Introduction
With the development of space technology, especially the improvement of manned spaceflight technology and the completion of the International Space Station, space life science has become a hot area of research around the world. However, the space contains extreme factors such as microgravity, radiation, temperature, magnetic field and vacuum, which differ substantially from the environments on the earth's surface (Thirsk et al. Reference Thirsk, Kuipers, Mukai and Williams2009; Horneck et al. Reference Horneck, Klaus and Mancinelli2010). In order to develop and utilize outer space, human beings must overcome these adverse conditions. Currently the space exploration is implemented by launching manned spacecraft. The microorganisms are inevitably sent to space along with human beings or aerospace facilities. However, the study of effect of space environment on microorganism depends on the development of manned space flight of each country. In the preliminary studied we could not expose the microorganisms outside the space station and mainly focus on effect of spaceflight inside the station.
Multiple studies have reported that spaceflight had significant effects on various microorganisms’ features including physiology, growth, metabolism, biofilm, virulence and drug resistance (for reviews, see Nickerson et al. Reference Nickerson, Ott, Wilson, Ramamurthy and Pierson2004; Lynch & Matin Reference Lynch and Matin2005; Horneck et al. Reference Horneck, Klaus and Mancinelli2010). For example, earlier investigations have demonstrated that the virulence of Salmonella was enhanced after spaceflight and modelled weightlessness, as well as global changes in gene and protein expressions (e.g. Nickerson et al. Reference Nickerson, Ott, Mister, Morrow, Burns-Keliher and Pierson2000; Wilson et al. Reference Wilson, Ott, Ramamurthy, Porwollik, McClelland, Pierson and Nickerson2002a, Reference Wilson, Ramamurthy, Porwollik, McClelland, Hammond, Allen, Ott, Pierson and Nickersonb, Reference Wilson2007; Rosenzweig et al. Reference Rosenzweig, Abogunde, Thomas, Lawal, Nguyen, Sodipe and Jejelowo2010). Gao et al. (2010) reported that space-related environments resulted in wide range of phenotypes for Streptomyces avermitilis. The features of bacteria were different after exposure to short- and long-term spaceflight (e.g. Juergensmeyer et al. Reference Juergensmeyer, Juergensmeyer and Guikema1999; Taylor & Sommer Reference Taylor and Sommer2005; Su et al. Reference Su2014; Wang et al. Reference Wang2014). Microgravity, one of important factor of space condition, is an environment where physical force of gravity attenuated close to zero which is commonly experienced in space station or during spaceflight (Nickerson et al. Reference Nickerson, Ott, Wilson, Ramamurthy and Pierson2004; Horneck et al. Reference Horneck, Klaus and Mancinelli2010). Both real microgravity and ground-based microgravity analogues have been showed to affect bacteria physiological features including morphology, virulence, growth, antibiotics resistance, biofilm formation, the substrate utilization efficiency and gene expressions (e.g. Horneck et al. Reference Horneck, Bucker, Dose, Martens, Bieger, Mennigmann, Reitz, Requardt and Weber1984a; Wilson et al. Reference Wilson, Ramamurthy, Porwollik, McClelland, Hammond, Allen, Ott, Pierson and Nickerson2002b; Lynch et al. Reference Lynch, Mukundakrishnan, Benoit, Ayyaswamy and Matin2006; Crabbe et al. Reference Crabbe, Pycke, Van Houdt, Monsieurs, Nickerson, Leys and Cornelis2010; Rosenzweig et al. Reference Rosenzweig, Abogunde, Thomas, Lawal, Nguyen, Sodipe and Jejelowo2010; Vukanti et al. Reference Vukanti, Model and Leff2012; Arunasri et al. Reference Arunasri, Adil, Charan, Suvro, Reddy and Shivaji2013). Meanwhile, many microorganisms are able to survive under harsh outer space conditions that are really external space environment (Schuerger et al. Reference Schuerger, Mancinelli, Kern, Rothschild and McKay2003; Horneck et al. Reference Horneck2012; Nicholson et al. Reference Nicholson, Moeller, Team and Horneck2012). Thus, the impact of space environment on microorganism, highly relevant to astronauts’ health and consequently the fulfilment of space expeditions, requires comprehensive investigation.
Escherichia coli (E. coli) is a motile, Gram-negative and short bacillus with rounded ends and no spores, which is always employed as an important model organism in biological research due to its simple structure, rapid reproduction and easiness to be cultured (Singleton Reference Singleton1999). Most of the E. coli strains are considered as opportunistic pathogens. Under normal circumstances, E. coli is riskless and even beneficial to hosts, where it resists the attack of other pathogens by competition and helps the host synthesize vitamin K2 (Bentley & Meganathan Reference Bentley and Meganathan1982; Hudault et al. Reference Hudault, Guignot and Servin2001; Reid et al. Reference Reid, Howard and Gan2001). However in special cases of reduced immunity or lack of stimulation in intestinal tract for a long time, E. coli will escape the intestinal tract to other organs such as gall bladder, urethra, bladder and appendix, causing local infections in corresponding areas or systematic disseminated infections (Croxen et al. Reference Croxen, Law, Scholz, Keeney, Wlodarska and Finlay2013; Lai et al. Reference Lai, Rosenshine, Leong and Frankel2013). In rare cases, virulent E. coli strains can induce severe illness, such as the haemolytic-uraemic syndrome occurred in an enterohemorhagic E. coli outbreak in Germany, 2011 (Mellmann et al. Reference Mellmann2011). Previous study suggested that simulated microgravity induced higher toxin production in virulent E. coli strains, and furthermore aggravated host reaction to the invasive bacteria probably due to changes in protein expression levels in these bacteria (Chopra et al. Reference Chopra, Fadl, Sha, Chopra, Galindo and Chopra2006). During space activities, these common intestinal floras are unavoidably taken into space with astronauts, which may increase the risk of infections. Other studies demonstrated that the immune system of astronauts was mis-regulated and the immune ability was greatly impaired during spaceflight, which could require weeks to reach full recovery (e.g. Knight et al. Reference Knight, Couch and Landahl1970; Cogoli et al. 1980; Sonnenfeld Reference Sonnenfeld2012; Chang et al. Reference Chang, Walther, Li, Boonyaratanakornkit, Galleri, Meloni, Pippia, Cogoli and Hughes-Fulford2012). Escherichia coli is one of the contaminants on spacecraft and related facilities (Taylor Reference Taylor1974; Venkateswaran et al. Reference Venkateswaran, Satomi, Chung, Kern, Koukol, Basic and White2001; Schuerger Reference Schuerger2004). Some previous studies on E. coli responses to spaceflight environment (mainly microgravity) were to some extent controversial. Pilot studies on E. coli reported prolonged growth phase, increased final cell population and higher production of secondary metabolites during spaceflight compared with strains cultured on the ground (Klaus et al. Reference Klaus, Simske, Todd and Stodieck1997; Kacena et al. Reference Kacena, Manfredi and Todd1999a, Reference Kacena, Merrell, Manfredi, Smith, Klaus and Toddb; Brown et al. Reference Brown, Klaus and Todd2002). Lynch et al. (Reference Lynch, Mukundakrishnan, Benoit, Ayyaswamy and Matin2006) reported that after 24 h incubation in a low-shear modelled microgravity vessel, E. coli strain formed thicker biofilm and exhibited higher tolerance to salt, ethanol and two antibiotics than the normal gravity control. Consistently, two subsequent studies identified up-regulated genes under reduced gravity conditions were relevant to starvation response, stress response, biofilm formation and lipid synthesis (Vukanti et al. Reference Vukanti, Mintz and Leff2008; Vukanti & Leff Reference Vukanti and Leff2012). Tucker et al. (Reference Tucker, Ott, Huff, Fofanov, Pierson, Willson and Fox2007) suggested that the responses of E. coli to microgravity could be medium-dependent. In rich Luria–Bertani (LB) medium, most significantly up-regulated genes were translation related (Tucker et al. Reference Tucker, Ott, Huff, Fofanov, Pierson, Willson and Fox2007). However, neither in rich nor in minimal medium, changes in stress responses or antibiotic sensitivity were observed compared to normal gravity environment (Tucker et al. Reference Tucker, Ott, Huff, Fofanov, Pierson, Willson and Fox2007). A more recent microgravity simulation study on E. coli reported up-regulation of stress-adaptation genes and DNA replication genes and down-regulation of membrane transporter genes, carbohydrate catabolic genes, and nucleotide metabolism genes (Arunasri et al. Reference Arunasri, Adil, Charan, Suvro, Reddy and Shivaji2013). Therefore it is crucial to investigate the changes of microbes after spaceflight and the underlying mechanisms.
In our study, a strain of E. coli LCT-EC106 was inoculated into the stab agar culture and sent to the space with the ShenZhou-8 spacecraft that travelled from 1 to 17 November 2011 for about 397 h. After the landing, the sample was subjected to phenotypic selection, including growth, drug resistance, microscopic examination and Biolog tests. Two mutant strains LCT-EC52 and LCT-EC59 were isolated and subsequently analysed at genomic, transcriptomic and proteomic level by next generation sequencing and iTRAQ to identify the changes of the bacteria occurred during spaceflight. To the authors’ knowledge, this is the first report of systematic changes of E. coli in response to spaceflight, especially microgravity and could shed light on the further study of E. coli–host interaction.
Materials and methods
Bacterial strains and media
Escherichia coli strain LCT-EC106 was obtained from China General Microbiological Culture Collection Center (CGMCC). The designed plastic containers were filled with semi-solid LB medium where E. coli was inoculated. Two copies of these plastic containers were prepared in this study. One group was sent into space on board of the Shenzhou-8 rocket on 1 November 2011, and returned after orbiting around the earth for approximately 397 h. The other was simultaneously put into an incubator (to simulate the temperature exposure of those in space) as a control experiment. According to the data from general headquarters of China manned space engineering the temperature in the capsule was reported per hour and ranged from 18 to 24°C. With the exception of spaceflight, all other culture conditions were identical between the two groups. After the return of the spacecraft to Earth, E. coli was taken out of the containers and the sample was immediately grown on solid agar plates with nutrients using the streaking method. Subsequently, 321 clones were randomly picked from each plate and screened through a series of phenotypic analyses, including Biolog assay for substrate utilization, morphological examinations, growth curve calculations and antibiotic susceptibility test. Two flight strains, LCT-EC52 and LCT-EC59, derived and differed from the original LCT-EC106 on the basis of the phenotypic characteristics were selected for further study.
The culture medium (LB medium) for all strains was consisted of tryptone (10 g l−1), yeast extract (5 g l−1), NaCl (10 g l−1) and agar (Difco) powder (15 g l−1). The pH of the culture medium was adjusted to 7.0–7.2. The culture broth (LB broth) was identical to the culture medium without the Bacto Agar.
Substrate utilization screening
The carbon utilization analysis of all strains with 95 substrates was determined with Biolog GEN III MicroPlate (Biolog, Hayward, CA) according to the manufacturer's protocol. Briefly, the three strains were streaked on LB plates and incubated at 37°C for 18 h. Colonies were then scraped from the plates and suspended into 15 ml IF-A inoculating fluid to a finale optical density at 600 nm (OD600) of 0.2. An aliquot of 100 μl bacterium suspension was added to each well on the microplate. The plates were incubated at 37°C for 48 h. Bacterial growth was assessed by the colour change from the reduction of the redox dye and corrected for the no-substrate control.
Morphological characteristics
All strains were incubated on LB medium under aerobic condition at 37°C for 16 h before colony phenotype observation. Gram reaction was determined via a non-staining method using 3% KOH solution (Smibert & Krieg Reference Smibert and Krieg1994). Cell morphology was examined using phase contrast microscope (Olympus BX51, Japan).
Antibiotic susceptibility test
Susceptibility tests were performed to determine the strains’ vulnerability to antibiotics by the disc diffusion method (Bauer et al. Reference Bauer, Kirby, Sherris and Turck1966). The LB agar plates were inoculated with cell suspension of a density of 107–108 CFU ml−1, and the antibiotics-impregnated discs were placed on the surface of the plate. After incubation (16 h at 37°C), the diameter of the zone of inhibition around each disc was measured. In total, 17 antibiotics including penicillin G, ampicillin, cefazolin, ceftazidime, ceftriaxone sodium, azithromycin, ciprofloxacin, lincomycin, vancomycin, cotrimoxazole, chloramphenicol, sulperazone, amikacin, streptomycin, minocycline, meropenem and piperacillin were tested on the three strains.
Bacteria growth curves
The three strains were grown overnight at 37°C in LB broth. An aliquot of 20 μl suspension was inoculated into a microtitre plate (honeycomb plate) containing 350 μl LB broth, which was monitored by Bioscreen C (Lab Systems, Helsinki, Finland) at 37°C with continuous shaking. OD600 of each strain was measured and the average of three measurements was reported. A well containing only 370 μl LB was also included as a blank control.
Genome sequencing and data analysis
Genomic DNA of the two flight strains and the control strain were extracted with the CTAB method and fragmented to the sizes of 500 bp and 6 kb with Covaris E-210 ultrasonicator. DNA libraries of 500 bp and 6 kb insertion size were constructed with Illumina kit according to the standard protocols and were sufficient for ~100 and ~50 genome coverage, respectively. Ninety bp paired-end sequencing of these libraries were performed on Illumina HiSeq2000 according to the manufacturers’ instructions.
Illumina base-calling pipeline (version HCS1.4/RTA1.12) was applied to process raw fluorescent images and call bases. To ensure accuracy of bioinformatics analysis, sequencing reads with three consecutive bases of quality ≤Q2 or four unknown bases were removed. After removal of adapter sequences and duplicates, remaining reads were assembled into contigs with SOAPdenovo (http://soap.genomics.org.cn). The structural information contained in 6 kb paired-end sequencing libraries was used to connect contigs into scaffolds.
Assembled scaffolds of the control strain LCT-EC106 were analysed with GlimmerV3.02 to predict coding sequences that were then aligned to NR, COG and KEGG databases for annotation. The rRNA and tRNA were identified with RNAmmer and tRNAscan-SE1.21, respectively.
Single nucleotide polymorphisms (SNPs) were identified by aligning reads from flight strains LCT-EC52 and LCT-EC59 to the control strain LCT-EC106 scaffolds. The software used in this analysis was MUMer 3.0 with default parameters. In order to ensure low false positive rate, candidate SNPs located on scaffold-gaps, scaffold-ends and repeat regions were discarded. Paired-end reads were mapped to the reference genome to validate candidate SNPs with the following criteria: Reads supporting most frequent (f1) and second frequent (f2) genotype on each SNP locus should count more than 10, and f1/f2 is required to be larger than 5; otherwise this candidate SNP will be discarded.
Insertion or deletion was identified by aligning LCT-EC52 and LCT-EC59 scaffolds to LCT-EC106 scaffolds. Gapped alignments were extracted as InDel candidates and were validated by aligning short reads to regions 150 bp around InDel loci with BWA.
Transcriptome sequencing
An aliquot of 20 ml of each E. coli strain LCT-EC106, LCT-EC52 and LCT-EC59 were grown in LB broth to stationary phase before harvest. Total RNA from the three strains were extracted with RNeasy Plant Mini Kit (Qiagen, CA, USA) and fragmented to ~200 nt. mRNA were reverse transcribed into cDNA with random hexamer primers. cDNA libraries were constructed according to the standard protocol (Nagalakshmi et al. Reference Nagalakshmi, Waern, Snyder and Frederick2010) and subsequently sequenced with Illumina HiSeq™ 2000 according to the manufacturer's instructions. Raw reads were similarly treated as the genomic sequencing data. Gene expression levels were expressed as Reads Per Kb per Million reads (RPKM; Mortazavi et al. Reference Mortazavi, Williams, McCue, Schaeffer and Wold2008).
Differentially expressed genes (DEGs) analysis
To identify DEGs between different samples, the following method was developed based on (Audic & Claverie Reference Audic and Claverie1997).
Denote the number of reads from gene A as x. p(x) is subject to Possion distribution:
$$p(x) = \displaystyle{{{\rm e}^{ - {\rm \lambda}} {\rm \lambda} ^x} \over {x!}}\,({\rm \lambda} \,\,{\rm is}\,\,{\rm the}\,\,{\rm real}\,\,{\rm transcripts}\,\,{\rm of}\,\,{\rm the}\,\,{\rm gene}).$$
The total read number of the sample 1 is N 1, and total read number of sample 2 is N 2; gene A holds x reads in sample 1 and y reads in sample 2. The probability that gene A expresses equally in the two samples can be calculated with:
$$\eqalign{& 2\sum\limits_{i = 0}^{i = y} {\,p(i\vert x)}, \cr & {\rm or}\,\,\, 2 \times \left( {1 - \sum\limits_{i = 0}^{i = y} {\,p(i\vert x)}} \right) \left( {{\rm if} \sum\limits_{i = 0}^{i = y} {\,p(i\vert x)} \gt 0. 5} \right), \cr & p(y\vert x) = \left( {\displaystyle{{N_2} \over {N_1}}} \right)^y \displaystyle{{(x + y)!} \over {x!y!\left( {1 + (N_2 /N_1 )} \right)^{(x + y + 1)}}}.} $$
P-value corresponds to differential gene expression test. Correction for false positive and false negative is performed using the False discovery ratio (FDR) method proposed by Benjamini & Yekutieli (Reference Benjamini and Yekutieli2001). FDR £0.001 was used as the threshold for the significance of gene expression difference.
Gene ontology (GO) analysis of DEGs
All DEGs were mapped to GO terms in the database (http://www.geneontology.org/). For each term, matched gene numbers were calculated, followed by the hyper geometric test for significantly enriched GO terms with a modified version of GO::TermFinder as the following:
$$P = 1 - \sum\limits_{i = 0}^{m - 1} {\displaystyle{{\left( \matrix{M \cr i } \right)\left( \matrix{N - M \cr n - i } \right)} \over {\left( \matrix{N \cr n } \right)}}}, $$
where N is the number of all genes with GO annotation; n is the number of DEGs in N; M is the number of all genes that are annotated to the certain GO terms; m is the number of DEGs in M. The calculated -value underwent the Bonferroni Correction, and corrected P-value £0.05 was selected as the threshold for significance. GO terms satisfying this condition were defined as significantly enriched GO terms in DEGs.
Pathway enrichment analysis of DEGs
Significantly enriched metabolic pathways or signal transduction pathways in DEGs were identified in KEGG database with the same algorithm as GO Analysis with minor modifications. The N in the above equation is the number of all genes with KEGG annotation, n is the number of DEGs in N, M is the number of all genes annotated to specific pathways, and m is the number of DEGs in M. Terms with corrected P-value £ 0.05 was considered as enriched in DEGs.
Proteomics analysis
Proteins from E. coli strains LCT-EC106, LCT-EC52 and LCT-EC59 were isolated and analysed using the iTRAQ combined with two-dimensional (2D) nano LC-MS/MS protocol. Isolated protein pools were quantified and identified with ProteinPilotÔ 4.0.8085. Paragon Algorith 4.0.0.0 was applied for protein spectrum search in the genome annotation database, allowing all post-translational modifications. All identified proteins with lower than 5% FDR were selected for further analysis according to following criteria: (1) peptides unique for a given protein; (2) proteins that contain at least one peptide with a confidence level greater than 95%; and (3) confidence level of protein sequence coverage (the number of matching amino acids/total number of amino acids for a given protein) greater than 95%.
Three biological replicates were performed for each strain. A given protein was considered differentially expressed between two strains only if its expression level showed a change greater than 1.2-fold in at least two replicates and a P value of <0.05.
The COG, GO functional analysis and KEGG Pathway analysis for proteomics were conducted in the same manner as transcriptome.
Results
Biochemical and morphological characteristics of two flight strains
Based on the substrate utilization screening, the utilization of a variety of substrate of strain LCT-EC52 and LCT-EC59 had changed after a 17 day flight in space compared with the control strain LCT-EC106. Carbon utilization analysis of the three strains was performed using Biolog assay (Biolog GEN III MicroPlate). Growth occurred in all wells; after 48 h incubation at 37°C, colour change was observed in all wells (Supplementary Fig. 1D-F online). The three strains shared identical responses to a total of 66 reactions; whereas the rest 30 wells were different (Table 1). Both flight strains lost response to the sole carbon source of D-cellobiose, sucrose, D-turanose, stachyose, D-raffinose, 1% sodium lactate, D-serine, myo-inositol, D-aspartic acid, L-arginine, L-histidine, L-pyroglutamic acid, pectin, quinic acid, citric acid and Tween-40. Noticeably, only the flight strain LCT-EC59 lost response to D-lactic acid methyl ester but showed enhanced response to dextrin. In contrast, the strain LCT-EC52 lost response to D-salicin, 3-methyl glucose, D-fucose, D-arabitol, gelatin, methyl pyruvate, potassium tellurite and sodium bromate. The extensive variation of sole-carbon-source utilization between the flight strains and the control strain indicated the metabolism pathways underwent substantial changes during space flight.
Biochemical characteristics of the three strains in this study
+, positive reaction; ++, significantly positive reaction; –, negative reaction.
A series of biologic studies were performed subsequently. Morphological studies showed no difference between the two flight strains LCT-EC52, LCT-EC59 and the control strain LCT-EC106, indicating the morphology had not changed appreciably (Supplementary Fig. 1A–C online). Furthermore, we did not observe any difference between the flight strains and the control strain in their resistance to 17 antibiotics (Table 2). Similarly, the comparison of growth curves showed that the flight strains exhibited an identical growth rate as the control strain during the entire growth period (Fig. 1).
Growth curves of the two flight strains LCT-EC52, LCT-EC59 and the control strain LCT-EC106. OD600 was measured in every hour for a period of 24 h and the average of three reads was plotted.
Antibiotic sensitivity test of the three strains in this study
Genome comparison of the flight strains and the control strain
The 500 bp and 6 kb insert size sequencing libraries were assembled to 227, 191 and 175 contigs for LCT-EC106, LCT-EC52 and LCT-EC59, respectively with SOAPdenovo (version 1.6). The contig sets were connected to 38, 37 and 33 scaffolds in LCT-EC106, LCT-EC52 and LCT-EC59, respectively by using 6 kb paired-end libraries information. N50 of the assembled scaffolds was 2676 640 bp and GC content was 50.37% in LCT-EC52, and correspondingly 5512 598 bp and 50.38% in LCT-EC106. The N50 of LCT-EC59 was 2712 414 bp and GC content was 50.37%.
Synteny analysis was subsequently performed by aligning the assembled genomes with MUMmer v3.22. The three strains were very similar at genomic level (Fig. 2), indicating their phenotype variation was attributed to changes in transcriptome.
Comparison of genome structures of LCT-EC52 and LCT-EC59 (a) and LCT-EC59 and LCT-EC106 (b) by the whole-genome synteny analysis.
To identify single nucleotide polymorphism (SNP), we aligned scaffolds with MUMmer v3.22. Mismatches in the alignment were reported as candidate SNPs, which were further filtered with stringent criteria. Briefly, we first aligned sequencing reads to scaffolds with BLAT, and discarded reads alignments that (1) have candidate SNPs in both 5 bp ends, and/or (2) have sequencing quality less than 20 on the candidate SNP sites. After discarding these reads alignments, candidate SNPs with more than ten read alignments support were finally reported. Taking the control strain LCT-EC106 as reference, we identified a non-synonymous SNP on 178 505 of Scaffold 6_2 in LCT-EC59, which is associated with a galactarate transporter gene. We also identified one intergenic SNP located at locus 55 270 of Scaffold 5 in LCT-EC52 and LCT-EC59. Finally, both flight strains harboured a deletion located in an intergenic region corresponding to 173 775 locus on Scaffold 5 of the control strain (Table 3).
The comprehensive difference between the flight strains and the control strain in genome, transcriptome and proteome
↑, up-regulated; ↓, down-regulated; DEGs, differentially expression genes; DEPs, differentially expression proteins.
Transcriptome comparison of the flight strains and the control strain
A large set of genes were differentially expressed between LCT-EC52 and LCT-EC106. In total, 1217 genes had significantly different expression levels (FDR≤0.001 and fold change > 2) between the two strains, among which 1001 genes were up-regulated and 216 were down-regulated in LCT-EC52 (Table 3). GO analysis showed that 584, 734 and 756 DEGs were successfully mapped to cellular components, molecular function, and biological processes, respectively (Fig. 3).
Gene ontology analysis of differentially expression genes between the two flight strains and the control strain. P, process ontology; F, function ontology; C, cellular ontology. Each bar represent the percentage and number of genes in each gene ontology category for strain LCT-EC52 (red bar) and LCT-EC59 (blue bar).
To identify enriched GO terms in DEGs, we applied the hyper geometric test to compare the enrichment of each GO term in DEGs and total genes (see the Materials and Methods section for details). In cellular component, 4.3% of DEGs were mapped to the term flagellum compared with only 1.3% of total genes in this category (P = 7.34×10− 10), indicating flagellum associated genes were enriched in DEGs. Similarly, we found other important terms that were enriched in DEGs, including bacterial-type flagellum (P = 5.22 × 10− 9), and cell projection (P = 4.15 × 10− 8). In function ontology, the term transmembrane transporter activity (P = 4.09 × 10− 5) and transporter activity (P = 0.00026) were overrepresented in DEGs. For process ontology, 27 out of 756 genes (3.6%) were mapped to cellular component movement, while 32 out of 3101 genes with process ontology (1.0%) (P = 4.04 × 10− 10). Twenty-one DEGs were mapped to cell projection organization, whereas 22 of total genes were mapped to this term (Fig. 3).
The comparison of LCT-EC59 and LCT-EC106 showed that 246, 313 and 329 DEGs were successfully mapped to cellular components, molecular function, and biological processes respectively. Enriched ontology terms in DEGs included flagellum (8.1% of DEGs, P = 4.97 × 10− 12), cell projection while (14.6% of DEGs, P = 9.40 × 10− 11), cellular component movement (6.1% of DEGs, P = 3.36 × 10− 10) and ion transmembrane transporter activity (8.9% of DEGs, P = 0.57514) (Fig. 3).
To shed light on metabolism difference between the flight strains and the control strain, we performed KEGG pathway annotation for DEGs of the three strains. For comparison of LCT-EC52 and LCT-EC106, 922 DEGs and a total of 3739 genes were mapped to KEGG pathways. Among all annotations, flagellar assembly was the most significantly enriched term in DEGs (4.56% of DEGs, 1.68% of total genes, P = 1.5 × 10− 12), followed by bacterial chemotaxis (1.95% of DEGs, 1.15% of total genes, P = 0.009463), and oxidative phosphorylation (2.49% of DEGs, 1.58% of total genes, P = 0.00993881). For comparison of LCT-EC59 and LCT-EC106, 397 DEGs and a total of 3739 genes were mapped to KEGG pathways. The significantly enriched KEGG terms in DEGs were flagellar assembly pathway (6.05% of DEGs, 1.68% of total genes, P = 6.68 × 10− 9) and bacterial chemotaxis (2.52% of DEGs, 1.15% of total genes, P = 0.012506).
Proteome comparison of the flight strains and the control strain
To compare protein expression levels between the flight strains and the control strain, we constructed a detailed reference protein expression profile. Briefly, we isolated and identified proteins from three biological replicates of the two flight strains LCT-EC52, LCT-EC59 and the control strain LCT-EC106. A total of 1752 proteins that appeared in at least two replicates of the three strains with a FDR lower than 5% were determined as expressed and were subsequently categorized based on COG, GO and KEGG classification (Table 3, Supplementary Fig. 2, 3 and 4 online; proteins without annotated GO term were removed from the figures).
To identify differentially expressed proteins (DEPs) between the flight strains and the control strain, we first screened the 1752 proteins for candidates that showed a fold change in the expression level greater than 1.2 between either of the flight strain and the control strain for at least two replicates. We then calculated the P-value for each candidate protein and filtered out ones that had a P-value greater than 0.05. A few different proteins were identified as differentially expressed between either of the flight strain and the control strain, and between the two flight strains as well (Fig. 4). In total, seven DEPs were discovered between flight strains and the control strain LCT-EC106 (Table 4). GO enrichment analysis indicated these proteins were involved in biological processes, including phospholipid biosynthesis, phospholipid metabolism, organophosphate metabolism, lipid biosynthesis, response to chemical stimulus, cellular lipid metabolism and lipid metabolism.
Numbers of proteins differentially expression between each flight strain and the control strain, and between two flight strains.
The differentially expressed proteins between the flight strains (LCT-EC52 and LCT-EC59) and the control strain (LCT-EC106)
Discussion
In our study, the bacteria were partially protected by the cabin from cosmic radiation, therefore the major challenge for the microbe was microgravity. Compared with the control strain LCT-EC106 on the ground, the morphology, growth kinetics, antibiotics sensitivity and genomes of the flight strains did not show significant difference, which may partially be attributed to the very limited cosmic radiation the bacterium experienced during the short spaceflight. The limited impact of radiation hitting some small number of individual bacterial cells is not likely to have an effect on the overall outcome. Furthermore, it has been suggested that non-motile bacteria tend to exhibit different growth kinetics after spaceflight or microgravity simulation, while the growth of motile bacteria did not seem to be affected under similar conditions (Bouloc & Dari Reference Bouloc and Dari1991). Although all of the measurements in this article were conducted in 1 g, with the only difference being that the experimental group had been previously exposed to spaceflight while the controls had remained on the Earth, Our study on E. coli again substantiated this hypothesis in some extent.
Despite the few changes in morphological and genomic properties, a large number of metabolic differences were identified in the two flight strains including utilization of multiple carbon sources. From the transcriptomic and proteomic perspective, the two flight strains diverged from each other and the original control strain in the expression of many different genes and proteins (online Supplementary Fig. 5), most of which were involved in processes such as cell motility, nutrient transportation, and lipid metabolism. The proteomic analysis identified seven statistically significant items of biological process including phospholipid biosynthetic process, phospholipid metabolic process, organophosphate metabolic process, lipid biosynthetic process, response to chemical stimulus, cellular lipid metabolic process, and lipid metabolic process. The lipid biosynthesis gene yfbE was identified up-regulated in both our study and Vukanti et al. (Reference Vukanti, Mintz and Leff2008). In our study, we also noticed significant changes in the expression of transporter genes as reported by others (Vukanti et al. Reference Vukanti, Mintz and Leff2008; Arunasri et al. Reference Arunasri, Adil, Charan, Suvro, Reddy and Shivaji2013). However, we did not observe significant enhancement of growth rate or stress related genes in the flight strains and their sensitivity to 17 antibiotics was not altered. Although ground-based microgravity simulation system to a large extent recapitulates the real spaceflight environment, subtle differences between these two conditions may be critical to induce significantly different responses (Horneck et al. Reference Horneck, Klaus and Mancinelli2010). Our study would be a valuable comparison with these simulation studies to model the responses of microbes to microgravity and understand the underlying mechanism.
Considering the massive transcriptional and translational changes in the metabolism pathways in the flight strains, it would be interesting to test in future whether these changes will affect the interaction between bacteria and the environment, such as the efficiency of nutrients transfer and utilization. More importantly, it is crucial to understand in future how these gene and protein expression changes may alter the relationship between E. coli and its host. Recent studies started to reveal the host–bacteria interaction under space flight environment (Duray et al. Reference Duray, Hatfill and Pellis1997; Nickerson et al. Reference Nickerson, Ott, Mister, Morrow, Burns-Keliher and Pierson2000; Chopra et al. Reference Chopra, Fadl, Sha, Chopra, Galindo and Chopra2006; Foster et al. Reference Foster, Khodadad, Ahrendt and Parrish2013). Using host squid Euprymna scolopes and bacterium Vibrio fischeri as a model system, Foster et al. (Reference Foster, Khodadad, Ahrendt and Parrish2013) found that while the host immune response was weakened under simulated microgravity, the host tissues underwent rapid apoptosis and regression. Therefore, as a common symbiosis bacterium with human, how the changes of E. coli in space environment may affect the physiology of its host requires extensive investigation.
Conflict of interest
The authors state no conflict of interest.
Supplementary materials
Supplementary material accompanies this paper on the Journal's website (http://journals.cambridge.org/IJA).
Acknowledgement
This work was supported by the National Basic Research Programme of China (973 Programme, No. 2014CB744400) and National Major Scientific and Technological Special Project for ‘Significant New Drugs Development’ (grant no. 2015ZX09J15102-003).
