Insights into the origins and evolution of mutualistic insect-bacterial symbioses

A diverse array of insect species harbor maternally transmitted mutualistic bacterial endosymbionts that perform a variety of functions within their hosts. Many of these associations are obligate in nature with the insect relying on the bacterial symbiont to provide nutrients that are lacking in the insect's natural diet. These obligate endosymbionts often show a highly reduced genome size and maintain only a small fraction of the gene inventory of free-living bacteria. Some of the smallest known bacterial genomes are from obligate endosymbionts that have been associated with their insect hosts for long periods of time. In addition to their small size, the genomes of ancient obligate symbionts also show an increased rate of DNA and polypeptide sequence evolution as well as a nucleotide composition bias that results in an increased ratio of adenine and thymine residues. Despite extensive study of these ancient endosymbionts, little is known about their origins. To address this issue and to better understand the forces shaping genomes in the early stages of an endosymbiotic association, this work focuses on two bacteria: strain HS, a recently characterized free-living bacterium that likely served as a progenitor to the Sodalis-allied clade of bacterial endosymbionts, and the Sitophilus oryzae primary endosymbiont (SOPE), a very recent established maternally transmitted obligate endosymbiont of the rice weevil. The complete genome sequencing of these two bacteria along with comparative genomic analyses revealed that SOPE has undergone a very rapid degeneration of its genome, losing nearly half of its coding capacity, a massive expansion of insertion sequence (IS) elements and numerous intragenomic rearrangements facilitated by the IS elements. Surprisingly, these changes have happened very recently since strain HS and SOPE shared a common ancestor approximately 28,000 years ago.

INSIGHTS INTO THE ORIGINS AND EVOLUTION OF MUTUALISTIC INSECT-BACTERIAL SYMBIOSES by Kelly F. Oakeson A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Biology The University of Utah May 2014 Copyright © Kelly F. Oakeson 2014 All Rights Reserved The University of Utah Graduate School STATEMENT OF DISSERTATION APPROVAL The following faculty members served as the supervisory committee chair and members for the dissertation of______________Kelly I7. Oakeson________________ Dates at right indicate the members' approval of the dissertation. C°°n Dale , Chair 12/13/2013 Date Approved Richard Clark , Member 12/13/2013 Date Approved .Ton Seger , Member 12/13/2013 Date Approved MicTael Shapi°o , Member 12/13/2013 Date Approved RobertWeiss , Member 12/13/2013 Date Approved The dissertation has also been approved by........... Chair of the Department of_____________ Biol°°y and by David B. Kieda, Dean of The Graduate School. ABSTRACT A diverse array of insect species harbor maternally transmitted mutualistic bacterial endosymbionts that perform a variety of functions within their hosts. Many of these associations are obligate in nature with the insect relying on the bacterial symbiont to provide nutrients that are lacking in the insect's natural diet. These obligate endosymbionts often show a highly reduced genome size and maintain only a small fraction of the gene inventory of free-living bacteria. Some of the smallest known bacterial genomes are from obligate endosymbionts that have been associated with their insect hosts for long periods of time. In addition to their small size, the genomes of ancient obligate symbionts also show an increased rate of DNA and polypeptide sequence evolution as well as a nucleotide composition bias that results in an increased ratio of adenine and thymine residues. Despite extensive study of these ancient endosymbionts, little is known about their origins. To address this issue and to better understand the forces shaping genomes in the early stages of an endosymbiotic association, this work focuses on two bacteria: strain HS, a recently characterized free-living bacterium that likely served as a progenitor to the Sodalis-allied clade of bacterial endosymbionts, and the Sitophilus oryzae primary endosymbiont (SOPE), a very recent established maternally transmitted obligate endosymbiont of the rice weevil. The complete genome sequencing of these two bacteria along with comparative genomic analyses revealed that SOPE has undergone a very rapid degeneration of its genome, losing nearly half of its coding capacity, a massive expansion of insertion sequence (IS) elements and numerous intragenomic rearrangements facilitated by the IS elements. Surprisingly, these changes have happened very recently since strain HS and SOPE shared a common ancestor approximately 28,000 years ago. iv I dedicate this dissertation to my entire family. Especially, to the memory of my late father, without whose support I would not have been able to complete my goal of obtaining a Ph.D. TABLE OF CONTENTS ABSTRACT........................................................................................................................... iii LIST OF TABLES................................................................................................................ viii LIST OF FIGURES................................................................................................................ ix ACKNOWLEDGEMENTS.................................................................................................. xi Chapters 1. INTRODUCTION.............................................................................................................. 1 2. A NOVEL HUMAN-INFECTION-DERIVED BACTERIUM PROVIDES INSIGHTS INTO THE EVOLUTIONARY ORIGINS OF MUTUALISTIC INSECT-BACTERIAL SYMBIOSES............................................................................................ 5 2.1 Abstract................................................................................................................. 6 2.2 Introduction...........................................................................................................6 2.3 Results................................................................................................................... 7 2.4 Discussion............................................................................................................ 11 2.5 Materials and Methods....................................................................................... 15 2.6 References........................................................................................................... 17 2.7 Acknowledgments.............................................................................................. 17 3. GENOME DEGENERATION AND ADAPTATION IN A NASCENT STAGE OF SYMBIOSIS.............................................................................................................. 19 3.1 Abstract...............................................................................................................20 3.2 Introduction.........................................................................................................20 3.3 Materials and Methods....................................................................................... 21 3.4 Results................................................................................................................. 23 3.5 Discussion...........................................................................................................31 3.6 Supplementary Material.....................................................................................34 3.7 Acknowledgements............................................................................................ 34 3.8 Literature Cited.................................................................................................. 34 4. FUTURE STUDIES AND CONCLUSIONS................................................................ 38 4.1 The Primary Endosymbiont of Sitophilus zeamais...........................................39 4.2 Summary and Conclusions.................................................................................42 4.3 References........................................................................................................... 46 APPENDIX. LIST OF PUBLICATIONS.......................................................................... 49 vii LIST OF TABLES 2.1. General feature of the strain HS, SOPE, and S. glossinidius genome sequences.....11 2.2 Allelic spectrum of pseudogene mutations in strain HS orthologs found in SOPE and S. glossinidius........................................................................................................ 15 3.1 Summary of the four major IS elements in the SOPE Genome.............................. 23 LIST OF FIGURES 2.1. Phylogeny of strain HS and related Sodalis-allied endosymbionts and free-living bacteria based on maximum likelihood analyses of a 1.46 kb fragment of 16s rRNA and a 1.68 kb fragment of groEL....................................................................... 8 2.2 Alignment between strain HS contigs (top) and chromosomes of SOPE (left) and S. glossinidius (right)...............................................................................................9 2.3 Retention of strain HS orthologs in S. glossinidius and SOPE according to COG functional category............................................................................................. 10 2.4 Alignments of three regions of the S. glossinidius, strain HS and SOPE chromosomes................................................................................................................ 10 2.5 Average size of strain HS orthologs classified as intact, pseudogenized and absent in SOPE (green) and S. glossinidius (red)....................................................... 12 2.6 Densities of disrupting mutations in SOPE and S. glossinidius pseudogenes......... 13 2.7 Numbers of cryptic pseudogenes in S. glossinidius and SOPE estimated using a Monte Carlo simulation................................................................................................ 14 2.8 Base composition bias and mutation rates observed in pairwise comparisons between strain HS, S. glossinidius and SOPE............................................................ 15 3.1 Microscopic images of SOPE...................................................................................... 23 3.2 Whole genome sequence alignment of strain HS and SOPE....................................24 3.3 IS element dense regions in SOPE..............................................................................25 3.4 Plot of dN vs. dS of orthologous genes in strain HS and SOPE.............................. 27 3.5 Overview of SOPE metabolism..................................................................................29 3.6 Comparison of type III secretion system gene clusters.............................................31 x ACKNOWLEDGEMENTS I would like to thank all the members of my lab, both past and present, and especially my advisor Colin Dale. I would also like to thank all the members of my advisory committee for their supportive comments, questions, and concerns. I also want to thank my wonderful fiance for all of the love, motivation, and support she has given me throughout these years; I couldn't have done it without her. CHAPTER 1 INTRODUCTION Insect-bacterial symbioses are widespread in nature, with an estimated 15% of all insect species harboring some type of bacterial symbiont. Numerous different types of insects harbor mutualistic endosymbionts, including many insects of agricultural importance, such as crop pests like weevils and aphids, and insects of medical importance that spread disease, such as the tsetse fly. These symbionts may allow their insect hosts to survive on new diets or on diets lacking essential nutrients, or may allow their host to survive under various environmental conditions or pressures. By performing such functions, these symbionts play important roles in the ecology and evolution of their insect hosts. Understanding how these associations arise and are maintained is of great interest. The nature of these mutualistic associations has long been the subject of molecular and evolutionary studies, focused primarily on those associations of ancient origin. Ancient symbionts share many features, including long-term host association, strict vertical transmission, residence inside specialized host tissues, supplements to the host diet, and very stable reduced genomes with an AT nucleotide bias. On the other end of the spectrum lie the recently acquired bacterial symbionts. These symbionts have only been associated with their respective host for a short period of time; they are mostly transmitted vertically, but there are some cases of horizontal transmission. Unlike ancient symbionts, these young symbionts reside in multiple host tissues and may perform diverse functions for the host. These recent symbionts also have genomes more similar to free-living bacteria in their size, nucleotide composition, and content. Another way to describe these symbionts is in the context of degenerative genome evolution. Degenerative genome evolution occurs when a free-living, non-host restricted bacteria with a large genome, few pseudogenes, and mobile elements degenerates to the reduced genome of an ancient, bacteriome associated symbiont. Once a bacterium becomes host-restricted, the genome starts to acquire pseudogenes, mobile elements, (i.e., bacterial insertion sequence elements), large and small deletions, and chromosome rearrangements. As the bacteria genome deterioration progresses down this host-restricted path, many pseudogenes and mobile elements are shed. The deletion of genes is accelerated, resulting in a very small and stable bacterial genome containing only a handful of genes with very little gene loss. Insect symbionts provide a great platform from which to study genome evolution. These bacteria live in a static and stable environment, they evolve in isolation with no lateral gene transfer from free-living bacteria, and they have a small population size with frequent population bottlenecks. These characteristics provide us with the opportunity to study the evolutionary process in the absence of gene exchange. 2 By studying these symbionts, particularly the more recent symbionts, we can start to answers some important questions about bacterial-insect symbioses. How do these symbioses originate; are they initially pathogenic bacteria that infect an insect host? Why do we see diverse insects maintaining symbiotic relationships with very closely related bacteria? As mentioned previously, there has been extensive work preformed on ancient symbionts with very reduced genomes, but little work has focused on the dynamics of genome degeneration shortly after the onset of host association. Is degeneration an aggressive process right from the onset or does it take place slowly over time? How do genes become inactivated and eventually lost and what is the role of repetitive DNA? In order to address these questions, my work has focused on a group of bacterial symbionts called the Sodalis-allied clade. These are closely related bacteria symbionts that are found in diverse insect hosts with wide geographical and ecological distributions. These symbiont are found in tsetse flies, mealybugs, spittlebugs, weevils, stinkbugs, and even bird lice. The focus of this thesis is on two members of the Sodalis-allied clade of symbionts, the bacterial symbiont of Sitophilus oryzae, Candidatus Sodalis pierantonius str. SOPE, Sodalis glossinidius (a secondary symbiont of the tsetse fly) and the putative environmental progenitor of both of these symbionts, strain HS. This dissertation is divided into four chapters. Chapter 2 focuses on the discovery of strain HS and how ancestral relatives of strain HS have served as progenitors for the independent acquisition and descent of Sodalis-allied endosymbionts found in diverse insect hosts. Chapter 2 also uses comparative genomic analyses to show that the gene inventory of two Sodalis-allied endosymbionts, the secondary symbiont of the tsetse fly Sodalis glossinidius and SOPE, 3 are both derived subsets of an ancestral strain HS genomic template. Chapter 3 presents the completed genome sequences and annotations of strain HS and SOPE. This chapter also shows how SOPE has rapidly undergone extensive adaptation towards an insect-associated lifestyle by the functional and evolutionary analyses of protein coding genes in SOPE. Chapter 4 consolidates the conclusions of this work and provides further perspectives on the topics addressed in Chapters 2 and 3. 4 CHAPTER 2 A NOVEL HUMAN-INFECTION-DERIVED BACTERIUM PROVIDES INSIGHTS INTO THE EVOLUTIONARY ORIGINS OF MUTUALISTIC INSECT-BACTERIAL SYMBIOSES Reprinted with permission from Clayton, A. L. et al. A Novel Human-Infection-Derived Bacterium Provides Insights into the Evolutionary Origins of Mutualistic Insect-Bacterial Symbioses. PLoS Genet 8, e1002990 (2012). 6 OPEN 3 ACCESS Freely available online PLOS I A Novel Human-Infection-Derived Bacterium Provides Insights into the Evolutionary Origins of Mutualistic Insect-Bacterial Symbioses Adam L. Clayton1*, Kelly F. Oakeson1, Maria Gutin1, Arthur Pontes1, Diane M. Dunn2, Andrew C. von Niederhausern2, Robert B. Weiss2, Mark Fisher3'4, Colin Dale1 1 Department of Biology, University of Utah, Salt Lake City, Utah, United States of America, 2 Department of Human Genetics, University of Utah, Salt Lake City, Utah, United States of America, 3 ARUP Institute for Clinical and Experimental Pathology, University of Utah, Salt Lake City, Utah, United States of America, 4 Department of Pathology, University of Utah, Salt Lake City, Utah, United States of America Abstract Despite extensive study, little is known about the origins of the mutualistic bacterial endosymbionts that inhabit approximately 10% of the world's insects. In this study, we characterized a novel opportunistic human pathogen, designated "strain HS," and found that it is a close relative of the insect endosymbiont Sodalis glossinidius. Our results indicate that ancestral relatives of strain HS have served as progenitors for the independent descent of Soc/a//s-allied endosymbionts found in several insect hosts. Comparative analyses indicate that the gene inventories of the insect endosymbionts were independently derived from a common ancestral template through a combination of irreversible degenerative changes. Our results provide compelling support for the notion that mutualists evolve from pathogenic progenitors. They also elucidate the role of degenerative evolutionary processes in shaping the gene inventories of symbiotic bacteria at a very early stage in these mutualistic associations. Citation: Clayton AL, Oakeson KF, Gutin M, Pontes A, Dunn DM, et al. (2012) A Novel Human-Infection-Derived Bacterium Provides Insights into the Evolutionary Origins of Mutualistic Insect-Bacterial Symbioses. PLoS Genet 8(11): e1002990. doi:10.1371/journal.pgen.1002990 Editor: David S. Guttman, University of Toronto, Canada Received June 6, 2012; Accepted August 10, 2012; Published November 15, 2012 Copyright: © 2012 Clayton 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: This research was supported by National Science Foundation (www.nsf.gov) grant EF-0523818 and National Institutes of Health (www.nih.gov) grant 1R01AI095736 (to CD). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: [e-mail] Introduction Obligate host-associated bacteria often have reduced genome sizes in comparison to related bacteria that are known to engage in free-living or opportunistic lifestyles [1]. This is exemplified by inspection of the genome sequences of mutualistic, maternally transmitted, bacterial endosymbionts of insects, many of which have been maintained in their insect hosts for long periods of evolutionary time [2]. Often these obligate endosymbionts maintain only a small fraction of the gene inventory that is found in related free-living counterparts [3 5], indicating that the obligate host-associated lifestyle facilitates genome degeneration and size reduction. At a simple level, the process of genome degeneration in obligate endosymbionts can be viewed as a streamlining of the gene inventory to yield a minimal gene set that is compatible with the symbiotic lifestyle. Genes that have no adaptive benefit are inactivated and deleted as a consequence of mutations that accumulate under relaxed selection at an increased rate in the asexual symbiotic lifestyle as a result of frequent population bottlenecks occurring during symbiont transmission [6]. Although we now have a detailed understanding of the mechanisms and evolutionary trajectory of genome degeneration in ancient obligate insect symbionts, the fundamental question of how these mutualistic associations arise remains to be answered. Studies focusing on insect-bacterial symbioses of recent origin show that closely related bacterial endosymbionts are often found in distandy related insect hosts [7,8]. This could be explained by the interspecific transmission of symbionts, mediated by parasitic wasps and mites that facilitate the transfer of symbionts between distandy related hosts [9,10]. Horizontal symbiont transmission could also be mediated by intraspecific mating, as demonstrated in the pea aphid [11]. Another possibility is that symbionts could be acquired de novo from an environmental source. Symbiont acquisition, at least initially, requires the symbiont to overcome or evade the insect immune response. Given that many insects are known to possess a potent immune system that repels invading microorganisms [12], it has been assumed that mutua­listic symbionts arise from pathogenic progenitors that have evolved specialized molecular mechanisms to facilitate evasion of the immune response and invasion of insect tissues [2]. In support of this notion, it has been shown that the genomes of recendy acquired mutualistic insect endosymbionts maintain genes similar to virulence factors and toxins that are found in related plant and animal pathogens [13 18]. In the current study we describe the discovery of a novel human-infective bacterium, designated "strain HS", isolated from a patient who sustained a hand wound following impalement with a tree branch. Phylogenetic analyses show that strain HS is a member of the Sodalis-allied clade of insect endosymbionts. Comparative analyses of the genome sequences of strain HS and related insect symbionts suggest that close relatives of strain HS gave rise to mutualistic associates in a wide range of insect hosts. PLOS Genetics | www.plosgenetics.org November 2012 | Volume 8 | Issue 11 | e l002990 7 Author Summary Many insects harbor symbiotic bacteria that perform diverse functions within their hosts. However, the origins of these associations have been difficult to define. In this study we isolate a novel bacterium from a human infection and show that this bacterium is a close relative of the Sodalis-allied clade of insect symbionts. Comparative genomic analyses reveal that this organism maintains many genes that have been inactivated and lost independently in derived insect symbionts as a result of rapid genome degeneration. Our work also shows that recently derived Soc/a//s-allied symbi­onts maintain a significant population of "cryptic" pseudo­genes that are assumed to have no beneficial function in the symbiosis but have not yet accumulated mutations that disrupt their translation. Taken together, our results show that genome degeneration proceeds rapidly following the onset of symbiosis. They also highlight the potential for diverse insect taxa to acquire closely related insect symbionts as a consequence of vectoring bacterial patho­gens to plants and animals. Results Isolation and Culture of Strain HS A 71 -year-old male presented to his primary care physician for a routine physical examination three days after sustaining a puncture wound to the right hand. The patient fell and was impaled between the thumb and forefinger by a ~ 1 cm diameter branch while removing branches from a dead crab apple tree. Upon presentation the patient denied fever or other constitutional symptoms and had a mild peripheral blood monocytosis (11.8%; reference range = 1.7 9.3%). A palpable cyst was noted in the right hand at the sight of impalement. Warm compresses were applied and cephalexin was prescribed at a dose of 500 mg four times daily for 10 days. The patient was evaluated again three days later due to continuing wound pain. The cyst was drained by aspiration and serosanguineous fluid was submitted for Gram stain and bacterial culture. The Gram stain showed scattered white blood cells, but no bacteria were visualized. A follow-up visit seven days later revealed the presence of an abscess, although the patient was afebrile and without local lymphadenopathy. The abscess was again drained by aspiration and the patient was advised to consult an orthopedic surgeon for evaluation. Subsequent surgery, approximately six weeks later, removed several foreign bodies from the wound and the patient recovered on a second course of cephalexin without incident. Two days after the original cyst aspiration, small numbers of gram negative rods resembling enteric bacteria were isolated on MacConkey agar at 35°C and 5% C 0 2. Colonies were wet, mucoid, variable in size, and slowly fermented lactose. The isolate could not be definitively identified by a manual phenotypic method (RapID ONE, Remel, Lenexa KS) and was misidentified as Escherichia coli at 98% confidence by an automated system (Phoenix, BD Diagnostics, Sparks, MD). Phylogenetic Analysis of Strain HS Phylogenetic analysis of 16S rRNA placed strain HS in a well-supported clade comprising «Wa/tr-a]lied insect endosymbionts sharing >97% sequence identity in their 16S rRNA sequences (Figure 1), which is a commonly used threshold for species-level conservation among bacteria [19]. Aside from strain HS, the closest non-insect associated relative of this clade is Biostraticola toji, which was isolated from a biofilm on a tufa deposit in a hard water rivulet [20]. However, B. toji shares only 96.5% sequence identity PLOS Genetics | www.plosgenetics.org 2 in 16S rRNA with its closest insect associated relative (S. glossinidius), while strain HS shares >99% sequence identity with the primary endosymbionts of the grain weevils Sitophilus oiyzae and S. zeamais and with recently discovered endosymbionts from the chestnut weevil, Curculio sikkvmensis and the stinkbug, Cantao occelatus [21 23]. Analysis of a protein-coding gene, groEL, corroborated these findings, confirming that strain HS is a dose relative of the grain weevils, chestnut weevil and stinkbug endosymbionts (Figure 1). Genome Sequences of Strain HS and Related Insect Symbionts To compare the genome sequences of strain HS and related (Sodafo-allied endosymbionts, we aligned a draft sequence assembly of strain HS, comprising a total of 5.15 Mb of DNA in 271 contigs, with the complete genome sequences of the tsetse fly secondary endosymbiont, S. glossinidius (4.3 Mb) [24,25], and the recently completed sequence of Sitophilus oiyzae primary endosym­biont (SOPE; 4.5 Mb). The resulting alignments (Figure 2) reveal a remarkable level of conservation in gene content and organization between strain HS, S. glossinidius and SOPE. To determine if this high level of conservation is simply a consequence of the close evolutionary relationship between these bacteria, we also con­structed a whole genome sequence alignment between strain HS and Dickeya dadantii, which represents the next most closely related free-living bacterium whose whole genome sequence is available (Figure SI). This alignment shows that strain HS and D. dadantii are substantially more divergent in terms of their gene inventories, consistent with the notion that they occupy distinct ecological niches. Considering the alignments between strain HS, S. glossinidius and SOPE, it is notable that while the genome sequences of strain HS and S. glossinidius display an increased level of co-linearity, the relationship between strain HS and SOPE is predicted to be closer based on the fact that they share a higher level of genome-wide sequence identity (Figure 2). The genome sequences of strain HS and S. glossinidius demonstrate a typical pattern of polarized nucleotide composition in each replichore (G+C skew, Figure 2), whereas the SOPE genome has numerous perturbations in G+C skew that must result from recent chromosome rearrangements. These rearrangements likely arose as a consequence of intragenomic recombination events between repetitive insertion sequence (IS)-elements, which are highly abundant in the SOPE genome (Figure S2), and have been documented as a causative agent of deletogenic rearrangements in other bacteria [26 28]. Although the gene inventories of strain HS, S. glossinidius and SOPE share many genes in common, as expected given their close evolutionary relationship, each bacterium also maintains a fraction of unique genes. In strain HS we identified a total of 1.9 Mb of DNA encoding genes not found in either S. glossinidius or SOPE that are classified in a wide range of functional categories (Figure 3). This indicates that strain HS has many unique genetic and biochemical properties, and is consistent with the observation that strain HS, unlike the fastidious and microaerophilic S. glossinidius [14], grows under atmospheric conditions on minimal media. In addition, strain HS maintains a number of unique genes sharing high levels of sequence identity with virulence factors found in both animal and plant pathogens, including an Hrp-type effector protein that is characteristically utilized by plant pathogenic bacteria [29] (Table SI). This may be indicative of the ability of strain HS to sustain infection in plant tissues. In comparison with strain HS, the unique fractions of the £. glossinidius and SOPE chromosomes are composed almost exclu­sively of components of mobile genetic elements, including integrated prophage islands and IS-elements. Following excision Origins of Mutualistic Insect-Bacteria I Symbionts November 2012 | Volume 8 | Issue 11 | e l002990 8 Origins of Mutualistic Insect-Bacterial Symbionts i- Et 100 moo r 16S rRNA Vibrio cholerae Enterobacter hormaechei Shigella flexneri Escherichia coli Yersinia pestis Dickeya dadantii - Biostraticola tofi E (Cantao ocellatus) --- E (Sitophilus granarius) r E (Sitophilus zeamais) *• E (Sitophilus oryzae) ---- E (Columbicola columbae) HS E (Curculio sikkimensis) - E (Craterina melbae) - Sodalis glossinidius ---E (Paracoccus northofagicola) ---E (Sitophilus rugicollis) ■ Candidatus Moranella endobia - Wigglesworthia glossinidia ■ Vibrio cholerae GroEL E (Pseudolynchia canariensis) Candidatus Blochmannia pennsylvanicus Candidatus Blochmannia floridanus - E (Tetropium castaneum) E (Bactehcera cockerelli) ------ Escherichia coli - Dickeya dadantii i---Shigella flexneri '----- Yersinia pestis Sodalis glossinidius E (Cantao ocellatus) E (Curculio sikkimensis) HS E (Sitophilus zeamais) E (Sitophilus oryzae) - E (Col um bicola columbae) ---Candidatus Blochmannia pennsylvanicus I------------Wigglesworthia glossinidia '- Candidatus Blochmannia floridanus 0.05 substitutions/site 0.3 substitutions/site Figure 1. Phylogeny of strain HS and related Sodalis-aHied endosymbionts and free-living bacteria based on maximum likelihood analyses of a 1.46 kb fragment of 16S rRNA and a 1.68 kb fragment of g ro E L Insect endosymbionts that do not have proper nomenclature are designed by the prefix "E", followed by the name of their insect host. The numbers adjacent to nodes indicate maximum likelihood bootstrap values shown for nodes with bootstrap support >70%. doi:10.1371/journal.pgen.1002990.g001 of these mobile genetic elements in silico prior to alignment, the resulting genome sequences of S. glossinidius (3.21 Mb) and SOPE (3.15 Mb) represent near-perfect subsets of the strain HS genome (Figure 2), indicating that S. glossinidius and SOPE are abridged derivatives of a strain HS-like ancestor. Independent Gene Inactivation and Deletion in S. glossinidius and SOPE To further understand genetic differences between strain HS, S. glossinidius and SOPE, we analyzed three genomic regions containing relatively high densities of pseudogenes in both S. glossinidius and SOPE (Figure 4). The most notable finding to arise from this comparison is the absence of pseudogenes in the three genomic regions of strain HS. Furthermore, our comparative analysis shows that S. glossinidius and SOPE each have a unique complement of pseudogenes. Indeed, even for orthologous genes that have been inactivated in both S. glossinidius and SOPE, mutations leading to gene inactivation in each insect symbiont genome are distinct, indicating that gene inactivation and loss took place independently in S. glossinidius and SOPE, mostly as a consequence of small frameshifting indels. However, it should also be noted that the reductions observed in the gene inventories of S. glossinidius and SOPE are very similar at the level of functional categories, indicating that the insect-associated lifestyle imposes similar constraints on the retention of genes encoding core functions such as replication, transcription, translation and energy generation (Figure 3). In order to determine the number of pseudo genes throughout the genome of strain HS, we performed a manual annotation and careful inspection of the complete draft strain HS sequence assembly. Out of a total of 4,002 intact candidate ORFs identified in the draft annotation (Table SI), only 48 (including phage and IS elements) were found to be translationally frameshifted or truncated by more than 10% of the size of their most closely related orthologs in the GenBank data­base (Table 1). This finding stands in stark contrast to the gene inventories of both S. glossinidius and SOPE, in which pseudogenes represent a substantial fraction of their total genomic coding capacity (Figure 2) [24,25]. Thus, for both S. glossinidius and SOPE, the predominant evolutionary trajectory following obligate insect association involved the inactivation an d /o r loss of a substantial component of the ancestral (strain HS-like) gene inventory. Evolution of Pseudogenes in 5. glossinidius and SOPE The close evolutionary relationships between strain HS, S. glossinidius and SOPE indicate that the respective insect symbioses are recent in origin. This raises the possibility that a subset of selectively neutral genes in the S. glossinidius and SOPE genomes have not yet accumulated mutations that lead to disruption of their open reading frames. Such "cryptic" pseudogenes are assumed to have no adaptive benefit in the symbiosis and are expected to accumulate nonsense and/or frameshifting mutations in the future [30]. To determine if the genomes of S. glossinidius and SOPE maintain cryptic pseudogenes, we compared the average size of all strain HS genes with the average sizes of strain HS orthologs that are classified either as intact, absent (lost via large deletion) or pseudo genes (visibly disrupted) in the <S. glossinidius and SOPE genomes (Figure 5). First, it is important to note that the average size of the absent strain HS orthologs in S. glossinidius and SOPE is not significandy different from the average size of all strain HS ORFs, indicating that large deletion events are not significandy biased with respect to size. However, in both S. glossinidius and SOPE, genes in the pseudogene class were found to have a larger average size in comparison to all strain HS orthologs. Similarly, genes in the intact class were found to have a smaller size in comparison to all strain HS orthologs. This can be explained by the fact that larger genes have an increased likelihood of PLOS Genetics | www.plosgenetics.org November 2012 | Volume 8 | Issue 11 | e l002990 9 10 11 Origins of Mutualistic Insect-Bacterial Symbionts T a b le 1. General features of the strain HS, SOPE, and S. glossinidius genome sequences. Number of Number of Chromosome size intact genes pseudogenes Mobile DNA G+C content Strain HS 5.16 Mba 4002 (4364)b 48 0.14 Mb 56.73% SOPE 4.51 Mb 1414 1194 1.36 Mb 56.06% S. glossinidius 4.17 Mb 1355 1376 0.96 Mb 54.69% The chromosome size of strain HS is estimated based on the combined size of non-redundant contigs in the draft sequence assembly. The number in parentheses indicates the total number of candidate genes identified in strain HS, including representatives that are fragmented in the current assembly. aEstimated based on current draft assembly. ^Total number of genes identified in strain HS draft genome. Genes containing gaps in the draft assembly were excluded from all comparative analyses. doi:10.1371/journal.pgen.1002990.t001 SOPE-weevil symbiosis originated more recently than the S. glossinidius-tsetst fly symbiosis. This is further supported by a comparison of the estimates of corrected mutation density derived from the simulation (Figure 7). While SOPE is estimated to maintain only 2 disrupting mutations/kb of pseudogenes, S. glossinidius is estimated to maintain more than twice that density of disrupting substitutions (4.39 disrupting mutations/kb). On a related note, we were unable to utilize dN/dS ratios to identify cryptic pseudogenes in SOPE or S. glossinidius. This is likely due to the fact that stochastic variation resulting from differences in expression level, codon bias and other factors greatly exceeds any signal resulting from a recent relaxation of selection. Accelerated Sequence Evolution and Base Composition Bias in SOPE and S. glossinidius The transition to obligate insect-association is also known to catalyze base composition bias and accelerated polypeptide sequence evolution on the part of the symbiont [31]. The results oudined in Table 1 show that the genomic GC-contents of S. glossinidius and SOPE are lower than that of strain HS. However, to avoid any bias arising from the differential gene content of these organisms, we also performed comparative analyses focusing solely on orthologous sequences. This facilitated the comparison of 1,355 intact genes and 1,376 pseudogenes shared between strain HS and S. glossinidius, and 1,414 intact genes and 1,194 pseudogenes shared between strain HS and SOPE. Although the symbioses in the current study are anticipated to be relatively recent in origin, comparisons focusing on these shared sequences also show that both S. glossinidius and SOPE have reduced GC-contents relative to strain HS (Figure 8). This effect is most notable at 4-fold degenerate (GC4) sites in S. glossinidius, which demonstrate the highest levels of sequence divergence and AT-bias in comparison to orthologs from strain HS. Assuming that the onset of AT-bias is coincident with the origin of symbiosis, this further supports the notion that the symbiosis involving -S', glossinidius is more ancient in origin. It is also notable that the number of substitutions at the 2nd codon position sites of pseudogenes (dGC2, Figure 8) is elevated by approximately the same extent (relative to intact genes) in S. glossinidius and SOPE. This implies that pseudogenes have been evolving under relaxed selection for approximately the same proportion of time since each symbiont diverged from strain HS. However, given that sequence divergence at silent sites (GG4) is greater between strain HS and £. glossinidius, this again invokes the interpretation that pseudogenes arose earlier in the S. glossinidius line of descent. It is also interesting to note that the level of divergence at GC2 sites (dGC2, Figure 8) relative to GG4 sites (dGC4, Figure 8) is greater in SOPE than in S. glossinidius. This can be explained by the fact that the pairwise comparison between strain HS and SOPE is expected to capture an increased PLOS Genetics | www.plosgenetics.org 6 proportion of mutations that are fixed in the insect-associated phase of life in which selection on polypeptide evolution is anti­cipated to be more relaxed. Mutational Dynamics of Gene Inactivation in SOPE and 5. glossinidius Considering only those mutations that have led to gene inactivation, we found that the relative ratios of truncating (large) indels, frameshifting (small) indels and nonsense mutations are similar in SOPE and S. glossinidius (Table 2). Inspection of the data reveals that small frameshifting deletions constitute the most abundant class of mutations leading to gene inactivation. However, it should be noted that the effects of large deletions are, for obvious reasons, not captured in our analyses. Another important point is that IS-element insertions appear to have contributed relatively little to the overall spectrum of mutations leading to gene inactivation in SOPE, representing only 10% of the total count. Indeed, the majority of IS-elements in SOPE are located either in intergenic regions or, more commonly, clustered inside other IS-elements. One potential explanation is that IS-element insertions in genic sequences might be more deleterious towards processes of transcription an d /o r translation in the cell, such that pseudogenes with IS-element insertions are preferentially deleted relative to pseudogenes with nonsense point mutations or small indels. However, it is conspicuous that clustering of IS-elements has also been reported for mobile DNA elements found in eukaryotes, including the MITE elements found in plants [32] and mosquitoes [33], and the Alu and LI elements found in the human genome [34]. The relative paucity of IS-elements in genic DNA is surprising given the fact that the SOPE genome has such a large number of pseudogenes that provide neutral space for IS-element colonization. However, the inability of IS-elements to occupy this territory can be rationalized as a consequence of an inherited adaptive bias that facilitates the avoidance of genic insertion. This makes sense when considering the perspective of an IS-element residing in a free-living bacterium that has relatively few dispensable genes. It also explains the propensity for IS-elements to insert themselves into the sequences of other IS-elements, because the safety of this approach has already been validated by natural selection. Clearly, in the case of SOPE, when the opportunity arose for expansion into novel territory (i.e. neutralized genic sequences), IS-elements were largely unable to overcome these basic evolutionary directives. Discussion Phylogenetic analysis of strain HS indicates that it shares a close relationship with the *Wa/2j-aIlied endosymbionts that are found in a wide range of insect hosts, including tsetse flies, weevils, lice and November 2012 | Volume 8 | Issue 11 | e l002990 12 13 Origins of Mutualistic Insect-Bacterial Symbionts SOPE disrupting mutations / kb (HS ortholog size) 100- o SOPE 0 2 4 6 8 10 12 14 disrupting mutations / kb (current ORF size) s.g. 0 2 4 6 8 10 12 disrupting mutations / kb (HS ortholog size) s.g. 0 2 4 6 8 10 12 14 disrupting mutations / kb (current ORF size) Figure 6. Densities of disrupting mutations in SOPE and S. giossinidiuspseudogenes. The numbers of frameshifting and truncating indels and nonsense mutations were computed from alignments of strain HSf SOPE and 5. giossinidius orthologs. Mutation densities were computed according to the original strain HS ORF sizes (left) or the current SOPE or S. giossinidius pseudogene sizes (right). doi:10.1371 /journa l.pgen.1002990.g006 stinkbugs. In terms of 16S rRNA sequence identity, strain HS is most closely related to endosymbionts found in the chestnut weevil, Curculio sikkimensis and the stinkbug, Cantao occelatus. In­terestingly, only limited numbers of these insects maintain Sodalis-allied endosymbionts in their natural environment [21 23], suggesting that they do not maintain persistent (maternally-transmitted) infections. Furthermore, it is notable that the sequences from strain HS, C. sikkimensis and C. occelatus are localized on very short branches in our phylogenetic trees, indicating that these particular lineages are evolving slowly in comparison to other <SWafo-allied endosymbionts. This low rate of molecular sequence evolution, along with the observation that the strain HS genome shows no sign of the characteristic degenerative changes that are known to accompany the transition to the obligate host-associated lifestyle, leads us to propose that strain HS represents an environmental progenitor of the &<afo/w-allied clade of insect endosymbionts. Closely related members of the *Wa&y-allied clade of insect endosymbionts have now been identified in a wide range of distandy related insect taxa, including some that are known to feed exclusively on plants and others that are known to feed exclusively on animals [8]. Although strain HS was isolated from the wound of a human host, it is difficult to assess the extent of its pathogenic capabilities, due to the fact that antibiotic treatment commenced three days prior to microscopic examination and culturing. In addition, the available evidence indicates that the original source of the infection was a branch from a dead crab apple tree. This implies that strain HS was present either on the bark or in the woody tissue of this tree, possibly acting as a pathogen or saprophyte. Furthermore, it is interesting to note that C. sikkimensis and C. occelatus, whose symbionts are most closely related to strain HS, are both known to feed on trees [35,36]. In addition, some wood and bark-inhabiting longhorn beedes, including Tetropium castaneum (Figure 1) have recendy been found to maintain Sodalis-allied endosymbionts [37]. Moreover, the ability of strain HS to persist in both plant and animal tissues is compatible with the observation that diverse representatives of both herbivorous and carnivorous insects have acquired &><f<2&-allied symbionts. In a comparative sense, relationships involving the Sodalis-allied endosymbionts are considered to be relatively recent in origin. Indeed, evidence of host-symbiont co-speciation only exists in the case of grain weevils, Sitophilus spp., which were estimated to have co-evolved with their -SWafo-allied endosymbionts for a period of around 20 MY, following the replacement of a more ancient lineage of endosymbionts in these insects [38,39]. The notion of a recent origin of the ^oafa/w-allied endosymbionts is further supported by the fact that the whole genome sequence of S. giossinidius is substantially larger than that of long-established mutualistic insect endosymbionts, and is close to the size of related free-living bacteria [24]. However, the S. giossinidius genome does have an unusually low coding capacity resulting from the presence of a large number of pseudogenes [24,25]. This suggests that S. giossinidius is at an intermediate stage in the process of genome degeneration, in which many protein coding genes have been inactivated by indels and nonsense mutations but have not yet been deleted from the genome. In the current study we show that the genome of the grain weevil symbiont, SOPE, is at a similar stage of degeneration as evidenced by the presence of a comparable number of pseudogenes and a large number of repetitive insertion sequence elements. In a comparative sense, it is interesting to note that SOPE and strain HS share a substantially higher level of sequence similarity, genome-wide, in comparison to S. giossinidius and strain HS (Figure 2). In the context of the progenitor hypothesis, the PLOS Genetics | www.plosgenetics.org November 2012 | Volume 8 | Issue 11 | e l002990 14 15 Origins of Mutualistic Insect-Bacterial Symbionts SOPE . HS GC2 = 42.1 GC2 = 42.8 GC2 = 41.9 GC4 = 81.8 GC4 = 82.4 GC4 = 78.3 GCI =49.3 GCI = 49.7 GCI = 48.0 - S. glossinidius dGC2 = 0.008 dGC2M, = 0.013 dGC4 = 0.043 dGC4^ = 0.049 dGC2 = 0.023 dGC2M, = 0.044 dGC4 = 0.281 dGC4ip = 0.313 Figure 8. Base composition bias and mutation rates observed in pairwise comparisons between strain HS, 5. glossinidius and SOPE. The evolutionary relationships between SOPE, strain HS and S. glossinidius are depicted by bold lines drawn to scale in accordance with levels of genome-wide divergence at 4-fold degenerate (GC4) sites. Upper boxes show genome-wide GC-percentages at 2nd codon position (GC2), GC4 and intergenic (GCI) sites. Lower boxes depict the number of substitutions per site for intact genes (dGC2 and dGC4) and pseudogenes (dGC2xP and dGC4»y). The data were obtained from pairwise analysis of point mutations in 1,355 intact genes and 1,376 pseudogenes shared between strain HS and S. glossinidius, and 1,414 intact genes and 1,194 pseudogenes shared between strain HS and SOPE. doi:10.1371 /journa l.pgen.1002990.g008 that irreversible degenerative changes, including gene inactivation and loss, in addition to base composition bias, commence rapidly following the onset of an obligate relationship. Indeed, the close relationship observed between strain HS and SOPE illustrates the potency of the degenerative evolutionary process at an early stage in the evolution of a symbiotic interaction. This is exemplified by the fact that SOPE is predicted to have lost 55% of its ancestral gene inventory (34% via gene loss and 21% via gene inactivation) in a period of time sufficient to incur a substitution frequency of only 4.3% at the highly variable GC4 sites of intact protein coding genes (Figure 8). Although estimates of genome wide synonymous clock rates vary by several orders of magnitude in bacteria [40], an estimate of fJ.& = 2 .2X 10-7, derived recently for another insect endosymbiont, Buchnera aphidicola [41], places the divergence of strain HS and SOPE at only c. 28,000 years, which is much more recent than previous estimates obtained for the origin of the SOPE symbiosis [38,39]. While the broad distribution of recendy derived endosymbionts in phylogenetically distant insect hosts has previously been attributed to interspecific symbiont transfer events [10,11], the results outlined in the current study indicate that diverse insect species can also acquire novel symbionts through the domestica­tion of bacteria that reside in their local environment. In the case of S. glossinidius and SOPE, our comparative analyses support the notion that these symbionts were acquired independendy, as evidenced by the presence of distinct mutations in shared pseudogenes. This also implies that symbionts rapidly become specialized towards a given host, likely restricting their abilities to switch hosts. Although the current study highlights the first description of a close free-living relative of the iSWafo-allied symbionts, it should be noted that environmental microbial diversity is vasdy undersampled [42]. Thus, it is conceivable that close relatives of extant insect endosymbionts, such as strain HS, are widespread in nature and provide ongoing opportunities for a wide range of insect hosts to domesticate new symbiotic associates. Furthermore, since many insects serve as vectors for plant and animal pathogens [43], it is conceivable that mutualistic associ­ations arise as a consequence of the domestication of vectored pathogens. This hypothesis is compelling because such pathogens are not expected to negatively impact the fitness of their insect vectors [44] and under those circumstances the transition to a mutualistic lifestyle could be achieved without any need to attenuate virulence towards the insect host. Materials and Methods Isolation and Phylogenetic Analysis of Strain HS Strain HS was isolated on MacConkey agar at 35°G and 5% CO2. 16S rRNA and groEL sequences were amplified from strain HS using universal primers. Following cloning of PCR products, eight clones were sequenced from each gene and consensus sequences were used in phylogenetic analyses. Sequence align­ments were generated for 16S rRNA and groEL using MUSCLE [45]. PhyML [46] was then used to construct phylogenetic trees using the HKY85 [47] model of sequence evolution with 25 random starting trees and 100 bootstrap replicates. T a b le 2. Allelic spectrum of pseudogene mutations in strain HS orthologs found in SOPE and 5. glossinidius. Disrupting Internal Internal Nonsense Pseudogenes mutations insertions deletions 5 deletions 3 deletions mutations3 IS elements SOPE (1194) 2249 19% 34% 7% 11% 19% 10% S. glossinidius (1376) 4316 13% 40% 12% 19% 16% Numbers in parentheses indicate the number of pseudogenes of strain HS orthologs found in the SOPE or S. glossinidius genome sequences. Nonsense mutations are classified as point mutations that catalyze the incorporation of a premature stop codon in the reading frame of a strain HS ortholog, independent of the presence of any frameshift resulting from an indel. aNonsense mutations are classified as point mutations that catalyze the incorporation of a premature stop codon in the reading frame of an HS ortholog, independent of the presence of any frameshifting indel. doi:10.1371/journal.pgen.1002990.t002 PLOS Genetics | www.plosgenetics.org 10 November 2012 | Volume 8 | Issue 11 | e l002990 16 Weevil Cultures and DNA Isolation Synchronous cultures of Sitophilus oiyzae and Sitophilus zeamais were reared on organic soft white wheat grains and corn kernels respectively, and maintained at 25°C with 70% relative humidity. Bacteriomes (containing the bacterial endosymbionts SOPE and SZPE) were isolated from 5th instar S. oiyzae and -S', zeamais larvae by dissection and homogenized at a sub-cellular level to release bacteria from host bacteriocyte cells; bacterial cells were then separated from host cells via centrifugation (2,000 xg, 5 min). Total genomic DNA was then isolated from bacteria using the Qiagen DNeasy Blood & Tissue Kit (Qiagen, Valencia, CA). SOPE and SZPE Shotgun Library Construction Six mg of genomic DNA was hydrodynamically sheared in 5 mM Tris, 1 mM EDTA, 100 mM NaCl (pH 8) buffer to a mean fragment size of 10 kb. The sample was washed and concentrated by ultrafiltration in a Centricon-100 (Millipore, Billerica, MA) and eluted in 250 |il of 2 mM Tris (pH 8). The fragments were end-repaired by treatment with T4 DNA polymerase (New England Biolab, Beverly, MA) to generate blunt ends. The DNA was then extracted with phenol/chloroform, ethanol precipitated, and 5f phosphorylated with T4 polynucleotide kinase (NEB). Ten mM of double-stranded, biotinylated oligonucleotide adaptors were blunt-end ligated onto the sheared genomic fragments at room temperature for 25 h using 10,000 cohesive end units of high concentration T4 DNA ligase (NEB). Unligated adaptors were removed by ultrafiltration in a Centricon-100. The adaptored fragments were bound to streptavidin-coated magnetic beads (Invitrogen), and after binding and washing, the adaptored genomic fragments were eluted in 10 mM TE (pH 8). Fragments in the 9.5 11.5 kb size range were gel purified after separation on a 0.7% l x TAE agarose gel, and the purified DNA was electroeluted from the agarose and desalted by ultrafiltration in a Centricon-100. SOPE and SZPE Shotgun Sequencing pWD42 vector (GenBank: AF129072.1) was linearized by digestion with BamHl (NEB) at 37°C for 4 h, extracted with phenol/chloroform, ethanol precipitated and resuspended in 100 ml of 2 mM Tris (pH 8.0). Ten picomoles of double-stranded, biotinylated oligo adaptors were ligated onto the i?amHI-digested vector at 25°C for 16 hrs using 4,000 units of T4 DNA ligase (NEB). Unligated adaptors were removed by ultrafiltration in a Centricon-100. The adaptored vector was bound to streptavidin-coated magnetic beads and the non-biotinylated adaptored vector was eluted in 10 mM TE (pH 8). One hundred ng each of adaptored vector and genomic DNA were annealed without ligase in 10 ml of T4 DNA ligase buffer (NEB) at 25°C for one hour. Two ml aliquots of the annealed vector/insert were transformed into 100 ml of XL-10 chemically competent E. coli cells (Agilent Technologies, Santa Lara, CA) and plated on LB agar plates containing 20 [ig/ml ampicillin. A total of 23,808 bacterial colonies were picked into 96-well microtiter dishes containing 600 ml of terrific broth (TB)+20 Jig/ml ampicillin and grown at 30°C for 16 h. Fifty ml aliquots were removed from the library cultures, mixed with 50 ml of 14% DMSO, and archived at - 80°C. The 200 ml cultures were diluted 1:4 in TB amp and runaway plasmid replication was induced at 42°C for 2.25 h. Plasmid DNA was purified by alkaline lysis, and cycle sequencing reactions were performed with forward and reverse sequencing primers using ABI BigDye v3.1 Terminator chemistry (Applied Biosystems, Foster City, CA). The reactions were ethanol precipitated, resuspended in 15 ul ofdH20 , and sequence ladders PLOS Genetics | www.plosgenetics.org 1 were resolved on an ABI 3730 capillary instrument prepared with POP-5 capillary gel matrix. SOPE Genome Sequence Assembly and Finishing Following elimination of any sequences encoding contaminating plasmid vector or host insect sequences, 38,755 shotgun reads were assembled using the Phusion assembler [48] using the paired-end sequences as mate-pair assembly constraints. Contig assem­blies were viewed and edited in Consed [49], and reads with high quality (Phred>20) discrepancies were disassembled. After inspection and manual assembly to extend contigs, gaps were closed by iterative primer walking (895 primer walk sequence reads) and gamma-delta transposon-mediated full-insert sequenc­ing of plasmid clones (6,165 sequence reads across 103 transposed plasmid clones) using an established protocol [50]. The average insert size of the plasmid library in the finished SOPE assembly was found to be 8.2 kb. SOPE Fosmid Library Construction and Genome Sequence Validation The SOPE fosmid library was constructed using the Epicenter EpiFOS Fosmid Library Production Kit (Epicentre Biotechnolo­gies, Madison, WI), using SOPE total genomic DNA. 1,404 paired-end reads were generated from 702 fosmid inserts and mapped onto the assembly derived from the plasmid shotgun sequencing for validation (Figure S2). Strain HS Sequencing Strain HS genomic DNA was isolated from liquid culture using the Qiagen DNeasy Blood & Tissue Kit (Qiagen, Valencia, CA). Five micrograms of total genomic DNA was used to construct a paired-end sequencing library using the Illumina paired-end sample preparation kit (Illumina, Inc. San Diego, CA) with a mean fragment size of 378 base pairs. This library was then sequenced on the Illumina GAIIx platform generating 26,891,485 paired-end reads of 55 bases in length. Strain HS Sequence Assembly Paired-end reads were quality filtered using Galaxy [51,52] and low quality paired-end reads (Phred<20) were discarded. The remaining 17,054,405 reads were then assembled using Velvet [53] with a k-mer value of 37, with expected coverage of 119 and a coverage cutoff value of 0.296. The resulting assembly consisted of 271 contigs with an N50 size of 231,573 and a total of 5,135,297 bases. No sequences were found to share significant sequence identity with genes encoding plasmid replication functions, suggesting that strain HS does not maintain any extrachromo-somal elements. Strain HS Annotation The assembled draft genome sequence of strain HS was annotated by automated ORF prediction using GeneMarkhmm [54]. The annotation was then adjusted manually in Artemis [55] using the published Sodalis glossinidius genome sequence [25] as a guide. ORFs were annotated as putatively functional only if (i) their size was ^90% of the most closely related ORF derived from a free-living bacterium in the GenBank database, and (ii) they did not contain any frameshifting indel(s). Genome Alignment, COG Classification, and Computational Analyses Curation of the strain HS genome sequence was performed in Artemis [55]. ORFs were classified into COG categories using the Origins of Mutualistic Insect-Bacterial Symbionts 1 November 2012 | Volume 8 | Issue 11 | e l002990 17 Origins of Mutualistic Insect-Bacterial Symbionts Cognitor software [56]. Syntenic links shown in Figure 2 were determined by pairwise nucleotide alignments between strain HS con tigs and S. glossinidius (GenBank: NC_007712.1) or the finished SOPE genome using the Smith-Waterman algorithm as imple­mented in the cross_match algorithm [49]. Figure 2 was prepared from data obtained from these alignments using CIRCOS [57]. The metrics depicted in Table 1, Table 2, and Figure 8 were computed from pairwise nucleotide sequence alignments of strain HS, S. glossinidius and SOPE ORFs using custom scripts. Candidate genes were classified as intact orthologs when their alignment spanned >99% of the HS ORF length (or 90% for ORFs <300 nucleotides in size) and did not contain frameshifting indels or premature stop codons. Monte Carlo Pseudogene Simulation A simple Monte Carlo approach was implemented to simulate the evolution of pseudo genes in £. glossinidius and SOPE. The simulation facilitated the progressive accumulation of random mutations in all strain HS orthologs of both intact genes and pseudo genes identified in the current S. glossinidius or SOPE gene inventories. Mutations accumulated in proportion to ORF size in a randomly selected class of neutral genes of user-defined size over a defined number of mutational cycles. At preset cycle intervals, the simulation recorded (i) the difference in size between intact and disrupted sequences, (ii) the number of neutral genes that have accumulated one or more disrupting mutations, and (iii) the density of disrupting mutations, which was calculated based on the cumulative size of all neutral genes. Accession Numbers The GenBank accession numbers for sequences used in Figure 1 are as follows: Endosymbiont of Circulio sikkimensis 16S rRNA, (AB559929.1), groEL, (AB507719); Vibrio cholerae 16S rRNA, (NC_002506.1), groEL, (NC_002506.1); Dickeya dadantii 16S rRNA (CPO020 38.1), groEL, (CP002038.1); Escherichia coli 16S rRNA, (NC_000913.2), groEL, (NC_000913.2); Candidatus Moranella endobia 16S rRNA, (NC_015735), groEL, (NC_015735); Sodalis glossinidius 16S rRNA, (NC_007712.1), groEL, (NC_007712.1); Yersinia pestis 16S rRNA, (NC_008150.1), groEL, (NC_008150.1); Wigglesworthia glossinidia 16S rRNA, (NC_004344.2), groEL, (NC_004344.2); Candidatus Blochmannia pennsylvanicus 16S rRNA, (NC_007292), groEL, (NC_007292); Endosymbiont of Cantao ocellatus 16S rRNA, (AB541010), groEL, (BAJ08314); Endosymbiont of Columbicola columbae 16S rRNA, (AB303387), groEL, (JQ063388); Sitophilus zeamais primary endosymbiont 16S rRNA, (AF548142), groEL (JX444567); Sitophilus oiyzae primary endosymbiont 16S rRNA, (AF548137), groEL (AF005236); Strain HS 16S rRNA, (JX444565), groEL QX444566). The GenBank References 1. Andersson SG, Kurland CG (1998) Reductive evolution of resident genomes. Trends Microbiol 6: 263 268. 2. Dale C, Moran NA (2006) Molecular interactions between bacterial symbionts and their hosts. Cell 126: 453 465. 3. Perez-Brocal V, Gil R, Ramos S, Lamelas A, Postigo M, et a l (2006) A small microbial genome: the end o f a long symbiotic relationship? Science 314: 312 313. 4. McCutcheon JP, Moran NA (2007) Parallel genomic evolution and metabolic interdependence in an ancient symbiosis. Proc Natl Acad Sci U S A 104, 19392 19397. 5. Nakabachi A, Yamashita A, To h H, Ishikawa H, Dunbar HE, et al. (2006) The 160-kilobase genome of the bacterial endosymbiont Carsonella. Science 314: 267. 6. Moran NA, McCutcheon JP, Nakabachi A (2008) Genomics and evolution of heritable bacterial symbionts. Annu Rev Genet 42: 165 190. 7. Novakova E, Hypsa V, Moran NA (2009) Arsenophonus, an emerging clade of intracellular symbionts with a broad host distribution. BMC Microbiol 9: 143. accession numbers for sequences used in Figure 4 are as follows: Strain HS Figure 4A (JX444569), Figure 4B (JX444571), Figure 4C (JX444572); Sitophilus oiyzae primary endosymbiont Figure 4A (JX444568), Figure 4B 0X444570), Figure 4C (JX444573). Supporting Information Figure S I Alignment between strain HS contigs (top) and the chromosome of Dickeya dadantii. The draft strain HS contigs are depicted in an arbitrary color scheme (outer top ring). On the upper track, grey bars depict genes unique to strain HS whereas green bars depict strain HS genes that share orthologs with the aligned D. dadantii chromosome. On the lower track, green and red bars represent intact and disrupted genes (respectively) in the D. dadantii chromosome, and blue bars indicate prophage and IS-element ORFs. (TIF) Figure S2 Genome assembly validation of the 4.5 Mb SOPE genome using plasmid and fosmid mate-pair coverage. The finished SOPE chromosome sequence assembly was obtained by paired-end plasmid shotgun sequencing, primer walking and directed transposon-based full-insert plasmid sequencing. The physical map of the plasmid paired-ends are represented in blue. Clones from underrepresented regions and IS-element clusters were completely sequenced by transposon-mediated sequencing (red; see Materials and Methods). The resulting finished assembly (circular chromosome: 4,513,139 bp) was validated by fosmid paired-end sequencing (orange). Depth of plasmid clone physical coverage is depicted in the histogram (yellow). The locations of the four most abundant families of IS-elements are depicted by the inner bars (red, IS903; blue, IS256; green, IS21; purple, ISL3). (TIF) T able S I List of complete strain HS gene products and status of orthologs in S. glossinidius and SOPE. Candidate virulence genes are highlighted in the column labeled "V", according to the presence of orthologs in animal (A) or plant (P) pathogens. (XLSX) Acknowledgments We thank Jon Scger and three anonymous reviewers for helpful comments. Author Contributions Conceived and designed the experiments: ALC KFO RBW MF CD. Performed the experiments: ALC KFO MG AP DMD ACvN RBW MF CD. Analyzed the data: ALC KFO MG AP DMD ACvN RBW MF CD. Contributed reagents/materials/analysis tools: RBW MF CD. Wrote the paper: ALC KFO RBW MF CD. 8. Snyder AK, McMillen CM, Wallenhorst P, Rio RV (2011) The phylogeny of ^o^flfo-like symbionts as reconstructed using surface-encoding loci. FEMS Microbiol Lett 317:143 151. 9. Huigens ME, de Almeida RP, Boons PA, Luck RF, Stouthamer (2004) Natural interspecific and intraspecific, horizontal transfer of parthenogenesis-inducing Wolbackia in Tritkogramma wasps. Proc Biol Sci 271: 509 515. 10. Jaenike J , Polak M, Fiskin A, Helou M, Minhas M (2007) Interspecific transmission of endosymbiotic Spiroplasma by mites. Biol Lett 3: 23 25. 11. Moran NA, Dunbar HE (2006) Sexual acquisition of beneficial symbionts in aphids. Proc Natl Acad Sci U S A 103: 12803 12806. 12. Hoffmann JA (1995) Innate immunity of insects. C u i t Opin Immunol 7: 4 10. 13. Pontes MH, Smith KL, De Vooght L, Van Den Abbeele J , Dale C (2011) Attenuation of the sensing capabilities of PhoQ_ in transition to obligate insect-bacterial association. PLoSGenet7: el002349. doi: 10.1371/journal, pgen.1002349. 14. Dale C, Young S, Haydon DT, Welburn SC (2001) The insect endosymbiont Sodalis glossinidius utilizes a type-III secretion system for cell invasion. Proc Natl Acad Sci U S A 98: 1883 1888. PLOS Genetics | www.plosgenetics.org 12 November 2012 | Volume 8 | Issue 11 | e1002990 18 Origins of Mutualistic Insect-Bacterial Symbionts 15. Dale C, Plague GR, Wang B, Ochman H, M oran NA (2002) Type III secretion systems and the evolution of mutualistic endosymbiosis. Proc Natl Acad Sci U SA 99: 12397 12402. 16. Degnan PH, Yu Y, Sisneros N, Wing RA, M oran NA (2009) Hamiltonella defensa, genome evolution o f protective bacterial endosymbiont from pathogenic ancestors. Proc Natl Acad Sci U S A 106: 9063 9068. 17. Novakova E, Hypsa V (2007) A new Sodalis lineage from bloodsucking fly Craterina melbae (Diptera, Hippoboscoidea) originated independently o f the tsetse flies symbiont Sodalis glossinidius. FEMS Microbiol Lett 269: 131 135. 18. Darby AC, Choi J-H, Wilkes T, Hughes MA, Weixen JH, et al. (2009) Characteristics of the genome of Arsenopkonus nasoniae, son-killer bacterium of the wasp Nasonia. Insect Mol Biol Suppl 1: 75 89. 19. Stackebrand E, Goebel BM (1994) Taxonomic note: a place for DNA-DNA reassociation and 16S rRNA sequence analysis in the present species definition in bacteriology. Int J Syst Bacteriol 44: 846 849. 20. Verbarg S, Friihling A, Cousin S, Brambilla E, Gronow S, et al. (2008) Biostraticola tofi gen. nov., spec, nov., a novel member of the family Enterobacteriaceae. Curr Microbiol 56: 603 608. 21. Kaiwa N , Hosokawa T , Kikuchi Y, Nikoh N, Meng XY, et al. (2010) Primary gut symbiont and secondary, .SWfl&r-allied symbiont of the Scutellerid stinkbug Cantao ocellatus. Appl Environ Microbiol 76: 3486 3494. 22. Toju H, Hosokawa T, Koga R, Nikoh N, Meng XY, et al. (2009) "Candidates Curculioniphilus buchneri," a novel clade of bacterial endocellular symbionts from weevils of the genus Curculio. Appl Environ Microbiol 76: 275 282­23. Toju H, Fukatsu T (2011) Diversity and infection prevalence of endosymbionts in natural populations of the chestnut weevil: relevance of local climate and host plants. Mol Ecol 20: 853 868. 24. Toh H, Weiss BL, Perkin SA, Yamashita A, Oshima K, et al. (2006) Massive genome erosion and functional adaptations provide insights into the symbiotic lifestyle of Sodalis glossinidius in the tsetse host. Genome Res 16: 149 156. 25. Belda E, Moya A, Bentley S, Silva FJ (2010) Mobile genetic element proliferation and gene inactivation impact over the genome structure and metabolic capabilities o f Sodalis glossinidius, the secondary endosymbiont of tsetse flies. BMC Genomics 11: 449. 26. Nevers P, Saedler H (1977) Transposable genetic elements as agents o f gene instability and chromosomal rearrangements. Nature 268: 109 115. 27. Bickhart DM, Gogarten JP, Lapierre P, Tisa LS, Normand P, et al. (2009) Insertion sequence content reflects genome plasticity in strains o f the root nodule actinobacterium Frankia. BMC Genomics 10: 468. 28. Parkhill J , Berry C (2003) Genomics: Relative pathogenic values. Nature 423: 23 25. 29. Tampakaki AP, Skandalis N, Gazi AD, Bastaki MN, Saixis PF, et al. (2010) Playing the "Harp" : evolution of our understanding of hrpJhrc genes. Annu Rev Phytopathol 48: 347 370. 30. Burke GR, Moran NA (2011) Massive genomic decay in Senatia symbiotica, a recently evolved symbiont of aphids. Genome Biol Evol 3: 195 208. 31. McCutcheon JP, Moran NA (2011) Extreme genome reduction in symbiotic bacteria. Nat Rev Microbiol 10: 13 26. 32. Tarchini R , Biddle P, Wineland R, Tingey S, Rafalski A (2000) The complete sequence of 340 kb of DNA around the rice Adhl adh2 region reveals interrupted colinearity with maize chromosome 4. Plant Cell 12: 381 391. 33. Feschotte C, Mouches C (2000) Recent amplification of miniature inverted-repeat transposable elements in the vector mosquito Culex pipiens characteriza­tion of the Mimo family. Gene 250: 109 116. 34. Jurka J , Kohany O, Pavlicek A, Kapitonov VV, Jurka MV (2004) Duplication, coclustering, and selection of human Alu retrotransposons. Proc Nad Acad S c iU S A 101: 1268 1272. 35. Toju H, Fukatsu T (2011) Diversity and infection prevalence of endosymbionts in natural populations of the chestnut weevil: relevance of local climate and host plants. Mol Ecol 20: 853 868. 36. Kaiwa N, Hosokawa T , Kikuchi Y, Nikoh N, Meng XY, et al. (2010) Primary gut symbiont and secondary, iWa&r-allied symbiont of the Scutellerid stinkbug Cantao ocellatus. Appl Environ Microbiol 76: 3486 3494. 37. Griinwald S, Pilhofer M , Holl W (2010) Microbial associations in gut systems of wood- and bark-inhabiting longhorned beetles [Coleoptera: Cerambycidae\. Syst Appl Microbiol 33: 25 34. 38. Lefevre C, Charles H, Vallier A, Delobel B, Farrell B, et al. (2004) Endosymbiont phylogenesis in the dryophthoridae weevils: evidence for bacterial replacement. Mol Biol Evol 21: 965 973­39. Conord C, Despres L, Vallier A, Balmand S, Miquel C, et al. (2008) Long-term evolutionary stability of bacterial endosymbiosis in curculionoidea: additional evidence of symbiont replacement in the dryophthoridae family. Mol Biol Evol 25: 859 868. 40. Morelli G, Didelot X, Kusecek B, Schwarz S, Bahlawane C, et al. (2010) Microevolution of Helicobacter pylori during prolonged infection of single hosts and within families. PLoS Genet 6: e 1001036. doi: 10.1371/journal.pgen. 1001036. 41. Moran NA, McLaughlin H, Sorek R (2009) The Dynamics and Time Scale of Ongoing Genomic Erosion in Symbiotic Bacteria. Science 323: 379 382. 42. Green J , Bohannan BJ (2006) Spatial scaling of microbial diversity. Trends Ecol Evol 21: 501 507. 43. Nadarasah G, Stavrinides J (2011) Insects as alternative hosts for phytopath-ogenic bacteria. FEMS Microbiol Rev 35: 555 575. 44. Stavrinides J , No A, Ochman H (2010) A single genetic locus in the phytopathogen Pantoea stewartii enables gut colonization and pathogenicity in an insect host Environ Microbiol 12: 147 155. 45. Edgar RC (2004) MUSCLE: multiple sequence alignment with high accuracy and high throughput Nucleic Acids Res 32: 1792 1797. 46. Guindon S, Gascuel O (2001) A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst Biol 52: 696 704. 47. Hasegawa M, Kishino H, Yamo T (1985) Dating of the human-ape splitting by a molecular clock of mitochondrial DNA. J Mol Evol 22: 160 174. 48. Mullikin JC, Ning Z (2003) The phusion assembler. Genome Res 13: 81 90. 49. Gordon D, Abajian C, Green P (1998) Consed: a graphical tool for sequence finishing. Genome Res 8: 195 202. 50. Robb FT, Maeder DL, Brown JR, DiRuggiero J , Stump MD, et al. (2001) Genomic sequence of hyperthermophile, Pyrococcus juriosus'. implications for physiology and enzymology. Methods Enzymol 330: 134 157. 51. Blankenberg D, Von Kuster G, Coraor N, Ananda G, Lazarus R, et al. (2010) Galaxy: a web-based genome analysis tool for experimentalists. C u i t Protoc Mol Biol 19: Unit 19.10.1 21. 52. Goecks J , Nekrutenko A, Taylor J ; Galaxy Team (2010) Galaxy: a comprehensive approach for supporting accessible, reproducible, and transpar­ent computational research in the life sciences. Genome Biol 11: R86. 53. Zerbino DR, Birney E (2008) Velvet: algorithms for de novo short read assembly using de Bruijn graphs. Genome Res 18: 821 829. 54. Lukashin AV, Borodovsky M (1998) GeneMark.hmm: new solutions for gene finding. Nucleic Acids Res 26: 1107 1115. 55. Rutherford K, Parkhill J , Crook J , Horsnell T, Rice P, et al. (2000) Artemis: sequence visualization and annotation. Bioinformatics 16: 944 945. 56. Tatusov RL, Fedorova ND, Jackson JD, Jacobs AR, Kiryutin B, et al. (2003) The COG database: and updated version includes eukaryotes. BMC Bioinfor­matics 4: 41. 57. Krzywinski M, ScheinJ, Birol I, Connors J , Gascoyne R , et al. (2009) Circos: an information aesthetic for comparative genomics. Genome Res 19:1639 1645. PLOS Genetics | www.plosgenetics.org 13 November 2012 | Volume 8 | Issue 11 | e l002990 CHAPTER 3 GENOME DEGENERATION AND ADAPTATION IN A NASCENT STAGE OF SYMBIOSIS Reprinted with permission from Oakeson, K. F. et al. Genome degeneration and adaptation in a nascent stage of symbiosis. Genome Biol Evol 6, 76-93 (2014). 20 GBE Genome Degeneration and Adaptation in a Nascent Stage of Symbiosis Kelly F. Oakeson1'*'1', Rosario Gil2,1, Adam L. Clayton1, Diane M. Dunn3, Andrew C. von Niederhausern3, Cindy Hamil3, Alex Aoyagi3, Brett Duval3, Amanda Baca1, Francisco J. Silva2, Agnes Vallier4, D. Grant Jackson1, Amparo Latorre2,5, Robert B. Weiss3, Abdelaziz Heddi4, Andres Moya2,5, and Colin Dale1 1Department of Biology, University of Utah 2Institut Cavanilles de Biodiversitat i Biologia Evolutiva, Universitat de Valencia, Spain 3Department of Human Genetics, University of Utah 4INSA-Lyon, INRA, UMR203 BF2I, Biologie Fonctionnelle Insectes et Interactions, Villeurbanne, France 5Area de Genomica y Salud, Fundacion para el Fomento de la Investigacion Sanitaria y Biomedica de la Comunitat Valenciana FISABIO - Salud Publica, Valencia, Spain Corresponding author: E-mail: [e-mail]. yThese authors contributed equally to this work. Accepted: December 20, 2013 Data deposition: The Candidatus Sodalis pierantonius str. SOPE genome sequence and annotation has been deposited at GenBank under the accession CP006568. The strain HS chromosome and plasmid sequence and annotation has been deposited at GenBank under the accession CP006569 and CP006570, respectively. Abstract Symbiotic associations between animals and microbes are ubiquitousin nature, with an estimated 15% of all insectspecies harboring intracellular bacterial symbionts. Most bacterial symbionts share many genomic features including small genomes, nucleotide com­position bias, high coding density, and a paucity of mobile DNA, consistent with long-term host association. In this study, we focus on the early stages of genome degeneration in a recently derived insect-bacterial mutualistic intracellular association. We present the complete genome sequence and annotation of Sitophilusoryzae primary endosymbiont (SOPE). We also present the finished genome sequence and annotation of strain HS, a close free-living relative of SOPE and other insect symbionts of the Sodalis-allied clade, whose gene inventory is expected to closely resemble the putative ancestor of this group. Structural, functional, and evolutionary analyses indicate that SOPE has undergone extensive adaptation toward an insect-associated lifestyle in a very short time period. The genome of SOPE is large in size when compared with many ancient bacterial symbionts; however, almost half of the protein-coding genes in SOPE are pseudogenes. There is also evidence for relaxed selection on the remaining intact protein-coding genes. Comparative analyses of the whole-genome sequence of strain HS and SOPE highlight numerous genomic rearrangements, duplications, and deletions facilitated by a recent expansion of insertions sequence elements, some of which appear to have catalyzed adaptive changes. Functional metabolic predictions suggest that SOPE has lost the ability to synthesize several essential amino acids and vitamins. Analyses of the bacterial cell envelope and genes encoding secretion systems suggest that these structures and elements have become simplified in the transition to a mutualistic association. Key words: recent symbiont, degenerative genome evolution, IS elements, pseudogenes, comparative genomics. Introduction Intracellular mutualistic bacteria are notable among cellular life forms because they maintain extremely small genomes. Many examples exist (Nakabachi et al. 2006; P^rez-Brocal et al. 2006; McCutcheon and Moran 2007, 2010; McCutcheon et al. 2009) and the smallest is currently the symbiont of the phloem-feeding insect pest Macrosteles quadrilineatus, "Candidatus Nasuia deltocephalinicola," with a 112 kb genome encoding just 137 protein-coding genes (Bennett and Moran 2013). Such small genomes are derived from a degenerative process that is predicted to take place over sev­eral hundred million years and is accompanied by an increased © The Author(s) 2014. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 76 Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 21 Nascent Stage of Symbiosis GBE rate of DNA and polypeptide sequence evolution (Perez-Brocal et al. 2006), and often a dramatic nucleotide composition bias that results in an increased ratio of adenine and thymine res­idues (Andersson JO and Andersson SGE 1999). Because endosymbiotic bacteria are isolated inside specialized cells (bacteriocytes) within their host, opportunities to engage in parasexual genetic exchange are greatly reduced in compari­son to free-living bacteria. The resulting evolutionary trajectory is therefore characterized by irreversible gene inactivation and loss; a process that is predicted to be accelerated by a reduced efficiency of selection resulting from frequent population bot­tlenecks that reduce the effective population size (Ne) during host reproduction (Moran 1996; Mira et al. 2001; Silva et al. 2003; Schmitz-Esser et al. 2011). Although many highly reduced endosymbiont genomes have now been sequenced and analyzed, little research has focused on recently derived examples and the forces shaping genome evolution in the early stages of an endosymbiotic association. To address this issue, we conducted a compara­tive analysis of the genome sequences of two recently derived insect symbionts, Sitophilus oryzae primary endosymbiont (SOPE) and Sodalis glossinidius (a secondary symbiont of tsetse flies) and a closely related free-living bacterium, desig­nated " strain HS" (Clayton et al. 2012). The characterization of strain HS and related Sodalis-allied insect symbionts re­vealed that genome degeneration is extremely potent in the early stages of a symbiotic association. In the case of SOPE, genome degeneration catalyzed the loss of over 50% of the symbiont gene inventory in a very short period of time (Clayton et al. 2012). Strain HS was discovered as a novel human-infective bac­terium, isolated from a hand wound following impalement with a tree branch. Phylogenetic analyses placed strain HS on a clade comprising the Sodalis-allied insect endosymbionts, including SOPE and So. glossinidius. Preliminary genomic com­parative analyses of the gene inventories of strain HS, SOPE, and So. glossinidius were compatible with the notion that strain HS has a gene inventory resembling a free-living common ancestor that has given rise to mutualistic bacterial symbionts in a wide range of insect hosts (Clayton etal. 2012). In this study, we report the complete genome sequence and annotation of both SOPE and strain HS. We also propose the formal nomenclature Candidatus Sodalis pierantonius str. SOPE to replace the more commonly used name SOPE. Although SOPE shares characteristics with ancient obligate intracellular symbionts, including strict maternal inheritance, residence in bacteriocytes, and nutrient provisioning, it has a relatively large genome with many pseudogenes and mobile genetic elements, consistent with the notion that it is a re­cently derived symbiont. We describe the predicted metabolic capabilities of SOPE and explain how an expansion of insertion sequence (IS) elements has mediated large-scale genomic rearrangements, some of which may be adaptive in nature. Further comparisons between the genomes of SOPE, So. glossinidius, and strain HS shed light on the adaptive changes taking place early in the evolution of insect symbionts. Materials and Methods SOPE Shotgun Library Construction and Sequencing Shotgun library construction and sequencing was performed as described by Clayton et al. (2012), briefly, 60 mg of genomic DNA was sheared to a mean fragment size of 10 kb, end repaired, and adaptors were blunt-end ligated to the frag­ments. Fragments in the size range of 9.5-11.5 kb were gel purified after separation in a 1 % agarose gel. Fragments were ligated into a plasmid vector and transformed into chemically competent Escherichia coli cells. Runaway plasmid replication was induced, and plasmid DNA was purified by alkaline lysis, and cycle sequencing reactions were performed. The reactions were ethanol precipitated, resuspended, and then sequenced on an ABI capillary sequencer. SOPE Genome Sequence Assembly, Finishing, and Validation Genome sequence assembly, finishing, and validation were performed as described by Clayton et al. (2012). Filtered reads were assembled using the Phusion assembler (Mullikin and Ning 2003), and after inspection of the initial contigs, gaps were closed using a combination of iterative primer walking and gamma-delta transposon-mediated full-insert sequencing of plasmid clones. Validation was performed by mapping 1,404 paired-end sequence reads generated from a SOPE fosmid library on to the finished genome assembly. SOPE Genome Annotation The assembled genome sequence of SOPE was submitted to the National Center for Biotechnology Information (NCBI) Prokaryotic Genomes Automatic Annotation Pipeline (PGAAP) for annotation. The resulting candidate open reading frames (ORFs) were then aligned to the HAMAP database (Lima et al. 2009) and classified according to their percent protein identity and length. ORFs that had more than 90% protein identity and more than 8 0% of the length of the database match and did not contain frameshifts or premature stop codons were classified as intact ORFs. The remaining candidate ORFs were then classified as intact or pseudogenes by generating a Blast database from the top HAMAP result for each candidate ORF, then two nucleotide query files were generated: one based on the PGAAP annotation and another including 2,500 nucleotides on either end of the candidate ORF. BlastX searches against the database generated from the top HAMAP result were then performed on each query file, and the output was parsed to search for extended protein matches that indicated either truncated candidate ORFs or possible frameshifted candidate ORFs. The annotation was then aligned to the draft genome sequence and annotation Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 77 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 22 Oakeson etal. GBE of strain HS using the Smith-Waterman algorithm imple­mented in cross_match (Gordon et al. 1998). Custom Perl scripts were used to identify any ORFs not identified by PGAAP as well as refine and classify ORFs as intact or pseu-dogenized. ORFs not identified by PGAPP but spanning more than 9 9 % of the orthologous strain HS ORF or more than 9 0% of ORFs smaller than 300 nucleotides in size were anno­tated as intact only if no inactivating mutation were present. Additional manual curation was preformed with extensive use of the Bacterial Annotation System (BASys) (van Domselaar et al. 2005) and EcoCyc databases (Keseler et al. 2013). IS elements were identified and annotated using ISSaga (Varani et al. 2011). The resulting annotation was also manu­ally curated and adjusted in Artemis (Rutherford et al. 2000). Strain HS Genome Sequence Finishing and Annotation The strain HS draft genome sequence and annotation was generated as described previously (Clayton et al. 2012). To close the genome sequence of strain HS, 16.5 mg of genomic DNA was submitted to Macrogen Inc. (Macrogen, Inc. Seoul, South Korea) for sequencing on the Roche 454 GS-FLX (454 Life Sciences, a Roche company. Branford, CT) platform. A total of 804,816 (543,754 paired-end) reads were gener­ated from a 5-kb insert size mate pair library. These reads along with 34 million paired-end Illumina reads of 55 bases in length were assembled with Newbler 2.7 (Margulies et al. 2005). The resulting assembly consisted of two scaffolds con­taining 47 contigs covering 5.1 Mbp. These gaps were then closed computationally (by incorporating gap filling reads) or by Sanger sequencing of polymerase chain reaction products derived from gaps yielding a closed circular chromosome of 4.7 Mbp and a circular megaplasmid of 449.8 kb. 16S rRNA Mutation Analysis Sequence alignments were generated using MUSCLE (Edgar 2004)forthe 16S rRNAgenesfromstrain HS, SOPE, SZPE, and So. glossinidius. PhyML (Guindon et al. 2010) was then used to construct a phylogenetic tree using the HKY85 (Hasegawa et al. 1985) model of sequence evolution with 25 random starting trees and 100 bootstrap replicates. Classification of mutations in the 16S rRNA stem regions was preformed as described by Pei etal. (2010) and classification of mutations in the entire 16S rRNA sequence was preformed as described by Wuyts (2001). Nucleotide Substitution Rates and Predicted Cryptic Pseudogenes Orthologous genes in SOPE, So. glossinidius, and strain HS were determined using OrthoMCL (Li 2003) with recom­mended parameters (Fischer et al. 2002). Before input into OrthoMCL all pseudogenes, IS elements, and phage se­quences were removed from the sets of SOPE and So. glossi­nidius genes. The output of OrthoMCL was then screened, and any nonorthologous genes or low-quality matches were discarded, and a total of 1,601 strain HS orthologous genes were obtained for SOPE and 1,734 for So. glossinidius. Sequence alignments for each orthologous gene pair was generated using MUSCLE (Edgar 2004). Pairwise estimates of the synonymous (dS) and nonsynonymous (dN) substitution rates were obtained from the YN00 program of the PAML 4.6 package (Yang 2007). The Processing Development Environment (www.processing.org, last accessed January 3, 2014) was used to plot dN and dS for each strain HS-SOPE pairwise comparison and to compute mean ORF sizes. A plot was also generated to compare the dN/dS values of all intact orthologs maintained by SOPE and So. glossinidius. Functional Analysis The Artemis Comparison Tool (Carver etal. 2005) was used to perform a pairwise comparison between the genomes of SOPE, strain HS, and So. glossinidius, to explore conservation of synteny, and to help in the identification of orthologous genes and pseudogenes, to identify similarities and discrepan­cies in the functional capabilities of these organisms. The rean­notated genome of So. glossinidius was used in this comparison (Belda et al. 2010). Metabolic capabilities were analyzed with Blast2Go (Conesa et al. 2005) and KAAS (Moriya et al. 2007) programs and manually curated. Functional information was retrieved from the BioCyc (Caspi et al. 2010), KEGG (Ogata et al. 1999), and BRENDA (Scheer et al. 2011) databases and extensive literature searches. Genomic Rearrangements Between SOPE and Strain HS To identify all genomic rearrangements between SOPE and strain HS, we performed a fully recursive search of the SOPE genome using all 20-mers derived from the complete strain HS genome sequence. Both the search and subsequent data plot­ting were performed in the Processing Development Environment (www.processing.org, last accessed January 3, 2014). The consensus FtsK orienting polar sequences (KOPS) site used for the plot in figure 2 was GGGNAGGG and the dif (the chromosomal site where the XerCD recombinase deca-tenates and resolves chromosome dimers) site is AGTACGCAT AATACATATTATGTTAAAT. Rendering Genomic Features Scalar diagrams of 1) two chromosomal clusters containing large numbers of IS elements and 2) regions encoding type III secretion systems (TTSS) in So. glossinidius,SOPE, and strain HS were rendered in the Processing Development Environment (www.processing.org, lastaccessedJanuary3, 2014). Data Availability The Candidatus Sodalis pierantonius str. SOPE genome se­quence and annotation was deposited in GenBank under 78 Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 23 Nascent Stage of Symbiosis GBE the accession number CP006568. The strain HS chromosome and plasmid sequence and annotation were deposited in GenBank under the accession numbers CP006569 and CP006570, respectively. Results General Features of the SOPE and Strain HS Genome Sequences SOPE is an intracellular, bacteriome-associated symbiont that resides in host bacteriocytes (fig. 1) that has a genome con­sisting of one circular chromosome of 4,513,140 bases with an average GC content of 56.06%. A total of 4,080 candidate protein-coding sequences (CDSs) were annotated of which 2,309 (56.6%) are predicted to be intact based on the absence of frame shift mutations, premature stop codons, or truncating deletions, whereas 1,771 (43.4%) candidate CDSs are predicted to be pseudogenes maintaining one or more these mutations. Mutations were identified by aligning 2,731 homologous CDSs shared between SOPE and strain HS, excluding mobile genetic elements such as integrated pro­phage islands and IS elements. Since the gene inventory of SOPE is known to be a subset of strain HS, and the sequences of strain HS and SOPE are very closely related (having only ~ 2% synonymous divergence genome wide [Clayton et al. 2012]), the genome sequence of strain HS provided a unique opportunity to accurately identify all the mutations leading to predicted ORF inactivations in SOPE. The complete genome sequence of strain HS consists of a circular chromosome of 4,709,528 bases and one mega plas­mid of 449,897 bases. The average GC content of the chro­mosome is 57.47%, and the plasmid GC content is 53.22%. There are a total of 3,993 CDSs encoded on the chromosome with only 61 pseudogenes and 365 CDSs encoded on the mega-plasmid with 14 predicted pseudogenes. Pseudogenes in strain HS were identified based on alignments of the CDSs with homologs from closely related bacteria in the NCBI database. helpers of transposition that are encoded within 804 IS ele­ments. This expansion consists of four major IS elements, ISSoEn 1 to 4 (previously described as ISsope1 to 4) (Gil Garcia et al. 2008), belonging to the IS families named IS5 (ssgr IS903), IS256, IS21, and ISL3, respectively. These four families constitute 795 of the 804 total IS element CDSs in the genome. Within each family, the percent nucleotide iden­tity was greater than 94%, indicating recent expansion within the SOPE chromosome, or maintenance of a high level of se­quence identity through gene conversion. Table 1 summarizes the number of intact and disrupted copies of each of the main IS types. The remaining nine IS elements consist of six copies of the ISPlu15 family, one copy of the IS418 family, and five copies of a Mu-like transposase. Extensive expansions of IS elements have been docu­mented in a number of bacteria undergoing lifestyle transi­tions (Parkhill et al. 2003; Moran and Plague 2004), implying that they are a typical component of the process of degener­ative evolution. Indeed, it has been proposed that such IS An Epidemic of IS Element Expansion in SOPE The genome of SOPE is notable because it has undergone a massive expansion of bacterial IS elements, which accounts for a total of 0.83 Mbp (18%) of the chromosome. The genome contains a total of 822 CDSs encoding either transposases or Fig. 1.-Microscopic images of SOPE. The main panel shows cells from a 5th instar bacteriome of SOPE stained with FM4-64 (red) and DAPI (blue). Rod-shaped bacteria are densely packed into the cytoplasm of the insect cells, whose nuclei display extensive DAPI staining. The inset panel shows isolated SOPE cells stained with DAPI, illustrating their fila­mentous morphology. Table 1 Summary of the Four Major IS Elements in the SOPE Genome ISSoEnl ISSoEn2 ISSoEn3 (Transposase) ISSoEn3 (Helper of Transposition) ISSoEn4 IS family IS5 (ssgr IS903) IS256 IS21 IS21 ISL3 IS elements with intact ORFs 189 104 49 61 9 IS elements with disrupted ORFs 122 160 68 38 10 Total 311 264 117 99 19 Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 79 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 24 Oakeson etal. GBE Fig. 2.-Whole-genome sequence alignment of strain HS and SOPE. Bezier curves highlight regions of synteny shared between strain HS and SOPE. Uninterrupted blocks of synteny are rendered in the same hue. Matches occurring on the same DNA strand are rendered in the purple spectrum, whereas those occurring on different strands are rendered in the yellow spectrum. Thus, to maintain replicational symmetry, matches highlighted in purple should remain on the same replichore, whereas those highlighted in yellow should switch replichore. Sequences shared between the strain HS megaplasmid and SOPE chromosome are rendered in gray. The outer plots represent GC skew with positive skew depicted in red and negative skew depicted in blue. The strand-specific locations of KOPS (FtsK orienting polar sequences) sites are shown as tickmarks overlaid on the plot of GC skew. Inside the plots, the pink and blue tick marks correspond to the positions of phage and IS element sequences, respectively. These and any other forms of repetitive DNA were masked in the generation of the alignment. The location of the dif (terminus) sequence in strain HS is highlighted and intersects with the switch in GC skew, as expected. element proliferations take place as a consequence of the imposition of relaxed selection on a large number of genes (Moran and Plague 2004; Plague et al. 2008), facilitating the expansion of IS elements into genomic space encompassing genes evolving under relaxed selection. However, in the case of SOPE, only a small proportion of IS elements were found to occupy genic sequences with the majority of these elements clustering in intergenic regions (fig. 2). Despite the high level of nucleotide identity within IS families, the sequence strategy facilitated assembly of chromosomal regions harboring both dense IS clusters and genome duplications (Clayton et al. 2012). Examples of two IS-dense intergenic clusters in the SOPE genome are depicted in figure 3. We previously ratio­nalized the clustering of IS elements on the basis that it might be favored by natural selection to avoid the interruption of vital genic sequences (Clayton et al. 2012). At face value, this appears to contradict the notion that IS element expansions occur as a direct consequence of the emergence of neutral space. However, the expansion of IS elements into just a small proportion of neutral space might be sufficient to precipitate an epidemic of activity, as a simple consequence of increasing IS element copy number. Indeed, it has been shown that IS element transposases have the capability to act in trans (Derbyshire et al. 1990; Derbyshire and Grindley 1996), such that a transposase derived from one element could cat­alyze the transposition of other elements in the genome. To this end it is interesting to note that many of the transposase genes in the IS elements of SOPE are pseudogenized. Although this could be taken as a sign that the epidemic of transposition is waning, it is also conceivable that those IS elements with inactive transposases are still being mobilized in trans. Similar to many nonhost-associated bacteria, strain HS has very few IS elements (Wagner et al. 2007). The existing pre­dicted transposase CDSs are either small fragments of IS ele­ments or ORFs that have been disrupted by inactivating mutations. Strain HS and SOPE contain homologs of the ISNCY ssgr ISPlu15 family IS element; however, this element only accounts for a very small fraction of the high number of IS elements in the SOPE genome. Of the four major IS element families in SOPE, none are found in the strain HS genome, but So. glossinidius does maintain the IS element belonging to the IS5 family (named ISSgll in this species; Belda et al. 2010), which is the most abundant in SOPE. Thus, the IS5 element may have been present in the last common ancestor of all three bacteria and then subsequently lost in strain HS, or it 80 Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 25 Nascent Stage of Symbiosis GBE Fig. 3.-IS element dense regions in SOPE. Scalar illustration of two IS element dense regions in the SOPE chromosome. The top row corresponds to the region encompassing SOPEG_ps0144-SOPEG_0160, and the bottom row corresponds to SOPEG_2674-SOPEG_2683. IS element ORFs are colored accord­ing to their family (see key). ORFs are shaded in accordance with their positional synteny in comparison with a full length HS ortholog. The full spectrum of shading (5'-3') is depicted in the key. ORFs disrupted by an IS element insertion are connected by curved lines. All non-IS element ORFs are labeled with their SOPE locus tags. is also possible that SOPE and So. glossinidius independently acquired this element. Evidence for Extensive Recent Intragenomic Rearrangements in SOPE Although strain HS and SOPE share a very high level of se­quence identity, consistent with the notion of recent common ancestry (Clayton et al. 2012), a genome wide alignment of homologous sequences in strain HS and SOPE revealed a sur­prisingly low level of genome-wide synteny (fig. 2). Although this lack of synteny could be explained as a consequence of rearrangements in either lineage, there are two compelling lines of evidence indicating that the rearrangements have pre­dominantly taken place in SOPE. First, it is notable that the majority of rearranged regions in SOPE are bounded by IS elements in SOPE (fig. 2), and IS elements have been impli­cated previously in driving intragenomic rearrangements in other endosymbionts and obligatory intracellular pathogens (Song et al. 2010). Second, SOPE, but not strain HS, has a highly disrupted (nonpolarized) pattern of GC skew that is atypical among prokaryotic genomes (fig. 2) (Francino and Ochman 1997; Frank and Lobry 1999) (Rocha 2004). Such perturbations in GC skew are expected to arise when rearran­gements occur that violate the conservation of strand-specific replicational symmetry. These perturbations can be visualized in figure 2, where the color scheme highlights rearrangements involving strand switching. In addition to perturbing GC skew, figure 2 also shows that the intragenomic rearrangements in the SOPE chromosome have disrupted the distribution of FtsK orienting polarized sequence (KOPS) motifs (Bigot et al. 2005; Levy et al. 2005). KOPS sites are short DNA sequences (GGGNAGGG) that are polarized from the replication origin to the dif site on the leading strands of the chromosome and serve to direct FtsK translocation of chromosomal DNA to daughter cells at the septum during chromosome replication and cell division. As is the case for GC skew, the distribution and strand bias of KOPS in strain HS is typical. It is also note­worthy that the SOPE genome has lost a portion of the ter­minus region of the chromosome that contains the dif site (which is clearly identifiable in strain HS). The dif site is recog­nized by the Xer recombination system that facilitates the resolution of concatenated chromosomes that are generated during replication (Carnoy and Roten 2009). It should be noted that SOPE shares the same morphology as dif mutants in E. coli (fig. 1), which are characterized by cells that form long filaments (Kuempel et al. 1991; Blakely et al. 1993). These mutants are unable to decatenate interlocked chromo­somes resulting from recombination between chromosome copies during replication (Kuempel et al. 1991). Thus, not only is there evidence that SOPE has undergone a large number of recent genomic rearrangements but it also seems likely that these rearrangements have had a deleterious impact upon the replication system. Gene Duplication Events in SOPE In addition to genome wide rearrangements, IS elements appear to have mediated partial genome duplications in SOPE. We detected a total of seven duplicated chromosomal regions comprising more than one CDS, ranging in size from 2.5 to 23.8 kb (supplementary file S1, Supplementary Material online). The most striking duplication in SOPE is 13,476 bases in length and encompasses the genes encoding the molecular chaperone GroEL and its cochaperone GroES as well as the adjacent genesyjdC, dipZ, cutA, aspA, fxsA, yjel, yjeK, and efp. Nucleotide alignments of the whole duplicated region show that the two copies are 99.9% identical, differing by only four single base indels and 13 nucleotide substitutions, indicating that the duplication took place recently. Notably, both copies of groEL and groES maintain intact ORFs and are therefore predicted to be functional. The duplicated regions are bounded by IS256 elements, implying that an IS element-mediated recombination event catalyzed the duplication. In other mutualistic symbionts, including SOPE, groEL and groES have been shown to be expressed at very high levels to compensate for the presence of aberrant polypeptides and/ or the absence of alternative repair pathways that function to rescue misfolded proteins (Ishikawa 1984; Moran 1996; Charles et al. 1997; Fares et al. 2004; Vinuelas et al. 2007; Stoll et al. 2009). Because SOPE has a very large complement of disrupted genes that are expected to yield truncated poly­peptides with folding constraints, we hypothesize that the duplication of the groEL region facilitated an adaptive benefit. Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 81 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 26 Oakeson etal. GBE This could also be true for other genomic duplications that have taken place in SOPE, although the nature of such bene­fits is not immediately obvious when considering the genes involved (supplementary file S1, Supplementary Material online). Evolution of Ribosomal RNA Genes in the Transition to Symbiosis Although the numbers of ribosomal RNA (rRNA) operons in bacteria can reach as many as 15 copies, it has been noted that many long-established primary endosymbionts maintain only one or two operon copies (Moran et al. 2008). The number of rRNA operons in bacteria has been shown to in­fluence growth rate (Stevenson and Schmidt 2004) and the ability to respond quickly to nutrient availability (Klappenbach et al. 2000). Neither of these traits is expected to be of great value for insect symbionts due to the fact that they inhabit a relatively static, competition-free environment. The genome of strain HS was found to maintain seven rRNA operons in total comprising the 16S, 23S, and 5S rRNA, with two operons maintaining an additional copy of the 5S rRNA gene. In contrast, the SOPE genome was found to maintain only two complete rRNA operons, along with three additional complete copies and a partial copy of the 16S rRNA gene. Not surprisingly, one of the complete rRNA operon copies maintains an intergenic tRNAGlu, whereas the other encodes tRNAIle and tRNAAla, ensuring that all three rRNA operon-associated tRNAs have been retained. Although the retention of the three isolated copies of 16S rRNA is intriguing and may have some cryptic adaptive value, it is equally conceivable that it simply reflects stochastic events inherent in the process of genome degeneration. To this end, it is notable that all the rRNA genes in SOPE occupy positions in the genome that correspond contextually to the positions of the complete rRNA operons in strain HS. Thus, it appears that the isolated copies of 16S rRNA resulted from deletion events, rather than gene duplications. Although many bacteria maintain near-identical copies of their rRNA genes as a consequence of gene conversion (Vetrovsky and Baldrian 2013), it was shown previously that SOPE maintains unusually divergent copies, presumably re­flecting a loss of this activity (Dale et al. 2003). In addition, rRNA genes typically evolve at a very low rate, due to the fact that their sequences are highly constrained by structure and function. In a previous study, we noted that the level of se­quence divergence between the 16S rRNA genes of SOPE and strain HS was unexpectedly high in comparison to the level of pairwise sequence identity observed between orthologous protein-coding genes in these bacteria (Clayton et al. 2012). Thus, we elected to further investigate the nature of muta­tions in the 16S rRNA genes of SOPE along with (for context), its sister species, the primary endosymbiont of the maize weevil S. zeamais (SZPE), and the closely related insect symbiont So. giossinidius. This was achieved by classifying mu­tations in the context of a 16S rRNA secondary structure model (Pei et al. 2010). This analysis facilitated the classifica­tion of mutations in stem regions of the 16S rRNA molecule as either structurally conservative or structurally disruptive. The results showed that both SOPE and SZPE have an unusually high ratio of disruptive to conservative mutations in their 16S rRNA genes, relative to strain HS and So. giossinidius (supple­mentary file S2, Supplementary Material online). W e then per­formed a second analysis using the 16S rRNA variability map that was derived from a large number of bacterial species (Wuyts 2001). This analysis facilitated the classification of mu­tations in the entire 16S rRNA molecule according to rarity. Conspicuously, the 16S rRNA genes of SOPE and SZPE were found to maintain a high number of substitutions at sites that typically display low variability. Taken together, these results indicate that the 16S rRNA genes in SOPE and SZPE are evolv­ing under relaxed functional constraints, despite the fact that these symbionts have a recent symbiotic origin. Nucleotide Substitution Rates and Prediction of Cryptic Pseudogenes The number of nonsynonymous substitutions per nonsynon-ymous site (dN) and the number of synonymous substitutions per synonymous site (dS) values were calculated for 1,602 orthologous genes in strain HS and SOPE. The graph of dN versus dS depicts each gene as an individual point, with the radius of each point proportional to ORF size (fig. 4). The gene with the highest dN value encodes a predicted ankyrin repeat domain protein. Ankyrin repeat domain proteins in eukaryotes have been shown to function in protein-protein interactions (Sedgwick and Smerdon 1999), and it has been hypothesized that ankyrin repeat domain proteins have a role in the cyto­plasmic incompatibility generated by Wolbachia in its insect host (Tram and Sullivan 2002; Tram et al. 2003). Analysis of the plot of dN versus dS revealed that genes with the highest dN/dS ratios were smaller in size. The number of genes with a dN/dS ratio > 0.3 (plotted in red) (supplementary file S3, Supplementary Material online) is ap­proximately equivalent to the number of genes predicted to be " cryptic pseudogenes" in a previous study (Clayton et al. 2012). The mean ORF size of these genes is significantly smaller than those with a dN/dS ratio less than 0.3 (plotted in green). This size difference supports the notion that the genes with a dN/dS ratio greater than 0.3 are a subset of genes evolving under relaxed selection that have not yet been disrupted by a mutation (Clayton et al. 2012). Of course due to the extremely close relationship between strain HS and SOPE, some estimates of dN and dS (especially from genes of small size) yield relatively large standard errors (supplementary file S3, Supplementary Material online). Thus, the results presented in this study should be taken 82 Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 27 Nascent Stage of Symbiosis GBE Fig. 4.-Plot of dN versus dS of orthologous genes in strain HS and SOPE. (A) Plot of dN versus dS of 1,601 orthologous genes in strain HS and SOPE. Each point depicted on the plot represents a single gene with the radius of each point proportional to gene size. Mean dN/dS ratio is plotted as a gray dotted line. Genes with a dN/dS ratio greater than 0.4 are annotated with their product. Genes with a dN/dS ratio greater than 0.3 are depicted in red and those genes with dN/dS ratios less than 0.3 are depicted in green. Mean ORF size was calculated for genes with dN/dS greater than 0.3 (red) and dN/dS less than 0.3 (green). (B) Plot of SOPE dN/dS versus Sodalis glossinidius dN/dS for 1,229 orthologous genes in strain HS, SOPE, and So. glossinidius. Each point on the plot represents a single orthologous gene. with " a grain of salt" and not considered to provide a de­finitive inventory of cryptic pseudogenes in SOPE. Functional Predictions of the Protein-Coding Genes in SOPE Genetic Machinery The predicted protein-coding gene inventory of SOPE was analyzed in comparison with that of So. glossinidius, taking advantage that both of them appear to be unique subsets of the gene complement found in their close relative strain HS (Clayton et al. 2012) and using the abundant functional infor­mation available for the orthologous genes in E. coli. This analysis revealed that the essential machinery needed for the storage and processing of genetic information is well pre­served in both SOPE and So. glossinidius, with a nearly com­plete set of genes needed for DNA replication, transcription, and translation. The only two genes absent in SOPE that have been considered essential for DNA replication in E. coli are dnaC and dnaT. However, only dnaC is present in So. glossinidius, and neither are universally present in endo­symbiont genomes (Gil et al. 2004). A general feature of endosymbionts with highly reduced genomes is the loss of DNA repair and recombination machin­ery. The loss of recF, a gene involved in DNA recombinational repair, was previously identified in SOPE and SZPE (Dale et al. 2003). The complete genome analysis revealed that other genes involved in this pathway are also pseudogenized in SOPE (ruvA, ruvB, and recG). Nevertheless, a minimal set of genes required for the mechanisms of base excision, nucleotide excision, and mismatch repair appear to remain intact. In contrast with more ancient endosymbionts, SOPE has maintained a significant number of genes associated with regulatory functions, a characteristic that it shares with So. glossinidius, although the preserved genes are not identical. In addition to many transcriptional and posttranscriptional reg­ulators, SOPE retains three intact sigma factors: rpoD (ct70, primary sigma factor during exponential growth) (Jishage et al. 1996), rpoH (ct32, primary sigma factor controlling the heat shock response during log-phase growth) (Grossman et al. 1984; Yura et al. 1984), and rpoS (alternative master regulator of the general stress response) (Maciag et al. 2011). In So. glossinidius, rpoS is a pseudogene, however, it retains rpoE (<s24, a minor sigma factor that responds to the effects of heat shock and other stresses on membrane and periplas-mic proteins) (Erickson et al. 1987; Wang and Kaguni 1989; Ades et al. 2003) and rpoN (<s54, which controls the expres­sion of nitrogen-related genes, also involved in the nitric oxide stress response) (Hirschman et al. 1985; Hunt and Magasanik Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 83 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 28 Oakeson etal. GBE 1985; Reitzer et al. 1987; Gardner et al. 2003), both of which have been pseudogenized in SOPE. Metabolic Reconstruction The detailed analysis of the predicted metabolic capabilities of SOPE (fig. 5) indicates that it should be able to synthesize most essential amino acids. However, the genes responsible for the synthesis of tryptophan and methionine are pseudogenized, and the complete operon involved in the biosynthesis of his­tidine has been lost. Regarding genes involved in the biosyn­thesis of cofactors and vitamins, SOPE should to be able to synthesize most of them, including the complete pathways for the synthesis of riboflavin, nicotinamide adenine dinucle­otide (NAD+), nicotinamide adenine dinucleotide phosphate (NADP+), coenzyme A, thiamine, and folate. The biosynthesis of lipoic acid, ubiquinone, and siroheme could also be per­formed by SOPE. However, it has lost the edp gene encoding the enzyme needed to perform the first step in the synthesis of pyridoxine. The complete pathway for biotin synthesis (exclud­ing bioH) is pseudogenized. SOPE appears to retain complete pathways for energy me­tabolism, as well as lipid and nucleotide biosynthesis, similar to what is found in So. glossinidius. However, alternative path­ways for nucleotide biosynthesis appear to be disrupted. For example, the purine metabolism pathway that appears to be intact in So. glossinidius suffers from a pseudogenized purH gene, responsible for the synthesis of inosine monophosphate, although the rest of the pathway remains intact. Therefore, SOPE probably employs the same solution as Mycoplasma genitalium, using the enzyme hypoxanthine phosphoribosyl-transferase (EC 2.4.2.8, encoded by the gene hpt) for the synthesis of guanosine monophosphate and adenosine mo­nophosphate from phosphoribosyl pyrophosphate (PRPP) and guanine or adenine. This alternative pathway has been inacti­vated in So. glossinidius. Pyrimidine biosynthesis appears to be complete in So. glossinidius, but the pseudogenization of udk and tdk genes in SOPE likely forces the biosynthesis of cytosine and thymine nucleotides from uracil. A previous comparative genomics study, using genome arrays hybridization (Rio et al. 2003), suggested that SOPE had retained many D-glucosidases, which can catabolize com­plex plant sugars. However, the availability of the whole genome revealed that most genes encoding such enzymes are pseudogenized. Nevertheless, it has retained malP, allow­ing the degradation of starch (the major constituent of rice) to obtain glucose-1-phosphate. As in other endosymbionts, SOPE is undergoing reduction in the number and diversity of transport-associated genes. It still retains intact genes encoding ABC transporters for several cell envelope precursors including, N-acetyl-D-glucosamine, lipopolysaccharides (LPS), lipoproteins, and phospholipids. SOPE also contains an ABC transporter for hydroxymethylpyr-imidine, which is needed for the synthesis of thiamine diphosphate. In contrast, So. glossinidius, which is unable to synthesize thiamine, lacks the transporter for a thiamine pre­cursor hydroxymethylpyrimidine but retains a thiamine trans­porter. Also present in SOPE are ABC transporters for polyamines and several amino acids, such as glutamate, as­partate, and D-methionine as well as transporters for sulfate, iron complexes, and zinc. In addition to its ABC transporter, N-acetylglucosamine can also be internalized through a phos­photransferase system (PTS). N-acetylglucosamine can be used as a carbon source by So. glossinidius and probably by SOPE as well. The only additional PTS that has been preserved in SOPE facilitates the intake of glucose. SOPE has lost those PTSs predicted to be used for the intake of maltose, mannose, and mannitol that are preserved in So. glossinidius. Additionally, SOPE possesses several electrochemical potential-driven transporters for various nitrogenous bases, aromatic and branched amino acids, as well as glutamate, aspartate, gluconate, and glycerate. Cell Envelope and Host-Symbiont Interactions The comparative analysis of the genes involved in peptidogly-can (PG) biosynthesis and turnover in SOPE, So. glossinidius, and strain HS reveals that all three likely retain a canonical cell wall. All genes involved in the initial stages of PG biosyn­thesis are preserved, and only slight differences in enzymes required during the final biosynthetic stage (Scheffers and Pinho 2005) are found in the three analyzed species. All of them have retained mrcB, encoding penicillin-binding protein 1B (PBP1B), one of the bifunctional, inner membrane enzymes catalyzing the transglycosylation and transpeptidation of PG precursors in the formation of the murein sacculus. The gene encoding the second enzyme with this same function, peni­cillin- binding protein 1A (PBP1A, encoded by mrcA), is pseu­dogenized in SOPE. Additionally, in E. coli, there are two outer membrane lipoproteins that are critical for PBP1 function, LpoA and B, acting on MrcA and B, respectively (Paradis- Bleau et al. 2010; Typas et al. 2010). As expected, lpoA and lpoB are intact in strain HS and So. glossinidius, but only lpoB remains intact in SOPE. Although a PBP1B-PBP1A double mu­tation is lethal in E. coli (Spratt 1975; Suzuki et al. 1978; Kato et al. 1985; Wientjes and Nanninga 1991), a single functional PBP1 is sufficient for murein synthesis, since PBP1A mutants do not exhibit defects in growth or cell morphology (Spratt and Jobanputra 1977). Therefore, it appears that SOPE may not have any serious defects in PG formation, which may explain in part why it can trigger the immune response of the rice weevil when injected in the hemolymph following its isolation from the bacteriome or after heat treatment (Anselme et al. 2008; Vigneron et al. 2012). Specific hydrolases, classified as muramidases, glucosami-nidases, amidases, endopeptidases, and carboxypeptidases, are involved in breaking the covalent bonds of the existing PG sacculus, to enable the insertion of new material for cell 8 4 Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 29 Nascent Stage of Symbiosis GBE ZZ______ Giu GlcNac-6P alucose-6P - I zwf+ppggll++rrpi iA] uracil H/dH xanthine A A A - uraA - nupC- xanP-V V V uracil H/dH xanthine CTP-[ndk+nrdAB+ dCTP [ PyrG| nagAB + glmMSU ^ ■ ribose-5P- prs>P RP P - upp>UMP- cmk + ndk>UTP-ddc + ndk + nrdAB^dTTP * D-Glu uDP-GlcNac fructose-6P tktAitalA I V „ ____________________ " " --- I I Pfl<A e^ p I hpt-^ xm p - guaB + purAB + adk-► / Ala ■ ddl» D-Ala- air -> D-Ala-D-Ala J I midDse.1iW* hosphate ^ pep----- guaA+gmk ^ GDP- nrdAB+nd -------- v > I aroFBQEKAC A w nuAD^, fba + 3^ r - rfe + rffEDMT ► giycerai kdsDACB)- ri bul o se-5P 1,3-biphosphoglycerate -------- Gly 3-phosphoglycerate- serACB Ser ^ ® ^ r gmPA k r « - ' CS G T P - folEBK+nudB ribABDHC riboflavin-(ribFl»FMN, FAD 7,8-dihydropteroate Pyridoxal-5P<p^- pyridoxal 2-phosphoglycerate eno PEP pyk^pykFl pyruvate h2s cysNDCHIJ I ^sufABCDSE lipoate A oxidizedthioredoxine ^ , . . . , ' NADP fabHBFGAZI . trxB_1 - MAnOU _ palmitate M - acyl-ACPJ fabH - aoetyl-OoA lipBA 4 * s o r [a t p] ______. acety I p hosohate-f^ckA acetate aceEFApdA p / _____ g f oxalacetate-| aspCj-^Asp---[nadBACD^- ► NAD-[ppnlf-► NADP+ fumC reduced thioredoxine H+ malonyl-ACP fumarate I fabD i sdhABCD + asd| j r L-homoserine , „ thrA thrBC citrate aspartate-semialdehyde dapABDEF + lysA isocitrate ______________ Lys (carAB+aralGHl si ro heme malonyl-CoA succinate ^ NADT NADH * su04 D 2-oxoglutara t y gln^argABCDI succinylCoA.*- sucAB/lpdA SdhA* G|u_ g|tx+hemALBCD> uroporphynogen CoA <coaADE + dpf - panthotanata-* . [Fe-S] cluster assembly i|VE + panBCD + t ' ThiaminePPi-* thiDEL. HMP-* argIGH Gin ornithine Transporters: ^ electrochemical potential driven; ^ channel-type; ABC; phosphotransferase system Fig. 5.-Overview of SOPE metabolism. The names in the yellow boxes indicate the genes predicted to be responsible for a given reaction. The generation of ATP is indicated. Abbreviations (besides the accepted symbols): CMP-KDO, CMP-3-deoxy-D-manno-octulosonate; DHF, dihydrofolate; ECA, enterobacterial common antigen; GlcNac-6P, N-acetyl glucosamine-6-phosphate; H/dH, nucleoside not G; DHAP, dihydroxyacetonephosphate, hydroxy-methylpyrimidine; PEP, phospoenolpyruvate; PG, peptidoglycan; PRPP, phosphoribosyl pyrophosphate; SAM, S-adenosylmethionine; THF, tetrahydrofolate; UQ: ubiquinone. growth and division (Scheffers and Pinho 2005). Many genes encoding proteins with these functions appear pseudogenized or are absent in SOPE and So. glossinidius, when compared with strain HS. Nevertheless, most organisms have redundant enzymes involved in these functions, and although many of them are still poorly characterized, some that would be essen­tial for the maintenance of a well-structured cell wall are still present in SOPE and So. glossinidius. SOPE, So. glossinidius, and strain HS are thought to be able to synthesize simplified LPS. All three lack enzymes to synthe­size the O-antigen, and some genes involved in the modifica­tion of the core region, which have been associated with virulence (Heinrichs et al. 1998; Yethon 1998; Regue et al. 2001) are absent or pseudogenized. Although most genes encoding lipid A-modifying enzymes that have been found in So. glossinidius are present in SOPE, pagP is absent. The PagP protein mediates the palmytoylation of lipid A, a struc­tural modification associated with bacterial resistance to alpha-helical antimicrobial peptides (AMPs) such as cecropin (Pontes et al. 2011). All three analyzed species may be able to synthesize the enterobacterial common antigen, a family-specific surface an­tigen restricted to the Enterobacteriaceae that is shared by almost all members of this family, although it is not present in some endosymbionts. Its biological function remains un­known, although it has been suggested that in Salmonella enterica, it is associated with bile resistance (Ramos-Morales et al. 2003). Genome Biol. Evol. 6(1):76-93. doi:10.1093/gbe/evt210 Advance Access publication January 8, 2014 85 Downloaded from http://gbe.oxfordjournals.org/ at Eccles Health Sci Lib-Serials on January 16, 2014 30 Oakeson etal. GBE In Gram-negative bacteria, up to six different specialized systems have been described for the translocation of proteins through the inner and outer membranes. Some proteins can be directly exported in a single step through the cell wall, whereas others are first exported into the periplasm through the Sec translocation and twin-arginine translocation (Tat) pathways. Both the Sec and Tat translocation systems are present and appear to be intact in SOPE, although several genes encoding accessory subunits have been lost or appear pseudogenized. Recent data have provided strong evidence that outer membrane proteins (OMPs) fulfill pivotal functions in host-symbiont interactions (Weiss et al. 2008; Login et al. 2011; Maltz et al. 2012). Therefore, we focus not only on the OMPs that are encoded in the SOPE and So. glossinidius genomes but also their ability to properly produce and place these pro­teins in the cell envelope. The signal transduction system EnvZ/ OmpR that regulates porin expression in E. coli is present both in SOPE and So. glossinidius. The OMP assembly complex BamABCDSmpA is also complete in SOPE and So. glossinidius. At least one member of the AsmA family, needed for the correct assembly of OMPs, is present in each genome. How