Core endophyte communities of different citrus varieties from citrus growing regions in China

The native microbiomes of citrus trees play important roles in plant health, with good communication between the native microbiome and the host plant. Here, we report on the native endophytes in 24 citrus varieties in nine citrus growing regions in China; some of the trees were healthy and others had asymptomatic or symptomatic huanglongbing, which is caused by the pathogen Candidatus Liberibacter asiaticus (CLas). We used culture-dependent analysis and characterized the isolates by partial 16S rRNA gene sequencing. The endophytes were compared between different citrus varieties, regions, and disease states (healthy, asymptomatic, and symptomatic). The total number of endophytes isolated from most of the citrus varieties was 104–106 CFU/g of leaves, but it differed significantly by disease state, with the highest numbers in the healthy leaves and the lowest in the symptomatic leaves (p < 0.05). Among the citrus varieties, the Valencia variety had the maximum number of endophyte species (22). The most dominant endophytes were Bacillus subtilis, B. velezensis, Curtobacterium luteum, and Microbacterium testaceum. The higher frequency of B. subtilis in the healthy/asymptomatic plants compared to the symptomatic plants suggests that it has a role in huanglongbing resistance. Native endophyte communities in various citrus varieties could be used to improve citrus growth and combat CLas.

Most plants are hosts to a diverse group of bacteria, known as endophytes, that do not harm the host and colonizing the internal tissues of plants without causing any immediate and overt negative symptoms 1 . Plant physical and physiological barriers need to be overcome for successful colonization by endophytes, with the exception of colonization by vertically transmitted endophytes. There is a clear distinction between pathogens and endophytes, as the latter do not harm or destroy plant cells to obtain resources. Putative pathogen effector proteins can act as important signatures of the divergence underlying host specialization [2][3][4] . However, there have been no significant molecular studies on the host specialization of endophytes that differentiate endophytes from pathogens 5 .
The rhizosphere contains rhizodeposits and root exudates, which are important residues for attracting microorganisms from the surrounding environment 6,7 . Bacterial endophytes mainly enter the host plant via colonization of root hairs 8 , and another important route involves attraction of endophytes by the exudates of leaf and stem surfaces 6 . Only adapted bacteria have the ability to enter plants through hydathodes, stomata, and wounds. Reductions in surface bacteria colonization can occur due to a lack of nutrients, ultraviolet light and, most importantly, desiccation 6,9 . Various bacterial endophytes with different colonization routes and specific bacteria-host interactions have been described in detail 9,10 . Several active and passive mechanisms are involved in the movement of endophytic bacteria from the rhizoplane to the cortical cell layer, and further colonization involves crossing the endodermis 6,11 . The internal plant compartments can be systematically colonized by bacterial endophytes using the xylem vascular system as the main route, and some bacterial endophytes colonize locally via intercellular spaces 10 . The holes in the perforation plates between xylem elements are large enough for endophytes to pass through 12 , but it may take several weeks for the vertical spread of bacteria through specific plants, and it remains unclear why this dissemination is so slow 13 . The optimal mechanisms by which specific endophytes reach specific parts or tissues of plants remains unknown. Sample processing and DNA extraction for detection of Candidatus Liberibacter asiaticus (CLas) pathogen. Samples were subsequently processed for PCR template preparation. DNA was extracted using the cetyl trimethylammonium bromide (CTAB) method with slight modifications 36,37 . Briefly, the midribs of leaves were separated and frozen in liquid nitrogen. The midribs of the nine leaves from a single sample pool were then macerated with a sterile pestle and mortar. The resulting powder was transferred into sterile Eppendorf tubes, CTAB buffer (1 ml) was added, and the samples were incubated in a water bath (65 °C) for 30-60 min. Next, phenol: chloroform: isoamyl alcohol (1:1:5, v/v) was added to each tube, which was then centrifuged (12,000 rpm, 10 min). After transferring the supernatant into a fresh tube, a mixture of isopropanol and sodium acetate (3 M) (1:1 v/v) was added and centrifugation was performed again (12,000 rpm, 10 min). Ethanol (70%) was used to wash the resulting pellet. The centrifugation was performed one additional time to remove impurities from the DNA. The residual ethanol was allowed to evaporate, and sterile distilled water was used to resuspend the DNA pellet. The DNA was stored at −20 °C until further use. qpcR for detection of CLas pathogen. CLas was detected using SYBR Green I reagent (Bio-Rad). PCR was performed in a 25-µl reaction mixture containing DNA template (5 µl), 1 × PCR buffer (SYBR Green Master Mix; Bio-Rad), and 0.8 µm CQULA04R and CQULA04F primers (which amplify the CLas-specific sequence of the ribosomal protein L12 [rplL] gene) 38 . A StepOne Real-Time PCR System (Applied Biosystems) was used with the following program: 95 °C for 1 min and 45 cycles each of 95 °C for 15 s, 59 °C for 15 s, and 72 °C for 45 s. During the extension step (72 °C for 45 s) of each cycle, the instrument collected the fluorescent signal generated by SYBR Green I (Bio-Rad) nonspecifically bound to any dsDNA. To analyze the specificity of the PCR amplification, a melt curve analysis was subsequently conducted using the following program: 95 °C for 1 min, 55 °C for 1 min, and then the temperature was increased by 0.5 °C every 10 s from 55 °C to 95 °C. The melt curve was plotted according to the manufacturer's instructions (Bio-Rad).
Construction of recombinant plasmid and standard curve. A target fragment (382 bp) of CLas was amplified using the CQULA03F/CQULA03R primer set 38 and electrophoresed on 2% agarose gel with UltraPower DNA stain (Bioteke Corporation). The target band was cut out and purified using a gel extraction kit (Omega) according to the manufacturer's instructions. The DNA was eluted in 20 μl Millipore pure water and 7.5 μl DNA with 2.5 μl Solution 1 (Takara) was used for ligation with the pMD18-T vector (Takara). The recombinant plasmid (pUC18-382) solution (10 µl) was used to transform 50 µl competent Escherichia coli TG1 cells (Takara) for 8-10 h at 4 °C. The resulting mixture was spread onto Luria Bertani (LB) plates containing ampicillin and positive clones were confirmed by PCR using CLas-specific primers 37 . The positive recombinant plasmid was extracted using a HiBind DNA Mini Column (Omega), following the instructions for the Plasmid DNA Mini Kit (Omega). Enzyme digestion with 4 µl Q-Pst1 and electrophoresis on 1% agarose gel was performed to confirm the insertion in the plasmid (Fig. S1). The recombinant plasmid was then sequenced and aligned using BLASTn. The plasmid standard solution was quantified with a NanoDrop spectrophotometer (Ultrospec 2100 pro, Amersham Biosciences). The solution was then diluted using 10-fold serial dilutions (to 10 −10 ) and the dilutions were subjected to RT-PCR to generate a standard curve (Fig. S2). Subsequently, a real-time thermal cycler could be used to automatically calculate the pathogen titer in the field samples. The unit of detection was fg µl −1 , which was converted into pathogen copy number/g of leaves. A melt curve was plotted according to the manufacturer's instructions (Bio-Rad) (Fig. S3).
Isolation of citrus native endophytes. The native endophytes were isolated from the healthy, asymptomatic, and symptomatic citrus leaves. Briefly, the citrus leaves were washed three times with tap water and then surface sterilized as reported previously 36 to avoid contamination of the analysis by surface bacteria. To confirm that the endophytes analyzed were native to the citrus leaves, a sterility check was performed by plating 100 μl of the water from the third rinse on Luria Bertani (LB)/Tryptic Soya agar medium. The leaves were then cut into four pieces (5-6 mm long), plated on LB/TSA) plates, and incubated for 48-96 h at 37 °C. Single colonies of bacteria recovered from each leaf fragment were selected, purified by repeated streaking, and stored in 50% glycerol in a −80 °C freezer. The bacterial endophytes were selected for further analysis by choosing all endophytes with unique morphology (based on colony shape and color). The total number of endophytes (CFU/g of leaves) for each citrus tree was also recorded. For each citrus variety, six plates each containing four leaf pieces were analyzed. Additionally, using citrus seedlings in a greenhouse, we confirmed that the endophytes could easily disseminate inside citrus leaves. www.nature.com/scientificreports www.nature.com/scientificreports/ Endophyte DNA extraction and PCR amplification. The bacteria were grown until midto late-log-phase (0.5-0.7 at OD 600 ) and 1 ml of the culture was centrifuged at 7,500 rpm for 10 min. DNA was extracted using the CTAB method with slight modifications 36 . The pellet was resuspended in Tris-ethylenediaminetetraacetic acid (EDTA) buffer and 525 μl phenol: chloroform: isoamyl solution (25:24:1, v/v) was added to the tube followed by centrifugation at 12,000 rpm for 15 min. An equal volume of chilled isopropanol was added to the resulting supernatant and the solution was centrifuged again at 12,000 rpm for 15 min. The pellet was resuspended in 50 μl distilled water and then left overnight at 4 °C. The presence and concentration of bacterial DNA was confirmed by running 5 μl of product on a 1.5% agarose gel. Purified DNA appeared as a defined band when visualized under ultraviolet light.
Identification of the endophytes was performed by amplifying the 16S rRNA gene from the selected bacteria 39 using IDB-PO 5′-GAAGAGTTTGATCCTGGCTCAG-3′ and 5′-IDB-P6 CTACGGCTACCTTGTTACGA-3′ primers 27 . The amplification conditions (repeated twice) were as follows: initial denaturation at 94 °C for 4.5 min followed by 30 cycles of denaturation at 94 °C for 40 s, annealing at 55 °C for 30 s, and extension at 72 °C for 1 min followed by a final extension at 72 °C for 10 min. The positive controls were the pathogenic bacteria of maize top rot, Klebsiella pneumoniae KpC4, and the clinical strain Kp138. Purified amplified products were cloned into the vector Top10 (Tiangen) for sequence analysis.
Phylogenetic analysis. Phylogenetic trees were constructed to determine the taxonomic relationships using Molecular Evolutionary Genetics Analysis software (MEGA7.0.21) and the maximum likelihood method based on the Kimura two-parameter model 40 . The dominant endophytic sequences obtained in this study have been submitted to the GenBank database (accession numbers MK618592-MK618638).
Statistical analysis. The data obtained from different provinces and different citrus varieties were subjected to analysis of variance (ANOVA) followed by Duncan's multiple range test. p < 0.05 was considered statistically significant. SPSS v21 (IBM) was used for the statistical analysis.

Results
Endophytic diversity in citrus varieties in different provinces. The native bacterial endophytic communities of healthy, symptomatic, and asymptomatic citrus leaves in nine citrus growing regions was assessed in 2016-2018, and Fig. 2 show the endophytic microbial diversity of different citrus varieties from different citrus growing regions in China. The endophyte communities varied between specific varieties from different sites and between different varieties from the same site. Diverse endophytic bacteria were recovered from the same and different varieties located in multiple site or individual sites, respectively. Fujian province had a large range of citrus varieties, which resulted in the maximum endophyte isolation frequency, with C. reticulata Blanco having more endophytes in Fujian. However, the number of endophytes depends on the number of citrus varieties in each specific location; for example, the high endophyte isolation frequency in Fujian province was due to the collection of samples from 11 citrus varieties. Fewer endophytes were recovered from C. grandis (L.) Osbeck cv. Guanximiyou in Fujian province. Chongqing province had only 30 endophytes. Among all of the citrus varieties www.nature.com/scientificreports www.nature.com/scientificreports/ from different provinces, Valencia Orange had maximum endophyte species (22). In addition, citrus varieties in different and same provinces were different in terms of endophyte recovery. The total number of endophytes isolated from most of the citrus varieties was 10 4 -10 6 CFU/g of leaves, with no significant differences between citrus varieties (Table S1).  Figure S4 shows how pure cultures of dominant endophytes were obtained by streaking individual colonies on LB agar. Phylogenetic analyses of the dominant isolates are shown in Fig. 3.

Detection of CLas.
All collected citrus leaves were investigated for the presence of CLas with CLas-specific primers 37 . The number of CLas copies was different between symptomatic and asymptomatic plants, with symptomatic plants having lower cycle threshold (CT) values and more CLas copies. No CLas was observed in healthy (uninfected) plants. After visually assessing the leaves as symptomatic or healthy/asymptomatic, the healthy/ asymptomatic leaves were distinguished based on CT values and the number of CLas copies, with CT values >32, 25-32, and <25 (10 0 -10 1 , 10 4 , and 10 6 CLas copies/g leaves) representing healthy, asymptomatic, and symptomatic states, respectively ( Table 2).
Comparison of endophytes based on disease state. The endophyte isolation frequency was significantly different between leaves with different disease states, with the highest numbers in the healthy leaves and the lowest in the symptomatic leaves (p < 0.05) (Fig. 4). Among the 114 bacterial endophytes isolated from the healthy leaves, the most dominant were B. subtilis, B. velezensis, C. luteum, S. endophytica, B. tequilensis, P. amylolyticus, and M. testaceum. Among the 41 endophytes isolated from the symptomatic trees, Bacillus sp. was the most dominant endophyte followed by Curtobacterium, and among the 58 endophytes isolated from the asymptomatic leaves, Bacillus sp. and B. megaterium were the most dominant endophytes. None of the other endophytes in healthy leaves were found in the symptomatic or asymptomatic leaves. Although Bacillus sp. was frequently isolated from all leaves, the isolation frequencies were considerably lower in the asymptomatic and symptomatic leaves compared to the healthy leaves. Most of the endophyte species were isolated at low frequencies from the various citrus varieties and regions. Endophyte species in leaves with different disease states. The dominant endophytes in different citrus varieties were compared, and B. subtilis was found to be the most frequent species in the healthy citrus plants and it was also recovered from a few asymptomatic and symptomatic trees (Fig. 5). B. subtilis showed significant differences between healthy trees and both symptomatic and asymptomatic trees (p < 0.05). The other dominant species across all disease states were Bacillus sp., B. velezensis, B. amyloliquefaciens, B. megaterium, B. tequilensis, C. luteum, Gammaproteobacterium symbiont of Plautia stali, and M. testaceum. In contrast, C. luteum, S. endophytica, P. amylolyticus, M. osloensis, and P. septica were frequently isolated only from healthy leaves. Interestingly, only Bacillus sp. was frequently recovered from asymptomatic leaves (p < 0.05), and it may provide resistance against CLas. (2020) 10:3648 | https://doi.org/10.1038/s41598-020-60350-6 www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
To our knowledge, this is the first study regarding the biogeographical diversity of endophytes isolated from citrus trees in most of the citrus growing regions in China. We isolated 213 endophytes from 24 citrus varieties. Healthy trees had more endophytes compared to symptomatic and asymptomatic trees, which may have been because the www.nature.com/scientificreports www.nature.com/scientificreports/ healthy trees were free of CLas. Previous studies reported that microbial communities in citrus trees are negatively affected by CLas 37,41,42 . In addition, microbial colonization of the branches, stems, roots, and leaves are affected by various factors. Microbial communities are present in the spatial environment inside plants, depending on their interactions inside the plants and the presence of pathogens 43,44 . Moreover, plant conditions pose a threat to the native microbial communities 36 . Similar findings using clone library and qPCR techniques were reported previously 45 . Pathogen infection of a plant drastically changes the native microbial communities and other potential beneficial microorganisms in the host. A previous study involving a clone library analysis revealed the various culturable bacteria in both CLas-infected and uninfected citrus roots with respect to recovery and frequency of   www.nature.com/scientificreports www.nature.com/scientificreports/ isolated bacteria. In our analysis, only citrus leaves were used to assess the endophytic diversity, which resulted in higher proportions of Actinobacteria and Firmicutes than those reported in other studies 46,47 .
We assessed the microbial diversity in leaves from all sampled citrus varieties and the endophyte isolation frequencies were maximum. Citrus leaves (rather than branches) from sweet orange and tangerine are the preferred niche from which to isolate endophytic bacteria 48 . The endophytic bacterial population native to citrus leaves has been reported to be diverse 49 . A previous microbial diversity assessment of symptomatic and asymptomatic CLas-infected Citrus sinensis groves revealed that citrus leaves have a large core microbiome 50 . In our study, Bacillus and Curtobacterium were dominant in symptomatic plants. Previous studies also found these bacteria in symptomatic and asymptomatic citrus plants, and they had promising effects on plant growth along with biocontrol abilities [51][52][53] .
Although CLas can colonize some plants without inducing any apparent HLB symptoms, the total number of endophytes were higher in these asymptomatic trees than in symptomatic trees. In contrast, a previous study reported lower endophytic diversity in Xylella fastidiosa-infected asymptomatic citrus plants compared to Xylella fastidiosa-infected symptomatic plants due to X. fastidiosa resistance in the former 48 . The Bacillus sp., Curtobacterium sp., Enterobacter sp., and Pantoea sp. found in our study were also reported in sweet orange and tangerine infected with X. fastidiosa (which causes citrus variegated chlorosis) in Brazil 36 , but the species in our study were different from those reported in HLB-infected citrus trees in Florida 50 . The differences could be due to differences in tissue samples (leaf, midrib, or branch), different environmental conditions (such as the weather), and the dominant HLB pathogen in each geographical area. We found that the dominant endophytic genus in most of the citrus varieties was Bacillus (few of these bacteria can fix nitrogen 54,55 , and they can colonize a diverse range of plants 36,48 ), and higher Bacillus density was associated with lower HLB severity (from healthy to asymptomatic and finally to symptomatic leaves). Endophyte colonization of citrus plants may depend on the HLB disease state (related to the CLas strain), with a potential synergistic interaction between endophytic Bacillus and CLas in order to mitigate HLB. However, as CLas is non-culturable in axenic cultures, we did not explore the interactions between the endophytes and CLas.
Several interesting bacteria were recovered from asymptomatic and healthy plants, indicating their potential association with HLB resistance. Another important endophyte recovered from healthy and symptomatic plants   www.nature.com/scientificreports www.nature.com/scientificreports/ was Curtobacterium, which has an important role against several plant pathogens due to antimicrobial production and its ability to induce systemic resistance 56,57 . Curtobacterium may be useful for limiting the CLas infection of symptomatic plants, but the mechanism needs further exploration. Bacillus and Curtobacterium in the inner parts of healthy and asymptomatic plants may enhance HLB resistance by producing antimicrobials or triggering a degree of CLas resistance. The frequent recovery of Bacillus and Curtobacterium from healthy and asymptomatic plants supported our hypothesis regarding possible colonization of the citrus plants with these endophytes.
Furthermore, we tested various citrus varieties to find the one with the largest number of endophyte species and to identify the endophyte species that were common among many citrus varieties. Valencia orange (C. sinensis) had the maximum number of endophyte species (22 species), followed by C. reticulata cv. Ponkan (14 species ., B. velezensis, B. amyloliquefaciens, B. megaterium, B. tequilensis, C. luteum, Gammaproteobacterium symbiont of Plautia stali, and M. testaceum. These dominant bacteria were also observed previously in maize 58 and citrus 59 . In contrast, C. luteum, S. endophytica, P. amylolyticus, M. osloensis, and P. septica were frequently isolated only from healthy citrus leaves. Several factors including environmental conditions, inoculum density, host developmental stage, and host species influence the endophytes found in specific plants 60 . A previous study also recovered Microbacterium, Lysinibacillus, Brevibacillus, and Variovorax more frequently 46 . We found that the total number of endophytes isolated from most of the citrus varieties was 10 4 -10 6 CFU/g of leaves. Endophyte colonization is regulated by factors such as plant growth stage, environmental conditions, seasonal variation, plant cultivar and, most importantly, plant genotype 61,62 . The role of specific endophytic communities in nature is an important clue to select endophytes with potential beneficial bioactivity.
Taking the results together, this is the first large-scale study in China showing diverse bacterial endophytic communities in various citrus varieties from nine citrus growing regions. The diversity of the citrus microbiomes in trees from different geographical areas was assessed. Bacillus subtilis and Bacillus sp. was the dominant genus in healthy and asymptomatic trees, respectively. The number of endophytes depended in each region on the number of citrus varieties. Moreover, the endophytic communities in healthy, asymptomatic, and symptomatic leaves varied, as reported previously 41 , revealed interactions among pathogenic and beneficial microbial communities inside citrus plants. The functional influences of endophytic communities on citrus plants need to be explored.