Physiological and transcriptomic analyses reveal mechanistic insight into the adaption of marine Bacillus subtilis C01 to alumina nanoparticles

An increasing number of studies have investigated the effects of nanoparticles (NPs) on microbial systems; however, few existing reports have focused on the defense mechanisms of bacteria against NPs. Whether secondary metabolism biosynthesis is a response to NP stress and contributes to the adaption of bacteria to NPs is unclear. Here, a significant induction in the surfactin production and biofilm formation were detected by adding Al2O3 NPs to the B. subtilis fermentation broth. Physiological analysis showed that Al2O3 NP stress could also affect the cell and colony morphogenesis and inhibit the motility and sporulation. Exogenously adding commercial surfactin restored the swarming motility. Additionally, a suite of toxicity assays analyzing membrane damage, cellular ROS generation, electron transport activity and membrane potential was used to determine the molecular mechanisms of toxicity of Al2O3 NPs. Furthermore, whole transcriptomic analysis was used to elucidate the mechanisms of B. subtilis adaption to Al2O3 NPs. These results revealed several mechanisms by which marine B. subtilis C01 adapt to Al2O3 NPs. Additionally, this study broadens the applications of nanomaterials and describes the important effects on secondary metabolism and multicellularity regulation by using Al2O3 NPs or other nano-products.

then caused a decrease in cell viability of Bacillus licheniformis 12 . However, these reports mainly focused on the antibacterial properties of NPs, whether or how microorganism adapt to NP stress remains unclear.
Bacillus species as biological control agents are receiving increased attention because of their ability to produce various antimicrobial substances. Additionally, these species are commonly used as a model Gram-positive strain for drug-resistance analysis. As a result, the antimicrobial effects of NPs have been explored with B. subtilis. Toxicity of Ag and ZnO NPs towards B. subtilis is significantly less due to the presence of a thicker peptidoglycan layer 13 . Previous studies have addressed the role of a limited sub-set of B. subtilis genes in response to Al 2 O 3 NPs but the potentially pan-metabolic action of Al 2 O 3 NPs on cells alludes to large-scale genetic regulation 14 . For Al 2 O 3 NPs, the toxic mechanism may be enhanced by association of the nanoparticle and bacterial surface and subsequent cell wall binding followed by the enhancement of permeability 15 , however, how B. subtilis adapt to the Al 2 O 3 NPs remains unknown. In our earlier study, we reported that Al 2 O 3 NPs can be used as effective flocculants for flocculating B. subtilis, and the possible attachment mechanisms of Al 2 O 3 NPs to the B. subtilis surface may be electrostatic 16 . Whether this electrostatic attachment could affect or change the physiological phenotype and development or affect secondary metabolism remained unclear.
Nearly 30 years ago, James A. Shapiro proposed multicellularity as a general bacterial trait 17 , and B. subtilis is now one of the classical and best-studied bacterial species 18 . Given that Al 2 O 3 NPs damage the bacterial cell wall and increase permeability, resulting in growth inhibition, we wondered whether or how Al 2 O 3 NPs affect multicellularity and secondary metabolism of B. subtilis, and how B. subtilis adapt to a certain concentration of Al 2 O 3 NPs. To test this aim, various concentrations of Al 2 O 3 NPs were added during the culturing and fermentation of surfactin of B. subtilis. We noted a significant induction in the surfactin production and biofilm formation by adding Al 2 O 3 NPs in the fermentation broth. Al 2 O 3 NPs also influenced the motility, colony morphology, and sporulation. Furthermore, a suite of toxicity assays testing membrane damage, cellular ROS generation, electron transport activity and membrane potential was used to determine the molecular mechanisms of toxicity of Al 2 O 3 NPs compared to their microsized analogues. To capture the overall genetic response to Al 2 O 3 NPs and bulk-Al 2 O 3 and to explore the mechanism of adaption to Al 2 O 3 NPs, whole transcriptomic analysis was used. These results reveal a new mechanism of how marine B. subtilis C01 adapted to alumina NPs. Additionally, this study broadens the potential applications of nanomaterials and has important implications for secondary metabolism and multicellularity regulation by using Al 2 O 3 NPs and for exploring other nano-products useful in product fermentation or bio-medical applications.

Results
Effect of Al 2 O 3 NPs on biofilm formation. In our previous study, it was reported that Al 2 O 3 NPs can be used as effective flocculants for flocculation of B. subtilis 16 . During the flocculation by using different concentrations of 40 nm Al 2 O 3 NPs, which have been characterised in our earlier study 15 , biofilm formation was also found to be influenced to varying degrees (Fig. 1A). After treating B. subtilis with 0.3, 1, 3, or 10 mM of 40 nm Al 2 O 3 NPs and continuing shake culturing for 60 h, biofilm formation of B. subtilis was enhanced as the concentration of Al 2 O 3 NPs increased, although high concentrations of Al 2 O 3 NPs could inhibit the growth of planktonic cells (Figs 1B and 2A). The quantitative analysis of biofilm formation using crystal violet was similar to the phenotypic analysis (Fig. 1A,B).
However, it remained unknown whether Al 2 O 3 NPs had the same effect on B. subtilis when stationary culturing. To test this, biofilm formation was monitored when Al 2 O 3 NPs were added in the liquid fermentation broth followed by stationary culturing. In contrast to shake culturing, Al 2 O 3 NPs prevented biofilm formation in stationary culturing (Fig. S1), which was probably due to the flocculation effect of Al 2 O 3 NPs ( Fig. 2A), which resulted in the restriction of motility. However, the exact mechanisms need to be determined in subsequent studies. Taken together, Al 2 O 3 NPs appear to be involved in the regulation of biofilm formation.
Effect of Al 2 O 3 NPs on surfactin production. Surfactin was quantified using HPLC to determine whether the surfactin production changed after the flocculation by Al 2 O 3 NPs. The results showed 3 mM Al 2 O 3 NPs could induce the surfactin production ( Fig. 2B). High concentrations of Al 2 O 3 NPs inhibit the growth of microorganisms 15 ; therefore, organisms require a suitable concentration and induction time of Al 2 O 3 NPs to adjust the growth and metabolism. To evaluate this effect, the concentration of Al 2 O 3 NPs was varied from 0 to 10 mM. Surfactin accumulation increased with increasing concentrations of Al 2 O 3 NPs until a limiting maximum concentration (3 mM or 4 mM) was reached ( Fig. 2A,B), and growth was reduced when the dosage was 10 mM. In response to varying Al 2 O 3 NPs induction times (0-72 h), the highest yield (33.5 mg/L) was at 12 h (Fig. S2). It is well knows that NP size holds an intriguing role on its physico-chemical property and subsequent effect on microbial system. In our study, we found that 3 mM small size (40 nm) of Al 2 O 3 NPs could induce more surfactin production compared with that induced by large size (110 nm and 280 nm) of Al 2 O 3 NPs (data not shown). As a result, the addition of 3 mM Al 2 O 3 NPs (40 nm) at 12 h to the fermentation media is the suggested treatment condition for further physiological study.

Al 2 O 3 NPs influence morphogenesis and motility.
To evaluate the effect of Al 2 O 3 NPs on B. subtilis C01 morphogenesis, C01 was grown on 2216E medium with or without 3 mM Al 2 O 3 NPs. As shown in Fig. 3, C01 control populations on the agar medium develop colonies with robust morphology. However, the robustness of colony morphology was dramatically diminished on the 3 mM Al 2 O 3 NP-containing agar medium. Colonies grown on the bulk-Al 2 O 3 -containing agar medium were similar to those on control agar medium (Fig. 3).
To investigate the effect of Al 2 O 3 NPs on surface motility, bacteria were spotted onto the centers of 2216E soft agar swarm plates (0.3% of agar in the 2216E medium with or without 3 mM Al 2 O 3 NPs). Within 6-8 h of incubation, the bacteria formed a colony of 2-4 cm diameter in control swarm plates (Fig. 3). However, the presence of 3 mM Al 2 O 3 NPs significantly altered the motility of bacterial cells compared to control in which no Al 2 O 3 NPs were present (Fig. 3). The results showed that Al 2 O 3 NPs could influence the morphogenesis and motility due to the nano-size of this particle, as bulk-Al 2 O 3 could not influence these phenotypes.
To determine whether Al 2 O 3 NPs could affect the flagella, we carried out microscopic studies of cells collected from the control fermentation broth and broth treated by 3 mM Al 2 O 3 NPs (Fig. 3). TEM results showed that cells from the control or bulk-Al 2 O 3-added fermentation broth have a number of peritrichous flagella. However, under Al 2 O 3 NPs treatment, Al 2 O 3 NPs could attach to the membrane and cause the flagellar damage, because cells from such treatment showed flagella missing or agglomerate flagella, and floccules was found attached to the cells (Fig. 3). In addition, the cell morphology was significantly different in the presence and absence of Al 2 O 3 NPs, and Al 2 O 3 NPs could attach to the cell membrane (Fig. 3).
A consensus has emerged that swarming motility by B. subtilis requires or must be facilitated by the production of the lipopeptide surfactin 19,20 . Figure 3 shows that exogenously adding commercial surfactin in 2216E agar media could significantly restore the swarming motility and cell morphology compared to the Al 2 O 3 NPs treatment. However, in the fermentation broth, surfactin did not reduce the flagellar damage, as the flagella of the cells in such treatment was also missing or agglomerate according to the TEM analysis (Fig. 3). Taken together, a relatively high surfactin production could restore and enhance swarming motility under Al 2 O 3 NP stress, as exogenously adding commercial surfactin in Al 2 O 3 NP treatment media could enhance the motility (Fig. 3).

The mechanisms of toxicity for Al 2 O 3 NPs on B. subtilis.
To elucidate the mechanisms of toxicity for the Al 2 O 3 NPs, a suite of assays measuring membrane potential, membrane damage, cellular ROS generation, and electron transport activity was employed in B. subtilis (see the Supporting Information for methods).  Table S1 assay for the overview of cellular processes regulated at the transcriptional level).
Two-component signal systems (TCS) are one means that bacteria have to respond to external stimuli. The transcriptome shows that the LiaRS TCS was involved in the adaptation to Al 2 O 3 NPs stress, as most of the related genes were highly upregulated (Fig. 4A). The response regulator/histidine kinase pair LiaRS of Bacillus subtilis, together with its membrane-bound inhibitor protein LiaF, constitutes an envelope stress sensing module that is conserved in Firmicutes bacteria 21 . LiaRS strongly responds to the presence of a number of cell wall antibiotics, such as bacitracin 22 . The exact physiological role of LiaI and LiaH is not well understood, but the proteins seem to be involved in sensing and counteracting membrane damage 23 . Additionally, genes in BceRS TCS, which mainly regulated the cell envelope stress response 24,25 , were also upregulated in the adaptation to Al 2 O 3 NP stress.
The membrane proteins are the integral part of the bacterial cell membrane to maintain the cell integrity. The transcriptome shows that many proteins integral to membrane were up-regulated, such as membrane protein OxaA, mechanosensitive channels. While inner membrane proteins involved in sporulation were down-regulated Fatty acids biosynthesis and lipid metabolism gene response to Al 2 O 3 NP stress. The transcriptome showed that the genes related to fatty acid biosynthesis (FabF, FabHa) were upregulated in response to the Al 2 O 3 NPs stress (Fig. 4E), and FapR, involved in negative regulation of the fatty acid biosynthetic process 26 , was down-regulated (Fig. 4E). Increased abundance of the FabF elongation enzyme can increase the chain length of the resulting fatty acids 27 . Another key gene, extracytoplasmic function σ factor σ W , was also induced (Fig. 4E), which was reported have a function in regulation of FabHa and FabF 28 . Additionally, expressions of some genes involved in the lipid metabolism were influenced by Al 2 O 3 NP stress (Fig. 4E). For example, long-chain acyl-CoA synthetase was induced, which was reported to activate fatty acids by thioesterification with coenzyme A. Fatty acyl-CoA molecules are then readily utilized for the biosynthesis of storage and membrane lipids 29 .
To test whether Al 2 O 3 NPs could affect the fatty acids profiles, fatty acid profiles of cultures after exposure to Al 2 O 3 NPs were analyzed and compared with the profiles of non-exposed cultures or cultures exposed to bulk-materials. Analysis revealed changes in membrane composition exclusively when cells were exposed to Al 2 O 3 NPs at a concentration of 3 mM but not in cultures exposed to bulk-material ( Table 1). The major changes were observed in proportions of i-C 13:0 , C 15:0 , i-C 14:0 -3OH, and C 18:0 ( Table 1).

Expression of genes involved in flagellar assembly and chemotaxis enhanced in the presence of
NPs. The transcription of flagellar biosynthesis genes (e.g., FlhA, FlhB, FlhF, and flgBCG) were highly upregulated, as well as most of the flagellar assembly genes (e.g., FliE-I) (Fig. 4B). Most prominent among the upregulated genes are the large flgB operon and the hag gene, both involved in motility. Flagella are constructed from over 20 different proteins that must be assembled with the correct order and in the correct stoichiometry 30 . To ensure proper assembly, flagellar gene expression is organized in at least two hierarchical levels defined here as "early-class" genes, recognized by σ 70 , and "late-class" genes, recognized by the alternative sigma factor σ 28 31 . FliE, encoding the flagellar hook-basal body complex protein, is a classic Early-class flagellar genes, was upregulated at the point of 60 min (Fig. 4B). MotA and MotB, encoding the protein components of the stator of proton-driven motors, were also upregulated (Fig. 4B). Generally, FliM secretion liberates its cognate σ 28 to direct expression of the late-class flagellar genes. However, FliM was also upregulated at this time (Fig. 4B). Considering that Al 2 O 3 NPs could eliminate the flagella when flocculation occurred, the types of regulated genes indicated that The transcriptomic data also show that many genes belonging to the chemotaxis system [e.g., chemoreceptor proteins (methylated chemotaxis proteins, MCPs), chemotaxis histidine kinase (CheA), and chemotaxis proteins (CheW and CheC)] are upregulated in the Al 2 O 3 NP treatment (Fig. 4B). The response of the MCPs is transmitted via the CheA histidine kinase that is complexed, together with adaptor protein CheW, on the cytoplasmic side 32 . In addition, other genes that improve the cell motility and secretion were also upregulated such as ClpX (encoding ATP-dependent Clp protease proteolytic subunit) (Fig. 4B). Al 2 O 3 NPs affect the biofilm-and sporulation-related genes. Biofilm formation is a social behavior that generates favorable conditions for sustained survival in the natural environment 33 . Generally, regulatory pathways that control biofilm formation include the Spo0A pathway, the SlrR-SinR epigenetic switch system [the YwcC and SlrA pathway and the Abh-extracytoplasmic function (ECF) RNA polymerase σ -factors pathway], and the DegS-DegU two-component system. The relationship of the four pathways is shown in Fig. 4D. The transcriptome analysis showed that most genes related to the Spo0A pathway (e.g., KinA-C and spo0A) were down-regulated, as well as tapA-sipW-tasA and SinR (Fig. 4D). Spo0A is a central transcriptional regulator that controls the expression of more than 100 genes, including those necessary for biofilm matrix gene expression and sporulation, by controlling the activity of the master regulator SinR, a repressor of the eps and tapA-sipW-tasA operons 34 .
Another pathway mediated by the TetR-type transcriptional repressor (YwcC) was down-regulated 35 . When YwcC receives an as-yet-unknown signal, slrA is derepressed and the matrix genes are induced by SlrA-mediated inactivation of SinR 35 . Meanwhile, the Abh-ECF σ factors pathway also showed a response to Al 2 O 3 NP stress (Fig. 4D); the Abh protein regulates the transcription of slrR, further inactivates SinR and then induces the expression of matrix genes 36 . The transcription of abh is controlled by several extracytoplasmic function (ECF) RNA polymerase σ -factors, includingσ M , σ W and σ X 36 . ECF σ -factors are activated by a variety of external stimuli, including cell wall stress and specific antibiotics 37 .
At the same time, genes involved sporulation (e.g., KinA-C, spo0F, spo0B, and spo0A) were also suppressed in the Al 2 O 3 NP treatment at 60 min. the percentage of sporulated cells was quantified, and the results showed that a high concentration of Al 2 O 3 NPs could suppress the formation of spores within 72 h (Fig. S6).
Upregulation of DNA damage repair-associated genes. Following Al 2 O 3 NP treatment for 60 min, three group of genes (homologous recombination, mismatch repair, and nucleotide excision repair) belonging to the DNA damage repair system were upregulated (Fig. 4C). For example, RecA, which is central to genome integrity and is important for strand exchange during homologous recombination, stabilizing stalled replication forks, and induction of the SOS transcriptional response to DNA damage 38 , was upregulated in the treatment. Meanwhile, genes encoding DNA mismatch repair and nucleotide excision repair proteins were also upregulated (Fig. 4C).
A plasmid-based in vitro DNA damage assay 39 was used to study the intrinsic potential of the Al 2 O 3 NPs to damage double-stranded DNA. The gel electrophoresis results for various treatments are shown in Fig. 4F. The positive control used in this study was a UV-treated plasmid, which was completely degraded and appeared as smeared. Severe DNA damage was observed for Al 2 O 3 NPs, which induced complete degradation of plasmid DNA (Fig. 4F). By contrast, bulk-Al 2 O 3 had no effect on DNA damage, appearing similar to the negative control. This might be one way to induce the DNA repair response in vivo, because a previous study had demonstrated the attachment of Al 2 O 3 NPs to the surface of the cell membrane and also their presence inside the cells due to formation of irregular-shaped pits and perforation on the surfaces of bacterial cells 40 . Furthermore, Al 2 O 3 NPs also induced the ROS scavenge system response, as some of the catalase-related family genes and super oxide dismutase (SOD) were upregulated (Fig. 4G). The physiological analysis also showed that high concentrations of Al 2 O 3 NPs could induce ROS in cells compared with bulk-Al 2 O 3 treatment and controls (Fig. 4B) and also might result in DNA damage in vivo 9 . However, the exact mechanisms for DNA damage in vivo need further study.

Systematic validation of transcriptome data using real-time PCR.
To validate the transcriptome data and systematically analyze the expression of key genes during treatment with Al 2 O 3 NPs and to further elucidate the mechanism of adaption of marine Bacillus subtilis C01 to Al 2 O 3 NP stress, eighteen key genes were selected on the basis of their possible role in two-component signal systems, membrane integrality, stress response, flagellar assembly, motility, biofilm formation, DNA repair, transcriptional regulation, and surfactin biosynthesis for real-time PCR analysis (Fig. S7). The real-time PCR used the control treatment and bulk-Al 2 O 3 treatment as references, separately. Meanwhile, we detected the gene transcripts at 60 min, 12 h, and 24 h after treatment with Al 2 O 3 NPs. The quantitative gene expression results correlated with the trend of regulation observed in the transcriptome experiment. In general, the relative expression of genes in Al 2 O 3 NPs/control were similar to those in Al 2 O 3 NPs/bulk-Al 2 O 3 , and the genes showed different responses to Al 2 O 3 NPs at different time points (Fig. S7). During the early stages of Al 2 O 3 NP stress, LiaR, as a key gene in LiaRS two-component signal systems mainly regulating the cell envelope stress response, was highly upregulated. Prolonged Al 2 O 3 NPs stress for 24 h leads to normal levels of LiaR. Genes involved in membrane integrality, SOD genes, and catalase genes were up-regulated last for a long time (last for 24 h). Genes involved in biosynthesis of flagellar components (flgB and fliE) exhibited a dramatic enhancement in their expression at 60 min after Al 2 O 3 NP treatment; however, these genes recovered to the normal level of expression for continued exposure to Al 2 O 3 NPs at 12 h and 24 h (Fig. S7).
RecA, involved in DNA repair, was also highly induced in the early stage of Al 2 O 3 NP stress (60 min); however, in the middle stage, RecA showed a dramatically decreased level of expression and then recovered to normal levels during the prolonged Al 2 O 3 NP exposure for 24 h (Fig. S7).
Most of biofilm formation and sporulation-related genes (DegS, SinR, KinA, Spo0A, tasA, epsA, and abrB) were down-regulated in the early stages of Al 2 O 3 NP stress; however, most of these genes were upregulated with prolonged with Al 2 O 3 NP exposure at 12 h or 24 h. Interestingly, the expression of SinR showed sustained down-regulation. Among those genes, both SinR and AbrB could repress tapA-sipW-tasA and epsA expression. The prolonged stress leading to down-regulation of SinR might induce the expression of tasA and epsA, and the repression from AbrB to tasA-epsA might be inhibited by high expression of spo0A (Fig. S7), thereby inducing biofilm formation in shake cultures ( Fig. 2A). However, the exact mechanisms need to be clarified in future studies.
More importantly, in this experiment, surfactin production was enhanced during the prolonged Al 2 O 3 NP stress for 60 h (Fig. 1C). However, the transcriptome showed the genes involved in the surfactin biosynthesis (srfAA-D) showed no significant change in expression, as even the regulation gene (comA) was down-regulated in the early stages (data not shown). The real-time PCR results showed that comA and srfA were indeed down-regulated during the early stages of stress but were induced with prolonged stress, and both were upregulated at 24 h; however, comA also showed down-regulation at 12 h treatment (Fig. S7). These results were consistent with the phenotypes in the former experiments (Fig. 2B).

Discussion
Recently, it has been widely accepted that NPs can offer a new strategy to tackle multidrug-resistant bacteria 41,42 . Many studies have focused on the antibacterial properties of NPs, and described the toxicity mechanisms of NPs against bacteria and drug-resistant bacteria 9,10 . However, the few existing reports on the defense mechanisms of tolerant bacteria against NPs are limited to Mycobacterium smegmatis with Cu-doped TiO 2 NPs 43 , B. subtilis and Pseudomonas putida with nC 60 44 , and Cupriavidus metallidurans CH34 with Al 2 O 3 NPs 45 and do not provide mechanistic insights by using full transcriptional analysis. Whether secondary metabolism biosynthesis could respond to NP stress and enhance the adaption of bacteria to NPs was unknown.
The Al 2 O 3 NPs have been shown to attack the bacterial cell membrane, alter membrane permeability 15,46 , and even accumulate inside the bacterial cell 40 . In agreement with this report, in our study, Al 2 O 3 NPs could attach to the cell membrane, affect the cell morphology and even cause membrane damage (Fig. 3). Meanwhile, toxicity analysis showed that relatively high concentrations of Al 2 O 3 NPs could cause membrane damage compared with bulk-Al 2 O 3 . Further transcriptional analysis showed that at least 200 genes encoding proteins related to membrane components were regulated by Al 2 O 3 NPs at the early stage of stress (Fig. S4), including genes involved in the LiaRS TCS 23 and the BceRS TCS 24,25 , both of which were involved in sensing and counteracting membrane damage (Figs 4A and S7). Extracytoplasmic function (ECF) RNA polymerase σ -factors (σ W , σ M ), involved in stress responses elicited by compounds that affect membrane integrity and/or fluidity, were also upregulated in the early stage of Al 2 O 3 NP stress. Cells have evolved the ability to modify membrane lipid composition to acclimatize to membrane stress 28 . In this sense, Kingston et al. 28 suggested that the σ W -dependent stress response in B. subtilis could regulate the fatty acids biosynthesis and reduce the membrane fluidity. Accordingly, our results show that the genes involved in fatty acids biosynthesis and the fatty acid profiles were changed with Al 2 O 3 NP stress, and i-C 13:0 , C 15:0 , i-C 14:0 -3OH, and C 18:0 were induced, which contribute to antioxidant stress and improve the stability of the membrane 28,47 , however, the results were different to nC 60 treatment, under which B. subtilis showed an increase in membrane fluidity 44 . In this study, ROS generation was also induced in the early stage, which might be related to the membrane damage. In other bacteria, Al 2 O 3 NP was also found to cause the membrane damage Scientific RepoRts | 6:29953 | DOI: 10.1038/srep29953 and induce the intracellular ROS in E. coli MG1655 and Cupriavidus metallidurans CH34 45 . Taken together, these results suggest that membrane-related stress was one of the most important effects caused by Al 2 O 3 NPs.
It was reported that some NPs such as Cu NPs 9 and Ag NPs 48 could damage DNA and cause growth inhibition of bacteria. However, whether Al 2 O 3 NPs have such properties remains unclear. In this study, the transcriptional analysis showed DNA damage repair-associated genes were highly upregulated in the early stage of Al 2 O 3 NP stress. Additionally, we found that a high concentration of Al 2 O 3 NPs could damage DNA in vitro (Fig. 4F). Furthermore, NPs generated ROS could cause DNA damage and induce DNA repair genes 9 . Coupled with the membrane damage and the induction of cellular ROS, Al 2 O 3 NPs probably induce the DNA damage in vivo. This result is in accordance with previously published data, which Cu NPs induce DNA damage in E. coli 9 , however, the exact mechanism for DNA damage needs further study.
Flagella enable Bacillus subtilis to move towards favorable environments or avoid harmful stimuli during swimming 20 . To avoid the stress of Al 2 O 3 NPs, B. subtilis had to regulate the expression of flagellar biosynthesis genes as well as most of the flagellar assembly genes (Fig. 4B). In this study, TEM analysis showed that Al 2 O 3 NPs could attach to the membrane and flagella, causing flagellar damage. As a result, B. subtilis upregulates the expression of the flagellar biosynthesis genes and flagellar assembly genes to repair the impaired flagella. Although the flagellar-related genes showed high expression following Al 2 O 3 NP treatment, the motility was not enhanced. This is probably because flagellar damage restricted the motility or the attachment of Al 2 O 3 NPs to cells increased the resistance to motility.
Swarming and sliding motility by B. subtilis were shown to require or be facilitated by the production of the lipopeptide surfactin 20 . In agreement with their report, our data showed that surfactin could restore and enhance swarming motility under Al 2 O 3 NP stress (Fig. 3). During Al 2 O 3 NP-treated fermentation, the induced surfactin production may alleviate the motility restriction to enhance chemotaxis in an attempt to eliminate the Al 2 O 3 NPs, as chemotaxis genes were also induced in this experiment. Interestingly, the surfactin produced could also reduce the surface tension 49 , which may help to wash Al 2 O 3 NPs away from cells. However, this needs to be determined in further studies.
Bacterial biofilms are multicellular communities in which cells are held together by an extracellular matrix to strengthen the adaption to various environmental factors 50,51 . In full agreement with these reports, our results showed a series of genes involved in biofilm formation were induced in different stages of Al 2 O 3 NP treatment. In the early stage, the YwcC-SlrA pathway and the Abh-ECF σ factors pathway were induced for a quick response to Al 2 O 3 NPs stress, while in the late stage, the spo0A pathway was activated and up-regulated tasA and epsA genes (Figs 4D and S7). Similar results were found in Shewanella oneidensis MR-1 response to Cu-doped TiO 2 NPs. S. oneidensis MR-1 could produce a large amount of extracellular polymeric substances (EPS) under NP stress, especially extracellular protein 43 . Surfactin, produced by constituent cells of the biofilm, was the first molecule identified as an inducer of matrix gene expression 52 . A recent study showed that surfactin could trigger the biofilm formation of B. subtilis in melon phylloplane 51 . The induction of surfactin in the late stage of Al 2 O 3 NP stress might contribute to the biofilm formation to alleviate Al 2 O 3 NP stress by separating B. subtilis from the Al 2 O 3 NPs (Fig. S8).
In summary, our transcriptome and physiological analyses suggest that the attachment of Al 2 O 3 NPs to the membrane of B. subtilis along with flagellar damage would initiate sensing and counteracting membrane damage. These responses also lead to changes in the membrane fatty acids profile, induction of DNA repair and the ROS scavenger system, and enhancement of flagellar biosynthesis to repair the organism, resulting in optimal conditions for B. subtilis growth and adaptation to Al 2 O 3 NPs in the early stage of stress. Furthermore, biofilm formation and surfactin biosynthesis were induced in the late stage to adapt to or avoid the stress (Fig. S8).

Methods
Strains and culture conditions. Marine Bacillus sp. strain C01, isolated from Weihai, was grown at 30 °C in Landy medium 16 . A 1% inoculum volume of culture was used to inoculate a 250 mL flask containing 100 ml of Landy medium, which was incubated for 72 h on an orbital shaker (150 rpm) at 30 °C. Al 2 O 3 NPs were added as an inductor at different times. Isolation of surfactin. Crude surfactin was isolated by adding concentrated hydrochloric acid to the Landy media after removing the biomass by centrifugation. A precipitate was formed at pH 2 which could be collected, dried, and extracted with dichloromethane. The solvent was removed under reduced pressure to give an off-white solid. Further purification was achieved by recrystallization. The dichloromethane extract was dissolved in distilled water containing sufficient NaOH to produce a pH of 8. This solution was filtered and titrated to pH 2 with concentrated HCl. The white solid was collected as a pellet after centrifugation. Quantitative analysis of surfactin by HPLC. The isolated surfactin was dissolved in 1 mL of methanol followed by charcoal treatment and passed through a 0.22-μ m-pore filter. The filtrate was subjected to HPLC on a reversed-phase column (RP-C18, 5 μ m, 4* 250 mm; Merck). The column was eluted at a flow rate of 1.0 mL/min with acetonitrile-water (80:20, v/v) and monitored at 214 nm. The concentration of surfactin was determined with a calibration curve made with authentic surfactin purchased from Sigma (S3523).

Preparation of Al
Scientific RepoRts | 6:29953 | DOI: 10.1038/srep29953 Quantitative analysis of biofilm formation. At specific times, planktonic cells were removed, biofilm cells were stained with 2 ml of 0.3% crystal violet for 10 min, washed with distilled water, and air dried. The crystal violet in the biofilm cells was solubilized with 2 ml of 70% ethanol, and the optical density at 570 nm (OD 570 ) was measured 53 .
Cell morphology and motility study. The effect of Al 2 O 3 NPs on Bacillus cell morphology and the attachment of Al 2 O 3 NPs to B. subtilis were studies using transmission electron microscopy (TEM) 15 .
Bacterial motility over a surface was analyzed by spotting 2 μ l culture of each strain grown overnight (~10 5 cells) onto the center of soft agar plates (2216E media with 0.3% agar) with different treatments. Plates were incubated at 30 °C in a humidified chamber for 6-8 h.