of atmospheric CO2 (350 or 700 ppm). In addition to the analysis of bacterial community, the use of DNA‐SIP allowed the comparison of bacterial groups assimilating roots exudates (based on 13C‐labeled DNA) in both conditions, in a period of 8 days after the CO2 pulse. The separation of 13C‐DNA indicated the low but increasing frequency of labeling in the rhizosphere, as averages of 0.6, 2.4 and 5.0% of total DNA were labeled after 2, 4, and 8 days after the 13CO2 pulse, respectively. Based on large‐scale sequencing of the V6 region in the gene 16S rRNA, we found an increase in the bacterial diversity in the 13C‐DNA along the sampling period. We also describe the occurrence of distinct bacterial groups assimilating roots exudates from sugarcane cultivated under each CO2 concentration. Bacilli, Gammaproteobacteria, and Clostridia showed high affinity for the C‐sources released by sugarcane under 350 ppm of CO2, while under elevated concentration of CO2, the assimilation of roots exudates was prevalently made by members of Bacilli and Betaproteobacteria. The communities became more similar along time (4 and 8 days after CO2 pulse), in both concentrations of CO2, electing Actinobacteria, Sphingobacteriia, and Alphaproteobacteria as the major cross‐feeders on sugarcane exudates. In summary, we described the bacterial groups with higher affinity to assimilate roots exudates in the rhizosphere of sugarcane, and also demonstrated that the rhizosphere community can be differentially assembled in a future scenario with increased contents of CO2.


Introduction
The increase in the contents of CO 2 in the atmosphere is one of the major components of climate change, first acting as a triggering component, leading to increases in average temperature of Earth, what consequently affects the climate and the rain regimes worldwide (Chen et al., 2013;Vasskog et al., 2015;Castello & Macedo, 2016;Worn & Paine, 2016;Runting et al., 2017).
The rapid increase in the concentration of atmospheric CO 2 can affects the plant metabolism, changing photosynthetic rates (Jia & Zhou, 2011;Philippot et al., 2013). It is postulated that it can change plant physiology, for example, leading to higher accumulation of biomass (Guenay et al., 2013) or determining differential patterns of diseases and pests occurrences . In addition, as plants invest part of the assimilated carbon to feed microbes in their rhizosphere (Jia & Zhou, 2011), climate change can interfere in the interactions between plants and beneficial microbes in the soil.
Rhizosphere is the ultimate soil portion in the roots vicinity, where microbial community assemblage and activity are modulated by the release of roots exudates (Drigo et al., 2013;Jiang et al., 2017). Plants pose a selection in this niche, aiming to increase the abundance of beneficial organisms, as those that prompt plant development, either by nutritional supply or protection against stressors (Medeiros et al., 2006;Mendes et al., 2013). Changes in the concentration of atmospheric CO 2 have been shown to influence the rhizosphere composition in native fields composed by grass plants (Drigo et al., 2010). However, it is important to make similar inferences in plants used for agriculture, what can serve for a better knowledge about the microbiomes of cultivated plant species, directly connecting climate change with food supply efficiency.
Sugarcane (Saccharum officinarum L.) is an important crop in Brazil, serving the energetic matrix, and used for the production of sugar and ethanol (Carvalho et al., 2017). Its high potential to produce biomass also includes this plant use for electrical and/or thermal energy (Lozano & Lozano, 2017). It presents a C4 metabolism, characterized by the linear increase in photosynthesis rate along the augmentation of the atmospheric CO 2 concentration (Eisenhauer et al., 2012. Recent studies have suggested the potential role of several bacterial groups influencing sugarcane development (Bhardwaj et al., 2017). These characteristics elect sugarcane as an important model plant to study the responsiveness of microbial communities in the rhizosphere for the increase in the CO 2 contents in the atmosphere.
Here, we used it to build an experimental design, complemented by the application of proper analytical methods to achieve the following objectives: (i) the description of bacterial groups that rapidly respond to the exudation of sugarcane roots; (ii) the detection of variations in the frequencies of these groups in a scenario with higher contents of atmospheric CO 2 . These objectives derived from the central hypothesis that changes in atmospheric concentration of CO 2 , as promoted by climate change, can influence the composition of bacterial community in the rhizosphere of sugarcane.

Soil and plant sources
The soil used in this experiment was collected at the farm located in 'Luiz de Queiroz' College of Agriculture (ESALQ) farm, a campus of the University of São Paulo, located in Piracicaba, Brazil (22°41 0 42.96″S; 47°38 0 28.68″L). The soil was classified as Kandiudalfic Eutrudox with clayey texture (NVef) (Santos et al., 2006) and characterized by 44.7% of clay, 26.4% of silt and 28.9% of sand, density of 1.29 g cm À3 ; pH 5.0 in CaCl 2 ; 2,7% of organic matter, and 86% of bases saturation.
Sugarcane seedlings (variety RB86-7515) were generated form original material also collected in the same farm. The selection of this variety was also based on its wide cultivation along the last 10 years in the main region for sugarcane production in Brazil (the State of São Paulo) (Costa et al., 2014).
Experimental setup, 13 C labeling, and samples collection The experiment was set using sugarcane seedlings with approximately 60 days after stalks germination in vases containing 3 kg of soilapproximately 30-40 cm height, with 3-4 leaves. These seedlings were then incubated into two sealed cabinets, made of acrylic (polymethyl methacrylate -PMMA), located in a greenhouse with temperature control at 24°C, providing the internal temperature in the cabinets between 26 and 28°C (monitored by a internal placed thermometer), limiting temperature stresses for the plant. When temperature raises above 28°C cabinets were opened until temperature stabilization.
The seedlings were then acclimatized for 30 days under greenhouse conditions (26-28°C, 16-h photoperiod), in treatments with two concentrations of CO 2 : 350 (AE17.5) or 700 (AE35) ppm (daily checked and adjusted). During acclimatization, cabinets were opened for 1 h every day, and the amount of CO 2 inside chambers were monitored by gaseous chromatography, calculating the average rate of CO 2 fixation per plant. The amendments of CO 2 were made based on the amount needed by the plant to perform photosynthesis. These concentrations of CO 2 were adopted following a previous study where such increase in the CO 2 contents was able to induce significant changes in the structure and composition of the microbial communities in the rhizosphere of pristine plants (Drigo et al., 2010).
Plants were then labeled with pulses of 13 CO 2 ( 13 C, 99% CO 2 , Cambridge Isotope Laboratories) or 12 CO 2 (for control plants), injected in the period of greatest solar intensityconnecting it with higher assimilation by photosynthesis. In total, 36 plants were used; each nine plants receiving a pulse of 13 CO 2 or 12 CO 2 , at concentrations of 350 or 700 ppm. Three rhizosphere samples (one per plant) were collected 48, 96, and 192 h after pulses of CO 2 (either 12 CO 2 or 13 CO 2 ). During this period plants did not shown any visual differences concerning its development. The rhizosphere was considered as the soil fraction highly adhered to roots. To proper access this soil portion, plants were removed from soils, roots were vigorously shaken and rinsed. Approximately 1.0 g gram of soil portions adhered to the roots were then removed using a spatula and stored in 2 mL plastic microtubes.

DNA extraction and separation of 'light' and 'heavy' fractions
The rhizosphere samples were subjected to total DNA extraction using the commercial kit Mobio Powersoil TM DNA Isolation (Mobio Laboratories Inc, Carlsbad, CA, USA). The quantity and integrity of the DNA were checked in agarose gel electrophoresis 1% (w/v) in 19 TAE buffer (Tris, acetic acid, EDTA), stained with in Sybr Green I (Invitrogen, Waltham, MA, USA), and further visualized under UV light.
The separation of labeled DNA ( 13 C-DNA) was performed by the ultra-centrifugation of 500 ng (in volume of 10 lL) of soil DNA in a gradient of cesium trifluoroacetate (CsTFA; Amersham Biosciences) (Manefield et al., 2002). Briefly, aliquots of DNA from each sample were added to 2.2-mL tubes containing 2.0 mL of CsTFA solution with a density of 1.62 g mL À1 . Control tubes were also prepared containing 10 lL of ultrapure water. All tubes were centrifuged in a bench Optima TLX ultracentrifuge 120 000 (Beckman, Brea, CA, USA) with a fixed angle rotor TLA 120.2 at 64 000 rpm for 40 h at 20°C.
After centrifugation, samples were fractionated by sample depletion from the bottom to the top of tubes using fractionation recovery system (Beckman), with a syringe attached to an infusion pump (flow rate of 0.2 mL min À1 ). A total of 25 fractions of 80 lL of solution were collected for each sample.
The density of each fraction was determined at the AR200 Refractometer (Reichert, Depew, NY, USA) as a measure of the refractive indices (RI), further used to the construction of the density curves (g mL À1 ). Total DNA present in each fraction was also quantified in each fraction using NanoDrop â ND-1000 UV-Vis Spectrophotometer (Uniscience, Hialeah, FL, USA). Fractions containing light and heavy portions of DNA were subjected to precipitation with isopropanol at 25°C for 2 h followed by 15 min of centrifugation (17 000 g at 4°C). Pelleted DNA was rinsed twice in chilled ethanol 70%, dried on air, and suspended in ultrapure sterilized water (10 lL). The total amount of DNA in each of these purified fractions was determined by quantification with Qubit â . 2.0 Fluorometer using Quant-iT dsDNA HS kit (Thermo â , Waltham, MA, USA).
Due to the low amount of DNA obtained in heavy fractions, aliquots of 10 to 30 ng of DNAeither from light or heavy samples -were submitted to amplification of the complete genome, promoted by the Whole Genome Amplification kit (WGA4) (Sigma, St. Louis, MO, USA), following the manufacturer instructions.
Sequencing the V6 region of the of bacterial gene 16S rRNA Fractionated DNA (light and heavy fractions) was subjected to sequencing of amplicons containing the V6 region of the gene 16S rRNA. The amplification reactions were performed using primers Bac967fF (combined with different 5-nucleotide tag for each ample) and Bac1046R as previously described (Sogin et al., 2006). Amplicons were purified with ChargeSwitch â PCR Clean-Up Kit (Life Technologies, Carlsbad, CA, USA) and quantified using the Qubit â 2.0 Fluorometer (Life Technologies). Sequencing was performed using the Ion Torrent personal genome machine (PGM) (Ion Torrent, Life Technologies) system with Ion 316 Chip (Ion Sequencing Kit User Guide v2.0, Life Technologies). See Kavamura et al. (2013) for a detailed description of this method.
The sequences analysis was carried out using the software QI-IME (Quantitative Insights Into Microbial Ecology) (Caporaso et al., 2010). The removal of poor quality sequences (below 50 pb) and primers, as well as the separation of samples by their tags (barcodes), was carried out from the split_libraries.py command. High quality sequences were binned into OTUs using a 97% nucleotide identity threshold with the uclust method (Edgar, 2010). The OTU table was rarified 17 000 sequences per samples (the lowest number found in one sample), decreasing the effects of sampling effort on the analysis. The representative sequences for each OTU were subjected to taxonomic assignments using the Basic Local Alignment Search Tool (BLAST) against the Caporaso Reference OTUs (Greengenes, 2014).

Data analysis of OTUs and taxa matrices
QIIME was used to generate weighted/unweighted UniFrac distance matrices (Lozupone et al., 2006) and a-diversity metrics, including Shannon-Weaver diversity indices. Values of bacterial diversity were compared by variance analysis (ANOVA), and average values were compared by Tukey test (P < 0.05).
To assess the variation in the phylogenetic beta-diversity among different samples we used the Principal Coordinate Analysis (PCoA) derived from the weighted UniFrac distances of bacterial communities. The separations between the groups were tested by the similarity analysis (ANOSIM) based on the distance matrix of OUT's transformed by Gower's algorithm. The intervals of R-values can indicate total separation (R > 0.75), separation with overlap (0.5 < R < 0.75), and approximation (R < 0.5) (Clarke & Gorley, 2001).
We used data from phyla and class taxa summary matrices to determine by Similarity Percentages (SIMPER) (Clarke, 1993) the taxa that primarily respond to variations of the communities in each concentration of CO 2 , or between light and heavy fractions of the DNA.
Using data from groups named as responsive by SIMPER, the percentages of the marked phylum/classes ( 13 C-DNA) were divided by the respective percentages of unlabeled ( 12 C-DNA), and this ratio was subtracted from 1. The same was done between taxa assessed in standard atmosphere and with high CO 2 concentrations. The data were normalized in interval À1 and 1, representing the values of predominance (P) of each taxon as a function of the isotopic and atmospheric variations. P > 0 indicates a predominance of the contribution of the marked or cultivated taxon in 700 ppm of CO 2 over the unmarked and cultivated in 350 ppm of CO 2 , respectively. P < 0 indicates the inverse of these predominance, and P = 0 refers to equivalent contributions of the taxon between the isotopic and atmospheric variants. The VEGAN package (Oksanen et al., 2011) was used to run all these analyzes in R environment.

Obtaining light and heavy fractions of rhizosphere DNA
The quantification of sugarcane carbon assimilation by the injection of CO 2 determined a peak of activity where plants fix approximately 210 mg CO 2 h À1 (Fig. S1). This information was later used to provide the CO 2 pulse for plant assimilation on appropriate period of the day (around noon).
We found similar percentages of 13 C labeled DNA in pulses with 350 or 700 ppm of CO 2 , with average values of 0.6%, 2.4%, and 5.0% of total DNA at two, four, and eight days after pulses, respectively. These percentages correspond to 13 C-DNA concentrations between 0.02-1.00 mg mL À1 , collected from fractions 0.161 to 0.166 g mL À1 within a density of CsTFA solution (Fig. S2). Similarly, lightweight DNA was concentrated in fractions of density ranging from 0.157 to 0.160 g mL À1 .
12 C-DNA-and 13 C-DNA-labeled bacterial structure and diversity A total of 445 853 valid sequences were used to support the further inferences (170 566 from 12 C-DNA and 275 287 from 13 C-DNA). These sequences were binned in 41.130 OTUs, which revealed a lower bacterial diversity in 13 C-DNA (3.92-6.68) than in 12 C-DNA (7.61-7.78) (Fig. 1). Comparing the labeling patterns in 350 or 700 ppm, only at 4 days after CO2 pulse, a difference is observed among treatments, with higher values of bacterial diversity for plants cultivated under 700 ppm of CO 2 (Fig. 1). Considering 12 C-DNA samples, the diversity was similar along the period of the experiment, and also independently on CO 2 concentrations. Evaluating 13 C-DNA samples, the tendency of diversity was similar between samples on different CO 2 concentrations, but it was possible to observe lower values on the beginning of incubation (2 days) and the progressive increase in diversity along the experimental period (4 and 8 days after the CO 2 pulse).
The PCoA shows a clear separation between the light and heavy fractions of DNA, represented in the first axis with 31% of variations explanation. Additionally, differences between samples from plants cultivated under distinct CO 2 concentrations were detected in the second axis, where 9.3% of the variation was plotted (Fig. 2).
The analysis of ANOSIM confirmed these differences in the community structures, with differentiation between light and heavy bacterial communities in both concentrations of CO 2 (R > 0.94). In the same context, the differences between communities derived from atmospheres with 350 ppm or 700 ppm of CO 2 were confirmed for 12 C-DNA (R = 0.70), and described as subtle for analysis based on the 13 C-DNA (R = 0.14) ( Table 1).

Bacterial succession driven by C consumption in the rhizosphere of sugarcane
We succeed to detect differential bacterial communities acting as early assimilators in the rhizosphere of sugarcane cultivated under distinct CO 2 concentrations. After 2 days of the 350 ppm 13 CO 2 pulse, the most frequent bacterial classes were Bacilli (32.6%), Gammaproteobacteria (27.5%) and Clostridia (29.3%), while prevalent groups cultivated under 700 ppm of CO 2 were affiliated to Bacilli (36.6%) and Betaproteobacteria (30.9%) (Fig. 3a).
The temporal variation in the taxonomic composition of the labeled bacteria in sugarcane rhizosphere was also observed. Either in samples from plants cultivated under 350 or 700 ppm of CO 2 , communities revealed similar dynamics in samples collected at 4 and 8 days Fig. 1 Values of bacterial diversity, based on Shannon-Weaver estimator, for light (12C-DNA) and heavy (13CDNA) fractions of DNA used to depict bacterial communities in the rhizosphere of sugarcane cultivated under standard (350 ppm) or elevated (700 ppm) concentrations of CO2, collected 2, 4, and 8 days after the CO2 pulse. Data represent means of three repetitions, and different letters indicate statistical differences as determined by the Tukey test (P < 0.05). after the 13 CO 2 pulse (Fig. 3a). In both cases, it is remarkable the increase in the frequencies of Actinobacteria, Sphingobacteriia, and Alphaproteobacteria, elected as the major cross-feeders, or secondary assimilators of the sugarcane exudates (Fig. 3a). The dynamics revealed by the labeled fraction of the DNA was not evident in the community derived from 12 C. Data from 12 C-DNA suggest a constant pattern of bacterial communities along the sampling period of 2, 4, and 8 days after the CO 2 pulse (Fig. 3b). These patterns were also similar in plants cultivated under 350 or 700 ppm of CO 2 . Remarkable is the frequent occurrence of Anaerolinea in samples of 12 C-DNA, a group not found in labeled DNA (Fig. 3b).
The SIMPER analysis revealed the complete set of bacterial groups supporting the differentiation of the communities described above (Table 2). Comparing the communities under distinct CO 2 concentrations, it is possible to describe the response of 17 bacterial groups along time (Table 2) or the differentiation between 12Cand 13C-DNA (Table 3).

Discussion
Plants rely on their associate microbiota to accomplish several of the functions responsible for their proper and healthy development. The rhizosphere is the main sensor of this interaction, where plants and microbes complement themselves nutritionally and protect each other from stressors (Pietersen et al., 2016). The knowledge on major players in this arena is essential either for better use this resource in technological innovations, and to predict changes in the plant association in future scenarios, as those promoted by climate change.
Several studies have shown the consequences of the increase in the CO 2 contents in plant metabolism and connections to soil. For example, an increase in the CO 2 to 850 ppm is correlated with increases in plant biomass (Garc ıa-Palacios et al., 2015). Also, Eisenhauer et al. (2012Eisenhauer et al. ( , 2013 showed that the increase in CO 2 concentrations increases the microbial biomass in soils. Together, such differentiations could induce shifts in plant metabolism, modulating the composition of roots exudates either quantitative or qualitatively (Van Veen et al., 1991;Medeiros et al., 2006;Drigo et al., 2010Drigo et al., , 2013, leading to differentiation on plant associated microbiota (Jia & Zhou, 2011;Philippot et al., 2013). Drigo et al. (2010) have used SIP-DNA to determine this effect in the composition of the rhizosphere microbial community in pristine gramineous species. Here, we used a similar analysis to infer on the bacterial communities found in the rhizosphere of sugarcane, a widely cultivated crop, responding for more than 20 million hectares of land use worldwide (http://www.fao.org/faostat/en/). In this case, consequences of climate change may possibly have a strong impact on plant development and viability, besides possible impact on the carbon cycle and storage in agricultural soils.
Results indicate the efficacy of the SIP approach to track the bacterial groups assimilating C direct from sugarcane roots. The increases along the 8 days of sampling in the amount of DNA labeled, and also in the diversity of bacterial community (Fig. 1) validate our approach. The sensitiveness of the SIP approach is indicated by the results, where bacterial community revealed by 12 C-DNA is similar, but 13 C-DNA showed very distinct patterns of bacterial community along samplings. The faster increase in bacterial diversity with labeled DNA also suggests that C release in the rhizosphere of plants cultivated under 700 ppm of CO 2 is more intense than those cultivated under 350 ppm of CO 2 . Similar analyzes of microbial communities in the rhizosphere have also used small periods of high CO 2 treatment to make inferences on this issue. For example, Gschwendtner et al. (2015) showed that incubation of beech trees (Fagus sylvatica L.) for only two days in high CO 2 concentrations (1100 ppm) significantly changes the diversity of bacteria in the rhizosphere. Modification of quantity and quality of plant exudation pattern may lead on a recruitment of differential groups of bacteria in their roots vicinity (Van Veen et al., 1991;Drigo et al., 2010Drigo et al., , 2013. We identified differential microbial assemblages in plants cultivated under distinct CO 2 concentration. The prevalent groups assimilating 13 C in the rhizosphere of plants grown under 350 ppm of CO 2 were Bacilli, Gammaproteobacteria, and Clostridia, while in plants under higher contents of CO 2 , most of bacteria labeled by 13 C were Bacilli and Betaproteobacteria (Fig. 3). The major bacterial groups found in the rhizosphere of sugarcane in this study were similar to those previously described as associated with this plant species (Dini-Andreote et al., 2010;Rachid et al., 2013;Durrer et al., 2017). Besides the major responders, the approach used here also named the major cross-feeders in the rhizosphere of sugarcane (Fig. 3). Members of Actinobacteria, Sphingobacteriia, and Alphaproteobacteria seem to properly get carbon from metabolites generated by the primary assimilators in the sugarcane rhizosphere, either under 350 or 700 ppm of CO 2 . Besides the composition of roots exudates, the differential dynamics of microbial communities observed in rhizosphere emerging from distinct concentrations of CO 2 could be also explained by lifestyle of some bacterial groups identified (Ho et al., 2017). Bacilli presented a behavior classified as stress-tolerant life style (Ho et al., 2017), as these organisms tolerate variable substrate variability, as indicated by the known features found in this groups of bacteria, such as desiccation and heat-resistant cells (Grime, 1977). Betaproteobacteria are classified as r-type life strategy or competitorsruderal (Ho et al., 2017). In this case, bacterial groups are proned to multiply in the presence of abundant sources of nutrients (Fierer et al., 2007;Ho et al., 2017). This group was responsive for the increase in the CO 2 contents in the atmosphere, what possibly increased plant exudation in the roots vicinity. The major cross-feeders members of Actinobacteria, Sphingobacteria, and Alphaproteobacteriaare possibly versatile in substrate utilization, able to use nutrient from recalcitrant soil organic matter, or derived from primary assimilators, a feature associated with stress-tolerantruderal life strategy (k-type) (Ho et al., 2017). Considering that some of these groups (e.g., Betaproteobacteria) occur differentially in the rhizosphere of plants under distinct concentrations of CO 2 , we need further to investigate the consequences of this shift for plant development.
Unfortunately, the sequencing approach used in this survey does not allow a better inference on the taxonomical identification of bacterial cells living in distinct rhizosphere, what could support more detailed inferences on the roles of bacteria when recruited by plant roots. Whether the changes in taxonomical composition identified here lead to a differential functioning of the rhizosphere, promoting or inhabiting plant growth remain to be targeted in future studies. Metagenomics also could support deeper studies of taxonomical identification and function of bacterial cells living in distinct rhizosphere.
In summary, this study is the first attempt to show changes in the rhizosphere of cultivated plants due to an increase in the CO 2 contents in the atmosphere. By approaching it using sugarcane as a model, we can suggest that in future scenario cropped plants may differentially assemble their rhizosphere. Intriguingly, we do not have information about the maintenance of the environmental services, or about the plant development supported by microbial communities in the rhizosphere in this new condition. Based on the model proposed here, we can move forward in this area of study in order to properly predict changes in the interactive system between plants and microbes in the rhizosphere due to the climate change. Table 3 The SIMPER determination of major bacterial groups contributing to distinctions of bacterial communities, and its percentage in communities depicted based on light ( 12 C-DNA) and heavy ( 13 C-DNA) fractions of DNA