Variation of the seed endophytic bacteria among plant populations and their plant growth‐promoting activities in a wild mustard plant species, Capsella bursa‐pastoris

Abstract Recent studies have revealed that some bacteria can inhabit plant seeds, and they are likely founders of the bacterial community in the rhizosphere of or inside plants at the early developmental stage. Given that the seedling establishment is a critical fitness component of weedy plant species, the effects of seed endophytic bacteria (SEB) on the seedling performance are of particular interest in weed ecology. Here, we characterized the SEB in natural populations of Capsella bursa‐pastoris, a model species of weed ecology. The composition of endophytic bacterial community was evaluated using deep sequencing of a 16S rDNA gene fragment. Additionally, we isolated bacterial strains from seeds and examined their plant growth‐promoting traits. Actinobacteria, Firmicutes, Alpha‐, and Gammaproteobacteria were major bacterial phyla inside seeds. C. bursa‐pastoris natural populations exhibited variable seed microbiome such that the proportion of Actinobacteria and Alphaproteobacteria differed among populations, and 60 out of 82 OTUs occurred only in a single population. Thirteen cultivable bacterial species in six genera (Bacillus, Rhodococcus, Streptomyces, Staphylococcus, Paenibacillus, Pseudomonas) were isolated, and none of them except Staphylococcus haemolyticus were previously reported as seed endophytes. Eight isolates exhibited plant growth‐promoting traits like phosphate solubilization activity, indole‐3‐acetic acid, or siderophore production. Despite the differences in the bacterial communities among plant populations, at least one isolated strain from each population stimulated shoot growth of either C. bursa‐pastoris or its close relative A. thaliana when grown with plants in the same media. These results suggest that a weedy plant species, C. bursa‐pastoris, contains bacterial endophytes inside their seeds, stimulating seedling growth and thereby potentially affecting seedling establishment.


| INTRODUC TI ON
Successful establishment of seedlings in novel environments is a major fitness component of weedy plant species and a critical factor determining species' geographic range (Crawley, 1997). Soil microbiota, especially microorganisms in plants' rhizosphere, is proposed to form a mutualistic interaction with plants and consequently facilitate weed establishment (Coats & Rumpho, 2014;Trognitz et al., 2016). Notably, plant seeds harbor diverse seed endophytic bacteria (SEB) (Truyens et al., 2015), so seeds and SEB of weed plants would disperse simultaneously. Since SEB would constitute microbiota inside or in the rhizosphere of seedlings (Kaga et al., 2009;Puente et al., 2009), SEB, in addition to soil microorganisms, likely influence seedling establishment in novel environments.
Despite their plausible ecological significance Jeong et al., 2021;White et al., 2018), relatively little information is available on the characteristics of SEB in weed plant species.
Most information on the SEB has come from studies using agriculturally important crop plants (Card et al., 2015;Pal et al., 2019;Shahzad et al., 2018). It should be noted that crop and wild plant species tend to have distinctive characteristics. Many crop plant species have a limited number of genotypes due to strong artificial selection, and their seeds are intensively managed to have consistent quality with pathogen-free status (Pérez-Jaramillo et al., 2016;White et al., 2019). Consequently, SEB of crop species are suggested to have lower diversity than wild plant species. Studies on the indigenous SEB of wild plant species are required to evaluate the ecological significance of SEB (Wassermann et al., 2019).
This study examined the bacterial community of seeds in one of the most common weedy species, Capsella bursa-pastoris (Brassicaceae) (Neuffer, 2011). C. bursa-pastoris is an annual weedy plant species worldwide, except for regions near the equator (Neuffer, 2011). As a model system of weed ecology, the adaptive divergence of C. bursapastoris along environmental gradients like altitude and latitude has been well documented (Huang et al., 2012;Neuffer, 2011).
Studies testing seed endophytes have often used DNA fingerprinting methods such as polymerase chain reaction-denaturing gradient gel electrophoresis (PCR-DGGE) or next-generation sequencing (NGS) Liu et al., 2019;Xu et al., 2014).
While such techniques have the advantage of detecting uncultivable bacteria, they provide scant information on the functional role of microorganisms in plant performance. In contrast, culturedependent methods can characterize the biological functions of isolated bacteria, although cultivable bacteria are only a subset of the whole microbiome. In particular, diverse assays can be conducted to evaluate whether SEB are able to produce so-called plant growthpromoting (PGP) molecules like indole-3-acetic acids (IAA), siderophores, 1-aminocyclopropane-1-carboxylic acid (ACC) deaminases, and phosphate-solubilizing molecules (Santoyo et al., 2016;Shahzad et al., 2018). Since culture-independent and -dependent methods provide complementary information on the SEB, this study used both methods.
Previous studies using crop plants suggested that the composition and function of SEB likely depend on cultivar genotypes and growing environments (Hardoim et al., 2012;Xu et al., 2014). Weedy plants occur in a broad geographic range with diverse environmental conditions. Distinctive genotypes tend to constitute natural populations in different environments (Linhart & Grant, 1996;Neuffer, 2011), implying that the composition and function of SEB in weedy species may be population specific. Thus, for a more general conclusion on the ecological role of SEB in weedy species, a comparative study at the population level was conducted.
Here, we collected seeds of C. bursa-pastoris populations along a latitudinal gradient in South Korea and characterized their SEB using both culture-independent and culture-dependent methods.
NGS technique was used to compare SEB community among C bursa-pastoris natural populations. In addition, SEB were isolated from seeds, and their PGP activities were assessed by examining the production of PGP molecules and the effect of SEB on seedling growth. Specifically, we addressed the following questions: (i) What kinds of endophytic bacteria inhabit inside seeds of C. bursapastoris? (ii) Do C. bursa-pastoris natural plant populations have differential seed bacterial communities? (iii) Do populations contain SEB-exhibiting PGP traits and promote seedling growth? 2 | MATERIAL S AND ME THODS

| Study populations and seed sources
We randomly chose four C. bursa-pastoris natural populations along a latitudinal gradient in South Korea in 2015 (Figure 1, Appendix S1).
Seeds were collected from at least 20 maternal genotypes in each population. Germinants of field-collected seeds were grown for one generation in a growth chamber, and seeds from these plants were used to examine SEB. In detail, field-collected seeds were sown in pots (8 cm × 7.5 cm × 6 cm) containing commercial soil medium (Shinsung Mineral Co. LTD, Kyunggi-do, Korea). The soil medium consisted of cocopeat (51.5% v/v), peat moss (10.0% v/v), zeolite (10.0% v/v), perlite (15.0% v/v), and vermiculite (13.0% v/v) with final pH of 5.0-7.0. Pots were maintained in a growth chamber at 25°C and a 12-hour light/dark photoperiod with 100 μmol/s/m 2 photosynthetically active radiation (PAR). Humidity inside the chamber was not controlled.
We randomly chose seven to eight maternal genotypes from each of the four C. bursa-pastoris populations and sterilized seed

T A X O N O M Y C L A S S I F I C A T I O N
Evolutionary ecology; Microbial ecology; Population ecology; Seed ecology surfaces following Johnston-Monje and Raizada (2011) with modification. Briefly, seeds were soaked in 0.1% Triton X-100 for 1 min, 3% sodium hydrochloride for 1 min, 70% ethanol for 1 min, and then washed three times with sterile distilled water. Surface sterilization was confirmed by incubating the sterilized seeds on potato dextrose agar (PDA) (# 213400, Difco Laboratories, Franklin Lakes, NJ, USA), R2A (# 218263, Difco), and Luria-Bertani (LB) agar (#7279, Acumedia, Lansing, MI, USA) at 25°C for a week. Only seeds that did not produce any microbial colonies were used to isolate bacterial endophytes and examine endophytic communities.
In order to visually confirm the occurrence of SEB, we used a fluorescence in situ hybridization (FISH) technique following Hewitson et al. (2010). Surface-sterilized seeds were fixed with 4% paraformaldehyde.
After adding 50 ng of oligodeoxynucleotide probe (EUB338) with dye, the slide glass was incubated in the hybridization buffer for 2 h at 45℃.
The slide was in the washing solution (0.9 M NaCl, 0.1% SDS, 20 mM Tris hydrochloride (pH 7.2)) for 30 min at 48℃ and washed twice with sterile distilled water. Pictures were taken using a Fluorescence microscope (EPI-Fluorescence & DIC Microscope, Carl Zeiss, Oberkochen, Germany) and edited using the iSolation FL auto program.

| Amplicon sequencing and clustering into operational taxonomic units (OTUs)
Equal amounts of seeds from seven individuals of each population were mixed to make three 500-mg samples. Each seed sample was homogenized using TissueLyser II (QIAGEN, Hilden, Germany).
DNA was extracted from each sample using DNeasy Plant Mini Kit (QIAGEN), and bacterial 16S rDNA was amplified. We used two methods to reduce chloroplast and mitochondrial DNA in the amplicon. First, we enriched bacterial cells in the homogenized seed mixture as described by Ikeda et al. (2009). Second, we amplified 16S rDNA using 799F (AACMGG-ATTAGATACCCKG) and 1193R (ACGTCATCCCCACCTTCC). As shown by Beckers et al. (2016), the 799F -1193R primer set amplifies mitochondrial and bacterial 16S rDNA but does not amplify chloroplast 16S rDNA. In addition, the size of amplified mitochondrial DNA (800 base pairs) was longer than Sequence reads were processed with the Mothur v.1.43 pipeline following MiSeq SOP (Schloss et al., 2009). To trim sequence reads, we used screen.seqs with the options of zero maxambig, five maxhomop, and 430 maxlength. This procedure produced 86,980 high-quality reads from 1,307,899 raw reads. Contigs were clustered into OTUs with 97% sequence identity and aligned using the SILVA nr_132 database (Quast et al., 2012). Approximately 20% of the raw sequence reads were removed because they were chimeric or nonbacterial sequences including chloroplast, mitochondria, archaea, and eukaryotes. Singletons were removed by remove.seqs in Mothur (Allen et al., 2016). OTUs other than cyanobacteria were designated in the phyla or classes of Proteobacteria for further statistical analyses following Beckers et al. (2016).

| Bacterial isolation and identification
To isolate SEB (Figure 2), surface-sterilized seeds were ground using a sterile mortar and pestle. The material was spread on five different solid media: LB agar, LGI (50 g/L sucrose, 0.01 g/L ferric chloride hexahydrate, 0.8 g/L potassium phosphate tribasic, 0.2 g/L magnesium sulfate heptahydrate, 0.002 g/L sodium molybdate dihydrate, 7.5 g/L agar; pH 7.5) for diazotrophic bacteria, R2A for oligotrophic bacteria, King's B agar (#60786, Sigma-Aldrich, St. Louis, MO, USA), and commercial agar used for cooking for slow-growing bacteria based on previous studies (Gagne-Bourgue et al., 2013;Johnston-Monje & Raizada, 2011). The plates were incubated at 25°C for a month during which a bacterial colony larger than 2 mm was picked F I G U R E 1 A photograph of Capsella bursa-pastoris (a) and seed sources (b) in the Korean Peninsula. The GPS coordinates of source populations are given in Appendix S1 for subculture (Gagne-Bourgue et al., 2013). A total of 18 morphologically different colonies from the plates were selected, subcultured twice, and then preserved in 20% glycerol stock solutions at -80°C until required.
Each PCR reaction contained a total of 50 μl consisting of 5 μl of 10× nTaq buffer, 5 μl of 5 μM dNTP, 1 U of nTaq polymerase (Enzynomics, Daejeon, Korea), 5 μl of each primer set, and 200 ng of template DNA. The reaction conditions were previously described (Coombs & Franco, 2003). PCR products were purified using the EZ-pure PCR purification kit (Enzynomics) and sequenced (Macrogen Inc.). To identify bacterial isolates, we aligned nucleotide sequences using MEGA 6.0 (Tamura et al., 2013) and compared them with previously reported sequences of bacterial type strains using EZBiocloud (Chunlab, Seoul, Korea). A bacterial isolate was assigned to a species with the highest 16S rDNA sequence similarity. The 16S rDNA sequences of the isolates have been deposited in NCBI GenBank, and their accession numbers are given in Appendix S2.

| Plant growth-promoting traits
We assessed four microbial traits that were suggested to promote plant growth directly (Glick, 2012). For assays, isolated bacterial strains were individually grown in test tubes containing 7 ml LB medium (#7178, Acumedia) at 28°C in a shaking incubator (210 rpm) for three days. Because all PGP assay procedures include incubation of bacterial strains at 30°C (see below), cells were cultured at 28°C before the assay to acclimate cells to 30°C. The bacterial cells were harvested by centrifugation and rinsed with sterilized deionized water (DW) twice. The cells were suspended in DW to the optical density (OD) of 1.2 at 600 nm, and triplicates of cell suspension were used for all assays.
The ability to solubilize inorganic phosphate was examined following Nautiyal (1999). Briefly, 20 μl of prepared bacterial solution was inoculated in 8 ml of National Botanical Research Institute's Phosphate (NBRIP) growth medium (10 g/L glucose, 5 g/L MgCl 2 ·6H 2 O, 0.25 g/L MgSO 4 ·7H 2 O, 0.2 g/L KCl, 0.1 g/L (NH 4 ) 2 SO 4 ; pH 6.75) containing insoluble tricalcium phosphate (5 g/L Ca 3 (PO 4 ) 2 ), and incubated at 30°C in a shaking incubator (150 rpm) for two days. NBRIP without bacterial inoculation was prepared as a negative control. Clear supernatant (100 μl) after centrifugation was transferred to a new clean glass tube with 4.2 ml of sterile DW, 500 μl of 2.5% ammonium molybdate in 5 N sulfuric acid, and 200 μl of α-amino-naphthol solution. The mixture was incubated at room temperature for 30 min, and their absorbance at 660 nm was measured using BioSpectrometer basic (Eppendorf, Hamburg, Germany). The phosphate level was estimated based on a standard curve that ranged from 0.1 to 2 mM of phosphate.
Siderophore production was quantitated following Schwyn and Neilands (1987) and Murakami et al. (2021). Prepared bacteria solution (20 μl) was inoculated in 8 ml of the 10 −2 diluted LB broth and incubated at 30°C in a shaking incubator (150 rpm) for four days. The supernatant (1 ml) was transferred to a glass tube with 700 μl of Chrome Azurol S (CAS) solution (0.165 g/L CAS, 0.082 g/L FeCl 3 , 0.397 g/L hexadecyltrimethylammonium bromide (HDTMA), in 100 mM piperazine buffer (pH 6.0) with 4 mM 5-sulfosalicylic acid). After incubation for one hour at room temperature, the absorbance was measured at 630 nm using the Biospectrometer basic. The percent siderophore unit (psu) was calculated as (Ar/As)/Ar × 100 where Ar is the absorbance of CAS solution mixed with uninoculated media and As is the absorbance of CAS solution mixed with the supernatant of each sample culture.
We quantitated IAA following Patten and Glick (2002) with modification. We inoculated 20 μl of a bacterial solution to 8 ml of LB broth containing 1000 mg/L of L-tryptophan (Sigma-Aldrich) and incubated the broth at 30°C in a shaking incubator (150 rpm) for two days. LB broth without bacterial inoculation was prepared as a negative control. After centrifugation, 1 ml of supernatant was recovered and mixed with 4 ml of Salkowski's reagent (11.5 M H 2 SO 4 , 9.2 mM FeCl 3 , in sterilized DW). After incubation for 30 min at room temperature, the absorbance at 535 nm was measured. IAA concentration was estimated based on a standard curve that ranged from 1 to 50 μg/ml IAA (Duchefa Biochemie, Haarlem, The Netherlands).
F I G U R E 2 Images of fluorescent in situ hybridization (FISH) to detect seed endophytic bacteria inside seeds of DEM population. Seeds were sectioned using Cryomicrotome, and the EUB338 probes (Macrogen, Daejeon, Korea) were hybridized. The white circle in the right image indicates the position of endophytic bacteria. (a) is an image magnified 400 −1 times by a light microscope. (b) is an image magnified 400× times with a fluorescence microscope ACC deaminase activity was determined following Penrose and Glick (2003) with modification. We inoculated 20 μl of bacterial solution into 7 ml of DF media (6 g/L Na 2 HPO 4 , 4 g/L KH 2 PO 4 , 2 g/L gluconic acid, 2 g/L citric acid, 0. HCl) was added and incubated at 30°C for 30 min. After the addition of 2 ml of 2 N NaOH, the absorbance at 540 nm was measured using the BioSpectrometer basic. The concentration of α-ketobutyrate was estimated based on a standard curve that ranged from 0 to 2 mM of αketobutyrate (Sigma-Aldrich). The quantity of the whole protein was estimated using Lowry's methods (Lowry et al., 1951).

| In vitro growth assay
To confirm whether the isolated endophytic bacteria are able to promote plant growth, we grew plant seedlings and isolated bacteria together in a single plate. Seedlings germinated from surface-sterilized seeds were used following previous studies (Dovana et al., 2015;Maggini et al., 2020). First, we examined the effects of bacteria on the model plant species, Arabidopsis thaliana Col line, as other studies often tested bacterial PGP activity using model plant species. After surface sterilization, seeds of A. thaliana were sown on a square plate with 0.2× Murashige and Skoog (MS) agar (#M022, Duchefa Biochemie, Haarlem, The Netherlands) containing 0.02% glycine, 0.5% inositol, and 0.5% sucrose. The plates were covered with aluminum foil and incubated at 25°C in the light for 16 h and 15°C in the dark for 8 h. Five days after incubation, 16 seedlings with a length of 1-2 cm were transferred to a new MS agar plate. The plates were incubated under the same conditions of the seed germination. Each bacterial isolate was streaked 7 cm below the root endpoint of seedlings in each plate, so that bacterial colonies did not contact plant seedlings (Figure 5e). Among 13 isolated bacteria, nine isolates were used for the experiment since four of them seldom grew in the MS agar. A plate that was not inoculated with any bacterial strain was a control. One-third of a Petri dish was not covered with aluminum foil for plants to receive light. Shoot and root lengths of seedlings were measured using a digital caliper at the time of streaking bacteria in each plate and after five days of incubation. The growth was calculated as the shoot or root length measured five days after bacterial inoculation minus those measured before inoculation.
We conducted the same procedures using C. bursa-pastoris seedlings. Seeds of ten maternal genotypes from each natural population were mixed and germinated in the same agar plates used for A. thaliana. Each plate had four seedlings from each plant population, resulting in a total of 16 seedlings. The plate inoculated with each bacterial isolate and a control plate without bacterial strain was replicated five times.

| Statistical analyses
All statistical analyses were performed using R version 3.6.1 (R Core

| Isolated endophytic bacteria and their PGP traits
A total of 13 cultivable bacterial species in six genera were isolated from four C. bursa-pastoris natural populations based on the similarities of partial 16S rDNA sequences (1203 -1439 nucleotides) (

| Differences in seed endophytic communities among natural populations
The major phyla inside the seeds of C. bursa-pastoris were Actinobacteria, Firmicutes, and Proteobacteria, which are also major bacterial phyla in other plant species (Card et al., 2015;Truyens et al., 2015). The number of OTUs and sequence reads were comparable to the values in previous studies examining seed bacterial communities in other plant species (Liu et al., 2019;Truyens et al., 2016).
Although plant species tend to have variable seed endophytic communities (Ikeda et al., 2009;Johnston-Monje & Raizada, 2011), it has been inconclusive whether natural populations of the same species also harbor differential seed endophytic communities. In C. bursa-pastoris, similar phyla constituted the seed bacterial community, but their proportion differed among natural populations ( Figure 3). Sixty of 82 OTUs from the sequence analyses occurred only in a single population (Figure 4b), and so did 11 of 13 cultivable seed bacteria (Table 1). These results suggest that bacterial species composition and their proportion in the seed endophytic community differ among plant natural populations.
Diverse factors affect the seed endophytic community. Plant genotype has been proposed to potentially shape the seed bacterial community (Gagne-Bourgue et al., 2013;Granér et al., 2003;Xu et al., 2014), while some studies could not detect its effect (Fürnkranz et al., 2012;Kukkurainen et al., 2005). In this study, the proportion of Actinobacteria differed between populations MOO and GUM ( Figure 3). Notably, different morphological and physiological traits were observed between populations MOO and GUM when plants were grown in a common environmental condition, suggesting that the two populations likely consisted of different genotypes (Choi et al., 2019). Genotypes of C. bursa-pastoris seem to affect the endophytic community of seeds.
Habitat difference in natural populations is another factor that potentially alters the seed endophytic community. For instance, environmental conditions such as soil type or environmental stress can change seed bacterial communities (Hardoim et al., 2012;Truyens et al., 2013). Soil microbial communities in plant habitats can additionally influence the seed endophytic community since bacteria in the soil can enter the plant body and migrate into the seeds (Bertani et al., 2016;Cope-Selby et al., 2017). In this study, we grew C. bursa-pastoris populations in the same growth chamber for one generation and used their seeds to examine endophytic bacterial communities. Thus, environmental factors would have a limited effect on the variable seed endophytic communities. In contrast, we cannot exclude the possibility that the natural habitats of our testing plant population might have differential soil bacterial communities, which might cause distinctive seed endophytic communities.

| Isolated endophytic bacteria from C. bursapastoris seeds and their PGP activities
Similar to previous studies, isolated cultivable bacterial strains par-  Despite this limitation, only one (S. haemolyticus D4) out of fifteen isolates was assigned to bacterial species that have been previously reported as seed endophytes (Najnin et al., 2014). The other strains except P. agarexedens G5 were known to inhabit the rhizosphere or plant tissues other than seeds (Beneduzi et al., 2008;Gopalakrishnan et al., 2015;Hong et al., 2015;Hu et al., 2012;Kim et al., 2012;Qin et al., 2017;Verma et al., 2016;Wang et al., 2018).
Given that SEB would be a subset of the bacterial community of phyllosphere or rhizosphere (Bertani et al., 2016;Cope-Selby et al., 2017), our result suggests testing more plant species would expand the current database of SEB.  . This study examined whether isolated strains could dissolve tricalcium phosphate to determine phosphate-solubilizing activity.
A limitation of this method is that it cannot measure the ability to dissolve metal phosphate, another common form of insoluble phosphate in soil (Bashan et al., 2013). Despite this limitation, the tricalcium phosphate method has been widely adopted in many studies as a useful screening method to identify candidate bacterial strains facilitating plant use of insoluble phosphates (Mahdi et al., 2021;Varga et al., 2020).
All bacterial isolates with phosphate-solubilizing activity produced 20.66-28.92% unit siderophores. Pseudomonas strains producing a similar quantity of siderophore were shown to promote the shoot and root growth in rice, wheat, or bottle gourd (Agrawal et al., 2017). Siderophores excreted from bacterial cells can sequester ferric ions to facilitate plant absorption (Glick, 2012). In addition, siderophores can act as chelating agents for Ca 2+ or metal ions that make phosphate ions insoluble (Wang et al., 2018). Siderophore production of isolates could synergistically contribute to their capacity of phosphate solubilization.
Six isolates exhibited a positive activity in the ACC deaminase assay. However, the value of ACC deaminase activity was lower than that of other Streptomyces spp. with 57-354 nmol α-KB/mg protein/h or Pseudomonas spp. with 1400-40870 nmol α-KB/mg protein/h (Agrawal et al., 2017;El-Tarabily, 2008). In response to environmental stresses, the ethylene level in a plant increases to reduce plant growth (Pierik et al., 2006). ACC deaminase can block the ethylene production of plants by cleaving ACC, the immediate precursor of ethylene, into α-ketobutyrate and ammonia (Glick, 2012;Penrose & Glick, 2003).
Consequently, ACC deaminase can ameliorate an adverse effect of environmental stress on plant growth. Notably, winter annuals like C.
bursa-pastoris often exhibit the stress avoidance strategy, such that they finish their life cycle before facing stressful environmental conditions like hot and dry summer (Crawley, 1997 Figure 5). In previous studies, strains of those species were isolated from the rhizosphere of plants and shown to promote the growth of crop plant species (Bhattacharyya et al., 2017;Boubekri et al., 2021;Subramaniam et al., 2020;Wang et al., 2018). This study showed that some strains of four species could migrate into seed.
It is well acknowledged that soil microbiome can facilitate weed establishment (Trognitz et al., 2016). However, the ecological significance of SEB in weedy species has gained attention just recently. For instance, SEB can contribute to the seed germination and growth in invasive Phragmites australis (White et al., 2018), drought resistance in Lactuca serriola (Jeong et al., 2021), and the competitive ability by antagonistic effects on the competitor species .
Although soil microbiome has been postulated as a major source of mutualistic microorganisms for weedy plant species (Trognitz et al., 2016), our results imply that SEB might be another source of mutualistic microorganisms. Even though they are phylogenetically close relatives (Beilstein et al., 2008), seedlings of A. thaliana and C. bursa-pastoris reacted differently to the same isolated strains. Given the complex interaction between host plant species and endophytic bacteria (Carvalho et al., 2016), the PGP activity of endophytic bacteria has been hypothesized to depend on the host plant species (Long et al., 2008).
Interestingly, a few bacterial strains isolated from C. bursa-pastoris exhibited the ability to promote the seedling growth of A. thaliana, suggesting that some endophytic bacteria can promote the growth of nonhost plant species. Endophyte host specificity does not necessarily exclude the possibility that endophytic bacteria can stimulate the growth of nonhost plant species. Although some endophytic bacteria could have PGP activities on a broad range of plant species, the mechanism of PGP activity likely differs among plant species (Long et al., 2008;Ma et al., 2011).
This study conducted in vitro seedling tests and showed that seven out of fifteen bacterial isolates stimulated shoot growth of either C. bursa-pastoris or A. thaliana. However, it should be noted that in vitro tests might have a limited power to evaluate the diverse effects of endophytic bacteria on plant growth. For instance, we did not directly infect bacteria into individual plants, but the effects of endophytic bacteria might depend on their location inside the plant body and the developmental stage of the host plants (Truyens et al., 2015). In addition, the in vitro seedling test cannot evaluate plant tolerance to biotic and abiotic environmental stresses. We are conducting additional in vivo studies to examine plant performance in natural environments using seeds directly inoculated by isolated bacteria. The results are expected to provide more clear evidence on the interaction between endophytic bacteria and plant performance.

| CON CLUS IONS
This study revealed that the bacterial communities inside seeds differed among C. bursa-pastoris natural populations, indicating the diversity of SEB should be evaluated at the population level.
Isolated bacteria from seeds produced a relatively high amount of PGP substances, and four strains stimulated shoot growth of C.
bursa-pastoris. This result implies that SEB in addition to soil microorganisms can be a source of mutualistic microorganisms facilitating seedling establishment that is a critical fitness component of weedy plant species.

ACK N OWLED G M ENTS
The

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