Distinct microbial communities among different tissues of citrus tree Citrus reticulata cv. Chachiensis

Plant microbiota colonize all organs of a plant and play crucial roles including supplying nutrients to plants, stimulating seed germination, promoting plant growth, and defending plants against biotic and abiotic stress. Because of the economic importance, interactions between citrus and microbes have been studied relatively extensively, especially citrus-pathogen interactions. However, the spatial distribution of microbial taxa in citrus trees remains under-studied. In this study, Citrus reticulata cv. Chachiensis was examined for the spatial distribution of microbes by sequencing 16S rRNA genes. More than 2.5 million sequences were obtained from 60 samples collected from soil, roots, leaves, and phloem. The dominant microbial phyla from all samples were Proteobacteria, Actinobacteria and Acidobacteria. The composition and structure of microbial communities in different samples were analyzed by PCoA, CAP, Anosim and MRPP methods. Variation in microbial species between samples were analyzed and the indicator microbes of each sample group were identified. Our results suggested that the microbial communities from different tissues varied significantly and the microenvironments of tree tissues could affect the composition of its microbial community.

Plants host diverse microbes that colonize to, on and in their tissues 1 . Based on their habitats, plant-associated microbial communities are referred to as rhizosphere microbiome, rhizoplane microbiome, phyllospher microbiome, and endosphere microbiome, respectively [2][3][4][5] . Different microbiomes interact with host plants and impact plants in various ways. In general, plant-associated microbial organisms can potentially have both positive and negative impacts on plant growth, development, and health 1 . Direct impacts on plant growth and development by microorganisms include improved nutrient accessibility such as nitrogen fixation and phosphate solubilization; altered microenvironments such as changed acidity (pH); and hormonal stimulation (phytohormone production) 3,6 . Microorganisms are also involved in the promotion or suppression of plant diseases either directly (such as antibiotics production) or indirectly (via disease resistance) 7 . Accordingly, characterization of microbial compositions and dynamics in different plant tissues and elucidation of functions associated with specific microbes or microbial communities will provide the basis for developing commercially probiotics for plants or designing strategies to manipulate microbes or microbial communities for economic or environmental benefits [8][9][10] .
Recent advances on culture-independent, high-throughput sequencing of the variable regions of 16 S rRNA genes in microbes and total genome assembly from metagenomic sequence reads has revolutionized our research on the structures and diversity of microbiomes in different plant tissues. As a result, microbiome data on various types of plants have been accumulated rapidly in recent years. Two independent studies on the microbial community in the root of the model plant Arabidopsis thaliana revealed consistent results on the core microbiome, with Actinobacteria and a few families from Proteobacteria enriched consistently in the endosphere compared with rhizosphere 11,12 . Microbiomes from various tissues of many crops such as sugarcane, rice, tomato, maize, sorghum, soybean, and cotton have been characterized to various degrees [13][14][15][16][17][18][19] . Microbiomes from certain tissues of fruits and trees including pear, banana, and apple have initially analyzed as well [20][21][22] .
Orange (Citrus × sinensis) is one of the most important fruits for humans worldwide. In Brazil alone, 35.6 million tons of oranges were produced based on the most recently available data in 2013. Like other crops and fruits, citrus production faces many hurdles and challenges that need to be overcome to achieve high yield and high quality of oranges. Characterization of microbiomes in different tissues of citrus trees may yield useful information for the improvement of orange production. There is fragmented information available on microbiomes of citrus β diversity analysis of samples. Principal co-ordinates analysis (PCoA) of samples from different parts of citrus trees were performed. The samples with the closer distance of Unweighted Unifrac indicated higher similarity between microbial communities. As shown in Fig. 2A, the microbial community in soil samples was more similar to that in root samples with all samples clustered together. The microbial communities in leaf and phloem samples differed significantly. The microbial communities differed significantly among the groups of

Similarity and difference among microbial community structures. The analytical methods Anosim
and Multi-Response Permutation Procedure (MRPP) were used to compare potential similarities and differences among community structures of different sample groups. ANOSIM analyses (analysis of similarities) revealed that all R-values were greater than 0 and all P-values were 0.001, indicating significant differences between microbial community structures among different sample groups (Table 3). MRPP analyses also revealed significant differences with values of expecting delta ranged from 0.08305 to 0.3433 and values of significance 0.001 (Table 4).

Microbial species analyses.
The metastat software based on Fisher exact test was used to analyze the differences in microbial species between different sample groups. The species with significant differences between different groups were filtered out according to the P and Q values (Table S2). Rhodopila was unique in the phloem group. Halomonas and Ralstonia were enriched in the leaf and phloem groups (P < 0.01). The abundance of these two microbes in the leaf group was significantly higher than that in the phloem group (P < 0.01). Methylobacterium and Sphingomonas were also very abundant in both the leaf and phloem groups (P < 0.01), with higher abundance in the phloem group. Streptomyces, Burkholderia-Paraburkholderia, and Acidibacter were mainly in the root and soil groups with higher abundance in the root group between the two (P < 0.05). Nitrosopumilus was enriched in the soil group compared to the other three groups (P < 0.01). Rhizobium was more abundant in the root group (P < 0.01).   Table 4. Inter-group difference analyzed by MRPP. R: Roots, S: Soils, P: Phloem, L: Leaves.

Discussion
Holobionts with high diversity residing in different plant tissues are regarded as plant microbiota 1 . The structure and function of microbial communities have attracted a great deal of attention because of increasing evidence suggesting critical roles of microbiomes in plant development and survival 34 . Plant-associated microorganisms could directly provide nutrition to plants via nitrogen fixation and phosphate solubilization 35 . Microbes can also induce resistance to other biotic and abiotic stresses through producing or degrading phytohormone 35 . Hence, it is necessary to investigate local microbial communities associated with plants to examine their impact on various aspects of a given plant species. Different plants host different microbial communities. Meanwhile, the microbial communities of plants could be affected by many factors. For example, the bacterium 'Candidatus Liberibacter asiaticus' (CLas) which induced the citrus devastating disease named Huanglongbing (HLB) could alter the root microbial community structure of Citrus limon and Citrus sinensis 36 . Administration of antibiotics including penicillin disrupted the interspecies microbiological connections and induced major changes in root bacterial community structure of grapefruit trees (Citrus paradise Macf.) 37 . Besides, the diversity of fungal endophyte communities in leaf, stem, trunk, and root tissues of C. reticulata cv. Siyahoo was observed 38 . However, a high similarity of dominant bacterial communities of Pericarpium Citri Reticulatae 'Chachiensis' obtained from different orchards in Xinhui District was observed 39 . In order to assess differences and dynamic changes in microbial communities among different tissues of C. reticulata cv. Chachiensis, 16S rRNA gene sequencing has been adapted to analyze microbes in samples derived from nearby soil, roots, leaves, and phloem. Our results demonstrated that the numbers and types of OTUs varied greatly among different sample groups. Among the microbes identified in this study, some genera have been well studied in the past few decades, whereas others were novel and unique. Identification of these microbes will facilitate comparative research on known microbes and functional studies on unique species. Furthermore, the microbial communities of many plant species were varied by geographical location. For example, the microbial diversity of C. reticulate blanco var. clementine in different regions was confirmed by 16S rDNA fingerprinting analysis 40 . For spruce (Picea spp.) trees, a significant difference of the microbial taxonomical composition in the phyllosphere was observed at three locations 41 . Moreover, the microbial community structure of malts from different cropping zones was variable. Among them, the effects of geographical location on the fungal community were more obvious than bacterial community 42 . Therefore, to study the microbial community of C. reticulata cv. Chachiensis more systematically, the effect of different geographical locations should be concerned and further studied. www.nature.com/scientificreports www.nature.com/scientificreports/ Large differences in community structures in different microbiota have been observed in other plants from previous studies 14,43 . Microenvironments of tree tissues are likely one of the major factors driving alteration in the composition of microbial communities 44 . Microenvironments in tissues of citrus trees might also play a major role in the variation of microbial communities observed here. As an open system, plants constantly obtain minerals and water from the surrounding soil. In plant-microbe interactions, a plant may play an active role in shaping its associated microbial communities based on its needs. In turn, microbes may also affect plant physiology. The dynamic interactions among plants, microbes, and other environmental factors determine the functioning of an ecological system [45][46][47] . Soil by far harbors the most biodiversity and is the largest reservoir of microbes. Soil microbes impact plant microbiota profoundly 48 . Roots of a plant are the initial and main sites for plant-microbe intimate interactions. Rhizoplane, one of the root-associated layers, serves as a critical gate that regulates microbial entry into roots 49 . A comprehensive analysis indicated that no difference existed in the microbial community structure of rhizosphere and associated bulk soil samples collected from twelve citrus varieties which distributed on six continents 50 . In this study, the microbial community structure in roots was also similar to that in soil, suggesting that root-associated microbes were mainly derived from the soil biome 35 . Alternatively, the exudates containing sugars, organic acids and amino acids secreted by plant roots have strongly affected the composition of microbes in surrounding soil. Proteobacteria, Actinobacteria, Acidobacteria and Bacteroidetes have been reported as the primary consumers of plant exudates and the predominant taxa of citrus 50,51 . Other reports supported the above results and the microbial community of other plants including maize, arabidopsis and tamarix consists of a few dominant phyla, such as Proteobacteria, Actinobacteria, and Bacteroidetes 16,52,53 . In this study, Proteobacteria, Actinobacteria, Acidobacteria, Bacteroidetes, and Firmicutes were ranked as the top five core microbes in soil and root samples, which was similar to the global citrus rhizosphere microbiome 50 . However, the core genera of global citrus rhizosphere microbiome including Pseudomonas, Agrobacterium, Cupriavidus, Bradyrhizobium, Rhizobium, Mesorhizobium, Burkholderia, Cellvibrio, Sphingomonas, Variovorax and Paraburkholderia differed with our results, and the different growth area, seasonal variation and environmental factors could be the reasons for the difference. Furthermore, the genus from these phyla including Dyella, Rhizobium, Kribbella, Streptomyces, Granulicella, Actinospica, Amycolatopsis, Nocardia, Burkholderia. Paraburkholderia and Novosphingobium were identified as the indicator microbes in the root group. These indicator microbes could play crucial functions to citrus trees, for example, supplying nutrients, conferring resistance against pathogens and parasites 47 .
The microbial communities in citrus leaves and phloem are far smaller in comparison with those in roots and soil in terms of the number of OUTs in each group based on our results. The significant differences could be related to microenvironments in different tissues. Microbes can be originated from different sources, including aerosols or dust 45 . Insects are another important source for plants to gain microbes, including viruses, phytoplasmas, fungi, and bacteria, which are commonly transferred by insects. For example, Diaphorina citri Kuwayama (Hemiptera) transmit the bacterium Candidatus Liberibacter asiaticus (CLas), which causes a destructive disease called Huanglongbing 54 . Therefore, different sources for plant tissues to gain microbes could be another reason for large differences in their microbial communities. As ground material, the central bacterial community in Pericarpium Citri Reticulatae 'Chachiensis' separated from the fresh fruit of C. reticulata cv. Chachiensis were investigated and Chloroplast and mitochondria were identified as the most abundant bacterial phylum 39 . The significant differences that occurred between our results and the results obtained from Pericarpium Citri Reticulatae 'Chachiensis' might due to the massive mitochondria and chloroplast of C. reticulata existed during the processes of microbes enrichment and DNA extraction 39 . In our knowledge, this is the first time to investigate the microbial community of citrus phloem, which could helpful for understanding the distribution of microbial communities in entire citrus plants. Besides, Methylobacterium and Amnibacterium were identified as the major genus in citrus phloem. Methylobacterium was demonstrated to reduce the proliferation of citrus pathogens including CLas 55 , while the function of Amnibacterium should be explained in further study. For the leaf samples, Proteobacteria, Firmicutes and Bacteroidetes were ranked as the top three phyla and counted for more than 96.9% of all the microorganisms. Moreover, Alphaproteobacteria, Betaproteobacteria, and Gammaproteobacteria were the most abundant bacterial classes in leaf samples, which was similar to the results of microbial communities of leaves collected from citrus trees across Florida 29 . Furthermore, some microbes with uniqueness were specific to tissue and were taken as the indicators of that tissue. For example, Methylobacterium, Amnibacterium, Rhodopila, and Terriglobus are indicators for the phloem group. Ralstonia, Bacteroides and Prevotella are indicators for the leaf group. The roles of these indicator microbes remain to be clarified.

conclusion
In this study, the microbes in different parts of citrus trees were investigated by sequencing 16S rRNA genes. More than 2.5 million sequence reads were obtained from 60 samples and were assembled into OTUs. In total, 4733, 3520, 421, and 583 OTUs, respectively, were identified in samples from soil, roots, leaves, and phloem. The dominant microbial phyla of all samples from citrus trees were Proteobacteria, Actinobacteria, and Acidobacteria. The composition and structure of microbial communities in different plant tissues were analyzed using PCoA, CAP, Anosim and MRPP methods, and the species with a significant difference between groups were identified according to the P and Q values. Our results indicated that the microbial community in different groups were heterogeneous and complex. Indicator microbes for each group were identified based on their uniqueness among different sample groups. The microbial communities in different parts of citrus trees revealed in this study laid a foundation for future studies on microbial diversity and impact on citrus trees.

Materials and methods
Sampling sites and collection. The citrus orchard for sampling in this study was in Xinhui, Guangdong Province, China (22°47′N, 113°03′E). C. reticulata cv. Chachiensis was planted in this orchard with strict water and fertilizer management and pest control for three years. A total of 60 samples were obtained from citrus tissues including roots, leaves, and phloem as well as surrounding soil. Fifteen samples were obtained from each tissue, www.nature.com/scientificreports www.nature.com/scientificreports/ and samples from each tissue were referred to as a group. Samples were collected in the spring (March 28) of 2017. Fresh leaf and phloem samples were obtained from trees selected randomly. After collection, samples were immediately frozen in liquid nitrogen and stored at −80 °C. For root samples, roots were cut from trees and washed with sterilized water to remove the sediment for 5 times. After removing water with paper towels, roots were frozen liquid nitrogen. Soil was collected from 3-5 cm underground and immediately frozen in liquid nitrogen.

DNA extraction and PCR amplification.
Citrus tissues were ground with liquid nitrogen to powder for DNA extraction. DNA was isolated and purified using an E.Z.N.A. ® Stool DNA Kit (Omega Bio-tek, Norcross, GA, U.S.) according to the manufacturer's protocols 56 . The DNA quality and concentration were checked on a 1.0% agarose gel after electrophoresis and a NanoDrop ND-2000 spectrophotometer (Thermo Fisher Scientific, United States), respectively. For the amplicon library preparation, DNA was used as the template and the amplification of 16 S rRNA genes was performed by an amplified method 57  Bioinformatics analysis. After removing the adaptors, primers and low-quality reads, the pair-end reads were assembled into final sequences based on overlapping alignments. The criterion for overlapping was at least 10 bp overlap with a mismatch ratio of less than 0.2. Chimera tags were filtered out using the Gold database by UCHIME (version 4.2.40) 59 . Operational taxonomic unit (OTU) analysis was performed using the Uparse package (version 7.0.1001) with a 97% sequence identity 60 . Each OTU was taxonomically assigned based on the silva database using the RDP classifier 61 . OTUs matched to chloroplast sequences, chondriosome sequences, and unclassified sequences were removed. Only those OTUs with relative abundance > 0.001% (above three tags in at least one sample) in at least one sample were retained. Similarities and differences among microbial communities from different groups were analyzed by principal co-ordinates analysis based on the distance of Unweighted Unifrac 62 . Canonical analysis of principal coordinates was chosen for diversity analyses among different groups 63 . Differences in microbial community structures between groups were examined with the methods of Anosim and Multi Response Permutation Procedure 64 . Differences in microbial species between groups were identified using metastat software based on the Fisher exact test 65 .