Comparison of specific endophytic bacterial communities in different developmental stages of Passiflora incarnata using culture‐dependent and culture‐independent analysis

Abstract Plants and endophytic microorganisms have coevolved unique relationships over many generations. Plants show a specific physiological status in each developmental stage, which may determine the occurrence and dominance of specific endophytic populations with a predetermined ecological role. This study aimed to compare and determine the structure and composition of cultivable and uncultivable bacterial endophytic communities in vegetative and reproductive stages (RS) of Passiflora incarnata. To that end, the endophytic communities were assessed by plating and Illumina‐based 16S rRNA gene amplicon sequencing. Two hundred and four cultivable bacterial strains were successfully isolated. From the plant’s RS, the isolated strains were identified mainly as belonging to the genera Sphingomonas, Curtobacterium, and Methylobacterium, whereas Bacillus was the dominant genus isolated from the vegetative stage (VS). From a total of 133,399 sequences obtained from Illumina‐based sequencing, a subset of 25,092 was classified in operational taxonomy units (OTUs). Four hundred and sixteen OTUs were obtained from the VS and 66 from the RS. In the VS, the most abundant families were Pseudoalteromonadaceae and Alicyclobacillaceae, while in the RS, Enterobacteriaceae and Bacillaceae were the most abundant families. The exclusive abundance of specific bacterial populations for each developmental stage suggests that plants may modulate bacterial endophytic community structure in response to different physiological statuses occurring at the different plant developmental stages.


| INTRODUC TI ON
Microbial endophytes are part of the plant micro-ecosystem, where they live internally without causing any damage or apparent symptom of a disease. These endophytes are ubiquitously associated with almost all plants (Nair & Padmavathy, 2014;Sharma, Kansal, & Singh, 2018). Endophytic bacteria colonize plant's intracellular or intercellular spaces and may originate from the phyllosphere, rhizosphere, or even seeds, existing in both free-living or endophytic states (Farrar, Bryant, & Cope-Selby, 2014). For the establishment of this plant-microbe interaction, plants constituted a complex ecosystem where they can provide necessary nutrients for microbial colonization. In return, endophytes perform diverse beneficial functions for the host-plant. They may directly affect the plant's development by making essential nutrients more available and modulating levels of phytohormones (Ryan, Germaine, Franks, Ryan, & Dowling, 2008;Tsavkelova, Klimova, Cherdyntseva, & Netrusov, 2006), or, as an indirect effect, through the synthesis of biomolecules, they may provide protection against abiotic and biotic stresses (Guo, Wang, Sun, & Tang, 2008;Strobel & Daisy, 2003). Thus, the plant may select its internal microbial population toward a specific ecological role to be played in this ecosystem (Hardoim et al., 2015;da Silva, Armas, Soares, & Ogliari, 2016). The plant-related factors known to determine the structure and composition of endophytic communities are the plant genotype, developmental stage, and crop environmental conditions (İnceoğlu, Salles, Overbeek, & Elsas, 2010; Overbeek & Van Elsas, 2008;Ren, Zhang, Lin, Zhu, & Jia, 2015;Yu, Yang, Wang, Li, & Yuan, 2015). Considering phenological aspects of plants, endophytic communities may also respond to seasonal conditions, as their hosts go through different developmental stages with each season.
Several methods have been progressively developed for analyzing the structure and composition of the host-associated microbial communities. Culture-dependent methods are suitable for functional studies of native species but are limited as it is estimated to recover <1% of the total bacterial diversity. It is known that conventional microbiological techniques select for specific groups that are able to grow under preestablished isolation conditions (Stewart, 2012;Vartoukian, Palmer, & Wade, 2010). In contrast, culture-independent methods may detect the occurrence of uncultivable, slowgrowing, or less abundant bacteria. These methods, generally based on 16S rRNA gene sequencing (Tringe & Hugenholtz, 2008), can be high throughput to assess the composition of bacterial communities in soil, water, air, or any environmental sample (An, Sin, & DuBow, 2015;Doherty et al., 2017;Janssen, 2006;Shokralla, Spall, Gibson, & Hajibabaei, 2012).
Passionflower is a tropical plant of the family Passifloraceae, mainly distributed throughout North and South America (Dhawan, Dhawan, & Sharma, 2004). In Brazil, the species grows into the vegetative stage (VS) from December to January, and the reproductive stage (RS) is (flowering and fruiting) from April to November (Fuentes, Lemes, & Rodríguez, 2000). It grows preferentially in well-drained soil, forming a climbing stem, three-lobed leaves, ovoid or globose fruits, and, due to the exotic appearance of its flower, it is recognized as the symbol for the "Passion of the Christ" (Miroddi, Calapai, Navarra, Minciullo, & Gangemi, 2013;Patel, Verma, & Gauthaman, 2009). P. incarnata has been widely used in traditional herbal medicine for treating anxiety, nervousness, constipation, dyspepsia, and insomnia. Nowadays, it is officially included in the national pharmacopeias from France, Germany, and Switzerland, also being monographed in the British Herbal Pharmacopoeia and the British Herbal Compendium (Dhawan et al., 2004).
Although its therapeutic aspects have been widely reported, only one study on P. incarnata fungal endophytes was performed (Seetharaman et al., 2017). The present study is the first one to evaluate the bacterial endophytic diversity associated with this medicinal plant.
The analysis of the structure of plant-associated bacterial communities in their different stages of development may establish a correlation between the occurrence of specific bacterial populations and physiological changes throughout the plant's development.
These plant-related conditions may play a critical role in the modulation of the endophytic communities. This study aimed at determining and comparing the structure and composition of cultivable and uncultivable bacterial endophytic communities to be found in the vegetative and RS of P. incarnata. This area is mountainous, with altitudes ranging between 700 and 900 m, displaying a humid subtropical climate with a mean temperature of 22°C in January and 22.6°C in April. Regarding its phenology, P. incarnata is typically in the VS in January, while in April it develops into its RS. Thirty healthy plants were randomly sampled in April 2015 for the RS, and these plants were flagged for the next sampling. In January 2016, sampling from the previously flagged plants was carried out, but these were in the VS. Sterilized gloves and scalpels were used to collect the whole leaves; the blades were changed between each collection. The samples were placed in sterilized polythene bags, transported to the laboratory on ice, and stored at 4°C until they were ready to be processed up to 72 hr afterward. The leaves were detached with a sterilized scalpel, washed with purified distilled water, and left 10-15 min to drain. Surface sterilization was performed on whole leaves according to Azevedo, Maccheroni, Pereira, and Araújo (2000), with modifications. Leaf tissues were treated with 100% ethanol for 3 min, followed by 2% sodium hypochlorite for 2 min, and 70% ethanol for 3 min. The disinfected leaves were washed three times with sterilized distilled water, and the last washing was inoculated on nutrient agar plates to validate the effectiveness of the surface sterilization procedure. Control agar plates incubated at 28 ± 2ºC were inspected for 48 hr to check the occurrence of any bacterial growth.

| DNA extraction and 16S rRNA gene sequencing
The genomic DNA of the endophytic bacterial strains was extracted using the methods described by Van Soolingen, Haas, Hermans, Groenen, and Embden (1993), with modifications. The 16S rRNA gene was amplified using universal bacterial 16S ribosomal gene primers 10F (5′-AGTTTGATCCTGGCTCAG-3′) and 1525R (5′-AGTTTGATCCTGGCTCAG-3′) (Lane, 1991) targeting the V1-V9 region. The 25 µl PCR reaction mixture contained 10 ng of DNA, 0.5 µl of dNTP mix (10 mM; Applied Biosystems), 2.5 µl of 10X PCR Buffer with 15 mM MgCl 2 (Applied Biosystems), 0.5 µM of each primer, one unit of Taq DNA polymerase (Applied Biosystems). The PCR conditions consisted of initial denaturation at 95ºC for 2 min, followed by 35 cycles of 94ºC for 1 min, 60ºC for 1 min, 72ºC for 3 min, and a final extension at 72ºC for 5 min. Agarose gel electrophoresis separated the PCR products, purified using GFX TM PCR DNA and Gel Band Purification kit (GE Healthcare Life Sciences, Germany), and sequenced on an ABI3500XL Series (Applied Biosystems) sequencer.
The primers above mentioned were used to assembly the 16S rRNA gene sequence, and the 1100R (5′-AGGGTTGGGGTGGTTG-3′) was used as an internal sequencing primer. For taxonomic assignment of bacterial strains, the 16S rRNA gene sequences were compared with the EZBiocloud 16S Database using the "Identify" service (Yoon et al., 2017), and species assignment was based on closest hits (Kim & Chun, 2014).

| DNA extraction and Illuminabased sequencing
The leaf samples obtained from the same plants used for the isolation of cultivable bacterial communities were sterilized using the same conditions described previously. Sterilization was confirmed by running a PCR with the same primers previously used on the last washing, and if no DNA was detected after the amplification, the sterilization was considered successful. The sterilized leaf tissues were homog-

| Processing of sequencing data
Raw sequence data were checked with sequence quality filters in FastQC software (Andrews, 2012). Sequences of lengths < 150 bp were removed, and the adapter, barcodes, and primers were trimmed using Trimmomatic software (version 0.36) (Bolger, Lohse, & Usadel, 2014). The sequencing data were processed using Quantitative Insights into Microbial Ecology (QIIME) software version 1.9.1 (Caporaso et al., 2010). All sequences that passed quality controls were clustered in operational taxonomic units (OTUs) at 97% genetic identity using an open reference approach (UCLUST algorithm) (Edgar, 2010). A representative sequence for each OTU was classified using Ribosomal Database Project classifier (Wang, Garrity, Tiedje, & Cole, 2007) and PyNast aligner (Caporaso et al., 2009) against the SILVA database (128 release) for taxonomy assignment (Quast et al., 2012). The chimeras were checked and filtered out by UCHIME (Edgar, Haas, Clemente, Quince, & Knight, 2011). OTUs assigned to chloroplasts or of mitochondrial origin were excluded.
Only OTUs of bacterial origin were considered for further analysis.

Rarefaction curve, alpha-diversity indices (Shannon-Wiener
Index, Simpson's evenness Index) and richness estimators (Abundance-based Coverage Estimator and Chao1) were calculated using the QIIME pipeline. The index estimator Chao was used to estimating the richness of the bacteria. The Shannon diversity and Simpson index were used to estimate the biodiversity of the bacterial communities. In order to calculate the diversity indices, each sample was rarified to an average sequences' depth, due to the variation in number obtained per sample (de Cárcer, Denman, McSweeney, & Morrison, 2011). In this study, the OTU table was rarefied to 404 sequences, corresponding to the sample with the lowest number of sequences (RS). We normalized this table of good reads by dividing the reads per OTU in a sample by the sum of good reads in that sample, resulting in a table of relative abundances (frequencies). All diversity indices and richness estimators were calculated 10 times.
Unassigned sequences were excluded from the determination of contributions of taxonomic groups in each bacterial community.
The structure of bacterial endophytic communities was visualized in Krona graphs, plotted using the Krona web interface software (Ondov, Bergman, & Phillippy, 2011).

| Statistical analysis
In order to search for biologically meaningful differences in the taxonomic distribution between endophytic bacterial communities in VS and RS, the two-way Fisher's exact test with a Storey False Discovery Rate multiple test correction analysis (adjusted qvalue < 0.05 and ratio of proportions effect size < 2.00) was carried out using the graphical software package Statistical Analysis of Taxonomic and Functional Profiles (STAMP) (Parks & Beiko, 2010

| Culture-dependent diversity analysis
In total, 204 pure cultures showing different colony morphologies were obtained; 146 were retrieved from the RS and 58 from the VS. No colonies emerged from the final washing of the sterilization procedure, an assurance that the surface sterilization procedure was effective. All isolates were identified based on 16S rRNA gene sequencing and alignment. The sequence data for these isolates have been submitted to the GenBank database under accession numbers from MG778707 to MG778907. Most of the sequences showed > 99% similarity to the reference strains of EzBiocloud database; only 10.3% showed a similarity between 97% and 99%. The results revealed a high diversity, distributed in 84 different bacterial taxa (Appendix Table A1). In the RS, Proteobacteria was the most abundant phylum, comprising 68.5% of total isolates. Alphaproteobacteria was represented mainly by Sphingomonadaceae (32.9%) and Methylobacteriaceae (13.7%), followed by Rhodobacteraceae (4.6%), Bradyrhizobiaceae (1.4%), and one strain of Caulobacteraceae.  (Figure 3).

| 5 of 16
GOULART eT AL.  The main disadvantage of culture-based techniques is that they typically allow for the detection of no more than 0.1%-10% of true bacterial diversity within an ecosystem (Handelsman & Smalla, 2003;Pace, 1997), compared to the diversity obtained from culture-inde-  (Eevers et al., 2015). In the RS, the highest species richness was obtained on the Glycerol-Asparagine medium, which contains glycerol as carbon source, asparagine as amino acid source, and various trace minerals; this combination of ingredients allowed for the isolation of a wide range of bacterial species belonging to the Sphingomonadaceae, Methylobacteriaceae, and Microbacteriaceae families (Huang et al., 2012;Li et al., 2005;Veyisoglu et al., 2013).
Their growth rate may also have contributed to the predominance of these taxonomic groups in the culture media mentioned above.
Regarding the culture-independent approach, rarefaction analyses from both samples suggested that the bacterial diversity in the RS was lower than that in the VS. Another possible bias affecting the culture-independent analysis was the high percentage of sequences of plastid and mitochondrial origin yielded, though specific primers were used to avoid the amplification of plant organelle sequences The coverage percentage, richness estimators (ACE and Chao1) and diversity indices (Shannon and Simpson) were calculated using Quantitative Insights into Microbial Ecology pipeline.
TA B L E 1 Number of OTUs and alpha-diversity indices of the endophytic bacterial communities associated with Passiflora incarnata F I G U R E 3 Krona plots on 16S rRNA sequences of the bacterial communities associated with Passiflora incarnata leaves. The data represent taxonomic hierarchies of bacterial communities in the (a) vegetative stage and in the (b) reproductive stage in a multilevel diagram (Chelius & Triplett, 2001). Similar results were found in a study on endophytic bacterial communities from banana shoot-tip tissues (Yashiro et al., 2011). Additionally, 15% of the overall sequences from both the vegetative and RS corresponded to unassigned OTUs.  1990). Additionally, a study led by Andreolli, Lampis, Zapparoli, Angelini, and Vallini (2016) showed that species richness in an endophytic bacterial community associated with Vitis vinifera cv. Corvina is higher on 3-year-old grapevines than on 15-year-old ones. These changes in diversity may occur due to the loss of "passenger" endophytic populations within senescent leaves and, consequently, they may lead to the permanence and establishment of endophytes with critical ecological roles for the most advanced plant developmental stages. On the other hand, the decrease of nutrients in more advanced developmental stages may also make the host-plant a less attractive niche for endophytic colonization, since in many conifers it was reported that mineral and sugar contents change as leaves age (Distelbarth, Kull, & Jeremias, 1984).
The study of host-associated microbial community composition and structure may elucidate the ecological role that each microbial group plays within the phytobiome. Moreover, host development and health are dependent on the presence of an entire microbial community (Robinson, Bohannan, & Young, 2010 (Rhoden et al., 2015). Besides being commonly characterized as endophytes (Govindasamy et al., 2010), Bacillus and Pseudomonas play a critical role in the promoting plant growth (Adesemoye, Obini, & Ugoji, 2008;Mercado-Blanco & Bakker, 2007;Pérez, Collavino, Sansberro, Mroginski, & Galdeano, 2016 than half of all bacteria isolated from the RS. The dominance of these three genera was also found in a study on cultivable endophytic bacteria associated with yerba mate (Ilex paraguariensis) (Araújo et al., 2002). Additionally, some previous studies showed that the occurrence of Curtobacterium and Methylobacterium has a particular influence on the acquisition of resistance to diseases caused by the phytopathogenic bacteria Xylella fastidiosa (Lacava, Araújo, Marcon, Maccheroni, & Azevedo, 2004;Sturz & Matheson, 1996) and Erwinia caratovora var. atroseptica (Ren et al., 2015), which means they may contribute to host-plant health.  Sturz & Matheson, 1996). This might explain why predominant taxonomic groups in the RS are heavily related to bacterial groups that have previously shown influence on resistance to some infectious diseases (Araújo et al., 2002;Lacava et al., 2004;Pérez et al., 2016;Sturz & Matheson, 1996). Further studies are needed to assess host-endosymbiont metabolomics at different developmental stages and determine whether the structure and composition of the endophytic bacterial communities could correlate with the plant phenological patterns.

| CON CLUS IONS
This study revealed the existence of differentiated communities according to the developmental stage of the plant. Both the culture-dependent and culture-independent approaches showed that specific bacterial populations were exceptionally abundant for each developmental stage, which may be due to endophyte selection being driven by physiological changes (such as nutritional requirements or susceptibility to pathogens) occurring during the host development.

ACK N OWLED G M ENTS
This study was funded by Fundação de Amparo à Pesquisa do

CO N FLI C T O F I NTE R E S T S
The authors declare that there is no conflict of interest.

AUTH O R CO NTR I B UTI O N S
FF-G, DA-A supervision, funding acquisition and resources. MCG, LGCY, KJHM formal analysis. MCG, LGCY conceptualization, visualization, writing-original draft preparation and writing-review and editing.

E TH I C S S TATEM ENT
None required.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data are provided in the results section of this paper. The sequence data for the isolates are available at www.ncbi.nlm.nih.gov/ genba nk/ under accession numbers from MG778707 to MG778907.