Environmental interactions are regulated by temperature in Burkholderia seminalis TC3.4.2R3

Burkholderia seminalis strain TC3.4.2R3 is an endophytic bacterium isolated from sugarcane roots that produces antimicrobial compounds, facilitating its ability to act as a biocontrol agent against phytopathogenic bacteria. In this study, we investigated the thermoregulation of B. seminalis TC3.4.2R3 at 28 °C (environmental stimulus) and 37 °C (host-associated stimulus) at the transcriptional and phenotypic levels. The production of biofilms and exopolysaccharides such as capsular polysaccharides and the biocontrol of phytopathogenic fungi were enhanced at 28 °C. At 37 °C, several metabolic pathways were activated, particularly those implicated in energy production, stress responses and the biosynthesis of transporters. Motility, growth and virulence in the Galleria mellonella larvae infection model were more significant at 37 °C. Our data suggest that the regulation of capsule expression could be important in virulence against G. mellonella larvae at 37 °C. In contrast, B. seminalis TC3.4.2R3 failed to cause death in infected BALB/c mice, even at an infective dose of 107 CFU.mL−1. We conclude that temperature drives the regulation of gene expression in B. seminalis during its interactions with the environment.

blight and anthracnose spots in addition to postharvest rot 23 . Pineapple disease, wilt and postharvest black rot are plant diseases caused by Ceratocystis fimbriata and Ceratocystis paradoxa in economically important crops such as sugarcane 24,25 . Therefore, the use of Burkholderia species as a biocontrol agent may prevent economic losses in a diverse range of crops with important global implications. To date, the interactions of B. seminalis with its several hosts have been poorly studied.
Many Bcc members are human pathogens; therefore, testing the virulence of the TC3.4.2R3 strain is crucial to validate its possible use as an agricultural biocontrol agent. Larvae of the greater wax moth Galleria mellonella have been widely used as a non-mammalian model to investigate bacterial pathogenesis because the innate immune system of this insect is similar in many aspects to that of the mammalian innate system 26 . Comparison of bacterial pathogenicity in G. mellonella and mouse infection models is important to develop our understanding of the interactions of B. seminalis with different hosts.
Temperature is a central trigger for the control of gene expression in bacterial pathogens 27 . For example, differentially expressed virulence factors, such as toxins, are affected by temperature in Pseudomonas syringae 28 and Yersinia entomophaga 29 . Motility, flagellar expression and biofilm production are often thermoregulated, for example, P. syringae 28 and Campylobacter jejuni 30 . Furthermore, surfactant and exopolysaccharide (EPS) production appears to occur in a temperature-dependent manner in Pseudomonas putida 31 and Erwinia amylovora 32 , respectively. Many genes involved in plant-microbe interactions 28 , insect-microbe interactions 29 , microbe-microbe interactions 33 and biocontrol potential 34 are also thermoregulated. Given that B. seminalis is a member of the Bcc that could have biocontrol applications, this study was undertaken to investigate the thermoregulation of B. seminalis strain TC3.4.2R3 in various hosts.

Thermoregulation of biofilm formation, EPS and motility in B. seminalis TC3.4.2R3.
To understand the role of temperature variation in strategies used by B. seminalis TC3.4.2R3 during ecological interactions, biofilm and EPS production, growth rate and swimming motility were evaluated (Fig. 1). We assessed the ability of B. seminalis TC3.4.2R3 to form biofilms using the MBEC (Minimum Biofilm Eradication Concentration) biofilm device of Innovotec Inc. (Edmonton, AB, Canada). Biofilm formation at 28 °C was 2.5-fold greater than that at 37 °C (Fig. 1a), and EPS production was 40% greater (p < 0.05) at 28 °C than at 37 °C (Fig. 1b). However, Effect of temperature on antagonism against plant pathogenic fungi. The ability of B. seminalis TC3.4.2R3 to inhibit the phytopathogenic fungi F. oxysporum, C. paradoxa, C. fimbriata and Colletotrichum spp. was evaluated at 28 °C and at 37 °C (Fig. 2). The ability of B. seminalis to inhibit in vitro phytopathogenic fungi was significantly higher at 37 °C than at 28 °C (Fig. 2a,b). In addition, we observed that B. seminalis produced less yellow pigment at 37 °C than at 28 °C (Fig. 2c) after 48 h of growth (Fig. 2a,b). Although fungal inhibition was observed when the bacteria were grown on plates for 24 h, the inhibition halo was twofold higher when grown for 48 h (Fig. 2a,b). www.nature.com/scientificreports www.nature.com/scientificreports/ higher virulence in the wax moth larvae infection model at 37 °C than at 28 °C, we proceeded to use the BALB/c mouse infection model to evaluate virulence in a mammalian model.

Effects of
The pathogenicity of B. seminalis to BALB/c mice was evaluated up to 30 days with intraperitoneal inoculations of 10 7 , 10 6 , 10 5 , 10 4 or 10 3 CFU.mL −1 as the final concentration of bacteria. For all experiments, no signs of disease were observed, even at the highest bacterial inoculum.
RNAseq analysis of B. seminalis TC3.4.2R3 under different temperatures. We used RNA sequencing and comparative transcriptomic analysis to identify genes and their respective expression levels at 28 °C and 37 °C in B. seminalis TC3.4.2R3 and to identify genes that potentially contribute to temperature and phenotype changes in B. seminalis TC3.4.2R3. Although we used environmental (28 °C) and clinical (37 °C) temperatures, the gene expression should be more related to antifungal production than to pathogenesis, since after 48 h of growth, the bacterial culture was in stationary phase. As previously presented, although there was fungal inhibition by B. seminalis grown for 24 h, the higher antifungal production occurred after 48 h of growth ( Fig. 2b) when the 28 °C and 37 °C cultures were in stationary stage. As a consequence of these observations, RNAseq was performed at these time points and temperatures.
Through the generation of more than 35 million reads per sample, we were able to map approximately 2 Gb of the TC3.4.2R3 genome sequence to the corresponding sequences in the B. seminalis TC3.4.2R3 genome (GenBank Accession Number LAEU01000000) that includes three genetic elements: chromosome 1 (3.50 Mb), chromosome 2 (3.05 Mb) and a mega-plasmid (1.09 Mb). Table 1 displays the details of the yield and quality of Illumina sequencing per sample. The average depth was 235X coverage for 28 °C samples and 233X coverage for 37 °C samples.
The RNA libraries were of high quality (Fig. 4), showing that 6912 (99.9%) of the 6917 genes were expressed. Only the pyochelin cluster was completely repressed at both 28 °C and 37 °C. To classify differentially expressed genes, we adopted a log2 fold-change (FC) above 1. Using this cutoff, 589 genes were differentially expressed, with 583 upregulated and six downregulated at 37 °C relative to 28 °C. Among these, 141 differentially regulated genes encoded hypothetical proteins, suggesting that these unknown genes could be related to adaptation to various environmental conditions. The downregulated genes encoded flagellin (Bsem_00099) and five hypothetical proteins (Supplementary Table S2). Functional classification of upregulated genes showed that these genes encode core functions such as energy metabolism, transport, regulatory proteins, cellular processes, and aromatic compound degradation and detoxification (Fig. 5).
The genes upregulated at 37 °C included 65 genes encoding translation and ribosomal proteins, 53 related to transcription regulation, 52 related to energy metabolism, 34 related to transport/secretion systems and 31 related to amino acid biosynthesis and metabolism. Growth at 37 °C also resulted in the upregulation of genes involved in responses to stress and detoxification (26 genes), DNA replication, repair and structure (20 genes), posttranslational modification and protein folding (19 genes), fatty acid biosynthesis and metabolism (15 genes),  www.nature.com/scientificreports www.nature.com/scientificreports/ nucleotide biogenesis and metabolism (15 genes) and aromatic compound degradation (12 genes). Furthermore, 12 heat shock proteins and chaperones were upregulated at 37 °C, as well as genes encoding cofactors and vitamin metabolism (10 genes), terpenoids and polyketide metabolism (nine genes), virulence factors (eight genes), cold shock (six genes), signal transduction systems (six genes), toxin-antitoxin system and toxins (five genes), sulfur metabolism (four genes), flagella and pili (four genes), polyhydroxyalkanoate (PHA) biosynthesis (four genes), polysaccharide biosynthesis and metabolism (four genes), quorum-sensing (two genes) and lipopolysaccharide (LPS) biosynthesis (two genes) (Fig. 5). Furthermore, other genes, such as methyltransferases, cell division protein FtsB, GTP-binding proteins and AAA-ATPases, were also upregulated.

Secondary metabolite clusters of B. seminalis and temperature-dependent expression.
A homology search for genes encoding known proteins with motifs involved in secondary metabolism and antibiotic production in antiSMASH revealed four putative clusters in chromosome 1, including genes encoding a nonribosomal peptide synthetase (NRPS) cluster that shares 23% amino acid similarity to cupriachelin, a terpene cluster, an arylpolyene cluster and a type I polyketide synthase (T1PKS) cluster that shares 16% amino acid similarity to a galactoglucan protein. Expression of the NRPS cluster was not regulated by temperature. The terpene cluster had 40% of the upregulated genes (1.30-to 3.21-fold), arylpolyene and T1PKS clusters, present in the wcb gene cluster, had 10% (1.00 to 1.60-fold) and 20% (1.16-to 1.66-fold) of upregulated genes, respectively (Supplementary Table S5).
The downregulated hypothetical protein Bsem_04337 composed an operon of five hypothetical proteins within the bacteriocin biosynthetic cluster. Inside the NRPS cluster, pyochelin biosynthetic genes, siderophores with antimicrobial activity were found, and their expression was not differentially regulated under the test conditions. Interestingly, five genes of the core pyochelin cluster were no-reads (Bsem_05509 to Bsem_05513), while the other six genes had no variation in expression level at the evaluated temperature (Supplementary Table S7).

Discussion
Significant changes in gene expression in B. seminalis TC3.4.2R3 were observed in morphological traits comparing bacteria grown at 28 °C and 37 °C (Fig. 7). The bacteria increased biofilm formation, reduced motility, increased antimicrobial production and capsule biosynthesis and decreased virulence to G. mellonella at 28 °C, suggesting that the genes associated with these phenotypes could be important during interactions with the soil microbiota or during association with the plant host.
In the present study, B. seminalis TC3.4.2R3 produced more biofilm at 28 °C than 37 °C (Fig. 1a). Previous studies have shown that biofilm formation precedes endophytic colonization in plant tissues 35 , which should occur at temperatures less than 30 °C. In contrast, Burkholderia species, such as B. pseudomallei strain 1026b, produce more biofilm at 37 °C in comparison to 30 °C 36 , suggesting that for pathogenic bacteria, biofilm formation should be induced close to host temperature. These results suggest that for the endophytic B. seminalis TC3.4.2R3, the increased biofilm production at 28 °C could have a beneficial role in plant host colonization but not in virulence against G. mellonella.
The synthesis of EPS by B. seminalis TC3.4.2R3 also occurred in a temperature-dependent manner (Fig. 1b) and was greater at 28 °C than 37 °C. EPS is a major component of the biofilm matrix in many bacteria. EPS production was consistent with greater biofilm production at 28 °C. The production of EPS by B. kururiensis is regulated in response to growth and external conditions 37 . Similar results were observed for B. tropica, with biofilm formation affected in a temperature-dependent manner up to 24 h 38 . Interestingly, even though the amount of biofilm was greater at 28 °C, bacterial growth was higher at 37 °C in B. seminalis strain TC3.4.2R3 (Fig. 1d), indicating no association between higher growth and increased biofilm formation.
Motility can be beneficial for bacteria in terms of nutrient acquisition, translocation in specific hosts to access optimal colonization sites, avoidance of noxious agents and dispersal in the environment 39 . We found that B. seminalis was more motile at 37 °C than at 28 °C, suggesting that flagellar synthesis or function was thermoregulated (Fig. 1c). The flagellar apparatus presents a complex modulation, as suggested for B. thailandensis, which contains flagellar genes up-and downregulated at 37 °C 27 . In B. seminalis TC3.4.2R3, the flagellin gene fliC was downregulated, while the flhD, flhC and fliL genes were upregulated at 37 °C, suggesting differential regulation of these genes. Similar results were found in B. pseudomallei strain 153: more motility was observed at 37 °C than at 25 °C and at 30 °C more so than at 42 °C, although fliC was downregulated (−2.60-fold) 40 .
Increased flagellar activity and adaptations to warmer temperatures require a significant amount of energy to support bacterial growth and function. Interestingly, not only energy production under aerobic conditions was favoured, characterized by the over-expression of the F0/F1 ATP synthase but also anaerobic pathways, such as alcohol fermentation. Since cytochrome genes are induced, it is expected that a microaerobic environment is created because cytochromes have high affinity for oxygen, triggering oxygen depletion in Burkholderia 41 . For B. seminalis TC3.4.2R3 grown at 37 °C, genes encoding chaperons and heat-shock proteins, universal stress proteins and several genes involved in cellular detoxification were upregulated to protect and ensure cell survival, as also observed in B. pseudomallei strain 153 at increased temperature 40 .
Furthermore, we found a cluster of PHA biosynthesis upregulation (Bsem_01251 to Bsem_01254) at 37 °C. Polyhydroxyalkanoates (PHA) are energy-reserve polymers produced by bacteria for surviving periods of starvation in natural habitats and possibly confer a competitive advantage by allowing storage and mobilization of carbon in the face of environmental changes 41,42 . This result, in combination with the energy generation, motility and growth curve (Fig. S1), suggests that at 37 °C, B. seminalis TC3.4.2R3 generates more energy to sustain increased motility, but as the growth rate remains similar, the surplus energy could be stored in the form of PHA. www.nature.com/scientificreports www.nature.com/scientificreports/ The increased expression of genes involved in primary metabolism, including nucleotide biosynthesis, cell division, RNA polymerases, ribosomal proteins and protein processing, indicates the high activity and energy-requiring cellular processes needed at 37 °C. Due to the accelerated activity, it is possible that cells enter a starvation period when grown at 28 °C; therefore, protein synthesis may be required during the starvation period to mediate nutrient scavenging, heat-shock response and recovery from starvation 43 . The rapid response to the availability of nutrients requires that the starved cells have an efficiently expressed transport system for nutrient uptake 44 , a result in agreement with our finding that many transmembrane and outer membrane transporters that are highly expressed at 37 °C. Moreover, the production of cell envelope structures, such as peptidoglycan, LPS and lipoproteins, was induced at 37 °C. This may be anticipated as they could confer protection to the cell during adaptation phases 45 .
Non-mammalian infection models have become particularly attractive because they are cheap, easy to assay, not limited by ethical concerns and can survive at 37 °C, thus enabling high-throughput in vivo screening. In this context, G. mellonella larvae have been used as a model organism to study bacterial pathogenicity, including members of the Bcc, because of the similarities between the innate immune systems of insects and mammals [46][47][48] . B. seminalis strains isolated from water, fruits and the rhizosphere are considered weak pathogens or non-virulent to G. mellonella in comparison with clinical isolates 49 . B. seminalis TC3.4.2R3 killed 100% of larvae in 24 h at 37 °C with 10 7 CFU.mL −1 , but the virulence was attenuated at 28 °C, and all larvae died in 65 h (Fig. 3). The results showed that although B. seminalis TC3.4.2R3 can kill wax moth larvae, this virulence is significantly increased at 37 °C. For B. cenocepacia, the largest number of virulence genes was induced at 37 °C in comparison to 20 °C, suggesting that temperature adaptation during host infection is associated with pathogenesis 50 .
However, in the BALB/c model, B. seminalis TC3.4.2R3 was not virulent, even at high concentrations (10 7 CFU.mL −1 ), indicating that this strain is not pathogenic to mice and that the results for G. mellonella and mice infection models were not equivalent. Similar results were observed for B. thailandensis, which is avirulent in mouse infection models 51 but is highly virulent in insects and nematoid infection models 52 . These results suggest that for some environmental strains, such as B. seminalis TC3.4.2R3, temperature plays a role mainly in the interaction with soil-and plant-associated microbiota, including bacteria, fungi and insects, increasing the capability of this bacterium to compete with other microorganisms and to colonize the host plant.
Many factors may be implicated in virulence, depending on the infection model. For B. cenocepacia, AHL-mediated quorum sensing, siderophores and LPS were found to be important virulence factors in mammals and G. mellonella; however, several virulence factors were host-specific for these bacteria 53,54 . In addition, it is likely that in G. mellonella, other virulence factors, such as flagellar motility, facilitate disease progression. Flagellar motility may be related to increased virulence in B. pseudomallei 55 , but pathogenic Burkholderia species appear to have adopted very different strategies in terms of the regulation of flagellar motility during host infection 27 .
Among the probable virulence factors, we found that capsular polysaccharide production was significantly induced at 37 °C. Capsular polysaccharides are often associated with host colonization and virulence 56 . In B. seminalis TC3.4.2R3, the genes Bsem_02944 to Bsem_02953, present in the central part of the WCB cluster, were upregulated at 37 °C (Fig. 6), suggesting that the capsule could be a key factor associated with virulence in G. mellonella but has minor roles in soil and plant interactions. A previous study showed that these genes were variably expressed in Burkholderia, which were associated with different lifestyles 7 .
Phenylacetic acid (PAA) is a signal molecule that could be involved in microbial interactions with the host 57 . The phenylacetic acid (PAA) catabolic pathway was shown to be required for the full pathogenicity of B. cenocepacia in a Caenorhabditis elegans nematode infection model 58 . Nevertheless, B. seminalis TC3.4.2R3 does not possess all genes of the PAA pathway. Clusters paaGHIJK and paaABCE are the two core functional units of the PAA catabolic pathway common in many organisms 59 . The paaE gene is absent in TC3.4.2R3 and has been shown to be necessary for the survival of B. cenocepacia in rats 60 . Therefore, the pathway for PAA degradation present on chromosomes 1 and 2 of B. seminalis TC3.4.2R3 may allow the partial degradation of PAA, and the accumulation of intermediates might have toxic effects on G. mellonella larvae, but the lack of paaE could explain the non-virulence in mice.
Indole-acetic acid (IAA) and phosphate solubilization are important characteristics in plant growth-promoting bacteria. IAA is an auxin that promotes cell expansion and plant growth. B. seminalis TC3.4.2R3 is able to produce IAA and solubilize phosphate 15 , but the expression of these genes was not thermoregulated. Additionally, in this study, B. seminalis had no effect on cotton and maize seed germination, regardless of the temperature.
B. seminalis TC3.4.2R3 presents the ability to control orchid necrosis 7 , which seems to be temperature dependent, since we observed less antimicrobial activity at 37 °C. However, the expression of the pyochelin and rhamnolipid gene clusters was not affected by temperature shift. Indeed, previous studies showed that the production of these antimicrobial compounds is induced by the presence of phytopathogenic fungi and oomycetes 16,17 . Interestingly, transcriptional factors related to repression of antibiotic biosynthesis were upregulated at 37 °C. The AbrB family transcriptional regulator inhibited bacitracin 61 and lantibiotics 62 biosynthesis in Bacillus. The TetR family of transcriptional regulators can also repress antibiotic biosynthesis, such as kanamycin 63 , auricin 64 and the peptidyl nucleoside antibiotic, gougerotin 65 . In this work, we found four TetR genes upregulated at 37 °C, which could be associated with repression of antibiotic production in B. seminalis TC3.4.2R3. Additionally, less pigment was produced at 37 °C, which could also be related to a secondary metabolite.
The behaviour and pathogenesis of B. seminalis TC3.4.2R3 are strongly influenced by temperature. At 28 °C, the temperature that reflects conditions that the bacterium encounters in the environment, more biofilm and EPS were produced, as well as potential biocontrol capabilities such as inhibition of phytopathogen growth. In terms of colonization and plant protection, this points to optimum adaptation at 28 °C. At 37 °C, the temperature reflecting infection of a mammalian host, the bacterium was more motile and produced fewer antimicrobial compounds, suggesting that a set of genes are regulated in a temperature-specific way, which could be related www.nature.com/scientificreports www.nature.com/scientificreports/ to environmental/host conditions. We observed that genes that encode EPS are induced at 37 °C and that this capsular polysaccharide is an important virulence factor for larvae infection, but this bacterium failed to induce mouse death, which could be associated with the lack of genes associated with virulence. In fact, the Type 3 (T3SS), T4 (T4SS) and T6 (T6SS) secretion systems gene clusters were not completely identified in the genome of this TC3.4.2R3 strain 7 . Although B. seminalis TC3.4.2R3 is within the Bcc group, it does not cause disease in the murine model, which is an important consideration given its biocontrol potential in crops. However, to corroborate the biosafety of this strain as a biocontrol agent, intranasal inoculations in mice and other mammalian models should be performed to confirm the avirulence of this endophytic strain.

Materials and Methods
Bacterial strain and growth conditions. B. seminalis strain TC3.4.2R3 was isolated from inside sugarcane root tissues 15 where it resides as an endophytic bacterium. The whole genome of the strain was previously sequenced and described 7 . B. seminalis was maintained at −80 °C in Luria-Bertani (LB) broth (Difco Laboratories, Sparks, USA) and 20% glycerol and was recovered on Luria-Bertani agar or broth (Difco Laboratories) with incubation for 24 h at 28 °C or 37 °C.
Swimming motility assay. An overnight bacterial culture was harvested and resuspended in phosphate buffered saline (PBS) to an OD 600 of 1.0. Bacterial suspensions were inoculated on the centre of LB with 0.3% w/v agar and incubated at 28 °C and 37 °C. Motility was evaluated by the appearance of growth rings around the inoculum area, and the halo diameter (mm) was measured after 24, 48 and 120 h 66 . The experiment was performed in triplicate.
Biofilm assays. Biofilm formation was quantified using a 96-well plate and accompanying peg-lid of the MBEC (Minimum Biofilm Eradication Concentration) assay device (Innovatech, Edmonton, Canada). Wells received 150 µL bacterial suspension of B. seminalis standardized to 10 7 CFU.mL −1 in Tryptone Soya Broth (TSB) (Difco Laboratories). Eight wells were inoculated in at least three independent experiments. The peg-lid was placed on the plate and incubated at 28 °C or 37 °C for 24 h. The peg-lid was transferred to a fresh 96-well plate containing pre-warmed TSB and incubated for more than 24 h. The peg-lid was rinsed with PBS and then baked at 60 °C for 20 minutes, followed by staining with crystal violet 0.1% (w/v) (200 µL per well) and incubation for 30 minutes at room temperature. Three wash plates containing PBS (200 µL per well) were used to rinse the pegs following staining; subsequently, the crystal violet was solubilized with 95% ethanol prior to measurement at OD 595 67 . In addition, a growth curve was prepared. Flasks containing 100 mL of TSB were each inoculated with a bacterial suspension to reach OD 595 equal to 0.05. Then, flasks were shaken at 150 rpm at 28 °C or 37 °C. Aliquots were removed at periods of 2 h until the stationary phase, and OD 595 readings were measured.
Measurement of exopolysaccharide (EPS) weight. EPS was purified from cell culture media 68 with modifications. Briefly, B. seminalis was grown for 5 days at 28 °C or at 37 °C in 10 mL of mannitol medium (0.2% yeast extract and 2% mannitol). The bacterial cultures were vortexed and centrifuged at 2,300 × g for 10 minutes. The supernatant was transferred to new tubes, and phenol was added at a 10% final concentration and then incubated at 4 °C for 5 h and further centrifuged for 15 minutes. The water phase was collected, 4 volumes of isopropanol was added and incubated at −20 °C for 18 h. The precipitated exopolysaccharide was centrifuged at 9,100 × g for 15 minutes and suspended in 150 µL of distilled water. The extract was lyophilized, and the dry weight was measured.
Antagonism against plant pathogenic fungi. An overnight B. seminalis culture was adjusted to 10 7 CFU.mL −1 and 10 µL was cultivated for 48 h at 28 °C or at 37 °C on potato dextrose agar (PDA) (Difco Laboratories). After bacterial growth, disks of phytopathogenic fungi (6 mm) were transferred to the centre of PDA medium and incubated at 28 °C. After fungi growth, the inhibition zone was measured and compared with the control, which consisted of fungi plates without B. seminalis. The antagonism was evaluated against Ceratocystis fimbriata, C. paradoxa, Colletotrichum falcatum and Fusarium oxysporum. The experiment was performed in triplicate 69 . Detection of putative secondary metabolism and antibiotic genes in the three chromosomes of B. seminalis was performed by antiSMASH version 3.0 70 .
Effects of Burkholderia on seed germination. B. seminalis was grown in LB broth at 28 °C or at 37 °C, shaking at 180 rpm for 48 h. Bacteria were harvested, and the absorbance was adjusted to OD 600 1.0 with sterile distilled water. Seeds of maize (Zea mays) and cotton (Gossypium hirsutum) were surface sterilized in 70% alcohol for 1 minute and 3% sodium hypochlorite for 1 minute and rinsed twice in sterile water. Subsequently, the seeds were immersed in the B. seminalis suspension at 28 °C or at 37 °C under agitation at 150 rpm for 1 h. Control seeds were treated with sterile water (untreated control). A replicate consisting of 50 seeds was placed on moist filter paper and incubated in a germination chamber at 28 °C or at 37 °C and a 12 h photoperiod for five days. During the experiment, filter paper was maintained humid, adding a suitable volume of sterile water daily and wrapping containers in transparent plastic bags. The experiment was carried out in triplicate. The mean germination percentage was calculated at the end of the experiment 71 . Galleria mellonella infection model. Larvae were obtained from LiveFoods UK and stored in woodchips at 10 °C prior to use. B. seminalis TC3.4.2R3 was inoculated in the hindmost larvae proleg using a 25 μL 22s-gauge gas-tight Hamilton syringe with 10 μL of 10 8 CFU.mL −1 and incubated at 28 °C or at 37 °C. PBS and non-inoculated larvae were used as controls and were incubated at the same temperature as the treatments. For each strain and treatment, ten larvae were inoculated per experiment in two independent experiments. The larval survival was monitored up to 72 h 48 . (2019) 9:5486 | https://doi.org/10.1038/s41598-019-41778-x www.nature.com/scientificreports www.nature.com/scientificreports/ BALB/c mouse infection model. To assess the pathogenic potential of B. seminalis TC3.4.2R3 in a mammalian host, six-to eight-week-old female BALB/c mice were infected. Experiments were approved by the Ethics Committee on the Use of Animals (CEUA) from University of Campinas under Registration No. 4305-1 and were conducted according to institutional standards. B. seminalis TC3.4.2R3 used for challenge was grown in LB broth at 37 °C under agitation (180 rpm) for 18 h. After growth, 1 mL of bacterial suspension was transferred to 50 mL of LB and incubated under the same conditions until OD 600 1.0 was reached. The culture was harvested for 15 minutes, and the pellet was washed with 10 mL of cold PBS. After centrifugation, the pellet was resuspended in PBS. Groups of three BALB/c mice were challenged by an intraperitoneal route with 100 µL of B. seminalis suspensions (10 9 , 10 8 , 10 7 , 10 6 or 10 5 CFU.mL −1 ). PBS was used as a control. The delivered doses of bacterial cells were then verified by plate counts on TSA. Mice were monitored for clinical signs and symptoms for 30 days.
RNA extraction and sequencing. B. seminalis was grown for 48 h with shaking (180 rpm) at 28 °C or at 37 °C in TSB in triplicate. These conditions were set up based on the time (48 h) where the greatest difference in the ability to inhibit fungi between the two evaluated temperatures was observed, in which temperature was correlated with the environmental (28 °C) or clinical (37 °C) conditions.
For RNA extractions, after growth, samples were frozen with liquid nitrogen and macerated. Total RNA was extracted using the Illustra RNAspin Mini Kit (GE Healthcare, Chicago, United States) according to the manufacturer's instructions. Samples were tested for quality in agarose gel, and 60 ng/mL of each sample was dissected with RNAstable (Biomatrica, San Diego, United States). RNA quality was evaluated by Qubit and Bioanalyzer, and only samples with intact RNA were processed. Removal of rRNA was performed using the Ribozero Bacteria (Gram-Negative) kit. The sequencing library was prepared according to the Illumina Truseq Stranded RNAseq protocol and was sequenced on the Illumina HiSeq 2000/2500 platform 46 to generate single-end reads with 50 bp. Reads were mapped on B. seminalis TC3.4.2R3 chromosomes (GenBank Accession Number LAEU01000000).

Mapping and analysis of Illumina reads.
Sequence reads from each sample had rRNA and tRNA subtracted with Bowtie 72 . The reference sequences used were E. coli and Burkholderia. The resulting sequences were mapped on the B. seminalis TC3.4.2R3 genome (GenBank Accession Number LAEU01000000) with BWA software 73 . The Samtools package 74 was used to convert sam into bam format files, sorted and indexed. To determine the differential expression of known transcripts, the resulting aligned reads were analysed by TopHat 75 and Cuffdiff package 76 . Transcripts mapping to the B. seminalis TC3.4.2R3 genome with a P-value of ≤0.05 were considered differentially expressed transcripts, and those with log2(FC) above 1 were analysed further. Relative expression graphs and tables were assembled using the R program and the CummeRbund library 77 . Volcano plots were made in R. Functional and Gene Ontology annotations of the genome were performed with Blast2GO software 78 . Furthermore, transcripts were analysed by dbCAN (carbohydrate-active enzyme annotation) 79 , SignalP for the presence and location of signal peptide 80 , EffectiveDB for prediction of bacterial protein secretion 81 and PSORTb for bacterial protein subcellular localization prediction 82 .
Statistical analysis. All in vitro assays were performed at least in triplicate, with subsequent statistical analysis by one-way ANOVA with Tukey's test (GraphPad Prism 7.0) and multiple comparisons. For growth curves, linear regression was used to calculate line equations, and curves were compared by the F test. P < 0.05 was deemed statistically significant. A Kaplan-Meier survival plot with log-rank (Mantel-Cox) test and Bonferroni correction (GraphPad Prism 7.0) was used to compare survival of G. mellonella.