Quorum sensing primes the oxidative stress response in the insect endosymbiont, sodalis glossinidius

Sodalis glossinidius, a maternally transmitted bacterial endosymbiont of tsetse flies (Glossina spp.), uses an acylated homoserine lactone (AHL)-based quorum sensing system to modulate gene expression in accordance with bacterial cell density. The S. glossinidius quorum sensing system relies on the function of two regulatory proteins; SogI (aLuxI homolog) synthesizes a signaling molecule, characterized as N-(3-oxohexanoyl) homoserine lactone (OHHL), and SogR1 (a LuxR homolog) interacts with OHHL to modulate transcription of specific target genes.

Quorum Sensing Primes the Oxidative Stress Response in the Insect Endosymbiont, Soda/is glossinidius Mauricio H. Pontes*, Markus Babst, Robert Lochhead, Kelly Oakeson, Kari Smith, Colin Dale Department of Biology, University of Utah, Salt Lake City, Utah, United States of America OPEN 3 ACCESS Freely available online ' PL P S O I 1 0 Abstract Background: Sodalis glossinidius, a maternally transmitted bacterial endosymbiont of tsetse flies (Glossina spp.), uses an acylated homoserine lactone (AHL)-based quorum sensing system to modulate gene expression in accordance with bacterial cell density. The S. glossinidius quorum sensing system relies on the function of two regulatory proteins; Sogl (a Luxl homolog) synthesizes a signaling molecule, characterized as N-(3-oxohexanoyl) homoserine lactone (OHHL), and SogR1 (a LuxR homolog) interacts with OHHL to modulate transcription of specific target genes. Methodology/Principal Findings: We used a tiling microarray to analyze the S. glossinidius transcriptome in the presence and absence of exogenous OHHL. The major finding is that OHHL increases transcription of a large number of genes that are known to be involved in the oxidative stress response. We also show that the obligate symbiont of the rice weevil, Sitophilus oryzae (SOPE), maintains copies of the quorum sensing regulatory genes that are found in S. glossinidius. Molecular evolutionary analyses indicate that these sequences are evolving under stabilizing selection, consistent with the maintenance of their functions in the SOPE symbiosis. Finally, the expression studies in S. glossinidius also reveal that quorum sensing regulates the expression of a cryptic, degenerate gene (car A) that arose from an ancient deletion in the last common ancestor of S. glossinidius and SOPE. Conclusions/Significance: This oxidative stress response is likely mandated under conditions of dense intracellular symbiont infection, when intense metabolic activity is expected to generate a heavy oxidative burden. Such conditions are known to arise in the bacteriocytes of grain weevils, which harbor dense intracellular infections of symbiotic bacteria that are closely related to S. glossinidius. The presence of a degenerate carA sequence in S. glossinidius and SOPE indicates the potential for neofunctionalization to occur during the process of genome degeneration. Citation: Pontes MH, Babst M, Lochhead R, Oakeson K, Smith K, et al. (2008) Quorum Sensing Primes the Oxidative Stress Response in the Insect Endosymbiont, Sodalis glossinidius. PLoS ONE 3(10): e3541. doi:10.1371/journal.pone.0003541 Editor: Raphael H. Valdivia, Duke University Medical Center, United States of America Received August 20, 2008; Accepted October 6, 2008; Published October 28, 2008 Copyright: © 2008 Pontes et al. 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 author and source are credited. Funding: Research in our laboratory is supported by a National Science Foundation award EF-0523818 to CD. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [e-mail] Introduction All living organisms are dependent on their ability to sense the physical properties of their environment and respond accordingly through behavioral changes. In the context of symbiosis, the establishment and maintenance of a symbiotic relationship is dependent on the ability of both parties to perceive one another and coordinate their activities. In several recent studies it has been shown that secreted chemicals provide a means for communica­tion between symbiotic partners. For example, chemical commu­nication between legumes and rhizobia is known to mediate many important symbiotic interactions. These include initiation of nodulation [1], evasion of legume defenses [2-5], elongation of infection threads [6,7] and development of nitrogen fixing bacteroids [4,5,8]. The marine symbiosis between Vibrio fischcri and the squid Eupiymna scolopcs is also dependent on chemical communication. In this system, chemical signaling mediates the induction of mucus secretion by the squid [9], bacterial attachment and aggregation to the host mucus [10,11], bacterial migration to the squid light organ [12], bacterial light emission [13], and the induction of physiological [14] and developmental [15] changes in the infected squid. Models systems, such as the plant-rhizobia and squid- Vibrio symbioses have provided a high level of insight into the molecular processes underlying interactions between symbiotic partners. However, this knowledge is lacking for the majority of symbiotic systems, where sophisticated experimental approaches are not applicable. Insects from many different taxonomic groups are known to maintain beneficial associations with maternally transmitted bacterial symbionts [16], Many of these associations are ancient in origin and obligate in nature; in these cases bacterial symbionts often provide their insect hosts with essential nutrients that are lacking in the host's specialized diet [17], More recently derived associations are typically facultative in nature; in these cases bacterial symbionts often provide ancillary benefits such as protection from natural enemies or enhanced tolerance towards conditional environmental stresses [18]. Regardless of the nature of the association, success in the symbiosis depends upon a complex interplay between bacterial symbionts and their insect hosts throughout the course of insect development and reproduc­tion. For example, many symbiotic bacteria preferentially colonize specialized insect tissues or cells, where they reach extremely high infection densities. This feature is most pronounced in obligate symbionts, which are often found exclusively in specialized organs PLoS ONE | www.plosone.org October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing called bacteriomes [19]. However, facultative symbionts also tend to colonize specific host cell and tissue types at high infection densities (20.21.22]. In both cases, it seems likely that specific colonization patterns provide physiological benefits for both partners in the association. In this study we describe the function of a quorum sensing system in the maternally transmitted symbiont of tsetse flies. Sodalis gfossinidius. Quorum sensing is a chemical signaling process that allows bacteria to monitor their local population density and induce changes in behavior through the modulation of gene expression (23], Our experiments demonstrate that S. gfossinidius utilizes an acylated homoserine lactone (AHL]-based quorum sensing system to prime the expression of genes that ameliorate the deleterious effects of oxidative stress that are encountered at high densities of symbiont infection in host tissues. Results Characterization of the Quorum Sensing Machinery of S. glossinidius The whole genome sequence of S. gfossinidius was found to contain a single candidate AHL synthase gene (GenBank locus tag SG0284; designated sogl in this report] and two candidate AHL- binding regulatory genes (SG0283 and SGI 740. designated sogRl and sogR2 respectively]. Whereas sogl and sogRl are located adjacent to one another. sogR2 is found at a distant location in the S. glossinidius chromosome. To assess the functionality of the S. gfossinidius quorum sensing system we performed thin layer chromatography (TLC] overlay bioassays with culture supernatant extracts from S. gfossinidius and a recombinant E. coli strain expressing the S. gfossinidius Sogl protein. The bioassay, performed using both Agrobacterium tumefaciens (24] and Chromobacterium violaceum (data not shown] (25] reporter strains, indicated the presence of a single AHL molecule in the supernatant extracts from both S. gfossinidius and the recombinant E. coli strain (Figure IA], To avoid complications associated with the purification of small molecules from the complex S. gfossinidius culture medium, we elected to perform mass spectrometric characterization of the S. gfossinidius AHL on the culture supernatant extracted from the recombinant K. coli strain expressing the S. gfossinidius Sogl protein. An ethyl acetate extract of culture supernatant was fractionated on a Cih column by reverse-phase high performance liquid chromatography. The fractions containing the putative AHL molecule were then identified using the A. tumefaciens reporter strain (24] and analyzed by electrospray ionization and collision induced dissociation mass spectrometry (CID-MS], The fragmen­tation pattern of the candidate AHL ion obtained following CID- MS was consistent with the structure of N-(3-oxohexanoyl] homoserine lactone (OHHL; Figure IB] (26]. This result is supported by the fact that the AHL molecules produced by S. gfossinidius and the recombinant E. coli strain co-migrated with synthetic and natural forms of OHHL produced by Pseudomonas aeruginosa strain FAOl in our TLC assays (Figure 1A] (27]. Interactions Between OHHL and the SogRl and SogR2 Transcriptional Regulators In many bacterial species that utilize AHL quorum sensing systems, the transcription of the lu xl homolog is autoregulated by quorum sensing (1 3.28.29.30], In addition, many quorum-sensing regulated genes contain an 18 to 20-bp long imperfect palindromic sequence upstream of their initiator codon. This inverted repeat sequence, known as lux box. serves as the binding site for LuxR transcriptional response regulators (31], The presence of a canonical lux box (ACCTGAACTTTAGTACAGGT] 86 bp upstream from the S. glossinidius sogl start codon suggested that the transcription of this gene is autoregulated by quorum sensing. To test this hypothesis and characterize interactions between OHHL and the SogRl and SogR2 transcriptional regulators, three recombinant plasmids were constructed (pMF2-4; Supple­mentary Table SI and Figure 2A], Each of these plasmids harbors an active sogl-lac? reporter fusion, and in addition pMP3 and pMP4 have sogRl and sogR2 expressed from their native promoters. When these plasmids were transformed into isogenic strains of K. coli. only cells harboring pMP3 (expressing SogRl] demonstrated increased P-galactosidase activity when synthetic OHHL was added to the culture medium (Figure 2A], This shows Figure 1. Characterization of S. glossinidius AHL. A. AHL samples were separated by TLC and overlaid with a live culture of an A tumefaciens AHL reporter strain. Dark spots indicate the presence of AHL molecules. Lane 1: synthetic N-(3-oxohexanoyl)-homoserine lactone (OHHL); lane 2: S. glossinidius culture supernatant extract; lane 3: culture supernatant extract from recombinant E. coli expressing S. glossinidius Sogl; lane 4: culture supernatant extract from an isogenic E. coli strain lacking Sogl; lane 5: culture supernatant extract from P. aeruginosa PA01 known to produce OHHL, N-(3-oxooctanoyl)-AHL (OOHL), N-3-(oxodecanoyl)-AHL (ODHL) and N-3-(oxododecanoyl)-AHL (OdDHL). B. Mass spectrum of the S. glossinidius candidate AHL ion following collision induced dissociation. The fragmentation pattern of this molecule is consistent with the structure of OHHL. doi:10.1371/journal.pone.0003541.g001 PLoS ONE | www.plosone.org October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing Figure 2. Interactions of S. glossinidius SogR-OHHL Complexes with sogland carA Promoters. A. (S-galactosidase activity of isogenic strains of E. coli harboring pMP2 (light blue), pMP3 (orange), pMP4 (light green) following exposure to different concentrations of exogenous OHHL. B. (?- galactosidase activity of strains of E. coli harboring pMP5 (dark blue), pMP6 (red) or pMP7 (dark green) following exposure to different concentrations of OHHL. Bars indicate standard deviation. doi:10.1371/journal.pone.0003541.g002 that the interaction between OHHL and the .S', glossinidius SogRl protein facilitates the binding of the SogR 1 -OHHL complex to the sogl promoter, resulting in an increase in the rate of transcription of sogl (Figure 2A). It should also be noted that cells harboring pMP4 (expressing SogR2) demonstrated no significant increase in P-galactosidase activity when synthetic OHHL was added to the culture medium. This is likely due to a lack of interaction between SogR2 and OHHL and/or the sogl promoter. Identification of Quorum Sensing Regulated Genes in S. glossinidius The .S', glossinidius tiling microavray was used to identify genes (including putative pseudogenes! that were differentially expressed in the presence and absence of exogenous OHHL. Data was obtained from four replicate microarray experiments and analyzed using a Bayesian analysis of posterior probability. Quantitative PCR (qPCR) assays were then performed to validate the results obtained from the microavray experiments. The data obtained from the qPCR assays closely matched the data obtained from the microarray experiment. The complete dataset obtained from the microarray and qPCR experiments is presented in Supplementary Table S2.' The microarray and qPCR experiments demonstrated that expression of the .S', glossinidius sogl gene (SG0284) was increased (4.6-fold and 7.5-fold, respectively! in response to the addition of exogenous OHHL. These results are in agreement with those results obtained from the promotcr-probc experiment (Figure 2B), validating the experimental approaches used in the microarray and qPCR experiments. Only two other candidate genes (SG0585 and SG0586! displayed a >5-fold increase or decrease in expression in response to OHHL. According to the microarray data, a substantial number of candidate genes show a statistically significant increase or decrease in expression in the range of 1.2 to 5-fold. In order to understand the biological significance of these changes in gene expression, the data was analyzed according to the clusters of orthologous groups (COG) classification (Figure 3). This revealed a bias in the representation of genes showing > 1.2-fold increase in expression in response to the addition of OHHL within COG categories C (energy production and conversion!, L (DNA replication, recom­bination and repair!, O (posttranslational modification, protein turnover, chaperones! and P (inorganic ion transport and metabolism!. Furthermore, a bias was also detected among those genes showing > 1.2-fold decrease in expression in response to the October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing Figure 3. COG-Based Analysis of Microarray Expression Data. Distribution of S. glossinidius genes among different COG categories. Blue bars correspond to all protein coding genes in the S. glossinidius genome, whereas orange and green bars correspond to genes whose expression increased or decreased (respectively) by at least 1.2-fold in response to addition of OHHL. The COG classifications are described in the inset. doi:10.1371/journal.pone.0003541 ,g003 addition of OHHL. The downregulated genes were overrepre­sented in COG categoriesj (translation). K (transcription). P and I (lipid transport and metabolism). Many of the genes that were upregulated in response to OHHL are predicted to play a role in the oxidative stress response. These include genes predicted to encode proteins involved in the breakdown of reactive oxygen species (ROS; e.g.. SG0017, SG0642, SGI609 and SG2101), repair of oxidatively damaged cellular components (SGI 106 and SGI 348). transport of iron and manganese (SG1 5 16. SG1517. SG1518 and SG1519) and protein folding (e.g.. SG0409, SG0584, SG0692 and SG2325; Supple­mentary Table S2). The addition of OHHL also increased the expression of bacterioferritin (SG2280) and genes involved in iron sidcrophorc biosynthesis (SGP1_0041-46) [32]. Based on the fact that the sidcrophorc biosynthetic genes of Erwinia chrysanthemi and S. glossinidius share high levels of sequence identity, it seems likely that S. glossinidius synthesizes an achromobactin-like, citrate-based sidcrophorc [32]. With this in mind, it should be noted that the addition of OHHL stimulated an increase in transcription of a number of citric acid cycle enzymes, including citrate synthase (SG0871), aconitase (SG0477), isocitrate dehydrogenase (SG0700), a-ketoglutarate dehydrogenase (SG0876, SG0877), succinyl-CoA synthase (SG0878, SG0879) and succinate dehydrogenase (SG0872-75; Supplementary Table S2). It is possible that the resulting increase in TGA cycle activity might be required to provide sufficient citric acid for sidcrophorc biosynthesis at high cell density. Among those genes showing decreased expression in the presence of OHHL. the largest representational bias was found in COG category J (Figure 3). This includes several genes encoding subunits of the 30S (SG0380, SG0412 and SG2269) and 50S ribosomal proteins (SG0133, SGI420, SGI421, SGI572, SG2207, SG2252, SG2270, SG2271 and SG2273). In addition, genes encoding a 16S rRNA pseudouridylate synthase A (SG1570), a tRNA/rRNA mcthyltransfcrasc (SGI908) and a putative ribosome modulation factor (SGI025) also displayed reduced expression in the presence of OHHL. This leads to an interesting hypothesis - perhaps quorum sensing restricts the growth rate of S. glossinidius at high infection density by reducing the activities of enzymes involved in translation. However, we were unable to detect any difference in the growth rate of laboratory cultures of S. glossinidius in the presence and absence of exogenous OHHL (data not shown). The S. glossinidius gene showing the largest reduction in expression in response to OHHL is annotated as a dcthiobiotin synthase (bioD; SGI466). Sodatis glossinidius is unusual because it maintains two phylogcnctically distinct ORFs (SG0906 and SGI466) encoding putative bioD homologs. SG0906, whose expression is not affected by OHHL, is located within an opcron alongside other genes known to be involved in biotin biosynthesis PLoS ONE | www.plosone.org 4 October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing (biojWFC: SG0902-SG0905). Thus, SG0906 appears to be a component of the canonical biotin biosynthesis gene cluster found in a wide range of bacteria [33], SGI 466 is also unusual because the C-terminal end of the predicted translation product maintains a sugar transporter domain that is not found in any of the other dethiobiotin synthetases in the GenBank database. Furthermore, SG1466 is not located within a cluster of genes involved in biotin biosynthesis. Thus, SG 1466 may have evolved to provide a novel function in S. glossinidius. Biochemical Detection of Siderophores Iron siderophore assays were performed to quantitate side- rophore production in S. glossinidius cultures maintained at low cell density in the presence and absence of exogenous OIIIIL. The addition of OIIIIL provides an "artificial" quorum that stimulates a 5-fold increase in siderophore production in culture media over the course of 24 h (Figure 4). After 48 h, this difference is reduced to 3-fold as the uninduced cultures reach a cell density sufficient to provide a "natural" quorum. These results demonstrate that S. glossinidius increases production of iron siderophores in a cell density-dependent manner in response to the quorum sensing signaling molecule, OIIIIL. This is in agreement with the microarray and qPCR data showing that S. glossinidius increases transcription of genes involved in siderophore biosynthesis in response to the addition of OIIIIL. Degeneration of Carbapenem Biosynthesis Genes According to the microarray data, two candidate genes (SG585 and SG586) demonstrated a >5-fold increase in transcription in response to the addition of OIIIIL. The GenBank annotation of the S. glossinidius genome sequence indicates that SG585 and SG586 are genic components of a biosynthetic pathway that leads to the production of a P-lactam antibiotic called carbapenem. Carbapenem is produced by a number of Gram-negative bacteria including Envinia spp., which is one of the closest free-living relatives of S. glossinidius [34-37], Furthermore, carbapenem production is often controlled by quorum sensing, and in Envinia carolovora the signaling molecule is known to be OIIIIL [38], The genes involved in the biosynthesis of carbepenem (canWCDE) are Figure 4. Influence of OHHL on Iron Siderophore Production in S. glossinidius. Siderophore units were estimated from samples of S. glossinidius culture supernatant in the presence (orange) and absence (blue) of exogenous OHHL. doi:10.1371/journal.pone.0003541.g004 normally localized in a cluster alongside genes that are known to confer a carbapenem resistance phenotype (carFCr. Figure 5) [36], In S. glossinidius. most of the genes required for carbapenem biosynthesis and resistance have been lost as a result of a deletion between the C-terminal domain of Car A (SG0586! and the N- terminal domain of CarG (SG0585). Therefore, only the N- terminal domain of Car A and the C-terminal domain of CarG remain. However, several lines of evidence indicate that the genic remnant of carA retains functionality. First, it should be noted that the weevil symbiont, SOPE, also maintains truncated copies of carA and card that share an almost identical intergenie deletion (Figure 5). Although it is possible that such deletions occurred independently in the lineages leading to S. glossinidius and SOPE, it seems more likely (based on parsimony! that a single deletion took place in the common ancestor of these symbionts. Furthermore, if the carA genes of S. glossinidius and SOPE were inactive, we would anticipate a high ratio of nonsynonymous (nonsilent) to synony­mous (silent! substitutions (dN/dS). However, our molecular evolutionary analysis (detailed below! shows that carA has a low f/N/rfS ratio, compatible with the preservation of gene function under stabilizing selection. Second, the truncation in the S. glossinidius and SOPE carA sequences occurs at a point in the conceptual CarA protein that links an N-terminal nucleophilic hydrolase domain and a C-terminal synthase domain (Figure 5). Thus, the carA sequences in S. glossinidius and SOPE retain only the nucleophilic hydrolase domain of the ancestral carA. Third, the conceptual nucleophilic hydrolase domain in the S. glossinidius and SOPE carA sequences retain many of the key residues known to be important for amino acid amidohydrolysis (data not shown!. Thus, we postulate that the truncated carA sequences in S. glossinidius and SOPE have acquired a novel functionality, unrelated to antibiotic biosynthesis, as a result of deletion. Interestingly, E. camtovora is known to maintain at least two homologs of LuxR, one of which (CarR! is dedicated to the regulation of the carbapenem gene cluster [39,40], Since S. glossinidius is also known to maintain two LuxR homologs, we decided to investigate interactions between SogRI, SogR2 and the carA promoter. Plasmids pMP5-7 were constructed by replacing the sogl-lac£ fusions from pMP2-4, respectively, with a carA-lac£ fusion. Escherichia coli strains harboring these plasmids were then tested for P-galactosidase activity in the presence of increasing concentrations of OIIIIL (Figure 2B). Interestingly, only cells harboring pMP6 (expressing SogRI! demonstrated increased P~ galactosidase activity in response to OIIIIL. This shows that the canonical SogRI protein is involved in the cell density dependent regulation of carA. and that SogR2 is therefore not a functional homolog of CarR. Molecular Evolutionary Analyses In order to understand the molecular evolutionary genetics of the quorum sensing system in S. glossinidius. we obtained homologs of sogl. sogRl. sogR2 and the truncated carA ORF from an unfinished (6-8 x coverage! genome sequence of the Sitophilus oryzae primary endosymbiont (SOPE), which is known to be closely related to S. glossinidius [41,42]. The putative coding sequences of sogl. sogRl and sogR2 were found to be intact in SOPE, indicating their potential to encode proteins that serve as regulators of quorum sensing in the weevil symbiosis. The tuxland tuxR homologs from S. glossinidius. SOPE and other Gram negative bacteria were used to construct maximum likelihood (ML) phylogenetic trees. Both the tuxl and luxR trees were supported by more than 50% of ML bootstrap resamples at every node. Four major clades were resolved in the tuxl tree (Figure 6A), each with >95% bootstrap support. The tuxl PLoS ONE | www.plosone.org October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing Figure 5. Degeneration of a Carbapenem Biosynthesis Gene Cluster in S. glossinidius and SOPE. Carbapenem production (carABCDE) and resistance (carFG) genes are clustered in the genomes of £ carotovora, P. luminescens and Serratia sp. Sodalis glossinidius and SOPE maintain truncated copies of carA and carG that likely arose through the deletion of intervening DNA. do i: 10.13 71/jo u rna I .po ne.0003541 .g005 homologs from S. glossinidius and SOPE were placed in a elade with 99% bootstrap support alongside luxl homologs from Envinia spp. and Yersinia spp.. which are the closest free-living relatives of S. glossinidius [43], Furthermore, the S. glossinidius and SOPE sequences were placed together in a sub-clade with 100% bootstrap support, indicating their close ancestry. The same pattern of relationships was resolved in the luxR tree (Figure 6B). Both the luxR homologs from S. glossinidius and SOPE were placed in a elade along with luxR homologs from Envinia spp. and Yersinia spp. with 97% bootstrap support. Interestingly, the S. glossindius Figure 6. Common Ancestry of S. glossinidius and SOPE Quorum Sensing Regulatory Genes. Maximum likelihood phylogenetic analyses provide support for the common ancestry of the sogl (A), sogRl and sogR2 (B) sequences in S. glossinidius and SOPE. Bootstrap values >50% are shown adjacent to each node. GenBank accession numbers for the sequences used in these analyses are provided in Supplementary Information S1. doi:10.1371/journal.pone.0003541 .g006 h. PLoS ONE | www.plosone.org October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing and SOPE sogRl and sogR2 sequences were placed in separate sub- clades with 100% and 98% bootstrap support, respectively. Thus, it seems likely that these two genes in .S', gbssimdm and SOPE have independent phylogenetic origins and did not arise from a recent gene duplication event. Estimating dN/d& (the number of nonsynonymous base substi­tutions per nonsynonymous site versus the number of synonymous base substitutions per synonymous site) provides a means to assess the strength and direction of natural selection acting upon genic (coding) sequences. In bacteria, pairwise estimates of dN/d& typically fall within the range of 0.04-0.2 for functional genes evolving under stabilizing selection [44], Conversely, genic sequences that have been rendered inactive (pseudogenes) are expected to have fl'N/flS ratios that approach parity. Since we made no direct assessment of the functionality' of the truncated carA homologs in S. glossinidius and SOPE, or the lux gene homologs in SOPE, we reasoned that it would be useful to perform tests of selection on these putative coding sequences. Thus, we obtained pairwise estimates of dN/d& for the lux], luxR and carA sequences from S. glossinidius, SOPE and close free-living relatives ('Fable I). Notably, the estimates of dN/d& in pairwise sequence comparisons between S. glossinidius and SOPE for 500/(0.0 71 ), sogRl (0.061), sogR2 (0.1 16) and the truncated carA homolog (0.095) are all within the range expected for genes evolving under stabilizing selection. This suggests that sogl, sogRl, sogR2 and the truncated carA all remain functional in S. glossinidius and SOPE. Discussion Sodalis glossinidius, an endosym biotic bacterium found in tsetse (lies (Cllossim spp.), uses an AIIL-based quorum sensing system to modulate gene expression according to cell density. Since S. glossinidius lias no known lifestyle component outside of the insect host, the quorum sensing system must play an important role in tlie regulation of bacterial gene expression during symbiosis. The Table 1. dNrdS ratios computed from pairwise comparisons of genes involved in quorum sensing in S. glossinidius, SOPE and related free-living bacteria. Sequence Pairs dti cfS dNIcfS Ratio S. glossinidius carA- SOPE carA 0.058 0.610 0.095 S. glossinidius carA- £ carotovora carA 0.698 1.504 0.464 SOPE carA- £ carotovora carA 0.694 1.766 0.393 S. glossinidius sogl- SOPE sogl 0.036 0.506 0.071 S. glossinidius sogl- Y. pestis ypel 0.466 1.800 0.259 SOPE sogl- Y. pestis ypel n/d* n/d* n/d* S. glossinidius sogRl- SOPE sogRl 0.039 0.635 0.061 S. glossinidius sogRl- Y. pestis ypeR 0.533 1.484 0.359 SOPE sogRl- Y. pestis ypeR n/d* n/d* n/d* S. glossinidius sogR2- SOPE sogR2 0.045 0.388 0.116 S. glossinidius sogR2- E. carotovora expR2 n/d* n/d* n/d* S. glossinidius sogR2- E. carotovora carR n/d* n/d* n/d* SOPE sogR2- £ carotovora expR2 0.497 1.970 0.252 SOPE sogR2- E. carotovora carR n/d* n/d* n/d* £ carotovora expR2- £ carotovora carR n/d* n/d* n/d* *n/d - not determined due to saturation (dS>2) at synonymous sites. GenBank accession numbers for the sequences used in these analyses are provided in Supplementary Information SI. doi: 10.1371/journal.pone.0003541 ,t001 .S', glossinidius quorum sensing system utilizes at least two distinct regulatory proteins, designated Sogl and SogRl. Sogl is an AIIL synthase that is responsible for the synthesis of a signaling molecule, OIIIIL. SogRl is a DNA-binding transcriptional regulator that interacts with OIIIIL and modulates the expression of target genes in accordance with bacterial population density. Sodalis glossinidius also maintains a second LuxR homolog (SogR2) but tlie function of this protein was not elucidated in this study. In the current study, we used a tiling microarray to identify genes in the .S', glossinidius genome whose expression is modulated in tlie presence of the quorum sensing signaling molecule, OIIIIL. The most striking finding was that OIIIIL increased the expression of a large number of genes that are known to be involved in the cellular response to oxidative stress. These includes oxyR, the positive regulator of hydrogen peroxide inducible genes [45], and numerous genes whose protein products are known to be involved in the direct detoxification of ROS [46,47], including eatalase (kaUi), manganese superoxide dismutase (sodii), peroxidase (ahpC) and glutathione S-transferase (jfd!). We also detected increased expression of genes encoding proteins known to be involved in the repair of cellular components following ROS- mediated damage [48-54], These include thioredoxin reductase (trxB), methionine sulfoxide reductase B (msrE) and several chaperones known to be involved in protein refolding and maturation under oxidative stress (e.g., clpB, dnaK, grpE, hlpG, ihpA, nfiui). The microarray data also shows that OIIIIL modulated the expression of genes involved in metal ion transport and storage. In many bacteria, the expression of genes involved in iron and manganese transport and metabolism is coordinately regulated with genes involved in the oxidative stress response [55,56], Both manganese and iron are known to serve as essential co-factors in a number of enzymes that detoxify ROS (e.g., eatalase and superoxide dismutase). Also, soluble iron is known to contribute directly to oxidative stress by catalyzing the formation of free radicals through the Fenton reaction [46], OIIIIL was also found to increase the expression of genes involved in the production of iron siderophores and baeterioferritin, which have been shown to play a role in ameliorating oxidative stress by removing toxic iron from solution and (in the case of baeterioferritin) serving as iron storage complexes for heme enzymes [57-60], In .S', glossinidius, OIIIIL also increased expression of genes encoding an iron/ manganese transport system (siLWCD) that is known to be associated with resistance to oxidative stress in other bacteria [61,62,63], Furthermore, biochemical assays showed that .S', glossinidius produced and secreted increased amounts of iron siderophores into culture media in the presence of exogenous OIIIIL. Finally, it should be noted that OIIIIL increased transcription of a cluster of genes encoding a polvol metabolism and transport system in .S', glossinidius (SG0608-SG0614: 1.4-3 fold upregulation). This is conspicuous because polvols are known to be abundant in animal cells under conditions of hyperglycemia [64], Many insect endosymbionts, including .S', glossinidius, are known to maintain extremely high infection densities within host tissues and cells over prolonged periods of time [65,66,67], In a recent study, Ileddi et al. [68] compared the metabolic and transcrip­tional profiles of maize weevil baeterioeytes in the presence and absence of a symbiotic bacterium (SZPE) that is another close relative of .S', glossinidius. Notably, symbiont-infected cells were found to display an unusual profile of carbohydrate transport and metabolism characterized by the induction of the polvol pathway through increased expression of aldose reductase enzymes. This adaptation is predicted to occur in response to the high carbohydrate uptake and intense metabolic activity in densely October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing infected bacteriocytes, which generates conditions analogous to those found in mammalian diabetic cells. One critical consequence of this intense metabolic activity is the generation of increased concentrations of ROS [64], Under these conditions, the weevil bacteriocytes w'ere found to display increased expression of genes encoding proteins that are expected to ameliorate the deleterious effects of ROS. Many bacteria are know'n to undergo an "adaptive response" upon challenge with a sublethal dose of ROS [69], The sublethal challenge serves to prime the expression of a large number of genes involved in the oxidative stress response, rendering the cells more resistant to a subsequent lethal dose. Our findings suggest that quorum sensing serves to prime the cellular response to oxidative stress in S. glossinidius in a similar way. The use of a quorum sensing system for this purpose makes sense because the concentrations of ROS are expected to increase in accordance with the density of bacterial infection in host cells and tissues. Thus it seems that symbionts and host cells w'ork together to modulate their gene expression profiles and metabolic activities to minimize the deleterious effects of oxidative stress encountered during their symbiotic interactions. This likely represents a key adaptation in the symbiotic relationship because it allows host cells to sustain dense intracellular symbiont infections. The discover)/ of complementary responses towards oxidative stress in the weevil bacteriocytes and in S. glossinidius prompted us to search for the presence of quorum sensing regulator)/ genes in the grain weevil symbiont, SOPE. The SOPE genome sequence w'as indeed found to contain intact homologs of the S. glossinidius sogl, sogRl and sogR'2 genes. Phylogenetic analyses indicate that the lux genes w'ere present in the last common ancestor of S. glossinidius and SOPE, so their acquisition likely predates the origin of the symbiotic associations involving these bacteria. Although the functions of the SOPE lux genes have not been defined in this study, molecular evolutionary analyses indicate that sogl, sogRl and sogR'2 are evolving under stabilizing selection, consistent with the maintenance of gene function in the weevil symbiosis. Given that many bacterial symbionts, including S. glossinidius and SOPE, are know'n to attain extremely high infection densities in host tissues in spite of the burden of oxidative stress, it seems likely that such high infection densities confer important advantages within the symbiosis. For this reason, we expect other insect-bacterial associations to have similar adaptations to deal the challenge of oxidative stress. Furthermore, it is noteworthy that insect symbionts such as S. glossinidius and members of the candidate genus Arsenophonus have proved difficult to isolate and manipulate in pure culture because they demonstrate a high level of sensitivity towards oxidative stress w'hen cultures are maintained on agar plates at low' cell densities [70], Thus, increasing our understand­ing of the oxidative stress response in symbiosis may also contribute to the development of improved techniques for the culture and manipulation of symbionts. Following the establishment of symbiosis, the switch to a strictly vertical mode of transmission and the adoption of a static lifestyle leads to a process of genome streamlining in symbiotic bacteria that is characterized by the progressive inactivation and deletion of genes evolving under relaxed selection [18], This evolutionary trajectory/ is largely degenerative and irreversible because symbionts have little or no opportunity to replenish their genetic inventory/ through parasexual recombination. For example, in S. glossinidius, only 50% of the w'hole genome sequence is anticipated to encode functional proteins [43], The remaining DNA is composed largely of pseudogenes - genes with reading frames truncated by more than 50% as a result of frameshifts and/or premature stop codons. In the current study we found that one such "pseudogene," carA (formally a component of an antibiotic biosynthesis gene cluster), unexpectedly displayed a 5-fold increase in transcription in response to OHHL. Pairwise analyses of dN/dS ratios show' that carA is evolving under strong stabilizing selection in S. glossinidius and SOPE. Taken together, these observations suggest that carA is maintained as a functional protein coding gene, albeit in a truncated form. This illustrates the potential for adaptive neofunctionalization to occur as a component of the process of genome degeneration and streamlining. Methods Bacterial Strains and Culture Conditions A complete list of plasmids and strains used in this study is provided in Supplementary Table SI. Sodalis glossinidius w'as maintained at 25°C in the semi-defined liquid Mitsuhashi- Maramorosch (MM) medium as described previously [65] or in a defined liquid medium containing 0.15 g/L CaCl2, 0.046 g/L MgCl2, 0.2 g/L KC1, 7.0 g/L NaCl, 4.0 g/L glucose and 6 g/L deferrated casamino acids. Escherichia coli strains w'ere maintained at 37°C in either Luria-Bertani (LB) medium or in M9 liquid minimal medium salts supplemented with 0.24 mg/L MgS04, 0.01 mg/L CaCl2, 4 g/L lactose and 2 g/L casamino acids. Agrobacterium tumefaciens KYC55 (pJZ372) (pJZ384) (pJZ410) w'as maintained at 28°C in AT minimal medium [24], Chromobacterium violaceum CV026 and P. aeruginosa PAOl w'ere maintained at 30°C and 37°C, respectively, in LB medium [25], Where appropriate, antibiotics w'ere added to the media at the following concentra­tions: 100 (Ig/ml of ampicillin, 100 (Ig/ml of gentamycin, 30 (lg/ ml (chromosomal integrations) or 50 (Ig/ml (high copy number plasmids) ofkanamycin, 100 (Ig/ml of spectinomycin and 1.5 (lg/ ml of tetracycline. Extraction of Culture Supernatants Acylated homoserine lactones w'ere extracted from culture supernatants as described by Shaw' et al. [71], Briefly, 500 ml of a stationary/ phase culture of S. glossinidius (grow'n in liquid MM medium), E. coli [TOPIO and TOPIO (pSGI) grow'n in defined liquid medium] and P. aeruginosa w'ere pelleted by centrifugation (8,000 xg, 20 min., 4°C). The resulting culture supernatants w'ere filtered through 0.2 (im pore-size membrane filters (Nalgene Labware, Cat. No. 154-0020) and extracted twice with an equal volume of ethyl acetate. The extracts w'ere combined, dried with anhydrous magnesium sulfate, filtered, evaporated using a vacuum centrifuge and resuspended in 500 (iL of acetonitrile. Thin Layer Chromatography Overlay Assay Samples w'ere chromatographed on Ci« reverse-phase TLC plates (Whatman, Cat. No. 4801-800) using methanol:water (60:40). Following development, plates w'ere dried and overlaid with live cultures of A. tumefaciens KYC55 (pJZ372) (pJZ384) (pJZ410) or C. violaceum CV026 indicator strains as described previously [24,25], High Performance Liquid Chromatography and Mass Spectrometry Analysis S. glossinidius and E. coli TOPIO (pSGI) culture supernatant extractions w'ere fractionated using reverse-phase high perfor­mance liquid chromatography. The samples w'ere chromato­graphed on a Cm column with a linear w'ater-acetonitrile gradient (0-40%) containing 0.1% trifluoroacetic acid (v/v). The eluted fractions w'ere screened for the presence of AHL with the A. tumefaciens biosensor strain [24]. Bioactive fractions from the E. coli TOPIO (pSGI) sample w'ere evaporated using a vacuum centrifuge October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing and rcsuspcndcd in methanol. The samples were analyzed by electrospray ionization and CID-MS using a Micromass Quattro II - Triple Quadrupole Mass Spectrometer under positive-ion conditions. Fractions were injected into the mass spectrometer in a solvent containing 50% methanol, 49.9% water and 0.1% formic acid at a flow rate of 5 |J.L/min. P-Galactosidase Assay E. coli strain BW25113, harboring specific reporter plasmids, w'ere grow'n overnight in LB medium. Overnight cultures w'ere diluted 1:500 in fresh medium containing 0, 4 n.M. 40 n.\I. 0.4 |0.M, 4 |0.M, 40 |XM, 100 |XM or 200 |XM of N-(3-oxohexa- noyl)-homoserine lactone (Sigma Aldrich, Cat. No. K3255). The cultures w'ere maintained for 3 hours at 37°C and P-galactosidase activity w'as measured as described by Miller [72]. All assays w'ere conducted in triplicates. Artificial Induction of Quorum and RNA Isolation A culture of S. glossinidius w'as grow'n to early log phase (ODgoo~0.04, approximately 1.5 xlO7 CFU/ml) in MM liquid medium. This culture w'as separated to yield eight cultures of equal volume and 100 |jM of N-(3-oxohcxanoyl)-homoscrine lactone (Sigma Aldrich) w'as added to four of the replicates. After 12 hours of incubation at 25°C, the cells w'ere harvested by centrifugation (5,000 xg, 10 min., 4°C). RNA w'as prepared using the SV Total RNA Isolation System (Promega, Cat. No. Z3100), according to the kit instructions. Aliquots of the RNA samples w'ere treated with DNase I (Ambion, Cat. No. AM1907) to remove contaminating DNA. RNA samples w'ere then reverse transcribed using the TaqMan Reverse Transcription Reagents (Applied Biosystems, Part No. N808-0234), according to the manufacturers instructions. Artificial induction of quorum w'as verified by measuring relative transcript levels of sogl in induced versus uninduced samples using qPCR. Microarray Expression Analyses and Quantitative PCR A custom tiling microarray w'as designed using the eArray Design Creation online application (Agilent Technologies). The array consisted of approximately 40,000 sixty-mer oligonucleotide probes, one probe per 200 bp, covering the entire Sodalis glossinidius genome (GenBank accession number AP008232.1). Standard positive and negative control features w'ere also included on the array. Microarrays w'ere printed using Agilent SurePrint technol­ogy in the 4 x44 k slide format. Microarray hybridization w'as performed according to the Agilent Technologies protocol, with the following modifications. Briefly, poly(A) tails w'ere added to the 3' end of the RNA molecules using the Ambion Poly(A) tailing kit (Cat. No. 1350). The polyadenylated RNA w'as then used as a template to generate lluorescently labeled cRNA using the Agilent Two-Color Low' RNA Input Linear Amplification Kit, labeling OHHL+ RNA with cyanine 3-cytosine triphosphate (CTP) and OHHL- RNA with cyanine 5-CTP, and vice versa. Fluorescently labeled cRNA samples (825 ng each), in addition to Agilent RNA spike-in controls, w'ere then fragmented and hybridized to the tiling microarray using the Agilent 2-color GE Hybridization/Wash protocol. Hybridized slides w'ere then scanned in an Agilent Technologies G2505B Microarray Scanner at 5 |J.m resolution, performing a simulta­neous detection of Cyanine-3 and Cyanine-5 signal on the hybridized slide. An extended dynamic range scan w'as then accomplished by performing a primary scan at 100% laser pow'er and a secondary scan at 10% pow'er: the former used to calculate intensities for non-saturating features, and the latter used to calculate intensities for saturating features. The scanned micro­array image files w'ere then loaded into Agilent Feature Extraction Software (v. 9.5.1), w'hich w'as used to perform calculations that included feature intensities, background measurements, and statistical analyses. To control for variation in individual probe hybridization efficiency and fluorescence intensity, mechanically sheared w'hole genomic DNA w'as used as a template for cRNA hybridization on a separate array. Differences in individual probe intensities w'ere used to normalize the raw' experimental data. Hierarchical clustering analyses correctly broke dow'n normalized data by sample treatment (+/- OHHL, dye type: r2 values >0.95)." Statistical analysis of the results w'as performed using Tiling Microarray Analysis Tools 2 (http://sourccforgc.net/projccts/ timat2). CyberT w'as used to estimate a confidence in the differential expression by calculating the posterior probability of differential expression [73], Aflymetrix's Integrated Genome Brow'ser w'as used to visualize the microarray analysis data (http:// w'ww.aftymetrix.com/support/developer/tools/dow'nload_igb.aftx). Microarray results w'ere verified by qPCR. Briefly, RNA samples from the four biological replicates w'ere pooled together, and subjected to DNase I treatment (Ambion, Cat. No. AM 1907) until no DNA could be detected by qPCR. RNA samples w'ere then reverse transcribed using the TaqMan Reverse Transcription Reagents (Applied Biosystems, Part No. N808-0234), according to the manufacturers instructions. Reactions w'ere performed in triplicates using iQ.Supermix (Bio-Rad, Cat. No. 170-8862), and the samples w'ere quantitated using an iCycler iQ^Multicolor Real­Time PCR Detection System (Bio-Rad). Relative transcript levels w'ere estimated using the standard curve method described previously by Dale et al. [74], In order to validate the microarray results, 10 different loci that w'ere either upregulated, dow'nreg- ulated or displayed no changed in expression profile in the microarray experiment w'ere selected for the qPCR verification. Siderophore Quantitative Assay A 150 ml culture of S. glossinidius w'as grow'n to early log phase (ODgoo~0.04, approximately 1.5xl07 CFLT/ml) in MM liquid medium. The cells w'ere pelleted by centrifugation (4,500 Xg, 10 min., 15°C) and w'ashed twice in an equal volume of 0.85% NaCl solution. After a third centrifugation, the cells w'ere rcsuspcndcd in 15 ml of 0.85% NaCl solution and 1 ml of this cell suspension w'as used to inoculate six replicates of 30 ml of defined medium. 100 |jM N-(3-oxohcxanoyl)-homoscrine lactone (Sigma Aldrich, Cat. No. K3255) w'as added to three of the replicates. After 24 and 48 hours of incubation at 25°C, culture aliquots w'ere pelleted by centrifugation (8,000 Xg, 20 min., 4°C), supernatants w'ere filtered through 0.2 |Xm pore-size membrane filters (Millipore, Cat. No. SLGP033RB) and siderophore units w'ere estimated using the chrome azurol S liquid assay [75], To control for the effects of N-(3-oxohcxanoyl)-homoscrine lactone as a potential iron chelating agent [76], 100 |jM of N-(3-oxohcx- anoyl)-homoscrine lactone w'as added to all supernatants and reference readings prior to quantitation. Phylogenetic and Molecular Evolutionary Analyses Phylogenetic analyses w'ere performed on homologs of laxl and luxR from S. glossinidius, a closely related grain w'eevil endosym­biont (SOPE) [67], and other closely related Gram negative bacteria know'n to maintain AHL-based quorum sensing systems. The nucleotide sequences of laxl and luxR w'ere aligned according to the corresponding protein sequence alignments generated in CLLTSTAL, to provide in-frame nucleotide sequence alignments. All third codon position characters w'ere then excluded from subsequent analyses to improve the signal to noise ratio (third October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing position characters were deemed uninformative due to high levels of substitution at synonymous sites). Maximum likelihood (ML) analyses were performed in PAUP 4.0 [77], using the heuristic tree-bisection-reconnection algorithm. ML parameters were estimated from an initial neighbor-joining tree and optimized in the construction of ML trees using variable base frequencies, a symmetrical substitution matrix and gamma distributed rate variation among sites. Bootstrap values were obtained from analysis of 100 replicates. Pairwise molecular evolutionary sequence analyses were performed in MEGA 4.0 [78]. Nucleotide sequence alignments were generated in frame, as described above. The frequencies of synonymous and nonsynonymous substitutions (dS and N, respectively) were estimated using the Kumar method. Z-tests of selection were then used to estimate the probability of rejecting a null hypothesis of strict neutrality in favor of an alternative hypothesis of stabilizing (purifying) selection. Supporting Information Table SI Plasmids and Strains Used in this Study- Found at: doi:10.1371/journal.pone.0003541.sOOl (0.06 MB DOC) Table S2 Results from Microarray Expression Analyses Found at: doi:10.1371/journal.pone.0003541.s002 (2.68 MB DOC) Supplementary Information SI Found at: doi:10.1371/journal.pone.0003541.s003 (0.03 MB DOC) Acknowledgments The authors thank Serap Aksoy (Yale University) for providing a culture of S. glossinidius. Sang-Jin Suh (Auburn University) for providing a culture of'/'. aeruginosa PAOl. Stephen Winans (Cornell University) for providing the A. tumefaciens reporter strain. Robert Simons (University of California. Los Angeles) for providing the plasmid pRS415 and Elliot Rachlin (University of Utah) for assistance with the mass spectrometry. Author Contributions Conceived and designed the experiments: MHP CD. Performed the experiments: MHP MB KS. Analyzed the data: MHP RL KO CD. Contributed reagents/materials/analysis tools: CD. Wrote the paper: MHP CD. References 1. Perrel X, Slaehelin C, Broughton WJ (2000) Molecular basis of symbiotic promiscuity, Microbiol Mol Biol Rev 64: 180 201. 2. Broughton WJ, Hanin M, Relic B, Kopcinska J, Golinowski W, el al. (2006) riavonoid-inducible modificalions Lo rhamnan O antigens are necessary for Rhizobitim sp. strain NGR234-legume symbioses. J Bacleriol 188: 3654 3663. 3. Campbell GR, Reuhs BL, Walker GC (2002) Chronic intracellular infection of alfalfa nodules by Sinorhizobium mclibli requires correct lipopolysaccharide core. Proc Nall Acad Sci U S A 99: 3938 3943. 4. Fraysse N, Jabbouri S, Treilhou M, Gouderc F, Poinsol V (2002) Symbiotic conditions induce structural modifications of Sinorhizobium sp. NGR234 surface polysaccharides. Glycobiology 12: 741 748. 5. Niehaus K, Lagares A, Puhler A (1998) A Sinorhizobium mclibli lipopolysaccharide mutant induces elfeclive nodules on the host plant Mcdicago saliva (alfalfa) but fails Lo establish a svmbiosis with Mcdicago Iruncalula. Mol PlanL Microbe InLeracL 11: 906 914. 6. Dazzo FB, Truchel GL, Hollingsworth Rl, Hrabak EM, Pankralz HS, et al. (1991) Rhizobitim lipopolysaccharide modulates infecLion Lhread developmenL in whiLe clover rooL hairs. J BacLeriol 173: 5371 5384. 7. Skorupska A, Janczarek M, Marczak M, Mazur A, Krol J (2006) Rhizobial exopolysaccharides: genetic conlrol and symbiotic funcLions. Microb Cell FacL 5. 8. Quandt J, Hillemann A, Niehaus K, Arnold W, Puhler A (1992) An osmoreverLanL of Rhizobiutn mclibli ndvB dele Lion muLanL forms infecLion Lhreads buL is defective in bacLeroid developmenL. Mol PlanL Microbe InLeracL 5: 420 427. 9. Nyholm SV, Deplancke B, Gaskins HR, Apicella MA, MeFafl-Ngai MJ (2002) Roles of Vibrio fscheri and nonsymbioLic bacLeria in Lhe dynamics of mucus secreLion during symbionL colonizaLion of lhe Eupymna scobpcs lighL organ. Appl Environ Microbiol 68: 5113 5122. 10. McFall-Ngai M, Brennan C, Weis V, Lamarcq L (1998) Mannose adhesin- glvcan inLeracLions in Lhe Eaprynma scolopes-Vibrio fischcri symbiosis. In: Le Gal Y, Halvorson HO, eds. New developmenLs in marine bioLechnology. New York: Plenum Press, pp 273 276. 11. Yip ES, Geszvain K, DeLonev-Marino CR, Visick KL (2006) The symbiosis regulaLor RscS conLrols Lhe syp gene locus, biofilm formaLion and symbiotic aggregaLion by Vibrio fischcri. Mol Microbiol 62: 1586 1600. 12. DeLoney-Marino CR, Wolfe AJ, Visick KL (2003) Chemoallraclion of Vibrio fiscluri Lo serine, nucleosides, and N-aceLylneuraminic acid, a componenL of squid lighL-organ mucus. Appl Environ Microbiol 69: 7527 7530. 13. Kaplan HB, Greenberg EP (1985) Dilfusion of auLoinducer is involved in regulaLion of Lhe Vibrio fischcri luminescence sysLem.J BacLeriol 163: 1210 1214. 14. Visick KL, Foster J, DoinoJ, McFall-Ngai M, Ruby EG (2000) Vibrio fischri lux genes play an imporLanL role in colonizaLion and developmenL of Lhe hosL lighL organ. J Bacleriol 182: 4578 4586. 15. Koropalnick TA, Engle JT, ApiceUa MA, Slabb EV, Goldman WE, el al. (2004) Microbial faclor-medialed developmenL in a hosl-baclerial muLualism. Science 306: 1186 1188. 16. WernegreenJJ (2002) Genome evolulion in baclerial endosymbionLs of insecLs. Nal Rev Genel 3: 850 861. 17. Zienlz E, Dandekar T, Gross R (2004) Melabolic inLerdependence of obligale inLracellular bacLeria and Lheir insecl hosLs. Microbiol Mol Biol Rev 68: 745 777. 18. Dale C, Moran NA (2006) Molecular inLeracLions belween baclerial symbionls and Lheir hosls. Cell 126: 453 465. 19. Baumann P (2005) Biology of bacleriocyle-associaled endosymbionLs of planl sap-sucking insecls. Annu Rev Microbiol 59: 155 189. 20. -Aksoy S, Chen X, Hypsa V (1997) Phylogeny and polenlial Iransmission rouLes of midguL-associaLed endosymbionLs of Iselse (Diplera:Glossinidae). InsecL Mol Biol 6: 183 190. 21. Moran NA, Russell JA, Koga R, Fukalsu T (2005) Evolutionary relalionships of ihree new species of Enlcrobaclcriaccac living as symbionls of aphids and olher insecls. .Appl Environ Microbiol 71: 3302 3310. 22. Tsuchida T, Koga R, Meng XY, Malsumolo T, Fukalsu T (2005) Characlerizalion of a facullalive endosymbiolic bacLerium of lhe pea aphid .Acyrlhosiphon p'isum. Microb Ecol 49: 126 133. 23. Gonzalez JE, Keshavan ND (2006) Messing with Bacterial Quorum Sensing. MicrobiolMol Biol Rev 70: 859 875. 24. Zhu J, Chai Y, Zhong Z, Li S, Winans SC (2003) Asrobaclcritim bioassay strain for ultrasensitive detection of N-acylhomoserine laclone-lype quorum-sensing molecules: detection of auloinducers in Mcsorhizobiam haakaii. Appl Environ Microbiol 69: 6949 6953. 25. McClean KH, Winson MK, Fish L, Taylor A, Chhabra SR, el al. (1997) Quorum sensing in Chro/mbaclcrium violaceum: exploitation ofviolacein production and inhibition for lhe deleclion of N-acylhomoserine laclones. Microbiology' 143:3703 3711. 26. Morin D, Grasland B, Vallee-Rehel K, Dufau C, Haras D (2003) On-line high- performance liquid chromalography-mass speclromelric deleclion and quanli- ficalion of N-acylhomoserine laclones, quorum sensing signal molecules, in lhe presence of biological malrices. J Chromalogr A 1002: 79 92. 27. Winson MK, Camara M, Lalifi A, Foglino M, Chhabra SR, el al. (1995) Multiple N-acyl-L-homoserine laclone signal molecules regulale produclion of virulence delerminanls and secondary melaboliles in Pseudomonas aeruginosa. Proc Nall Acad Sci U S A 92: 9427 9431. 28. Egland KA, Greenberg EP (1999) Quorum sensing in Vibrio fischcri: elemenls of lhe luxl promoler. Mol Microbiol 31: 1197 1204. 29. Fuqua WC, Winans SC (1994) A LuxR-LuxI lype regulalory syslem aclivales Agrobacknum Ti plasmid conjugal Iransfer in lhe presence of a planl lurnor melabolile.J Bacleriol 176: 2796 2806. 30. Lewenza S, Sokol PA (2001) Regulation of ornibaclin biosynlhesis and N-acyl-L- homoserine laclone production bv CepR in BurUiolderia cepacia. ] Bacleriol 183: 2212 2218. 31. Fuqua C, Winans SC, Greenberg EP (1996) Census and consensus in baclerial ecosyslems: Lhe LuxR-LuxI family of quorum-sensing Iranscriplional regulalors. -Annu Rev Microbiol 50: 727 751. 32. Darby AC, Lagnel J, Matthew CZ, Bourlzis K, Maudlin I, el al. (2005) Exlrachromosomal DNA of lhe symbiont Sodalis glossinidius. J Bacleriol 187: 5003 5007. 33. Rolfe B, Eisenberg MA (1968) Genetic and biochemical analysis of Lhe biotin loci of Escherichia coli K-12. J Bacleriol 1968 96: 515 524. 34. Derzelle S, Duchaud E, Kunsl F, Danchin A, Berlin P (2002) Identification, characlerizalion, and regulation of a clusler of genes involved in carbapenem biosynlhesis in Pholorhabdtis Itimincsccns. -Appl Environ Microbiol 68: 3780 3789. PLoS ONE | www.plosone.org 10 October 2008 | Volume 3 | Issue 10 | e3541 Insect Symbiont Quorum Sensing 35. McGowan SJ, Sebaihia M, Porter LE, Stewart GSAB, Williams P, et al. {1996} Analysis of bacterial carbapenem antibiotic production genes reveals a novel beta-lactam biosvnthesis pathwav. Mol Microbiol 22: 415 426. 36. McGowan SJ, Sebaihia M, O'Leary S, Hardie KR, Williams P, el al. (1997) Analysis of the carbapenem gene cluster of Erwinia carobvora: definition of the antibiotic biosynthetic genes and evidence for a novel beta-lactam resistance mechanism. Mol Microbiol 26: 545 556. 37. Thomson NR, Crow MA, McGowan SJ, Cox A, Salmond GPC (2000) Biosynthesis of carbapenem antibiotic and prodigiosin pigment in Serratia is under quorum sensing control. Mol Microbiol 36: 539 556. 38. Bainlon NJ, Stead P, Chhabra SR, Bvcrofl BW, Salmond GP, el al. (1992) jV-{3- oxohexanoyl)-L-homoserine lactone regulates carbapenem antibiotic production in Erwinia carotovora. Biochem J 288: 997 1004. 39. Andersson RA, Eriksson ARB, Heikinheimo R, Mae A, Pirhonen M, et al. (2000) Quorum sensing in the plant pathogen Erwinia carotovora subsp. carotovora: the role of expR [Ecc). Mol Plant Microbe Interact 13: 384 393. 40. McGowan S, Sebaihia M, Jones S, Yu B, Bainlon N, el al. (1995) Carbapenem antibiotic production in Erwinia carotovora regulated by CarR, a homologue to the LuxR transcriptional activator. Microbiology 141: 541 550. 41. Heddii A, Charles H, Khalchadourian C, Bonnot G, Nardon P [1998} Molecular characterization of the principal symbiotic bacteria of the weevil Sitophilus orj'zae: a peculiar G+C content of an endocylobiolic DNA. J Mol Evol 47: 52 61. 42. Dale C, Plague GR, Wang B, Ochman H, Moran NA {2002} Type III secretion systems and the evolution ofmulualislic endosymbiosis. Proc Natl Acad Sci U S A 99: 12397 12402. 43. Toh H, Weiss BL, Perkin SA, Yamashita A, Oshirna K, el al. {2006} Massive genome erosion and functional adaptations provide insights into the symbiotic lifeslvle of Sodalis glossinidim in the tsetse host. Genome Res 16: 149 156. 44. Rocha EP, Smith JM, Hurst LD, Holden MT, Cooper JE, el al. {2006} Comparisons of dN/dS are time dependent for closely related bacterial genomes. J Theor Biol 239: 226 235. 45. Bauer CE, Elsen S, Bird TH {1999} Mechanisms for redox control of gene expression. Annu Rev Microbiol 53: 495 523. 46. Imlay JA {2003} Pathways of oxidative damage. Annu Rev Microbiol 57: 395 41*8. 47. Kanai T, Takahashi K, Inoue H {2006} Three distinct-type glutathione S- transferases from Escherichia coli important for defense against oxidative stress. J Biochem 140: 703 711. 48. Angelini S, Gerez C, Choudens SO-d, Sanakis Y, Fonlecave M, et al. {2008} NfuA, a new factor required for maturing Fe/S proteins in Escherichia coli under oxidative stress and iron starvation conditions. J Biol Chern 283: 1408 14091. 49. Hossain MM, Nakamolo H {2003} Role for the cyanobacterial HlpG in protection from oxidative stress. Curr Microbiol 46: 70 76. 50. Kitagawa M, Miyakawa M, Malsumura Y, Tsuchido T {2002} Escherichia coli small heat shock proteins, IbpA and IbpB, protect enzymes from inactivation by heal tmd oxidants. Eur J Biochem 269: 2907 2917. 51. Maluszewska E, Kwialkowska J, Kuczynska-Wisnik D, Laskowska E {2008} Escherichia coli heat-shock proteins IbpA/B are involved in resistance to oxidative stress induced by copper. Microbiology 154: 1739 1747. 52. Stewart EJ, Aslund F, Beckwith J {1998} Disulfide bond formation in the Escherichia coli cytoplasm: an in vivo role reversal for the thioredoxins. EMBO J 17: 5543 5550. 53. Takemolo T, Zhang QM, Yonei S {1998} Different mechanisms of ihioredoxin in its reduced and oxidized forms in defense against hydrogen peroxide in Escherichia coli Free Radical Biol & Med 24: 556 562. 54. Weissbach H, Resnick L, Brol N {2005} Methionine sulfoxide reductases: history and cellular role in protecting against oxidative damage. Biochim Biophys Acta 1703: 203 212. 55. Hanlke K {2001} Iron and metal regulation in bacteria. Curr Opin Microbiol 4: 172 177. 56. Touali D {2000} Iron and oxidative stress in bacteria. Arch Biochem Biophys 373: 1 6. 57. Carrondo MA {2003} Ferritins, iron uptake and storage from the baclerioferrilin viewpoint. EMBO J 22: 1959 1968. 58. Chen CY, Morse SA {1999} .Neisseria gomrrhoeae baclerioferrilin: structural heterogeneity, involvement in iron storage and protection against oxidative stress. Microbiology 145: 2967 2975. 59. Ma J-F, Ochsner UA, Klolz MG, Nanayakkara VK, Howell ML, el al. {1999} Baclerioferrilin A modulates calalase A {KatA} activity and resistance to hydrogen peroxide in Pseudomonas aeruginosa.} Bacleriol 181: 3730 3742. 60. Tindale AE, Mehrolra M, Ollem D, Page WJ {2000} Dual regulation of calecholale siderophore biosynthesis in Azobbacter vinelandii by iron and oxidative stress. Microbiology 146: 1617 1626. 61. Davies BW, Walker GC {2007} Disruption of sitA compromises Sinorhizobimn meliloti for manganese uptake required for protection against oxidative stress. J Bacleriol 189: 2101 2109. 62. Kilphali W, Ngok-ngam P, Suwanmaneeral S, Sukchawalil R, Mongkolsuk S {2007} jigrobacterimn hmiefaciens j'ur has important physiological roles in iron and manganese homeostasis, the oxidative stress response, and full virulence. Appl Envir Microbiol 73: 4760 4768. 63. Sabri M, Leveille S, Dozois CM {2006} A SilABCD homologue from an avian pathogenic Escherichia coli strain mediates transport of iron and manganese and resistance to hydrogen peroxide. Microbiology 152: 745 758. 64. Brownlee M {2001} Biochemistry and molecular cell biology of diabetic complications. Nature 414: 813 820. 65. Dale C, Maudlin I {1999} Sodalis gen. nov. and Sodalis glossinidius sp. nov., a microaerophilic secondary endosymbiont of the tsetse fly Glossirui morsiiam morsiiam. Ini J Svsl Bacleriol 49: 267 275. 66. Mira A, Moran NA {2002} Estimating population size and transmission bottlenecks in maternally transmitted endosymbiolic bacteria. Microb Ecol 44: 137 143. 67. Heddi A, Grenier AM, Khalchadourian C, Charles H, Nardon P {1999} Four intracellular genomes direct weevil biology: nuclear, mitochondrial, principal endosymbiont, and Wolbachia. Proc Nall Acad Sci U S A 96: 6814 6819. 68. Heddi A, Vallier A, Anselme C, Xin H, Rahbe Y, el al. {2005} Molecular and cellular profiles of insect bacleriocyles: mutualism and harm al the initial evolutionary step of symbiogenesis. Cell Microbiol 7: 293 305. 69. Crawford DR, Davies KJA {1994} Adaptive response and oxidative stress. Environ Health Perspecl io2{Suppl. 10}: 25 28. 70. Pontes MH, Dale C {2006} Culture and manipulation of insect facultative symbionts. Trends Microbiol 14: 406 412. 71. Shaw PD, Ping G, Daly SL, Cha C, CronanJE, el al. {1997} Detecting and characterizing N-acyl-homoserine lactone signal molecules by thin-layer chromatography. Proc Nall Acad Sci U S A 94: 6036 6041. 72. Miller JH {1992} A shorl course in bacterial genetics: a laboratory manual and handbook for Escherichia coli and related bacteria. New York: Cold Spring Harbor Laboratory Press, pp 72 74. 73. Baldi P, Long AD {2001} A Bayesian framework for the analysis of microarray expression data: regularized l-lesl and statistical inferences of gene changes. Bioinforinalics 17: 509 519. 74. Dale C, Jones T, Pontes M {2005} Degenerative evolution and functional diversification of type-III secretion systems in the insect endosymbiont Sodalis glossinidius. Mol Biol Evol 22: 758 766. 75. Payne SM {1994} Detection, isolation, and characterization of siderophores. Methods Enzymol 235: 329 344. 76. Kaulinann GF, Sartorio R, Lee SH, Rogers CJ, Meijler MM, el al. {2005} Revisiting quorum sensing: discovery of additional chemical and biological functions for 3-oxo-N-acylhomoserine lactones. Proc Natl Acad Sci U S A 102: 309 314. 77. Swofford DL {2000} PAUP*: Phylogenetic Analysis Using Parsimony and other methods. Sunderland, MA: Sinauer, 4bl0 ed:. 78. Tamura K, Dudley J, Nei M, Kumar S {2007} MEGA4: Molecular Evolutionary Genetics Analysis {MEGA} software version 4.0. Mol Biol Evol 24: 1596 1599. 79. Simons RW, Hournan F, Kleckner N {1987} Improved single and multicopy lac- based cloning vectors for protein and operon fusions. Gene 53: 85 96. 80. Dalsenko KA, Wanner BL {2000} One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci U S A 97: 6640 6645. PLoS ONE | www.plosone.org 11 October 2008 | Volume 3 | Issue 10 | e3541