Open Access

Parallel evolution of genome structure and transcriptional landscape in the Epsilonproteobacteria

  • Ida Porcelli1,
  • Mark Reuter1,
  • Bruce M Pearson1,
  • Thomas Wilhelm1 and
  • Arnoud HM van Vliet1Email author
BMC Genomics201314:616

DOI: 10.1186/1471-2164-14-616

Received: 4 June 2013

Accepted: 3 September 2013

Published: 12 September 2013



Gene reshuffling, point mutations and horizontal gene transfer contribute to bacterial genome variation, but require the genome to rewire its transcriptional circuitry to ensure that inserted, mutated or reshuffled genes are transcribed at appropriate levels. The genomes of Epsilonproteobacteria display very low synteny, due to high levels of reshuffling and reorganisation of gene order, but still share a significant number of gene orthologs allowing comparison. Here we present the primary transcriptome of the pathogenic Epsilonproteobacterium Campylobacter jejuni, and have used this for comparative and predictive transcriptomics in the Epsilonproteobacteria.


Differential RNA-sequencing using 454 sequencing technology was used to determine the primary transcriptome of C. jejuni NCTC 11168, which consists of 992 transcription start sites (TSS), which included 29 putative non-coding and stable RNAs, 266 intragenic (internal) TSS, and 206 antisense TSS. Several previously unknown features were identified in the C. jejuni transcriptional landscape, like leaderless mRNAs and potential leader peptides upstream of amino acid biosynthesis genes. A cross-species comparison of the primary transcriptomes of C. jejuni and the related Epsilonproteobacterium Helicobacter pylori highlighted a lack of conservation of operon organisation, position of intragenic and antisense promoters or leaderless mRNAs. Predictive comparisons using 40 other Epsilonproteobacterial genomes suggests that this lack of conservation of transcriptional features is common to all Epsilonproteobacterial genomes, and is associated with the absence of genome synteny in this subdivision of the Proteobacteria.


Both the genomes and transcriptomes of Epsilonproteobacteria are highly variable, both at the genome level by combining and division of multicistronic operons, but also on the gene level by generation or deletion of promoter sequences and 5′ untranslated regions. Regulatory features may have evolved after these species split from a common ancestor, with transcriptome rewiring compensating for changes introduced by genomic reshuffling and horizontal gene transfer.


While our appreciation of microbial diversity has been greatly increased by the exponential increase in the availability of genome sequences and by metagenomic approaches [1, 2], it has also highlighted our relative lack of understanding about what drives variation, and which limitations and constraints control the process of genome variation. Diversity at the level of gene order and genome content can be introduced via the reorganisation of the genome, through combinations of gene inversion, recombination, gene duplication, deletion and horizontal gene transfer [3, 4]. Such movement, deletion or introduction of genes or operons can create a problem for the cell, as the reorganisation of the genome may result in disruption of transcriptional circuitry controlling the expression levels of such genes. However, variability can also be introduced at the gene level, e.g. by generation of alternative transcription start sites, promoter recognition sequences or alterations in the 5′ untranslated regions affecting folding or stability.

The level of RNA in a cell is usually controlled at the transcriptional and post-transcriptional levels. In bacteria, transcriptional regulation is commonly mediated via control of transcription initiation by RNA polymerase (RNAP) at the promoter [5]. Alternatively, post-transcriptional gene regulation is often mediated by the (often combined) action of non-coding or antisense RNA [6], RNA chaperones [7] and the activity of ribonucleases [8]. In the last two years, the use of high-throughput sequencing of cDNA (RNA-seq) has revealed that the complexity of the microbial transcriptome is much higher than previously predicted [913]. However, the high level of phylogenetic diversity within the bacterial kingdom has so far limited the possibilities for interspecies transcriptome comparison, since the species for which high resolution transcriptome maps are available are either too closely related (e.g. the Enterobacteriaceae) or too distantly related to allow meaningful comparisons at the evolutionary level.

The Epsilon-subdivision of the Proteobacteria (Epsilonproteobacteria) is a lineage which contains both pathogenic and non-pathogenic bacteria. The best studied examples of the former category are the human pathogens Campylobacter jejuni and Helicobacter pylori, which belong to the order Campylobacterales[14]. However, next to these important human pathogens, the Epsilonproteobacteria also contain chemolithoautotrophic microorganisms isolated from deep-sea vents [15, 16], as well as the bovine rumen-colonising bacterium Wolinella succinogenes[17]. Despite the differences in ecological niches between the genera, and the genome sizes of Epsilonproteobacteria varying between 1.5 and 2.6 Mbp, genomic comparisons revealed that the Epsilonproteobacteria share similar transcription machinery including few sigma factors (with the notable exception of Arcobacter butzleri[18]), metabolic pathways and limitations, and have about half of the predicted genes in the genome in common with other Epsilonproteobacteria [14, 15]. However, while these genomes share functionality, genome architecture and often low G + C content, the gene order and genome organisation have diverged significantly. This raises the question on how the genome and associated transcriptome copes with such large scale reorganisations of the genome when genera and species evolutionary diverge over time. To address this question, we have mapped the primary transcriptome of C. jejuni at the single nucleotide resolution using differential RNA-seq, have compared it with the primary transcriptome map of H. pylori[10] and have used genome sequences of 40 other taxa of the Epsilonproteobacteria to assess conservation and evolution of transcriptional circuitry in this highly variable group of bacteria.

Results and discussion

Differential RNA-seq analysis of the C. jejuni primary transcriptome

The C. jejuni NCTC 11168 genome contains 1643 annotated coding sequences (CDS), with only few stable RNA molecules known outside the ribosomal and transfer RNA species [1921]. A single nucleotide resolution map of the C. jejuni transcriptome was generated by differential RNA-sequencing (dRNA-seq, [10]) using a motile variant of C. jejuni strain NCTC 11168 [22] and Roche 454 sequencing. To assess whether the dRNA-seq cDNA libraries are a good representation of transcribed sequences of C. jejuni, we compared the RPKM-values obtained for the CDSs from the non-enriched (−TEX) 454 cDNA sequencing with the previously published Illumina-based RNA-seq data for C. jejuni NCTC 11168 [21] and the signal intensity on a PCR-product based C. jejuni microarray [23, 24] normalised to a genomic DNA reference [25]. There was a good correlation between the RPKM values for the two RNA-seq experiments and the microarray data (Additional file 1: Figure S1), with the best correlation observed between the two RNA-seq based approaches.

Genome-wide identification of C. jejuni transcription start sites and promoters

The dRNA-seq data were subsequently used for the identification of transcription start sites (TSS) of primary RNAs, which are protected from digestion by Terminator Exonuclease (TEX) through their 5′-triphosphate modification [10, 26]. The dRNA-seq method is based on the comparison of read distribution between the two cDNA-library enriched in primary 5′ ends (+TEX), and the non-enriched cDNA library (−TEX). Read distribution in the -TEX library displayed distribution throughout coding sequences, whereas treatment of RNA with Terminator Exonuclease for the + TEX library led to a typical cDNA read distribution resembling a sawtooth-like profile with an elevated 5′ flank [10, 27] (Figure 1A, Additional file 2: Figure S2). TSS were annotated as primary, secondary, internal and antisense, based on their genomic location and association with annotated features, according to previously described criteria (Additional file 2: Figure S2) [10]. We identified a total of 992 TSS in the C. jejuni transcriptome (listed in Additional file 3: Table S1), which consisted of 510 primary and 11 secondary TSS located in intergenic regions which are associated with an annotated feature (CDS, pseudogene or stable RNA). A total of 12 genes are transcribed from two independent promoters (Additional file 4: Table S2), with 266 TSS located inside coding sequences or pseudogenes, and 206 TSS located antisense to coding sequences or annotated features.
Figure 1

Analysis of the primary transcriptome of C. jejuni NCTC 11168 by differential RNA-seq (dRNA-seq) analysis of transcription start sites (TSS). (A) Genome map of C. jejuni strain NCTC 11168 with the outer three rings showing the annotated features of the plus and minus strand and annotated pseudogenes [19, 20]. The inner rings show TSS associated with annotated features (green), antisense to annotated features (red), and non-coding RNAs (blue). The top inset shows the TSS and promoter of the porA gene and a putative non-coding RNA (NC3, CJnc140). The lower inset shows an operon with a primary TSS of a leaderless mRNA, an internal TSS and an antisense TSS. Red histograms (+TEX) show the cDNA read distribution in the cDNA library enriched for primary transcripts, dark blue histograms show cDNA read distribution in the non-enriched library (-TEX). B) WebLogo representations of the recognition sequences of σ28, σ54 and σ70, based on the sequences upstream of 26, 18 and 948 TSS, respectively (Additional file 5: Table S3). C) Identification of leaderless mRNAs of C. jejuni NCTC 11168. The graph shows the length distribution of 5′ untranslated regions (5′ UTRs) of 471 C. jejuni genes located downstream of a primary TSS. Red bars indicate proposed leaderless mRNAs, blue bars indicate a 5′ UTR with ribosome binding site (RBS). The insets show the motifs detected by MEME [28] upstream of the translational start codon of the C. jejuni genes with an RBS (right) and leaderless mRNAs (left). On the right side examples of 5′ UTRs are shown, with 3 proposed leaderless mRNAs, and 4 short 5′ UTRs containing a proposed RBS. The TSS (green, underlined), the σ70 -10 sequence (red) and the proposed RBS (blue) are highlighted. The full list of leaderless mRNAs is given in Additional file 13: Table S6.

Comparison of the dRNA-seq TSS with 53 previously published C. jejuni TSS, and 8 additional TSS determined by 5′ RACE analysis for this study (Additional file 5: Table S3), showed that 32/61 dRNA-seq TSS were identical, and an additional 17 were within 2 nt distance (81.8%, Additional file 6: Figure S3) of the previously described TSS, a difference which may be caused by the difficulty of 454 sequencing to accurately read long homopolymeric stretches [10]. In addition, due to the low number of TSS available for strain NCTC 11168, we used TSS from other reference strains and clinical isolates, and hence there may be strain differences in TSS as well [29]. The remaining 12 TSS were previously reported to lack recognisable promoter sequences, and as they were obtained by primer extension probably represent the 5′ end of processed RNA species rather than primary RNAs. This percentage match is similar as that described previously for H. pylori[10] and Salmonella enterica serovar Typhimurium [30, 31]. Comparison with an independently performed study using Illumina sequencing published during preparation of this manuscript [29], showed that 795 TSS described in Additional file 3: Table S1 match TSS described in that study, but also highlights the identification of 197 additional TSS not described in [29]. There are several possible explanations for this discrepancy, which includes the different sequencing technology used (454 vs Illumina), as well as difference in growth conditions or growth phase of the C. jejuni cultures. It does highlight that the C. jejuni Supergenome described in [29] will undoubtedly be further expanded by future RNA-seq based studies with C. jejuni.

Analysis of C. jejuni promoter sequences

C. jejuni has three sigma factors for promoter recognition, with σ28 and σ54 thought to be primarily involved in flagellar biogenesis, and σ70 to function as major vegetative sigma factor [20]. This was confirmed by dRNA-seq analysis, as only 26/992 (2.6%) of TSS were preceded by a putative σ28 recognition sequence (5′ CGATwt at 6–8 nt upstream of the TSS, Figure 1B) and 18/992 (1.8%) of TSS were preceded by a σ54 recognition sequence (5′ GGaa-N6-tTGCTt at 8–13 nt upstream of the TSS, Figure 1B) [32, 33]. The remaining 948/992 (95.6%) of TSS were preceded by a gnTAnaAT motif at 4–8 nt upstream of the TSS (Figure 1B), consistent with a −10 Pribnow box for σ70[5]. As previously predicted [34], a −35 sequence was not present, but the sequences upstream of the −10 box showed a periodic signal centering on the −7, −17, −27 and −38 residues upstream of the TSS (Figure 1B, Additional file 7: Figure S4). We further analysed the average profile for 99 physico-chemical and geometrical DNA properties of the aligned σ70 promoters from position +1 to −51 (Additional file 7: Figure S4), including the corresponding 50 overlapping dinucleotides [35, 36]. This highlighted the conservation of 16 dinucleotides individually and the overall nucleotide, dinucleotide and physical properties conservation in comparison (conservation measured by entropy). Furthermore, the overall nucleotide and dinucleotide conservation is quite similar, whereas some properties are partly higher conserved, especially the two measures slide and entropy at positions –27,–26. Slide is known to be indicative for DNA stiffness [37], which is related to the DNA entropy. This indicates that the right DNA stiffness at these positions might support promoter functioning. We also found a significant correlation of two physical properties (inclination, direction of deflection angle) of neighboured dinucleotides at positions –31/–32 –33/–34 (Additional file 7: Figure S4). Overall there was a good correlation between the nucleotide sequence and physico-chemical and geometrical DNA properties of the aligned σ70 promoters. In addition, there was no difference observed between σ70 promoters upstream of internal TSS and antisense TSS when compared to primary TSS and secondary TSS in intergenic regions (not shown).

Genome-wide antisense transcription in C. jejuni

Within the 992 TSS, 206 were on the antisense strand of annotated features (antisense TSS, Additional file 3: Table S1, Additional file 8: Table S4), which confirmed the presence of genome-wide cis-antisense transcription, as recently described in other microbes [10, 3841]. We subsequently confirmed four antisense TSS by 5′ RACE (Additional file 5: Table S3), thus ensuring that the antisense TSS identified are not an artifact of the dRNA-seq technology. Antisense transcripts were often relatively short (27–285 nt in our dataset, average 114 nt), and many display a low number of reads, which may indicate spurious or pervasive transcription [41]. The presence of antisense TSS was not related to the level of transcription of the gene in either microarray or dRNA-seq, nor is antisense transcription related to specific functional categories of the genes opposite to the antisense TSS (Additional file 7: Table S4). Some genes have multiple antisense TSS, and antisense transcription was also detected opposite to transcriptionally active C. jejuni pseudogenes, which may allow for silencing of these pseudogenes via the activity of the double strand-specific ribonuclease III [29, 42] or block the progress of RNA polymerase via transcriptional interference [43]. Antisense RNA may contribute to downregulation of parts of operons by post-transcriptional modification, without a requirement for transcriptional regulators. Alternatively, since some of the antisense TSS were located at the 3′ end of the coding sequence, they may function in transcript termination.

Non-coding RNA

The annotations of the C. jejuni NCTC 11168 genome sequence [19, 20] suggested the presence of several species of non-coding and stable RNAs, such as rRNAs, tRNAs, tmRNA, RNase P and the signal recognition particle (SRP) RNA. Furthermore, the presence of a thiamine pyrophosphate (TPP)-responsive riboswitch was predicted upstream of the thiC gene [19, 21], as well as a possible purine riboswitch upstream of the purD gene [44], but no other ncRNA species were predicted or recognised, consistent with the absence of the Hfq RNA chaperone commonly associated with ncRNA-dependent regulation in bacteria [7, 45]. A total of 29 putative non-coding and stable RNAs (ncRNAs) were identified in intergenic regions, scattered over the C. jejuni genome (Additional file 9: Table S5). We confirmed transcription of eight of these ncRNAs using Northern hybridisation (Additional file 10: Figure S5). Most of the ncRNAs detected were relatively short (30–100 nt), consistent with the relatively small and densily packed nature of the C. jejuni genome. Transcription of other predicted ncRNAs (tmRNA, RNase P and SRP RNA) was confirmed using dRNA-seq, with the SRP RNA also being detected by Northern hybridisation (Additional file 10: Figure S5). Comparison of the C. jejuni sRNAs recently described by Dugar et al.[29] and the earlier C. jejuni RNA-seq study by Chaudhuri et al.[21] showed a good correlation with the first study with 8 new ncRNAs described here, but only partial overlap with the second study, as two proposed non-coding RNAs matched (NC15/CJnc110 and NC8/CJnc190), with the rest either gene promoters (such as Intergenic_671549–671895 which encodes a selW ortholog [46]) or absent in our study.

The highest transcribed C. jejuni non-coding RNA (next to rRNA and tRNA) is located upstream of the purD (cj1250) gene, and a shorter version of this sequence was previously proposed as potential purine riboswitch [44]. However, the same region was recently identified in H. pylori to harbor a homolog of the abundant 6S RNA, a widespread regulator of RNA polymerase that was first described in E. coli[47]. Investigation of the 185 nt transcript (Additional file 11: Figure S6) showed that it started further upstream than the previously predicted purine riboswitch, and that it folds in a structure corresponding to that of bacterial 6S RNA [47, 48], with a closing stem, central bubble and terminal loops (Additional file 11: Figure S6). The E. coli 6S RNA accumulates during exponential growth, and regulates the activity of σ70-containing RNAP by mimicking its open complex promoter structure, thus complexing σ70-cofactored RNAP [49]. In E. coli, RNAP releases itself from 6S RNA after a nutritional upshift by the production of a small product RNA (pRNA, 14–20 nt), originating in the central bubble [50]. Our original analysis did not show any such pRNA, but as our cut-off for cDNA reads was <18 nt, we also searched the <18 nt cDNA reads for sequences on the complementary strand of 6S RNA, and indeed found a 13 nt RNA antisense to the 6S RNA (Additional file 11: Figure S6) at a similar position as detected for one of the two pRNAs of H. pylori 6S RNA [10]. In C. jejuni, 6S RNA transcription is not significantly regulated in the different phases of exponential growth, and is not significantly altered upon growth cessation after exposure to pH 5.0 or 3.6 (Additional file 11: Figure S6), suggesting its role in C. jejuni may be distinct from that reported for E. coli.

Leader peptides upstream of amino acid biosynthetic genes

For TSS that are >50 nt upstream of the annotated translation initiation codon (ATG, GTG or TTG), we searched the putative 5′ untranslated region (5′ UTR) for the presence of a small open reading frame (ORF) with a potential ribosome binding site (RBS, aAGGa) upstream, as was recently described for the mfrX gene upstream of the C. jejuni mfrABE genes [51]. Several small ORFs were thus identified and the length of the corresponding 5′ UTR was corrected. For three of these short ORFs, a functional prediction can be made based on their location upstream of the leucine, tryptophan and methionine amino acid biosynthetic operons in C. jejuni (Additional file 12: Figure S7) [52]. These three ORFs are likely to encode regulatory leader peptides, which couple transcription of amino acid biosynthetic genes to the availability of amino acid-coupled tRNAs [52], which has not been described for Epsilonproteobacteria. The small ORF (28 aa, tentatively named LeuL) upstream of the leuABCD (cj1719c-1716c) genes contains 5 Leu-codons at the C-terminal end of the polypeptide, which are all rare codons for leucine in C. jejuni (CUA, CUC and CUG), which together represent only 10.5% of the Leu codons in C. jejuni. Similarly, the short ORF (24 aa, tentatively named TrpL) upstream of the trpEDFBA (cj0345-0349) genes does contain a single Trp-codon at the C-terminal end of the polypeptide. Finally, a third short ORF (20 aa, tentatively named MetL) is located on a short RNA preceding the metBA (cj1727c-1726c) genes, with 3 Met-codons (Additional file 12: Figure S7). The RNAs encoding these ORFs all terminate shortly behind the stopcodon, and we propose that these polypeptides function as leader peptides, which allow transcription termination in the absence of ribosome stalling, and antitermination when the ribosome stalls due to the lack of availibility of tRNAs charged with the respective amino acid [52, 53].

5′ untranslated regions and leaderless mRNAs

The average length of the 5′ untranslated regions (5′ UTRs) from 471 primary TSS ranged from 0 to 158 nt (average 30.6 ± 17.8 nt). A motif search using the MEME Motif discovery tool identified the sequence of 5′-aAGGa as conserved RBS motif (Figure 1C). The relatively short 5′ UTRs of the other promoters in intergenic regions are consistent with the C. jejuni genome being tightly packed, since >93% of the genome is thought to contain functional regions [20, 54]. With the exception of the annotated TPP riboswitch upstream of the cj0453 (thiC) gene [19], there were no metabolite-sensing riboswitches detected in the collection of 5′ UTRs. This is consistent with a previous study predicting an absence of these structures in C. jejuni and related bacteria [44].

Nineteen of the 5′ UTRs were <10 nt in length, with 12/19 of the TSS starting on the first nucleotide of the translation initiation codon, and these 5′ UTRs lacked a recognisable Shine-Dalgarno (ribosome binding site, RBS) sequence, with all the connected genes having an ATG startcodon (Figure 1C, Additional file 13: Table S6) and are preceded by a TAnAaT σ70 promoter sequence (Figure 1C, Additional file 13: Table S6). Such mRNAs are known as leaderless mRNAs [55], and were previously thought to be rare in bacteria. Leaderless mRNAs allow for translation during a range of physiological conditions, without competition for 30S ribosomes [55, 56]. The genes translated from the C. jejuni leaderless mRNAs indeed encoded proteins predicted to be involved in stress-responses, like the DNA repair systems Nth endonuclease III (cj0595c) and MutY (cj1620c), the outer membrane efflux protein CmeD (cj1031) and the predicted multidrug efflux pump cj1257c (Additional file 13: Table S6).

Comparison of primary transcriptomes of C. jejuni and H. pylori

The availability of dRNA-seq datasets for H. pylori[10] and C. jejuni (this study) allowed for a direct transcriptome comparison between two relatively distant species within the order Campylobacterales. Both species are pathogenic to man, have a similar genome size (~1.7 Mbp) and cellular morphology, and colonise mucus layers within the mammalian and avian gastrointestinal tracts.

Gene order

We first used BLASTP to compare the ORFs annotated in both genomes, and found they share 881 ORFs when only counting the highest scoring ortholog. These are however not ordered similary, as the gene order- based genome synteny was very low (Figure 2A, B), with only 10 regions where 5 or more orthologs were in the same order. The longest regions containing a conserved gene order are a ribosomal operon (cj1708c-1688c, hp1320-1300), the operon encoding a putative NADH-ubiquinone oxidoreductase (cj1579c-1566c, hp1260-1274) and the operon containing the F1F0 ATPase (cj0098-0116, hp1141-1126), but the latter two still contain insertions of non-orthologous genes within the region. Other regions, such as the region containing the spoT gene [57, 58] are contiguous in the C. jejuni genome, but split over two regions in the H. pylori genome [59] (Figure 2A). We used one of these conserved regions (cj1274c-1271c, hp0777-0774) to directly compare the primary transcriptomes of both organisms (Figure 2B). Even though the gene order is conserved, the location of promoters is not. Both sets of genes are transcribed from a σ70-dependent promoter upstream of the pyrH gene, but the location of the internal promoters differs completely between C. jejuni and H. pylori, with the latter having an additional promoter upstream of rpoZ, and although both genomes have an internal promoter in the tyrS gene, its location is not conserved (Figure 2B).
Figure 2

The lack of genome synteny between C. jejuni and H. pylori at the level of gene order is reflected at the level of TSS/promoter location and non-coding RNAs. A) A scatter plot of annotated gene numbers shows the lack of genome synteny and gene order between C. jejuni NCTC 11168 (Cje) and H. pylori 26695 (Hpy). The red and blue circles represent the two regions compared below the graph, with arrows representing conserved genes (green), genes only found in either C. jejuni or H. pylori (white). Orange bars indicate homologous genes. Line arrows above the gene representations indicate TSS and associated promoters. B) Alternative visualisation of genomic location of orthologs between C. jejuni NCTC 11168 and H. pylori 26695, based on pairwise BLASTP comparisons of annotated features, using the Artemis Comparison Tool (ACT) [60]. Red lines indicate the presence of an ortholog between C. jejuni and H. pylori in the same transcriptional orientation, whereas blue lines indicate inverted transcriptional orientation. The spoT genomic region [57, 58] is used as example of a region with gene order conserved between the species, but a lack of conservation of transcriptional features. The dRNA-seq profiles of the C. jejuni cj1274c-1271c region and the H. pylori hp0777-hp0774 region (+TEX = enriched for TSS, -TEX = not enriched) are shown above and below, respectively. Promoters associated with TSS are indicated in black (primary TSS), blue (internal TSS for C. jejuni) and green (internal TSS for H. pylori). C) Non-coding RNAs of C. jejuni and H. pylori are not conserved between these two species with the exception of the 6S RNA (right), and are associated with a lack of conservation of flanking genes. One example of each category (C. jejuni-specific ncRNA, H. pylori-specific ncRNA and a conserved stable RNA) is shown. Orange bars connect orthologous genes.

Non-coding RNA

Similarly, non-coding RNAs are not conserved between C. jejuni and H. pylori[29], with the exception of the stable RNAs like the 6S RNA. When the genomic locations of one H. pylori ncRNA (nc5490 [10]) and one C. jejuni ncRNA (NC4, CJnc170) were compared (Figure 2C), this showed that the neighbouring genes are not conserved between the species, which may explain the species-specificity of the ncRNAs. One possible explanation for the uniqueness of the ncRNAs identified here, may be that these ncRNAs are generated and deleted during genome reorganisations and gene reshuffling. The exception is the 6S RNA, which is in both genomes upstream of the purD gene (Figure 2C), although the upstream gene differs, and also there is significant sequence difference between the 6S RNA genes of both organisms [10].

Antisense transcripts and internal promoters

Both the C. jejuni and H. pylori genomes only encode σ28, σ54 and σ70 sigma factors, with the large majority (>95%) of promoters being transcribed by σ70[10]. The H. pylori and C. jejuni σ70 promoters show a high degree of homology, both with the gnTAnaAT motif as −10 box [10], and higher conservation of the T-residues at −7, −17, −27 and −38 (Figure 3A). To compare for conservation of the transcriptional landscape between C. jejuni and H. pylori, we compared the genes in both genomes which have a) antisense transcription or an internal promoter and b) an ortholog in the comparator genome. This allowed for the comparison of 383 H. pylori genes with antisense transcription detected [10] with 82 C. jejuni genes for which antisense transcription was detected. Of these only 46 of these genes displayed antisense transcription in both C. jejuni and H. pylori (Figure 3B), however, for only two of these genes (cj0509/hp0264 and cj0774/hp1576) the location of the antisense TSS was conserved. The antisense RNA in the clpB (cj0509c/hp0264) gene is located in a part encoding a conserved sequence in the ClpB protein, with the antisense RNA being of identical length (105 nt), with both promoter and asRNAs being ~80% identical in DNA sequence. In contrast, 44/46 other asRNAs were located in different parts (albeit sometimes closely located) of the corresponding C. jejuni and H. pylori genes, as is shown in Figure 3C for the gyrB (cj0003/hp0501) gene, with the H. pylori asRNA being located in the first half of the gene, and the C. jejuni asRNA being located in the part encoding the C-terminal end of the protein. The σ70 promoter motifs were not conserved in these two genomes, with in both cases the highly conserved −7 T residue being altered, thus probably inactivating the σ70 recognition sequence. Similar results were obtained with the 218 internal promoters of H. pylori and 122 internal promoters of C. jejuni. Only 41 of these internal promoters were located in orthologs shared between H. pylori and C. jejuni (Figure 3B), and of these 41, only six have the internal promoter in the same position in the gene, thus highlighting the lack of conservation of the transcriptional landscapes between H. pylori and C. jejuni.
Figure 3

Lack of conservation of antisense promoters and internal promoters between C. jejuni and H. pylori . A) Comparison of the σ70 recognition sequences of C. jejuni and H. pylori, demonstrating that there is little difference between σ70 recognition sequences in the two species, with the σ70 -10 sequence of both species being gnTAnaAT. B) Comparison of the number of genes with antisense TSS and internal TSS in C. jejuni and H. pylori. Both species share 46 orthologs displaying antisense transcription in both species, whereas 41 shared orthologs have an internal promoter in both species. The lower number of antisense and internal promoters in C. jejuni is due to the lower coverage of 454 sequencing-based dRNA-seq. C) Examples of conservation and lack of conservation of antisense transcription in C. jejuni and H. pylori. Of the 46 genes with an antisense TSS shared between C. jejuni and H. pylori, only 2 out of 46 have the antisense TSS in the same position. This is shown for the clpB (cj0509c / hp0264) gene. The other 44 genes have the antisense promoters in different locations in the gene, as is shown for the gyrB (cj0003 / hp0501) gene. The boxes below the genes show the protein motifs encoded opposite to the antisense promoters, while the boxes below the protein sequence compare the antisense promoters. Asterisks indicate conserved residues, black and bold typeface indicates the antisense RNA sequence, the underlined residue is the transcript start site (TSS, +1), while red and bold typeface indicates the -10 sequence of the σ70 promoter. Complete information on antisense and internal promoters can be found in Additional file 19: Figure S9 and Additional file 18: Table S10 and Additional file 20: Table S11.

Predictive comparisons with other Epsilonproteobacterial genomes

To investigate whether the lack of conservation of transcriptional landscapes between C. jejuni and H. pylori were dependent on phylogenetic differences, we used comparative genomics and genome synteny analyses with the genomes of 40 other Epsilonproteobacteria (Additional file 14: Table S7). These included 18 members of the Campylobacteraceae (12 Campylobacter spp, 3 Arcobacter spp and 3 Sulfurospirillum spp), 18 members of the Helicobacteraceae (13 Helicobacter spp, 3 Sulfuromonas spp, Sulfuricurvum kujinense and W. succinogenes), and 6 species of other Epsilonproteobacteria found in deep-sea hydrothermal vents. A phylogenetic tree based on 16S rDNA sequences (Figure 4A) shows the phylogenetic relationships between the investigated species, and shows the subdivision of Campylobacter spp into thermophilic (C. jejuni to C. upsaliensis) and non-thermophilic species (other Campylobacter spp), and within the Helicobacter spp the subdivision into gastric Helicobacter spp (H. pylori to H. bizzozeroni) and enterohepatic Helicobacter spp (other Helicobacter spp with the exception of H. mustelae).
Figure 4

Comparative genomics of the Epsilonproteobacteria highlights a lack of gene order-based genome synteny throughout this lineage. A) Phylogenetic tree of 42 members of the Epsilonproteobacteria, based on comparison of 16S rDNA sequences. Members of the Campylobacteraceae (the genera Campylobacter, Sulfurospirillum (Sulfsp) and Arcobacter (Arcob)) are highlighted in orange, members of the Helicobacteraceae are highlighted in pink (the genera Helicobacter, Wolinella, Sulfurimonas (Sm) and Sulfuricurvum (Scu)), and other Epsilonproteobacteria are highlighted in green (the genera Nitratifractor (Nitfr), Sulfurovum, Nitratiruptor, Caminibacter (Camin) and Nautilia (Naut)). Sulfurovum sp. AR is not included in this tree as a full 16S rDNA sequence was not available from the genome sequence. B) Selected scatter plots of gene order-based genome synteny comparisons of C. jejuni NCTC 11168 with other members of the Campylobacteraceae, and H. pylori 26695 with other members of the Helicobacteraceae. The full comparisons are shown in Additional file 15: Figure S8, full information is given in Additional file 16: Table S8. Both the Campylobacteraceae and Helicobacteraceae show a rapid loss of genome synteny, even within the species Campylobacter and Helicobacter.

Gene order

We subsequently used the annotated features to compare genome synteny of all these species with C. jejuni NCTC 11168 and H. pylori 26695 (Additional file 15: Figure S8, Additional file 16: Table S8). From these analyses it is clear that gene order -based genome synteny is only conserved between closely related species (i.e. C. doylei and C. coli for C. jejuni, and H. acinonychis and H. cetorum for H. pylori), with genome synteny rapidly lost beyond these closely related species (Figure 4B). This progressive lack of genome synteny may explain the large differences between the experimentally determined C. jejuni and H. pylori transcriptomes, and may also explain the lack of conservation of non-coding RNAs within this phylogenetic clade (Figure 2C) [29].

Leaderless mRNA

We also assessed the difference between Epsilonproteobacteria at the gene level, using the difference between leaderless and leadered mRNAs (Figure 1C). An initial comparison of leaderless mRNAs between C. jejuni and H. pylori showed clear differences, as only 3/23 genes are leaderless in both organisms, with 12 leaderless in C. jejuni only, and 8 leaderless in H. pylori only. One example of such difference is shown in Figure 5A, for the C. jejuni cj0153c gene. This gene is located in a conserved five-gene region (cj0155c-0151c and hp0551-0555), but is leaderless in C. jejuni, effectively splitting the five genes into two separate operons, while there is only a single five-gene operon in H. pylori (Figure 5A). The intergenic regions showed clear differences with the σ70 -10 sequence (ggTAAAAT) in C. jejuni and an RBS in H. pylori (AAGGG). As the cj0155c-0151c genes are conserved throughout the Epsilonproteobacteria, we compared the intergenic regions between the cj0154c and cj0153c orthologs in all 42 genomes, and this showed that the majority of genomes (27/42) contain a predicted σ70 -10 sequence, while 13/42 contain a recognisable RBS sequence, with 2 genomes containing neither (Additional file 17: Table S9). The RBS was only present in Helicobacter spp, and interestingly also in C. upsaliensis, which is surprising in view of its close phylogenetic relationship to C. jejuni, and suggests a secondary, independent evolutionary change in C. upsaliensis or the effect of a natural transformation event (Figure 5A).
Figure 5

Prediction of leaderless mRNAs in the Epsilonproteobacteria suggests evolutionary conversion to and from leaderless mRNAs. A) The presence or absence of a leaderless mRNA can result in fusion or separation of cistrons, as shown for the leaderless cj0153c gene and the leadered H. pylori hp0553 ortholog. The dRNA-seq profiles for the cj0155c-cj0151c and hp0551-hp0555 genes show that cj0153c gene has a TSS with a σ70 -10 sequence, whereas hp0551-hp0555 is transcribed as a single multicistronic mRNA, and hp0553 has a potential ribosome binding site (RBS). The translational start site (ATG, GTG or TTG, brown), the proposed σ70 -10 sequence (red), and the proposed RBS (blue) are shown for examples from Campylobacteraceae (orange), Helicobacteraceae (pink) and other Epsilonproteobacteria (green). B) Prediction of leaderless mRNAs in the Epsilonproteobacteria, based on the sequences directly upstream of the translational start codon of 23 genes which are leaderless mRNAs in either or both C. jejuni NCTC 11168 and H. pylori 26695. The 5' UTRs were scored for the presence of gnTAnaAT or AG-rich RBS-like sequences, as shown in Figures 1C and 5A. The full set of annotated sequences and gene numbers is available in Additional file 17: Table S9. Predictions are colour-coded: yellow for of a leaderless mRNA, blue for a ribosome binding site, grey for the absence of an ortholog, and white when neither a RBS nor a suitable -10 sequence was detected. Members of the Campylobacteraceae are indicated in orange, members of the Helicobacteraceae in pink, and other Epsilonproteobacteria in green, and are ordered by the number of predicted leaderless mRNAs they have in common with C. jejuni. The genes are given at the top, with orange representing C. jejuni, pink H. pylori. A WebLogo representation of the upstream sequences and startcodons is shown for leaderless mRNAs and RBS-containing 5′ UTRs.

We expanded this search to all 23 genes which are leaderless in either C. jejuni or H. pylori in all 42 Epsilonproteobacterial genomes. All genomes were searched for orthologs of the C. jejuni and H. pylori genes, and the −30 to +3 sequences (counted until the translational startcodon) were searched manually for σ70 -10 box and ATG startcodon (gnTAnaAT-N5-9-ATG) [61] and potential ribosome binding site and all three possible startcodons (aAGGa-N3-10-aTG), and were also used for a MEME motif search (Figure 5B, Additional file 18: Table S10). From the overview presented in Figure 5B, it is clear that the predicted leaderless or leadered mRNAs do not strictly follow the 16S rDNA-based phylogenetic tree (Figure 4A). While C. jejuni, C. doylei (a subspecies of C. jejuni) and C. coli cannot be distinguished in this analysis, this is also mostly true for H. pylori, H. acinonychis and H. cetorum, mirroring the genome synteny analyses (Figure 4B, Additional file 15: Figure S8). Only a single gene (cj1247c in C. jejuni, hp0820 in H. pylori) is a predicted leaderless mRNA in all species containing this gene, which is always upstream of the uvrC DNA repair gene, again supporting an important role of leaderless mRNAs in stress responses [55]. Interestingly, the Sulfurimonas spp (members of the Helicobacteraceae) have more leaderless mRNAs in common with the Campylobacter spp than with other members of the Helicobacteraceae, while there is virtually no conservation of leaderless mRNAs within the genus Helicobacter (Figure 5B). Another surprise was that the enterohepatic Helicobacter spp had only one or two predicted leaderless mRNAs in common with other Epsilonproteobacteria, which suggests that they may contain a completely different set of leaderless mRNAs, something which was not followed up for this study.

Antisense transcription and internal promoters

Next we compared five internal and three antisense promoters conserved between C. jejuni and H. pylori (Figure 3B) in the other Epsilonproteobacterial genomes. BLAST-searches were used to identify the corresponding regions in the respective orthologs, and sequences were searched for potential σ70 -10 box both manually and using MEME (Additional file 19: Figure S9, Additional file 18: Table S10, Additional file 20: Table S11). As with the leaderless mRNAs, there was no full conservation of internal or antisense promoters, although for the promoter internal to cj0705 this can be linked to the genomic organisation: in W. succinogenes the cj0705 ortholog is fused to the downstream cj0706 ortholog (thus not requiring a promoter), whereas in Caminibacter mediatlanticus and Nautilia profundicola the downstream cj0706 ortholog is absent (Additional file 19: Figure S9, Additional file 18: Table S10). Similarly, there is a good albeit imperfect correlation between the presence of the cj0100 and cj0101 (parAB) orthologs and the presence of an internal promoter in cj0099 (Additional file 19: Figure S9, Additional file 18: Table S10). With regard to the antisense promoters, most of the cj0509c (clpB) orthologs in the Epsilonproteobacteria contain a predicted σ70 -10 box at the equivalent position (35/42, Additional file 20: Table S11), whereas the predicted antisense promoters in the cj0003 (gyrB) orthologs are confined to the phylogenetically closely related species, i.e. the C. jejuni, C. doylei and C. coli group vs the H. pylori, H. acinonychis and H. cetorum group, consistent with evolutionary relationships between these species (Figure 4A, Additional file 19: Figure S9, Additional file 20: Table S11).

Leader peptides

Finally, we also searched the Epsilonproteobacterial genomes for orthologs of the leucine, tryptophan and methionine amino acid biosynthetic genes, and whether they contained potential leader peptides upstream (Additional file 12: Figure S7, Additional file 21: Table S12). While LeuL orthologs (20–35 aa peptide with several leucines at the C-terminus) were found in most Epsilonproteobacterial genomes with leucine biosynthetic genes, TrpL and MetL orthologs were only found in C. jejuni, C. coli and C. doylei, and were absent from other Epsilonproteobacteria with tryptophan or methionine biosynthetic genes (Additional file 12: Figure S7, Additional file 21: Table S12). As with the previous examples, this suggests that there are clear differences between genomes and transcriptomes within the Epsilonproteobacteria, and also that changes in genome content and gene order have necessitated the development of differential forms of transcriptional, post-transcriptional and possibly translational regulation of gene and protein expression.


In this study, we present the primary transcriptome of Campylobacter jejuni at single nucleotide resolution, obtained by using differential RNA-sequencing analysis using 454 sequencing. Our analysis confirms that the original analyses of the C. jejuni genome [19, 20, 34] have indeed underestimated its versatility and complexity, with a wealth of non-coding and antisense RNAs, as well as intragenic promoters and leaderless mRNAs. All these features are likely to contribute to the success of C. jejuni as pathogen, allowing it to survive in the food chain and infect different hosts. Our analysis complements and supplements the previously released and reannotated genome sequence and protein interactome maps for C. jejuni[19, 20, 62], and RNA-seq analyses using Illumina sequencing [21, 29]. The large number of transcription start sites found in the relatively small C. jejuni genome supports the findings in other bacteria, where a much larger number of TSS have been detected than was expected, e.g. the >17,000 TSS identified in Sinorhizobium meliloti[63], which has a much larger genome, megaplasmids and multiple sigma factors when compared to the Epsilonproteobacteria.

The availability of the dRNA-seq datasets for two related members of the Epsilonproteobacteria has allowed for the first high resolution comparison of primary transcriptomes at the single-nucleotide level of related, but independent species (C. jejuni and H. pylori). All characterised members of the Epsilonproteobacteria have relatively small genomes (1.5 - 3 Mbp), and show high levels of variation, probably due to a relative scarcity of DNA repair mechanisms and the exchange of DNA by natural transformation and horizontal gene transfer [19, 64, 65]. Interestingly, despite the variability in both the genome and transcriptomes of these organisms, there were parts which showed high levels of conservation (like operons encoding ribosomal proteins) and others which showed no conservation at all (Additional file 15: Figure S8). With regard to transcriptome organisation, there were the large scale differences already predicted by the comparison of genome sequences (Additional file 15: Figure S8) [14, 66], but also very subtle differences with respect to coupling and uncoupling of transcriptional networks, for instance by the appearance and disappearance of promoters coupled to leaderless mRNAs (Figure 5), and generation and absence of internal and antisense promoters (Additional file 19: Figure S9).

Overall, there was very low synteny between the regulatory features of the C. jejuni and H. pylori transcriptomes with respect to the position and sequence of internal promoters, antisense RNAs and non-coding RNAs, with the exception of the ancestral 6S RNA. Orthologous sequences to the C. jejuni ncRNAs and asRNAs were only found in other C. jejuni strains and partially in closely related species (C. doylei, C. coli), and similarly conservation of H. pylori features was limited to other H. pylori strains, and partially with H. acinonychis and H. cetorum[29, 67]. All this suggests that many of these regulatory features of the transcriptomes of Epsilonproteobacteria will have developed after genera have evolutionary split from a common ancestor, and are likely to be in constant flux depending on their ecological niches and its influence on genome reorganisation, mutation frequency and horizontal gene transfer. The large differences observed between C. jejuni and H. pylori, and even the differences observed between C. jejuni strains [29] promises that future RNA-seq experiments with other Epsilonproteobacteria can be expected to show up many new and exciting features.


Bacterial strains and growth conditions

A motile variant of C. jejuni strain NCTC 11168 [20] was used throughout this study, and cultured in a MACS-MG-1000 controlled atmosphere cabinet (Don Whitley Scientific) under microaerobic conditions (85% N2, 5% O2, 10% CO2) at 37°C. For growth on plates, strains were grown on blood plates (Blood Agar Base 2 (BAB), 1% yeast extract, 5% horse blood (Oxoid) with Skirrow supplements (10 μg ml-1 vancomycin, 5 μg ml-1 trimethoprim, 2.5 IU polymyxin-B). Broth culture was carried out in Brucella broth (Becton, Dickinson & Company) [22].

RNA preparation, cDNA library construction and Roche 454 pyrosequencing

RNA was isolated from the motile C. jejuni strain NCTC 11168 [22], grown to late log phase (OD600 = 0.21). Total RNA was purified omitting size selection, to avoid the loss of small RNA molecules. The exclusion of rRNA and tRNA was also omitted, to avoid the potential loss of other RNA species. RNA was isolated using hot phenol [68], to ensure that small RNAs would not be removed by the extraction procedure. The RNA was treated with DNase I to remove residual genomic DNA, followed by optional treatment with Terminator Exonuclease (TEX, Peicentre Biotechnology) for enrichment of primary RNAs [10, 26], and treatment with Tobacco Acid Phosphatase (TAP, Cambio, UK) to generate 5′-P ends for downstream ligation of 454 adapters [10]. After ligation of an RNA oligonucleotide to the phosphorylated 5′-ends of RNA, and polyadenylation of RNA, first strand cDNA was generated using an oligo-dT containing 454-B primer (Additional file 22: Table S13). The cDNA fragments were barcoded and amplified, and used for generation of cDNA libraries for the 454 FLX system at Vertis Biotech, Germany. These libraries were subsequently analysed using a Roche FLX sequencer located at Liverpool University, UK, as previously described [10]. The enrichment procedure significantly reduced the level of 23S and 16S rRNA, but led to an increase in 5S rRNA, while tRNA levels were not altered [10].

Mapping of 454 reads and annotation of transcription start sites

Sequencing reads were grouped based on the barcode tag, the 5′ adapter was clipped, and reads of >70% A were removed. The remaining reads were aligned against the C. jejuni genome NCTC 11168 genome sequence using Segemehl version [69], and converted into number of reads per nucleotide position. Graphs representing the number of mapped reads per nucleotide were visualized using the Integrated Genome Browser software from Affymetrix [10, 70]. TSS were manually annotated based on a higher and characteristic cDNA coverage of the 5′-end of a given cDNA in the library constructed with terminator exonuclease-treated RNA. Genomes were annotated and analysed using Artemis [60]. Transcript levels of individual genes were expressed as Reads Per Kilobase per Million mapped reads (RPKM) values, calculated after mapping of reads using CLC Genomics Workbench v5 (CLC Bio). Thermodynamical and geometrical dinucleotide properties of DNA sequences were visualised using the DiProDB browser [36], whereas sequence conservation was visualised using the WebLogo program [71]. Sequence alignment was performed using ClustalX2 [72], phylogenetic analyses with Phylip v3.69 [73], and sequence motif searches were done using the MEME suite [28].

Comparative genomics

The complete genome sequences or contigs of 42 species of the Epsilonproteobacteria (Additional file 16: Table S8) were downloaded from the NCBI Genomes database ( or via the PATRIC website at the Virginia Tech University ( [74]. Incomplete genome sequences were concatenated into a single genome sequence using the Union program of the mEMBOSS suite [75] in the order of the contigs provided. Pairwise comparisons of annotated features were made using BLASTP [76], with a E-value threshold of 0.000001, and sorted to record the highest match with the annotated features of the C. jejuni NCTC 11168 or H. pylori 26695 genome. The respective gene numbers were extracted and used as X,Y coordinates in a scatterplot, essentially as described for the GeneOrder 4.0 program [77]. To identify orthologs of genes with leaderless mRNAs, internal promoters or antisense RNAs in C. jejuni NCTC 11168 and H. pylori 26695, the annotated features and genomic DNA sequence were probed with BLASTP and TBLASTN with the BioEdit program ( BLASTP alignments were used to identify corresponding regions in the genes for analysis of promoter conservation based on the -10 sequence both manually (for 5′-gnTAnaAT sequences) and with MEME motif searches.

Northern blot analysis

RNA was separated on 6% Tris-borate-EDTA polyacrylamide (PAA) gels, containing 8.3 M urea. Each lane contained 10 μg of total RNA, isolated from C. jejuni NCTC 11168 grown to early, mid and late logarithmic phase, or subjected to 30 min incubation in Brucella broth of pH 5.0 or pH 3.6. After separation, RNA was transferred onto HybondXL membranes (GE Healthcare) by electroblotting and cross-linked to the membrane. Membranes were prehybridized in Rapid-hyb buffer (GE Healthcare) at 42°C, followed by hybridization with 10 pmol [γ-32P]-ATP end-labeled oligodeoxynucleotides (Additional file 22: Table S13) for 1 h. After washing 3 times for 15 min in 5×, 1×, and 0.5× SSC–0.1% SDS solutions (42°C), signals were visualized on a phosphorimager (FLA-5000 Series, Fuji) [10].


RNA adapter (Additional file 22: Table S13) was ligated to the 5′ end of both TAP-treated and untreated RNA. 5′ RACE was performed as described previously [51, 78]. First-strand cDNA synthesis was performed using 2 pmol Random hexamer (GE Healthcare, USA) and Thermoscript RT (Invitrogen) according to manufacturer’s instructions. The RNA template was removed at the end by incubating the samples for 20 minutes at 37°C in the presence of 5 units RNase H (New England Biolabs, Ipswich, USA). PCR amplification was performed using gene-specific primers (Additional file 22: Table S13) and a 5′ adapter-specific DNA primer (Additional file 22: Table S13). The resulting PCR products were cloned into the pGEM-Teasy cloning vector (Promega, Leiden, The Netherlands) and the nucleotide sequence of the inserts was determined.

Availability of supporting data

The dRNA-seq histogram files and associated information have been deposited in the GEO database with accession number GSE49312 ( The raw sequencing data have been uploaded as 454 SFF files into the Short Read Archive with accession number SRX326863 ( An annotated Artemis entry has been created which contains the information of Additional file 3: Table S1 for use with the C. jejuni NCTC 11168 genome sequence (Accession number NC_002163) and is included with the article as Additional file 23.



Copy DNA


Transcription start site(s)


RNA polymerase


Differential RNA-sequencing


Terminator exonuclease


Tobacco acid phosphatase


Million base pairs




Rapid amplification of cDNA ends


Reads per kilobase per million mapped reads


Transfer-messenger RNA




Signal recognition particle


Thiamine pyrophosphate


Non-coding RNA


Small RNA


Product RNA

5’ UTR: 

5’ untranslated region


Ribosome binding site


Open reading frame


Blood agar base 2.



We gratefully acknowledge the support of the Biotechnology and Biological Sciences Research Council (BBSRC) via the BBSRC Institute Strategic Programme Grants IFR/08/3 and BB/J004529/1. We thank Cynthia Sharma and Jörg Vogel for support with dRNA-sequencing and for H. pylori dRNA-seq data, Charles Penn and Brendan Wren for support and helpful discussions, Margaret Hughes and the University of Liverpool Centre for Genomic Research for 454 sequencing, Vertis Biotech for cDNA library generation, Sacha Lucchini for assistance with microarray analysis, and the members of the IFR Campylobacter group for experimental support and suggestions.

Authors’ Affiliations

Gut Health and Food Safety Programme, Institute of Food Research


  1. Pallen MJ, Wren BW: Bacterial pathogenomics. Nature. 2007, 449 (7164): 835-842. 10.1038/nature06248.View ArticlePubMedGoogle Scholar
  2. Sorek R, Cossart P: Prokaryotic transcriptomics: a new view on regulation, physiology and pathogenicity. Nat Rev Genet. 2009, 11 (1): 9-16.View ArticlePubMedGoogle Scholar
  3. Raskin DM, Seshadri R, Pukatzki SU, Mekalanos JJ: Bacterial genomics and pathogen evolution. Cell. 2006, 124 (4): 703-714. 10.1016/j.cell.2006.02.002.View ArticlePubMedGoogle Scholar
  4. Boto L: Horizontal gene transfer in evolution: facts and challenges. Proc Biol Sci. 2010, 277 (1683): 819-827. 10.1098/rspb.2009.1679.PubMed CentralView ArticlePubMedGoogle Scholar
  5. Haugen SP, Ross W, Gourse RL: Advances in bacterial promoter recognition and its control by factors that do not bind DNA. Nat Rev Microbiol. 2008, 6 (7): 507-519. 10.1038/nrmicro1912.PubMed CentralView ArticlePubMedGoogle Scholar
  6. Papenfort K, Vogel J: Regulatory RNA in bacterial pathogens. Cell Host Microbe. 2010, 8 (1): 116-127. 10.1016/j.chom.2010.06.008.View ArticlePubMedGoogle Scholar
  7. Chao Y, Vogel J: The role of Hfq in bacterial pathogens. Curr Opin Microbiol. 2010, 13 (1): 24-33. 10.1016/j.mib.2010.01.001.View ArticlePubMedGoogle Scholar
  8. Deutscher MP: Degradation of RNA in bacteria: comparison of mRNA and stable RNA. Nucl Acids Res. 2006, 34 (2): 659-666. 10.1093/nar/gkj472.PubMed CentralView ArticlePubMedGoogle Scholar
  9. Wurtzel O, Sapra R, Chen F, Zhu Y, Simmons BA, Sorek R: A single-base resolution map of an archaeal transcriptome. Genome Res. 2010, 20 (1): 133-141. 10.1101/gr.100396.109.PubMed CentralView ArticlePubMedGoogle Scholar
  10. Sharma CM, Hoffmann S, Darfeuille F, Reignier J, Findeiß S, Sittka A, Chabas S, Reiche K, Hackermüller J, Reinhardt R: The primary transcriptome of the major human pathogen Helicobacter pylori. Nature. 2010, 464 (7286): 250-255. 10.1038/nature08756.View ArticlePubMedGoogle Scholar
  11. Perkins TT, Kingsley RA, Fookes MC, Gardner PP, James KD, Yu L, Assefa SA, He M, Croucher NJ, Pickard DJ: A strand-specific RNA-Seq analysis of the transcriptome of the typhoid bacillus Salmonella typhi. PLoS Genet. 2009, 5 (7): e1000569-10.1371/journal.pgen.1000569.PubMed CentralView ArticlePubMedGoogle Scholar
  12. Guell M, van Noort V, Yus E, Chen WH, Leigh-Bell J, Michalodimitrakis K, Yamada T, Arumugam M, Doerks T, Kuhner S: Transcriptome complexity in a genome-reduced bacterium. Science. 2009, 326 (5957): 1268-1271. 10.1126/science.1176951.View ArticlePubMedGoogle Scholar
  13. van Vliet AHM: Next generation sequencing of microbial transcriptomes: challenges and opportunities. FEMS Microbiol Lett. 2010, 302 (1): 1-7. 10.1111/j.1574-6968.2009.01767.x.View ArticlePubMedGoogle Scholar
  14. Eppinger M, Baar C, Raddatz G, Huson DH, Schuster SC: Comparative analysis of four Campylobacterales. Nat Rev Microbiol. 2004, 2 (11): 872-885. 10.1038/nrmicro1024.View ArticlePubMedGoogle Scholar
  15. Nakagawa S, Takaki Y, Shimamura S, Reysenbach AL, Takai K, Horikoshi K: Deep-sea vent epsilon-proteobacterial genomes provide insights into emergence of pathogens. Proc Natl Acad Sci USA. 2007, 104 (29): 12146-12150. 10.1073/pnas.0700687104.PubMed CentralView ArticlePubMedGoogle Scholar
  16. Voordeckers JW, Starovoytov V, Vetriani C: Caminibacter mediatlanticus sp. nov., a thermophilic, chemolithoautotrophic, nitrate-ammonifying bacterium isolated from a deep-sea hydrothermal vent on the Mid-Atlantic Ridge. Int J Syst Evol Microbiol. 2005, 55 (Pt 2): 773-779.View ArticlePubMedGoogle Scholar
  17. Baar C, Eppinger M, Raddatz G, Simon J, Lanz C, Klimmek O, Nandakumar R, Gross R, Rosinus A, Keller H: Complete genome sequence and analysis of Wolinella succinogenes. Proc Natl Acad Sci USA. 2003, 100 (20): 11690-11695. 10.1073/pnas.1932838100.PubMed CentralView ArticlePubMedGoogle Scholar
  18. Miller WG, Parker CT, Rubenfield M, Mendz GL, Wosten MM, Ussery DW, Stolz JF, Binnewies TT, Hallin PF, Wang G: The complete genome sequence and analysis of the epsilonproteobacterium Arcobacter butzleri. PLoS One. 2007, 2 (12): e1358-10.1371/journal.pone.0001358.PubMed CentralView ArticlePubMedGoogle Scholar
  19. Gundogdu O, Bentley SD, Holden MT, Parkhill J, Dorrell N, Wren BW: Re-annotation and re-analysis of the Campylobacter jejuni NCTC11168 genome sequence. BMC Genomics. 2007, 8: 162-10.1186/1471-2164-8-162.PubMed CentralView ArticlePubMedGoogle Scholar
  20. Parkhill J, Wren BW, Mungall K, Ketley JM, Churcher C, Basham D, Chillingworth T, Davies RM, Feltwell T, Holroyd S: The genome sequence of the food-borne pathogen Campylobacter jejuni reveals hypervariable sequences. Nature. 2000, 403 (6770): 665-668. 10.1038/35001088.View ArticlePubMedGoogle Scholar
  21. Chaudhuri RR, Yu L, Kanji A, Perkins TT, Gardner PP, Choudhary J, Maskell DJ, Grant AJ: Quantitative RNA-seq analysis of the Campylobacter jejuni transcriptome. Microbiology. 2011, 157 (Pt 10): 2922-2932.PubMed CentralView ArticlePubMedGoogle Scholar
  22. Reuter M, Mallett A, Pearson BM, van Vliet AHM: Biofilm formation in Campylobacter jejuni is increased under aerobic conditions. Appl Environ Microbiol. 2010, 76 (7): 2122-2128. 10.1128/AEM.01878-09.PubMed CentralView ArticlePubMedGoogle Scholar
  23. Holmes K, Mulholland F, Pearson BM, Pin C, McNicholl-Kennedy J, Ketley JM, Wells JM: Campylobacter jejuni gene expression in response to iron limitation and the role of Fur. Microbiology. 2005, 151 (Pt 1): 243-257.View ArticlePubMedGoogle Scholar
  24. Monk CE, Pearson BM, Mulholland F, Smith HK, Poole RK: Oxygen- and NssR-dependent globin expression and enhanced iron acquisition in the response of Campylobacter to nitrosative stress. J Biol Chem. 2008, 283 (42): 28413-28425. 10.1074/jbc.M801016200.PubMed CentralView ArticlePubMedGoogle Scholar
  25. Hinton JC, Hautefort I, Eriksson S, Thompson A, Rhen M: Benefits and pitfalls of using microarrays to monitor bacterial gene expression during infection. Curr Opin Microbiol. 2004, 7 (3): 277-282. 10.1016/j.mib.2004.04.009.View ArticlePubMedGoogle Scholar
  26. Zhang H, Ehrenkaufer GM, Pompey JM, Hackney JA, Singh U: Small RNAs with 5′-polyphosphate termini associate with a Piwi-related protein and regulate gene expression in the single-celled eukaryote Entamoeba histolytica. PLoS Pathog. 2008, 4 (11): e1000219-10.1371/journal.ppat.1000219.PubMed CentralView ArticlePubMedGoogle Scholar
  27. Jager D, Sharma CM, Thomsen J, Ehlers C, Vogel J, Schmitz RA: Deep sequencing analysis of the Methanosarcina mazei Go1 transcriptome in response to nitrogen availability. Proc Natl Acad Sci USA. 2009, 106 (51): 21878-21882. 10.1073/pnas.0909051106.PubMed CentralView ArticlePubMedGoogle Scholar
  28. Bailey TL, Boden M, Buske FA, Frith M, Grant CE, Clementi L, Ren J, Li WW, Noble WS: MEME suite: tools for motif discovery and searching. Nucleic Acids Res. 2009, 37 (Web Server issue): W202-W208.PubMed CentralView ArticlePubMedGoogle Scholar
  29. Dugar G, Herbig A, Forstner KU, Heidrich N, Reinhardt R, Nieselt K, Sharma CM: High-resolution transcriptome maps reveal strain-specific regulatory features of multiple Campylobacter jejuni isolates. PLoS Genet. 2013, 9 (5): e1003495-10.1371/journal.pgen.1003495.PubMed CentralView ArticlePubMedGoogle Scholar
  30. Ramachandran VK, Shearer N, Jacob JJ, Sharma CM, Thompson A: The architecture and ppGpp-dependent expression of the primary transcriptome of Salmonella Typhimurium during invasion gene expression. BMC Genomics. 2012, 13: 25-10.1186/1471-2164-13-25.PubMed CentralView ArticlePubMedGoogle Scholar
  31. Kroger C, Dillon SC, Cameron AD, Papenfort K, Sivasankaran SK, Hokamp K, Chao Y, Sittka A, Hebrard M, Handler K: The transcriptional landscape and small RNAs of Salmonella enterica serovar Typhimurium. Proc Natl Acad Sci USA. 2012, 109 (20): E1277-E1286. 10.1073/pnas.1201061109.PubMed CentralView ArticlePubMedGoogle Scholar
  32. Carrillo CD, Taboada E, Nash JHE, Lanthier P, Kelly J, Lau PC, Verhulp R, Mykytczuk O, Sy J, Findlay WA: Genome-wide expression analyses of Campylobacter jejuni NCTC11168 reveals coordinate regulation of motility and virulence by flhA. J Biol Chem. 2004, 279 (19): 20327-20338. 10.1074/jbc.M401134200.View ArticlePubMedGoogle Scholar
  33. Barrero-Tobon AM, Hendrixson DR: Identification and analysis of flagellar coexpressed determinants (Feds) of Campylobacter jejuni involved in colonization. Mol Microbiol. 2012, 84 (2): 352-369. 10.1111/j.1365-2958.2012.08027.x.PubMed CentralView ArticlePubMedGoogle Scholar
  34. Petersen L, Larsen TS, Ussery DW, On SL, Krogh A: RpoD promoters in Campylobacter jejuni exhibit a strong periodic signal instead of a −35 box. J Mol Biol. 2003, 326 (5): 1361-1372. 10.1016/S0022-2836(03)00034-2.View ArticlePubMedGoogle Scholar
  35. Friedel M, Nikolajewa S, Suhnel J, Wilhelm T: DiProGB: the dinucleotide properties genome browser. Bioinformatics. 2009, 25 (19): 2603-2604. 10.1093/bioinformatics/btp436.PubMed CentralView ArticlePubMedGoogle Scholar
  36. Friedel M, Nikolajewa S, Suhnel J, Wilhelm T: DiProDB: a database for dinucleotide properties. Nucleic Acids Res. 2009, 37 (Database issue): D37-D40.PubMed CentralView ArticlePubMedGoogle Scholar
  37. Calladine CR, Drew H, Luisi B, Travers A: Understanding DNA, the molecule and how it works. 2004, New York: Elsevier Academic PressGoogle Scholar
  38. Thomason MK, Storz G: Bacterial antisense RNAs: how many are there, and what are they doing?. Annu Rev Genet. 2010, 44: 167-188. 10.1146/annurev-genet-102209-163523.PubMed CentralView ArticlePubMedGoogle Scholar
  39. Toledo-Arana A, Dussurget O, Nikitas G, Sesto N, Guet-Revillet H, Balestrino D, Loh E, Gripenland J, Tiensuu T, Vaitkevicius K: The Listeria transcriptional landscape from saprophytism to virulence. Nature. 2009, 459 (7249): 950-956. 10.1038/nature08080.View ArticlePubMedGoogle Scholar
  40. Rasmussen S, Nielsen HB, Jarmer H: The transcriptionally active regions in the genome of Bacillus subtilis. Mol Microbiol. 2009, 73 (6): 1043-1057. 10.1111/j.1365-2958.2009.06830.x.PubMed CentralView ArticlePubMedGoogle Scholar
  41. Raghavan R, Sloan DB, Ochman H: Antisense transcription is pervasive but rarely conserved in enteric bacteria. MBio. 2012, 3 (4): e00156-12-10.1128/mBio.00156-12.PubMed CentralView ArticlePubMedGoogle Scholar
  42. Gan J, Tropea JE, Austin BP, Court DL, Waugh DS, Ji X: Structural insight into the mechanism of double-stranded RNA processing by ribonuclease III. Cell. 2006, 124 (2): 355-366. 10.1016/j.cell.2005.11.034.View ArticlePubMedGoogle Scholar
  43. Giangrossi M, Prosseda G, Tran CN, Brandi A, Colonna B, Falconi M: A novel antisense RNA regulates at transcriptional level the virulence gene icsA of Shigella flexneri. Nucleic Acids Res. 2010, 38 (10): 3362-3375. 10.1093/nar/gkq025.PubMed CentralView ArticlePubMedGoogle Scholar
  44. Weinberg Z, Barrick JE, Yao Z, Roth A, Kim JN, Gore J, Wang JX, Lee ER, Block KF, Sudarsan N: Identification of 22 candidate structured RNAs in bacteria using the CMfinder comparative genomics pipeline. Nucl Acids Res. 2007, 35 (14): 4809-4819. 10.1093/nar/gkm487.PubMed CentralView ArticlePubMedGoogle Scholar
  45. Valentin-Hansen P, Eriksen M, Udesen C: The bacterial Sm-like protein Hfq: a key player in RNA transactions. Mol Microbiol. 2004, 51 (6): 1525-1533. 10.1111/j.1365-2958.2003.03935.x.View ArticlePubMedGoogle Scholar
  46. Gursinsky T, Grobe D, Schierhorn A, Jager J, Andreesen JR, Sohling B: Factors and selenocysteine insertion sequence requirements for the synthesis of selenoproteins from a gram-positive anaerobe in Escherichia coli. Appl Environ Microbiol. 2008, 74 (5): 1385-1393. 10.1128/AEM.02238-07.PubMed CentralView ArticlePubMedGoogle Scholar
  47. Wassarman KM: 6S RNA: a regulator of transcription. Mol Microbiol. 2007, 65 (6): 1425-1431. 10.1111/j.1365-2958.2007.05894.x.View ArticlePubMedGoogle Scholar
  48. Barrick JE, Sudarsan N, Weinberg Z, Ruzzo WL, Breaker RR: 6S RNA is a widespread regulator of eubacterial RNA polymerase that resembles an open promoter. RNA. 2005, 11 (5): 774-784. 10.1261/rna.7286705.PubMed CentralView ArticlePubMedGoogle Scholar
  49. Wassarman KM, Storz G: 6S RNA regulates E. coli RNA polymerase activity. Cell. 2000, 101 (6): 613-623. 10.1016/S0092-8674(00)80873-9.View ArticlePubMedGoogle Scholar
  50. Wassarman KM, Saecker RM: Synthesis-mediated release of a small RNA inhibitor of RNA polymerase. Science. 2006, 314 (5805): 1601-1603. 10.1126/science.1134830.View ArticlePubMedGoogle Scholar
  51. Guccione E, Hitchcock A, Hall SJ, Mulholland F, Shearer N, van Vliet AHM, Kelly DJ: Reduction of fumarate, mesaconate and crotonate by Mfr, a novel oxygen-regulated periplasmic reductase in Campylobacter jejuni. Environ Microbiol. 2010, 12 (3): 576-591. 10.1111/j.1462-2920.2009.02096.x.View ArticlePubMedGoogle Scholar
  52. Livny J, Waldor MK: Mining regulatory 5′ UTRs from cDNA deep sequencing datasets. Nucl Acids Res. 2010, 38 (5): 1504-1514. 10.1093/nar/gkp1121.PubMed CentralView ArticlePubMedGoogle Scholar
  53. Merino E, Jensen RA, Yanofsky C: Evolution of bacterial trp operons and their regulation. Curr Opin Microbiol. 2008, 11: 78-86. 10.1016/j.mib.2008.02.005.PubMed CentralView ArticlePubMedGoogle Scholar
  54. Gaskin DJ, Reuter M, Shearer N, Mulholland F, Pearson BM, van Vliet AHM: Genomics of thermophilic Campylobacter species. Genome Dyn. 2009, 6: 91-109.View ArticlePubMedGoogle Scholar
  55. Moll I, Grill S, Gualerzi CO, Blasi U: Leaderless mRNAs in bacteria: surprises in ribosomal recruitment and translational control. Mol Microbiol. 2002, 43 (1): 239-246. 10.1046/j.1365-2958.2002.02739.x.View ArticlePubMedGoogle Scholar
  56. Kaberdina AC, Szaflarski W, Nierhaus KH, Moll I: An unexpected type of ribosomes induced by kasugamycin: a look into ancestral times of protein synthesis?. Mol Cell. 2009, 33 (2): 227-236. 10.1016/j.molcel.2008.12.014.PubMed CentralView ArticlePubMedGoogle Scholar
  57. Gaynor EC, Wells DH, MacKichan JK, Falkow S: The Campylobacter jejuni stringent response controls specific stress survival and virulence-associated phenotypes. Mol Microbiol. 2005, 56 (1): 8-27. 10.1111/j.1365-2958.2005.04525.x.View ArticlePubMedGoogle Scholar
  58. Wells DH, Gaynor EC: Helicobacter pylori initiates the stringent response upon nutrient and pH downshift. J Bacteriol. 2006, 188 (10): 3726-3729. 10.1128/JB.188.10.3726-3729.2006.PubMed CentralView ArticlePubMedGoogle Scholar
  59. Tomb JF, White O, Kerlavage AR, Clayton RA, Sutton GG, Fleischmann RD, Ketchum KA, Klenk HP, Gill S, Dougherty BA: The complete genome sequence of the gastric pathogen Helicobacter pylori. Nature. 1997, 388 (6642): 539-547. 10.1038/41483.View ArticlePubMedGoogle Scholar
  60. Carver T, Berriman M, Tivey A, Patel C, Bohme U, Barrell BG, Parkhill J, Rajandream MA: Artemis and ACT: viewing, annotating and comparing sequences stored in a relational database. Bioinformatics. 2008, 24 (23): 2672-2676. 10.1093/bioinformatics/btn529.PubMed CentralView ArticlePubMedGoogle Scholar
  61. Brock JE, Pourshahian S, Giliberti J, Limbach PA, Janssen GR: Ribosomes bind leaderless mRNA in Escherichia coli through recognition of their 5′-terminal AUG. RNA. 2008, 14 (10): 2159-2169. 10.1261/rna.1089208.PubMed CentralView ArticlePubMedGoogle Scholar
  62. Parrish JR, Yu J, Liu G, Hines JA, Chan JE, Mangiola BA, Zhang H, Pacifico S, Fotouhi F, DiRita VJ: A proteome-wide protein interaction map for Campylobacter jejuni. Genome Biol. 2007, 8 (7): R130-10.1186/gb-2007-8-7-r130.PubMed CentralView ArticlePubMedGoogle Scholar
  63. Schluter JP, Reinkensmeier J, Barnett MJ, Lang C, Krol E, Giegerich R, Long SR, Becker A: Global mapping of transcription start sites and promoter motifs in the symbiotic alpha-proteobacterium Sinorhizobium meliloti 1021. BMC Genomics. 2013, 14: 156-10.1186/1471-2164-14-156.PubMed CentralView ArticlePubMedGoogle Scholar
  64. Gilbreath JJ, Cody WL, Merrell DS, Hendrixson DR: Change is good: variations in common biological mechanisms in the epsilonproteobacterial genera Campylobacter and Helicobacter. Microbiol Mol Biol Rev. 2011, 75 (1): 84-132. 10.1128/MMBR.00035-10.PubMed CentralView ArticlePubMedGoogle Scholar
  65. Campbell BJ, Engel AS, Porter ML, Takai K: The versatile epsilon-proteobacteria: key players in sulphidic habitats. Nat Rev Microbiol. 2006, 4 (6): 458-468. 10.1038/nrmicro1414.View ArticlePubMedGoogle Scholar
  66. Fouts DE, Mongodin EF, Mandrell RE, Miller WG, Rasko DA, Ravel J, Brinkac LM, DeBoy RT, Parker CT, Daugherty SC: Major structural differences and novel potential virulence mechanisms from the genomes of multiple Campylobacter species. PLoS Biol. 2005, 3 (1): e15-10.1371/journal.pbio.0030015.PubMed CentralView ArticlePubMedGoogle Scholar
  67. Eppinger M, Baar C, Linz B, Raddatz G, Lanz C, Keller H, Morelli G, Gressmann H, Achtman M, Schuster SC: Who ate whom? Adaptive Helicobacter genomic changes that accompanied a host jump from early humans to large felines. PLoS Genet. 2006, 2 (7): e120-10.1371/journal.pgen.0020120.PubMed CentralView ArticlePubMedGoogle Scholar
  68. Mattatall NR, Sanderson KE: Salmonella typhimurium LT2 possesses three distinct 23S rRNA intervening sequences. J Bacteriol. 1996, 178 (8): 2272-2278.PubMed CentralPubMedGoogle Scholar
  69. Hoffmann S, Otto C, Kurtz S, Sharma CM, Khaitovich P, Vogel J, Stadler PF, Hackermuller J: Fast mapping of short sequences with mismatches, insertions and deletions using index structures. PLoS Comput Biol. 2009, 5 (9): e1000502-10.1371/journal.pcbi.1000502.PubMed CentralView ArticlePubMedGoogle Scholar
  70. Nicol JW, Helt GA, Blanchard SG, Raja A, Loraine AE: The integrated genome browser: free software for distribution and exploration of genome-scale datasets. Bioinformatics. 2009, 25 (20): 2730-2731. 10.1093/bioinformatics/btp472.PubMed CentralView ArticlePubMedGoogle Scholar
  71. Crooks GE, Hon G, Chandonia JM, Brenner SE: WebLogo: a sequence logo generator. Genome Res. 2004, 14 (6): 1188-1190. 10.1101/gr.849004.PubMed CentralView ArticlePubMedGoogle Scholar
  72. Larkin MA, Blackshields G, Brown NP, Chenna R, McGettigan PA, McWilliam H, Valentin F, Wallace IM, Wilm A, Lopez R: Clustal W and Clustal X version 2.0. Bioinformatics. 2007, 23 (21): 2947-2948. 10.1093/bioinformatics/btm404.View ArticlePubMedGoogle Scholar
  73. Felsenstein J: PHYLIP - phylogeny inference package (Version 3.2). Cladistics. 1989, 5: 164-166.Google Scholar
  74. Gillespie JJ, Wattam AR, Cammer SA, Gabbard JL, Shukla MP, Dalay O, Driscoll T, Hix D, Mane SP, Mao C: PATRIC: the comprehensive bacterial bioinformatics resource with a focus on human pathogenic species. Infect Immun. 2011, 79 (11): 4286-4298. 10.1128/IAI.00207-11.PubMed CentralView ArticlePubMedGoogle Scholar
  75. Rice P, Longden I, Bleasby A: EMBOSS: the European molecular biology open software suite. Trends Genet. 2000, 16 (6): 276-277. 10.1016/S0168-9525(00)02024-2.View ArticlePubMedGoogle Scholar
  76. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment search tool. J Mol Biol. 1990, 215 (3): 403-410.View ArticlePubMedGoogle Scholar
  77. Mahadevan P, Seto D: Rapid pair-wise synteny analysis of large bacterial genomes using web-based GeneOrder4.0. BMC Res Notes. 2010, 3: 41-10.1186/1756-0500-3-41.PubMed CentralView ArticlePubMedGoogle Scholar
  78. Shaw FL, Mulholland F, Le Gall G, Porcelli I, Hart DJ, Pearson BM, van Vliet AHM: Selenium-dependent biogenesis of formate dehydrogenase in Campylobacter jejuni is controlled by the fdhTU accessory genes. J Bacteriol. 2012, 194 (15): 3814-3823. 10.1128/JB.06586-11.PubMed CentralView ArticlePubMedGoogle Scholar


© Porcelli et al.; licensee BioMed Central Ltd. 2013

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.