## THE PATHOGENIC YERSINIAE – ADVANCES IN THE UNDERSTANDING OF PHYSIOLOGY AND VIRULENCE, VOLUME II

EDITED BY : Matthew S. Francis and Victoria Auerbuch PUBLISHED IN : Frontiers in Cellular and Infection Microbiology

#### Frontiers Copyright Statement

© Copyright 2007-2019 Frontiers Media SA. All rights reserved. All content included on this site, such as text, graphics, logos, button icons, images, video/audio clips, downloads, data compilations and software, is the property of or is licensed to Frontiers Media SA ("Frontiers") or its licensees and/or subcontractors. The copyright in the text of individual articles is the property of their respective authors, subject to a license granted to Frontiers.

The compilation of articles constituting this e-book, wherever published, as well as the compilation of all other content on this site, is the exclusive property of Frontiers. For the conditions for downloading and copying of e-books from Frontiers' website, please see the Terms for Website Use. If purchasing Frontiers e-books from other websites or sources, the conditions of the website concerned apply.

Images and graphics not forming part of user-contributed materials may not be downloaded or copied without permission.

Individual articles may be downloaded and reproduced in accordance with the principles of the CC-BY licence subject to any copyright or other notices. They may not be re-sold as an e-book.

As author or other contributor you grant a CC-BY licence to others to reproduce your articles, including any graphics and third-party materials supplied by you, in accordance with the Conditions for Website Use and subject to any copyright notices which you include in connection with your articles and materials.

All copyright, and all rights therein, are protected by national and international copyright laws.

The above represents a summary only. For the full conditions see the Conditions for Authors and the Conditions for Website Use. ISSN 1664-8714 ISBN 978-2-88963-051-6 DOI 10.3389/978-2-88963-051-6

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

## THE PATHOGENIC YERSINIAE – ADVANCES IN THE UNDERSTANDING OF PHYSIOLOGY AND VIRULENCE, VOLUME II

Topic Editors: Matthew S. Francis, Umeå University, Sweden Victoria Auerbuch, University of California, Santa Cruz, United States

<sup>&</sup>quot;Transmission electron microscopic observation of Yersinia flagella". Image: Ikenna Obi in collaboration with Umeå Core Facility for Electron Microscopy (Umeå University, Sweden).

Pathogenic Yersinia consist of the prominent human pathogens Y. pestis, Y. enterocolitica, and Y. pseudotuberculosis, the fish pathogen Y. ruckeri, as well as a number of insect pathogens. Facilitated by the ease of in vitro culturing, genetic tractability, and availability of relevant infection models, studies of pathogenic Yersinia have revealed a great deal about physiological processes at the molecular level that contribute to pathogen adaptation to the ever changing environments both inside and outside of the host. Comprehensive genome sequencing analyses has further benefitted understanding of this bacterial pathogen evolution. Critically, many of these detailed molecular studies also identified potential targets for the design and development of anti-bacterial therapeutic drugs that could help to fight the

ever-increasing problem of resistance to conventional antibiotics. New developments in several of these areas are highlighted in this edition of the Research Topic "The Pathogenic Yersiniae – Advances in the Understanding of Physiology and Virulence, Second Edition".

Citation: Francis, M. S., Auerbuch, V., eds. (2019). The Pathogenic Yersiniae – Advances in the Understanding of Physiology and Virulence, Volume II. Lausanne: Frontiers Media. doi: 10.3389/978-2-88963-051-6

# Table of Contents

*06 Editorial: The Pathogenic Yersiniae–Advances in the Understanding of Physiology and Virulence, Second Edition* Matthew S. Francis and Victoria Auerbuch

#### SECTION 1

#### PROTEIN SECRETION

*11 Chromosomally-Encoded* Yersinia pestis *Type III Secretion Effector Proteins Promote Infection in Cells and in Mice*

Sara Schesser Bartra, Cherish Lorica, Lianfen Qian, Xin Gong, Wael Bahnan, Henry Barreras Jr., Rosmely Hernandez, Zhongwei Li, Gregory V. Plano and Kurt Schesser

*20 Heterologous Complementation Studies With the YscX and YscY Protein Families Reveals a Specificity for* Yersinia pseudotuberculosis *Type III Secretion*

Jyoti M. Gurung, Ayad A. A. Amer, Monika K. Francis, Tiago R. D. Costa, Shiyun Chen, Anton V. Zavialov and Matthew S. Francis

*36 Type VI Secretion Systems Present New Insights on Pathogenic* Yersinia Xiaobing Yang, Junfeng Pan, Yao Wang and Xihui Shen

#### SECTION 2

#### NICHE ADAPTATION


Guillaume Le Baut, Claire O'Brien, Paul Pavli, Maryline Roy, Philippe Seksik, Xavier Tréton, Stéphane Nancey, Nicolas Barnich, Madeleine Bezault, Claire Auzolle, Dominique Cazals-Hatem, Jérome Viala, Matthieu Allez, The REMIND GROUP, Jean-Pierre Hugot and Anne Dumay

#### SECTION 3

#### GENE EXPRESSION CONTROL


Jeffrey K. Schachterle, Ryan M. Stewart, M. Brett Schachterle, Joshua T. Calder, Huan Kang, John T. Prince and David L. Erickson *104 BfvR, an AraC-Family Regulator, Controls Biofilm Formation and pH6 Antigen Production in Opposite Ways in* Yersinia pestis *Biovar Microtus* Haihong Fang, Lei Liu, Yiquan Zhang, Huiying Yang, Yanfeng Yan, Xiaojuan Ding, Yanping Han, Dongsheng Zhou and Ruifu Yang

#### SECTION 4

#### MICROBIOLOGY AND PATHOGENESIS OF NON-MAMMALIAN YERSINIA INFECTIONS

*115 The Infection Process of* Yersinia ruckeri*: Reviewing the Pieces of the Jigsaw Puzzle*

José A. Guijarro, Ana I. García-Torrico, Desirée Cascales and Jessica Méndez


Katharina Springer, Philipp-Albert Sänger, Christian Moritz, Angela Felsl, Thomas Rattei and Thilo M. Fuchs

#### SECTION 5

#### NEW FRONTIERS IN YERSINIA BIOLOGY RESEARCH

*152 An Experimental Pipeline for Initial Characterization of Bacterial Type III Secretion System Inhibitor Mode of Action Using Enteropathogenic*  Yersinia

Jessica M. Morgan, Hanh N. Lam, Jocelyn Delgado, Justin Luu, Sina Mohammadi, Ralph R. Isberg, Helen Wang and Victoria Auerbuch


# Editorial: The Pathogenic Yersiniae–Advances in the Understanding of Physiology and Virulence, Second Edition

#### Matthew S. Francis <sup>1</sup> \* and Victoria Auerbuch<sup>2</sup> \*

<sup>1</sup> Department of Molecular Biology, Umeå Centre for Microbial Research, Umeå University, Umeå, Sweden, <sup>2</sup> Department of Microbiology and Environmental Toxicology, University of California, Santa Cruz, Santa Cruz, CA, United States

Keywords: phenotypic and niche adaptation, protein secretion, biofilm, small regulatory RNAs, two-component systems, virulence blockers, Crohn's disease, fish- and insect-pathogen

#### **Editorial on the Research Topic**

#### **The Pathogenic Yersiniae–Advances in the Understanding of Physiology and Virulence, Second Edition**

#### Edited and reviewed by:

John S. Gunn, The Research Institute at Nationwide Children's Hospital, United States

#### \*Correspondence:

Matthew S. Francis matthew.francis@umu.se Victoria Auerbuch vastone@ucsc.edu

#### Specialty section:

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

> Received: 20 March 2019 Accepted: 03 April 2019 Published: 18 April 2019

#### Citation:

Francis MS and Auerbuch V (2019) Editorial: The Pathogenic Yersiniae–Advances in the Understanding of Physiology and Virulence, Second Edition. Front. Cell. Infect. Microbiol. 9:119. doi: 10.3389/fcimb.2019.00119 Of the 18 known Yersinia species, Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica are pathogenic to humans and animals and are widely characterized. The zoonotic obligate pathogen Y. pestis is the causal agent of plague, a systemic disease that is usually fatal if left untreated (Zietz and Dunkelberg, 2004; Zhou et al., 2006). Free-living Y. enterocolitica and Y. pseudotuberculosis are the agents of yersiniosis, a rarely systemic gastrointestinal disease (Galindo et al., 2011). The remaining species are mostly harmless to humans, although Y. ruckeri is an enteric fish pathogen affecting mainly salmonids, while a few others display toxicity toward insects (Sulakvelidze, 2000; Tobback et al., 2007; Fuchs et al., 2008; Chen et al., 2010). At the forefront of Yersinia research are studies of classical microbiology, pathogenesis, protein secretion, niche adaptation, and regulation of gene expression. In pursuit of these endeavors, new frontiers are being forged on waves of methodological and technological innovation. In this second edition of the special research topic on the pathogenic Yersiniae is a compilation of reviews and research articles that summarize current knowledge and future research directions in the Yersinia pathophysiology field.

### PROTEIN SECRETION

Type III secretion (T3S) is prominent protein delivery process in a large number of Gram-negative bacteria that confers to them an ability to interact in pathogenic or symbiotic relationships with either vertebrate or invertebrate hosts (Buttner, 2012; Deng et al., 2017). The Ysc-Yop T3S system (T3SS) is encoded on a virulence plasmid common to all human pathogenic Yersinia (Cornelis et al., 1998). This so-called "injectisome" has long been believed to provide a conduit through which a restricted set of just six or seven plasmid-encoded host-modulating Yop effectors are delivered from the bacterial cytoplasm into the eukaryotic cell cytosol (Pha and Navarro, 2016; Grabowski et al., 2017). Using a transposon site hybridization based genome wide screen, Schesser-Bartra et al. identified three chromosomally-encoded proteins that promote Y. pestis infection in cells and in mice. With features indicative of host-modulating Yop effectors, they identify the first non-plasmid encoded secretion substrates of the Ysc-Yop T3SS. In another study that performed heterologous complementation analyses with the YscX and YscY protein families, Gurung et al. reveal that the YscX and YscY protein complex produced by Y. pseudotuberculosis is specifically critical for biogenesis and/or function of the Ysc-Yop T3SS. The authors go on to discuss what might be the molecular basis for this specificity.

While pathogenic potential of Yersinia for humans and animals is heavily correlated to the plasmid encoded Ysc-Yop T3SS, Yang et al. provide new insight into the four independent Type VI secretion system (T6SS) copies present in human pathogenic Yersinia. The impact of these multiple systems on Yersinia physiology and pathogenesis is likely to be very large given how T6SSs have capacity to deliver multiple effectors into either prokaryotic or eukaryotic cells, and are known to affect diverse biological processes such as virulence, anti-virulence, stress resistance and competition (Alteri and Mobley, 2016; Lien and Lai, 2017).

#### NICHE ADAPTATION

Enteropathogenic Yersinia are foodborne pathogens. Therefore it comes as no surprise that they thrive at refrigeration temperatures (Brocklehurst and Lund, 1990; Goverde et al., 1994; Azizoglu and Kathariou, 2010; Keto-Timonen et al., 2018), and in this environment even remain primed for infection (Asadishad et al., 2013). To understand the molecular mechanisms by which psychotropic Yersinia thrive in cold environments may give rise to strategies by which growth can be restricted, and this would be a strategically important preventative measure for the food processing industry. To investigate the genome-wide cold adaptation behavior of Y. pseudotuberculosis, Virtanen et al. used RNA-Seq technology to identify genes that were significantly more expressed in a cell density specific manner at cold temperature. Among the many genes that were up-regulated were nutrient acquisition genes, cold shock protein genes, DEAD-box RNA helicase genes, genes handling compatible solutes, genes involved in transcription termination and translation initiation, and genes involved in cell wall modification. This suggests that Y. pseudotuberculosis establishes a core network of cold responsive proteins to drive ribosome biogenesis and function at low temperature.

It follows that psychotropic Yersinia are enriched in a variety of foods on a global scale (Hilbert et al., 2003; Ozdemir and Arslan, 2015; Le Guern et al., 2016). Moreover, changing food consumption practices and globalization of the international food trade have contributed to increased frequency of yersiniosis (Gupta et al., 2015). At the same time, orally ingested Yersinia have the potential to survive passage through the gastrointestinal tract. It has been postulated that surviving bacteria may contribute to the onset or persistence of gut inflammation (Hugot et al., 2003). Although experimental mouse models of Crohn's disease do not discount contributions made by infecting enteropathogenic Yersinia (Meinzer et al., 2008; Murthy et al., 2014; Fonseca et al., 2015; Han et al., 2017), support stemming from cohort studies of Yersinia infected clinical material is underwhelming (Kallinowski et al., 1998; Lamps et al., 2003; Knosel et al., 2009; Chiodini et al., 2013; Leu et al., 2013). To further investigate this issue, Le Baut et al. analyzed the prevalence of Yersinia species in a total of 470 illeal samples taken from Crohn's disease patients and healthy controls. Significantly, Yersinia species were detected with equal frequency in both disease and healthy ileum tissue, suggesting that they are well adapted to this niche. Hence, there is now a need to characterize the effect of resident Yersinia on maturation and regulation of the mucosal immune response.

### GENE EXPRESSION CONTROL

Behind every successful niche adaptation is a complex regulatory circuitry that controls specific gene expression profiles. For example, the two-component or histidineaspartate phosphorelay systems are vital for the monitoring of environmental and intracellular signals to produce changes in gene expression or behavioral responses (Stock et al., 2000; Laub and Goulian, 2007). In Yersinia, a large number of twocomponent systems are known (Marceau, 2005), with a few of them making recognized contributions to Yersinia survivability in the environment or in an infected host (Flamez et al., 2008; Reboul et al., 2014). A notable two component system is EnvZ/OmpR that enables many bacteria to alter gene expression in response to osmotic and acid stress (Walthers et al., 2005; Chakraborty and Kenney, 2018). The work of Jaworska et al. reports on OmpR-mediated control of iron acquisition via transcriptional repression of the HemR1 and HemR2 heme receptors. This regulatory circuit works in conjunction with the transcriptional repressor Fur to prevent over-accumulation of iron/heme by Y. entercolitica.

Another two component system is BarA/UvrY. Responding to metabolic end products such as short chain fatty acids, BarA/UvrY signaling is the primary regulator of the widespread Csr global regulatory system, and in this way can profoundly influence multiple metabolic, behavioral and virulence traits in many bacteria (Vakulskas et al., 2015). In the report by Schachterle et al. BarA/UvrY signaling was found to repress the formation of Y. pseudotuberculosis biofilms through activation of the CsrB regulatory RNA. It is likely that this is pleiotropic repression affecting multiple elements of biofilm formation and maintenance by Y. pseudotuberculosis.

The primary requirement for mature biofilm formation by Yersinia is the production of an exopolysaccharide (EPS) that requires the hmsHFRS locus to coordinate its synthesis and transport (Bobrov et al., 2008). Moreover, c-di-GMP enhances EPS production, and the levels of this signaling molecule are tightly controlled by the opposing actions of two diguanylate cyclases (encoded by hmsT and hmsD) and a phosphodiesterase (hmsP) (Kirillina et al., 2004; Bobrov et al., 2011). The study of Fang et al. describes a novel AraC-like transcriptional activator termed BfvR that controls Y. pestis biofilm formation via stimulating transcription from the hmsHFRS and hmsCDE operons to elevate EPS and c-di-GMP production. This identifies BfvR as the first AraC family transcription regulator reported to control biofilm formation in Yersinia.

### MICROBIOLOGY AND PATHOGENESIS OF NON-MAMMALIAN YERSINIA INFECTIONS

Although not known to be harmful to humans, the enteric fish pathogen Y. ruckeri is still a pathogen of great interest as it has capacity to causes significant economic losses in the aquaculture industry (Tobback et al., 2007). This is reflected by a recent surge of reports that offer improved understanding of the biological processes contributing to Y. ruckeri infection and pathogenicity. The review by Guijarro et al. assimilates this new knowledge to provide up-to-date insight into the molecular mechanisms of the Y. ruckeri infection process. Complementing this review is a report by Wrobel et al. that analyzed the complete DNA sequence of the unique pYR4 plasmid from a highly virulent isolate of Y. ruckeri. This cryptic plasmid has potential to impact positively on Y. ruckeri virulence since it encodes for a type IV pilus and a type IV secretion system that are well established virulence associated factors in other bacteria (Craig et al., 2004; Giltner et al., 2012; Gonzalez-Rivera et al., 2016; Grohmann et al., 2018).

Moreover, there has been great interest in the function and taxonomical distribution of insecticidal genes among Yersinia spp., owing in part to their potential in contributing new knowledge to the ecology, evolution and pathogenicity of human pathogenic Yersinia (Pinheiro and Ellar, 2007; Fuchs et al., 2008, 2011; Hares et al., 2008; Spinner et al., 2012, 2013; Alenizi et al., 2016). Using Y. frederiksenii as a model system that displays toxicity toward insects, Springer et al. were able to demonstrate a distinct contribution of the novel heat-stable cytotonic enterotoxin to oral and intrahemocoelic toxicity of infected insects. These findings led the authors to discuss how the ability to enter invertebrates may constitute a selective advantage to Yersinia isolates in environmental survival and evolution of virulence.

#### NEW FRONTIERS IN YERSINIA BIOLOGY RESEARCH

Conventional antibiotics have saved the lives of many by decreasing the morbidity and mortality of bacterial infectious diseases. However, the global emergence of bacteria resistant to these antibiotics means that they no longer work effectively, and this presents a major healthcare issue that creates tremendous global social and economic suffering (Aminov, 2010). Consequently, alternative solutions to this healthcare crisis that are effective and reliable must be swiftly identified. In recent years, one such alternate approach has been to isolate anti-bacterials that function by targeting a virulence determinant (Clatworthy et al., 2007; Maura et al., 2016). Ideally, these so called "anti-infectives" or "virulence blockers" would be specific for pathogenic bacteria and have a bacteriostatic effect that would synergize with the immune system to clear the infection. A classic example of this endeavor is the identification of novel chemical inhibitors of the T3SS (Keyser et al., 2008; Duncan et al., 2012). Despite the success of identifying chemical inhibitors of the T3SS, none of these have yet reached the market. This issue is addressed in a report by Morgan et al. which describes the development of an experimental pipeline that would help transition from high throughput screening to inhibitor validation and initial determination of their mode of action. In so doing, the authors consider important new possible modes of action for T3SS inhibitors.

Bacterial virulence regulation is exquisitely fine-tuned so that subsets of virulence factors are expressed only at times of need. Alterations in the local environment account for triggering changes in this virulence gene expression profile. Responses are rapid, and it is now clear that post-transcriptional regulatory effects, such as small non-coding RNAs contribute to the rapidity of this re-programming. Benefitting from progressive developments in genome-wide omics-based methods of exploration, several RNA-based regulatory systems have been discovered in pathogenic Yersinia. These discoveries have been reviewed by Knittel et al. in the context of Yersinia niche colonization, metabolic adaptation, acute and chronic infection, and evolution. By inference, at least some RNA-based regulatory systems could serve as a suitable target of anti-infective drug development.

The second edition of this research topic, provides many examples demonstrating the great capacity of Yersinia species to adapt and thrive in diverse environmental niches. This is reiterated by the timely review by Davis, which sheds light on the ability of Yersinia sub-populations to phenotypically diversify during an infection in order to balance the need to maintain bacterial growth while resisting attack from different cellular elements of an activated immune system. Underpinning this phenotypic diversification is the ability of subsets of bacteria to make temporal and spatial adjustments to their gene expression profiles in response to the microenvironment. Having the technology to detect gene expression profiles in distinct subpopulations of bacteria offers a unique opportunity to understand the yin and yang of interactions between individual bacteria and specific immune cell types. In turn, this may eventually enable the generation of more efficacious approaches to treat infections by having the option of tailoring novel antibacterials or their immunomodulatory counterparts that can favorably influence the outcome of this bacteria-immune cell interplay.

### AUTHOR CONTRIBUTIONS

MF developed the initial concept and outline. Both MF and VA contributed to the final version of the manuscript.

### FUNDING

MF received research funding from the Swedish Research Council (Vetenskapsrådet) under award numbers 2009-5628, 2014-2105 and 2018-02676, the Foundation for Medical Research at Umeå University and the Faculty of Science and Technology at Umeå University. VA received research funding from the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under award numbers R01AI106930 and R01AI119082.

#### REFERENCES


evidence for oscillation between invertebrates and mammals. BMC Genomics 12:168. doi: 10.1186/1471-2164-12-168


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Francis and Auerbuch. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Chromosomally-Encoded *Yersinia pestis* Type III Secretion Effector Proteins Promote Infection in Cells and in Mice

Sara Schesser Bartra<sup>1</sup> , Cherish Lorica1,2, Lianfen Qian<sup>3</sup> , Xin Gong<sup>4</sup> , Wael Bahnan<sup>1</sup> , Henry Barreras Jr. <sup>1</sup> , Rosmely Hernandez <sup>1</sup> , Zhongwei Li <sup>4</sup> , Gregory V. Plano<sup>1</sup> and Kurt Schesser <sup>1</sup> \*

<sup>1</sup> Department of Microbiology and Immunology, University of Miami Miller School of Medicine, Miami, FL, United States, <sup>2</sup> Division of Pediatric Infectious Diseases, Department of Pediatrics, University of Miami Miller School of Medicine, Miami, FL, United States, <sup>3</sup> Department of Mathematics, Charles E. Schmidt College of Science, Florida Atlantic University, Boca Raton, FL, United States, <sup>4</sup> Department of Biomedical Science, Charles E. Schmidt College of Medicine, Florida Atlantic University, Boca Raton, FL, United States

#### *Edited by:*

Matthew S. Francis, Umeå University, Sweden

#### *Reviewed by:*

Nikhil A. Thomas, Dalhousie University, Canada Andreas Diepold, Max-Planck-Institut für terrestrische Mikrobiologie, Germany Luís Jaime Mota, Universidade Nova de Lisboa, Portugal

> *\*Correspondence:* Kurt Schesser kschesser@med.miami.edu

#### *Specialty section:*

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

> *Received:* 09 August 2018 *Accepted:* 22 January 2019 *Published:* 22 February 2019

#### *Citation:*

Schesser Bartra S, Lorica C, Qian L, Gong X, Bahnan W, Barreras H Jr, Hernandez R, Li Z, Plano GV and Schesser K (2019) Chromosomally-Encoded Yersinia pestis Type III Secretion Effector Proteins Promote Infection in Cells and in Mice. Front. Cell. Infect. Microbiol. 9:23. doi: 10.3389/fcimb.2019.00023 Yersinia pestis, the causative agent of plague, possesses a number of virulence mechanisms that allows it to survive and proliferate during its interaction with the host. To discover additional infection-specific Y. pestis factors, a transposon site hybridization (TraSH)-based genome-wide screen was employed to identify genomic regions required for its survival during cellular infection. In addition to several well-characterized infection-specific genes, this screen identified three chromosomal genes (y3397, y3399, and y3400), located in an apparent operon, that promoted successful infection. Each of these genes is predicted to encode a leucine-rich repeat family protein with or without an associated ubiquitin E3 ligase domain. These genes were designated Yersinia leucine-rich repeat gene A (ylrA), B (ylrB), and C (ylrC). Engineered strains with deletions of y3397 (ylrC), y3399 (ylrB), or y3400 (ylrA), exhibited infection defects both in cultured cells and in the mouse. C-terminal FLAG-tagged YlrA, YlrB, and YlrC were secreted by Y. pestis in the absence but not the presence of extracellular calcium and deletions of the DNA sequences encoding the predicted N-terminal type III secretion signals of YlrA, YlrB, and YlrC prevented their secretion, indicating that these proteins are substrates of the type III secretion system (T3SS). Further strengthening the connection with the T3SS, YlrB was readily translocated into HeLa cells and expression of the YlrA and YlrC proteins in yeast inhibited yeast growth, indicating that these proteins may function as anti-host T3S effector proteins.

Keywords: *Yersinia*, type III secretion, pathogenesis, leucine rich repeats, plague, mice

## INTRODUCTION

Yersinia pestis, the etiologic agent of plague, causes a variety of serious diseases in humans and animals. The clinical syndromes in humans include bubonic, pneumonic, and septicemic plague (Perry and Fetherston, 1997). Fleas transmit Y. pestis from infected domestic rats to humans causing bubonic plague. Y. pestis also can be transmitted via respiratory secretions following contact with another infected human, leading to pneumonic plague. Currently, plague is still a public health threat in certain regions of Asia, Africa, North and South America (Centers for Disease Control Prevention., 2013). In recent years in the U.S., most cases of plague have occurred in children in whom diagnosis has been delayed. Among 183 U.S. pediatric cases from 1947 to 2001, 91% presented primarily as bubonic and one third of these cases developed secondary complications, such as sepsis, meningitis, and pneumonia. Children were more likely than adults to manifest with bubonic plague (91 vs. 79%), develop complications (32 vs. 27%), and to die (17 vs. 14%) (Dennis and Chow, 2004). Because plague is highly contagious, Y. pestis can be used in biological warfare and is considered a Category A agent of bioterrorism (Inglesby et al., 2000).

Among the virulence factors identified in Y. pestis, the plasmid pCD1 encoded type III secretion system (T3SS) is one of the best-studied. The T3SS is a complex protein machinery shared by numerous Gram-negative pathogens that injects effector proteins directly into the cytosol of an infected eukaryotic cell (Cornelis, 2002a). The translocation of effector proteins requires direct contact between the bacterium and the host cell. In Y. pestis, the T3S apparatus is called the Ysc injectosome and the effector proteins are termed Yops (Yersinia outer proteins). The Yops disrupt host signaling pathways that normally lead to bacterial phagocytosis and the production of pro-inflammatory cytokines (Cornelis and Van Gijsegem, 2000). To date, six Yop effectors have been identified—YopH, YopE, YopT, YpkA/YopO, YopP/YopJ, and YopM (Cornelis, 2002b). Following its identification as a virulence factor in the mouse model, YopM was initially described as an inhibitor of platelet aggregation (Leung and Straley, 1989). Later YopM was linked to a variety of activities associated with the host innate immune response including depletion of inflammatory monocytes and natural killer (NK) cells and induction of specific cytokines (Kerschen et al., 2004; Ye et al., 2009; McPhee et al., 2010). Once translocated into eukaryotic cells, YopM directly interacts with a number of host proteins including kinases and capase-1 to inhibit signaling pathways and inflammasome activation (McDonald et al., 2003; Hentschke et al., 2010; LaRock and Cookson, 2012; Chung et al., 2014).

Structurally YopM is primarily composed of a number of leucine-rich repeats (LRR) (Evdokimov et al., 2001). Recently it has been shown that Y. pestis, as well as a number of other Yersinia species (but not Y. enterocolitica), possesses a chromosomal locus that potentially encodes three YopMlike proteins (Hu et al., 2016). Sequence analysis showed that the plasmid-encoded YopM was evolutionarily distinct from these three chromosomal-encoded YopM-like proteins and that two of these YopM-like proteins possessed E3 ligase domains in their C-terminal regions (Hu et al., 2016). Here we describe the independent identification and characterization of the genes encoding these three chromosomally-encoded YopM-like proteins in an unbiased screen for Y. pestis factors that promote its survival following its infection of macrophages.

## MATERIALS AND METHODS

#### Bacterial Strains

The Escherichia coli DH5α and Y. pestis strains were routinely grown in heart infusion broth (HIB) or on tryptose blood agar (TBA) base plates (Difco, Detroit, MI) at 27◦C (Y. pestis) or 37◦C (E. coli). Appropriate antibiotics were added to culture media when needed with ampicillin (50µg/ml), kanamycin (50µg/ml), chloramphenicol (20µg/ml), and streptomycin (50µg/ml). Nonpolar deletion of Y. pestis KIM5 chromosomal DNA sequences ylrC (y3397; codons 29-515), ylrB (y3399; codons 29-261), and ylrA (y3400; codons 23-529) was accomplished using lambda Red recombination as described by Datsenko and Wanner (2000). PCR products used to construct the gene replacement were amplified using the template plasmid pKD4 (Km<sup>r</sup> ). The resulting PCR products were gel purified, ethanol precipitated, and resuspended in 10 µl of distilled water. Y. pestis KIM5 strain carrying plasmid pKD46, which encodes the Red recombinase, was induced with 0.2% L-arabinose for 2 h prior to harvest. Electrocompetent cells were electroporated with the purified PCR products. The transformations were plated onto TBA plates containing kanamycin (50µg/ml). Plasmid pCP20, which encodes the FLP recombinase, was electroporated into the Km<sup>r</sup> resistant strains to facilitate the removal of the FLP recognition target-flanked kan cassette and plasmid pKD46 simultaneously. Plasmid pCP20 was cured from the Y. pestis deletion strains by overnight growth at 39◦C. Additionally, a deletion strain of the three gene sequences ylrC-codon 29 to ylrAcodon 529 was constructed using lambda Red-recombination as described above.

### Transposon Site Hybridization (TraSH)-Based Screening

A TraSH-based approach was used to map and quantify the relative abundance of the different transposon mutants in order to identify transposon mutant insertion sites that are underrepresented in the population of mutants exposed to RAW264.7 murine macrophage-like cells. The details of the mutagenesis, infection, and TraSH-screen and data analysis are described in Bartra et al. (2012).

#### Infection Assays

Cell infection assays were performed essentially as described by Rosenzweig et al. (2005). RAW 264.7 murine macrophage-like cells and HeLa cells were cultured in Dubelcco's modified Eagle's medium (DMEM, Gibco) containing 10% heat-inactivated fetal bovine serum (Cellgro) at 37◦C in the presence of 5% CO2. Cells were seeded in 24-well plates at densities of 2.5–4.0 × 10<sup>5</sup> per well. Bacterial cell cultures were grown in tissue culture (TC) medium overnight at 27◦C, diluted into fresh TC media and incubated with shaking at 27◦C for 2 h and then shifted to 37◦C for 1 h prior to being added to the cultured macrophages at a multiplicity of infection (MOI) of 30. After a 30 min attachment period, fresh DMEM medium was added to each well. At the 0 h and 8 h time points, the infected cells were lysed with 500 µl of distilled H2O and a portion of which was plated on

TBA plates. Colony forming units (CFU) were counted after 2–3 days. For mouse infections, all procedures were in strict accordance with federal and state government guidelines for the Care and Use of Laboratory Animals of the National Institutes of Health and their use was approved for this entire study by the University of Miami institutional animal care and use committee (protocol number 15-081). Mice were infected intravenously with a total of 2000 CFU (1000 CFU each of the parental Y. pestis KIM5 and the isogenic 1yopB or 1ylrABC strains). Mice were humanely sacrificed at 48 h post infection, spleens were removed and homogenized in sterile water containing 0.05% triton X-100 by grinding through a fine wire mesh. The resulting homogenates were diluted and plated on media containing either chloramphenicol (CM) to select for the CM-resistant parental strain, as well as antibiotic-free media that allowed growth of both the parental strain and the CM-sensitive mutant strains. Two to three days later colonies were enumerated and the competition index (CI) for the parental/ 1yopB and parental/1ylrABC coinfected animals was computed by dividing the CFU of the mutant by the CFU of the parental strain.

### Construction of YlrA, YlrB, and YlrC Expression Plasmids

DNA fragments used encoding YlrA, YlrB, and YlrC were PCR amplified from chromosomal DNA of Y. pestis KIM5. The resultant DNA fragments were digested with HindIII and BglII and inserted into HindIII- and BglII-digested pFLAG-CTC vector (Sigma-Aldrich). These vectors express full-length C-terminal FLAG-tagged YlrA-FLAG, YlrB-FLAG, and YlrC-FLAG. In addition, DNA sequences predicted to encode the

FIGURE 1 | A TraSH screen indicates that the ylrABC locus promotes the survival of Y. pestis during infection. A pool of 90,000 unique Y. pestis transposon variants were used to infect cultured RAW 264.7 macrophages. Following infection bacterial DNA was isolated and the relative abundance of specific regions of the Y. pestis genome was determined. Shown in the abundance at the ylr locus indicating that Y. pestis variants possessing transposon insertion at this locus were under-represented at the termination of the infection period. Protein domains are labeled as Novel E3 Ubiquitin Ligase (NEL), Leucine-Rich Repeat (LRR), and T3SS Secretion and Translocation Signal (SS) domains.

YlrA, YlrB, and YlrC N-terminal T3S signal (SS) (amino acid residues 2 to 10) were deleted from each expression vector using whole plasmid PCR (Imai et al., 1991), generating plasmids pYlrA-FLAG-1SS, pYlrB-FLAG-1SS, and pYlrC-FLAG-1SS. Oligonucleotide pairs used were YlrASS-F and YlrASS-R, YlrBSS-F and YlrBSS-R, and YlrCSS-F and YlrCSS-R The resultant Ylr expression plasmids were transformed into Y. pestis KIM814 (Bartra et al., 2006).

#### Construction of Vectors for β-Lactamase Translocation Studies

Expression plasmids encoding full length YlrA, YlrB, and YlrC carrying a C-terminal β-lactamase gene were constructed by the PCR-ligation-PCR technique (Ali and Steinkasserer, 1995). Individual Y. pestis KIM genes and upstream sequences that include each gene's ribosomal binding site were amplified by PCR from vectors encoding full-length FLAG-YlrA, FLAG-YlrB, and FLAG-YlrC, respectively, using oligonucleotides primer pairs YlrA-KpnI-F and YlrA-FL-R, YlrB-KpnI-F and YlrB-FL-R, and YlrC-KpnI-F and YlrC-FL-R. The DNA fragments encoding the Bla gene were amplified from plasmid pBSKII- using primers Bla-25-F and Bla-STOP-HindIII-R. Obtained DNA fragments were gel purified, kinased and ligated. The reaction was used for a second PCR using primers YlrA-KpnI-F and Bla-STOP-HindIII-R, YlrB-KpnI-F and Bla-STOP-HindIII-R, and YlrC-KpnI-F and Bla-STOP-HindIII-R. The resulting DNA fragments were ethanol precipitated, digested with KpnI and HindIII, and inserted into KpnI and HindIII-digested pBad18-Cm<sup>r</sup> . The resultant plasmids were transformed into a Y. pestis strain lacking YopE, YopJ, YopM, YopT, YopH, and YpkA.

#### Yeast Studies

DNA fragments encoding YlrA, YlrB and YlrC were amplified by PCR from pFLAG-YlrA, pFLAG-YlrB, and pFLAG-YlrC, respectively, using oligonucleotides primer pairs YlrA-XhoI and YlrA-BamHI, YlrB-XhoI and YlrB-BamHI, and YlrC-XhoI and YlrC-BamHI listed in Table 2. The resultant DNA fragments were digested with BamHI and XhoI and inserted into BamHIand XhoI-digested pREP3X. The pREP-3X vector contains the inducible nmt1 promoter (Maundrell, 1993) and the presence of thiamine (15µM) in the medium represses the promoter. If thiamine is not present in the media, the promoter becomes activated resulting in the transcription of the ylr gene cloned downstream. Plasmids generated were pREP-3X-YlrA, pREP-3X-YlrB, and pREP-3X-YlrC. In addition, site-specific mutation of three catalytic residues of the putative NEL domain of YlrA was done using the whole plasmid PCR method (Imai et al., 1991) generating plasmids pREP-3X-YlrA-C386S, pREP-3X-YlrA-R389A, and pREP-3X-YlrA-S449F. Schizosaccharomyces pombe h <sup>−</sup> ade6-704 leu1-32 ura4-D18 was grown in Pombe Glutamate medium (PMG) at 32◦C. The plasmid vector pREP3x (nmt13x Thiamine repressible promoter) was used to express the Ylr genes in fission yeast. The plasmid vectors were transformed using the lithium acetate protocol (Morita and Takegawa, 2004). The yeast strains were grown in PMG supplemented with the appropriate amino acids as well as thiamine to maintain selection for 2 days in logarithmic growth to an optical density (OD) of 0.1–0.4 at 595 nm. After growth, the cells were centrifuged at 4,000 × g, washed 3 times with distilled sterile water, and diluted for an overnight growth in media lacking thiamine. The following day, the OD of the cultures at 595 nm was measured and the cultures were diluted to an OD of 0.1. Three 10-fold serial dilutions of the cultures were made, and 5 µL of each dilution was spotted onto an agar plate with and without thiamine and grown at 32◦C for 3 days.

#### Secretion Assay

For standard secretion assays, Y. pestis strains were grown in thoroughly modified Higuchi's (TMH) medium in the presence or absence of 2.5 mM CaCl<sup>2</sup> for 1 h at 27◦C and then shifted to 37◦C for the next 5 h as previously described (Jackson et al., 1998). FLAG-tagged Ylr proteins (pFLAG-CTC vector) were induced with 0.05 mM IPTG (isopropyl-β-Dthiogalactopyranoside) at the temperature shift.

#### SDS-PAGE and Immunoblotting

Cultures of bacteria were harvested by centrifugation at 14,000 × g for 10 min at room temperature. Pellets of whole-cell bacteria and trichloroacetic acid (TCA)-precipitated supernatant proteins were resuspended according to the harvest OD<sup>620</sup> and subjected to SDS-PAGE and immunoblotting as previously described (Jackson et al., 1998). LcrV was detected with rabbit polyclonal antisera (1:20,000) raised against the full-length LcrV proteins. FLAG-tagged proteins were detected with anti-FLAG M2 monoclonal antibody (Sigma-Aldrich) (1:1,000).

#### β-Lactamase Translocation Assay

HeLa cells were seeded to a six-well plate to achieve a 40– 50% confluence 1 day prior to infection. Y. pestis strains were grown at 27◦C for 2 h in HIB with antibiotics as appropriate and then shifted to 37◦C for 1 h to induce expression of T3SS proteins. At temperature shift, 0.2% L-arabinose was added to the cultures. Bacteria were added at MOI = 50 to HeLa cells. The infected cells were incubated at 37◦C with 5% CO2. Two negative controls (without Y. pestis cells and Y. pestis

immunoblot analysis with anti-FLAG and anti-LcrV (control) antibodies.

with pBAD18-pYscF1-Bla) and a positive control were used to determine the background blue and green fluorescence. After infection for 3 h, the growth medium from the cells was removed and the cells were washed with 500 µL of 1x of GIBCO Hank's Balanced Salt Solution (HBSS). 6X CCF-AM (Invitrogen) was added to each sample to obtain a final concentration of 1X per manufacturer's instructions. The plate was covered and incubated in room temperature for 1 h. Cells were then visualized by confocal-fluorescent microscopy (Leica TCS SP5 Inverted Confocal Microscope).

#### RESULTS

#### Pro-survival Activity of *Y. pestis* Leucine-Rich Proteins During Cellular Infection

Previously we described a transposon site hybridization (TraSH) based screen to identify Y. pestis loci that are required during infection of cultured mouse macrophage-like RAW 264.7 cells (Bartra et al., 2012). From a pool of approximately 90,000 unique variants, 44 Y. pestis ORFs were identified that appeared to be required for pathogen survival during its interaction with the macrophage. A number of these ORFs (17) are encoded on the extrachromosomal plasmids pCD1 and pPCP1 and express components of the T3SS and Pla protease that both play critical roles in Y. pestis virulence (Viboud and Bliska, 2005). One such pCD1-encoded locus among this set of 17 identified ORFs, YopM, possesses Leucine-Rich Repeats (LRRs), and is associated with a variety of infection phenotypes (see Introduction). Among the 27 chromosomally-encoded pro-survival ORFs, there were three, y3397, y3399, and y3400, that were previously predicted to possess LRRs (Hu et al., 2016) and here will be designated as Yersinia leucine-rich repeat A (ylrA), B (ylrB), and C (ylrC), respectively (**Figure 1**). In addition to LRR-encoding sequences, these three ORFs also possess predicted secretion signals (SS) for the T3SS at their N-termini and YlrA and YlrC are additionally predicted to encode novel ubiquitin E3 ligase (NEL) domains (Wang et al., 2013; Marchler-Bauer et al., 2017).

### Defective Infection Phenotypes of Engineered 1*ylr* Strains

The aforementioned TraSH screen involves infecting macrophages with a mixture of Y. pestis variants followed by a quantification of under-represented genetic loci. To test whether the Ylr-encoding loci play a pro-survival function for Y. pestis during a single-strain infection, individual 1ylrA, 1ylrB, and 1ylrC mutant strains were constructed and tested in a colony-forming unit (CFU)-based infection assay. Similar to what has been shown previously, the fold-increases of the parental Y. pestis KIM5 strain and the T3SS mutant stain (1yopB) during a 8 h infection, were 170 and 3, respectively (**Figure 2A**; Rosenzweig et al., 2005). The infectivity of the single gene deletion 1ylrA, 1ylrB, and 1ylrC mutant strains, as well as the triple gene deletion 1ylrABC mutant strain were intermediate between the parental and T3SS mutant strains (**Figure 2A**). This intermediate in vitro infection phenotype is similar to that observed in Yersinia mutants individually deleted

FIGURE 4 | YlrB is translocated into eukaryotic cells. HeLa cells were infected with Yp589 (Y. pestis strain with deletion of YopE, YopJ, YopM, SycT, and YpKA) carrying YscF1-Bla (possessing the yscF 5 ′ UTR and the initiation codon), pMM85 YopM-Bla, YlrA-Bla, YlrB-Bla, or YlrC-Bla and were then incubated at 37◦C for 3 h. CCF2-AM was added to HeLa cells and visualized by fluorescence microscopy. HeLa cells infected with the strain carrying the YlrB-Bla were positive for cytoplasmic β-lactamase activity (blue fluorescence), indicating that these cells had been injected with YlrB-Bla. Shown is a single representative experiment performed multiple times.

for the YopE and YopH T3SS effectors (Bartra et al., 2001). These data validated the findings of the TraSH screen indicating that the Ylr-encoding genes promote the survival of Y. pestis during their interaction with macrophages.

The Y. pestis 1ylrABC strain was tested in a mouse-based infection model to determine whether the Ylr-encoding loci promote the pathogen survival in vivo. A competition infection assay was used to compare the proliferation of the parental KIM5 strain to that of the mutant strain within individual mice. Accordingly, in the control experiment mice were infected with an equal mixture of the parental KIM5 and the isogenic T3SS mutant 1yopB strains and after 2 days the mice were humanely sacrificed and the presence of each strain in the spleen was determined by CFU assay. In contrast to the equal ratio of the parental and 1yopB stains in the "input" (i.e., the dosed inoculum), the parental strain was greatly over-represented in the "output" (i.e., the splenic homogenates derived from mice infected for 2 days) (**Figure 2B**). Mice similarly infected with an equal mixture of the parental KIM5 and the isogenic 1ylrABC strains also yielded a statistically significant overrepresentation of the parental strain following 2 days of infection (**Figure 2B**). There was no detectable differences in growth rates between the parental KIM5 and the isogenic 1ylrABC strains (**Supplementary Figure 1**). Collectively these data indicate that the ylr locus promotes Y. pestis infection both in cells and in mice.

### YlrA, Ylrb, and Ylrc Are Secreted by the *Y. pestis* Ysc Type III Secretion System

Sequence analysis of the ylrA, ylrB, and ylrC genes indicated that they each potentially encode at their amino-terminus a recognition sequence for the type III secretion system (T3SS)

(Hu et al., 2016). To determine if the YlrA, YlrB, and YlrC proteins are secreted by the T3SS, expression vectors containing full-length C-terminal FLAG-tagged proteins were constructed. In addition, DNA sequences predicted to encode YlrA, YlrB, and YlrC N-terminal T3SS signal residues 2 to 10 were deleted from each of the expression vectors. Generated plasmids were placed in Y. pestis KIM814, a strain containing a modified pCD1 virulence plasmid that contains all of the genes necessary to assemble a functional T3SS but lacks all other effector proteins (Bartra et al., 2006). The strains were tested for expression and secretion in the presence and absence of 2.5 mM of calcium and bacterial pellet and supernatant were analyzed. FLAGtagged YlrA, YlrB, and YlrC were expressed in the presence and absence of calcium, but were secreted only in the absence of calcium (**Figure 3**; red boxes; and **Data Sheet 1**). There was no secretion of the YlrA, YlrB, and YlrC proteins when the putative T3SS signal of each protein was deleted. Finally, there was no secretion of YlrA-FLAG, YlrB-FLAG, or YlrC-FLAG detected by Y. pestis strain KIM8 (pCD1-) or Y. pestis 1yscF, both of which are unable to assemble a functional Ysc injectisome (**Supplementary Figures 2–4**). Together, these results demonstrate that YlrA, YlrB and YlrC are T3S substrates and that they are recognized and secreted by the Y. pestis T3SS.

### YlrB Is Translocated Into Mammalian Cells

A β-lactamase reporter system was used to determine whether Ylr proteins are translocated into cultured cells during infection (Marketon et al., 2005). Upon diffusion into cultured cells, CCF2-AM is cleaved into the membrane-impermeable CCF2 by endogenous cytoplasmic esterases, emitting a green fluorescence. When β-lactamase fusion proteins are injected into the cells, CCF2 can be cleaved further, changing the fluorescent emission to blue (Charpentier and Oswald, 2004). Cells that were either uninfected or infected with a Y. pestis strain expressing YscF1- Bla, which is expressed but not secreted by the T3SS, retained their initial green fluorescence (**Figure 4**). In striking contrast, essentially all CCF2-labeled cells infected with a Y. pestis strain expressing YopM-Bla, which, as discussed in the Introduction, is a plasmid-encoded virulence factor, fluoresce blue indicating that YopM-Bla is readily translocated into the infected cell cytosol. Cells infected with Y. pestis strains expressing either YlrA-Bla or YlrC-Bla retained their green fluorescence indicating that these proteins were not detectably translocated as assayed by this method despite being readily expressed as assayed by western blotting (**Figure 4**; **Supplementary Figure 5**). In contrast, blue fluorescing cells were readily detected following infection with a Y. pestis strain expressing YlrB-Bla. These show that the YlrB is translocated by Y. pestis during infection.

### Intracellular Activity of Ylr

Previous studies of the Yersinia Yop virulence factors have shown a robust correlation between their growth inhibition in yeast and their activity in mammalian cells (Lesser and Miller, 2001; Wiley et al., 2009). Yeast provide a relatively simple genetic model system for the study of the eukaryotic genes and is based on the premise that yeast share many molecular, genetic, and biochemical features with mammalian

experiment performed three times with similar outcomes.

cells (Zhao and Lieberman, 1995). Accordingly, the fission yeast Schizosaccharomyces pombe was transformed with either an empty vector control plasmid or inducible expression plasmids encoding either YlrA, YlrB, or YlrC and plated either on media in which the Ylr proteins were not expressed ("uninduced") or on media in which Ylr expression occurs ("induced"). All S. pombe transformant strains grew equally well on uninduced media (**Figure 5**). However, on induced media, there was a marked differences between the relatively rapid growth of the empty vector and YlrB transformants (rows 1 and 6) and that of the relatively poorer growth of the YlrA and YlrC transformants (rows 2 and 7). Previous work has shown that IpaH effectors and SspH2 have a conserved catalytic cysteine residue that is absolutely required for ubiquitin ligase activity (Rohde et al., 2007; Quezada et al., 2009). To determine if this residue and other conserved residues of the NEL domain of YlrA will affect its growth-inhibiting activity in yeast, these residues were substituted generating plasmids expressing YlrA-C386S, YlrA-R389A, and YlrA-S449F. There was a clear reduction in growth inhibition for the S. pombe strain expressing the YlrA-C386S variant compared to the strain expressing wild-type YlrA (**Figure 5**, rows 2 and 3). In contrast, there was no detectable loss of growth inhibitory activity in the S. pombe strains expressing either the YlrA-R389A or YlrA-S449F (rows 4 and 5). These data suggest that the ubiquitin ligase activity of YlrA is required for its growth inhibitory activity in eukaryotic cells.

#### DISCUSSION

In this study, we demonstrate that the Y. pestis chromosomallyencoded proteins YlrA, YlrB, and YlrC are secreted and recognized by the T3SS and are required for the optimal survival of this pathogen both in the presence of cells as well as in mice. Although many chromosomally-encoded Y. pestis virulence factors have been described, to the best of our knowledge the Ylr are the first chromosomally-encoded virulence factors secreted by the pCD1-encoded T3SS.

YlrA, YlrB, and YlrC are leucine rich repeat proteins (LRR) with (YlrA and YlrC) or without (YlrB) an E3 ligase domain (NEL). The Ylr proteins belong to a diverse group of molecules distinguished by a consensus sequence consisting predominantly of leucines, hence the term leucine-rich repeat. Similar proteins are found in other Gram-negative bacteria that possess T3SSs, including Yersinia (YopM) (Kobe and Deisenhofer, 1994), Shigella (IpaH family proteins) (Ashida et al., 2007), Salmonella (SspH1, SspH2 and SlrP) (Miao et al., 1999; Tsolis et al., 1999; Bernal-Bayard et al., 2010) and other bacterial pathogens. In addition to possessing detectable membrane-penetrating activity, LRR domains offer adaptable structural frameworks that mediate protein-protein interactions, coupling enzymatic domains to substrate-binding domains, function to determine substrate specificity, and to suppress the enzymatic activity of the NEL domain prior to substrate binding (Kobe and Kajava, 2001; Quezada et al., 2009; Rüter et al., 2010). The NEL domains function as ubiquitin E3 ligases which mimic the activities of host E3 ubiquitin ligases and ubiquitinate specific target proteins (Rohde et al., 2007; Quezada et al., 2009). YlrA and YlrC contain an N-terminal LRR domain and a predicted C-terminal novel E3 ligase domain (NEL) (Hu et al., 2016). E3 ligase activity has been demonstrated for the Shigella effectors IpaH4.5 and IpaH9.8 and Salmonella effectors SspH1, SspH2, and SlrP (Rohde et al., 2007; Singer et al., 2008; Zhu et al., 2008; Bernal-Bayard and Ramos-Morales, 2009; Quezada et al., 2009). Bacterial E3 ubiquitin ligases transported into host cells mimic the activities of host E3 ubiquitin ligases and ubiquitinate specific target proteins. For example, IpaH 4.5 appears to dampen the innate immune system of human cells by inhibiting nuclear factor KB (NF-KB) signaling in response to an intracellular infection with Shigella (Ashida et al., 2010). The Salmonella virulence effector SlrP targets the mammalian thioredoxin-1 (Trx) leading to ubiquitination of Trx and causes a decrease in its redox activity (Bernal-Bayard and Ramos-Morales, 2009).

We have demonstrated that YlrB is translocated into the eukaryotic cell, whereas there was no evidence of translocation of the YlrA and YlrC using a β-lactamase reporter system. These results are difficult to interpret. Both full-length YlrA and YlrC are expressed and secreted, albeit at lower levels than YlrB (see **Figure 3** and **Supplementary Figure 5**); therefore we expect them to be translocated as well. In addition, the fact that they contain an E3 ligase domain suggests that these proteins work inside the eukaryotic cell. Lack of detectable translocation in our experiments could be due to the large size of the proteins (∼600 amino acids). Further experiments are needed using smaller-sized constructs expressing the translocation domains of both proteins. The most compelling data we present that YlrA and YlrC actually function within eukaryotic cells is the fact that they both disrupt the growth of the fission yeast S. pombe. Furthermore, to test if the conserved catalytic cysteine residue in YlrA is important to the activity of the protein, it was replaced with a serine residue. This resulted to the loss of the growth-inhibiting activity of YlrA. We speculate that the deleterious effect of YlrA and YlrC is due to the activity of their NEL domain. In a study using a yeast model, Rohde et al. (2007) have shown that S. flexneri IpaH9.8 inhibit yeast pheromone response signaling by ubiquitination and by promoting proteasome-dependent destruction of the mitogen-activated protein kinase kinase (MAPKK) Ste7. It was also demonstrated in the same study that the cysteine residue conserved in the NEL domain of all IpaH family members is required for the IpaH9.8 activities in yeast. The Salmonella effector SspH2 also has a conserved cysteine residue (Quezada et al., 2009). This catalytic cysteine residue and all other NEL active site residues are absolutely conserved between YlrA, YlrC and SspH2. We conclude that both YlrA and YlrC are strong candidates to join the list of bacterial virulence proteins that target the mammalian ubiquitination system.

To date, little is known about the leucine-reach repeat proteins in Yersinia with the exception of YopM. Miao et al. (1999) described three leucine-rich repeat open reading frames (ORFs) in Y. pestis that share significant homology with SspH2. These ORFs seem to be ylrA, ylrB and ylrC, since the ORFs were described to be present in an operon-like structure in the chromosome. Later, Evdokimov et al. (2001) described that these three ORFs share similar structure with Salmonella effectors SlrP, SspH1, SspH2; Shigella effector IpaH; and Yersinia effector YopM. Chou et al. (2012) PCR amplified Y. pestis y3400 (ylrA), expressed and purified the previously uncharacterized protein in E. coli and demonstrated that the purified protein possessed constitutive, not autoinhibited, ubiquitin E3 ligase activity. Finally, Soundararajan et al. (2011) were the first to identify and analyze the structure-function relation of a protein which appears to be the protein product of y3397 (YlrC) (McPhee and Bliska, 2011). More systematic studies are needed to characterize the expression, secretion, translocation, molecular functions and host target of these novel effector proteins. Research on these ylr gene products will lead to a better understanding of Y. pestis pathogenesis and possibly to the identification of the proteins as novel vaccine candidates or potential targets for therapeutics.

In conclusion, the ylrA, ylrB, and ylrC gene products represent the first chromosome-encoded T3S effector proteins of Y. pestis and are required for the optimal survival of this pathogen in the presence of macrophages. The gene products share significant similarity with T3S effector proteins from other

#### REFERENCES


bacterial pathogens and hence they are strong candidates to join the list of bacterial virulence proteins.

#### AUTHOR CONTRIBUTIONS

SS, CL, XG, WB, ZL, GP, and KS planned and performed the majority of the experiments. HB and RH provided technical assistance. LQ analyzed the results of the TraSH screen. SS, CL, and KS wrote the manuscript.

#### FUNDING

This study was funded in part by NIH AI119450 (GP/KS) and U.S. Army Research, Development and Engineering Command (RDECOM), Contract W911SR-07-C-0084 (ZL).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00023/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Schesser Bartra, Lorica, Qian, Gong, Bahnan, Barreras, Hernandez, Li, Plano and Schesser. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Heterologous Complementation Studies With the YscX and YscY Protein Families Reveals a Specificity for *Yersinia pseudotuberculosis* Type III Secretion

#### *Edited by:*

Sophie Bleves, Aix-Marseille Université, France

#### *Reviewed by:*

Luís Jaime Mota, Faculdade de Ciências e Tecnologia da Universidade Nova de Lisboa, Portugal Eric Faudry, CEA Grenoble, France

> *\*Correspondence:* Matthew S. Francis matthew.francis@umu.se

#### †*Present Address:*

Tiago R. D. Costa, MRC Centre for Molecular Bacteriology and Infection, Department of Life Sciences, Imperial College London, London, United Kingdom

> *Received:* 16 December 2017 *Accepted:* 28 February 2018 *Published:* 16 March 2018

#### *Citation:*

Gurung JM, Amer AAA, Francis MK, Costa TRD, Chen S, Zavialov AV and Francis MS (2018) Heterologous Complementation Studies With the YscX and YscY Protein Families Reveals a Specificity for Yersinia pseudotuberculosis Type III Secretion. Front. Cell. Infect. Microbiol. 8:80. doi: 10.3389/fcimb.2018.00080 Jyoti M. Gurung1,2, Ayad A. A. Amer 1,2, Monika K. Francis 1,2, Tiago R. D. Costa1,2† , Shiyun Chen<sup>3</sup> , Anton V. Zavialov <sup>4</sup> and Matthew S. Francis 1,2 \*

<sup>1</sup> Department of Molecular Biology, Umeå University, Umeå, Sweden, <sup>2</sup> Umeå Centre for Microbial Research, Umeå University, Umeå, Sweden, <sup>3</sup> Key Laboratory of Special Pathogens and Biosafety, Wuhan Institute of Virology, Chinese Academy of Sciences Wuhan, Wuhan, China, <sup>4</sup> Department of Chemistry, University of Turku, Turku, Finland

Type III secretion systems harbored by several Gram-negative bacteria are often used to deliver host-modulating effectors into infected eukaryotic cells. About 20 core proteins are needed for assembly of a secretion apparatus. Several of these proteins are genetically and functionally conserved in type III secretion systems of bacteria associated with invertebrate or vertebrate hosts. In the Ysc family of type III secretion systems are two poorly characterized protein families, the YscX family and the YscY family. In the plasmid-encoded Ysc-Yop type III secretion system of human pathogenic Yersinia species, YscX is a secreted substrate while YscY is its non-secreted cognate chaperone. Critically, neither an yscX nor yscY null mutant of Yersinia is capable of type III secretion. In this study, we show that the genetic equivalents of these proteins produced as components of other type III secretion systems of Pseudomonas aeruginosa (PscX and PscY), Aeromonas species (AscX and AscY), Vibrio species (VscX and VscY), and Photorhabdus luminescens (SctX and SctY) all possess an ability to interact with its native cognate partner and also establish cross-reciprocal binding to non-cognate partners as judged by a yeast two-hybrid assay. Moreover, a yeast three-hybrid assay also revealed that these heterodimeric complexes could maintain an interaction with YscV family members, a core membrane component of all type III secretion systems. Despite maintaining these molecular interactions, only expression of the native yscX in the near full-length yscX deletion and native yscY in the near full-length yscY deletion were able to complement for their general substrate secretion defects. Hence, YscX and YscY must have co-evolved to confer an important function specifically critical for Yersinia type III secretion.

Keywords: T3S chaperone, secretion hierarchy, substrate sorting, LcrH/SycD, YscV, protein–protein interaction

## INTRODUCTION

Type III secretion (T3S) is an effective means for many different Gram-negative bacteria to deliver proteins into diverse eukaryotic cell types. This is a common virulence mechanism of many harmful pathogens (Buttner, 2012; Portaliou et al., 2016), but is also useful for bacteria relying on a symbiotic lifestyle or for survival in the environment (Pallen et al., 2005), and for the biosynthesis of the flagella apparatus (Erhardt et al., 2010). Upon artificial induction, a T3S system (T3SS) can also secrete substrates into laboratory culture media and this requires a complex of ∼20 proteins that all together span both bacterial membranes (inner and outer), the peptidoglycan layer and which also protrudes out from the surface. This apparatus has been purified from a few different bacteria, taking on the appearance of a syringe equipped with needle (often termed the "needle complex") (Kubori et al., 1998; Blocker et al., 1999, 2001; Kimbrough and Miller, 2000; Tamano et al., 2000, 2002; Sekiya et al., 2001; Sukhan et al., 2001; Journet et al., 2003; Marlovits et al., 2004; Mueller et al., 2005). Generally, protein cargo is thought to be secreted through the needle complex (Radics et al., 2014), and these can be divided into three categories—"early" substrates that make up the outer needle, "middle" substrates comprising the pore-forming translocon proteins and "late" host-modulating effector proteins that with assistance from the translocon are targeted to the eukaryotic cell interior (Osborne and Coombes, 2011; Buttner, 2012; Dewoody et al., 2013).

About 10 proteins exist in the needle complex that are common to all known non-flagella and flagella T3SSs (Francis et al., 2004; Buttner, 2012; Portaliou et al., 2016). Moreover, thorough analysis of sequence similarities can classify all T3SSs into evolutionary distinct nodes (Troisfontaines and Cornelis, 2005; Abby and Rocha, 2012). Among the non-flagella T3SSs, a major node is the inv/spa system encoded by the Salmonella pathogenicity island 1 (SPI-1) of Salmonella enterica and including systems from Shigella sp. (mxi/spa), Escherichia coli (evi/epa) and Burkholderia sp. (inv/spa). Another major node is characterized by the ysc system, encoded on a common virulence plasmid of human pathogenic Yersinia sp., and includes systems found in Pseudomonas aeruginosa (psc), selected Aeromonas sp. (asc), Photorhabdus luminescens (lsc/sct) and certain Vibrio sp. (vsc). Unique to this node are a number of genetically distinct T3SS components not represented in other nodes. Two examples of this are the components YscX and YscY.

The functional role of YscX and YscY in T3S remains enigmatic. In Yersinia, a mutant lacking either yscX or yscY allele poorly synthesizes Yop substrates and their secretion is abolished (Iriarte and Cornelis, 1999; Day and Plano, 2000; Bröms et al., 2005). Moreover, YscX is a T3SS substrate that prior to secretion is stabilized in the cytoplasm by the YscY chaperone (Iriarte and Cornelis, 1999; Day and Plano, 2000). Interestingly, an in silico analysis revealed that YscY possesses three tandom tetratricopeptide repeats (Pallen et al., 2003) that are important for function of the translocator class of T3S chaperones (Bröms et al., 2006; Edqvist et al., 2006; Buttner et al., 2008; Lunelli et al., 2009; Job et al., 2010; Singh et al., 2013; Kim et al., 2014). These studies taken together indicate that YscX could be a structural component of the needle complex, while YscY would be necessary to stabilize pre-made pools of YscX and to ensure correct temporal YscX secretion. However, a recent study could find no support for YscX association with the needle, but rather together with YscY was found associated with YscV (alternatively known as LcrD), a core structural inner membrane component of all T3SSs (Diepold et al., 2012).

In a study of P. aeruginosa T3S, it was reported that the YscY homolog, Pcr4 (from here on termed PscY for consistency), is actually the secreted component, and not the YscX homolog, Pcr3 (here on termed PscX) (Yang et al., 2007). Despite this anomaly, the YscX/PscX and YscY/PscY components otherwise appear to serve somewhat analogous functions in their respective T3SSs (Bröms et al., 2005; Yang et al., 2007). Moreover, expression of yscX and yscY in the corresponding P. aeruginosa mutants can efficiently restore T3S (Bröms et al., 2005). Yet specific differences must exist since neither pscX nor pscY alone, or even when expressed in combination, could complement the T3S defects observed in Yersinia mutants lacking the corresponding alleles (Bröms et al., 2005). A known difference is the YscY interaction with the LcrH chaperone (also known as SycD) that has an undisclosed role in T3SS regulation in Yersinia (Francis et al., 2001; Bröms et al., 2005); the equivalent interaction between PscY and the PcrH chaperone in P. aeruginosa has not been detected (Broms et al., 2003; Bröms et al., 2005). Moreover, it is not yet known whether PscX and PscY complex with PscV (the YscV homolog) as do the three equivalent Yersinia components (Diepold et al., 2012).

Hence, in this study we further investigated the notion that an YscX-YscY complex has evolved molecular attributes unique to T3SS function in Yersinia. We analyzed T3S from Yersinia yscX or yscY mutants ectopically expressing in trans homologous alleles from Aeromonas salmonicida and A. hydrophila (termed ascXAs/ascXAh and ascYAs /ascYAh), P. luminescens (termed sctX or sctY), V. harveyi and V. parahaemolyticus (termed vscXVh /vscXVp and vscYVh /vscYVp) or P. aeruginosa (PscX and PscY). No combination resulted in restoration of T3S, despite confirmation of reciprocal interactions giving rise to the equivalent of YscX-YscY bipartite and YscX-YscY-YscV tripartite complexes. This suggested that the YscX-YscY complex serves an exclusive T3SS function in Yersinia.

#### MATERIALS AND METHODS

#### Strains, Plasmids, and Growth Conditions

Bacterial strains and plasmids used in this study are listed in Table S1. Bacteria were routinely cultivated in Lysogeny broth (LB) (Bertani, 2004) or LB agar at either 26◦C (Yersinia pseudotuberculosis) or 37◦C (E. coli) with aeration. When required, the antibiotics carbenicillin (Cb), kanamycin (Km) and gentamicin (Gm) were added to laboratory culture media at the final concentrations of 100µg per ml, 50µg per ml and 20µg per ml, respectively. Analysis of T3SS by Y. pseudotuberculosis occurred at 37◦C in Brain Heart Infusion (BHI) broth. Media containing Ca2<sup>+</sup> ions was the non-inducing condition (BHI supplemented with 2.5 mM CaCl2), while media devoid of Ca2<sup>+</sup> ions was the inducing condition (BHI supplemented with 20 mM MgCl<sup>2</sup> and 5 mM Ethylene glycol-bis-(β-aminoethyl ether)-N,N,N',N'-tetraacetic acid). To stimulate promoter activity from the expression vectors pMMB67EHgm and/or pMMB208, isopropyl β-D-1 thiogalactopyranoside (IPTG) at a final concentration of 0.4 mM was added.

The Saccharomyces cerevisiae reporter strain AH109 or Y190 was maintained by growth at 30◦C in YEP broth [2% (w/v) peptone, 1% (w/v) yeast extract, 2% (v/v) glucose] or agar [YEP broth with 2% (w/v) agar]. pGBKT7 derived plasmids were sustained in yeast by growth on SD synthetic minimal medium (0.67% (w/v) Yeast nitrogen base without amino acids, 300 µgml−<sup>1</sup> L-isoleucine, 1.5 mgml−<sup>1</sup> L-valine, 200 µgml−<sup>1</sup> L-adenine hemisulphate salt, 200 µgml−<sup>1</sup> L-arginine HCl, 200 µgml−<sup>1</sup> L-histidine HCl monohydrate, 1 mgml−<sup>1</sup> L-leucine, 300 µgml−<sup>1</sup> L-lysine HCl, 200 µgml−<sup>1</sup> L-methionine, 500µgml−<sup>1</sup> L-phenylalanine, 2 mgml−<sup>1</sup> L-threonine) while pBridge derived plasmids were grown on the same media minus methionine. Yeast containing pGADT7 derivatives were cultured on similar SD media with the exception that leucine was replaced with 200 µgml−<sup>1</sup> L-tryptophan.

#### PCR Amplification and Sequence Analysis

Amplified DNA fragments were obtained by PCR using the appropriate oligonucleotide combinations listed as online Supplementary Information (Table S2). These were synthesized by either DNA Technology A/S (Aarhus, Denmark), TAG Copenhagen A/S (Copenhagen, Denmark) or Sigma-Aldrich Sweden AB (Stockholm, Sweden). Amplified fragments were confirmed to be mutation free by first cloning into pCR <sup>R</sup> 4- TOPO TA (Invitrogen AB, Stockholm, Sweden) or pTZ57R using the InsTAclone PCR cloning kit (Thermo Fisher Scientific, Gothenburg, Sweden) and then via commercial sequencing (Eurofins MWG Operon, Ebersberg, Germany or GATC Biotech AB, Solna, Sweden).

#### Phylogenetic Analysis of YscX and YscY Protein Families

Protein sequences corresponding to YscX and YscY protein family were mined from National Center for Biotechnology Information (https://www.ncbi.nlm.nih.gov/sites/entrez) and phylogenetic trees obtained by the Neighbor-Joining method using MEGA6 (Molecular Evolutionary Genetics Analysis) (Tamura et al., 2013). Briefly, proteins were aligned by ClustalW using default settings and the unaligned regions and gaps were trimmed. The significance of the phylogenetic grouping was assessed using confidence level for 1,000 replicates. The trees were rooted with genetically related Bordetella pertusis I475 YscX-like or YscY-like protein as the respective out-groups.

### Generation of Bacterial Protein Expression Constructs

Constructs containing the yscX and yscY alleles from Y. pseudotuberculosis and pscX and pscY alleles from P. aeruginosa have been described previously (Bröms et al., 2005). ascX and ascY were amplified from both A. salmonicida subsp. salmonicida JF2267 (a gift from Joachim Frey, Universität Bern, Switzerland) and A. hydrophila AH-3 (Juan Tomás, Universidad de Barcelona, Spain). vscX and vscY were amplified from both Vibrio harveyi BB120 (Debra Milton, Umeå University, Sweden) and V. parahaemolyticus RIMD2210633 (Tetsuya Iida, Osaka University, Japan). Finally, sctX and sctY were amplified from P. luminescens TT01 (Rif<sup>R</sup> ) (David Clarke, University of Bath, United Kingdom). For the complementation assay in Y. pseudotuberculosis, with one exception all alleles were cloned individually or together as a dual expression construct into EcoRI-BamHI digested pMMB67EHgm. (The vscY allele from V. harveyi was digested by BamHI-PstI.) To examine stable protein expression in Y. pseudotuberculosis, with two exceptions all PCR amplified alleles with a 5-prime FLAGTM epitope were cloned individually into BamHI-EcoRI digested pMMB208. (The yscY allele from Y. pseudotuberculosis and the vscY allele from V. harveyi was digested by PstI-BamHI.) The same approach was utilized to generate expression constructs of codon-optimized gene variants.

### Construction of Yeast Plasmids

To investigate reciprocal binding between the products of yscY, yscX and related alleles, a series of constructs were generated in the yeast two-hybrid vectors pGBKT7 and pGADT7 (Clontech Laboratories, Palo Alto, CA). Constructs expressing the yscY and pscY alleles fused to the GAL4 DNA binding domain in pGBKT7 are already described (Francis et al., 2001; Bröms et al., 2005). PCR amplified ascY, vscY, and sctY alleles were fused to the GAL4 DNA binding domain by cloning into EcoRI-BamHI digested pGBKT7. Plasmids expressing the yscX and pscX alleles fused to the GAL4 activation domain in pGADT7 are already established (Bröms et al., 2005). PCR amplified ascX, vscX, and sctX alleles were fused to the GAL4 activation domain by cloning into EcoRI-BamHI digested pGADT7.

Reciprocal interaction between YscX, YscY, YscV, and related family members were assessed with a yeast three hybrid approach by utilizing a three hybrid GAL4 DNA binding domain pBridge vector (Clontech Laboratories, Palo Alto, CA) and Gal4 activation domain pGADT7 vector. Constructs expressing yscY and related alleles were expressed as Gal4 DNA binding fusion protein by introducing into EcoRI-PstI digested pBridge downstream of ADH1 promoter. Additionally, the coding sequences of yscX and related alleles were cloned into NotI-BglII sites of pBridge downstream of MET25 promoter to finally yield pBridge-yscY-yscX and its variants. Likewise, yscV and related members were expressed as GAL4 activation fusion protein from pGADT7. While PCR-amplified yscV, ascV and sctV alleles were cloned into EcoRI-XhoI sites, PCR-amplified pscV and vscV were inserted into NdeI-BamHI sites of pGADT7.

### Yeast Transformation and the n-Hybrid Assays

For yeast two-hybrid assay, transformation of the S. cerevisiae reporter strain AH109 was performed as described earlier (Francis et al., 2000; Bröms et al., 2005; Amer et al., 2016). Protein interactions from multiple independent transformations were determined by measuring the activation of the ADE2 reporter gene activation and the HIS3 reporter gene during growth on tryptophan and leucine minus SD synthetic minimal medium also lacking either adenine or histidine, respectively. The latter also required the addition of 4 mM 3-aminotriazole in the growth media to overcome any risk of false positives (James et al., 1996). Analysis of protein stability in yeast was performed as previously described (Francis et al., 2000).

The yeast three-hybrid assay was performed in S. cerevisiae reporter strain Y190 as described previously (Carlo et al., 2007; Glass et al., 2015). Briefly, yeast were co-transformed with pBridge and pGADT7-based vectors and a master plate established by initial growth at 30◦C on SD synthetic minimal medium lacking leucine, tryptophan and methionine. The extent of protein interactions were then measured via the activation of the HIS3 reporter gene upon growth on the above mentioned medium also lacking histidine, but supplemented with 40 mM 3-aminotriazole (3-AT) to neutralize the inherent leakiness of the reporter. Equal amount of colony from each of the transformants were resuspended in water and subjected to 5 fold serial dilutions. An aliquot of each serial dilution was then transferred onto two sets of plates: (1) master plate (SD -Leu, - Trp, -Met) and (2) selective plate (SD −Leu, −Trp, −Met, −His + 40 mM 3-AT). The growth of yeast on the selective plate indicated a positive interaction. As a control, 1 mM methionine was used for near complete repression of MET25 repressible promoter (Her et al., 2003).

### Analysis of Gene Transcription by Qualitative RT-PCR

Total RNA was isolated from T3SS-induced Yersinia cultures as previously described (Carlsson et al., 2007a,b). Briefly, overnight cultures of Yersinia grown in a secretion permissive condition (BHI minus Ca2+) were diluted into 5 ml fresh media to an optical density (OD) at 600 nm of 0.1. After incubating at 26◦C for 30 min, cultures were shifted to 37◦C and grown for approximately 90 min to an OD at 600 nm of 0.4 to 0.8. RNA was immediately stabilized by mixing two volumes of RNA protect bacterial reagent (QIAGEN GmbH, Hilden, Germany) with one volume of the bacterial culture. Total RNA was isolated using the NucleoSpin RNA II method (Macherey Nagel, Düren, Germany) that included an on-column DNase treatment. For reverse transcription, 0.2 µg of total RNA was used to generate cDNA by RevertAid H Minus Reverse Transcriptase (RT) system (Thermo Scientific, Vilnius, Lithuania). RNA samples that were not treated with RT were used as negative controls to confirm the absence of contaminating DNA. PCR to confirm the presence of specific transcript was performed using gene-specific primers as listed in Table S2 and the cDNA generated above as template (Carlsson et al., 2007a,b).

Qualitative RT-PCR to determine the stability of mRNA transcript was performed as described elsewhere (Okan et al., 2006; Jeters et al., 2009; Chen and Anderson, 2011) with some modifications. Briefly, Yersinia cultures were grown in BHI media lacking Ca2<sup>+</sup> at 37◦C to an OD at 600 nm of 0.4 to 0.8. At time point zero, one volume of bacterial culture was removed and rifampicin to a final concentration of 100µg/ml was added to the bacterial culture to inhibit de novo RNA synthesis. Thereafter, one volume of bacterial cultures was taken at 5, 10, and 20 min time points. RNA was always stabilized by the addition of two volumes of RNA protect bacterial reagent (QIAGEN GmbH, Hilden, Germany). Isolation of total RNA, synthesis of cDNA and qualitative presence of transcript by PCR was performed as described above.

### Synthesis and Secretion of Type III-Secreted Substrates

Induction of type III substrate synthesis and secretion from Y. pseudotuberculosis was performed as previously described (Francis et al., 2000, 2001). All protein samples were normalized to the amount of bacterial cells at an optical density of 600 nm. Protein associated with whole bacteria was assessed by sampling from pelleted bacterial cultures. Total fraction contained proteins associated within intact bacteria and secreted to the culture medium. Sampling of the cell-free supernatant assessed the secreted protein levels. All protein fractions were separated by SDS-PAGE and subjected to immunoblotting. Detection of Yersinia substrates used rabbit polyclonal antisera raised against secreted YopE, YopD, and YopB (AgriSera AB, Vännäs, Sweden).

#### Quantification of Protein Production in *Y. pseudotuberculosis*

Relative protein levels were quantified from protein bands on scanned western blot X-ray films using the gel analysis tool in ImageJ (Schneider et al., 2012). In every case, the lane profile plot area of each protein band of interest was normalized to the corresponding protein band appearing in the same lane in the loading control blot.

## RESULTS

### Amino Acid Identity Among the YscX and YscY Protein Families

In silico analysis has identified up to 9 core structural components in all T3SSs (Francis et al., 2004). Additional common components restricted to genetically related T3SS sub-families can also be found. A genetically related T3SSs exists in the notable human and animal pathogens Y. pestis, Yersinia enterocolitica, Y. pseudotuberculosis, and P. aeruginosa, the marine pathogens Aeromonas hydrophilia, A. salmonicida, V. harveyi, and V. parahaemolyticus, and the insect pathogen P. luminescens. Components common only to these systems include the YscX and YscY protein families. To ascertain genetic relatedness between representatives of the YscX and YscY protein families, amino acid sequences were retrieved from the sequenced genomes at http://www.ncbi.nlm.nih.gov/sites/entrez of the above mentioned bacteria and then aligned with ClustalW (http://www.ebi.ac.uk/Tools/clustalw/index.html). The evolutionary pattern was then examined by generating a phylogenetic tree rooted against the genetically related YscXlike and/or YscY-like protein from B. pertusis I475 by using the neighbor-joining method (Tamura et al., 2013; Bhattacharyya et al., 2017).

Based on amino acid sequence, the percent amino acid identity to YscX (122 aa) ranged from 54.1% over a 122 residue overlap (A. salmonicida) to 36.5% across a 126 residue overlap (V. parahaemolyticus) (**Figure 1A**). Apart from the near complete identity of YscX among the three Yersinia species, most relatedness occurred between the AscX proteins produced by the two Aeromonas species (∼97% identical) and the VscX proteins produced by the two Vibrio species (∼75%) (**Figure 1A**). Consistent with this, a phylogenetic tree analysis revealed that they each formed a separate clade distinct from other homologs (**Figure 1C**).

Similarly, amino acid sequence identity to YscY (114 aa) from Yersinia ranged between 52.6% (A. hydrophila) over a 116 residue overlap down to 32.5% (V. harveyi) over a 114 residue overlap (**Figure 1B**). Apart from the YscY proteins, the AscY proteins displayed most similarity to each other (∼91% identical), followed by the VscY proteins (∼68%), consistent with them forming separate phylogenetic clades (**Figure 1D)**. In addition, the P. luminescens proteins SctX and SctY were most similar to the AscX and AscY counterparts from Aeromonas species (∼63 and ∼61%, respectively) (**Figures 1A,B)** and they all grouped close together on the phylogenetic tree (**Figures 1C,D)**. Finally, among the three human pathogenic Yersinia species, YscY amino acid sequence was identical, whereas in YscX the amino acid glutamine at position 29 in both Y. pseudotuberculosis and Y. pestis existed as an aspartate residue in Y. enterocolitica (Figure S1). This subtle sequence divergence of YscX from Y. enterocolitica is reflected in the phylogenetic tree (**Figure 1C**).

In summary, this analysis reveals that YscX and YscY generally share most genetic relatedness with equivalent proteins from the two Aeromonas sp. Moreover, the YscX and YscY homologs that have diverged the most stem from the two Vibrio sp. This is in concordance with the phylogeny of non-flagella-T3SS of other core Ysc structural components (Troisfontaines and Cornelis, 2005; Romano et al., 2016).

### Reciprocal Chaperone-Substrate Interactions

A cornerstone of YscX and YscY function appears to be their ability to interact with each other (Day and Plano, 2000; Bröms et al., 2005), and this seems true also of PscX and PscY function in P. aeruginosa T3S (Bröms et al., 2005; Yang et al., 2007). The yeast two-hybrid system has proven to be a reliable tool to demonstrate reciprocal interactions of YscX/PscX with YscY/PscY (Francis et al., 2001; Bröms et al., 2005). Thus, we utilized this approach to examine the interaction reciprocity between the representative YscX and YscY family members detailed in **Figure 1**. The PCR amplified yscX-related alleles were all separately cloned into the pGADT7 vector to establish fusions to the C-terminus of the GAL4 activation domain (AD). Additionally, the PCR amplified yscY-related alleles were all individually cloned into the pGBKT7 vector to establish fusions to the C-terminus of the GAL4 DNA binding domain (BD). The various pairwise combinations of vectors expressing a AD and BD fusion were established in S. cerevisiae AH109 containing an ADE2 and HIS3 reporter gene. Interactions between YscX-like and YscY-like proteins were determined by the ability of transformed yeast to grow on either minimal media lacking histidine (but supplemented with 3-aminotriazole at a final concentration of 4 mM) or adenine. Interaction specificity was confirmed by the failure of a particular strain to grow on this same media after being cured of either the AD or BD expressing plasmid (data not shown). Reciprocal chaperone (YscY-like)-substrate (YscX-like) interactions could be observed for every possible combination (**Figure 2**, Figure S2). However, YscY-like variants derived from Y. pseudotuberculosis, P. aeruginosa, Aeromonas species, and P. luminescens all interacted better with YscX-like proteins also derived from these same bacteria. In contrast, these proteins consistently displayed weaker interactions with their more distantly related counterparts originating from Vibrio species (**Figure 2**, Figure S2). Importantly however, this was not a reflection of reduced protein expression, since cognate proteins derived from V. harveyi and V. parahaemolyticus demonstrated strong reciprocal binding with each other (**Figure 2**, Figure S2). Hence, reciprocal interactions between YscY- and YscXlike proteins are all generated in a yeast two-hybrid system. Conservation of these interactions is suggestive of a common functional requirement in T3S.

#### Reciprocal Ternary Interactions Between YscX, YscY and the Inner Membrane Component YscV Indicate Formation of a Conserved Substrate Secretion Sorting Complex

Together YscX and YscY co-interact with YscV, a highly conserved T3SS component that makes up part of the inner membrane export apparatus (Diepold et al., 2012). This YscX-YscY-YscV tripartite complex helps to facilitate the temporal export of early T3S substrates (Diepold et al., 2012). We utilized the yeast three hybrid approach to verify that YscX and YscY together form a ternary complex with YscV (**Figure 3**). Y190 yeast competent cells were co-transformed with pBridge vector carrying PADH1-BD-YscY/pMET25-YscX and pGADT7 carrying pADH1-AD-YscV. Transformed yeast cells were grown on SD minimal medium lacking histidine (but supplemented with 40 mM 3-aminotriazole) either in the absence or presence of methionine as a means to control output from the leaky MET25 promoter (Carlo et al., 2007). Expression of yscX under the control of MET25 promoter can be repressed in the presence of 1 mM methionine and activated in methioninefree media. Based on the extent of yeast growth in the absence of methionine, we could verify the formation of a YscY-YscX-YscV complex (**Figure 3**, panel 7). Importantly, the interaction was comparatively diminished upon growing yeast transformants on media containing 1 mM methionine. As expected, YscX did not bind to YscV in the absence of YscY (unpublished data), nor did YscY bind to YscV in the absence of YscX (**Figure 3**, panel 5). Moreover, transformation of empty vectors, YscV or YscX-YscY alone did not permit auto-activation of HIS3 reporter gene. These data demonstrated the suitability of the yeast three hybrid system for detecting the YscY-YscX-YscV interaction. Hence, we then tested if YscY-like, YscX-like, and YscV-like proteins from P. aeruginosa, A. hydrophila, V. parahaemolyticus, and P. luminescens have evolved to maintain the tripartite interaction with their specific counterparts. A significant amount

homologs. Genetically related YscX-like and YscY-like protein sequences from Bordetella pertusis I475 were used as an outgroup. Values next to each branch node

represent bootstrap percentages for 1,000 replicates. The scale bar represents the number of amino acid substitutions per site.

of interaction between the cognate partners was evident in all YscY, YscX, YscV homologs tested (**Figure 3**, panel 16– 19). Given the ability of YscY, YscX, and YscV homologs to maintain tripartite interactions, we then tested the possibility of reciprocal interactions involving the different combinations of family members. A reciprocal interaction between the members of ternary complex was observed in all combinations (**Figure 3**, panel 8–15). Thus, conservation of this interaction suggests that YscY-like, YscX-like, and YscV-like proteins have coevolved to perform a common T3S function, which is to facilitate the temporal export of early T3S substrates (Diepold et al., 2012).

### Production of YscX and YscY Is a Strict Requirement for T3SS Assembly and Function in *Y. pseudotuberculosis*

Next we asked whether members of the YscX and YscY protein families could restore T3S function to a respective 1yscX or 1yscY null mutant of Y. pseudotuberculosis. To achieve this, the various PCR-amplified yscX- and yscY-like alleles were cloned either separately or together under the control of an IPTG inducible promoter in the expression vector pMMB67EHgm and then conjugated into the appropriate Y. pseudotuberculosis background. Following standardization against bacterial cell number, the level of T3S substrates associated with the total bacteria (a mixture of proteins contained within intact bacteria and secreted to the culture supernatant) and secreted free into the culture media was examined from bacteria grown in both T3S restrictive (BHI plus Ca2+) and permissive (BHI minus Ca2+) conditions. Expectantly, YscX when expressed in a 1yscX null mutant (**Figure 4A**) and YscY when expressed in a 1yscY null mutant (**Figure 4B**) restored functional T3S in Y. pseudotuberculosis. Moreover, YscX derived from Y. enterocolitica could readily complement the T3S defect in the 1yscX null mutant (data not shown). Furthermore, dual expression of both YscX and YscY together in a 1yscX, yscY

an intrinsic leakiness with the HIS3 reporter, 4 mM 3-aminotriazole was added to histidine dropout media to suppress false positives (James et al., 1996).

double mutant also restored functional T3S (**Figure 5**). On the other hand, no expressed YscX-like or YscY-like protein could restore functional T3S to a 1yscX null mutant (**Figure 4A**) or to a 1yscY null mutant (**Figure 4B**), respectively. Additionally, coexpression of the cognate substrate-chaperone pair did not help to restore T3S to a 1yscX, yscY double mutant (**Figure 5**). These data are consistent with an earlier study which concluded that PscX and PscY from P. aeruginosa could not complement the Yop synthesis and secretion defect apparent in a Y. pseudotuberculosis mutant lacking the yscX and yscY alleles respectively (Bröms et al., 2005). Taken all together, these data suggest that YscX-YscY interplay has functional consequences unique to T3S by Yersinia.

#### Codon Optimization Fails to Improve the Complementation Efficiency of Non-endogenous *yscX-* and *yscY-*Like Alleles

Complementation may have failed because the non-cognate alleles were not sufficiently expressed. Due to the absence of available antibodies, production of YscX-like and YscY-like proteins in Y. pseudotuberculosis was assessed by appending an N-terminal FLAGTM epitope, followed by immunoblotting using an anti-FLAGTM monoclonal antibody. Accumulated levels of recombinant PscX, VscXVp, SctX and to a much lesser extent VscXVh could be detected (Figure S3A). Additionally, accumulated levels of recombinant PscY, VscXVh, VscXVp, and to a lesser extent SctY could be detected (Figure S3B). However, accumulated levels of all four recombinant Asc proteins were undetectable under these assay conditions (Figure S3). A number of reasons could contribute to the wide disparity in heterologous protein production in Yersinia, although it can often be attributed to species-specific GC content and species-specific codon usage (Gustafsson et al., 2004; Yu et al., 2015; Zhou et al., 2016). Indeed, the percent GC content of the individual genomes targeted here do vary considerably—Y. pseudotuberculosis (47.6% GC content), Aeromonas sp. (61.6%), P. aeruginosa (66.6%), Vibrio sp. (45.4%), and P. luminescens (42.8%) (http://archaea.ucsc.edu/) (Schneider et al., 2006; Chan et al., 2012), and this variance can be assumed to tailor the codon usage within each individual allele that may be incompatible with optimal translation in Y. pseudotuberculosis (Gustafsson et al., 2004; Yu et al., 2015; Zhou et al., 2016).

Interestingly, an analysis of codon usage scores in the yscX and yscY-like gene families based on the codon adaption index (CAI) revealed the presence of codon usage biases of different genes (Figure S4). Hence, we synthesized commercially a series of synthetic alleles that were optimized for production in Y. pseudotuberculosis by actively considering (i) codon usage bias, (ii) GC content, and (iii) mRNA secondary structure. In all cases the decoded amino acid sequence remained identical to that encoded by the native gene sequences. In the optimized synthetic genes, the number of unfavorable codons was reduced by upgrading the CAI value to ∼0.9 and by adjusting the GC content more closely resembling that of Yersinia (Figures S4, S5). Predicted mRNA secondary structure was not affected by optimization (data not shown). To assess protein production, we cloned both native and synthetically optimized gene sequences as N-terminal FLAGTM-tagged variants into pMMB208 vector

were grown on a synthetic dropout (SD) agar plates lacking tryptophan (Trp) and leucine (Leu) for maintenance of the plasmid pairs and methionine (Met) to induce production of YscX-like proteins. The strength of interaction was studied by growing yeast in equivalent media lacking Histidine (His). A concentration of 40 mM 3-aminotriazole was added to reduce leaky HIS3 expression. The result shown is a representative of three independent experiments.

under the control of an IPTG-inducible promoter. Following introduction into Y. pseudotuberculosis and subsequent growth in BHI minus Ca2+, bacterial pellets standardized against bacterial cell number were examined for accumulation of FLAGTMtagged protein. Except for the SctX and VscYVh variants, codon optimization led to improved production of all the YscXlike and YscY-like variants with accumulated levels ranging between 1.3- and 7.0-fold more abundant then their native (non-optimized) counterparts (Figure S6). Interestingly, despite the successful efforts to codon optimize, accumulated levels of some recombinant proteins such as those originally derived from Aeromonas sp. still failed to reach a level close to that achieved for the native Yersinia proteins (**Figure 6**).

Significantly, this low level protein production was not due to an inferior transcriptional output because accumulated mRNA levels transcribed from all yscX-like (Figure S7A) and yscY-like (Figure S7B) genes was similar, which is consistent with transcription being driven by an identical promoter and transcriptional start site architecture in every case. Furthermore, the stability of accumulated ascX mRNA transcripts (Figure S7C) and ascY mRNA transcripts (Figure S7D) was equivalent to yscX and yscY mRNA transcripts. Thus, additional hitherto unknown regulatory elements must be limiting AscX and AscY production in Y. pseudotuberculosis.

Nevertheless, given that in most cases elevated production arose from codon optimization, we assessed if these codon

pMMB67EHgm-encoded YscY family members. Strains: Parent (YPIII/pIB102); 1yscY (YPIII/pIB890); complemented YPIII/pIB890, pMMB67EHgm (Vector); complemented YPIII/pIB890, pJEB292 (YscY+); complemented YPIII/pIB890, pJEB296 (PscY+); complemented YPIII/pIB890, pMF721 (AscY<sup>+</sup> As); complemented YPIII/pIB890, pMF723 (AscY<sup>+</sup> Ah); complemented YPIII/pIB890, pMF796 (VscY<sup>+</sup> Vh); complemented YPIII/pIB890, pMF726 (VscY<sup>+</sup> Vp); complemented YPIII/pIB890, pMF728 (SctY+). Molecular mass values shown in parentheses were deduced from primary amino acid sequences.

optimized YscX-like and YscY-like protein variants could restore T3S to the 1yscX null mutant and 1yscY nullmutant, respectively. The optimized yscX-like and yscY-like genes were cloned individually into pMMB67EHgm and Yops production associated with total bacterial culture and Yops secretion associated with bacterial supernatants were monitored in secretion permissive growth conditions. Once again however, Yops production and secretion could be restored only when the 1yscX null mutant was complemented with native YscX protein (**Figure 6A**) and 1yscY null mutant with native YscY protein (**Figure 6B**).

To be sure that all other codon-optimized homologs were sufficiently produced to permit complementation, we compared these levels to the minimal amount of YscX and YscY sufficient to restore T3S to the 1yscX and 1yscY null mutants respectively. Critically, an undetectable amount of native YscX and YscY ectopically produced during growth in the presence of 0.02 mM of IPTG was already sufficient to restore high capacity T3S to the 1yscX null mutant (**Figure 6C**) and 1yscY null mutant (**Figure 6D**), respectively. These levels were quite noticeably much less than accumulated amounts detected for all of the non-complementing YscX-like (**Figure 6E**) and YscY-like (**Figure 6F**) variants when bacteria were grown in the presence of 0.4 mM IPTG. From this data we infer that insufficient expression of non-native YscX- and YscY-like family members cannot explain their inability to function in Y. pseudotuberculosis.

#### Codon Optimized Alleles Do Not Suppress a Functional *Yersinia* T3SS

Despite the YscX- and YscY-like homologs being unable to substitute for the respective loss of yscX or yscY in Y. pseudotuberculosis, we still wondered if they possess any activity at all when expressed in Y. pseudotuberculosis. To assess this, we examined for the ability of produced homologs to exert a dominant negative effect on Yersinia T3SS. All of the codonoptimized homologs of YscX and YscY under control of the IPTG inducible promoter of pMMB67EHgm were introduced in parental Y. pseudotuberculosis harboring an intact and fully functional T3SS. Firstly, we examined the protein expression profile of YscX and YscY family members in parental Yersinia background and found them to be comparable to what we had previously observed in the knockout strain backgrounds (data not shown). We then assessed the impact of this production on T3SS function as measured by Yops production and secretion. Significantly, neither the production of codon-optimized YscXlike (**Figure 7A**) nor YscY-like (**Figure 7B**) homologs interfered with T3SS activity by parental bacteria. The complete absence of any dominant negative effect implies that none of the YscX-like or YscY-like proteins are recognized by the Yersinia T3SS,

YPIII/pIB881, pJEB335 (PscX, PscY+); complemented YPIII/pIB881, pMF733 (AscXAs, AscY<sup>+</sup> As); complemented YPIII/pIB881, pMF734 (AscXAh, AscY<sup>+</sup> Ah); complemented YPIII/pIB881, pMF797 (VscXVh, VscY<sup>+</sup> Vh); complemented YPIII/pIB881, pMF735 (VscXVp, VscY<sup>+</sup> Vp); complemented YPIII/pIB881, pMF736 (SctX, SctY+). Molecular mass values shown in parentheses were deduced from primary amino acid sequences.

which underscores that these proteins are truly non-functional in Yersinia bacteria. Taken altogether, our data indicates that an undisclosed feature(s) inherent in the native YscX and YscY products is necessary for proper T3S function in Yersinia, and one or more of these features cannot be substituted for by a genetically related gene from another bacterial source.

#### DISCUSSION

On the basis that a number of core T3SS basal body components associated with bacterial inner and outer membranes are highly conserved, these have been the focus of studies to establish at least eight phylogenetically distinct T3SS sub-families among sequenced bacteria (Troisfontaines and Cornelis, 2005; Abby and Rocha, 2012). In addition, there exist critical components that are present only in a restricted T3SS sub-family. For example, the poorly studied proteins YscX and YscY exist only in the Ysc T3SS sub-family, and the prototype of this family is the plasmid encoded Ysc-Yop system of human pathogenic species of Yersinia. In this study, we have reported on the evolutionary conservation among the YscX and YscY protein families. We also explored their function in two ways. The first approach examined the binary interactions between all YscX-like and YscY-like family members, and the ability of this YscX-YscY complex to interact with the core basal body component YscV. The second approach examined the ability of all YscX-like and YscY-like family members to complement a type III secretion defect in Y. pseudotuberculosis brought about by a deletion of yscX and/or yscY. Our focus was on six different homologs sourced from P. aeruginosa, A. hydrophila, A. salmonicida, V. harveyi, V. parahaemolyticus and P. luminescens. Despite showing robust and interchangeable binary YscX-YscY interactions and ternary YscX-YscY-YscV interactions, none of the homologs could restore T3S to Y. pseudotuberculosis lacking functional YscX and/or YscY. Thus, it seems that this YscX-YscY-YscV system is functionally conserved in Ysc-like T3SSs, but that each set exhibits functional specificity to assemble their own apparatus, and this could include recognition of their own set of substrates.

Events allowing YscX to form stable complexes with the known cognate chaperone YscY, and for the YscX-YscY complex to interact with the core T3SS structural component YscV, are considered to constitute integral elements of T3SS assembly and function (Day and Plano, 2000; Diepold et al., 2012). Our nhybrid analyses suggests that these interactions are a key attribute of all YscX-like, YscY-like and YscV-like homologs, serving an important T3SS assembly function that is evolutionarily conserved. In fact, the need to form a bipartite and a tripartite interaction is most likely an ancient role of these protein families, and maybe the reason why YscX and YscY could functionally substitute for the absence of PscX and PscY in mutants of P. aeruginosa (Bröms et al., 2005). However, merely having the ability to establish these known reciprocal interactions was clearly not enough to permit any of the homologs to mediate functional exchange in Y. pseudotuberculosis. Indeed this suggests that both YscX and YscY possess other molecular targets that are arguably unique to Ysc-Yop T3SS function. An approach to identify these putative targets within soluble bacterial lysate or solubilized bacterial membrane fractions would be to combine pull down technology using the Flag-YscX or Flag-YscY as a bait with mass spectroscopy detection. We can only assume that this need for additional molecular interactions arose as a consequence of Yersinia evolving the ability to thrive in a particular ecological niche. This notion is consistent with previous findings that correlate a diversification of bacterial T3SSs with known bacterial-host interactions (Abby and Rocha, 2012).

Sequence alignments revealed several obvious regions of uniqueness within the YscX and YscY amino acid sequence, respectively. At first glance, these are too numerous to perform targeted mutagenesis to screen for loss-of-function in the Yersinia proteins or gain-of-function in the heterologous proteins. Hence, we are now attempting to first map potentially critical functional domains unique to YscX and YscY, prior to embarking on a site-directed mutagenesis strategy. A series of chimeric alleles between yscX and yscX-like homologs as well as

complemented YPIII/pIB880, pJMG263 (FLAGTM-AscX<sup>+</sup> Ah.opt); complemented YPIII/pIB880, pJMG264 (FLAGTM-VscX<sup>+</sup> Vh.opt); complemented YPIII/pIB880, (Continued)

(VscYVp. +

FIGURE 6 | pJMG265 (FLAGTM-VscX<sup>+</sup> Vp.opt); complemented YPIII/pIB880, pMF266 (FLAGTM-SctX.<sup>+</sup> opt); 1yscY (YPIII/pIB890); complemented YPIII/pIB890, pMMB208 (Vector); complemented YPIII/pIB890, pMF800 (FLAGTM-YscY+); complemented YPIII/pIB890, pJMG267 (FLAGTM -PscY.opt+); complemented YPIII/pIB890, pJMG268 (FLAGTM-AscY<sup>+</sup> As.opt); complemented YPIII/pIB890, pJMG269 (FLAGTM-AscY<sup>+</sup> Ah.opt); complemented YPIII/pIB890, pJMG270 (FLAGTM-VscY<sup>+</sup> Vh.opt); complemented YPIII/pIB890, pJMG271 (FLAGTM-VscY<sup>+</sup> Vp.opt); complemented YPIII/pIB890, pJMG272 (FLAGTM-SctY<sup>+</sup> .opt); complemented YPIII/pIB880, pJEB291 (YscX+); complemented YPIII/pIB880, pJMG293 (PscX.<sup>+</sup> opt); complemented YPIII/pIB880, pJMG294 (AscX<sup>+</sup> As.opt); complemented YPIII/pIB880, pJMG295 (AscX<sup>+</sup> Ah.opt); complemented YPIII/pIB880, pJMG296 (VscX<sup>+</sup> Vh.opt); complemented YPIII/pIB880, pJMG297 (VscX<sup>+</sup> Vp.opt); complemented YPIII/pIB880, pJMG298 (SctX<sup>+</sup> .opt); YPIII/pIB890, pJEB292 (YscY+); YPIII/pIB890, pJMG286 (PscY.<sup>+</sup> opt); YPIII/pIB890, pJMG287 (AscYAs. + opt); YPIII/pIB890, pJMG288 (AscYAh. + opt); YPIII/pIB890, pJMG289 (VscYVh. + opt); YPIII/pIB890, pJMG290 (VscYVp. + opt); YPIII/pIB890, pJMG291 (SctY.<sup>+</sup> opt). Molecular mass values shown in parentheses were deduced from primary amino acid sequences.

opt); YPIII/pIB102, pJMG288 (AscYAh.

yscY and yscY-like homologs are being created. We proposed that by generating large N-terminal and C-terminal chimeric sets that contain progressively more native YscX sequence and native YscY sequence presents an alternative tool to define a minimal genetic content that confers to the chimera a functional competence for Y. pseudotuberculosis Ysc-Yop T3SS activity, which can then be subsequently confirmed using site-directed mutagenesis.

opt); YPIII/pIB102, pJMG287 (AscYAs.

opt); YPIII/pIB102, pJMG291 (SctY.<sup>+</sup>

It is likely that such sequence would serve as an interaction interface for engaging with one or more unknown targets, making it useful as a bait to identify this unique contributor to T3SS activity in Y. pseudotuberculosis. Some unique nonreciprocal interactions are already known—YscY interacts with the translocator class chaperone LcrH/SycD in Y. pseudotuberculosis (Francis et al., 2001; Bröms et al., 2005), while PscY (termed Pcr4) interacts with Pcr1 (the equivalent of TyeA in Yersinia) in P. aeruginosa (Yang et al., 2007). However, neither of these two interactions are prerequisites for the physical assembly of their respective T3SSs, but rather function primarily in the fine-tuning of substrate secretion control. In particular, LcrH/SycD influences the ability of YopD to bind both yop mRNA and the 30S ribosomal subunit to modulate translation initiation of yop mRNA and accelerate its turnover (Anderson et al., 2002; Chen and Anderson, 2011; Kopaskie et al., 2013). The consequence of this LcrH-dependent YopD RNA binding activity is to post-transcriptionally regulate T3SS genes to impart a level of hierarchal secretion control. Significantly, a strain defective in LcrH does not loose capacity to secrete substrates (Wattiau et al., 1994; Francis et al., 2001; Edqvist et al., 2006), and neither do defined mutants in LcrH that are unable to interact with

opt); YPIII/pIB102, pJMG289 (VscYVh.

opt). Molecular mass values shown in parentheses were deduced from primary amino acid sequences.

opt); YPIII/pIB102, pJMG290

YscY (Francis et al., 2001; Bröms et al., 2005). Thus, LcrH-YscY complex formation is likely to feed into the post-transcriptionally active complex of YopD, LcrH, and yop mRNA with the sole purpose to further fine-tune temporal and spatial control of substrate secretion. It is not anticipated to assist in assembly of a secretion-competent apparatus per se. On this basis, we do not believe that the specificity of LcrH-YscY interaction in Yersinia is the reason why heterologous protein expression cannot support T3S. Hence, we will continue to pursue evidence for a unique non-reciprocal interaction involving YscX and YscY that is essential for the physical assembly of the Ysc-Yop T3SS, rather than one involved in substrate secretion control.

Through sequence alignments we did observe that only native YscX contained two cysteine residues in codon positions that were absolutely conserved. At this stage we do not know if these cysteine residues are reduced or participate in inter- or intradisulfide bond formation. If they do participate in disulfide bond formation, this would be a unique feature of native YscX, and gives a reason as to why heterologous YscX-like proteins that lack these two conserved cysteines are not functional in Yersinia. It is well-established that disulfide bonds are an important structural element of the secretome of bacterial pathogens (De Geyter et al., 2016; Bocian-Ostrzycka et al., 2017). This is true also for human pathogenic Yersinia sp., where a role for thiol bonding is reported for the Ysc-Yop T3SS, the Caf1 chaperone-usher system, and the invasin autotransporter (Leong et al., 1993; Zav'yalov et al., 1997; Jackson and Plano, 1999; Mitchell et al., 2017). Thus, a logical next step is to use site-directed mutagenesis to investigate possible inter- or intra-disulfide bond formation in YscX.

Following completion of the needle assembly in Y. enterocolitica, YscX secretion into the extracellular milieu occurs (Diepold et al., 2012). However, the purpose of YscX secretion is enigmatic. Our ongoing studies have so far revealed that secretion of YscX is absolutely required for a functional T3SS in Y. pseudotuberculosis (unpublished data). One reason for YscX secretion might be to free up a complex of YscY-YscV to recognize and secrete other later subsets of T3S substrates, namely the translocators that are necessary to trigger the next stage of injectisome assembly and the host-modulating effectors that are injected into eukaryotic cells (Dewoody et al., 2013). In parallel, secreted YscX might comprise a minor component of the external needle, although experimental evidence to support this is completely lacking. Alternatively, secreted YscX may comprise a structural component within the needle assembly platform, such as via a cooperation with the YscI inner rod. Hence, to understand the role of YscX secretion in T3SS function is important. This pursuit is made all the more intriguing on the basis that PscY is the secreted component of the T3SS of P. aeruginosa, not PscX as one may have expected (Yang et al., 2007).

Given that all indicators point to the fact that YscX secretion is essential for T3SS function in Yersinia, it follows that a functional requirement for complementation of the 1yscX deletion mutant would likely be secretion of the corresponding YscX-like homologs. Taking account of the already stated PscY secretion exception (Yang et al., 2007), we are aware of only one other YscX-like homolog that has been proven experimentally to be a T3SS substrate—the AscX substrate of A. salmonicida (Vanden Bergh et al., 2013). We attempted to examine secretion of all YscX-like proteins by parental Y. pseudotuberculosis containing a fully functional Ysc-Yop T3SS. However, this was complicated by the need to produce all YscX-like variants as a FLAGTM fusion to facilitate their immune-detection, for this tag may impact on fusion protein secretion efficiency. Indeed, the amount of FLAGTM-YscX secreted was only a fraction of the amount that was produced (data not shown). Furthermore, we could not corroborate the findings of Vanden Bergh et al. (2013), for the yields of AscX were routinely too low to assess its secretion. Consistent with this, most other FLAGTM-tagged variants were not even detected in the secretion fraction, including PscX that is not a T3S substrate (Yang et al., 2007) (data not shown). The exception to this was the observation that FLAGTM-VscXVp and FLAGTM-SctX were active substrates of a fully intact Ysc-T3SS of Y. pseudotuberculosis (data not shown). Hence, at present we cannot establish a clear correlation between lack of secretion and lack of functional complementation.

One final observation of interest stemming from this study concerned the poor production of the two AscX proteins and the two AscY proteins in the Y. pseudotuberculosis background. This occurred despite (1) YscX and AscX as well as YscY and AscY sharing a sequence identity above the 50 percentile, (2) an abundance of comparatively stable mRNA being initiated from an identical promoter used to support the expression of all other genes studied in this report, and (3) the use of synthetic genes that had been codon optimized to maximize efficient translation in Y. pseudotuberculosis. Hence, some hitherto unknown phenomenon is responsible for the lack of heterologous production of these proteins in Yersinia. A straightforward reason for this is that they are normally stabilized by speciesspecific protein–protein interactions. The YscY chaperone is thought to function as an YscX stabilizer (Iriarte and Cornelis, 1999; Day and Plano, 2000). Indeed, this motivated the dual expression complementation experiment performed in this study. However, co-expression of AscX with cognate AscY did not increase the production of either protein. A conclusion from this work is that AscY alone is probably insufficient for AscX stabilization, meaning that the complex of AscX-AscY requires stabilizing influence from additional protein–protein interactions not supported by expression in a heterologous host like Yersinia. Another reason for poor AscX and AscY yield could also be that some aspect of mRNA sequence and/or structure defined by the ascX and ascY open reading frames influences that rate of their translation in Yersinia. As a means to pin-point the reason for low yield, we are currently analyzing allelic chimeras between ascX and yscX as well as ascY and yscY. This approach has the possibility to locate the genetic regions in AscX and AscY that are either (1) the interface for species-specific protein– protein interactions that are necessary for protein stability, or (2) the site of novel posttranscriptional control mechanisms that have potential to influence the spatial and temporal control of T3SS.

In summary, on the basis of complementation assays measuring the capacity for in vitro synthesis and secretion of Yops, it appears that the function of native YscX and YscY is evolutionary optimized to contribute unique and essential information for the physical assembly of the Ysc-Yop T3SS by pathogenic Yersinia sp. Thus, the challenge that lies ahead is to define the function of both YscX and YscY, and to decipher whether this function requires the recruitment of additional molecular components. In contrast, from this work we now know that the established bipartite YscX-YscY interaction and the tripartite YscX-YscY-YscV interaction is not a unique feature restricted only to native components of Yersinia. Rather, these interactions are preserved among all heterologous YscX-like and YscYlike family members, and this probably reflects an ancient and evolutionarily conserved universal function of these proteins.

#### AUTHOR CONTRIBUTIONS

JG, AA, MKF, TC, and MSF conceived and planned the experiments. SC and MSF contributed essential materials. JG, AA, MKF, and TC carried out the experiments and contributed to sample preparation. JG, AA, MKF, TC, SC, AZ, and MSF contributed to the interpretation of the results. JG and MSF took the lead in writing the manuscript. All authors provided critical feedback and helped shape the research, analysis, and manuscript.

#### REFERENCES


#### FUNDING

This work has been supported by grants from the Swedish Research Council (2009-5628 and 2014-2105), the Foundation for Medical Research at Umeå University and the Faculty of Science and Technology at Umeå University.

#### ACKNOWLEDGMENTS

This work was performed within the framework of the Umeå Centre for Microbial Research at Umeå University. We appreciate the contributions made by undergraduate students of the Umeå University Life science and Biomedicine programs. Andreas Diepold is acknowledged for valuable discussion surrounding YscX and YscY function. We express gratitude to Hans Wolf-Watz for antibodies to YopE, YopD, and YopB. We are also grateful to Joachim Frey, Juan Tomás, Debra Milton, Tetsuya Iida, and David Clarke for the gift of strains and/or chromosomal DNA.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00080/full#supplementary-material


secretion chaperone SicA in Salmonella enterica serovar Typhimurium. FEMS Microbiol. Lett. 350, 146–153. doi: 10.1111/1574-6968.12315


machinery: the needle part is changeable in length and essential for delivery of effectors. EMBO J. 19, 3876–3887. doi: 10.1093/emboj/19.15.3876


to regulate co-translational protein folding. Mol. Cell 59, 744–754. doi: 10.1016/j.molcel.2015.07.018


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Gurung, Amer, Francis, Costa, Chen, Zavialov and Francis. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Type VI Secretion Systems Present New Insights on Pathogenic Yersinia

Xiaobing Yang1,2†, Junfeng Pan2†, Yao Wang<sup>2</sup> and Xihui Shen1,2 \*

<sup>1</sup> State Key Laboratory of Crop Stress Biology for Arid Areas, College of Life Sciences, Northwest A&F University, Yangling, China, <sup>2</sup> Shaanxi Key Laboratory of Agricultural and Environmental Microbiology, College of Life Sciences, Northwest A&F University, Yangling, China

The type VI secretion system (T6SS) is a versatile secretion system widely distributed in Gram-negative bacteria that delivers multiple effector proteins into either prokaryotic or eukaryotic cells, or into the extracellular milieu. T6SS participates in various physiological processes including bacterial competition, host infection, and stress response. Three pathogenic Yersinia species, namely Yersinia pestis, Yersinia pseudotuberculosis, and Yersinia enterocolitica, possess different copies of T6SSs with distinct biological functions. This review summarizes the pathogenic, antibacterial, and stress-resistant roles of T6SS in Yersinia and the ion-transporting ability in Y. pseudotuberculosis. In addition, the T6SS-related effectors and regulators identified in Yersinia are discussed.

Keywords: pathogenic Yersinia, type VI secretion system (T6SS), function, effectors, regulation

#### Edited by:

Matthew S. Francis, Umeå University, Sweden

#### Reviewed by:

Cristina Elisa Alvarez-Martinez, Universidade Estadual de Campinas, Brazil Erwan Gueguen, Université de Lyon, France Christopher Alteri, University of Michigan, United States

> \*Correspondence: Xihui Shen xihuishen@nwsuaf.edu.cn

†These authors have contributed equally to this work.

> Received: 02 May 2018 Accepted: 13 July 2018 Published: 31 July 2018

#### Citation:

Yang X, Pan J, Wang Y and Shen X (2018) Type VI Secretion Systems Present New Insights on Pathogenic Yersinia. Front. Cell. Infect. Microbiol. 8:260. doi: 10.3389/fcimb.2018.00260

### INTRODUCTION

The genus Yersinia comprises three species of bacterial pathogens, namely Y. pestis, Y. enterocolitica, and Y. pseudotuberculosis, that are causative agents of human diseases. Y. pestis is the etiological agent of plague, often transmitted by flea bites or aerosols. It infects regional lymph nodes or lungs and causes the highly lethal disease in humans. Y. pseudotuberculosis and Y. enterocolitica are enteric pathogens, which usually grow in the environment and can be transmitted to mammalian hosts through ingestion of contaminated food or water. They typically cause a broad range of gastrointestinal diseases, from enteritis to mesenteric lymphadenitis (Bibikova, 1977; Brubaker, 1991; Bottone, 1997; Putzker et al., 2001; Pujol and Bliska, 2005). These three bacterial species have been used in experimental models of infection to study their pathogenicity for the mammalian host.

Despite differences in pathogenesis, these virulent Yersinia species have several common virulence factors, including the 70-kb virulence plasmid (pCD1 in Y. pestis and pYV in enteropathogenic Yersinia) and the yersiniabactin (Ybt) system (Brubaker, 1991; Heesemann et al., 1993; Cornelis et al., 1998). The 70-kb plasmid contains dozens of genes encoding structural components of a type III secretion system (T3SS), and also encodes several T3SS effector proteins called Yersinia outer proteins (Yops) and their dedicated chaperones to subvert the innate immune system of the hosts (Bliska et al., 2013; Schwiesow et al., 2015). Since enzymes/toxins/effectors delivered by secretion systems play a crucial role in the interaction between pathogens and their hosts or competitors, the various types of secretion systems attract interest in the research on pathogenic bacteria.

Several virulence associated secretion (vas) genes in T6SS gene clusters were identified in bacterial pathogens before, but till the year of 2006 the Type VI secretion system was proposed and defined as a contact-dependent secretion mechanism (Mougous et al., 2006; Pukatzki et al., 2006). T6SS was found distributed in about 25% of all sequenced Gram-negative bacteria with high Yang et al. T6SS in Pathogenic Yersinia

conservation (Boyer et al., 2009; Salomon and Orth, 2015). The T6SS is a proteinaceous machinery that directly injects effector proteins into target cells in a one-step process with a bacteriophage-like cell-puncturing device (Zoued et al., 2014), although the membrane puncturing process has not been directly observed to date. The versatile T6SS weapons allow bacteria to compete with other bacteria or attack simple or higher eukaryotes. The main function of T6SS was implicated in virulence, commensalism or symbiosis, and interbacterial competition (Jani and Cotter, 2010). Apart from the conserved components and similar structures, various functions of T6SSs have been exploited in the past decade, and the newly discovered effectors and their dedicated chaperones secreted by T6SS are also being updated. In the three pathogenic Yersinia species, different series of T6SS seem to possess different functions. This review focuses on the detailed description of T6SSs function in Yersinia.

#### COMPARISON OF T6SS GENE CLUSTERS IN YERSINIA SPECIES

#### Component, Structure, and Energetics of T6SSs

The core components of T6SS contain 13 subunits, which comprise the typical T6SS structure similar to the T4 bacteriophage, with tail, spike, sheath, hub or baseplate proteins (Boyer et al., 2009; Cascales and Cambillau, 2012). The typical T6SS structure is composed of three subunits: the membrane complex, the baseplate complex, and the injection apparatus. The membrane complex is composed of proteins TssJLM, which anchor the baseplate complex (TssAEFGK) to the membrane and provide structural support. The injection apparatus contains needle sheath (TssBC), tail tube (TssD/Hcp) and spike complex (TssI/VgrG and PAAR motifs) (Silverman et al., 2012; Brunet et al., 2015; Cianfanelli et al., 2016). Generally, various effectors and chaperones could bind to this injection apparatus when they were needed to be secreted out through the T6SS apparatus, and the secretion of Hcp or VgrG is often regarded as the hallmark of a functional T6SS in many bacterial species (Mougous et al., 2006; Pukatzki et al., 2006; Wang et al., 2011). The PAAR (Proline-Alanine-Alanine-aRginine) repeat containing proteins were regarded as an additional component of the T6SS machinery. PAAR-repeat could form a sharp conical extension on the tip of the VgrG spike, which is further involved in the effector recruitment and attaching effector domains to the spike (Shneider et al., 2013; Bondage et al., 2016). The 13 core components of typical T6SS are listed in **Table 1.**

The secretion of T6SS is a dynamic cycle of assembly/extension, contraction/puncture, and disassembly of the sheath, accompanying the energy supply and effector transportation (Basler et al., 2012). The initial stage of the T6SS assembly is to form the membrane complex, followed by the recruitment of baseplate proteins for anchor and extension of the outer sheath and the inner tube (Brunet et al., 2015). This indicated that the T6SS shares a common assembly pathway of tail tubes with bacteriophage (Brunet et al., 2014). During the attack process, the sheath-like structure propels an inner tube through contraction, and the membrane-puncturing spike is pierced into the target cells. Then effectors and chaperones are delivered into target cells with this expelled structure in a one-step manner (Cianfanelli et al., 2016).

### Different Series of T6SS in Three Yersinia Species

T6SS displays a single copy in majority of the bacterial species, yet multiple distinct copies are found in several bacterial species (Boyer et al., 2009). In the Yersinia species, six and four T6SS clusters were identified in Y. pestis (Andersson et al., 2017) and Y. pseudotuberculosis (Zhang et al., 2011b), respectively, while only one copy was found in Y. enterocolitica (Jaakkola et al., 2015). The T6SS gene clusters in Yersinia are shown in **Figure 1**.

It is noteworthy that the multiple distinct T6SS copies are not functionally redundant; for example, T6SS-1 and T6SS-5 in B. thailandensis mediate the bacterial antagonism and macrophage infection, respectively (Schwarz et al., 2010). In Y. pestis, the T6SS region YPO0498-YPO0516 (the T6SS Cluster A in **Figure 1**) is preferentially expressed at 26 vs. 37◦C (Cathelyn et al., 2006; Robinson et al., 2009), which suggests this gene cluster may function in natural conditions rather than in its mammalian host. Deletion of the T6SS Cluster A locus reduced the uptake by J774.1 murine macrophages (Robinson et al., 2009). However, it had no effect on virulence in bubonic or pneumonic murine plague models compared to the parental Y. pestis CO92 strain (Robinson et al., 2009; Andersson et al., 2017). An Hcp-like protein encoded by ypo0502 was found to play important roles in autoagglutination (AA), indicating that the T6SS Cluster A is involved in intraspecies interaction of bacteria (Podladchikova et al., 2011). A similar thermoregulated gene cluster in Y. pseudotuberculosis was identified as T6SS4 (ypk\_3548- ypk\_3566), which is precisely regulated by temperature, growth phase, and AHL-dependent quorum sensing systems, and plays a crucial role in resistance to environmental stresses (Zhang et al., 2011b). The T6SS Cluster A in Y. pestis and T6SS4 in Y. pseudotuberculosis could both be induced by room temperature conditions (about 25◦C). The phylogenetic analysis based on TssL of Yersinia T6SSs showed a close genetic distance between T6SS Cluster A in Y. pestis and T6SS4 in Y. pseudotuberculosis (**Figure 2**). Furthermore, the two T6SS clusters have similar genomic organization, suggesting they may both play crucial roles in environment adaptability. From **Figure 1** it showed the T6SS-E and F clusters in Y. pestis only contain 4 and 9 T6SS genes, respectively. Considering these components seemed complementary, and the two T6SS clusters locate in vicinity to each other, it speculated they may represent an integral set of T6SS. It is worth noting that Y. pestis has 6 clusters but only one with a complete set of T6SS genes, while Y. pseudotuberculosis has 4 clusters with full set of genes. Note that some clusters encode an incomplete set of T6SS genes and could be non-functional. The function of these incomplete T6SSs should be experimental verified in the future.

#### TABLE 1 | The 13 conserved components of typical T6SS.


#### THE FUNCTIONS OF T6SS IN YERSINIA

#### Virulence

At the beginning, genes of T6SS cluster in Vibrio cholerae and Pseudomonas aeruginosa were identified as vas (virulenceassociated secretion) genes for their roles in pathogenesis (Mougous et al., 2006; Pukatzki et al., 2006). Afterwards, the virulence-related functions of T6SS in the host were verified in several organisms including Edwardsiella tarda (Zheng and Leung, 2007), Aeromonas hydrophila (Suarez et al., 2008), Salmonella enterica (Blondel et al., 2010), Campylobacter jejuni (Lertpiriyapong et al., 2012), and Burkholderia pseudomallei (Schwarz et al., 2014).

In 2008, five T6SS gene clusters in Y. pestis KIM were identified through a phylogenetic analysis (Bingle et al., 2008), while six T6SS clusters in Y. pestis CO92 were predicted based on in silico analysis (Boyer et al., 2009). Actually, six T6SS clusters are distributed in genome of Y. pestis KIM and CO92 according to the SecReT6 database, which is a web-based resource for type VI secretion systems found in bacteria (Li et al., 2015). However, the virulence of T6SS in Y. pestis to hosts was confirmed through laboratory experiments till 2015. Using an in vivo signaturetagged mutagenesis (STM) screening approach, Ponnusamy et al. identified three T6SS genes with virulence potential in Y. pestis CO92. In-frame deletion of vasK (YPO3603, a component of the T6SS cluster G) in Y. pestis CO92 resulted in significant attenuation of the bacterium in murine models of infection, and this attenuation could be fully complemented (Ponnusamy et al., 2015). In addition, the five deletion mutants of T6SS clusters (B, C, E-G) showed varying levels of attenuation when evaluated in

a mouse model of pneumonic plague, but not all T6SS clusters of Y. pestis are equally functional. Deletion mutants for Cluster C and Cluster F exhibited limited, 14%, or no attenuation, respectively, while deletion mutants for Cluster B, Cluster E, and Cluster G, exhibited significant levels of attenuation in infection of a bubonic plague model. A further augmented attenuation was detected in mutants lacking multiple T6SSs. The 1ypo2720-27331hcp3 (Cluster E and hcp homolog 3) double deletion mutant showed 60% of attenuation in comparison to WT CO92. This result further supports the involvement of T6SSs in Y. pestis virulence (Andersson et al., 2017). Moreover, most of the attenuated T6SS deletion mutant strains showed decreased host cell cytotoxicity and intracellular survival in comparison to WT CO92 (Andersson et al., 2017).

Y. pseudotuberculosis YPIII mutants lacking T6SS-4 or yezP (ypk\_3549, an effector gene in T6SS-4 cluster) are defective in virulence in mice (Wang et al., 2015). The enteropathogenic Y. enterocolitica causes infections similar to Y. pseudotuberculosis, but has several different characteristics related to epidemiology and ecology, and only one type of T6SS was found in the Y. enterocolitica genome. It is speculated that the lack of multiple T6SS might be beneficial for escaping the host immune system, because a T6SS with toxic effectors might encumber the invasion and survival of Y. enterocolitica cells in mammalian hosts (Jaakkola et al., 2015). This function can be regarded as anti-virulence, which has been discussed previously (Jani and Cotter, 2010).

#### Bacteria-Bacteria Interaction

Bacteria have evolved various mechanisms to kill bacterial competitors and defense against predators in their environment. The T6SS has membrane-penetrating tails similar to bacteriophages, suggesting that T6SS may be involved in cell-cell interactions of bacteria (Pell et al., 2009). Bacteria could utilize T6SS to intimately interface with other bacteria, efficiently killing or inhibiting competitors with T6SS toxins and protecting itself with immune proteins, which has been reported in P. aeruginosa (HSI-1) (Hood et al., 2010), B. thailandensis (T6SS-1) (Schwarz et al., 2010), V. cholerae (MacIntyre et al., 2010; Ishikawa et al., 2012), and Serratia marcescens (Murdoch et al., 2011). Different from the traditional offensive and defensive model of T6SS mediated bacterial competition (Russell et al., 2012; Lien and Lai, 2017; Yang et al., 2018), swarming Proteus mirabilis discriminate non-identical population via T6SS-dependent delivery of toxic effectors, thus forming a visible demarcation line (Dienes line) between different Proteus isolates (Alteri et al., 2013). Although this phenomenon of self-recognition during swarming in P. mirabilis has been known for decades it was only ascribed to the T6SS recently.

Through DNA microarrays analysis of Y. pestis CO92, an IAHP (IcmF-Associated Homologous Protein) locus (YPO0498- 0516) was found regulated by RovA, which was specifically required for bubonic plague (Cathelyn et al., 2006). However, another study indicated that mutation in this locus had no effect in laboratory models of infected oriental rat flea X. cheopis, or in virulence using bubonic or pneumonic murine plague models (Robinson et al., 2009). Similarly, in-frame deletion of YPO0498 did not affect virulence in vivo (Ponnusamy et al., 2015). Recently, this locus was classified as T6SS cluster A in Y. pestis CO92, and was confirmed not required for pathogenesis (Andersson et al., 2017). Nevertheless, deletion of the T6SS Cluster A locus resulted in increased uptake and intracellular growth of Y. pestis in the macrophage-like J774.A1 cells, indicating a role of T6SS in interaction with host cells (Robinson et al., 2009). Among this T6SS cluster, YPO0502, identified as a Hcp-like protein that forms pilus-like structure, determines Y. pestis autoagglutination (AA). This suggests that at least one (T6SS Cluster A) of the 6 T6SS clusters in Y. pestis is involved in bacterial interaction (Podladchikova et al., 2011). However, whether Yersinia T6SSs are involved in bacterial competition and Dienes line formation needs to be verified in the future.

#### Stress Response

Many of the earliest studies on T6SS focused on their roles in virulence and the interactions between pathogens and hosts. In 2009, it was reported that a T6SS in Vibrio anguillarum acts as a sensor for an unknown extracytoplasmic signal that modulates RpoS (Weber et al., 2009). RpoS is the main regulator of gene expression during stationary phase and stress conditions, and it positively regulates VanT linking quorum sensing (QS) to stress response and general physiology of V. anguillarum (Weber et al., 2008). Therefore, this study suggested a new function for T6SS in the ecology of bacteria, that T6SS played a crucial role in bacterial stress response and cell survival after exposure to various environmental challenges (Weber et al., 2009).

In Y. pseudotuberculosis, RpoS was found to be important in resistance to multiple stressors including oxidative stress, acid stress, osmotic stress, and 42◦C heat shock, and was also crucial for motility, biofilm formation, and T6SS expression. The electrophoretic mobility shift assay (EMSA) showed that RpoS regulates T6SS4 expression by directly binding to the T6SS4 promoter (Guan et al., 2015). However, compared with the two reported YpsRI QS system (Atkinson et al., 2008) and the OmpR regulator (Zhang et al., 2013), RpoS plays a less crucial role in T6SS4 activation. T6SS4 is only partly responsible for the stress-resistance activity of RpoS, consistent with the relatively weak role of RpoS in activating T6SS4 expression in Y. pseudotuberculosis (Guan et al., 2015). Similarly, the PppA-PpkA pair of H1-T6SS in P. aeruginosa plays roles in bacterial resistance to oxidative and osmotic stress, and pppA-ppkA deletion affects the expression of the RpoS and QS regulons (Goldova et al., 2011).

In 2013, an OmpR-T6SS4 regulation pathway for stress resistance in Y. pseudotuberculosis was characterized in two studies, both of which confirmed the direct binding of OmpR on the T6SS-4 promoter region in vitro (Gueguen et al., 2013; Zhang et al., 2013). In an earlier study, OmpR was found to be essential for low pH adaptation by positively regulating the expression of urease in Y. pseudotuberculosis (Hu et al., 2009). Here, it was indicated that T6SS4 contributed to acid survival by maintaining a steady-state intracellular pH in Y. pseudotuberculosis. The acid resistance ability is dependent upon the ATPase activity of ClpV4 that participates in H<sup>+</sup> extrusion, and this process is positively regulated by the global regulator OmpR (Zhang et al., 2013).

In addition, the T6SS-4 expression mediated by OmpR is also induced in high osmolarity conditions or in the presence of sodium deoxycholate. Besides T6SS4 gene cluster, OmpR activates additional genes involved in tolerance to high osmolarity, because the ompR transposon mutant cells were more severely affected than the tssF4 cells (Gueguen et al., 2013). Similarly, an earlier study showed that deletion of T6SS increased the susceptibility of C. jejuni to a bile salt, deoxycholic acid (DCA), and the reason for which was attributed to an increase in the intracellular influx of DCA mediated by T6SS (Lertpiriyapong et al., 2012).

#### Ion Transport

Recently, T6SS4 in Y. pseudotuberculosis was found to have a notable function of transporting zinc ions (Zn2+) from the environment into bacterial cells to mitigate the detrimental hydroxyl radicals induced by multiple stressors (Wang et al., 2015). Generally, bacteria utilize the classical ZnuABC transport system to acquire Zn2<sup>+</sup> to maintain zinc balance. In Y. pseudotuberculosis, the ZnuABC transporter contributes to reduce the intracellular ROS levels and thus prevents oxidative damage to cells (Wang et al., 2016). While in Y. pestis, the yersiniabactin (Ybt, a zincophore for Zn2<sup>+</sup> acquisition) synthetase HMWP2, together with ZnuABC, are critical for lethal infections in a mouse model of septicemic plague (Bobrov et al., 2014). Interestingly, in Y. pseudotuberculosis a T6SS4 mediated substrate YezP (YPK\_3549), a novel Zn2+-binding protein, has the capacity to rescue the sensitivity to oxidative stress exhibited by T6SS4 mutants when added to extracellular milieu. Zn2<sup>+</sup> acquisition achieved by YezP is co-regulated with T6SS4 by OxyR, which is a global oxidative stress regulator that senses diverse environmental cues (Wang et al., 2015).

Similar capability of ion transport by T6SS was found in B. thailandensis and P. aeruginosa. In B. thailandensis, a proteinaceous zincophore (TseZ) secreted through T6SS4 interacts with the outer membrane heme transporter HmuR to acquire zinc under oxidative stress (Si et al., 2017a). In addition, the T6SS4 in B. thailandensis facilitated the uptake of Mn2<sup>+</sup> by secreting a Mn2+-binding effector TseM during oxidative stress. The Mn2<sup>+</sup> load on TseM could be delivered to MnoT, a Mn2+-specific TonB-dependent outer membrane transporter, and then transported into cells to fulfill the increased cellular demand for Mn2<sup>+</sup> under oxidative stress (Si et al., 2017b). While in P. aeruginosa, a H3-T6SS secreted effector TseF (PA2374) is involved in iron uptake by interacting with outer membrane vesicles (OMVs) and the Pseudomonas Quinolone Signal (PQS) system (Lin et al., 2017). These studies highlight the newly characterized function of ion transport through T6SSs. Whether and how Yersinia T6SSs transport other metal ions besides Zn2<sup>+</sup> needs to be investigated in the future.

#### Other Potential Functions

More functions were identified in other bacterial species in addition to Yersinia. T6SS is also implicated in biofilm formation in several bacterial pathogens. For example, Hcp participates in biofilm formation in P. aeruginosa PAO1 (Southey-Pillig et al., 2005), and H1-T6SS is associated with biofilm-specific antibiotic resistance in P. aeruginosa PA14 (Zhang et al., 2011a). In Acidovorax citrulli, four mutants of T6SS components were reduced in biofilm formation (Tian et al., 2015). Recently, T6SS was shown to be associated with cell autophagy. VgrG2, a translocon of T6SS2 in Vibrio parahaemolyticus, induces autophagy in macrophages (Yu et al., 2015). A Type VI secretion PGAP1-like effector TplE, belonging to Tle4 phospholipase family in P. aeruginosa (Russell et al., 2013), induced autophagy through disruption of endoplasmic reticulum (ER) homeostasis (Jiang et al., 2016). In addition, the T6SS in P. mirabilis plays a role in self-recognition during swarming via delivery of toxic effectors (Alteri et al., 2013). Further research and analysis need to be performed to verify these potential T6SS functions in Yersinia species.

#### EFFECTORS SECRETED BY T6SS IN YERSINIA

With a bacteriophage tail-like structure, T6SS apparatus can deliver various effectors into bacterial competitors for


TABLE 2 | Putative PAAR motif-containing effectors in Yersinia T6SS clusters.

interbacterial competition (Cianfanelli et al., 2016), into eukaryotic hosts for pathogenesis (Durand et al., 2014) or into extracellular milieu for nutritional demands (Wang et al., 2015; Lin et al., 2017; Si et al., 2017a). The antibacterial T6SS effector usually has a chaperone immunity protein that provides protection to the organism itself. These effectors could be divided into three categories as targeting the membrane, cell wall, or nucleic acid (Russell et al., 2014). The eukaryotic hosttargeted T6SS effectors are diverse in biological and biochemical functions. For example, VgrG-1 in V. cholerae can cause actin crosslinking for intestinal inflammation (Ma and Mekalanos, 2010), VgrG2b in P. aeruginosa interacts with microtubules for successful invasion of epithelial cells (Sana et al., 2015), and VgrG-5 in B. thailandensis could induce the host cell membrane fusions by interacting directly with a neighboring host cell receptor or by engaging in homotypic interactions within or between cells (Schwarz et al., 2014). A recent publication summarizes the T6SS effectors identified in recent years (Lien and Lai, 2017).

Many T6SS secretion effectors have been identified in bacterial organisms. However, apart from the hallmark of T6SS effectors Hcp or VgrG, only two effectors were functionally characterized in Yersinia species according to published papers. They are YezP (YPK\_3549) in Y. pseudotuberculosis, which is a novel Zn2+-binding protein substrate (Wang et al., 2015), and YPO0502 in Y. pestis CO92, which is a low-molecularweight component with siderophore activity and regarded as the autoagglutination factor (Podladchikova et al., 2011). In addition, three T6SS associated proteins, y3673/Hcp, y3674 and y3675, were detected in the outer-membrane proteome in Y. pestis KIM6+ using 2DGE gel electrophoresis (Pieper et al., 2009). However, the y3674/y3675 genes possess VipB/VipA motif according to the KEGG database, which suggests these two proteins may comprise the contractile sheath of T6SS. With high-throughput signature-tagged mutagenesis (STM) approach, three T6SS components or effector-encoding genes (vasK/ypo3603, ypo0498, and ypo1484) were identified in Y. pestis CO92 (Ponnusamy et al., 2015). VasK was confirmed to be important during Y. pestis infection in mouse models of plague (Andersson et al., 2017). With a conserved "MxiM" motif, YPO0498 may be a lipoprotein, while YPO1484 is likely to be an RNase toxin with a sharp conical extension complex of "PAAR" (Proline-Alanine-Alanine-aRginine) (Shneider et al., 2013). However, more experiments need to be performed to explain the connection between these proteins and T6SS apparatus. Recently, Andersson et al generated three PAAR motif repeat-containing T6SS effector encoding gene deletion mutants (1ypo0873, 1ypo1484, and 1ypo3615). Only 1ypo1484 exhibited a limited level of attenuation, 20% survivability, in comparison to WT CO92 in a mouse model of pneumonic plague. This attenuation was similar to the level of attenuation, 14%, reported in the T6SS Cluster C (ypo1458-1484) deletion mutant, which contains the ypo1484 gene (Andersson et al., 2017). As multiple copies of T6SS have been confirmed in Yersinia species, it seems that more potential undiscovered effectors remain to be identified. Here, we presented the hypothetical effectors with "PAAR" motif in Yersinia T6SS clusters (as shown in **Table 2**).


TABLE 3 | The identified transcriptional regulators for T6SSs in Yersinia.

#### T6SS REGULATION IN YERSINIA

T6SS gene clusters are found both in pathogenic and non-pathogenic bacteria, which are distributed in various environments including marine, soil, rhizosphere, higher plants and mammalian hosts, implying their diverse functions (Boyer et al., 2009). As the assembly, contraction, and disassembly cycle of the T6SS is likely to be energetically costly to the bacterial cell, the T6SS organelle is tightly regulated when organisms alternate in various environmental conditions (Silverman et al., 2012). It has been reported that T6SS in bacterial species responds to various environmental factors including salinity (Salomon et al., 2013), iron-concentration (Chakraborty et al., 2011), temperature (Pieper et al., 2009), QS-dependent cell density (Zheng et al., 2010; Kitaoka et al., 2011), host immune system, and other stressors (Brooks et al., 2013).

Temperature is a common environmental factor for most pathogens, and changes widely when organisms alternate in various environmental conditions. The expression of T6SS cluster A (YPO0498-YPO0516 in Y. pestis CO92 or y3658-y3677 in Y. pestis KIM) is induced by lowering the temperature from 37 to 26◦C (Pieper et al., 2009; Robinson et al., 2009). In Y. pseudotuberculosis, all the four T6SS clusters are differentially thermoregulated. In addition to the induction of T6SS4 at 26◦C, the T6SS1 expression is obviously induced at 37◦C, while T6SS2, T6SS3, and T6SS4 was completely repressed at 37◦C, which suggested that the T6SS1 played a more important role in bacterial virulence during mammalian host infection than other 3 T6SS loci. In addition, T6SS is precisely regulated by growth phase and acylated homoserine lactones (AHLs) dependent quorum sensing (QS) systems (Zhang et al., 2011b). Furthermore, T6SS4 in Y. pseudotuberculosis has been shown to respond to oxidative stress (Wang et al., 2015).

T6SS could be activated by different transcriptional regulators to combat multiple stresses and host immunity, and several transcriptional regulators have been identified in Yersinia species. In Y. pseudotuberculosis, the OmpR-regulated T6SS4 plays important roles under acidic and osmotic stress conditions (Gueguen et al., 2013; Zhang et al., 2013). It was proposed that T6SS4 was activated by RpoS and partly mediated the roles of RpoS in osmotic and acid resistance (Guan et al., 2015). Furthermore, T6SS4 could be regulated by OxyR or ZntR to cope with oxidative stress by removing ROS produced by host immunity or harmful environments (Wang et al., 2015,

2017). We have summarized the identified regulators of T6SS in Yersinia in **Table 3**.

In addition to the research on Yersinia, other transcriptional regulators of T6SSs have been identified in more bacterial species. In P. aeruginosa, a T6SS is controlled by the global virulence regulator proteins RetS and LadS (Mougous et al., 2006). In Salmonella enterica, T6SS genes were controlled by the SsrA/SsrB two-component regulatory system (Parsons and Heffron, 2005). Furthermore, T6SS is regulated by AggR in enteroaggregative Escherichia coli (Dudley et al., 2006), by VirA/G in Burkholderia mallei (Schell et al., 2007), and similarly by quorum sensing mechanism in the plant pathogen Pectobacterium atrosepticum (Liu et al., 2008). By contrast, few post-transcriptional regulators have been identified so far. Lon, a posttranslational regulator, is known to affect a variety of physiological traits in many bacteria. It suggested that T6SS was regulated by LonA in V. cholerae, as well as biofilm formation, swimming motility, intracellular levels of cyclic diguanylate (Rogers et al., 2016). In P. aeruginosa, a screen to identify T6SS regulatory elements found that the posttranscriptional regulator RsmA imposes a concerted repression on all three T6SS clusters. RsmA and a transcriptional regulator AmrZ orchestrate the assembly of all three T6SSs in P. aeruginosa (Allsopp et al., 2017).

#### CONCLUSIONS

Over the past decade, pathogenic bacteria P. aeruginosa and V. cholerae have been widely used to study T6SS, and significant advancements have been made on the understanding of their structure, activation signals, and regulatory pathways (Chen et al., 2015; Joshi et al., 2017). As important pathogenic bacteria, three Yersinia species also attracted increasing attention because of their multiple T6SS copies. As a summary, we drew a schematic diagram based on the current achievement of research on T6SSs in Yersinia (**Figure 3**). However, many questions need to be addressed. We believe that the future work in Yersinia T6SS research will involve at least six points: 1, The T6SS activation conditions; 2, The T6SS regulators and the regulation network; 3, Identification of T6SS secreted effectors and their new functions; 4, Based on the knowledge of T6SS attack and protection ability,

#### REFERENCES


the ways to prevent the T6SS<sup>+</sup> pathogens; 5, The different functions and connection of different T6SS clusters; 6, The exploitation of new methods or techniques for further T6SS studies. The progress of Yersinia T6SS research will provide new knowledge and challenges, as well as further our understanding of the T6SS complexities in other important human pathogens. This research will also influence clinical medicine, microbial community ecology, and physiology.

#### AUTHOR CONTRIBUTIONS

XY, JP, and YW collected and assessed the references; YW and XS contributed in the proposal and guideline of the review; XY, JP, and XS wrote the paper.

#### FUNDING

The authors thank the National Natural Science Foundation of China (Nos. 31725003, 31670053, and 31370150) for the financial support.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Yang, Pan, Wang and Shen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Changes in Transcriptome of Yersinia pseudotuberculosis IP32953 Grown at 3 and 28◦C Detected by RNA Sequencing Shed Light on Cold Adaptation

Jussa-Pekka Virtanen, Riikka Keto-Timonen\*, Kaisa Jaakkola, Noora Salin and Hannu Korkeala

Department of Food Hygiene and Environmental Health, Faculty of Veterinary Medicine, University of Helsinki, Helsinki, Finland

#### Edited by:

Victoria Auerbuch, University of California, Santa Cruz, United States

#### Reviewed by:

Karl M. Thompson, Howard University, United States Michael Marceau, Université Lille Nord de France, France

> \*Correspondence: Riikka Keto-Timonen riikka.keto-timonen@helsinki.fi

#### Specialty section:

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 07 May 2018 Accepted: 09 November 2018 Published: 27 November 2018

#### Citation:

Virtanen J-P, Keto-Timonen R, Jaakkola K, Salin N and Korkeala H (2018) Changes in Transcriptome of Yersinia pseudotuberculosis IP32953 Grown at 3 and 28◦C Detected by RNA Sequencing Shed Light on Cold Adaptation.

Front. Cell. Infect. Microbiol. 8:416. doi: 10.3389/fcimb.2018.00416 Yersinia pseudotuberculosis is a bacterium that not only survives, but also thrives, proliferates, and remains infective at cold-storage temperatures, making it an adept foodborne pathogen. We analyzed the differences in gene expression between Y. pseudotuberculosis IP32953 grown at 3 and 28◦C to investigate which genes were significantly more expressed at low temperature at different phases of growth. We isolated and sequenced the RNA from six distinct corresponding growth points at both temperatures to also outline the expression patterns of the differentially expressed genes. Genes involved in motility, chemotaxis, phosphotransferase systems (PTS), and ATP-binding cassette (ABC) transporters of different nutrients such as fructose and mannose showed higher levels of transcripts at 3◦C. At the beginning of growth, especially genes involved in securing nutrients, glycolysis, transcription, and translation were upregulated at 3◦C. To thrive as well as it does at low temperature, Y. pseudotuberculosis seems to require certain cold shock proteins, especially those encoded by yptb3585, yptb3586, yptb2414, yptb2950, and yptb1423, and transcription factors, like Rho, IF-1, and RbfA, to maintain its protein synthesis. We also found that genes encoding RNA-helicases CsdA (yptb0468), RhlE (yptb1214), and DbpA (yptb1652), which unwind frozen secondary structures of nucleic acids with cold shock proteins, were significantly more expressed at 3◦C, indicating that these RNA-helicases are important or even necessary during cold. Genes involved in excreting poisonous spermidine and acquiring compatible solute glycine betaine, by either uptake or biosynthesis, showed higher levels of transcripts at low temperatures. This is the first finding of a strong connection between the aforementioned genes and the cold adaptation of Y. pseudotuberculosis. Understanding the mechanisms behind the cold adaptation of Y. pseudotuberculosis is crucial for controlling its growth during cold storage of food, and will also shed light on microbial cold adaptation in general.

Keywords: cold stress, stress tolerance, compatible solute, RNA helicase, cold shock protein, transcription factor

## INTRODUCTION

Yersinia pseudotuberculosis is an enteropathogenic bacterium that causes the foodborne infection yersiniosis. Although it has an optimum growth temperature of around 28◦C, Y. pseudotuberculosis can grow and multiply at coldstorage temperatures, even as low as 0◦C (Keto-Timonen et al., 2018). Long storage at low temperature favors the growth of psychrotrophs like Y. pseudotuberculosis, as they can proliferate without much competition. Outbreaks of Y. pseudotuberculosis have been associated with vegetables, raw milk, and drinking untreated water (Sato and Komazawa, 1991; Nuorti et al., 2004; Rimhanen-Finne et al., 2009; Pärn et al., 2015). In order to survive the various ecological niches through which it could enter the food chain, Y. pseudotuberculosis carries many tools in its genome, such as operons for the transport of substrates more abundant in plants and soil than animal tissue, as well as type VI secretion systems (Jaakkola et al., 2015).The bacterium has been found in the intestines of many animals like domestic pigs, goats, sheep, wild lagomorphs, birds, rodents, and shrews (Niskanen et al., 2003; Laukkanen et al., 2008; Giannitti et al., 2014; Le Guern et al., 2016; Joutsen et al., 2017). It can also thrive in soil as well as in certain protozoans and nematodes (Buzoleva and Somov, 2003; Gengler et al., 2015; Santos-Montañez et al., 2015).

Adaptation to cold and long-term growth at low temperatures poses many challenges to bacteria. Low temperature decreases the fluidity of cell membranes, thereby interfering with normal membrane protein function. It also slows down protein folding, ribosomes, and translation as well as excessively stabilizes nucleic acid structures (Palonen et al., 2010). Cold also induces radical oxygen species production by both slowing metabolism and increasing oxygen solubility (Chattopadhyay et al., 2011). Psychrotrophic bacteria have many methods to deal with these problems. For example, bacteria can increase membrane fluidity by introducing unsaturated lipids that no longer fit as snugly together (Suutari and Laakso, 1994). Unsaturated fatty acids dominate the fatty acid composition of Y. pseudotuberculosis at low temperatures (Bakholdina et al., 2004).

Cold shock proteins (Csp) are small proteins whose mRNA carries a cold shock domain that enables its translation at low temperatures. Csps, along with helicases and translation factors exhibiting similar stabilizing secondary structures, unwind mRNA and support the translational apparatus at low temperatures. Y. pseudotuberculosis has nine csp genes, homologous to those of Escherichia coli, five of which are induced at low temperature in E. coli (Keto-Timonen et al., 2016). The bacterium also has five helicases with a conserved DEADbox motif: CsdA (yptb0486), RhlE (yptb1214), RhlB (yptb0165), DbpA (yptb1652), and SrmA (yptb2900).

The main goal of this study was to determine how Y. pseudotuberculosis manages to thrive at refrigerator temperature. We identified which genes showed significantly more transcripts at 3◦C when compared to 28◦C, at each growth phase, mainly focusing on the beginning of growth and logarithmic phase. Of all cold shock proteins, those encoded by yptb1423, yptb3585, yptb3586, and especially yptb2414 and yptb2950, showed significantly more transcripts at low temperature, seemingly forming the backbone of cold acclimation of Y. pseudotuberculosis. Furthermore, we found that, in addition to CsdA, helicases RhlE, and DbpA were significantly upregulated at low temperature, which speaks to their importance in surviving low temperatures. The bacterium also seems to accumulate glycine betaine by uptake and biosynthesis, as the corresponding genes were upregulated at low temperature. Transcription termination factor Rho, along with IF-1 and RbfA, both acting on ribosomes, were also upregulated, which would seem to suggest that they play an important role. None of these genes and proteins have, to our knowledge, been linked to cold acclimation of Y. pseudotuberculosis before.

### MATERIALS AND METHODS

#### Bacterial Strain and Growth Conditions

Single Y. pseudotuberculosis IP32953 colonies grown on blood agar plates at 28◦C were inoculated and grown separately in LB broth (Luria-Bertani; Sigma-Aldrich, St. Louis, MO, USA) at 28◦C with shaking overnight. Overnight broths were diluted (1:100) in LB broth and divided into two groups so that four cultures (biological replicates) were grown with shaking both at 3 ◦C and at 28◦C. Biological replicates were used to better model true biological variability and improve the accuracy of statistical methods. Samples for total RNA extraction were collected at six corresponding points at different phases of growth across both temperatures (**Figure 1**). For total RNA extraction 1.25 ml of bacterial culture was mixed with 250 µl cold phenol-ethanol mixture (1:10) and kept on ice for 30 min. After incubation, samples were centrifuged at 4◦C at 13,200 rpm for 2 min and the resulting cell pellets were stored at −70◦C until RNA isolation.

#### RNA Isolation

Total RNA was isolated using GeneJET RNA Purification Kit (Thermo Fisher Scientific, Waltham, MA, USA) and treated with DNA-Free DNA Removal Kit (Ambion, Life Technologies, Carlsbad, CA, USA) according to manufacturers' instructions. The quantity and quality of RNA was examined with a Nanodrop ND-1000 Spectrophotometer (Thermo Fisher Scientific) and Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). RNA was stored at −70◦C until RNA-seq library preparation.

### Library Preparation and Sequencing

Ribosomal RNA was depleted from total RNA using Ribo-Zero rRNA Removal Kit for Bacteria (Epicenter, Madison, WI, USA) following the manufacturer's protocol. cDNA libraries were prepared using Script-Seq v2 RNA-Seq Library Preparation Kit (Epicenter) following the manufacturer's instructions. The libraries were amplified twice by PCR and barcoded. PCR was performed using Phusion High-Fidelity DNA Polymerase Kit (Thermo Fisher Scientific). The libraries were purified using AMPure XP System (Beckman Coulter, Brea CA, USA) after both PCR procedures. The libraries were sequenced (75 bp single read) using the Illumina NextSeq500 platform at the Institute of Biotechnology, University of Helsinki, yielding four sets of sequence reads (biological replicates) per growth point at both

3 and 28◦C. Raw sequences were deposited in the Sequence Read Archive (http://www.ncbi.nlm.nih.gov/sra) under accession number SRP144570.

### cDNA Synthesis and RT-qPCR

To confirm the differences of transcript levels identified in the RNA-sequence expression analysis, RT-qPCR validation was performed for selected genes (betB, csdA, fabF, infA, mdtI, proX, rhlE, and rho) at growth points I and III at both temperatures (3 and 28◦C). These genes or operons they represent are discussed in detail in this paper. A total of 500 ng of each RNA sample from three biological replicates per growth point was reverse-transcribed into cDNA in duplicate by using Maxima cDNA Synthesis Kit (Thermo Fisher Scientific) according to manufacturer's instructions. Primers for RT-qPCR were designed using Primer-BLAST software ((Ye et al., 2012); **Table 1**). Two replicate qPCR reactions for each cDNA sample were performed using the Dynamo Flash SYBR Green qPCR Kit (Thermo Scientific). Each reaction consisted of 1x Master Mix, 0,5 µM of forward and reverse primer and 4 µl of 1:20 (gene of interest) or 1:100,000 (16S rrn) diluted cDNA in a total volume of 20 µl. Rotor-Gene Q thermal cycler (Qiagen GmbH, Hilden Germany) was used in PCR runs with the cycling protocol consisting of initial heating step at 95◦C for 7 min, followed by 40 cycles of denaturation at 95◦C for 10 s, annealing at 60◦C for 15 s, extension at 72◦C for 20 s, and a final extension at 60◦C for 1 min. After each run a melt curve analysis was done to confirm specificity. Amplification reaction efficiencies for each primer pair were obtained from RT-qPCR standard curves prepared from serial dilutions of pooled cDNA samples. The duplicate C<sup>q</sup> values for PCR replicates were averaged. The relative quantification of gene of interest transcript levels at 3◦C, normalized to reference gene (16S rrn) transcript levels and calibrated to the samples taken at the same growth point at 28◦C, were calculated using the Pfaffl method (Pfaffl, 2001). Given near equal primer efficiencies, the logFC values of Pfaffl gene expression ratios between 3 and 28◦C should be proportional to the logFC values acquired by RNA-Seq. A linear regression analysis was performed on corresponding logFC values of RNA-Seq and RT-qPCR.

TABLE 1 | Primers used in quantitative real-time reverse transcription-PCR.


### Alignment and Annotation of RNA-Seq Data

We aligned, annotated, and analyzed the sequence reads with Bioconductor (Huber et al., 2015). The complete reference genome sequence and genomic features of Y. pseudotuberculosis IP32953 (Johnson et al., 2015) were acquired from the Pathosystems Resource Integration Center, PATRIC (Wattam et al., 2014). The sequence reads were both aligned to the reference genome, allowing a maximum of 10 hits, and annotated with Bioconductor package QuasR (Gaidatzis et al., 2015).

#### Differential Gene Expression Analysis

One of the replicates of growth point IV at 28◦C was discarded for its low alignment quality. The remaining annotated transcript counts were analyzed using Bioconductor package baySeq (Hardcastle and Kelly, 2010). Low count reads were filtered out. The replicate counts for each growth point at 3◦C were compared to the counts of the respective growth point at 28◦C. The prior distributions were acquired using a negative binomial distribution whose parameters were estimated by quasimaximum-likelihood methods with a sample size of 10,000. The posterior likelihoods were established using 10 iterations to re-estimate the priors. Results were normalized by library size and results with FDR >0.05 were discarded. Log<sup>2</sup> fold changes (logFC) across the two temperatures were calculated using transcript count averages of replicates, and genes with logFC ≥ 2 were considered significantly expressed at 3◦C. For ease of viewing, genes were further grouped into operons retrieved from ProOpDB (Taboada et al., 2012). Amino acid sequence similarities were derived from multiple sequence alignment by Clustal Omega (Sievers et al., 2011). Upregulated genes sharing similar functions were classified into functional units (modules), acquired from the KEGG database (Kanehisa and Goto, 2000), using Bioconductor package clusterProfiler (Yu et al., 2012). Pvalues were adjusted for multiple comparisons by controlling the false discovery rate, using the Benjamini & Hochberg method implemented in clusterProfiler (Benjamini and Hochberg, 1995).

### Clustering

Expression profiles of genes that were significantly more expressed at low temperature, at least at one growth point, were clustered using Pearson correlation distance. Count data was normalized for visualization by using median ratio normalization implemented in Bioconductor package DESeq2 (Love et al., 2014). Averages of replicate counts were used instead of individual counts. Normalized data for each gene (rows) across all growth points (columns) was then plotted in a heatmap (**Figure 2**).

### RESULTS

Expression profiles of Y. pseudotuberculosis IP32953 grown at 3 and 28◦C were compared at corresponding phases of the growth based on growth curves determined by the optical density. We found 570 genes in total that showed significantly more transcripts at 3◦C than 28◦C at least at one of the growth points I–VI (**Table 2**). Motility and chemotaxis genes were at the top throughout growth since Y. pseudotuberculosis is non-motile at 28◦C. The total number of upregulated genes was 482 when motility, chemotaxis, and tRNA genes were filtered out. Growth point VI held the fewest significantly expressed genes (N = 125) whereas growth points II and IV held the most (N = 162; **Table 2**). The difference in expression was largest for genes involved in motility, with a maximum logFC of 7.42.

A gene that was expressed both differentially and in great numbers at 3◦C probably plays an important role in cold growth. We clustered expression profiles of genes that at least once during growth showed more transcripts at 3◦C than at 28◦C. Five subclusters (A–E) could be identified in the expression profile heatmap (**Figure 2**). The genes of most interest in the context of cold growth express highly at the beginning and logarithmic phases of growth, clustered into subclusters B–E (see details in **Supplementary Material**). Subcluster A mostly held genes that were expressed highly at the stationary phase.

#### Validation of RNA Sequencing Results With RT-qPCR

The expression profiles of eight differentially expressed genes at two different growth points were evaluated by RT-qPCR and compared with the RNA-seq analysis results. In linear regression analysis between the RNA-seq and RT-qPCR logFC values (**Figure 3**), a 0.93 Pearson correlation coefficient value (R 2 = 0.85) was observed, which confirmed the reproducibility and reliability of the RNA-seq method (**Figure 3**).

### Differentially Expressed Genes by Growth Point

At growth point I, genes involved in acquiring compatible solutes and various nutrients showed significantly more transcripts at 3◦C, and this pattern continued until stationary phase (**Figure 4**). For example, a significant increase in transcripts at 3◦C was displayed in the following genes: genes encoding a glycine betaine transporter (yptb2959–61); phosphotransferase systems (PTS) to import fructose (yptb1329– 31; **Figure 5**), N-acetylglucosamine (yptb1120, yptb3075–82), and L-ascorbate (yptb2600–2; **Figure 5**); and ATP-binding cassette (ABC) transporters of maltose (yptb2521, yptb3095–102) and

aldopentoses (yptb3591–6). A significant portion of upregulated genes at this growth point were associated with different PTSs and glycine betaine transport (**Figure 4**). As expected, genes encoding chaperone molecules such as helicase RhlE (yptb1214) and Csps (yptb2950, yptb2414, yptb3585, and yptb3586), which destabilize nucleic acid secondary structures, showed more transcripts at 3◦C.

At growth point II, genes involved in further processing of compatible solutes, spermidine efflux, synthesizing desaturated membrane lipids, biosynthesis of ribosomes, securing translation



Flagellar assembly and chemotaxis genes have been disregarded as Y. pseudotuberculosis is not motile at 28◦C; and tRNA-genes have also been filtered out.

under cold stress, and posttranscriptional modification of RNA molecules showed significantly more transcripts at 3◦C. Modules of different amino acid transporters and betaine synthesis were overexpressed (**Figure 4**). Genes encoding translation factors IF-1 (yptb1395) and Rho factor (yptb0167) peaked at this growth point. New additions to differentially expressed chaperone genes were helicase gene dbpA (yptb1652) and a new Csp gene (yptb1423).

At growth point III, a urease operon (yptb2938–44) and PTSs to import fructose (**Figure 5**), N-acetylglucosamine, and mannose showed significantly more transcripts at 3◦C than 28◦C. The gene encoding superoxide dismutase (yptb3925) also showed more transcripts at 3◦C at this growth point. At growth point IV, it seems the nutrient strategy shifted as an operon involved in sulfur metabolism (yptb2309–14) was expressed more at 3◦C than 28◦C. In addition to genes encoding IF-1 and Rho factor, a new translation factor gene rbfA (yptb0481) showed more transcripts at 3◦C at this point of growth.

At growth points V and VI, the nutrient strategy continued shifting as operons involved in the metabolism of histidine (yptb1965–9, yptb3851–2; **Figure 5**), cystine (yptb1717–20; **Figure 5**), methionine (yptb2973–5; **Figure 5**), and nitrogen compounds (yptb0022–3) showed more transcripts at 3◦C. At the same time, Csp gene expression levels and differences tapered off. The gene encoding helicase CsdA (yptb0486) was significantly more expressed at 3◦C at growth point III, but at 28◦C at growth points V and VI. An operon encoding a system that utilizes autoinducer AI-2 showed significantly more transcripts at growth point V.

#### Expression of Cold Shock Protein Genes

Of all the Csp genes, yptb2414 showed most transcripts at 3 ◦C from the beginning of growth throughout the logarithmic growth phase (**Figure 6**). Its difference in expression between 3 and 28◦C was significant at all growth points and the largest of all Csp genes. Gene yptb1624 was expressed significantly more at 28◦C (**Figure 6**). Gene yptb2950 showed many more transcripts at 3◦C at growth points I–IV, with a peak logFC of over 7 (**Figure 6**). Its transcription counts dipped after growth point I and decreased toward stationary phase. Gene yptb1423 was expressed at lower levels, but showed significantly more transcripts at 3◦C throughout growth points II–IV (**Figure 6**).

Nearly identical genes, yptb3585 and yptb3586, were expressed highly at growth point I, but their expression levels dipped right after (**Figure 7**). However, the differences in their expression between 3 and 28◦C consistently favored cold at significant levels. Expression levels of yptb3587 and yptb1088 rose toward the stationary phase at 3◦C, but the difference between 3 and 28◦C was not significant. After starting growth, yptb3587 appeared slightly downregulated (**Figure 7**). Gene yptb1392 was expressed much in the same way, but showed significantly more transcripts at 28◦C at the stationary phase of growth (**Figure 7**).

#### Expression of DEAD-Box RNA Helicase Genes

Y. pseudotuberculosis has five helicases with a conserved DEADbox motif: CsdA (yptb0486), RhlE (yptb1214), RhlB (yptb0165), DbpA (yptb1652), and SrmA (yptb2900). All the helicase genes were expressed more at 3◦C throughout growth (**Figure 8**), but the difference was significant for csdA (yptb0486) at growth point III, rhlE (yptb1214) at growth points I–IV, and dbpA (yptb1652) at growth points II and IV.

#### Expression of Genes Handling Compatible Solutes

Genes yptb2959–61 form the proU operon that encodes a transport system for glycine betaine and proline; genes yptb1195– 98 form the betIBA-betT divergent operon that is involved in glycine betaine biosynthesis; and genes yptb2052 and yptb2051 form the mdtIJ operon that encodes a spermidine efflux

pump. Operon proU showed significantly more transcripts at 3 ◦C at growth points I–IV (**Figure 9**), operon betIBA-betT at growth points II–IV (**Figure 10**), and operon mdtIJ at II and IV (**Figure 11**). Corresponding functional modules were also overexpressed at growth points I–IV for operon proU, points II– V for operon betIBA-betT, and points II and IV for operon mdtIJ (**Figure 4**). Another glycine betaine and L-proline transporter encoding gene proP (yptb0608) showed more transcripts at 3 ◦C throughout the growth, but not quite at significant levels (**Figure 9**).

#### Expression of Genes Encoding Rho Factor, IF-1, and RbfA

Genes rho (yptb0167), infA (yptb1395), and rbfA (yptb0481) encode the homologs of Rho factor, the translation initiation factor IF-1, and ribosome binding factor RbfA, respectively. The three factor genes were expressed highly from the beginning of growth to the end of the logarithmic phase (**Figure 12**). Their difference in expression between 3 and 28◦C was significant at growth points II and IV, over 16-fold at the largest for yptb1395, but the three genes showed more transcripts at 3◦C consistently throughout the growth.

### Expression of Genes Involved in Modifying Membrane Lipid Composition

Genes yptb1450 and yptb2469, encoding FabA and FabF, were expressed highly at the beginning of growth whereas gene yptb2426 encoding FabB at stationary phase (**Figure 13**). Gene yptb2469 showed significantly more transcripts at 3◦C at growth point II, although the difference favored cold during most of the growth.

## DISCUSSION

We identified several upregulated genes that may be important to cold adaptation of Y. pseudotuberculosis. We also analyzed gene expression profiles to better understand what happens at different phases of growth in cold. To tackle problems

the beginning of growth. Toward the end of growth at 3◦C, amino acid utilization takes precedence when sugar importer expression tapers off.

at the beginning and logarithmic phase of growth.

posed by refrigerator temperatures, Y. pseudotuberculosis has an extensive toolbox that includes Csps, DEAD-box RNAhelicases, compatible solutes, transcription factors, and fatty acid saturases.

#### Nutrient Acquisition

Y. pseudotuberculosis expresses many genes involved in securing nutrients significantly more at 3◦C compared to 28◦C (**Figures 4**,**5**), as successful growth in low temperatures requires more energy. In our results, a shift in nutrient utilization over time is observed. At the beginning of growth, transporters of fructose, ascorbate, maltooligosaccharides, and maltose were upregulated at 3◦C. At the beginning of logarithmic phase, transporters of glutamate, aspartate, and arginine were upregulated. Toward the stationary phase, in addition to an operon turning histidine to glutamate, transporters of cystine, methionine, and sulfates were upregulated.

PTSs allow quick acquisition of carbohydrates from the environment (Postma et al., 1993). PTSs have also been linked to regulatory functions during cold stress (Wouters et al., 2000; Monedero et al., 2007). Y. pseudotuberculosis expresses several PTSs especially at the beginning of growth significantly more at 3◦C than 28◦C, as it is in dire need of carbon to sustain

its laborious growth. Another large peak in expression of a fructose PTS can be observed at growth point V at 3◦C, whereas the peak is located at growth point IV at 28◦C. However, this is probably due to slower regulatory processes at 3◦C. In Y. pestis, the malMBKEFG-operon, which is involved in maltose intake, was highly upregulated after cold shock (Han et al., 2005). Maltose can act as an cryoprotectant in addition to being a carbon source (Jain and Roy, 2009). In our results, parts of the mal-operon were upregulated especially at the beginning of growth.

E. coli has been shown to accumulate certain amino acids, including aspartic acid, glutamic acid, and methionine,

when subjected to suboptimal temperatures (Jozefczuk et al., 2010). Accumulation has been theorized to result from protein degradation. In our results, ABC transporters for cystine and methionine were upregulated in cold at the stationary phase.

### Cold Shock Proteins

Y. pseudotuberculosis seems to express Csps encoded by yptb2414 and yptb2950 almost exclusively at low temperature, i.e., with the highest significant difference of all Csps, and at high

levels. Additionally, yptb3585, yptb3586, and yptb1423 were upregulated throughout the growth. Csps encoded by yptb2950 and yptb3585/6 are comparable to CspG and CspB of E. coli, respectively (Keto-Timonen et al., 2016). Csp encoded by yptb2414 is 80% similar to CspC and CspE of E. coli K-12 W3110.

Csp encoded by yptb1088, which is almost identical to CspE of E. coli (Keto-Timonen et al., 2016), showed more transcripts at 3◦C, but not at significant levels. However, its transcript count rose toward stationary phase like has been shown to happen in E. coli (Czapski and Trun, 2014). In E. coli, CspE was also the most abundant of all csps after cold shock and present in the bacterium at all times. The expression of yptb1392, similar to CspD of E. coli, ramped up at stationary phase in 28◦C and a similar but less dramatic bump can be seen in expression profile in 3◦C. In E. coli, CspD is not cold-induced and is expressed mainly at stationary phase (Yamanaka et al., 2001). It is involved in persister cell and biofilm formation (Kim and Wood, 2010).

In E. coli, the genes cspA, cspB, cspE, and cspG all must be deleted to achieve a cold-sensitive phenotype, and the deletion of only one or two genes leads to overexpression of the rest (Xia et al., 2001). In addition, CspC and CspE seem to double as regulatory elements in stress responses (Phadtare and Inouye, 2001; Phadtare et al., 2006). The two proteins have been shown to be required for expression of fructose

PTS at suboptimal temperatures (Phadtare et al., 2006). E. coli expresses cspA, cspB, and cspG even when its protein synthesis is inhibited by antimicrobials (Etchegaray and Inouye, 1999). In Clostridium botulinum, deletion of cspB, or cspC leads to a cold-sensitive phenotype, and cspC deletion even hinders growth at 37◦C (Söderholm et al., 2011). Genes cspB and cspC also provide the bacterium resistance to NaCl, pH, and ethanol stress (Derman et al., 2015). In Staphylococcus aureus, cspB mutation renders the bacterium susceptible to cold and various antimicrobials (Duval et al., 2010). By contrast, CspC has been shown to induce more strongly in response to antimicrobials, hydrogen peroxide, and arsenate than cold (Chanda et al., 2009).

In our experiment, the expression of the genes yptb3585– 86, comparable to cspB of E. coli (Keto-Timonen et al., 2016), dipped after the beginning of growth at 3◦C but showed more transcripts at 3◦C throughout the growth. Pathogenic Yersinia have been shown to hold a tandem gene duplication cspA1/A2 (Neuhaus et al., 1999), which is very nearly identical to genes yptb3585–87 and needs to be downregulated after the initial cold shock for growth to continue (Neuhaus et al., 2000). In Yersinia enterocolitica, cspA and cspB were the very first genes expressed after cold-shock with their transcript levels decreasing after the beginning (Bresolin et al., 2006). Similarly, cspA and cspB expression fluctuated accordingly with temperature cycling in E. coli (Ivancic et al., 2013). In light of our

results, yptb3585–86 might play an important role in the initial regulation of cold-inducible genes also in Y. pseudotuberculosis. Genes yptb2414 and yptb2950 seem to be important at low temperature growth considering their differential expression and transcript levels. It is also probable that most Csps of Y. pseudotuberculosis are interchangeable. Csp encoded by yptb1423 is an odd one out, with a similarity of 54% at best to other Csps of Y. pseudotuberculosis (Keto-Timonen et al., 2016), with a peak similarity of 50% to the Csps of E. coli K-12 W3110.

### DEAD-Box RNA Helicases

Y. pseudotuberculosis is unable to grow at 3◦C without a functional RNA helicase gene csdA (yptb0486) (Palonen et al., 2012). Of other foodborne pathogens, Listeria monocytogenes and C. botulinum have been shown to require RNA-helicases, including a CsdA homolog, to grow at suboptimal temperatures (Markkula et al., 2012; Söderholm et al., 2015). In fact, CsdA was the most abundant protein extracted from the microbial mats of Lake Joyce in Antarctica (Koo et al., 2016). CsdA and SrmB of E. coli are directly involved in the biogenesis of large ribosomal subunits, presumably at subsequent steps with some functional overlap, and deletion of either gene stunts growth at low temperatures (Charollais et al., 2003, 2004). The deletion of dbpA in E. coli did not lead to similar accumulation of ribosome precursors and stunted growth at either 25 or 37◦C (Peil et al., 2008).

Our results suggest that DbpA (encoded by yptb1652) may indeed be important at low temperatures, at least in Y. pseudotuberculosis. DbpA is missing completely from Pasteurellales, an order of symbionts and parasites growing at unchanging, high-temperature, nutrient-rich niches, which suggests that DbpA, like other DEAD-box helicases, has become specialized to operate in adverse growth conditions (Iost and Dreyfus, 2006). Amino acid similarities between corresponding helicases of Y. pseudotuberculosis IP32953 and E. coli K-12 W3110 range from 69 to 90%.

Gene yptb1214 coding RhlE had the largest and most consistent significant difference in expression in favor of 3◦C. It has been suggested by Jain (2008) that RhlE may modulate the function of CsdA and SrmB in priming immature ribosomal RNA. This would also help explain the gene's prominence at low temperature in our results at the beginning and during logarithmic phase when ribosomal activity is at its highest. In E. coli, RhlE has been shown to complement the function of both CsdA and RhlB (Khemici et al., 2004; Awano et al., 2007), but deletion of rhlE did not affect growth at 37◦C in E. coli (Phadtare, 2011).

### Compatible Solutes

Our results show that Y. pseudotuberculosis expresses genes involved in glycine betaine intake and biosynthesis significantly more at 3◦C during critical points of growth, i.e., the beginning and logarithmic phase. It has been shown that Y. enterocolitica, with almost identical proU and proP operons, accumulates glycine betaine at low temperature, but glycine betaine only protects it against osmotic stress (Park et al., 1995; Annamalai and Venkitanarayanan, 2009). However, glycine betaine has been shown to also grant cold protection to gram-positive L. monocytogenes, B. subtilis, and gram-negative Vibrio anguillarum (Ko et al., 1994; Hoffmann and Bremer, 2011; Ma et al., 2017). Furthermore, choline/carnitine/betaine transporter gene of C. botulinum was upregulated in cold shock (Dahlsten et al., 2014). The precise

mechanism behind glycine betaine meditated cold protection is unclear, but the molecule has been suggested to hinder ice crystal formation and displace water around macromolecules, reducing aggregation and denaturation (Hoffmann and Bremer, 2011).

In our results, an operon encoding spermidine efflux pump MdtIJ was expressed significantly more at 3◦C. Bacteria need spermidine for various purposes, but at low temperatures excess spermidine is detrimental (Higashi et al., 2008). Spermidine displaces magnesium-ions from ribosomes, thus deactivating them, but it also prolongs synthesis of Csps after cold shock (Limsuwun and Jones, 2000). It has been shown that prolonged synthesis of Csps after cold shock occupies all available ribosomes effectively stopping growth in Y. enterocolitica (Neuhaus et al., 2000). Our results suggest that also Y. pseudotuberculosis ejects excess spermidine to avoid these problems.

#### Rho Factor, IF-1, and RbfA

Translation factors like IF-1 and RbfA act directly on ribosomes, securing protein synthesis. Initiation factor IF-1 separates the overly stabilized large and small subunits of ribosomes so that they can fuse and begin translating again elsewhere (Giangrossi et al., 2007), while RbfA primes 16S rRNA for new small subunits (Xia et al., 2003). Both IF-1 and RbfA have been shown to be significantly more expressed during cold shock in E. coli (Xia et al., 2003; Giangrossi et al., 2007). E. coli IF-1 resembles the CspA homologs (Phadtare et al., 2007), whereas in Y.

pseudotuberculosis they are only 42–50% similar in nucleic acid and 14–21% in amino acid sequence.

The rho gene is associated with bacteria that are subjected to various stresses in their environment (D'heygère et al., 2013). Rho factor is needed at low temperature to clear the bacterial DNA of frozen ribosomes and polymerases (D'heygère et al., 2013), and the structure of its RNA-binding subunit is similar to that of a cold shock domain (Briercheck et al., 1996). Rho factor also coordinates homeostasis of magnesium ions, which are needed by ribosomes, by controlling transporter genes (Kriner and Groisman, 2015). The Rho factor has been shown to be upregulated at low temperature in at least Acidithiobacillus ferrooxidans (Mykytczuk et al., 2011), Pseudomonas haloplanktis strain TAC125 (Piette et al., 2010), and Bacillus subtilis (Quirk et al., 1993). To our knowledge, none of these factors have been linked to cold growth in Yersinia genus before.

Y. pseudotuberculosis expressed the genes encoding IF-1, RbfA, and Rho all through the critical phases of growth at high levels but also significantly more at 3◦C. Massive difference of IF-1 (yptb1395) expression in favor of cold during logarithmic phase suggests that the initiation factor is important for continued growth at low temperature. It is probable that IF-1 and RbfA of Y. pseudotuberculosis provide the bacterium with working ribosomes at low temperature. Y. pseudotuberculosis might need Rho factor not only to terminate inefficient transcription but also to provide ribosomes with much needed magnesium ions. The upregulation of mgt transporter genes at 3◦C throughout the growth supports this theory.

#### Cell Membranes

Lipid synthesis and cell growth cease in cyanobacteria until an adequate membrane lipid composition has been achieved (Sinetova and Los, 2016). However, gram-negative cyanobacteria and gram-positive Bacillus subtilis possess a two-component system, DesK-DesR, that senses increasing membrane rigidity at low temperatures (Beranová et al., 2010; Sinetova and Los, 2016). Although a corresponding two-component system has not been identified in Y. pseudotuberculosis (Palonen et al., 2011), the ratio of unsaturated to saturated fatty acids in its membranes rises at low temperature (Bakholdina et al., 2004). In another gramnegative bacterium, E. coli, temperature drop directly affects activity of the FabF enzyme in its cytosol and this enzyme is involved in unsaturated lipid synthesis with FabA and FabB (Mansilla et al., 2004). Cold stress has also been shown to induce the transcription of fabF in Y. pestis (Han et al., 2005).

It is also possible that the csp encoded by yptb2414, a homolog of E. coli CspC and CspE, is involved in cold adaptation of membranes in Y. pseudotuberculosis. CspC and CspE bind specific uracil-rich segments of mRNA, and these segments often encode hydrophobic structures of membrane proteins (Benhalevy et al., 2015). It has been suggested that bacteria use these segments as addresses for directing membrane protein mRNA to where they are needed in the cell (Benhalevy et al., 2015).

#### CONCLUSIONS

Y. pseudotuberculosis has extensive tools to keep its protein synthesis running at low temperature. It is probable, that the functions of its Csps, RNA helicases and factors acting on ribosomes overlap greatly and thus form a robust network to protect nucleic acids from cold damage. Csps encoded by yptb1423, yptb2414, yptb2950, yptb3585–86, and RNA helicases CsdA, RhlE, and DbpA, seem to form the backbone of cold survival of Y. pseudotuberculosis with their regulatory and nucleic acid unwinding functions. IF-1, RbfA, and by extension, Rho, keep the ribosomes of Y. pseudotuberculosis running even at suboptimal temperatures. Rho factor also terminates frozen ribosomes and RNA polymerases, freeing scant resources under cold stress.

#### REFERENCES


Increased expression of motility, chemotaxis, and nutrient uptake genes at low temperature suggests that Y. pseudotuberculosis actively tries to find and secure sufficient resources to grow in refrigerator temperatures. The nutrients acquired are diverse and the nutrient profile seems to shift as growth progresses. The bacterium also possibly changes its membrane lipid composition with Fab enzymes, much like E. coli, to battle increasing membrane rigidity.

The exact role of glycine betaine in cold resistance is uncertain, but Y. pseudotuberculosis seems to accumulate it during cold growth like many other bacteria. Genes involved in defense against foreign DNA as well as oxidative stress were upregulated at 3◦C, which would seem to suggest that stress responses are linked in some way.

#### AUTHOR CONTRIBUTIONS

RK-T and HK designed the study. NS and RK-T performed the experiments. J-PV and KJ performed the transcriptome analysis. J-PV, KJ, RK-T, and HK contributed to the data analysis and interpretation. J-PV drafted the manuscript. RK-T, NS, and HK contributed to manuscript revision. All authors have read and approved the final manuscript.

#### FUNDING

This work was funded by the Finnish Veterinary Foundation and the Walter Ehrström Foundation.

### ACKNOWLEDGMENTS

We gratefully acknowledge Dr. Elisabeth Carniel, Institut Pasteur, Paris, France for providing the Y. pseudotuberculosis IP32953 strain. We thank Henna Niinivirta, Erika Pitkänen, and Kirsi Ristkari for technical assistance.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00416/full#supplementary-material

with cold shock proteins in Escherichia coli. PLoS ONE 10:e0134413. doi: 10.1371/journal.pone.0134413


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Virtanen, Keto-Timonen, Jaakkola, Salin and Korkeala. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Prevalence of Yersinia Species in the Ileum of Crohn's Disease Patients and Controls

Guillaume Le Baut 1,2, Claire O'Brien3,4, Paul Pavli 3,4, Maryline Roy <sup>1</sup> , Philippe Seksik <sup>5</sup> , Xavier Tréton1,6, Stéphane Nancey 7,8, Nicolas Barnich<sup>9</sup> , Madeleine Bezault <sup>10</sup> , Claire Auzolle<sup>10</sup>, Dominique Cazals-Hatem1,6, Jérome Viala1,11, Matthieu Allez <sup>10</sup> , The REMIND GROUP, Jean-Pierre Hugot 1,11 \* † and Anne Dumay 1†

#### Edited by:

Matthew S. Francis, Umeå University, Sweden

#### Reviewed by:

Maristela Gomes De Almeida, Hospital Edmundo Vasconcelos, Brazil Wojciech Marlicz, Pomeranian Medical University, Poland

#### \*Correspondence:

Jean-Pierre Hugot jp.hugot@orange.fr

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 13 July 2018 Accepted: 30 August 2018 Published: 21 September 2018

#### Citation:

Le Baut G, O'Brien C, Pavli P, Roy M, Seksik P, Tréton X, Nancey S, Barnich N, Bezault M, Auzolle C, Cazals-Hatem D, Viala J, Allez M, The REMIND GROUP, Hugot J-P and Dumay A (2018) Prevalence of Yersinia Species in the Ileum of Crohn's Disease Patients and Controls. Front. Cell. Infect. Microbiol. 8:336. doi: 10.3389/fcimb.2018.00336 <sup>1</sup> UMR1149 INSERM, Research Centre on Inflammation, Université Paris Diderot-Sorbonne Paris-Cité, Paris, France, <sup>2</sup> Department of Gastroenterology and Nutrition, Centre Hospitalier Universitaire de Caen, Caen, France, <sup>3</sup> IBD Research Group, Canberra Hospital, Canberra, ACT, Australia, <sup>4</sup> Australian National University Medical School, Canberra, ACT, Australia, <sup>5</sup> Gastroenterology Unit, CNRS, INSERM, ERL 1157, LBM, APHP, Saint Antoine Hospital, Sorbonne Universités, UPMC Univ Paris 06, Ecole Normale Superieure, Paris, France, <sup>6</sup> Departments of Gastroenterology and Pathology, Hôpital Beaujon, Assistance Publique Hopitaux de Paris, Paris, France, <sup>7</sup> Department of Gastroenterology, Hospices Civils de Lyon, Lyon-Sud Hospital, Pierre-Bénite, France, <sup>8</sup> INSERM U1111, International Center for Research in Infectiology, Lyon, France, <sup>9</sup> UMR 1071 Inserm/Université d'Auvergne, USC-INRA 2018, Microbes, Intestin, Inflammation et Susceptibilité de l'Hôte (M2iSH), CRNH Auvergne, Clermont-Ferrand, France, <sup>10</sup> Department of Gastroenterology, Saint-Louis Hospital, APHP, INSERM U1160, University Denis Diderot, Paris, France, <sup>11</sup> Department of Pediatric Gastroenterology, Hôpital Robert Debré, Assistance-Publique Hôpitaux de Paris, Paris, France

Yersinia are common contaminants of food products, but their prevalence in the human gut is poorly documented. Yersinia have been implicated in Crohn's Disease (CD, an inflammatory bowel disease) however their role in CD is controversial. We performed highly sensitive PCR assays of specific sequences for the gyrB gene of Y. aldovae, Y. bercovieri, Y. enterocolitica, Y. intermedia, Y. mollaretii and the inv gene of Y. pseudotuberculosis. We analyzed a total of 470 ileal samples taken from 338 participants (262 CD patients and 76 controls) belonging to three independent cohorts. All patients and controls were phenotyped and genotyped for the main CD susceptibility variants: NOD2, ATG16L1, and IRGM. Yersinia were found in 7.7% of ileal samples (respectively 7.9 and 7.6% in controls and CD patients) corresponding to 10% of participants (respectively 11.8 and 9.5% in controls and CD patients). Y. enterocolitica, Y. pseudotuberculosis and Y. intermedia were the most frequently identified species. The bacteria were more frequent in resected specimens, lymph nodes and Peyer's patches. Yersinia were no more likely to be detected in CD tissues than tissues from inflammatory and non-inflammatory controls. CD patients treated with immunosuppressants were less likely to be Yersinia carriers. In conclusion, this work shows that Yersinia species are frequently found at low levels in the human ileum in health and disease. The role of Yersinia species in this ecosystem should now be explored.

Keywords: yersinia, Crohn's disease, gut microbiota, ileal mucosa, innate immunity, mucosal immune system, molecular test

## INTRODUCTION

The genus Yersinia comprises 18 species, 3 of which are well-recognized human pathogens: Y. pestis, the causative agent of plague, and the two enteropathogenic species Y. pseudotuberculosis and Y. enterocolitica. The 15 other species are Y. aldovae, Y. aleksiciae, Y. bercovieri, Y. entomophaga, Y. frederiksenii, Y. intermedia, Y. kristensenii, Y. massiliensis, Y. mollaretii, Y. nurmii, Y. pekkanenii, Y. rohdei, Y. ruckeri, Y. similis, and Y. wautersii.

Various culture-dependent, immunological, and molecular techniques are currently available for the detection of pathogenic Yersinia (Gupta et al., 2015). The presence of Y. enterocolitica and Y. pseudotuberculosis can be determined quantitatively by direct culture on selective agar plates but confirmatory tests require a combination of cold enrichment and subcultures. Cultureindependent methods have been developed to detect pathogenic Y. enterocolitica and Y. pseudotuberculosis. These PCR assays usually target the yadA or virF gene located on the pYV plasmid, and thus only detect virulent strains. Consequently, little is known about the presence of non-pathogenic Yersinia species in most of studied samples. Yersinia prevalence in food products, especially for low virulence species is thus undoubtedly underestimated, especially considering that Yersinia surveillance is not systematic in industrial processes.

Despite these limitations, Yersinia (mainly enterocolitica) have been associated with a variety of foods, including milk and milk products, raw meat (beef, pork, chicken, and lamb), poultry, eggs, vegetables, bean sprouts, tofu, seafood, and others. Yersinia species are able to propagate in vacuum-packed foods and at refrigeration temperature. Refrigeration of contaminated foods at manufacturing and consumer sites may provide Yersinia species an opportunity to survive and thrive in food. A recent shift toward the increased consumption of processed foods wherein contamination can occur after pasteurization has also potentiated the risk of outbreaks. Expansion in international food trade and changes in livestock farming and food industry have also led to the emergence of yersiniosis globally (Gupta et al., 2015).

As an example, a recent study performed in Turkey found 84 (28%) out of 300 food samples were contaminated by Yersinia species (Özdemir and Arslan, 2015). Of the food samples analyzed, 41% of 120 meat products and 19% of 180 milk products contained Yersinia species. Y. enterocolitica was the most frequently detected species (N = 18), followed by Y. rohdei (N = 15), Y. intermedia (N = 14), Y. pseudotuberculosis (N = 12), Y. ruckeri (N = 12), Y. mollaretii (N = 5), and Y. bercovieri (N = 4). In another study from Austria, 90/120 (75%) of meat or fish samples contained Yersinia (Hilbert et al., 2003). In France, Yersinia was detected in eggs, vegetables, pastries, and others (Le Guern et al., 2016). Exposure to Yersinia in our food products appears to be common across continents.

Following food ingestion, about 10% of bacteria survive the acidic gastric environment. If they do survive, they may translocate the gut barrier via the follicle-associated epithelium, which comprises Peyer's patches in the small bowel, and isolated lymphoid follicles in the large bowel. Following translocation, Yersinia may be phagocytosed by macrophages and drain to neighboring lymph nodes via lymphatic vessels, and less often to the portal blood stream.

It has been proposed that Yersinia species may contribute to the occurrence or persistence of gut inflammation in Crohn's Disease (CD) patients (Hugot et al., 2003), which is a chronic relapsing inflammatory condition of the digestive tract. CD can occur anywhere along the gastrointestinal tract, but most commonly affects the terminal ileum. Like yersiniosis, CD inflammation is initially focused around the follicle-associated epithelium (Fujimura et al., 1996; Krauss et al., 2012) and may progress into deeper, more diffuse ulcerations. Lymphatic vessels have been shown to play a key role in the dissemination in both yersiniosis (Von Der Weid and Rainey, 2010) and CD (Randolph et al., 2016). CD is characterized by an increased and persistent reactivity against the gut microbiota and similar observations have been made in mice infected with Yersinia (Fonseca et al., 2015). Several NOD2 gene variants and a variant on the ATG16L1 gene are strongly associated with CD. In mice, Nod2 mutations homologous to the human CDassociated mutations favor the survival of animals orally infected with Y. pseudotuberculosis (Meinzer et al., 2012). Similarly, the CD-associated Atg16l1 mutation has been associated with decreased clearance of Y. enterocolitica and a concurrent increase in inflammatory cytokine production (Murthy et al., 2014).

Yersinia infection in CD patients has been explored using conventional methods. Sera from CD patients were found to be reactive to enteropathogenic Yersinia more often than control sera (Saebo et al., 2005). Five studies used Yersiniaspecific sequence primers to detect Yersinia species in CD tissues (Kallinowski et al., 1998; Lamps et al., 2003; Knösel et al., 2009; Chiodini et al., 2013; Leu et al., 2013). Overall 0–63% of CD samples were positive compared to 0–31% of inflammatory or non-inflammatory controls. In all studies, Yersinia were found in CD and/or controls suggesting that Yersinia is a common inhabitant of the human gastrointestinal tract. Differences in patient and control cohorts, tissues analyzed, and PCR methods may explain the large variations reported. Of note, Knosel et al. found more Yersinia species in control than CD archived tissues. In their study, PCR signals were weak suggesting that a non-specific sequence may have been amplified or alternatively that contaminations may have occurred. In none of the above studies, except for Chiodini's work, were the PCR products sequenced in order to confirm their specificity.

To further document the presence of Yersinia in intestinal tissues of CD patients and controls, we analyzed samples from three cohorts, representing a total 470 samples from 338 participants. All patients and controls were phenotyped and genotyped for the most strongly associated CD susceptibility gene variants. All samples were analyzed under the same conditions, using a validated sensitive PCR method capable of detecting six Yersinia species. PCR products were systematically sequenced to confirm the presence of Yersinia. Correlations between Yersinia positivity and clinical or genetic parameters were assessed.


CD, Crohn's Disease; UC, Ulcerative Colitis; IBD, Inflammatory Bowel Disease; <sup>a</sup>at diagnosis; <sup>b</sup>at enrolment; \*one or more of the R702W, G908R, or L1007fs-insC mutations. Cohort 1 and 3: post-surgical resected tissues, cohort 2: endoscopic mucosal biopsies. Montreal Classification is reported for CD patients (Satsangi et al., 2006).

#### MATERIALS AND METHODS

#### Patients

Three cohorts were used in this study. Cohort 1 is an Australian cohort of 65 patients [34 CD, 7 Ulcerative Colitis (UC), 24 others] previously published by O'Brien et al. (2014). Samples were taken from involved (n = 59) and uninvolved (n = 53) areas on ileal resected specimens and mesenteric lymph nodes (n = 53). Cohort 2 is a French cohort of 62 patients (17 CD, 14 UC and 31 non-inflammatory controls). Mucosal biopsies were collected at diagnosis, and represent 45 Peyer's patches and 51 non-inflammatory ileal mucosa. Cohort 3 consisted of 211 CD resected bowel specimens from the REMIND study (Fumery et al., 2017), representing 211 inflamed ileal samples. The main characteristics of the three cohorts are summarized in **Table 1**.

All participants provided a written informed consent and the study was approved by the relevant ethic committees in France and Australia [Australian National University Human Ethics Committee approval (ref 2012/596), ACT Health Human Research Ethics Committee approval (ref ETH.5.07.464), AFFSAPS approval (ref IDRCB: 2009-A00205-52) and French ethic committee Hôpital Saint Louis (ref 2009/17)].

#### DNA Extraction and Amplification

For cohort 1, DNA was extracted using DNeasy Blood and Tissue kit (Qiagen) as previously described (O'Brien et al., 2014). For cohort 2, DNAs were extracted using QIAamp DNA mini kit (QIAgen) according to standard protocols. For the third cohort, DNA extraction was performed from specimens preserved in RNAlater, using Trizol reagent (Invitrogen life technologie) with NucleoSpin Tissue (Macherey-Nagel).

DNA quantity and quality were checked for all samples using a Nanodrop ND-1000 spectrophotometer (NanoDrop Technologies, USA). Of the total of 509 DNA samples, 39 with DNA concentrations less than 20 ng/µl were removed. DNA samples were pre-amplified using the GenomiPhi V2 DNA Amplification kit (GE Healthcare) according to the manufacturer's instructions. Briefly, 25 ng of each DNA sample was heat-denatured at 95◦C. After addition of the enzyme solution, the sample was incubated at 30◦C for 90 min. DNA polymerases were then heat-inactivated at 65◦C for 10 min and samples were stored at −20◦C until use.

#### Yersinia Identification

Unlike of previous studies, we wanted to identify several Yersinia species frequently encountered in contaminated food. We focused on Yersinia-specific sequences present in the genome regardless their pathogenicity and ability to differentiate Yersinia species. In silico analyses directed us to focus on gyrB, a gene coding for a subunit of DNA gyrase. Proteins encoded by gyrB exhibit considerable variation between species (Mun Huang, 1996). Their substitution rates have been estimated to be 4-fold higher than RNA 16S and gyrB has therefore been used to differentiate species and genera phylogenetically. **Figure 1A** shows the main differences between DNA sequences of different species of Yersinia and E. coli. In addition to gyrB, we determined the presence of inv, a virulence gene present in 100% of Y. pseudotuberculosis strains (Thoerner et al., 2003), and ail, a gene which is a marker of pathogenicity of Y. enterocolitica.

PCR assays were performed with 1.2 µl of pre-amplified DNA in a 20 µl volume containing Platinum blue reagent (Invirogen life technologie) together with reverse and forward primers (0.2µM of each). Our PCR assays were specific for the following six species: Y. aldovae, Y. bercovieri, Y. enterocolitica, Y. intermedia, Y. mollaretii, and Y. pseudotuberculosis. The primer sequences are provided in **Table 2**. PCR reactions were carried out using an Eppendorf Mastercycler (Eppendorf, Germany). An initial denaturation was performed at 95◦C for 10 min, followed by 35 cycles of 95◦C for 30 s, annealing temperature for 30 s and 72◦C for 30 s. A second PCR reaction of 35 cycles was performed using 1.2 µl of the first-round PCR products, using the same conditions described above. All PCR reactions included negative controls.

PCR products were then analyzed on 1.5% agarose gels and stained with SYBR-safe (Invitrogen). Sequencing was

FIGURE 1 | Validation of the PCR methods. (A) Comparison of GyrB sequences between different Yersinia species and E.coli. In red are indicated the nucleic acids present in all species. In blue is indicated the consensus sequence (Corpet, 1998). YA, Yersinia aldovae; YB, Yersinia bercovieri; YE, Yersinia enterocolitica; YI, Yersinia Intermedia; YM, Yersinia Mollaretii. (B) Specificity of the PCR methods developed for Y. frederiksenii (GyrB). DNAs from several species of Yersinia were amplified with the PCR technique and loaded on an agarose gel. A band indicates the presence of the specific PCR product specific to Y. frederiksenii. (C) Sensitivity of the PCR method developed for Y. enterocolitica (GyrB). The quantity of amplified DNA is expressed as genome equivalents. The method was highly sensitive when two cycles of 35 amplifications were performed.


TABLE 2 | Primers used for Yersinia species identification.

subsequently performed on all gel-purified PCR products (QIAquick extraction kit, Qiagen). Bi-allelic direct DNA sequencing was applied using BigDye Terminator v1.1 reagent (Applied Biosystems, USA), on an ABI 3730 automated genetic analyzer and analyzed using the 4peaks software. Sequences were then compared with those present in the NCBI public databases using BLAST. The presence of Yersinia was assumed if BLAST returned a homology score with Yersinia of at least 95%.

#### NOD2, ATG16L1 and IRGM Genotyping

Allelic discrimination assays were used to genotype patients and controls for five susceptibility variants: NOD2 SNPs rs2066844 (assay ID: C-11717468\_20), rs2066845 (assay ID: C\_11717466\_20), and rs2066847 (custom assay); ATG16L1 SNP rs2241880 (assay ID: C\_9095577\_20); and IRGM SNP rs10065172 (assay ID: C\_30593568\_10). These SNPs were analyzed using real-time PCR as recommended by the manufacturer (Applied Biosystems, USA). Reactions were carried out using a LightCycler 480 Systems (Roche). An initial denaturation step was performed at 95◦C for 10 min, followed by 40 cycles of 92◦C for 15 s and 1 min at 60◦C.

#### Statistics

Comparisons between groups were done using Fisher exact test for quantitative data, combined with a student T-test for qualitative data. Tests were performed using Xlstat V1.2. A Pvalue <0.05 was retained as significant.

#### RESULTS

#### Validation of the PCR Methods

Validation of the PCR techniques was based on the analysis of DNA from 10 Yersinia species (Y. intermedia, Y. bercovieri, Y. pseudotuberculosis, Y. entomophaga, Y. aldovae, Y. frederiksenii, Y. mollaretii, Y. kristensenii, Y. ruckeri, and Y. rohdei) kindly provided by Elisabeth Carniel from the international reference center of Yersinia, Institut Pasteur, Paris, France. PCRs were performed using 5 ng of DNA (corresponding to ∼10<sup>6</sup> genomes). All species except Y. rohdei were detected by the selected primers. **Figure 1B** shows the results obtained with gyrB specific primers for Y. enterocolitica.

To evaluate the sensitivity of the PCR methods, we serially diluted the DNA samples between 1 to 10,000 genomes based on an estimated genome weight of ∼4.7 × 10−<sup>6</sup> ng. The serial DNA dilutions underwent 35 PCR cycles, but no PCR products were observed in the majority of cases (**Figure 1C**). We thus performed a second round of 35 cycles, which allowed us to detect the bacteria. In these conditions, the thresholds were estimated between 1 and 100 equivalent genomes.

We also determined if the PCR techniques were efficient in detecting Yersinia in a DNA sample containing human genomic DNA. The presence of eukaryotic DNA did not alter the sensitivity of the methods (data not shown). We also confirmed that genomic pre-amplification did not significantly change the capacity of the PCR to detect Yersinia DNA (data not shown). Of note, when performed on human samples, the specificities of the PCR methods were lower than previously described with purified Yersinia DNA only. Indeed, several non-specific PCR products were evident for ileal samples, indicating that sequencing was always necessary to validate the presence of Yersinia-specific sequences in tissues.

Ultimately, we retained the PCR assays for Y. aldovae, Y. bercovieri, Y. enterocolitica, Y. intermedia, Y. mollaretii, and Y. pseudotuberculosis. For the other species, the sensitivity and/or specificity of the techniques were deemed to be insufficient.

#### Yersinia in CD and Controls

We analyzed a total of 140 samples from 76 controls (ischemic colitis, intestinal obstructions, diverticular diseases and others) (**Table 3**). Seventy nine samples were from cohort 1 and consisted of mucosal tissues and lymph nodes. We found 6/54 specimens and 3/25 lymph nodes positive for Yersinia corresponding to a



CD, Crohn's Disease; NIA, Non-inflammatory area; IA, inflammatory area; LN, lymph node; PP, Peyer's patch; IM, ileal mucosa; YE, Y. enterocolitica; YP, Y. pseudotuberculosis; YB, Y.bercovieri; YM, Y. molaretii; YA, Y. aldovae; YI, Y. intermedia. \*one patient was positive for 3 Yersinia species.

total of 11%. When several samples were available for the same individual, some were found positive, while others were not. This result indicated that the status of a given individual was highly dependent on the number of samples studied but also on the number of PCR reactions conducted on the same DNA sample (data not shown). In cohort 2, we found Yersinia DNA in 2/26 Peyer's patches (8%) and 0/35 mucosal biopsies, corresponding to a total of 3%. Thus a total of 8% of positive samples were found in the combined cohorts.

In cohorts 1 and 2, Yersinia species identified in controls were Y. enterocolitica (n = 9), Y. intermedia (n = 2), Y. pseudotuberculosis (n = 1), and Y. molaretii (n = 1). No differences were found between inflammatory and noninflammatory controls.

In CD patients (all cohorts), as for the control group, biopsies were less often positive (1/33, 3%) than resected specimens or lymph nodes (24/297, 8%). On average, no clear difference was observed between CD cases and inflammatory or noninflammatory controls. The Yersinia species encountered in CD patients were Y. enterocolitica (n = 13), Y. intermedia (n = 3), Y. pseudotuberculosis (n = 4), Y. molaretii (n = 1), and Y. aldovae (n = 1) (**Table 3**). These results are similar to those of the control groups. Of note, only 3/13 Y. enterocolitica contained the ail gene suggesting that most of the strains present in the mucosa were not fully virulent (Sihvonen et al., 2011).

#### Impact of Genetic and Clinical Parameters on the Presence of Yersinia Species

To look for an association between clinical, genetic or therapeutic parameters and the presence of Yersinia in CD patients we analyzed the data from cohorts 1 and 3 which were fully phenotyped and genotyped (**Table 4**). Positivity rates were roughly identical when gender, Montreal classification [age at diagnosis, behavior, and location) (Satsangi et al., 2006)] and genotypes were included in the analysis. Most of the therapeutic regimens also did not affect the frequency of detection, however a borderline association between the use of immunosupressants and the absence of Yersinia was found in cohorts 1 (P = 0.08) and 3 (P = 0,08). This association reached significance when pooling the two cohorts (P = 0.035).

Most patients from cohort 3 were followed prospectively for at least 6 months after surgery. It was thus possible to evaluate the prognostic value of the presence of Yersinia at surgery on disease recurrence risk. The Rutgeerts endoscopic severity index, based on the presence of small bowel ulcerations, is widely used to monitor the presence of post-operative recurrence (i > 1) or not (i < 2). The Rutgeerts index calculated at least 6 months after surgery was not associated with the presence of Yersinia at surgery, suggesting that a sample positive for Yersinia is not predictive of an early recurrence.

#### DISCUSSION

This study is the largest report of the presence of Yersinia in the human ileum. It is based on a total of 470 samples, taken from controls (140 samples) and CD patients (330 samples). The samples consisted of non-inflammatory ileal mucosa, Peyer's patches, resected bowel specimens, and lymph nodes. We looked for six different species of Yersinia with a highly sensitive PCR method, and sequenced all of the PCR products to confirm the presence of Yersinia-specific DNA.

This study proposes a new set of PCR tests based on the amplification of gyrB sequences, capable of identifying Y. aldovae, Y. bercovieri Y. enterocolitica, Y. intermedia, and Y. mollaretii. Despite a very high specificity of the tests in vitro, PCR products must always be sequenced to confirm the presence of Yersinia in tissues, because non-specific sequences are frequently encountered in vivo. These tests are highly sensitive, with detection thresholds between 1 and 100 bacteria per sample.

Given the low thresholds of bacterial detection, sampling plays an important role in the assay's results: increasing the number TABLE 4 | Presence of Yersinia in ileal samples according to clinical and genetic classification of CD patients.


\*one or more of the R702W, G908R or L1007fs-insC mutations; 5-ASA, 5 aminosalycilates.

of samples tested increases the number of positive results on a per patient basis. This point is well illustrated by cohort 1 where three samples were tested for most patients with a positivity of 21% at the patient level but only 10% at the sample level. As a consequence, it is possible that Yersinia species could be found in the ileum of many people (may be all) if one performs enough tests. As a whole, the percentage and significance of negative results in patients remains to be evaluated. The concordance between this new test and other methods like serological tests or immunohistochemistry methods also needs to be explored.

If our work suggests that Yersinia carriage is common, it also indicates that there are generally low numbers in the ileum. This result is consistent with the fact that Yersinia are rarely found in stool and mucosal samples from either CD patients or controls in gut microbiota analyses (Sihvonen et al., 2011). These are designed to explore the dominant species, and are not appropriate for detecting low abundance of Yersinia species.

Low bacterial loads are also in line with a common exposure to the few bacteria assumed to be present in contaminated food. The main bacteria encountered in the ileum are Y. enterocolitica which represent about 2/3 of the identified bacteria. Y. pseudotuberculosis and Y. intermedia are the second most commonly found here. Even if we did not test all the panel of Yersinia species, these relative frequencies seem to reflect the relative frequencies of Yersinia species in food products (Özdemir and Arslan, 2015). Among Y. enterocolitica, most of the strains did not carry the ail gene indicating that they are not fully virulent. In other words, bacteria appear to be mainly non-virulent, or transient (allochthonous), and their retrieval from the gut may simply be a marker of food contamination. Additional works looking at a relationship between food habits and the presence of Yersinia in the gut would help to resolve this question.

On the other hand, looking at Yersinia species as inoffensive bystanders passively transported by the luminal flow can be questioned. Yersinia species are able to survive in the ileum due to their facultative anaerobic properties and tolerance to the bile acids (Bottone, 1997). They are able to interact with the gut associated lymphoid tissue and we confirm here the tropism of Yersinia for M-cells, the bacteria being mainly found in superficial biopsies centered on Peyer's patches. However, superficial biopsies tended to be positive less often than resected bowel specimens (P = 0.058). This may be related to technical differences in sample collection and storage. But it may also indicate that the bacteria are more commonly detectable within the inner intestinal layers. This point is in accordance with a previous study showing the presence of Yersinia in the submucosal layer of the intestine (Chiodini et al., 2013). The relatively high frequency of bacteria in mesenteric lymph nodes also supports this idea. If true, this finding indicates that Yersinia, even low virulence strains, are able to translocate through the gut epithelial barrier and putatively the gut-vascular barrier (Spadoni et al., 2015).

It is important to note that bacteria were equally frequent in non-inflammatory, and inflammatory controls and CD patients, and that Yersinia could be found in inflammatory and non-inflammatory parts of the ileum in CD patients. Thus, CD cannot be seen as an infection by Yersinia species: the disease is not defined by its presence in the intestinal tissues. However, it cannot be dismounted that CD patients have an abnormal response toward Yersinia. Indeed, Yersinia have been associated with long-term sequelae after acute infection in mice, including destruction of the lymphatic network and loss of antigen tolerance, similar to CD (Fonseca et al., 2015). Furthermore, CD is characterized by increased adipose tissue around the lesions (fat wrapping). White adipose tissue has been recently shown to play a key role in the long-term memory of infection by Yersinia (Han et al., 2017). Thus CD could be characterized by an excessive response toward Yersinia. The abnormal inflammatory response to Yersinia in mice carrying either Nod2 or Atg16l1 CD-associated mutations (Meinzer et al., 2008; Murthy et al., 2014) further supports this idea.

This work failed to detect an association between phenotypic or genotypic subgroups of patients. Mutations in NOD2 (a gene involved in the innate immunity and bacterial sensing) or ATG16L1 and IRGM, (two genes involved in autophagy and xenophagy) are not associated with Yersinia positivity despite their roles in the host response to the bacteria. The age at onset is also not associated with the presence of Yersinia. It is more difficult to demonstrate an association between disease location and behavior (L and B values of the Montreal classification). Indeed, fistulizing, and penetrating ileal diseases are over-represented in surgical patients, which represented the largest contingent of the studied cohorts.

Common medications used to treat CD patients, including antibiotics, are not associated with the presence of Yersinia. This finding is not surprising if we consider that they are often administered prior to surgery. As a result, bacteria, including dead cells, remain detectable by PCR. Steroids, immunosuppressants or anti-TNF antibodies could be expected to decrease bacterial clearance and promote Yersinia detection. However, the presence of bacteria at very low rates in the mucosa must be distinguished from an invasive infection and the impact of immunosuppressive agents is likely to be different in these two situations. At the opposite, immunosuppressants tended to be associated with a lower rate of positive samples in two cohorts. An hypothesis to explain this finding is that azathioprine (which is the most commonly used immunosuppressant in CD), modulates intracellular GTP levels and thus interferes with the function of Rac1, a Rho GTPase (Tiede et al., 2003). Rho GTPase activities are important for bacterial internalization by host cells and subsequent dissemination and survival of bacteria (Schweer et al., 2013). A defect of bacterial internalization could be disadvantageous for bacterial survival and azathioprine could thus have a beneficial role in CD, not only via its immunosuppressive properties, but also via the reduction of Yersinia burden in the gut mucosa.

In summary, this work provides for the first time the demonstration that Yersinia species are common in human ileum. The presence of different species seems to be proportional to assumed rates of exposure in contaminated foods. Yersinia are found in the same proportions in controls and CD patients, whatever their clinical presentation. It is important now to characterize the mucosal immune response toward Yersinia species in CD patients.

#### AUTHOR CONTRIBUTIONS

GL, AD, and J-PH were involved in study design, data acquisition, analyses and writing. CO, PP were involved in study design, data acquisition, and analyses for cohort 1. JV, J-PH, DC-H, and MA were involved in data acquisition for cohort 2. PS, XT, SN, MB, CA, and The REMIND Group were involved in study design, data acquisition and analyses for cohort 3. NB, MR, and AD were involved in participants genotyping PCR analyses.

#### ACKNOWLEDGMENTS

This work has been supported by Investissements d'Avenir programme ANR-11-IDEX-0005-02, Sorbonne Paris Cite, Laboratoire d'excellence INFLAMEX. We thank all investigators (researchers, clinicians and surgeons) of the REMIND group who collected the ileal samples. We thank Rida Zouak for technical assistance and Elisabeth Carniel who gave us the reference Yersinia strains.

#### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Le Baut, O'Brien, Pavli, Roy, Seksik, Tréton, Nancey, Barnich, Bezault, Auzolle, Cazals-Hatem, Viala, Allez, The REMIND GROUP, Hugot and Dumay. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# OmpR-Mediated Transcriptional Regulation and Function of Two Heme Receptor Proteins of *Yersinia enterocolitica* Bio-Serotype 2/O:9

Karolina Jaworska, Marta Nieckarz, Marta Ludwiczak, Adrianna Raczkowska\* and Katarzyna Brzostek

Department of Applied Microbiology, Institute of Microbiology, Faculty of Biology, University of Warsaw, Warsaw, Poland

#### *Edited by:*

Victoria Auerbuch, University of California, Santa Cruz, United States

#### *Reviewed by:*

Michael Marceau, Université Lille Nord de France, France Xihui Shen, Northwest A&F University, China

> *\*Correspondence:* Adrianna Raczkowska araczko@biol.uw.edu.pl

#### *Specialty section:*

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 10 July 2018 *Accepted:* 29 August 2018 *Published:* 20 September 2018

#### *Citation:*

Jaworska K, Nieckarz M, Ludwiczak M, Raczkowska A and Brzostek K (2018) OmpR-Mediated Transcriptional Regulation and Function of Two Heme Receptor Proteins of Yersinia enterocolitica Bio-Serotype 2/O:9. Front. Cell. Infect. Microbiol. 8:333. doi: 10.3389/fcimb.2018.00333 We show that Yersinia enterocolitica strain Ye9 (bio-serotype 2/O:9) utilizes heme-containing molecules as an iron source. The Ye9 genome contains two multigenic clusters, hemPRSTUV-1 and hemPRST-2, encoding putative heme receptors HemR1 and HemR2, that share 62% amino acid identity. Expression of these proteins in an Escherichia coli mutant defective in heme biosynthesis allowed this strain to use hemin and hemoglobin as a source of porphyrin. The hemPRSTUV-1 and hemPRST-2 clusters are organized as operons, expressed from the phem−<sup>1</sup> and weaker phem−<sup>2</sup> promoters, respectively. Expression of both operons is negatively regulated by iron and the iron-responsive transcriptional repressor Fur. In addition, OmpR, the response regulator of two component system (TCSs) EnvZ/OmpR, represses transcription of both operons through interaction with binding sequences overlapping the −35 region of their promoters. Western blot analysis of the level of HemR1 in ompR, fur, and ompRfur mutants, showed an additive effect of these mutations, indicating that OmpR may regulate HemR expression independently of Fur. However, the effect of OmpR on the activity of the phem−<sup>1</sup> promoter and on HemR1 production was observed in both iron-depleted and iron-replete conditions, i.e., when Fur represses the iron-regulated promoter. In addition, a hairpin RNA thermometer, composed of four uracil residues (FourU) that pair with the ribosome-binding site in the 5′ -untranslated region (5′ -UTR) of hemR1 was predicted by in silico analysis. However, thermoregulated expression of HemR1 could not be demonstrated. Taken together, these data suggest that Fur and OmpR control iron/heme acquisition via a complex mechanism based on negative regulation of hemR1 and hemR2 at the transcriptional level. This interplay could fine-tune the level of heme receptor proteins to allow Y. enterocolitica to fulfill its iron/heme requirements without over-accumulation, which might be important for pathogenic growth within human hosts.

Keywords: OmpR, Fur, *Yersinia enterocolitica*, HemR1, HemR2

### INTRODUCTION

Yersinia enterocolitica, a member of the genus Yersinia in the family Enterobacteriaceae is a human enteropathogen which causes yersiniosis, i.e., gut-associated diseases such as enteritis, diarrhea, and mesenterial lymphadenitis (Bottone, 1997). Multiple virulence factors encoded by chromosomal and plasmid pYV-located genes are involved in Y. enterocolitica virulence (Cornelis, 2002). In addition, iron-acquisition systems are considered important pathogenicity determinants. The concentration of iron in the environment is critical for the control of bacterial metabolism. Iron limitation in the host can abolish bacterial growth, whereas a high intracellular iron concentration may damage bacterial cells due to the formation of harmful reactive oxygen species (ROS). Thus, the transport, storage, and metabolism of iron have to be tightly controlled to maintain iron homeostasis (Hantke, 2001).

A variety of mechanisms are employed by Y. enterocolitica to take up iron from the host body, including siderophore-mediated uptake and systems for acquiring iron from abundant heme or hemoproteins (Caza and Kronstad, 2013). Pathogenic strains can be divided into two groups: those producing the siderophore yersiniabactin (biotype 1B) and those unable to produce this siderophore but able to use ectogenic siderophores released by other bacteria, such as ferrioxamin B and E or ferrichrome (biotypes 2–5) (Heesemann, 1987; Bäumler et al., 1993). About 70% of the iron in the human host is present within heme and/or hemoproteins (hemoglobin, myoglobin, cytochromes). A heme uptake system was identified previously in Y. enterocolitica bio-serotype1B/O:8 strain WA-C (Stojiljkovic and Hantke, 1992, 1994). This system, involved in the acquisition and transport of the entire heme moiety into the cytoplasm, consists of the receptor HemR, an ATP-binding cassette (ABC) transporter HemTUV, and a putative heme-degrading protein HemS. The energy for heme uptake is transferred from the inner to the outer membrane via the TonB/ExbB/ExbD system (Krewulak and Vogel, 2011). The protein TonB spans the periplasm and can physically interact with a highly conserved region of the receptor HemR called the TonB box. A conformational change in the heme receptor caused by this interaction permits the transport of the ligand across the OM into the periplasm (Nau and Konisky, 1989; Braun et al., 1991). Gene clusters responsible for the uptake and transport of heme have been identified in several bacterial species, including Yersinia pestis (hmuRSTUV) (Thompson et al., 1999), Yersinia pseudotuberculosis (hmuRSTUV) (Schwiesow et al., 2018), Shigella dysenteriae (shu genes) (Wyckoff et al., 1998), and Vibrio cholerae (hut genes) (Occhino et al., 1998), although their organization varies. The heme gene cluster of Y. enterocolitica bio-serotype 1B/O:8 strain WA-C contains six open reading frames (ORFs) hemPRSTUV, organized in an operon whose expression is enhanced by iron deprivation (Stojiljkovic and Hantke, 1992, 1994). To avoid high concentrations of free intracellular iron, the ferric uptake regulator (Fur) with Fe2<sup>+</sup> cofactor efficiently represses the transcription of appropriate genes. Fur is a global transcriptional regulator that tightly controls the transport, storage, and metabolism of iron in many Gram-negative bacteria (Hantke, 2001). In addition, Fur represses the expression of genes involved in the regulation of multiple cellular functions such as the oxidative stress response, energy metabolism, acid tolerance, and virulence (Hassett et al., 1996; Ochsner and Vasil, 1996; Bijlsma et al., 2002; van Vliet et al., 2003).

Bacterial cells are constantly challenged by various environmental stresses in their natural habitats. Y. enterocolitica faces several different challenges during infection and colonization of the human body. Efficient adaptation to changing environmental conditions is possible due to the activity of sensory regulators such as TCSs, e.g., the EnvZ/OmpR signaling pathway. This prototype TCS, first identified and characterized in non-pathogenic Escherichia coli K-12, consists of the transmembrane histidine kinase EnvZ and the response regulator OmpR. EnvZ senses changes in the environment, undergoes autophosphorylation and then a phosphate group is transferred to OmpR. Conformational changes in the phosphorylated OmpR allow it to bind to DNA as a dimer and modulate gene expression (Kenney, 2002). EnvZ/OmpR is involved in the transcriptional regulation, in a positive or negative manner, of several genes/operons of E. coli in response to changes in osmolarity, pH, temperature, and the concentration of nutrients in the environment (Slauch and Silhavy, 1989; Higashitani et al., 1993; Shin and Park, 1995; Vidal et al., 1998; Yamamoto et al., 2000; Jubelin et al., 2005). OmpR was recognized as a pleiotropic regulator that controls the expression of genes involved in many different cellular processes such as chemotaxis, motility, drug sensitivity, or acid resistance, and virulence of pathogenic bacteria (Bernardini et al., 1990; Chatfield et al., 1991; Shin and Park, 1995; Bang et al., 2002; Feng et al., 2003; Stincone et al., 2011).

The relationship between virulence and the activity of the OmpR protein has also been described for Y. enterocolitica O:8 (Dorman et al., 1989) and Y. enterocolitica 2/O:9 (our studies). We have shown that OmpR is involved in the control of various cellular processes and functions in Y. enterocolitica, including adhesion, invasion, motility, Yop production, biofilm formation, and multidrug and serum resistance (Brzostek et al., 2003, 2007, 2012; Raczkowska et al., 2010, 2011a,b, 2015; Brzóstkowska et al., 2012; Skorek et al., 2013). Taken together, these findings have revealed the important role of OmpR in remodeling the cell surface and in the adaptation of Y. enterocolitica to different environmental niches, including the host body.

Comparative proteomic LC-MS/MS analysis of outer membranes prepared from Y. enterocolitica bio-serotype 2/O:9 strain Ye9 and its isogenic ompR deletion mutant AR4 identified HemR, an ortholog of enterobacterial heme receptor proteins, including HemR of Y. enterocolitica strain 8081 of bio-serotype 1B/O:8 (99% amino acid identity), as subject to negative regulation by OmpR. Moreover, preliminary data suggested an indirect role for OmpR in regulating hemR expression (Nieckarz et al., 2016).

Here, we demonstrate that the Y. enterocolitica Ye9 genome contains two multigene clusters, hemPRSTUV-1 and hemPRST-2, encoding homologous heme receptor proteins HemR1 and HemR2. Furthermore, we show that both clusters are organized as operons that are negatively regulated by Fur. More importantly, OmpR directly represses the expression of hemR1 and hemR2 through interaction with binding sequences located in the promoter regions of their respective operons. In addition, a zipper-like RNA structure closely resembling a FourU RNA thermosensor was recognized within the hemR1 5 ′ -untranslated region (5′ -UTR), but hemR1 expression was not subject to thermoregulation at the post-transcriptional level. Our results suggest that OmpR controls HemR1 expression in both the absence and presence of the Fur repressor. Finally, we evaluated the importance of both heme receptor proteins in iron/heme acquisition by Y. enterocolitica and E. coli cells.

### MATERIALS AND METHODS

#### Bacterial Strains and Growth Conditions

The bacterial strains used in this study are described in **Table S1**. E. coli strains for plasmid manipulation and propagation were grown at 37◦C in LB medium (10 g/l tryptone, 5 g/l yeast extract, 5 g/l NaCl). E. coli strain SASX77 (1hemA mutant), was grown in the presence of 50 mM aminolevulinic acid (ALA; Sigma-Aldrich) unless otherwise indicated. Y. enterocolitica strains were cultured at 26 or 37◦C in LB medium. To achieve irondepleted conditions, cultures were grown in LBD medium, i.e., LB supplemented with an inhibitory concentration of 150µM 2,2′ -dipyridyl (DPD; Sigma-Aldrich). Antibiotics were used at the following concentrations: nalidixic acid (Nal), 30µg/ml; chloramphenicol (Cm), 25µg/ml; kanamycin (Km), 50µg/ml; gentamicin (Gm), 40µg/ml; tetracycline (Tet), 12.5µg/ml; spectinomycin (Sp), 100µg/ml.

#### Assay for Utilization of Heme-Containing Compounds as an Iron Source

Y. enterocolitica strain Ye9 grown under iron-depleted conditions was tested for the ability to obtain iron from heme-containing compounds. Initially, the growth yield (OD600) of this strain in LB supplemented with the iron chelator (DPD) to final concentrations of 0, 100, 150, 200, and 300µM, was assessed. Cultures were incubated at 26 or 37◦C for 28 h. Subsequently, heme and hemoglobin at final concentrations of 10 and 2.5µM, respectively, were added to LBD (150µM DPD) and the growth yield determined as above.

#### Growth of *E. coli* Strain SASX77 (1*hemA* Mutant) Carrying Plasmids Expressing *Y. enterocolitica* HemR1 or HemR2 Proteins

The E. coli 1hemA strain SASX77, which is defective in ALA synthetase, was transformed with the plasmid pACYC184 or its derivatives pHEM1 and pHEM2, and maintained on LB agar supplemented with 50µM ALA and Tet. Overnight cultures of these strains were adjusted to an OD<sup>600</sup> of 0.1 and 1 ml was harvested by centrifugation (2,400 × g, 2 min, RT). The pelleted cells were resuspended in 100 µl of 0.9% (w/v) NaCl, then mixed with 3 mL of 0.75% (w/v) agarose, and spread on LB agar plates. After solidification of the agarose, 5-mm diameter circles of Whatman filter paper (thickness 0.88 mm) were placed on the plates and these were wetted with 10 µl of different test solutions: 10 mM hemin, 0.1 mM hemoglobin, 0.9% (w/v) NaCl, or 50 mM ALA. The plates were photographed after incubation at 37◦C for 48 h.

### CAS Assay

Chrome azurol S (CAS) agar plates, prepared as described previously (Schwyn and Neilands, 1987; Neilands, 1994), were used to monitor siderophore production. CAS is an iron-dye complex that changes color from blue to orange when iron (Fe3+) is removed by the action of siderophores. Strains were grown overnight in LBD medium at 26◦C and 10 µl aliquots were spotted onto the plates. The plates were photographed after incubation at 26◦C for 48 h.

### Molecular Biology Techniques

All DNA manipulations, polymerase chain reactions (PCRs), restriction digests, ligations, and DNA electrophoresis, were performed as previously described (Sambrook and Russell, 2001). Plasmid and genomic DNA were isolated using a Plasmid Miniprep DNA purification Kit and Bacteria & Yeast Genomic DNA Purification Kit (EurX), respectively. When the amplified fragments were used for cloning, PCR was performed using DreamTaq DNA polymerase or Phusion High-Fidelity DNA polymerase (Thermo Scientific). Oligonucleotide primers for PCR and sequencing were purchased from Sigma Aldrich and are listed in **Table S2**. DNA fragments amplified by PCR were purified using a PCR/DNA Clean Up kit (EurX). The plasmids used in this study are described in **Table S1**. DNA sequencing was performed by Genomed S.A. (Warsaw, Poland).

#### Semi-quantitative Reverse Transcription (RT)-PCR Gene Expression Analysis

Cultures of Y. enterocolitica Ye9 were grown until OD<sup>600</sup> ∼1 in LBD medium at 37◦C and then total RNA was isolated from 10<sup>9</sup> cells using a High Pure RNA Isolation Kit (Roche). Following treatment with RNase-free DNase I (TURBO DNA-freeTM Kit, Invitrogen), the RNA was reversetranscribed using Maxima H Minus reverse transcriptase (Thermo Scientific) primed with random hexamers. The cDNA was used as the template in PCRs (RNA as a negative control) with primer pairs RThPR1F/RThPR1R, RThRS1F/RThRS1R, and RThPR1F/RThPV1R (**Table S2**), specific for hemPR-1, hemRS-1, and hemPRSTUV-1, respectively, or RThPR2F/RThPR2R, RThRS2F/RThRS2R, RThRS2F/RThPV2R, and RThPR2F/RThPV2R specific for the hemPR-2, hemRS-2, hemRST-2, and hemPRST-2 mRNAs, respectively. The amplified fragments were mixed with RunSAFE stain (Cleaver Scientific), resolved by electrophoresis on 2% (w/v) agarose gels and visualized with a GE Healthcare AI600 Imager.

### Construction of Transcriptional Fusion Plasmids phem-1::*lacZ* and phem-2::*lacZ*

To obtain transcriptional fusions of the phem−<sup>1</sup> and phem−<sup>2</sup> promoters with the lacZ gene, fragments of the hemPRSTUV-1 and hemPRST-2 operons containing the predicted promoters were first amplified from Y. enterocolitica Ye9 genomic DNA by PCR using the primer pairs hemP1F/hemP1R and hemP2F/hemP2R, respectively (**Table S2**). The obtained amplicons were digested with the restriction endonucleases EcoRI/KpnI and cloned into the corresponding sites of reporter vector pCM132Gm [derivative of plasmid pCM132 (Marx and Lidstrom, 2001) containing a Gm resistance cassette] upstream of a promoterless lacZ gene. The resulting constructs were verified by PCR using the primer pair pCM132GmSpr1/pCM132Spr2 (flanking the EcoRI and KpnI sites) followed by sequencing of the amplicons. The two constructs, named pCM1 (phem−<sup>1</sup> ::lacZ) and pCM2 (phem−<sup>2</sup> ::lacZ), were introduced into E. coli BW25113. These constructs were also introduced into E. coli S17-1 λpir and transferred by conjugation into Y. enterocolitica Ye9N and the ompR mutant AR4. Selection of transconjugants was carried out on LB plates containing Gm and Nal (Ye9N) or Gm and Km (AR4). The presence of the plasmids in the Y. enterocolitica strains was confirmed by plasmid isolation and PCR with the primer pair pCM132GmSpr1/pCM132Spr2.

#### Construction of a HemR1′ - ′GFP Translational Fusion

To measure post-transcriptional regulation of hemR1 expression, a translational fusion with GFP was constructed in plasmid pFX-P (Schmidtke et al., 2013) using the Golden Gate technique (Engler et al., 2008). A DNA fragment carrying the 5′ -UTR of hemR1 plus the first 16 codons of the gene was amplified by PCR using the primer pair hemR-fw/hemR-rev and Ye9 genomic DNA as the template. In a separate PCR a fragment containing the lacZ promoter was amplified using primer pair OR181-placfw/OR182-plac-rv and plasmid pFX-P as template. The PCR primers contained BsaI sites and additional sequences designed to generate ends compatible with BsaI-cleaved vector pFX-P (**Table S2**). In a 20 µl Golden Gate cloning reaction, 40 fmol of vector were mixed with 40 fmol of each PCR product, 5 units of Eco31I (BsaI isoschizomer, Thermo Scientific), and 4.5 units of T4 DNA ligase (Thermo Scientific) in ligase buffer. The reaction was incubated at 37◦C for 1 h, 5 min at 50◦C, followed by 5 min at 80◦C, and then it was used to transform E. coli BW25113. Clones were selected on LB agar supplemented with Sp and their identity confirmed by PCR and plasmid DNA sequencing.

#### Construction of *fur* Deletion Mutants

The 1fur::Gm deletion mutants of Y. enterocolitica Ye9N and the ompR mutant AR4 were obtained by homologous recombination using suicide vector pDS132 (Philippe et al., 2004). A fur gene mutated by the insertion of a Gm<sup>R</sup> cassette was constructed by overlap extension PCR using primers listed in **Table S2**. As the first step, three DNA fragments were PCR-amplified using Y. enterocolitica genomic DNA (for flanking regions) or plasmid pBBR1MCS-5 Gm<sup>R</sup> (for the Gm<sup>R</sup> cassette) as the templates. Fragment A, extending from 403 bp upstream of fur to 6 bp within the ORF was amplified using primer pair Fur1/Fur2; fragment B, an 803-bp Gm<sup>R</sup> cassette was amplified using primer pair Fur2/Fur3; fragment C, comprising the last 20 bp of the fur ORF plus 557 bp downstream of this ORF was amplified using primer pair Fur5/Fur6. A mixture of these three amplicons was then used as the template with flanking primers (Fur1/Fur6) in a PCR to generate the mutagenic fragment (1,860 bp). This was purified, digested with XbaI, and then cloned into the corresponding site of suicide vector pDS132. The resulting plasmid pDSfur was propagated in E. coli S17-1 λpir, with selection on Cm and Gm, and then sequenced to confirm the absence of errors. This plasmid was introduced into two Y. enterocolitica strains, Ye9N and AR4, by biparental mating. Transconjugants created by single crossovers to integrate the allelic exchange plasmid into the Ye9N or AR4 genomes were selected in LB supplemented with Cm, Gm plus Nal (Ye9N), or Km (for AR4). Plasmid integration after a single crossover was verified by PCR. To force the second recombination, the transconjugant strains were plated on LB agar containing Gm and 10% (w/v) sucrose, and incubated at room temperature for 48 h. Sucrose-resistant colonies were screened for the loss of Cm resistance (encoded by the vector). The correct allelic exchange was verified for these fur mutants by PCR using the primer pair Fur0/Fur7 (**Table S2**) and by sequencing.

#### Gel Electrophoresis, Preparation of Cell Extracts, and Western Blotting

For immunological detection of the HemR1 and HemR1′ - ′GFP proteins, Y. enterocolitica strains were grown under the desired conditions and extracts prepared from equal numbers of cells. The protein concentrations in the extracts were determined using the RC-DC protein assay (Bio-Rad) and if necessary they were diluted in Laemmli buffer to achieve equal loading (Sambrook and Russell, 2001). Samples were separated by electrophoresis on 12% TGX Stain-Free FastCast Acrylamide gels. Each gel was then Stain-Free activated by a 5 min exposure to UV and imaged using a GE Healthcare AI600 Imager. In some experiments, the loading of equivalent amounts of protein was controlled by Coomassie blue staining of an identical gel run in parallel. Next, the proteins were transferred to nitrocellulose membrane (Amersham Protran Western blotting membrane, nitrocellulose, pore size 0.2µM; GE Healthcare) using a wet electroblotting system (Bio-Rad). The blots were probed with polyclonal rabbit antibodies against HemR1 (1:10,000, generous gift of Jürgen Heesemann) or GFP (Sigma Aldrich, 1:10,000). Sheep anti-rabbit IgG, conjugated to horseradish peroxidase (HRP, Promega) was used as the secondary antibody (1:15,000). Positive immunoreaction was visualized using Clarity ECL Blotting Substrate (Bio-Rad) for HRP-based chemiluminescent detection (GE Healthcare AI600 Imager).

### β-Galactosidase Assays

β-Galactosidase assays were performed essentially as described by Thibodeau et al. (2004), using 96-well microtiter plates (Nest Sc. Biotech.) and a Sunrise plate reader (Tecan). Briefly, cultures grown to stationary phase were diluted to an OD<sup>600</sup> of 0.3–0.5 and 80 µl of each cell suspension were then mixed with 20 µl of POPCulture Reagent (EMD Millipore Corp) and incubated for 15 min to cause cell lysis. In the wells of a microtiter plate, 20 µl of each cell lysate were mixed with 130 µl of Z-Buffer and 30 µl of ONPG (4 mg/ml), as described by Miller (1992). For kinetic assays, the absorbance at 415 nm (relative to a blank) was measured at time intervals of 10 s, with 2 s of shaking before each reading. The assays were performed at 26◦C and monitored for up to 20 min. The β-galactosidase activity was expressed in Miller units calculated as described previously (Thibodeau et al., 2004). Each assay was performed at least in triplicate.

#### Electrophoretic Mobility Shift Assays (EMSAs)

For in vitro DNA-binding studies, recombinant OmpR-His<sup>6</sup> was expressed and purified as described previously (Nieckarz et al., 2016). Briefly, the N-terminal His-tagged OmpR protein (OmpR-His6, 29.78 kDa) was expressed from plasmid pETOmpR in E. coli BL21(DE3) and purified using Ni-NTA resin (Qiagen). The concentration of the purified OmpR protein was determined using the RC DC protein assay (Bio-Rad). DNA fragments comprising the regulatory regions upstream hem-1 and hem-2 clusters of the Y. enterocolitica Ye9 were amplified by PCR using primers listed in **Table S2** with Ye9 genomic DNA as the template. The amplicons were purified using a Gene Matrix PCR/DNA Clean-Up kit (EurX) and the concentration of DNA was determined with a NanoDrop 2000. EMSA reactions (10 µl), contained 0.05 pmol of each DNA fragment, OmpR binding buffer (50 mM Tris-HCl pH 8.0, 100 mM KCl, 1 mM EDTA, 1 mM DTT, 20 mM MgCl2, 12% glycerol, 100µg/ml BSA, 0.1% Triton X-100), and increasing amounts of OmpR-His6. After incubation at room temperature for 15 min, 2 µl of 30% (v/v) glycerol were added and the reactions were loaded onto a 4.2% native polyacrylamide gel (19:1 acrylamide/bisacrylamide, 0.2X TBE, 2% glycerol) that had been pre-run in 0.2X TBE running buffer for 40 min at 80 V. After loading, the electrophoresis was continued for 3 h at 110 V. The gels were then soaked in SYBRgreen 1X solution (Invitrogen) and visualized using a GE Healthcare AI600 imager.

#### Sequence Alignments and *in silico* Modeling

Bioinformatic analyses examined the complete genome sequences of Y. enterocolitica subsp. palearctica Y11 and Y. enterocolitica subsp. enterocolitica 8081 (GenBank; http:// www.ncbi.nlm.nih.gov/genbank/). Homology searchers were performed with BLAST software (https://blast.ncbi.nlm.nih.gov/ Blast.cgi). Promoter prediction was conducted using the webbased software BPROM in the Softberry package (http://linux1. softberry.com/berry.phtml?topic=bprom&group=programs& subgroup=gfindb; Solovyev and Salamov, 2011). Predicted secondary structures of the partial and complete hemR1 5 ′ - UTR were obtained using Mfold software (http://mfold.rna.

albany.edu; Zuker, 2003). Western blot images were analyzed with Amersham Imager 600 Analysis Software V1.0.0 (GE Healthcare).

#### Statistical Analyses

Statistical analyses were performed using Prism 7 software (v. 7.02, GraphPad). One-way ANOVA was used to determine statistically significant differences.

#### RESULTS

### *Y. enterocolitica* Ye9 Utilizes Heme-Containing Molecules as an Iron Source

The wild-type strain Ye9 is a low-level pathogenic strain belonging to Y. enterocolitica bio-serotype 2/O:9. In contrast to more highly pathogenic Y. enterocolitica strains from bioserotype 1B/O:8, it lacks the high-pathogenicity island (HPI), a chromosomal cluster of iron-regulated genes involved in the biosynthesis of siderophore yersiniabactin (Ybt) and the uptake of Fe-Ybt into cells (Pelludat et al., 1998; Carniel, 2001). To study the ability of Y. enterocolitica strains to produce siderophores, the strains Ye9 and JB580v (bio-serotype 1B/O:8) were spotted onto agar containing CAS and incubated at 26◦C for 48 h (**Figure 1A**). As anticipated, an orange halo was observed around the colonies of JB580v, demonstrating its ability to synthesize yersiniabactin (**Figure 1A**), while Ye9 growth failed to produce any color change on the CAS plates, suggesting that it does not produce siderophores. Strain Ye9 was further characterized for its ability to utilize heme-containing molecules as an iron source. The growth yield of Ye9 incubated at 26 or 37◦C in LB medium supplemented with 100–300µM of the iron chelator 2,2′ -dipyridyl (LBD) was significantly lower than in LB medium or LB with 10µM FeCl3, and varied according to the chelator concentration (**Figure 1B**). However, the addition of hemin (He, 10µM) or hemoglobin (Hb, 2.5µM) caused moderate stimulation of the growth of Ye9 in LBD (150µM 2,2′ -dipyridyl) at 37◦C, demonstrating that this strain can utilize heme-bound iron from hemin or hemoglobin (**Figure 1C**).

### Identification of Two Multigenic Clusters Encoding Putative Outer Membrane Heme Receptor Proteins HemR1 and HemR2 in the Genome of *Y. enterocolitica* Ye9

The hemR gene was previously recognized within the multigenic hemPRSTUV cluster responsible for the uptake and transport of hemin in highly pathogenic Y. enterocolitica (1B/O:8) strain WA-C (Stojiljkovic and Hantke, 1992). We established the presence of the hemPRSTUV cluster of Y. enterocolitica Ye9 (2/O:9) by PCR amplifications using primers designed from the Y11 (4/O:3) genome sequence (GenBank Acc. No. FR729477). Sequencing of these amplicons confirmed the presence of six ORFs hemPRSTUV-1 (hem-1 locus) (**Figure S1**). The sequence of this gene cluster was identical to that found in the Y11 (4/O:3) genome and it shared 99% sequence identity with hemPRSTUV of strain 8081 of bio-serotype 1B/O:8 (data not shown). The Y. enterocolitica hemR1 gene encoding the outer membrane heme receptor protein HemR1 (687 amino acids) is situated downstream of the hemP1 ORF encoding a small protein HemP1 (64 amino acids) of uncharacterized function (**Figure 2A**). The amino acid sequences of the HemR1 protein of Y. enterocolitica (2/O:9) strain Ye9 and Y. enterocolitica (1B/O:8) strain 8081 share 99% identity (data not shown). The deduced amino acid sequences of the proteins encoded by hemSTUV-1 in Ye9 are similar to the equivalent sequences of proteins of the 8081 strain.

HemS1 is possibly involved in the release of iron from heme, and HemTUV-1 comprise subunits of a periplasmic/inner membrane ABC heme permease (Stojiljkovic and Perkins-Balding, 2002).

Bioinformatic analysis using BLAST software revealed the presence of a second multigenic cluster hemPRST-2 (hem-2 locus), encoding putative hemin transporter protein HemR2, in the genomes of Y. enterocolitica strains of low (2/O:9) and high pathogenicity (8081 strain, 1B/O:8) (**Figure 2A**). As for hem-1, we determined the sequence of the hem-2 locus in Y. enterocolitica strain Ye9 (**Figure S1**). Analysis of the structure of this novel heme transport locus revealed four intact ORF hemPRST-2 organized in an order similar to the hemin transport cluster hem-1. The predicted translation products of the hem-2 locus genes exhibit a high degree of amino acid sequence similarity to those encoded by the corresponding ORFs of hem-1 (HemP 52%, HemR 62%, HemS 53%, HemT 60% identity, respectively) (**Figure S2**, **Figure 2B**). The nomenclature for the hem-2 genes at locus 2 follows that of the highly homologous hem genes at locus 1, and the putative proteins encoded are named HemP2, HemR2, HemS2, and HemT2, respectively. Interestingly, the lack of hemUV2 genes in this cluster suggests that the putative hem-2 heme permease is non-functional. This does not exclude the possibility that HemR2 might function as a receptor protein involved in heme uptake by utilizing components of the ABC transport system encoded by the hem-1 locus.

Alignment of HemR2 with the well-characterized HemR1 showed high amino acid sequence identity. These homologous proteins have typical TonB boxes at their amino-termini, conserved histidine residues, plus region V, which is characteristic of TonB-dependent proteins. These conserved motifs and residues were shown to be essential for heme transport through the receptor channel (Thompson et al., 1999) (**Figure 2B**).

#### Genes Within the *hemPRSTUV*-1 and the *hemPRST-*2 Clusters Are Organized as Operons

There are six and four genes within the hemPRSTUV-1 (hem-1) and the hemPRST-2 (hem-2) clusters of Y. enterocolitica Ye9, respectively (**Figure 2A**). To determine whether the genes within these two loci might be transcribed as one transcriptional unit, sqRT-PCR analysis was performed on total RNA isolated from Y. enterocolitica Ye9 grown in iron-depleted medium (LBD). The locations of the primers used in this analysis are shown in **Figures 3A,C** and their sequences are given in **Table S1**. For the hemPRSTUV-1 cluster, the obtained RT-PCR products (815 and 739 bp) confirmed co-transcription of the hemPR-1 and hemRS-1 genes, respectively. The operon organization of hemPRSTUV-1 was conclusively demonstrated by the amplification of a 5,612-bp RT-PCR product (**Figure 3B**).

The operon organization of the hemPRST-2 cluster was also examined by sqRT-PCR analysis using the primers (**Table S1**) shown in **Figure 3C**. RT-PCR products of the expected sizes (528 and 804 bp) confirmed the co-transcription of the hemPR-2 and hemRS-2 genes, respectively. However, a 3,692-bp RT-PCR product spanning the entire hemPRST-2 operon could not be obtained, despite repeated attempts. Therefore, co-transcription of the hemRST-2 genes was tested. The presence of a faint RT-PCR product of the expected size (2,082 bp) confirmed the operon organization of the hemPRST-2 cluster (**Figure 3D**). Based on these data, it seems that the hemPRST-2 transcript might be particularly unstable or expressed at low levels at 37◦C. Control PCRs with the tested primers and Ye9 genomic DNA as template gave positive results in all cases (data not shown).

proteins are boxed. Conserved amino acids are indicated by asterisks.

FIGURE 3 | sqRT-PCR analysis of the hemPRSTUV-1 and hemPRST-2 gene clusters of Y. enterocolitica Ye9. (A,C) Scheme showing primers used in sqRT-PCR analysis of hem clusters 1 and 2. The sequences of the following primers are listed in Table S2. 1, RThPR1F; 2, RThPR1R; 3, RThRS1F; 4, RThRS1R; 5, RThPV1R; 6, RThPR2F; 7, RThPR2R; 8, RThRS2F; 9, RThRS2R; 10, RThPT2R. (B,D) Agarose gel electrophoresis of sqRT-PCR products from hem clusters 1 and 2. Total RNA isolated from strain Ye9 grown in iron-depleted LBD medium (LB with the iron chelator 2,2′ -dipyridyl) at 37◦C was DNase I treated and then reverse transcribed with random hexamers. The obtained cDNA was used as a template in PCR reaction with pairs of primers shown in (A) and (C). RNA was used as the template in negative control reactions. M, GeneRuler 1 kb DNA Ladder.

#### Role of HemR1 And HemR2 Proteins in Heme and Hemoprotein Utilization by *E. coli* 1*hemA* Strain SASX77

Previous studies have shown that a plasmid containing hemPR, part of the hemin operon of Y. enterocolitica WA-C O:8, permits an E. coli strain lacking a functional hemA gene to use hemin as a porphyrin source (Stojiljkovic and Hantke, 1992, 1994). To examine the importance of the two Y. enterocolitica Ye9 HemR proteins in the utilization of hemin and hemoglobin as sources of porphyrin, they were introduced into the E. coli 1hemA strain SASX77, which is defective in ALA synthetase (auxotrophic for ALA, the heme biosynthetic precursor). The hemPR genes of the hem-1 and hem-2 operons with their native promoters were cloned into vector pACYC184 to obtain plasmids pHEM1 and pHEM2, respectively. These plasmids and the parent vector were introduced into E. coli SASX77. All transformant strains grew well in the presence of ALA, but no growth was observed on LB agar lacking this precursor.

To examine the effect of hemin or hemoglobin on the growth of strain SASX77 harboring plasmids containing hemPR1 or hemPR2, a plate assay was used to detect utilization of these porphyrin sources in iron-replete (LB) and iron-deficient conditions (LBD). The strains expressing the receptors HemR1 or HemR2 were able to utilize exogenously supplied hemin and hemoglobin as porphyrin sources (**Table 1**, **Figure 4**). However, growth of the SASX77/pHEM1 (HemR1+) strain around the hemin and hemoglobin discs was significantly stronger than that of strain SASX77/pHEM2 (HemR2+). Moreover, the HemR2 expressing bacteria grew only as single colonies. Control strain SASX77/pACYC184 was unable to use either hemin or hemoglobin (**Figure 4A**). In iron-depleted conditions, the zone of growth stimulation was broad (up to 30 mm in diameter), but the colonies were very small and needed more time to develop (up to a week, in contrast to 1–2 days for the strains growing on LB) (data not shown). These results strongly suggest that both HemRs participate in the acquisition of heme and hemoproteins in Y. enterocolitica Ye9, and that HemR1 is significantly more efficient than HemR2.

Next, the level of HemR1 receptor protein in the E. coli strain carrying pHEM1 was determined by Western blot analysis with anti-HemR1 polyclonal antibody (**Figure 4B**). Cells of Y. enterocolitica Ye9 and of the E. coli strain carrying pACYC184 were used as a positive and negative control, respectively. All strains were grown in both iron-replete (LB+FeCl3) and irondeficient (LBD) conditions. Extracts of cells grown to stationary phase at 37◦C were immunoblotted and a strong HemR1 signal was produced by the samples from the Ye9 strain and E.

TABLE 1 | Assessment of HemR1 and HemR2 activity in hemin and hemoglobin transport in E. coli 1hemA strain SASX77.


<sup>−</sup>No growth.

<sup>+</sup>∼10 mm diameter growth zone around the discs.

++20–25 mm diameter growth zone around the discs.

+++>30 mm diameter growth zone around the discs.

*<sup>a</sup>*Growth as single colonies.

coli DH5α/pHEM1 grown in LBD medium. As expected, no immunoreactive band was observed in the case of E. coli/pHEM2. This result indicated the absence of cross-reactivity between anti-HemR1 antibodies and the protein HemR2. Since no HemR2 antibody was available it was not possible to confirm expression of this protein in the E. coli strain carrying plasmid pHEM2. This experiment also demonstrated the induction of HemR1 protein in Y. enterocolitica grown under iron starvation conditions, confirmed that this gene can be expressed in a heterologous system from a recombinant plasmid, and showed that this expression is repressed by iron.

#### OmpR Inhibits the Activity of the *hemPRSTUV*-1 and *hemPRST*-2 Promoters

The putative promoters phem−<sup>1</sup> and phem−<sup>2</sup> , located 42 and 43 bp upstream of the hemP gene in the hemPRSTUV-1 and hemPRST-2 operons, respectively, of Y. enterocolitica Ye9, were identified using the program BPROM (Solovyev and Salamov, 2011). Sequence analysis of the hemPRSTUV-1 and hemPRST-2 regions also led to the identification of an internal, putative promoter in the intergenic region between hemR and hemS. These findings corroborate the previously described locations of hemPRSTUV operon promoters in Y. enterocolitica WA-C (Stojiljkovic and Hantke, 1992), Y. pestis (hmu locus, Thompson et al., 1999), and Y. pseudotuberculosis (hmu locus, Schwiesow et al., 2018). In silico analysis revealed the presence of potential

were placed on these plates and zones of growth around these discs were analyzed after 1–4 days of incubation at 37◦C. (B) Immunodetection of HemR1 in Y. enterocolitica Ye9N (wt) and E. coli SASX77 carrying plasmids pHEM1 (HemR+), pHEM2 (HemR2+), or pACYC184. Cell extracts from overnight cultures were analyzed by Western blotting with a polyclonal antibody directed against HemR1 (upper) and by TGX Stain free (lower, loading control). MW, 3-Color Prestained Protein Marker; kDa. This result is representative of at least three independent experiments.

Jaworska et al. OmpR Inhibits Heme Receptor Expression

OmpR binding sites within the phem−<sup>1</sup> and phem−<sup>2</sup> promoters (see below), but not within the putative internal promoters, so their activity was not investigated further.

To examine the activity of the phem−<sup>1</sup> and phem−<sup>2</sup> promoters and the influence of iron and the regulator OmpR, we generated transcriptional fusions of the putative promoter sequences with the lacZ gene in vector pCM132Gm. The resulting plasmids pCM1 (phem−<sup>1</sup> ::lacZ) and pCM2 (phem−<sup>2</sup> ::lacZ) were introduced into E.coli BW25113 and two strains of Y. enterocolitica: the wildtype strain Ye9N (both fusions) and the isogenic ompR mutant AR4 (only fusion phem−<sup>1</sup> ::lacZ). The transformed strains were grown to stationary phase in LB or LBD and β-galactosidase activity was measured (**Figure 5**).

In the E. coli strains grown at 37◦C in both iron-replete or iron-depleted conditions the expression of the phem−<sup>1</sup> ::lacZ transcriptional fusion (pCM1) was significantly higher than that of phem−<sup>2</sup> ::lacZ (pCM2), indicating that phem−<sup>2</sup> is a weaker promoter. In addition, the expression of both phem−<sup>1</sup> ::lacZ and phem−<sup>2</sup> ::lacZ was upregulated in the presence of the iron chelator (∼3-fold and ∼2.5-fold, respectively), demonstrating inhibition by inorganic iron (**Figure 5A**).

To investigate whether the hem-1 and hem-2 operons are subject to OmpR-dependent regulation, phem−<sup>1</sup> ::lacZ and phem−<sup>2</sup> ::lacZ activity was measured in the Y. enterocolitica strains differing in the amount of OmpR grown at 26 or 37◦C in iron-replete or iron-depleted conditions. With the phem−<sup>1</sup> ::lacZ fusion, a ∼2-fold (26◦C) and ∼1.5-fold (37◦C) upregulation was observed under conditions of iron deficiency, confirming iron-dependent regulation of phem−<sup>1</sup> . Moreover, βgalactosidase activity was increased in the absence of OmpR at both studied temperatures in iron-replete or iron-depleted conditions (**Figure 5B**). The effect of complementation of the ompR mutation in strain AR4 or overexpression of OmpR in the wild-type strain were studied by introducing plasmid pBOmpR carrying the wild-type copy of ompR. An increase in the level of OmpR caused a clear reduction in the expression of phem−<sup>1</sup> ::lacZ in cells grown in LB with FeCl<sup>3</sup> or LBD, showing that OmpR negatively regulates expression of this fusion. These results indicated that hem-1 expression is controlled by an ironand OmpR-dependent promoter located upstream of this operon. Both iron and OmpR participate in the negative regulation of hem-1.

To test the role of iron and OmpR in the regulation of phem−<sup>2</sup> ::lacZ, the wild-type strain Ye9N transformed with plasmid pCM2 was grown at 26 or 37◦C in iron-replete or iron-depleted conditions. A ∼1.5-fold increase in β-galactosidase activity was detected in wild-type cells grown at 26◦C under irondeficiency (**Figure 5C**). However, the inhibitory effect of iron on phem−<sup>2</sup> ::lacZ expression was not observed at 37◦C, suggesting a more complex regulatory mechanism compared to the hem-1 promoter. Attempts to study the activity of the phem−<sup>2</sup> ::lacZ reporter fusion in the ompR mutant background were thwarted by the inability to transform strain AR4 with plasmid pCM2. Therefore, the wild-type ompR allele was introduced in trans (pBOmpR) into the Ye9N cells to monitor the effect of overexpression. The observed decrease in phem−<sup>2</sup> activity in the presence of ompR in multicopy in the wild-type strain grown

Jaworska et al. OmpR Inhibits Heme Receptor Expression

at 26 or 37◦C, suggested a negative role for OmpR in hem-2 expression. Interestingly, the inhibitory effect of iron was absent in the wild-type strain containing plasmid pBOmpR grown at 26◦C (**Figure 5C**).

### Impact of Fur and OmpR on HemR1 Receptor Production

The analysis of strains carrying the phem−<sup>1</sup> ::lacZ reporter fusion construct suggested that iron and OmpR negatively influence the expression of the hem-1 operon at the transcriptional level. We next examined whether this regulatory effect was observable at the protein level. Western blot analysis with a polyclonal anti-HemR1 antibody was used to detect the HemR1 protein in cell extracts prepared from the following Y. enterocolitica strains: wild-type Ye9N, ompR mutant AR4, Ye9N/pBOmpR, AR4/pBOmpR, and strains with empty vector pBBR1. All strains were grown to stationary phase at 37◦C in LB supplemented with FeCl<sup>3</sup> (iron-replete) or LBD (iron-depleted). As shown in **Figures 6A,B**, HemR1 protein production was significantly influenced by iron content, because HemR1 was undetectable in strain Ye9N grown under iron-replete conditions, and highly abundant in iron-depleted conditions. Moreover, HemR1 production was notably increased in the ompR mutant compared to the wild-type strain in both media. Plasmid pBOmpR carrying the wild-type ompR allele, used to complement the ompR mutation in strain AR4, caused a clear reduction in the level of HemR1, confirming that OmpR negatively regulates its expression. In addition, the strain Ye9N/pBOmpR, which overexpresses OmpR, produced less HemR1 than strain Ye9N in LBD medium (**Figure 6B**). The presence of the empty vector pBBR1 in the wild-type strain and the ompR mutant, grown in LBD medium (**Figure 6B**) did not have any impact on the level of HemR1. However, for some unknown reason, a difference in the abundance of HemR1 was observed in the ompR mutant vs. the ompR strain carrying the empty vector pBBR1, when grown in iron-replete conditions (**Figure 6A**).

These data corroborated the results of the transcriptional fusion analysis and suggested that the Fur may be involved in iron-dependent inhibition of hem-1 operon expression, and thus HemR1 production.

In order to determine whether OmpR controls the level of HemR1 expression independently of the Fur repressor,

FIGURE 6 | Effect of iron, Fur, and OmpR on HemR1 production. The level of HemR1 in cell extracts of strains grown overnight in LB+FeCl<sup>3</sup> or LBD at 37◦C analyzed by immunoblotting with a polyclonal HemR1 antibody. (A,B) The HemR1 level in the wild-type strain (wt), wt/pBBR1 (empty vector), wt/pBOmR (overexpression), the ompR mutant AR4, ompR/pBOmpR (complementation), and ompR/pBBR1 (empty vector) shown on Western blots (upper) with the stained gels as a loading control (bottom). (C) The HemR1 level in the fur mutant background, i.e., fur mutant, furompR double mutant, and furompR/pBOmpR. The percentage values indicating the HemR1 band intensities in the tested strains relative to the wt (B) and fur mutant (C), were (Continued) FIGURE 6 | determined using Amersham Imager 600 Analysis Software V1.0.0 (GE Healthcare). Stained gels shown as a loading control: (A,C) TGX Stain-Free gels (BioRad), (B) Coomassie blue stained gel. MW, 3-Color Prestained Protein Marker, kDa. The size of the HemR1 band is approximately 75 kDa on the Western blots. This result is representative of at least three independent experiments.

the fur gene was deleted in both the wild-type strain Ye9N (Ye9fur) and the ompR mutant (AR4fur). HemR1 levels were then examined by immunoblotting of cell extracts prepared from the fur, furompR and furompR/pBOmpR strains grown to stationary phase in LB supplemented with FeCl<sup>3</sup> or LBD at 37◦C (**Figure 6C**). The Western blot data showed a high level of HemR1 production in the fur mutant background (independently of iron concentration), confirming that the Fur represses the biosynthesis of HemR1 in Y. enterocolitica. Interestingly, the strain also lacking OmpR (furompR mutant), grown in LBD medium, displayed a further increase in HemR1 protein abundance (ca. 70%), which suggested that OmpR inhibits the production of HemR1 independently of Fur. This effect was restored by complementation with plasmid pBOmpR carrying the wild-type ompR allele. Under iron-replete conditions the difference in HemR1 level between the fur and furompR mutants was less evident. These data suggested that iron may influence OmpR-dependent HemR1 synthesis irrespective of Fur.

#### Interactions of OmpR With Putative OmpR-Binding Sequences in the Promoter Regions of *hem*-1 and *hem*-2

We next tested whether OmpR influences hem-1 and hem-2 operon expression directly or if other regulatory factors are involved. In silico analysis of the Y. enterocolitica Ye9 hem-1 promoter region phem−<sup>1</sup> , conducted using the consensus Fur binding site sequence of E. coli [FBS, 5′ - GATAATGAT(A/T)ATCATTATC-3′ ] (Stojiljkovic et al., 1994; Escolar et al., 1999), led to the identification of the element FBS-1, with 74% homology to the E. coli FBS, that overlaps the −10 element of phem−<sup>1</sup> (**Figure 7A**). In addition, a putative OmpR binding site OBS-1, overlapping the −35 element of phem−<sup>1</sup> was identified by comparison with the consensus OmpR-binding sequence of E. coli (OBS, 5′ -TTTTACTTTTTG(A/T)AACATAT-3 ′ , Maeda et al., 1991). OBS-1 exhibits 60% identity to the E. coli OBS (**Figure 7A**).

In silico analysis of the hem-2 promoter region phem−<sup>2</sup> revealed a putative Fur binding site (FBS-2), as well as a potential OmpR binding site (OBS-2) in similar locations to FBS-1 and OBS-1. FBS-2 and OBS-2 are 74 and 45% identical to the E. coli consensus sequences, respectively (**Figure 7B**).

The ability of OmpR to directly interact with the predicted OBS elements was tested in electrophoretic mobility shift assays (EMSAs) using DNA fragments with or without the putative OmpR-binding sites (**Figures 7C–E**). Three phem−<sup>1</sup> fragments were PCR-amplified with the pairs of primers listed in **Table S2**. Fragment F1 (415 bp) represented a sequence upstream of OBS-1, F2 (326 bp) carried OBS-1, while F3 (107 bp) was a sequence downstream of OBS-1. When purified recombinant OmpR was mixed with equimolar amounts of these three phem−<sup>1</sup> fragments, shifted OmpR/DNA complexes started to appear at an OmpR concentration of 0.65µM, and their amount increased as the concentration of the protein was raised (**Figure 7D)**. The formation of these complexes was related to the disappearance of fragment F2, so were attributed to specific binding to OBS-1.

The interaction of OmpR with the putative OBS-2 within the promoter region of hem-2 was examined by incubating increasing amounts of recombinant OmpR with fragment F4 (303 bp), the region of phem−<sup>2</sup> containing OBS-2, or phem−<sup>1</sup> fragment F2 (326 bp), carrying OmpR binding site OBS-1. Specific OmpR/F4 complexes were observed at an OmpR concentration of 1.31µM, with the simultaneous disappearance of fragment F4. In comparison, the appearance of OmpR/F2 complexes occurred at the lower OmpR concentration of 0.65µM (**Figure 7E**). A 16S rDNA fragment (amplified with primer pair 16SF/16SR, **Table S2**), to which OmpR is unable to bind, was used as a negative control in both experiments. These results suggested that transcription of the hem-1 and hem-2 gene clusters is directly regulated by OmpR, and this regulation requires OmpR binding to specific sites in their promoter regions.

#### The Impact of Temperature on Post-transcriptional HemR1 Expression

Y. enterocolitica HemR1 is a homolog (69% amino acid identity) of the S. dysenteriae ShuA receptor required for the utilization of heme as a source of iron (data not shown). Recently, thermoregulation of ShuA protein expression under iron-limited conditions was demonstrated, with low ShuA abundance at 25◦C increased to high levels at 37◦C (Kouse et al., 2013). Within the shuA 5 ′ -UTR a regulatory element was recognized that closely resembles a FourU RNA thermometer: a zipperlike RNA structure that occludes the Shine-Dalgarno sequence at low temperatures, thus blocking translation of the shuA transcript.

Both HemR1 and ShuA are encoded within hemin transport loci, but the organization of these multigenic clusters differs considerably (Wyckoff et al., 1998). The hemR1 gene of Y. enterocolitica is located 172 bp downstream of hemP1, the first gene of the hemPRSTUV-1 operon. There is no hemP1 homolog in the hemin transport locus in S. dysenteriae. Moreover, the intergenic non-translated hemP1-hemR1 region (172 nt) is present in polycistronic mRNA of hem-1. In comparison, the predicted promoter of shuA is located 328 bp upstream of the translation start site in a monocistronic transcript.

To investigate the possible thermoregulation of hemR1 expression we analyzed an extended 5′ -UTR of Y. enterocolitica hemR1 for the presence of secondary structures using Mfold software (Zuker, 2003). This analysis identified a hairpin sequestering the hemR1 ribosomal binding site, containing four consecutive uracil residues characteristic of a FourU RNA thermometer (Waldminghaus et al., 2007; Böhme et al., 2012). However, compared to the FourU base-pair region in the 5′ - UTR of shuA there was moderate sequence variation in the

FIGURE 7 | OmpR binding to fragments of the hem-1 and hem-2 promoter regions. (A,B) OmpR- and Fur-binding sites identified in promoter regions of the two heme uptake loci. The −35 and −10 promoter elements are underlined and putative regulator-binding sites are boxed. The ATG start codon of the first ORF of each operon is shown in bold. FBS, Fur-binding site; OBS, OmpR-binding site. (C) Schematic representation of fragments used in EMSA experiments presented in (D) and (E). The (Continued)

FIGURE 7 | numbered arrows represent the primers used to amplify the DNA fragments used in EMSAs: 1, hem1F; 2, hem1R, 3, hem1-aF; 4, hem1-aR; 5, hem1-bF; 6, hem2F; 7, hem2R. (D) EMSA with fragment F2 (326 bp) containing OBS-1 and neighboring fragments F1 (415 bp, upstream) and F3 (107 bp, downstream). A 16S rDNA fragment (211 bp) was used as a negative control. The reaction mixtures contained 0.05 pmol of each fragment and increasing amounts of purified OmpR-His6 were added: lane 1, no protein; lanes 2–5, 0.65, 1.31, 2.61, and 3.92µM of OmpR-His6, respectively. (E) Binding of purified OmpR-His6 to fragments containing OBS-1 (F2, lanes 1–4) and OBS-2 (F4, lanes 5–8). The 16S rDNA fragment was used as a negative control. The reaction mixtures contained 0.05 pmol of each fragment and increasing amounts of purified OmpR-His6 were added: lanes 1 and 5, no protein; lanes 2 and 6, 0.65µM; lanes 3 and 7, 1.31µM; lanes 4 and 8, 3.92µM.

surrounding regions (**Figure 8A**). No similar secondary structure was detected in the 5′ -UTR of hemR2 (data not shown).

The impact of temperature on HemR1 expression was assessed by immunoblotting of cell extracts prepared from the fur mutant following growth in LB medium to stationary phase at 26 or 37◦C (**Figure 8B**). Identical levels of HemR1 were detected in the tested strain, irrespective of the growth temperature, which indicated that HemR1 production is not thermoregulated in Y. enterocolitica Ye9.

To further examine the potential thermoregulation of HemR1 expression at the post-transcriptional level, a plasmid pFX-Plac-hemR1, which carries the plac promoter linked to the 5′ - UTR and first 16 codons of hemR1 fused in frame with gfp, was constructed. Plasmid pFX-Plac-hemR1 and control plasmids pFX-0, containing promoterless gfp, and pFX-1, carrying gfp fused to the plac promoter (Schmidtke et al., 2013), were introduced into E. coli BW25113. Cultures of these strains were grown to stationary phase in LB medium at 26 or 37◦C and green fluorescence intensity was measured (**Figure 8C**). Analysis of the strain carrying the hemR1′ - ′ gfp fusion revealed a decrease in hemR1 expression at 26◦C relative to 37◦C, however a slight reduction in fluorescence at this lower temperature was also observed for the control strain carrying pFX-1. In parallel, the impact of growth temperature on the GFP level was assessed by immunoblotting with an anti-GFP antibody (**Figure 8D**). The Western blot data corresponded well with the results of the fluorescence assay. A higher level of HemR1′ - ′GFP fusion protein was produced at 37◦C compared to 26◦C, which may indicate that temperature-dependent post-transcriptional regulation of hem-1 occur in E. coli. However, we could not exclude that observed effect may be explain also by the increased metabolic activity at the higher temperature.

### DISCUSSION

The acquisition of iron/heme is necessary for pathogenic bacteria to grow within the host and to survive during the progression of an infection (Mietzner and Morse, 1994; Lee, 1995; Braun et al., 1998). On the other hand, both iron and heme are toxic at high concentrations (Anzaldi and Skaar, 2010). Thus, tight regulation of iron/heme acquisition is essential to fulfill the iron requirements of a pathogen while preventing harmful overaccumulation.

Studies on the role of OmpR in pathogenic Yersinia strains have provided evidence that this response regulator might be involved in the adaptation of yersiniae to diverse environmental conditions, which allows the bacteria to grow in distinct niches within and outside the host body (Brzostek et al., 2003, 2007; Raczkowska et al., 2011a,b, 2015; Brzóstkowska et al., 2012; Skorek et al., 2013). In a recent shotgun proteomic study we revealed that the regulator OmpR affects, both positively and negatively, the production of over 100 membrane proteins in Y. enterocolitica strain Ye9 of bio-serotype 2/O:9, a low pathogenicity strain that is not lethal for mice at low doses (Nieckarz et al., 2016). The loss of OmpR positively affected the production of heme receptor HemR1 in Y. enterocolitica cells grown in LB medium at 37◦C and preliminary data suggested indirect repression of hemR1 by OmpR.

In this study, we investigated the molecular mechanisms underlying OmpR-dependent production of two potential heme receptor proteins encoded within hemPRSTUV-1 (hem-1 locus) and hemPRST-2 (hem-2 locus) gene clusters identified in Y. enterocolitica strain Ye9. Firstly, we demonstrated that strain Ye9 utilizes heme-containing molecules as an iron source. The Y. enterocolitica hemin uptake system involved in transporting the entire heme moiety into the cytoplasm was described previously in strain WA-C, a mouse-virulent strain of bio-serotype 1B/O:8 (Stojiljkovic and Hantke, 1992, 1994). In contrast to strain Ye9, this strain is able to produce and utilize the siderophore yersiniabactin (Ybt), encoded on a chromosomal HPI (36 kb) (Heesemann et al., 1993; Rakin et al., 1994, 2012; Pelludat et al., 1998).

Heme uptake systems have been identified in several pathogenic bacteria, including Y. pestis (Thompson et al., 1999), Y. pseudotuberculosis (Schwiesow et al., 2018), S. dysenteriae (Mills and Payne, 1995, 1997; Wyckoff et al., 1998), and V. cholerae (Henderson and Payne, 1994; Occhino et al., 1998). Partial redundancy of heme-acquisition systems is often observed in Gram-negative bacteria due to the presence of uptake systems with a specific outer membrane receptor for heme and hemeproteins and/or the production of secreted hemophores (Ochsner et al., 2000; Runyen-Janecky, 2013).

In this study, two multigenic clusters, hemPRSTUV-1 and hemPRST-2 were identified in the genome of Ye9. The genetic organization of these clusters is identical except that the hem-2 locus lacks two ORFs, hemU2 and hemV2, encoding potential components of the heme transport system. Evidence from sqRT-PCR analysis revealed that both the hem-1 and hem-2 clusters are organized as operons, and expressed from the promoters phem−<sup>1</sup> and phem−<sup>2</sup> located upstream of the respective hemP genes. Experiments with a Y. enterocolitica strain carrying a plasmid-located phem−<sup>1</sup> ::lacZ reporter fusion showed that phem−<sup>1</sup> promoter activity is inhibited by iron, which suggested the involvement of Fe2+-Fur in this repression. It has been previously demonstrated in diverse microorganisms that Fur, activated upon binding ferrous iron, binds to specific genes to inhibit

FIGURE 8 | Influence of temperature on hemR1 expression. (A) Potential FourU thermometer RNA secondary structures in the 5′ -UTRs of Y. enterocolitica Ye9 hemR1 and S. dysenteriae shuA, composed of a stretch of FourU and the ribosomal binding site (RBS), revealed by Mfold analysis (http://mfold.rna.albany.edu). Boxes indicate the location of the FourU motif (shaded) and putative RBS, the translation start codon is marked red, and unpaired nucleotides within the predicted secondary structure are in green. (B) Temperature-dependent HemR1 expression in Y. enterocolitica examined by immunoblotting. The analyzed samples were cell lysates of the fur mutant grown at 26 or 37◦C in LB medium. The top panel shows the Western blot probed with a polyclonal antibody against HemR1, the bottom panel shows the Coomassie blue-stained gel as a loading control. The percentage values on the blot, indicating the HemR1 band intensities relative to that of cells grown at 26◦C, were determined using Amersham Imager 600 Analysis Software V1.0.0 (GE Healthcare). (C) Expression of a hemR1′ - ′gfp translational fusion at different temperatures, monitored by fluorescence intensity. E. coli BW25113 harboring the reporter plasmid pFX-Plac-hemR1 or control plasmids pFX-0 and pFX-1 were grown in LB medium at 26 or 37◦C. The GFP fluorescence intensity (RFU) of overnight cultures was determined. The data represent the averages ± SD from at least three experiments with duplicate cultures. Significance was calculated using one-way ANOVA [P > 0.05, \*P < 0.05, \*\*\*P < 0.001]. (D) GFP abundance examined by immunoblotting. The analyzed samples were cell lysates of E. coli BW25113 harboring the plasmids pFX-0, pFX-1 or pFX-Plac-hemR1. The top panel shows the immunoblot probed with antibody against GFP, the bottom panel shows the TGX Stain-Free gel as a loading control. MW, 3-Color Prestained Protein Marker; kDa. These results are representative of at least three independent experiments.

their transcription (Bagg and Neilands, 1987; de Lorenzo et al., 1987; Andrews et al., 2003). Based on experimental and/or bioinformatic analyses, almost all iron-acquisition systems of pathogenic yersiniae, i.e., Y. pestis, have been shown to be Furcontrolled (Gao et al., 2008). More importantly, the results we obtained using transcriptional fusions with the lacZ gene and by Western blot analysis demonstrated that OmpR negatively regulates phem−<sup>1</sup> activity and thus HemR1 protein expression under conditions of iron starvation. The precise role of OmpR in the repression of HemR1 expression was revealed in a fur mutant background. While the ompR and fur mutations alone led to an increase in HemR1 synthesis, the combination of the two produced an additive effect. These data showed that OmpR inhibits HemR1 production in the absence of Fur. However, the OmpR-dependent regulation of HemR1 expression seems to be much more complex, since the effect of OmpR was also observed at the transcriptional and HemR1 protein level in iron-replete conditions, when the Fur repressor is active.

EMSA analysis revealed that OmpR acts directly at the phem−<sup>1</sup> promoter. We observed that OmpR specifically recognizes and binds a phem−<sup>1</sup> promoter fragment containing the in silicopredicted OmpR-binding sequence. This result confirmed a direct role for OmpR in mediating hemR1 repression. In the hem-1 promoter, the predicted 19-bp Fur box (FSB-1) overlaps the −10 promoter motif, and is only 10 bp downstream of the 20-bp putative OmpR-binding site (OBS-1) at the −35 motif. Based on the location of these binding sites, we propose the following model for Fur- and OmpR-dependent repression of hem-1. Consistent with the notion that under iron-starved conditions Fur-dependent repression is abolished, OmpR may interact with OBS-1 and prevent the polymerase binding to the promoter, leading to inhibition of hem-1 expression. In iron-replete medium, there are a number of possible ways in which OmpR could regulate hem-1 expression. It is likely that binding of OmpR to the phem−<sup>1</sup> promoter leads to a DNA structure that is favorable for the function of Fe2+-Fur. Recent reports have suggested that the OmpR of Salmonella enterica may influence DNA topology to control target promoter activity (Cameron and Dorman, 2012; Quinn et al., 2014). We do not rule out the possibility that OmpR-Fur protein interactions could stimulate the binding of Fur to the hem-1 promoter, which would lead to inhibition of transcription initiation. Further research is required to characterize the molecular mechanisms modulating the activity of Fur at the hem-1 promoter in the presence of OmpR.

Similarly to the hem-1 locus, the hem-2 gene cluster hemPRST-2 is organized as an operon. However, phem−<sup>2</sup> promoter activity was much lower than phem−<sup>1</sup> and iron repression was detected only at 26◦C. Although a sequence displaying 74% identity to the Fur box consensus was recognized in phem−<sup>2</sup> (at the −10 motif), the potential regulatory role of Fe2+-Fur might be dependent on other thermoregulated factors. The role of OmpR in the activity of phemR2 was studied by analyzing the expression of a phem−<sup>2</sup> ::lacZ fusion in the wild-type strain with a normal or raised OmpR content. This experiment revealed that OmpR negatively regulates hem-2 expression. Interestingly, in the presence of an increased level of OmpR, iron regulation was abolished. Moreover, in EMSAs, we were able to detect the specific interaction of OmpR with the putative OmpR binding site in phem−<sup>2</sup> , but with lower affinity than with OBS-1 in the hem-1 promoter. The higher binding sequence degeneracy in phem−<sup>2</sup> (45% identity to the consensus sequence for OBS-2 vs. 60% identity for OBS-1) might explain the weaker interaction of OmpR with this promoter.

Taken together, these results demonstrated the importance of the OmpR regulator in hem-1 and hem-2 expression, and hence the production of the HemR1 and HemR2 heme receptors in Y. enterocolitica. Our findings suggest that the effects of OmpR on hem-1 and hem-2 transcription are likely to be direct, i.e., produced by binding to specific DNA sequence elements in the promoter regions. The negative effect of OmpR on hem-1 transcription occurs in both the absence and presence of the Fur repressor. The role of OmpR in regulating hem-2 expression is difficult to assess and needs further study.

Thermo-induced structural changes in mRNA play a fundamental role in temperature sensing in bacteria and influence virulence gene expression in many pathogens, including those of the genus Yersinia (Narberhaus et al., 2006; Narberhaus, 2010; Böhme et al., 2012). Analysis of the intergenic region of the hemP1-hemR1 transcript in locus hem-1, but not in locus hem-2, in Y. enterocolitica Ye9, suggested the formation of a secondary structure in the hemR1 5 ′ -UTR with a FourU element that may sequester the ribosome binding site. Such an RNA structure was previously shown to be sufficiently stable at moderate temperatures (25◦C) to inhibit expression of the ShuA receptor in S. dysenteriae. Melting of the FourU motif at 37◦C permits access of ribosomes and initiates shuA translation (Kouse et al., 2013). However, Western blot analysis revealed that HemR1 of Y. enterocolitica is not subject to thermoregulation. Despite the presence of a FourU sequence in the 5′ -UTR of hemR1, the unpaired G-G sequence within the hairpin may destabilize this secondary structure. In S. dysenteriae, only a single nucleotide replacement within the shuA hairpin led to destabilization of this inhibitory structure and resulted in increased expression of this gene at the non-permissive temperature of 25◦C (Kouse et al., 2013). A stretch of four uracils located within an intergenic region of the yscW-lcrF transcript in a pathogenic Yersinia has been linked with the thermally regulated expression of the transcript encoding LcrF, the transcriptional activator of yop and ysc genes of the type III secretion system (Böhme et al., 2012). It is notable that analyses of the expression of a hemR1′ - ′ gfp translational fusion and the level of a HemR1′ - ′GFP fusion protein in E. coli showed a degree of temperature dependence, suggesting that a mechanism of thermoregulation, unconnected with the presence of a FourU RNA thermometer, is active in this heterologous genetic background.

HemR1 and the newly identified HemR2 protein exhibit a high degree of amino acid sequence similarity (62% identity), including the presence of signature motifs like the TonB box, the conserved V region associated with all TonB-dependent OM proteins, and histidine residues required for heme transport via the receptor channel (Kadner, 1990; Bracken et al., 1999).

Significant similarities between HemR1/HemR2 and heminbinding proteins suggest that these two Y. enterocolitica proteins are involved in the binding and utilization of hemin and hemeproteins. The expression of the HemR1 or HemR2 protein from their native promoters in E. coli SASX77, a 1hemA mutant defective in the biosynthesis of a heme precursor and naturally lacking a heme binding protein in the outer membrane (Stojiljkovic and Hantke, 1994; Mills and Payne, 1995), permitted hemin and hemoglobin utilization. The strain expressing HemR2 was less efficient in utilizing hemin or hemoglobin, which might reflect lower activity of the phem−<sup>2</sup> promoter. It is also possible that structural or functional differences in these OM receptors may influence the transport of the heme moiety.

Taken together the findings of this study indicate that Y. enterocolitica HemR1 and HemR2 are outer membranereceptors that play an important role in hemin and hemoglobin utilization. We hypothesize that HemR1 and HemR2 together with HemTUV-1, a periplasmic/inner membrane ABC heme transporter also encoded by the hem-1 locus, may constitute a system involved in the acquisition of the heme moiety from host hemoproteins under the varied conditions encountered by Y. enterocolitica during an infection. It is also possible that HemR2 together with an as yet unidentified cytoplasmic transporter constitute an alternative system for the utilization of heme and host hemoproteins. Such a system might be able to counteract the loss/inactivation of the Y. enterocolitica heme uptake system based on the HemR receptor. Pathogenic bacteria frequently possess more than one transport system for iron or heme/hemoprotein uptake (Braun et al., 1998), so additional

#### REFERENCES


systems besides those involving the TonB-dependent HemR1 and HemR2 receptors may help perform this vital function in Y. enterocolitica.

In conclusion, this study has demonstrated that the regulators Fur and OmpR participate in a complex mechanism governing the negative regulation of Y. enterocolitica hemR1 and hemR2. This interplay might be responsible for fine-tuning the expression of HemR receptor proteins mediating iron/heme acquisition during infection, to permit rapid growth while avoiding toxicity.

#### AUTHOR CONTRIBUTIONS

KJ, AR, and KB designed the study. KJ, MN, and ML performed the experiments. KB, AR, KJ, and ML analyzed the data. AR, KB, and KJ wrote the manuscript. KJ and MN designed and prepared the figures. KB provided financial support.

#### FUNDING

This work was supported by a grant from the National Science Center, Poland (OPUS grant UMO-2016/21/B/N26/011009). Publication co-financed by the University of Warsaw.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00333/full#supplementary-material


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Jaworska, Nieckarz, Ludwiczak, Raczkowska and Brzostek. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Yersinia pseudotuberculosis BarA-UvrY Two-Component Regulatory System Represses Biofilms via CsrB

Jeffrey K. Schachterle<sup>1</sup> , Ryan M. Stewart <sup>1</sup> , M. Brett Schachterle<sup>1</sup> , Joshua T. Calder <sup>1</sup> , Huan Kang<sup>2</sup> , John T. Prince<sup>2</sup> and David L. Erickson<sup>1</sup> \*

#### Edited by:

Matthew S. Francis, Umeå University, Sweden

#### Reviewed by:

Yi-Cheng Sun, Institute of Pathogen Biology (CAMS), China Gregory Anderson, Indiana University, Purdue University Indianapolis, United States Nadim Majdalani, National Institutes of Health (NIH), United States

> \*Correspondence: David L. Erickson david\_erickson@byu.edu

#### Specialty section:

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 14 June 2018 Accepted: 24 August 2018 Published: 18 September 2018

#### Citation:

Schachterle JK, Stewart RM, Schachterle MB, Calder JT, Kang H, Prince JT and Erickson DL (2018) Yersinia pseudotuberculosis BarA-UvrY Two-Component Regulatory System Represses Biofilms via CsrB. Front. Cell. Infect. Microbiol. 8:323. doi: 10.3389/fcimb.2018.00323 <sup>1</sup> Department of Microbiology and Molecular Biology, Brigham Young University, Provo, UT, United States, <sup>2</sup> Department of Chemistry and Biochemistry, Brigham Young University, Provo, UT, United States

The formation of biofilms by Yersinia pseudotuberculosis (Yptb) and Y. pestis requires the hmsHFRS genes, which direct production of a polysaccharide extracellular matrix (Hms-ECM). Despite possessing identical hmsHFRS sequences, Yptb produces much less Hms-ECM than Y. pestis. The regulatory influences that control Yptb Hms-ECM production and biofilm formation are not fully understood. In this study, negative regulators of biofilm production in Yptb were identified. Inactivation of the BarA/UvrY two-component system or the CsrB regulatory RNA increased binding of Congo Red dye, which correlates with extracellular polysaccharide production. These mutants also produced biofilms that were substantially more cohesive than the wild type strain. Disruption of uvrY was not sufficient for Yptb to cause proventricular blockage during infection of Xenopsylla cheopis fleas. However, this strain was less acutely toxic toward fleas than wild type Yptb. Flow cytometry measurements of lectin binding indicated that Yptb BarA/UvrY/CsrB mutants may produce higher levels of other carbohydrates in addition to poly-GlcNAc Hms-ECM. In an effort to characterize the relevant downstream targets of the BarA/UvrY system, we conducted a proteomic analysis to identify proteins with lower abundance in the csrB::Tn5 mutant strain. Urease subunit proteins were less abundant and urease enzymatic activity was lower, which likely reduced toxicity toward fleas. Loss of CsrB impacted expression of several potential regulatory proteins that may influence biofilms, including the RcsB regulator. Overexpression of CsrB did not alter the Congo-red binding phenotype of an rcsB::Tn5 mutant, suggesting that the effect of CsrB on biofilms may require RcsB. These results underscore the regulatory and compositional differences between Yptb and Y. pestis biofilms. By activating CsrB expression, the Yptb BarA/UvrY two-component system has pleiotropic effects that impact biofilm production and stability.

Keywords: Yersinia pseudotuberculolisis, biofilm, carbon storage regulator system A (CsrA), two-component regulation, fleas

## INTRODUCTION

Like many bacteria, Yersinia pseudotuberculosis (Yptb) in varied environments such as soil and water face temperature extremes, desiccation, nutrient deprivation, or nematode predation, and their survival is enhanced by efficient biofilm production. Transmission of Yersinia pestis by fleas is also significantly influenced by biofilm production. One transmission mechanism exhibited by some fleas requires Y. pestis to form biofilm on the spines that line the interior surface of the flea's proventriculus (Hinnebusch et al., 1996; Jarrett et al., 2004). Biofilm formation in both Yptb and Y. pestis is aided by the hmsHFRS gene products, which together direct the synthesis of an extracellular matrix (ECM) containing poly-β-1,6 linked N-acetyl-D-glucosamine (β-1,6-GlcNAc) (Bobrov et al., 2008; Erickson et al., 2008; Hinnebusch and Erickson, 2008). HmsR and HmsS are inner-membrane proteins that may assemble polymers from UDP-N-acetylglucosamine in the cytoplasm. The polymer is likely deacetylated by HmsF prior to export and transport through the outer membrane porin HmsH. Hms-ECM production by Yptb and Y. pestis can be visualized as Congo red binding (pigmentation) on agar plates. All four of the hmsHFRS gene products are required for pigmentation and biofilm, and inactivation of the periplasmic, deacetylase, and glycosyl transferase domains of HmsH, F, and R, respectively, reduces Congo red binding and in vitro biofilm formation (Forman et al., 2006).

Recent efforts have focused on understanding how biofilm production in Y. pestis is regulated. Multiple regulatory influences of the Hms-ECM in Y. pestis have been identified including temperature (Perry et al., 2004), polyamines (Wortham et al., 2010), and the second messenger cyclic-diguanylate (cdi-GMP) (Kirillina et al., 2004; Bobrov et al., 2011). Hms-ECM is produced at growth temperatures ≤26◦C (Perry et al., 1990; Hinnebusch et al., 1996; Darby et al., 2002; Jarrett et al., 2004). Temperature regulation is achieved in part by the reduced translation or stability of HmsH, HmsR and the diguanylate cyclase HmsT (Perry et al., 2004) at temperatures above 28◦C. High levels of cyclic diguanylate in the cell may increase the glycosyl transferase activity of HmsR (Bobrov et al., 2008). Y. pestis contains an additional diguanylate cyclase (HmsD) that enhances biofilm formation in the flea digestive tract (Bobrov et al., 2011; Sun et al., 2011).

Less is known about the regulation of biofilm in Yptb. Although Yptb, like Y. pestis, forms Hms-dependent biofilms in vitro and on the outer mouthparts of Caenorhabditis elegans nematodes (Darby et al., 2002; Joshua et al., 2003), the majority of Yptb strains do not form pigmented colonies on Congo-red agar. Additionally, Yptb never forms biofilm on the flea proventriculus to cause blockage of the digestive tract, even though it can colonize the flea midgut (Erickson et al., 2006). Compositional changes in the ECM of Y. pestis and Yptb may contribute to this difference but it is now clear that during the evolution of Y. pestis, mutations in genes that affect the regulation of Hms protein synthesis or production and stability of the ECM have been selected. Functional regulatory mechanisms that repress Hms-dependent ECM production in Yptb such as the RcsA transcriptional regulator (Sun et al., 2008; Guo et al., 2015), the NghA glycosyl hydrolase (Erickson et al., 2008), and additional enzymes that produce or degrade cyclic diguanylate (Bobrov et al., 2011) have been lost or altered during the emergence of Y. pestis, which have enhanced its biofilm production within fleas.

In this study we sought to identify additional regulatory mechanisms that repress production of extracellular polysaccharides in Yptb. A screen for mutants with enhanced Congo Red pigmentation phenotypes suggested a role for the BarA/UvrY two-component regulatory system. We tested the effect of barA, uvrY and csrB mutation in Congo Red binding and biofilm stability. We sought to correlate the changes we observed in biofilm stability and Congo-red binding with production of specific carbohydrates using fluorescently-labeled lectins and flow cytometry. Finally, we employed a proteomic approach to identify multiple downstream targets of the CsrB regulatory RNA that could influence biofilm formation.

### MATERIALS AND METHODS

#### Bacterial Strains and Growth Conditions

Y. pseudotuberculosis strain IP32953 and Y. pestis KIM6+ were routinely grown at 28◦C in Terrific broth (TB) or at 21◦C in 1% heart infusion broth supplemented with 0.2% galactose (HIG). The transposon mutants in strain IP32953 were generated using the pRL27 Tn5 donor plasmid as described previously (Erickson et al., 2016). The mutants were plated onto HIG agar containing 0.01% Congo-Red dye. After growth on Congo-red agar plates for 48 h, mutants with increased pigmentation were selected and the location of their transposon insertions determined using arbitrary PCR and sequencing as previously described (Erickson et al., 2016).

To complement the transposon insertion mutants, copies of barA, uvrY, rcsA, and rcsB were amplified by PCR from IP32953 genomic DNA (all primers are listed in **Supplementary Table 1**) and inserted into pJET1.2. The CsrB expression plasmid contains the csrB sequence in plasmid pACYC184 and was generously supplied by Petra Dersch. The csrA gene was also overexpressed by cloning into pJET1.2. Plasmids were created in E. coli strain DH5α and transferred to Yptb by electroporation.

#### Biofilm Tests

Liquid biofilms were grown in 24-well plates in 1 ml of HIG broth. The plates were incubated at 21◦C for 48 h with shaking at 100 rpm. The unattached cells were removed and the wells were gently washed with 1 ml water, and then the plates were dried for 30 min at 80◦C. The attached cells were stained for 20 min with 0.1% crystal violet and rinsed 3 times with water. The bound dye was solubilized by adding 1 ml DMSO and then the absorbance was measured at 590 nm.

To measure biofilm stability, biofilms were grown on 0.2µm polycarbonate filters (25 mm) placed on HIG agar plates. After growth in liquid media overnight, 5 µl of saturated culture was added to the filters and incubated for 72 h. The filters containing the biofilms were transferred to 50 ml conical tubes containing 40 ml of phosphate-buffered saline (PBS). The tubes were placed horizontally and shaken at 100 rpm for 1 h. The density of the cells that were dislodged during this time was measured by spectrophotometry at 600 nm. The tubes were then vortexed vigorously for 5 min to suspend all of the bacteria that were present on the filters, and the density of the total population was measured. The absorbance of the biofilm cells that were initially dislodged was divided by the absorbance after vortexing to calculate the percent disruption after 1 h.

#### Lectin Binding Flow Cytometry

Bacteria were grown on HIG agar plates for 24 h. Single colonies were washed and resuspended in 300 µl of 20 mM HEPES buffer containing 0.1 mM each of MgCl2, CaCl2, MnCl2. Washed bacteria were diluted 1:10 into HEPES buffer containing 0.2 mg/ml FITC-labeled lectin (Wheat germ agglutinin or Ulex europaeus agglutinin, Sigma). Samples were incubated on ice in the dark for 30 min and then washed twice in HEPES buffer and finally resuspended in HEPES containing 1% formaldehyde. The fluorescence of individual bacterial cells was measured using a BD FACSCanto II flow cytometer. Negative control samples contained bacteria with no lectin added. The specificity of the lectin binding was assessed by adding 10 mM N,-N′ - Diacetylchitobiose or L-(-)-Fucose (Sigma) to the WGA or UEA reactions, respectively.

#### Flea Infections

X. cheopis were orally infected with Yptb through shaved mouse skins using a membrane feeder apparatus and monitored for survival and proventricular blockage as previously described (Erickson et al., 2006, 2007; Zhou et al., 2012). For each experiment, ∼200 fleas were allowed to feed on 5 ml of heparinized human blood containing 5–8 × 10<sup>8</sup> colony-forming units (CFU) Yptb wild type IP32953 or uvrY::Tn5 per ml. Feedings lasted 1 h, and the blood was kept at 37◦C during the feeding. Control fleas were fed under the same conditions, on the same blood source but without bacteria. All of the female fleas that ingested blood meals were separated and monitored for 24 h to determine acute toxicity. All of the male fleas were visually inspected for development of proventricular blockage characterized by appearance of fresh blood in the esophagus but not the midgut during twice-weekly feedings for 4 weeks. After 28 days, 20 fleas from each group were homogenized and plated on TB agar containing 1µg/ml irgasan, 0.5µg/ml crystal violet, and 1 mg/ml bile salts to determine the proportion of fleas that remained infected throughout the experiment.

### Proteome Samples

Samples for label-free proteomic analysis of wild-type and csrB::Tn5 mutant were prepared using a filter aided sample preparation (FASP) protocol as described previously (Wi´sniewski et al., 2009). Bacteria were grown for 24 h at 21◦C on solid HIG media and then washed and frozen. Frozen cell pellets were dropped into 95◦C SDT lysis buffer (4% sodium dodecyl sulfate, 100 mM Tris-HCl, 0.1M dithiothreitol) and incubated at 95◦C for 10 m. Cell suspensions were sonicated for 30 s (3 Watts of power) and incubated for an additional 10 m at 95◦C. Cell extracts were clarified, and clarified lysate was mixed with UA (8 M urea, 100 mM Tris-HCl pH 8.5) in a ratio of 1:5 and applied to Vivacon-500 30,000 MWCO filters (Sartorius, Göttingen, Germany). The filters were washed twice with 100 µL buffer UA. Filters were then incubated 20 min in the dark with 100 µL buffer UA amended with 50 mM iodoacetamide and washed twice with UA and twice with 100 µL 50 mM ammonium bicarbonate. Samples were then digested with trypsin (1 µg proteomic grade trypsin added to 40 µg of total protein in 50 mM ammonium bicarbonate). The filters were incubated at 37◦C for 16 h. Peptides were eluted with 50 mM ammonium bicarbonate and acidified to 1% formic acid.

#### Ultra-Performance Liquid Chromatography Coupled With Mass Spectrometry (UPLC-MS/MS) and Data Analysis

Tryptic peptides from the wild-type and the mutant csrB::Tn5 were analyzed with reversed phase chromatography (RPC) on an Ultra Performance Liquid Chromatography (UPLC) Eksigent NanoLC system coupled with an LTQ Orbitrap XL mass spectrometer (Thermo Scientific) equipped with an electrospray ionization source (ESI). Specifically, 4 µL of digested proteins were loaded on to a C18-trap column (Waters Corporation), and were desalted with 97% mobile phase A (0.1% formic acid in H2O, Optima) and 3% of mobile phase B (0.1% formic acid in CH3CN, Optima) at a flow rate of 4 µL/min for 10 min. A linear gradient at a flow rate of 325 nL/min changing from 3 to 35% of mobile phase B was then applied to load the peptides onto an online Waters Peptide Separation Technology C18 column in 90 min, and further sprayed into LTQ Orbitrap XL MS. The mass spectrometer was operated in the positive ion mode, and MS1 was analyzed at a resolution of 60,000 at m/z 400. MS/MS fragmentation analysis was performed on the top 10 most abundant MS1 precursor ions with collision-induced dissociation at a collision energy of 35 V. The online XCalibur software (Thermo Fisher Scientific) was used for data collection. MaxQuant software (Max Plank Institute of Biochemistry) was utilized for MS-based proteomics data analysis, e.g., protein identification and quantification. A Yptb IP32953 FASTA file downloaded from Uniprot (The UniProt, 2017) was applied for protein identification with 1% false discovery rate (FDR). A label free quantification (LFQ) approach was utilized for screening differently expressed proteins, with LFQ min. ratio count set as 2.

#### Urease Assay

Urease activity was quantified as previously described (Onal Okyay and Frigi Rodrigues, 2013) with modifications. Bacteria grown in HIG were pelleted, washed, and resuspended in PBS to an absorbance (600 nm) of 1.0. Stuart's broth (20 g/L urea, 0.1 g/L yeast extract, 0.095 g/L disodium phosphate, 0.091 g/L monopotassium phosphate, 0.01 g/L phenol red) was used as the assay medium. Bacterial suspensions (20 µl) were added to 1 ml Stuart's broth and incubated at 21◦C. Every 20 min for 4 h, the absorbance at 560 nm was measured, which corresponds to an increase in pH due to the hydrolysis of urea detected by the color change from yellow to pink of the phenol red indicator.

### RESULTS

### Yptb Mutants That Form Pigmented Colonies on Congo-Red Agar

To identify genes that repress the formation of the Hmsdependent ECM, we used random Tn5 mutagenesis of Yptb strain IP32953, which normally forms white or very light-pink colonies on Congo-red agar. From a screen of ∼15,000 colonies from four different mutagenesis experiments, we isolated 35 mutants that consistently produced colonies that were noticeably darker than the wild-type when grown at 21◦C on Congo-red agar. We identified the transposon insertion sites by sequencing arbitrary PCR products and comparing these sequences to the Yptb IP32953 genome. The locations of the insertions are listed in **Table 1**. As expected, we obtained insertion mutants in several known negative regulators of Hms expression, including the rcsA and rcsB transcriptional regulators of the Rcs phosphorelay system and the phosphodiesterase hmsP. Insertions were also found in or near genes not previously reported to affect Hms function, including transcriptional regulator nhaR, the chaperone djlA, and the DNA-binding protein dps. NhaR is a transcriptional activator of the E. coli pgaABCD genes, which are orthologs of the hmsHFRS operon (Goller et al., 2006; Cerca and Jefferson, 2008). DjlA is a member of the DnaJ family of inner membrane co-chaperone proteins, whose overexpression leads to increased production of E. coli colanic acid capsule via activation of the Rcs phosphorelay system (Clarke et al., 1997; Chen et al., 2001; Shiba et al., 2006).

We obtained several independent insertions in the genes encoding the BarA sensor kinase, the UvrY response regulator, and the CsrB small RNA that resulted in increased pigmentation on Congo-red plates. The CsrA regulatory protein is an RNA-binding protein that primarily represses translation of target mRNA molecules by binding at sites containing A/UCANGGANGU/A motifs, often at or near the Shine-Dalgarno sequence (Timmermans and Van Melderen, 2010). The pgaABCD transcript in E. coli contains three CsrA-binding sites in its 5′ untranslated region (Wang et al., 2005), and mutants that lack CsrA produce higher than normal levels of β-1,6-GlcNAc polysaccharide. CsrB and CsrC are small RNAs with high affinity for CsrA that when expressed can relieve CsrAmediated inhibition of translation. In response to accumulation of by-products of glycolysis such as acetate and formate (Chavez et al., 2010), BarA phosphorylates UvrY, which then activates transcription of genes including the CsrB regulatory RNA (Suzuki et al., 2002). Thus, mutations in barA, uvrY, or csrB decrease β-1,6-GlcNAc production in E. coli. Our transposon mutagenesis results suggested that the opposite is true in Yptb, which led us to focus our investigation on this regulatory system.

#### The Role of BarA/UvrY/CsrB on Biofilm Stability and Extracellular Polysaccharide Production

In order to verify that the transposon insertions did not affect expression of other genes, we complemented uvrY::Tn5 and barA::Tn5 strains with plasmids containing functional barA TABLE 1 | Yptb strain IP32953 Tn5 insertion mutants with increased Congo-red binding.


or uvrY genes. Overexpression of barA restored the wildtype phenotype (non-pigmented colonies) in the barA::Tn5 background. However, overexpression of uvrY in this strain resulted in an intermediate pigmentation phenotype, suggesting that phosphorylation of UvrY by BarA is required for UvrY to fully activate the relevant downstream targets. Overexpression of CsrB in both the barA::Tn5 and uvrY::Tn5 strains restored the non-pigmented phenotype, suggesting that the regulatory effect of BarA/UvrY occurs largely through CsrB and its regulation of CsrA. When we overexpressed the CsrA regulatory protein in wild type Yptb the cells were highly pigmented, but they grew more slowly and many non-pigmented colonies appeared in the population (data not shown) suggesting a toxic effect. Thus, the effect of the BarA/UvrY/CsrB system most likely affects pigmentation by regulating CsrA activity.

To define the role of this system in biofilm production, we first tested the ability of Yptb mutants lacking BarA or UvrY to form biofilms in 24-well polystyrene plates (**Figure 1**). The uvrY::Tn5 and csrB::Tn5 strains formed slightly thicker biofilms than the wild type, whereas the barA mutant strain was not significantly different from wild type in this assay. Similar to the Congo-red binding, overexpression of CsrB was sufficient to reduce biofilm production in the barA::Tn5 and uvrY::Tn5 mutants.

We used an additional assay specifically to investigate the cohesiveness of the biofilms formed by these strains. We allowed biofilms to form on polycarbonate filters and then measured the tendency of the biofilm to become dislodged and cells to become suspended in solution. The biofilms were placed in tubes with saline and agitated. The proportion of the biofilm that was dislodged and suspended in the solution was measured after a 1-h period (**Figure 2**). This assay showed that the wild type biofilms were easily disrupted, whereas the csrB::Tn5 and uvrY::Tn5 biofilms strongly adhered together. The barA::Tn5 biofilms were less cohesive than the uvrY::Tn5 or csrB::Tn5 biofilms but more than the wild type, similar to the Congo-red binding results.

Since the uvrY::Tn5 mutant produces strongly cohesive biofilms in vitro, we considered whether this strain could form biofilms in fleas and cause proventricular blockage similar to Y. pestis, or to Yptb mutants that lack rcsA, hmsT and hmsD (Sun et al., 2014). We infected Xenopsylla cheopis fleas with wild type Yptb or the uvrY::Tn5 mutant, and monitored the fleas for 28 days for signs of proventricular blockage. Although both strains were able to colonize and maintain stable infections, none of the fleas were observed to develop blockage over this period. However, more fleas infected with the mutant strain survived the first 24 h after infection than with the wild type strain (**Figure 3**), suggesting that UvrY may regulate acute toxicity of Yptb within the flea digestive tract.

Pigmentation of Yersinia colonies on Congo-red agar is correlated with Hms-ECM production, but additional proteins and carbohydrates are also able to bind the dye. To more specifically assess the surface carbohydrate differences in these strains, we employed fluorescently-labeled lectins with affinity to specific sugars. Flow cytometry was used to measure the fluorescence of individual bacteria after labeling with wheatgerm agglutinin (WGA, **Figures 4A,B**), which binds to dimers or trimers of N-acetylglucosamine, or to Ulex europaeus agglutinin (UEA-1, **Figures 4C,D**), which binds to α-linked fucose residues. We measured the percentage of bacteria within each sample with detectable lectin binding (% positive, **Figures 4A,C**) as well as the mean fluorescence of the entire population (**Figures 4B,D**). We found that the binding of WGA to Y. pestis KIM6+ was very high, and this binding was strongly inhibited by addition of competing

FIGURE 1 | BarA, UvrY, and CsrB negatively regulate biofilm production in Yptb IP32953. Congo-red binding phenotypes were qualitatively assessed and categorized (light pink indicates little to no binding and red indicates strongly pigmented colonies). Biofilm attachment after 48 h to polystyrene plates was quantified by crystal violet staining (A590nm). Results are means and standard deviations for a representative experiment (n = 3) that was performed three times. One-way ANOVA analysis was performed with Tukey's correction for multiple comparisons and strains significantly different from the wild type strain are indicated (\*adjusted P-value < 0.05, \*\* < 0.01).

FIGURE 2 | Loss of uvrY or csrB increases the cohesiveness of Yptb biofilms. Biofilms of each strain were grown on polycarbonate filters and then transferred to saline and agitated for 1 h. The proportion of the biofilm that was disrupted by the agitation was measured. Results are means and standard deviations of three independent experiments. One-way ANOVA analysis was performed with Tukey's correction for multiple comparisons. Columns with the same letter (a, b, c) are not significantly different from each other (95% confidence interval).

N-acetyl-chitobiose. UEA-1 bound only modestly to Y. pestis, consistent with the majority of the extracellular carbohydrate consisting of Hms-ECM. Surprisingly, the wild type Yptb strain did not bind detectably to WGA and only moderately to UEA-1.

FIGURE 3 | Loss of uvrY does not affect Yptb colonization or blockage due to biofilm but does reduce acute oral toxicity toward X. cheopis fleas. Fleas were fed blood meals containing 10<sup>8</sup> CFU/ml wild type or uvrY::Tn5 mutant strain. (A) The proportion of fed fleas that died within 24 h of infection was significantly less for the uvrY::Tn5 mutant compared to the wild type strain (\*\*p < 0.01 by unpaired T-test) in three independent feeding experiments. (B) Proportions of fleas that remained infected with the uvrY::Tn5 mutant or wild type strain after 28 days. No signs of proventricular blockage were observed in any Yptb-infected fleas (wild type or mutant) during twice-weekly feedings.

This binding was not inhibited by the addition of competing chitobiose (for WGA) or fucose (for UEA-1), suggesting the possibility of secondary or tertiary sugar preferences for lectin binding to these bacteria. The csrB::Tn5 mutant strain bound more WGA lectin than the wild type strain, which is consistent with greater Hms-ECM production, but again the addition of competing chitobiose did not reduce lectin binding. Surprisingly, the rcsB::Tn5 mutant strain, which is known to express high levels of cyclic diguanylate and Hms-ECM, bound more strongly to UEA-1 than to WGA. These results suggest differences in the extracellular matrix between Y. pestis and Yptb.

#### Proteomic Investigation of BarA/UvrY Regulation Through CsrB

Phosphorylated UvrY upregulates CsrB expression in E. coli and Yersinia. Multiple copies of CsrA protein are bound by CsrB RNA, thus preventing it from binding its target mRNAs. In E. coli, CsrA binds to pgaABCD transcripts to repress their translation. CsrA is most often a translational repressor, but enhances translation of some target mRNAs (Potts et al., 2017). Unlike the pgaABCD upstream region, no obvious CsrA binding sites were identified within the coding sequences or in the 500 bp region immediately upstream of the hmsH start codon. We therefore considered whether the CsrA protein of Yptb could instead repress other negative regulators of Hms activity. We applied a proteomic approach on whole cell lysates prepared from wild type Yptb strain and the csrB::Tn5 mutant using LC-MS/MS

FIGURE 4 | Loss of rcsB and csrB alters lectin binding to bacterial surfaces. Bacteria were incubated with FITC-labeled WGA (A,B) or UEA (C,D) lectins. Flow cytometry was used to measure the proportion of bacteria with any detectable binding (A,C) or the mean fluorescence of the population (B,D). For each strain that was tested, a control with no lectin was included to establish baseline fluorescence, as well as samples with competing sugars (chitobiose or fucose). Results are means and standard deviations for a representative experiment that was performed four times. To see if the csrB::Tn5 or rcsB::Tn5 mutants bound lectin differently than the wild type strain, one-way ANOVA analysis was performed with Tukey's correction for multiple comparisons using the lectin-treated samples (blue bars). The csrB::Tn5 and rcsB::Tn5 mutants bound significantly more WGA than the wild type strain, but only the rcsB::Tn5 mutant bound more UEA than wild type. Y. pestis binding was tested as a control, and it bound to WGA far more efficiently than Yptb (\*\*p < 0.01, \*\*\*p < 0.001, \*\*\*\*p < 0.0001).

analysis and mapping the peptide fragments to the Yptb IP32953 database. We selected proteins that were at least 1.5-fold changed in abundance between the two strains (**Supplementary Table 2**). As expected, the vast majority of the differentially expressed proteins were less abundant in the csrB mutant, consistent with the role of CsrA protein as mainly a repressor of translation.

The global changes in transcript abundance in a Yptb strain YPIII csrA mutant were recently reported by Bücker et al. (2014). Many of the genes that are upregulated in their study showed lower protein levels in this analysis of strain IP32953 csrB::Tn5 mutant. The RovA (SlyA) protein is known to be repressed by CsrA (Heroven et al., 2008), and our data indicated a 28-fold decrease in RovA abundance. RovA is controlled by CsrA through the RovM protein, which is activated by CsrA (Heroven et al., 2008). RovM peptides were not detected among the wild type samples, but were in the csrB::Tn5 mutant (**Supplementary Table 2**). Several additional regulatory proteins that could regulate biofilm production were repressed in the csrB::Tn5 mutant strain. Nucleoid-associated proteins, including Integration Host Factor alpha (IHFα, 11.8-fold reduction) and histone-like nucleoid structuring protein (H-NS, 24-fold reduction) function to silence and activate gene expression on a global scale (Dillon and Dorman, 2010). IHFα was also identified in ourinitial transposon screen for mutants that exhibit enhanced Congo-red binding (**Table 1**). H-NS represses expression of horizontally acquired genes (Lucchini et al., 2006), and the hmsHFRS operon is contained within the mobile Yersinia highpathogenicity island (Lillard et al., 1997). H-NS also represses expression of pgaABCD and biofilm production in Actinobacillus pleuropneumoniae (Bossé et al., 2010). H-NS is believed to be essential for Yersinia survival (Ellison and Miller, 2006), so we were unlikely to identify it in our transposon screen.

We identified transposon insertions in both rcsA and rcsB in our screen for Congo-red binding mutants (**Table 1**). The Yptb RcsB transcriptional regulatory protein, together with RcsA, represses transcription of the diguanylate cyclases hmsT and hmsD, thereby reducing biofilm production. Y. pestis rcsA has a frameshift mutation, thus eliminating this repression and enhancing biofilms (Sun et al., 2008, 2014; Sun Y. C. et al., 2012; Guo et al., 2015). To ensure that the transposon insertions in rcsA and rcsB were not affecting other genes, we complemented both mutants via plasmids containing functional copies of these genes. The rcsB::Tn5 mutant was complemented with a plasmid containing rcsB. The rcsA::Tn5 mutant was able to be complemented via rcsA from Yptb but not from Y. pestis, consistent with the frameshift mutation in Y. pestis rcsA. Our proteomic data revealed that csrB::Tn5 produced 8-fold less RcsB protein than the wild type strain (**Supplementary Table 2**). CsrA has not been previously shown to repress expression of RcsB directly or indirectly. To test the relationship between rcsB and CsrB further, we overexpressed CsrB in the Yptb rcsA::Tn5 or rcsB::Tn5 mutant strains (**Figure 5**). Unlike in the uvrY::Tn5 or barA::Tn5 mutants, CsrB expression did not reduce Congored binding in the rcsA or rcsB mutants. This suggests the possibility that CsrB competes with rcsB transcripts for CsrA binding. The predicted secondary structure of the rcsB transcript

FIGURE 5 | Biofilm and Congo-red binding by rcsA and rcsB mutants is not affected by overexpression of CsrB. Congo-red binding phenotypes and biofilm attachment in polystyrene plates after 48 h were measured in wild type Yptb and in rcsA::Tn5 or rcsB::Tn5 mutants. The effects of plasmids containing rcsA from Yptb (pRcsAYPTB) or Y. pestis (pRcsAYP) were determined in the rcsA::Tn5 mutant strain. The rcsB::Tn5 mutant was complemented with a plasmid containing rcsB from Yptb. Overexpression of CsrB (pCsrB), which reduced Congo-red binding and biofilm production in barA::Tn5 and uvrY::Tn5 mutants (Figure 1), was also tested in the rcsA and rcsB mutants. Results are means and standard deviations for a representative experiment (n = 3) that was performed four times. One-way ANOVA analysis was performed with Tukey's correction for multiple comparisons. Values that were significantly different from the wild type are indicated (\*\*\*adjusted P-value < 0.001, \*\*\*\* adjusted P-value < 0.0001).

contains a stem-loop with an exposed GGA motif overlapping the Shine-Delgarno sequence (**Supplementary Figure 1**), which is characteristic of CsrA targets (Romeo and Babitzke, 2018). Alternatively, the absence of a functional Rcs phosphorelay could overpower the negative regulatory effects of CsrB.

Urease subunit proteins were also substantially reduced in the csrB::Tn5 mutant (**Supplementary Table 2**). The urease structural proteins (UreA 59-fold, UreB 21-fold, UreC 21-fold), and the accessory proteins (UreE 1423-fold, UreF undetected, UreG 45-fold, UreD 71-fold), make up the multimeric urease enzyme. Y. pestis lacks a functional UreD subunit, which is sufficient to eliminate urease activity (Sebbane et al., 2001). Unlike Y. pestis, Yptb strains are acutely toxic to X. cheopis fleas (Erickson et al., 2007), and restoration of urease also restores toxicity to Y. pestis (Chouikha and Hinnebusch, 2014). Since the uvrY::Tn5 mutant was also less toxic to fleas, we tested whether this strain produces less urease enzyme. The wild type strain produced much more urease than the uvrY::Tn5 mutant (**Figure 6**). The csrB::Tn5 mutant also produced much less urease, and no urease activity was detected using Y. pestis KIM6+. During the preparation of this manuscript, it was also reported by Dai et al. that CsrA represses urease production in Yptb strain YPIII, which is consistent with our results (Dai et al., 2018).

### DISCUSSION

Switching between planktonic and sessile modes of growth is a tightly controlled process involving multiple levels of transcriptional and post-transcriptional regulation. Successful navigation of this transition is critically important to the success of pathogenic yersiniae (Chen et al., 2016). In this study we demonstrated that the BarA/UvrY two component system as well as the small RNA CsrB repress biofilm production in Yptb. Disruption of CsrB increased biofilm production while at the same time reducing urease-mediated toxicity toward fleas. Many bacterial pathogens, including Yersinia (Bobrov et al., 2011; Zhao et al., 2013; Redelman et al., 2014), exhibit an inverse relationship between biofilm production and acute virulence. The regulatory networks that govern transition between highly toxic phenotypes and the biofilm mode of growth are of interest as possible therapeutic targets. CsrB is known to exert its effects by occupying CsrA proteins, thereby altering expression of CsrAcontrolled regulatory proteins and cell-envelope structures that have pleiotropic effects.

Previous work has shown that in Yptb strain YPIII, UvrY (and CsrB) expression is normally repressed by the cyclic AMP-Crp complex except during times of nutrient deprivation (Heroven et al., 2012). Starvation conditions would therefore be predicted to promote motility and reduce biofilm formation by reducing the concentration of free CsrA proteins. The downstream targets of CsrA that mediate the biofilm-specific effects are not completely known and may be different between individual strains of Yptb or in Y. pestis. In Yptb strain YPIII, CsrA indirectly activates expression of the RovM transcriptional regulatory protein (Heroven et al., 2008). RovM binds the promoter regions of the flhDC flagellar regulatory genes, the hmsHFRS operon, and the rovA virulence regulator (Heroven et al., 2008; Zhao et al., 2017). Binding of RovM represses flhDC and enhances hmsHFRS and rovA transcription. Others have also shown that CsrA enhances biofilms in Y. pestis (Willias et al., 2015). However, loss of CsrA did not change Y. pestis hmsHFRS, hmsT, or hmsP transcription as occurs in Yptb. It will be interesting to determine if RovM activates hmsHFRS in Y. pestis, or whether RovM is controlled by CsrA in the same fashion as in Yptb. We did not identify any transposon insertions in the rovM gene (YPTB2588) in our screen for Congored binding mutants. This could be because the expression of rovM is very low in nutrient-rich conditions such as were used in our study (Heroven and Dersch, 2006). Indeed, RovM protein levels in the wild type strain were below the limit of detection in our proteomic analysis. However, RovM peptides were detected in the csrB mutant samples, consistent with CsrA being an activator or RovM expression in Yptb strain IP32953. We also did not obtain any transposon insertion mutants in the CsrC regulatory RNA gene in our screen. In contrast to E. coli, expression of CsrC is not activated by UvrY in Yptb strain YPIII (Heroven et al., 2008). Rather, when CsrB expression is triggered by UvrY, CsrC is repressed by a mechanism that is yet to be determined.

In addition to RovM, our results suggest that other regulatory proteins that control biofilm attachment and stability are under the influence of CsrA. These include the RcsA and RcsB transcriptional regulatory proteins. The Rcs phosphorelay system contains a sensor kinase RcsC that autophosphorylates upon detection of signals that include cellular stress. The phosphate is then transferred through RcsD to RcsB, which then dimerizes and binds to specific promoter sequences. RcsB may also form heterodimers with RcsA, and RcsAB binding represses transcription of diguanylate cyclases hmsD and hmsT, as well as the hmsHFRS operon (Sun Y. C. et al., 2012; Fang et al., 2015). The rcsA gene has been disrupted in Y. pestis, which was a key event in the evolution of Y. pestis transmission by fleas because it greatly enhanced cyclic di-GMP and biofilm production (Sun et al., 2008, 2014). We found that when rcsA or rcsB were nonfunctional, overexpression of CsrB did not reduce Congo-red pigmentation or prevent biofilm formation. In our analysis of the csrB::Tn5 mutant proteins, RcsA and RcsB proteins were both decreased compared to the wild type strain. These results are suggestive of a possible direct interaction between CsrA and rcsA or rcsB transcripts, particularly in the case of rcsB.

Disruption of uvrY or csrB had a very strong effect on the cohesiveness of Yptb biofilms. We considered that the most likely explanation for this would be increased abundance of extracellular polysaccharide contained in the biofilm matrix, specifically the Hms-ECM in these mutants. Yersinia biofilms are often described as bacteria encased in a homopolymer of N-acetyl-D-glucosamine (Hms-ECM) (Zhou and Yang, 2011; Sun F. et al., 2012). Hms-ECM is almost certainly present in Y. pestis and Yptb biofilms, based on antigenic cross-reactivity of Y. pestis biofilms with antibodies directed against purified poly-N-acetylglucosamine (Erickson et al., 2006) and from the extensive genetic data showing the impact of hmsHFRS on biofilms in both species. However, our results also suggest that loss of CsrB could control biofilm production by affecting expression of other cell surface molecules. The binding of WGA lectin to Yptb mutants that were strongly pigmented on Congo-red agar was far less than what we observed for Y. pestis, and binding was barely detectable for the wild type IP32953 strain. Binding of UEA-1 lectin, which is specific for fucose, was increased in the rcsB mutant to a greater degree than WGA lectin, which is specific for N-acetylglucosamine. However, these results are complicated by the fact that addition of competing sugars did not reduce lectin binding, except for WGA and Y. pestis. The possibility of other proteins, carbohydrates, or nucleic acids within Yersinia biofilms deserves further investigation, particularly in Yptb. It is likely that Hms-ECM is increased on the surface of uvrY, csrB and rcsB

#### REFERENCES


Yptb mutants, but other macromolecules within the biofilm matrix may interfere with efficient WGA lectin binding. The presence of additional carbohydrates within Yersinia biofilms is suggested by previous lectin binding experiments in other Yptb strains (Tan and Darby, 2004) as well as genetic studies showing that lipopolysaccharide genes are required for efficient biofilm production in Y. pestis (Liu et al., 2016). Yptb biofilm extracellular matrix also contains extracellular DNA (Atkinson et al., 2011). The protein component of Yersinia biofilms has not been well-studied, but even when lacking the Hms polysaccharide, Y. pestis can form biofilm-like aggregates that resist killing by neutrophils (Jarrett et al., 2004). The biochemical nature of the components within these aggregates is unknown, but may suggest that specific Yersinia proteins that perform stabilizing and attachment (Fong and Yildiz, 2015) or immune evasion functions within the ECM await discovery.

#### AUTHOR CONTRIBUTIONS

JS, JP, and DE conceived and designed the experimental procedures, JS, RS, MS, JC, HK, and DE performed the experiments and carried out data analysis, JS and DE wrote the manuscript.

#### FUNDING

This work was funded by a Mentoring Environment Grant from Brigham Young University.

#### ACKNOWLEDGMENTS

We thank Petra Dersch for the CsrB expression plasmid.

### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00323/full#supplementary-material


in the flea vector by two c-di-GMP diguanylate cyclases. PLoS ONE 6:e19267. doi: 10.1371/journal.pone.0019267


for Yersinia pestis biofilm formation. Environ. Microbiol. 12, 2034–2047. doi: 10.1111/j.1462-2920.2010.02219.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Schachterle, Stewart, Schachterle, Calder, Kang, Prince and Erickson. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# BfvR, an AraC-Family Regulator, Controls Biofilm Formation and pH6 Antigen Production in Opposite Ways in Yersinia pestis Biovar Microtus

Haihong Fang1,2†, Lei Liu3†, Yiquan Zhang<sup>4</sup> , Huiying Yang<sup>1</sup> , Yanfeng Yan<sup>1</sup> , Xiaojuan Ding<sup>5</sup> , Yanping Han<sup>1</sup> , Dongsheng Zhou<sup>1</sup> \* and Ruifu Yang<sup>1</sup> \*

<sup>1</sup> State Key Laboratory of Pathogen and Biosecurity, Beijing Institute of Microbiology and Epidemiology, Beijing, China, <sup>2</sup> Division of Biology, Beijing Academy, Beijing, China, <sup>3</sup> Department of Blood Transfusion, Wuhan General Hospital of PLA, Wuhan, China, <sup>4</sup> School of Medicine, Jiangsu University, Zhenjiang, China, <sup>5</sup> Department of Microbiology, Anhui Medical

#### Edited by:

Matthew S. Francis, Umeå University, Sweden

#### Reviewed by:

Vladimir L. Motin, The University of Texas Medical Branch at Galveston, United States Michael Marceau, Université Lille Nord de France, France David Erickson, Brigham Young University, United States

#### \*Correspondence:

Dongsheng Zhou dongshengzhou1977@gmail.com Ruifu Yang ruifuyang@gmail.com

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 06 May 2018 Accepted: 11 September 2018 Published: 02 October 2018

#### Citation:

Fang H, Liu L, Zhang Y, Yang H, Yan Y, Ding X, Han Y, Zhou D and Yang R (2018) BfvR, an AraC-Family Regulator, Controls Biofilm Formation and pH6 Antigen Production in Opposite Ways in Yersinia pestis Biovar Microtus. Front. Cell. Infect. Microbiol. 8:347. doi: 10.3389/fcimb.2018.00347 Biofilm formation is critical for blocking flea foregut and hence for transmission of Y. pestis by flea biting. In this study, we identified the regulatory role of the AraC-family transcriptional regulator BfvR (YPO1737 in strain CO92) in biofilm formation and virulence of Yersinia pestis biovar Microtus. Crystal violet staining, Caenorhabditis elegans biofilm assay, colony morphology assay, intracellular c-di-GMP concentration determination, and BALB/c mice challenge were employed to reveal that BfvR enhanced Y. pestis biofilm formation while repressed its virulence in mice. Further molecular biological assays demonstrated that BfvR directly stimulated the expression of hmsHFRS, waaAE-coaD, and hmsCDE, which, in turn, affected the production of exopolysaccharide, LPS, and c-di-GMP, respectively. In addition, BfvR directly and indirectly repressed psaABC and psaEF transcription, respectively. We concluded that the modulation of biofilm- and virulence-related genes by BfvR led to increased biofilm formation and reduced virulence of Y. pestis biovar Microtus.

#### Keywords: Yersinia pestis, BfvR, biofilm, virulence, transcriptional regulation

## INTRODUCTION

University, Hefei, China

Yersinia pestis, which causes severe and even fatal zoonotic diseases, has been responsible for the three plague pandemics throughout human history. Y. pestis can form biofilm not only in vitro, but also in the infected fleas that leads to a blockage in the proventriculus and enhances the flea-borne transmission (Hinnebusch et al., 1996). Studies on biofilm formation in Y. pestis have focused on the synthesis and degradation of: (i) extracellular matrix (Bobrov et al., 2008); (ii) intracellular secondary messengers, such as 3′ ,5′ -cyclic diguanylic acid (c-di-GMP) (Kirillina et al., 2004; Bobrov et al., 2011); and (iii) lipopolysaccharide (LPS) (Knirel and Anisimov, 2012). Transcription regulation can also modulate the expression of biofilm-related genes through multiple cellular pathways (Sun et al., 2009; Sun Y. C. et al., 2012; Willias et al., 2014, 2015).

Bacterial exopolysaccharide (EPS) is the primary component of the biofilm extracellular matrix. The hmsHFRS operon, located in the 102-kb pgm locus, coordinates the synthesis and transport of EPS in Y. pestis (Bobrov et al., 2008). The secondary signaling molecule c-di-GMP promotes EPS production in Y. pestis. This signaling molecule is catalyzed by diguanylate cyclases (DGC) and degraded by phosphodiesterase (PDE). In Y. pestis, DGC is encoded by two genes, hmsT and hmsD (derived from the threegene operon hmsCDE); while PDE is encoded by hmsP (Kirillina et al., 2004). In general, a high concentration of intracellular c-di-GMP can stimulate biofilm formation and interfere with virulence (Hengge, 2009). DGC and PDE inversely regulate biofilm formation, which is dependent on control of HmsHFRSbased poly-β-1, 6-N-acetylglucos-amine synthesis in Y. pestis (Bobrov et al., 2011).

LPS also plays a vital role in biofilm formation and is synthesized by a three-gene operon waaAE-coaD, containing genes waaA, waaE, and coaD, in Y. pestis. Deletion of waaA, which encodes a 3-deoxy-D-manno-octulosonic acid transferase involved in the synthesis of LPS (Tan and Darby, 2005), results in a biofilm-defective phenotype in Y. pestis (Liu et al., 2014).

It was reported that the expression of 214 genes significantly increased in the flea compared with all in vitro growth conditions, including y2570 (YPO1737 in CO92 and bfvR in Microtus strain 91001). Whole-genome microarrays in Y. pestis KIM6+ showed that y2570 had significantly higher expression levels in the flea gut but not the rat bubo (Vadyvaloo et al., 2010). We predicted and identified that bfvR, located between the bases 1,639,340 and 1,639,726 on the genome of Microtus strain 91001, can participate in biofilm formation in vivo (Song et al., 2004; Mao et al., 2016). In this research, we also studied the regulation mechanisms on biofilm formation for the mentioned genes by BfvR in Y. pestis.

Moreover, we found bfvR could inhibit the virulence of BALB/c mice in 91001. The AraC family transcription regulator typically binds to the target DNA and regulates bacterial virulence by sensing small molecule inducers such as urea, bicarbonate, or cellobiose etc. The molecular inducers were abundant at the sites where the bacterial pathogen colonizes and damages its host (Yang et al., 2011).

The protein Psa, known as pH 6 antigen, is one of the surface proteins involved in adhesion to the host cellular surface during initiation of Y. pestis pathogenesis. Psa proteins result from the polymerization of single PsaA pilin subunits (Bao et al., 2012), and the assembly of pH 6 antigen is mediated by the secretion of PsaB and PsaC, which constitutes the chaperone/usher machinery on the cell surface (Price et al., 1995). PsaABC is highly expressed following a temperature shift from 26 to 37◦C and acidic media, while PsaEF is responsible for the transcriptional activation of psaABC (Yang and Isberg, 1997). pH 6 antigen mediates the entry of Y. pestis into human pulmonary epithelial cells (Liu et al., 2006; Galvan et al., 2007) and promotes the delivery of Yops (effectors of the plasmid pCD1-encoded type III secretion system) into target host cells via cell-to-cell contact between Y. pestis and eukaryotic cells (Felek et al., 2010). pH 6 antigen does not strengthen the attachment to mouse macrophages, however, the resistance to phagocytosis can be developed independent of Yersinia outer proteins and capsule antigen in Y. pestis KIM5 (Huang and Lindler, 2004). However, loss of Psa has no effect on the virulence of Y. pestis strain 231 following subcutaneous challenge of naive and pH 6 antigenimmunized mice (Anisimov et al., 2009). In our opinion, the role of pH 6 antigen in Y. pestis virulence appears to be dependent on the strains tested and the animals challenged. In our study, we showed that by decreasing the transcription of psaABC and psaEF, BfvR could influence the survival of mice infected with strain 91001.

There are 7–19 AraC- or AraC-like-family transcription regulatory genes deposited in the GenBank database from Yersinia genomes. Y. pestis biovar Microtus strain 91001, Orientalis strain CO92, and Medievalis strain KIM D27 possess 19, 13, and 14 AraC family transcription regulators, respectively. YbtA, a previously reported AraC family transcription regulator in Y. pestis, can promote the transcription of psn (receptor for yersiniabactin) and irp2 (yersiniabactin biosynthesis gene), but inhibit the transcription of ybtA itself (Fetherston et al., 1996). MarA can affect bacterial drug sensitivity (Gillette et al., 2000), both MarA47 and MarA48 in the KIM strain of Y. pestis can regulate drug sensitivity and virulence (Lister et al., 2010). LcrF, similar to ExcA (an AraC family transcription regulator in Escherichia coli), can also regulate the transcription of the Ysc-Yop genes of type III secretion system in Y. pestis (King et al., 2013; Li et al., 2014). However, none of the AraC family transcription regulators have been reported to regulate biofilm formation in Y. pestis.

In this study, we firstly revealed the regulation of biofilm formation by BfvR in Y. pestis. In addition, our findings offer insights into the coordination of virulence and biofilm formation in Y. pestis.

### MATERIALS AND METHODS

#### Bacterial Strains and Mutant Preparation

The wild-type (WT) Y. pestis strain 91001 is avirulent in humans but highly lethal to mice (Zhou et al., 2004). The entire coding region of bfvR was replaced with the kanamycin resistance cassette using the one-step inactivation method based on the lambda Red recombination system (Datsenko and Wanner, 2000) and the Y. pestis bfvR null mutant was designated as ∆bfvR. All primers used in this study were listed in **Table S1**.

A PCR-generated DNA fragment containing the bfvR coding region with its 609-bp upstream promoter-proximal region and 267-bp downstream transcriptional terminator region was cloned into the pACYC184 vector (GenBank accession number X06403). The recombinant plasmid was introduced into ∆bfvR, yielding the complementary strain C-bfvR.

For psaABC-related gene regulation experiments, supplemented BHI (sBHI) broth [3.7% Bacto brain heart infusion (BD Biosciences), 0.5% Oxoid yeast extract, 2.5 mM CaCl2, 0.2% xylose, pH 6.0] was used for bacterial cultivation (Lindler et al., 1990). An overnight bacterial culture with an optical density (OD620) value of about 1.0 was diluted 1:20 into fresh sBHI for further growth at 26◦C with 230-rpm shaking. The cell culture with an OD620 of about 0.6 was transferred to 37◦C for a further 3-h culture prior to the harvesting of cells.

For bfvR gene regulation and phenotypic experiments, an overnight cell culture in Luria–Bertani (LB) broth with an OD620 of about 1.5 was immediately diluted 1:20 into fresh LB broth for the second generation cultivation at 26◦C to reach an OD620 of about 1.0 and the culture was stored at 4◦C for up to 24 h. The

tail vein (E) and the subcutaneous injection at the inguinal region (F) and evaluated the virulence by the survival curve and P value with the GraphPad Prism 5.0 software. This is a representative experiment. P1 and P2 represented the P values between WT and ∆bfvR or C-bfvR, respectively. \*\*\* Shows significant difference, ns indicates no significant difference.

culture was also diluted 1:20 into fresh LB broth for the third cultivation at 26◦C for 6–8 h to reach an OD620 of about 1.0.

#### Biofilm-Related Assays

Four different methods (Fang et al., 2013, 2015) of biofilmrelated assays were employed: (i) Crystal violet staining of biofilms, the in vitro biofilm masses attached to the well walls were stained by crystal violet when bacteria were grown in polystyrene microtiter plates; (ii) Caenorhabditis elegans biofilm assay, the nematode eggs were inoculated onto Y. pestis lawns and the percentages of fourth-stage larvae and adults (L4/adult) of C. elegans were determined for evaluating the bacterial ability to produce biofilms; (iii) the colony morphology assay, the rugose colony morphology of bacteria grown on LB agar plates was observed at different times for assessing the bacterial ability to synthesize biofilm matrix exopolysaccharide;

FIGURE 2 | Regulation of hmsH transcription by BfvR. Lanes G, A, T, and C represented Sanger sequencing reactions. Symbols + and – indicated the nucleotide positions downstream and upstream of the translation start site of target gene, respectively. (A) Primer extension. A(−128) represents the transcription start site of hmsH. (B) Quantitative RT-PCR. The relative mRNA expression was determined by calculating the threshold cycle (Ct) of hmsH via the classic 2-11Ct method through standardized by the 16S rRNA gene. (C) LacZ fusion. LacZ reporter fusion with the promoter-proximal region (−451…+103) of hmsH and β-galactosidase in cellular extracts were determined. (D) EMSA. Purified His-BfvR protein bound to the promoter-proximal fragment of hmsH labeled radioactively and the retarded DNA band were shown in the amount-dependent manner.

and (iv) determination of intracellular c-di-GMP levels by a chromatography-coupled tandem mass spectrometry method (Spangler et al., 2010).

### Murine Infection Model

All animal experiments were carried out in accordance with the principles of the Basel Declaration and recommendations of

the Guidelines for Welfare and Ethics of Laboratory Animals of China, the Committee on Animal Research of the Academy of Military Medical Sciences. The protocol was approved by the Committee on Animal Research of the Academy of Military Medical Sciences. Bacterial cultures were washed twice with PBS (pH 7.2), and then subjected to serial 10-fold dilutions with PBS. Appropriate dilutions were plated onto He's agar plates (1 L powder containing pig blood peptone 12 g, K2HPO41 g, NaCl 4 g, and agar 13 g, LandBridge,China) to calculate the number of colony-forming units (CFU). For each strain tested, 0.1 ml of the 10<sup>3</sup> CFU/ml bacterial suspensions were inoculated by subcutaneous injection at the inguinal region or by intravenous injection via the tail vein into each of 10 female BALB/c mice (aged 6 to 8-week-old). The numbers of mice that died at specified times were recorded and used to create a survival curve with the GraphPad Prism 5.0 software. P-values were calculated with the log-rank (Mantel–Cox) test and the Gehan–Breslow– Wilcoxon test affiliated in this software. P < 0.01 was considered to indicate statistical significance.

#### Primer Extension (PE) Assay

For the PE assay (Ghosh et al., 1978; Sun F. et al., 2012; Zhang et al., 2013), an oligonucleotide primer complementary to a portion of the RNA transcript of each indicated gene was used to synthesize cDNAs. About 10–30 µg of total RNA from each strain was annealed with 1 pmol of [γ−32P] end-labeled reverse primer using a Primer Extension System (Promega) in accordance with the manufacturer's instructions. The same labeled primer was also used for sequencing with the fmol <sup>R</sup> DNA Cycle Sequencing System (Promega). The PE products and sequencing DNAs were concentrated and analyzed in a 6% polyacrylamide/8 M urea gel. The results were detected by autoradiography (Kodak film) and then analyzed.

#### Quantitative RT-PCR

Gene-specific primers were designed to produce amplicons for the target genes. Contaminating DNA in the RNA samples was removed using the Ambion DNA-freeTM Kit (Applied Biosystems). cDNAs were generated using 5 µg of RNA and 3 µg of random hexamer primers. Real-time PCR was performed using the LightCycler system (Roche) and the SYBR Green master mix (Takara) (Zhan et al., 2008; Sun et al., 2014). Based on the standard curves of 16S rRNA gene expression, the relative mRNA level was determined by calculating the threshold cycle (Ct) of target genes via the classic 2-11CT method. Negative controls used cDNA generated without reverse transcriptase as templates. Reactions containing primer pairs without template were also included as blank controls. The 16S rRNA gene was used as an internal control for normalization.

#### LacZ Reporter Fusion and β-Galactosidase Assay

The promoter-proximal DNA region of each gene tested was prepared by PCR with Takara ExTaq DNA polymerase using Y. pestis strain 91001 genomic DNA as the template. This purified fragment was then cloned directionally into the HindIII–BamHI site of the transcriptional fusion vector pRW50 (El-Robh and Busby, 2002) that contained a promoterless lacZ reporter gene. The clone was verified by DNA sequencing. Each Y. pestis strain tested was transformed with the recombinant plasmid and the empty plasmid pRW50 as a negative control. β-galactosidase activity in extracts was measured from the cells cultivated as described above according to the β-Galactosidase Enzyme Assay System (Promega) (Sun F. et al., 2012; Zhang et al., 2013).

### Preparation of 6×His-Tagged BfvR Protein

The preparation of purified BfvR protein was performed as previously described (Heroven and Dersch, 2006). The entire coding region of bfvR was amplified from Y. pestis strain 91001 and cloned directionally into the BamHI and HindIII sites of plasmid pColdI (Qing et al., 2004) (Novagen, GenBank accession number AB186388). The recombinant plasmids were transformed into E. coli BL21 (DE3) cells (Novagen). Expression of His-BfvR protein was induced by the addition of 1 mM or 2 mM isopropyl-beta-D-thiogalactoside at 16◦C overnight. His-BfvR was purified under native conditions using QIAexpressionistTM Ni-NTA affinity chromatography (Qiagen). The purified eluted protein was concentrated with an Amicon Ultra-15 column (Millipore) to a final concentration of 0.5– 0.7 mg/ml in storage buffer containing 10 mM Tris-HCl, 1 mM EDTA, 0.1 mM DTT, and 5 mM β-mercaptoethanol (pH 7.0) plus 20% glycerol. The protein purity was verified by sodium dodecyl sulfate–polyacrylamide gel electrophoresis with Coomassie brilliant blue staining.

### Electrophoretic Mobility Shift Assay (EMSA)

For EMSA, we have made minor modifications according to the Hellman's method (Hellman and Fried, 2007). Promoterproximal DNA region was prepared by PCR amplification. The 5 ′ end of the DNA was labeled using [γ-<sup>32</sup>P] ATP and T4 polynucleotide kinase. DNA binding was performed in a 10 µl volume containing binding buffer [100µM MnCl2, 1 mM MgCl2, 0.5 mM DTT, 50 mM KCl, 10 mM Tris-HCl (pH 7.5), 0.05 mg/ml sheared salmon sperm DNA, 0.05 mg/ml BSA, and 4% glycerol], labeled DNA (1,000 to 2,000 c.p.m/µl), and increasing amounts of His-RovA or His-BfvR. Two control reactions were included: one contained the specific DNA competitor (unlabeled promoter DNA regions; cold probe), while the other was the nonspecific protein competitor (rabbit anti-F1-protein polyclonal

IgG antibody). After incubation at room temperature for 30 min, the products were loaded onto a native 4% (w/v) polyacrylamide gel and electrophoresed in 0.5×Tris-borate buffer containing 100µM MnCl<sup>2</sup> for 30 min with 220 V at 8◦C. Radioactive species were detected by autoradiography.

#### Experimental Replicates and Statistical Methods

For real-time RT-PCR, the β-galactosidase activity assay, crystal violet staining of biofilms, the determination of L4/adult nematodes, c-di-GMP experiments and murine infection were performed with at least three independent bacterial cultures/lawns, and values were expressed as the mean ± standard deviation. The paired Student's t-test was performed to determine significant differences; P < 0.01 was considered to indicate statistical significance. For primer extension, EMSA, and colony morphology observations, representative data from at least two independent biological replicates were shown.

#### RESULTS

### Involvement of BfvR in Biofilm Formation and Virulence in Mice

To investigate the potential role of BfvR in biofilm formation, the phenotypes of relevant mutant strains were analyzed. We selected four experimental strains: (i) wild-type (WT); (ii) bfvR mutant strain (1bfvR); (iii) pACYC184-1bfvR complementary strain (C-bfvR); and (iv) a strain containing a mutation in the biofilm-related gene hmsS (1hmsS), a typical biofilm negative strain.

Deletion of bfvR had no effect on the growth in LB (**Figure S1**) but led to a dramatic reduction in biofilm formation in Y. pestis strain 91001. Compared with the WT and C-bfvR strains, the 1bfvR mutant strain showed less crystal violet staining (**Figure 1A**), with smoother colony morphology (**Figure 1B**), similar to the 1hmsS mutant. After the nematode eggs were fed with 1hmsS strain, 93% of the eggs developed into L4/adult nematodes, compared with 19.3, 47.3, and 20.3% for the WT, 1bfvR, and C-bfvR strains, respectively (**Figure 1C**), indicating bfvR might affect C. elegans development. In general, when fed with WT strain, about 20% or fewer larvae grew and developed into L4/adult nematodes (Zhou and Yang, 2011; Fang et al., 2013); however, the exact ratio of nematode adults did not directly correlate with the biofilm formation ability of Y. pestis. The concentration of intracellular c-di-GMP was quantitatively detected by UPLC-MS/MS analysis (Spangler et al., 2010). Lower production of cellular c-di-GMP was observed in strain 1bfvR (11.45 nmol/µg) compared with the WT (19.68 nmol/µg) and C-bfvR (32.70 nmol/µg) strains (**Figure 1D**). These phenotypes suggested that deletion of bfvR resulted in an obvious reduction in biofilm/c-di-GMP production.

Wild type bacteria, the BfvR mutant and the complemented strain were used to infect mice through both subcutaneous injection and intravenous injection. This was repeated three times independently for WT and the BfvR mutant, and repeated two times for the complemented strain. Finally, we calculated all the results and drew the survival curves with Graphpad 5.0. Compared with WT, ∆bfvR displayed a significant increase in virulence in BALB/c mice after intravenous (P = 0.0003) or subcutaneous (P = 0.0008) injection. Moreover, compared to WT, the complementary effect of C-bfvR also showed no significant differences during the intravenous (P = 0.065) or subcutaneous (P = 0.4017) routes of infection (**Figures 1E,F**).

Taken together, BfvR could enhance biofilm formation and inhibit the virulence of Y. pestis strain 91001 in BALB/c mice.

#### BfvR Acts as an Activator of EPS and c-di-GMP

In order to reveal the mechanisms by which BfvR modulated the synthesis of EPS, we analyzed expressions of the related genes including hmsH and waaA. The genes encoding DGC and PDE, namely hmsC, hmsT, and hmsP in Y. pestis, were also evaluated by qRT-PCR, PE, β-galactosidase activity assays and EMSA.

BfvR directly promoted hmsH expression. The mRNA levels of hmsH in WT and ∆bfvR strains were analyzed by primer extension. The levels of mRNA from the two transcriptional start sites (Sun F. et al., 2012) were found to be lower in ∆bfvR than in the WT strain (**Figure 2A**). Quantitative realtime PCR (qRT-PCR) assay indicated a significant reduction in relative mRNA expression in ∆bfvR compared with the WT strain (**Figure 2B**). A β-galactosidase activity assay also indicated that the promoter activity of hmsH (**Figure 2C**) was attenuated in ∆bfvR relative to the WT. EMSA showed that His-BfvR could bind to the promoter-proximal regions of hmsH in a dosedependent manner (**Figure 2D**). These results indicated that BfvR in Y. pestis strain 91001 could promote biofilm formation by directly stimulating hmsH expression.

BfvR directly induced waaAE-coaD expression. The transcription of waaA (Liu et al., 2014), the first gene in the waaAE-coaD operon, was investigated in both WT and ∆bfvR strains. Both primer extension (**Figure 3A**) and qRT-PCR assays (**Figure 3B**) revealed a significant reduction in waaAEcoaD expression in ∆bfvR compared with the WT strain. The promoter activity of waaA (**Figure 3C**) was attenuated in ∆bfvR relative to the WT, as shown by the β-galactosidase activity assay. EMSA revealed that His-BfvR could bind to the upstream regions of waaA- promoter (**Figure 3D**) in a dose-dependent manner. These results showed that BfvR in Y. pestis could promote biofilm formation by directly stimulating the expression of waaAE-coaD.

The c-di-GMP synthesis- and degradation-related genes hmsT, hmsCDE, and hmsP were also regarded as major targets of biofilm formation at the post-transcriptional level in Y. pestis. The mRNA levels and binding activities to His-BfvR of these three genes were investigated by PE, qRT-PCR, βgalactosidase activityassay, and EMSA. Compared withWT, ∆bfvR only showed a significant decrease in mRNA levels of hmsCDE (**Figures 4A–C**) but not hmsT (**Figures S2A–C**) or hmsP (**Figures S3A–C**). EMSA also showed that only the promoter-proximal regions of hmsC (**Figure 4D**) but not hmsT (**Figure S2D**) or hmsP (**Figure S3D**) could bind His-BfvR in vitro. These results indicated that BfvR in Y. pestis could promote c-di-GMP synthesis and regulate biofilm formation by directly stimulating the expression of hmsCDE.

#### Regulation of the psa Loci by BfvR

Two loci, psaABC and psaEF located upstream of psaA, responsible for expression of pH 6 antigen (Psa) in Y. pestis, were selected as target genes for investigating the regulatory action of BfvR in virulence. PE, qRT-PCR, and β-galactosidase activity assays showed that deletion of bfvR led to elevated expression of psaA (**Figures 5A–C**) and psaE (**Figures S4A–C**). EMSA showed that His-BfvR could not bind the promoter-proximal region of psaE (**Figure S4D**) but could bind that of psaA (**Figure 5D**). In our opinion, BfvR might regulate Y. pestis virulence through acting on many unknown factors at the same time, but BfvR could at least influence Y. pestis virulence by controlling the expression of psaABC.

### DISCUSSION

The AraC/XylS family of polypeptides is broadly distributed among Gram negative and positive bacteria (Tobes and Ramos, 2002). Biofilm formation modulation mediated by the AraC-type regulator has been reported for various types of bacteria but, prior to this study, not for Y. pestis. In this work, we showed for the first time that BfvR could similarly regulate biofilm formation in Y. pestis strain 91001.

Production of extracellular matrix was an important mechanism modulated by the AraC-type transcriptional regulator during biofilm formation. Polysaccharide intercellular adhesin (PIA)/poly-N-acetylglucosamine production was shown to be vital for Staphylococcus epidermidis biofilm formation, and this was encoded by the icaADBC operon, which in turn was modulated by the AraC-type regulator Rbf (Rowe et al., 2016). In Enterococcus faecium, the putative AraC type of regulator EbrB, regulated by esp operon expression, was shown to be responsible for the production of the enterococcal surface protein and was implicated in biofilm formation (Top et al., 2013). Another putative AraC-type transcriptional regulator PerA in a clinical isolate of E. faecalis strain E99 was detected to be involved in survival within macrophages, pathogenesis in mice, and biofilm formation (Coburn et al., 2008). Similarly, data also showed that BfvR could promote biofilm formation by increasing the synthesis of bacterial EPS and LPS, by directly stimulating the expression of hmsHFRS and waaAE-coaD, respectively.

WaaA plays an important role in the synthesis of lipid A (Dentovskaya et al., 2011) in Yersinia, which is responsible for endotoxin and virulence in mice in Y. pestis KIM6+ (Sun et al., 2013). It seems to be contradiction that there could be an ∼75% reduction in waaA gene expression (**Figure 3B**) and therefore presumably reduced LPS core oligosaccharide production, yet the virulence of the mutant is still enhanced. In fact, the direct effect of WaaA on mice virulence of Y. pestis strain 91001 has not been specifically reported. We found that deletion of PhoP or RcsB dramatically induced the downregulation or upregulation of waaA, respectively (Liu et al., 2014). However, loss of PhoP or RcsB had no significant effect on Y. pestis strain 91001 virulence tested by us (data not shown). In our opinion, downregulation of waaA in the BfvR mutant will not necessarily led to a notable alteration of Y. pestis strain 91001 virulence in mice. The mechanisms by which WaaA modulates strain 91001 virulence in mice still needs further investigation.

Within current sequence databases, the AraC/XylS family proteins are the most common positive regulators (Tobes and Ramos, 2002). It has been found that deletion of bfvR in Y. pestis strain 91001 resulted in a dramatic reduction in biofilm production through biofilm-related phenotypic assays. Until now, there has been no report to indicate that c-di-GMP production is regulated by an AraC-family protein. In this study, we demonstrated that deletion of bfvR in Y. pestis affected the intracellular c-di-GMP levels shown in the biofilm phenotypic assay. Similar to most AraC/XylS family regulators, BfvR displayed positive regulation during biofilm formation in Y. pestis.

The AraC/XylS family of transcription factors have been shown to play an important role in regulating bacterial virulence by modulating the expression of virulence-related genes in Citrobacter rodentium (Hart et al., 2008) and Salmonella enterica serovar Typhimurium (Bailey et al., 2010). Work reported here also indicated that a similar mechanism existed on virulence modulation by BfvR in Y. pestis. BfvR can repress Y. pestis virulence in mice by directly and indirectly inhibiting the expression of psaABC and psaEF at the transcriptional level, respectively. This work further extended our knowledge of regulation mechanisms under the control of the AraC regulator and their biological relevance in vitro (**Figure 6**).

Based on the results of an NCBI conserved domain assay (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi), the structure of BfvR was found to be homologous to that of SoxS of Xenorhabdus ishibashii with 100% coverage and 78% identity aligned by BlastP. SoxS protein was a direct transcriptional activator of the oxidative stress genes of the SoxRS regulon (Li and Demple, 1994), and was widely investigated in strains of family Enterobacteriaceae (Gallegos et al., 1997). SoxS regulated the expression of ompW gene, which was involved in osmoregulation as a minor porin, in Salmonella enterica serovar Typhimurium (Gil et al., 2009). ydbK gene (involved in superoxide resistance) and ompN (participating in the minor porins like ompW), which were coexpressed in an operon, were indirectly activated by SoxS in a multidrug-resistant E. coli strain NorE5 (Fabrega et al., 2012). Redox-cycling drugs rather than superoxide can directly activate the SoxRS response in E. coli (Gu and Imlay, 2011). Whether there is a similar SoxRS response shown by BfvR in Y. pestis remains to be investigated.

Taken together, our findings suggested that the AraC-family regulator BfvR was firstly found to be involved in the modulation of both biofilm formation and pathogenesis in Y. pestis strain 91001. In this study we only focused on biofilm formation and pathogenesis regulated by this regulator, the regulatory role of BfvR in other processes, such as sensing the redox pressure in Y. pestis, requires elucidation in future studies.

### AUTHOR CONTRIBUTIONS

DZ and RY conceived the study and designed the experimental procedures. HF, LL, YZ, HY, YY, XD, and YH performed the experiments and carried out data analysis. DZ, HF, LL, YZ, and RY wrote the paper.

### FUNDING

This work was supported by the National Natural Science Foundation of China (31430006), the National Special Project on Research and Development of Key Biosafety Technologies Grant (2016YFC1200100), the China Postdoctoral Science Foundation (2015M572755), the Hubei Province health and family planning scientific research project (WJ2018H0070), and the Natural Science Research Project from Department of Education of Anhui Province (KJ2015A019).

### ACKNOWLEDGMENTS

We thank Kate Fox, DPhil, from Liwen Bianji, Edanz Group China (www.liwenbianji.cn/ac) for critically reading and language editing the manuscript.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00347/full#supplementary-material

#### REFERENCES


is enhanced by the carbon storage regulator protein, CsrA. PLoS ONE 10:e0135481. doi: 10.1371/journal.pone.0135481


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Fang, Liu, Zhang, Yang, Yan, Ding, Han, Zhou and Yang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# The Infection Process of Yersinia ruckeri: Reviewing the Pieces of the Jigsaw Puzzle

#### José A. Guijarro\*, Ana I. García-Torrico, Desirée Cascales and Jessica Méndez

Área de Microbiología, Departamento de Biología Funcional, Facultad de Medicina, Instituto de Biotecnología de Asturias (IUBA), Universidad de Oviedo, Oviedo, Spain

Finding the keys to understanding the infectious process of Yersinia ruckeri was not a priority for many years due to the prompt development of an effective biotype 1 vaccine which was used mainly in Europe and USA. However, the gradual emergence of outbreaks in vaccinated fish, which have been reported since 2003, has awakened interest in the mechanism of virulence in this pathogen. Thus, during the last two decades, a large number of studies have considerably enriched our knowledge of many aspects of the pathogen and its interaction with the host. By means of both conventional and a variety of novel strategies, such as cell GFP tagging, bioluminescence imaging and optical projection tomography, it has been possible to determine three putative Y. ruckeri infection routes, the main point of entry for the bacterium being the gill lamellae. Moreover, a wide range of potential virulence factors have been highlighted by specific gene mutagenesis strategies or genome-wide transposon/plasmid insertion-based screening approaches, such us in vivo expression technology (IVET) and signature tagged mutagenesis (STM). Finally, recent proteomic and whole genomic analyses have allowed many of the genes and systems that are potentially implicated in the organism's pathogenicity and its adaptation to the host environmental conditions to be elucidated. Altogether, these studies contribute to a better understanding of the infectious process of Y. ruckeri in fish, which is crucial for the development of more effective strategies for preventing or treating enteric redmouth disease (ERM).

#### Edited by:

Victoria Auerbuch, University of California, Santa Cruz, United States

#### Reviewed by:

Petra Dersch, Helmholtz-Zentrum für Infektionsforschung (HZI), Germany Gokhlesh Kumar, Veterinärmedizinische Universität Wien, Austria

#### \*Correspondence:

José A. Guijarro jaga@uniovi.es

Received: 13 April 2018 Accepted: 08 June 2018 Published: 26 June 2018

#### Citation:

Guijarro JA, García-Torrico AI, Cascales D and Méndez J (2018) The Infection Process of Yersinia ruckeri: Reviewing the Pieces of the Jigsaw Puzzle. Front. Cell. Infect. Microbiol. 8:218. doi: 10.3389/fcimb.2018.00218 Keywords: Yersinia ruckeri, infection route, virulence genes, comparative genome, proteome analysis

## INTRODUCTION

Yersinia ruckeri is able to infect different fish species such as carp, catfish, sturgeon, perch and burbot (Tobback et al., 2007; Kumar et al., 2015), although it mainly affects salmonids. Interestingly, Y. ruckeri has recently been isolated as an unusual microorganism in a human wound infection (De Keukeleire et al., 2014) as well as in milk, cheese, chicken and minced meat (Özdemir and Arslan, 2015). In salmonids it causes enteric red mouth disease (ERM), which is important owing to the economic losses it causes in the aquaculture industry. ERM is a systemic disease affecting fish in all stages of development, causing a high degree of mortality. The disease occurs acutely in juveniles and has a tendency to become chronic in adult fish (Tobback et al., 2007). Some infected fish are asymptomatic, becoming carriers of the pathogen and acting as reservoirs of the bacterium. It is important to highlight that Y. ruckeri is a facultative intracellular bacterium, capable of surviving inside macrophages (Ryckaert et al., 2010).

Y. ruckeri strains have been classified into four serotypes (Romalde et al., 1993), the main outbreaks in fish farms being caused by serotype O1 biotype I, although other serotypes, particularly serotype O2, can also be involved (Romalde et al., 2003). Vaccination is the most effective way to control the disease. ERM biotype 1 vaccine, which was the first to be available for fish (Amend et al., 1983), has been used since the early 80s and is very effective in the prevention of the disease. However, during recent years worldwide outbreaks have been produced mainly by novel non-motile and lipase-negative strains, belonging to biotype II (Austin et al., 2003; Fouz et al., 2006; Arias et al., 2007; Calvez et al., 2014). These phenotypes have no appreciable effect on the virulence of the pathogen (Evenhuis et al., 2009; Welch et al., 2011), but the biotype 2 strains are able to elude the protection provided by the vaccine, probably due to the antigenic differences existing in their O-antigen with respect to biotype 1 strains (Tinsley et al., 2011). This has promoted the development and subsequent commercialization of a new vaccine which simultaneously confers protection against the two biotypes (Tinsley et al., 2011).

Owing to the prompt development of the vaccine, the study of the virulence mechanisms of this bacterium has been side-lined for a long time. Nevertheless, during the last few years a considerable number of reports related to the route of Y. ruckeri infection or its virulence, as well as several comparative genomic and proteomic analyses have been published. All of this knowledge about the infectious process of this bacterium is condensed and critically reviewed in the present article.

### YERSINIA RUCKERI INFECTION ROUTE

Studies aimed at determining the infection path and invasion mechanism of fish by Y. ruckeri have been carried out with different strains, modes of infection and experimental conditions, which sometimes makes it difficult to draw a conclusion. In addition, various tools have been developed and used for elucidating the interaction between Y. ruckeri and rainbow trout, its natural host.

Bacterial counting in fish tissues at different times after experimental infection (Tobback et al., 2009, 2010); GFP tagging of Y. ruckeri and further visualization by epifluorescence or flow cytometry in tissues (Welch and Wiens, 2005); in situ hybridization (Khimmakthong et al., 2013); bioluminescence imaging using strains harboring the luxCDABE operon from Photorhabdus luminescens (Méndez and Guijarro, 2013) and optical projection tomography (OPT) (Ohtani et al., 2014), a recently developed three-dimensional bioimaging technique combined with immunohistochemical assay (IHC) (Tobback et al., 2009; Ohtani et al., 2014), have all been used. Each of these technologies provides a different approach to determining the route of entry and tissue distribution of Y. ruckeri. Although all of them have limitations, if we analyse the results obtained as a whole, it seems that the gills are the portal of entry for the bacterium into the fish. In effect, most studies have established this organ as the first to be infected (Tobback et al., 2009; Khimmakthong et al., 2013; Ohtani et al., 2014, 2015), and more specifically the pavement cells of the gill lamellae, from which the bacterium would spread to blood (Ohtani et al., 2014). Thus, Y. ruckeri was detected only 1 min post infection in the blood, where the number of bacteria increased for 40 min, reaching 2.3x10<sup>5</sup> CFU/ml (Ohtani et al., 2014). Although gills seem to be the main point of entry of Y. ruckeri, other sites in the fish cannot be excluded. In effect, most studies establish the presence of Y. ruckeri in the intestine shortly after infection (Khimmakthong et al., 2013; Ohtani et al., 2014, 2015). It is known that the mucus of the gills is twice as efficient as that of the intestine in enabling the adherence of the bacterium (Tobback et al., 2010). Moreover, after infection, twice as many bacteria were detected in the gills as in the intestine (Tobback et al., 2009; Khimmakthong et al., 2013). However, bacteria have been detected in the intestine from 1 min (Khimmakthong et al., 2013) to 30 min post-infection (Ohtani et al., 2014). Additionally, intestinal dissemination is the main sign of the pathology (**Figure 1**) (Méndez and Guijarro, 2013; Ohtani et al., 2014). This, together with the fact that it is unusual for an intestinal pathogen to enter the host by any other route than the oral one means that the digestive tract cannot be excluded as another portal of entry. In addition, Ohtani et al. (2014) used OPT analysis to establish a third route of infection for Y. ruckeri, through the skin surface on the lateral line, from which it would penetrate to internal cell layers, a similar result to that found by Khimmakthong et al. (2013). A recent study reported the ability of two isolates of Y. ruckeri to invade different types of salmon epithelial cells (Menanteau-Ledouble et al., 2018), the biotype 1 isolate (ATCC 29473) being more infectious than the non-motile biotype 2 strain (A7959-11). The authors suggested that the bacterium could enter host cells by means of several mechanisms that take advantage of cytoskeletal systems. The capacity to invade the epithelium would undoubtedly facilitate entry into the host.

In conclusion, all the results indicate that Y. ruckeri can access the interior of rainbow trout by at least three different paths: the gills, lateral line and digestive tract. From these locations bacteria would spread to the blood circulation system to further colonize

expresses the highest number of counts in a pixel inside the region analyzed

(Figure taken from Méndez and Guijarro, 2013).

and infect other internal organs, causing in many cases the death of the fish.

#### DEFINED GENES RELATED TO VIRULENCE

Despite its being a microorganism responsible for a high proportion of losses in continental aquaculture, there are few studies on Y. ruckeri virulence factors, perhaps as a consequence of the early development of a vaccine, which made their study unattractive. Many of the virulence factors described so far refer to extracellular factors (ECF) that are common to a wide range of gram-negative pathogenic bacteria, particularly the enterobacteria group (**Table 1**, **Figure 2**). Indeed, extracellular proteases, haemolysins and siderophores are among the main extracellular molecules involved in the Y. ruckeri infection process.

Yrp1 is a serralysin family extracellular metalloprotease, widely characterized from an enzymatic point of view (Secades and Guijarro, 1999) and secreted through an ABC transport system (Fernández et al., 2002). It is able to degrade different matrix and muscle proteins such as laminin, fibrinogen, gelatin, actin, and myosin (Fernandez et al., 2003). Although it was shown not to be present in all virulent Y. ruckeri strains (Secades and Guijarro, 1999), its presence makes the bacteria more virulent. Transcriptional fusion between the luxCDABE operon and the yrp1 promoter demonstrated that yrp1 is induced more in vivo than in vitro (Méndez and Guijarro, 2013). This is consistent with the expression of yrp1 in fish tissues and the observation that an yrp1 mutant strain is significantly attenuated in virulence (Fernández et al., 2002). Interestingly, a heat-inactivated Yrp1 toxoid was able to elicit a strong protection against the ERM disease (Fernandez et al., 2003). All of this established a clear and important role for this protease in the pathogenic process.



Additional degrading enzymes were identified in Y. ruckeri, such as two putative peptidases belonging to the U32 family which are encoded by the yrpAB operon (Navais et al., 2014b). Both genes, yrpA and yrpB, are present in a similar genetic organization in different pathogens of the enterobacteria group, including human pathogenic Yersinia species (Navais et al., 2014b). A synergistic effect occurred in the induction of the yrpAB operon under low oxygen conditions when peptones were also present in the culture medium (Navais et al., 2014b). Infection studies with a yrpA mutant strain suggest that the YrpA peptidase contributes to virulence in a similar way to the U32 peptidases of Proteus mirabilis (Zhao et al., 1999) and Helicobacter pylori (Kavermann et al., 2003). A more detailed analysis of the role of these peptidases in the virulence of pathogenic Yersiniae could be interesting from a public health point of view.

Two different genetic approaches have been applied to allow the selection and further identification of Y. ruckeri genes related to the infection process: in vivo expression technology (IVET) (Fernandez et al., 2004) and signature-tagged mutagenesis (STM) (Dahiya and Stevenson, 2010a). Each of these has advantages and disadvantages, but undoubtedly, their application has led to the identification of a number of genes related to Y. ruckeri pathogenesis.

IVET allowed the identification of up to 14 genes specifically induced in vivo, during the infection process of Y. ruckeri. Among them, those related to two-component and type IV secretion systems, adherence, haemolytic activity and iron acquisition were further analyzed. It has been found that the genes involved in iron acquisition through the catechol siderophore ruckerbactin have a genetic organization similar to those of the E. coli enterobactin gene cluster (Fernandez et al., 2004). Regulation of this cluster depends on iron availability and temperature in such a way that low iron conditions and low temperature (18◦C in relation to 28◦C) induced its expression (Fernandez et al., 2004). It was shown that production of the siderophore is upregulated during the infection of fish (Fernandez et al., 2004; Fernández et al., 2007a) and it is also involved in virulence since 100-fold attenuation, in relation to the parental strain, was obtained when an isogenic mutant in rucC, a gene from the ruckerbactin cluster, was used in LD<sup>50</sup> experiments (Fernandez et al., 2004).

A putative additional system of Y. ruckeri for obtaining iron is the yhlBA cluster, also identified by IVET. The yhlA gene encodes a Serratia-type haemolysin, whereas the yhlB product is related to the secretion/activation of YhlA (Fernández et al., 2007b) (**Figure 2**). Analysis of yhlB::lacZY transcriptional fusion indicated that higher levels of expression of yhlB were obtained under iron-starvation conditions, and also at 18◦C than at 28◦C (Fernández et al., 2007b). Interestingly, the involvement of the two genes in bacterial virulence is indicated not only by the virulence attenuation obtained when yhlA and yhlB insertional mutants were used, but also by the cytolytic properties of YhlA against the BF-2 fish cell line (Fernández et al., 2007b). In vivo monitoring of yhlBA promoter showed that it was highly expressed (Méndez and Guijarro, 2013), as was expected, since the system had been previously isolated as an IVET clone. In

contrast to yrp1, the yhlAB operon is present and the haemolytic activity is detectable in different Y. ruckeri strains from different origins and geographical locations. This indicates that this operon is of importance for virulence in this species (Fernández et al., 2007b).

The analysis of a chromosomally located traHIJKCLMN operon of Y. ruckeri that was identified by IVET showed that it was structurally related to the DNA transfer system present in the pADAP virulence-related plasmid of Serratia entomophila (Hurst et al., 2003). The cluster was present in different Y. ruckeri strains but it was not found in the genomic analysis of human pathogenic Yersinia (Méndez et al., 2009). This operon is composed of at least eight genes displaying homology with type IV secretion systems (T4SS). Indeed, traH, traI, traJ, and traK genes encoded for proteins similar to those encoded by the dot/icm genes of the T4SS of Legionella pneumophila (Komano et al., 2000). Thus, it is likely that the gene products of the tra operon constitute parts of the T4SS involved in the transfer of virulence factors. In fact, insertional mutagenesis of the traI gene resulted in attenuation of the virulence (Méndez et al., 2009). Like other virulence factors, the tra operon was more highly expressed at 18◦C than at 28◦C.

A novel two-gene operon cdsBA involved in the transport and further degradation of L-cysteine was also selected by IVET. Attenuation of a cdsA mutant supported a role in virulence (Méndez et al., 2011). The cdsB gene encoded an Lcysteine desulfidase involved in cysteine degradation, previously characterized in the archaeal Methanocaldococcus jannaschii (Tchong et al., 2005), whereas cdsA is a cysteine permease (Méndez et al., 2011). Interestingly, the operon was shown to be present in several anaerobic and facultative bacterial groups, particularly in some species of enterobacteria, including Y. enterocolitica, although it was absent in Y. pseudotuberculosis and Y. pestis (Méndez et al., 2011). This operon was specifically induced by L-cysteine under low oxygen conditions and the uptake of this amino acid was an energy-dependent process (Méndez et al., 2011). Transcriptional fusion of the cdsBA promoter with the luxCDABE operon showed that the cdsA and cdsB genes were barely expressed under routine laboratory conditions but were found to have high activity inside the fish. Therefore, virulence attenuation of the cdsA mutant underlines the importance of this operon for the bacterium during the infection process. Although the precise role of the CdsAB system remains unknown, it was suggested that it could be involved in the generation of iron-sulfur-centers for FE-S proteins or in the accumulation of glutathione, an important detoxification and redox buffer molecule in the cell (Méndez et al., 2011).

Interestingly, more of these IVET-selected genes are induced at 18◦C, the temperature around which outbreaks of the disease take place, than at 28◦C, the optimal temperature for growth of the bacterium (Fernández et al., 2007b; Méndez et al., 2009). This finding points to the existence in Y. ruckeri of specific mechanisms (thermo-induced changes of DNA supercoiling, RNA and protein thermometer, etc.) for the temperature regulation of virulence at temperatures below the optimal for its growth, as seems to occur in a good number of microorganisms that infect ectothermic animals (Guijarro et al., 2015).

STM was also able to select Y. ruckeri mutants that can survive in vitro but not in the fish host (Dahiya and Stevenson, 2010a). Further identification of the inactivated genes in these mutants revealed up to 25 different ORFs (Dahiya and Stevenson, 2010a), among which the znuA gene, coding a zinc binding protein, and the uvrY gene, coding a response regulator, were chosen for deeper study. The znuA gene forms part of the znuABC operon encoding a high-affinity zinc transporter, which has a genetic organization similar to that found in E. coli (Patzer and Hantke, 2000) and S. Typhimurium (Campoy et al., 2002). In fact, the introduction of the Y. ruckeri znuABC locus was able to restore the growth of a znuABC mutant of E. coli in zinc-deficient media (Dahiya and Stevenson, 2010b). The znuABC system is widely distributed in bacteria such as Brucella abortus (Kim et al., 2004), Campylobacter jejuni (Davis et al., 2009) and Neisseria gonorrhoeae (Lim et al., 2008), contributing to their virulence. Interestingly, mutations in the znuABC system of Y. pestis did not affect virulence (Bobrov et al., 2014, 2017). The reason is that this bacterium has an additional zinc transport system mediated by some components of the yersiniabactin siderophore (Ybt) which is able to sequestrate zinc in addition to iron (Bobrov et al., 2014, 2017). In mouse models of bubonic and pneumonic plague this system compensates for the loss of znuABC. However, the simultaneous absence of both systems leads to a decrease in Y. pestis virulence (Bobrov et al., 2014, 2017). However, in Y. ruckeri it seems that the znuABC system is the main Zn acquisition system, since the znuA mutant had nearly 150- to 350-fold lower infection loads in the kidney of rainbow trout than the parental strain, which demonstrates that this Zn transport system has an important role in the infection process (Dahiya and Stevenson, 2010b).

In another mutant selected by STM, the BarA-UvrY twocomponent system was inactivated. This mutant was recovered from fish in 5- to 15-fold lower numbers compared to the parental strain at the beginning of the infection (Dahiya and Stevenson, 2010c), suggesting that this system is involved in the invasion of gills or gut tissues, as occurred in the uvrY mutants of Y. pseudotuberculosis (Heroven et al., 2008), E. coli (Herren et al., 2006) and S. Typhimurium (Johnston et al., 1996). In addition, the Y. ruckeri uvrY mutant was more sensitive to H2O2, which suggests an increasing susceptibility to oxidative killing by phagocytic cells. All of this resulted in a lower recovery (10<sup>3</sup> - to 10<sup>5</sup> -fold) of the uvrY mutant than the parental strain at the end of co-infection experiments which were carried out by immersion of rainbow trout in water containing both strains (Dahiya and Stevenson, 2010c).

As expected, LPS is important for the pathogenesis of Y. ruckeri (Altinok et al., 2016). Specific deletion of the lpxD gene, involved in the lipid A biosynthesis, resulted in an LPS-deficient strain. This lpxD mutant showed a significant attenuation of virulence (10<sup>3</sup> -fold) in relation to the parental strain when LD<sup>50</sup> determination was carried out by injection (Altinok et al., 2016). It is important to highlight that lpxD mutation reduced the overall number of invasive bacteria, which in turn, may influence the process of invasion and colonization, leading to a decrease in virulence. Special attention must be paid to a virulence factor related to legionaminic acid, a component of the Y. ruckeri LPS that will be addressed in the next section.

Another interesting virulence factor that has been described in Y. ruckeri is the Afp18 toxin, although it should be noted that its effect has hitherto only been tested in zebrafish embryos (Jank et al., 2014). This is encoded by a gene that is part of the antifeeding prophage gene cluster (afp), consisting in a 25 Kb DNA segment involved in the production of a prophage contractile tail. This cluster is similar to the one found in Serratia entomophila (Rybakova et al., 2013; Jank et al., 2014), which has common characteristics with the R-pyocins type VI secretion system delivery apparatus of Pseudomonas aeruginosa (Heymann et al., 2013). This contractile tail is thought to capture the Afp18 toxin and inject it into the cell cytoplasm of zebrafish blastoderm cells (Jank et al., 2014). The Afp18 protein has a glycosyltransferase domain. Its activity alters the early phase of zebrafish embryogenesis, whereas a non-functional Afp18 lacking the glycosyltransferase domain has no effect on development (Jank et al., 2014). The effect of the Afp18 glycosyltransferase is through a process of specific tyrosine mono-O-GlcNAcylation of RhoA, a regulatory GTPase associated with cytoskeleton development, resulting in actin depolymerization and the abrogation of early zebrafish development (Jank et al., 2014). Interestingly, genome analysis indicates that this antifeeding prophage system is widely distributed among prokaryotes and archaea (Sarris et al., 2014). In S. entomophila the Afp18 toxin acts as a virulence factor during the infection of insects (Rybakova et al., 2013), and could also be important as a virulence factor in human pathogenic bacteria. Since all this work was carried out on zebrafish embryos, it could be interesting to study the role of Afp18 on Y. ruckeri infection of rainbow trout by generating a mutant strain and following the progression of the disease.

Finally, two elements that should be taken into account are the flhDC operon, whose inactivation results in a more virulent strain and the heat sensitive factor (HSF), which has been considered as a virulence factor for a good number of years. The flhDC is a flagellar master operon of Y. ruckeri involved in the regulation of motility and phospholipase secretion, among other processes (Jozwick et al., 2016). In fact, mutation of the flhD gene resulted in significant changes in the bacterial transcriptome (Jozwick et al., 2016). Interestingly, no differences were found between the flhD mutant and the wild type strain when a standard experimental challenge model with mortalities as an endpoint was used. However, there was a significant increase in the flhD mutant density within the spleen in relation to the wild type strain in competition experiments, a characteristic which was reverted in the complemented strain (Jozwick et al., 2016). The result indicated that mutation of the flhDC operon gives a competitive advantage over the parental strain during the infection process. Although it is not known how the mutation in this operon affects bacterial virulence, the result is similar to that found in the S. Typhimurium flhD mutant in the sense that it was more virulent than the parental strain (Schmitt et al., 2001).

It is also worth mentioning that for many years it was thought HSF was required for virulence in Y. ruckeri (Furones et al., 1993). However, this assumption was recently discarded, since HSF was found to be an alkylsulphatase enzyme encoded by the yraS gene, whose mutation did not produce any modification in the virulence of the strain in relation to the wild type (Navais et al., 2014a).

#### PREDICTED VIRULENCE FACTORS DEDUCED FROM PROTEOMIC AND GENOMIC ANALYSIS

During recent years the development of alternative technologies based on proteomic, transcriptomic and genomic analyses has provided new tools to identify potential virulence factors in bacteria. It is important to consider that this type of analysis is based on comparison and only predicts the potential role of some of the genes or proteins identified as virulence factors. In Y. ruckeri, proteomic studies have been carried out under very specific culture conditions. Thus, differentially regulated proteins were identified and quantified in normal and ironlimited media using a label-free, gel-free shotgun proteomic approach (Kumar et al., 2016). Iron is an essential nutrient for bacteria that is difficult to obtain during the infection process, so a limited-iron environment is expected to induce a set of genes, some of which are related to pathogenesis. Under reduced iron availability, the majority of up-regulated proteins in biotype 1 and biotype 2 Y. ruckeri strains were related to iron-binding proteins (YfuA, YiuA and YfeA), iron-binding transporters (HemS and Feo), hemin transport (YhlBA) and TonB-dependent iron binding receptor, the latter probably being involved in the transport of ruckerbactin into the periplasm of cells and siderophore biosynthesis proteins. YfeA and YhlBA systems have previously been described as important virulence factors in Y. pestis (Bearden and Perry, 1999) and Y. ruckeri (Fernández et al., 2007b), respectively. Another up-regulated protein under iron-limited conditions was a superoxide dismutase, involved in protective mechanisms against oxidative stress, and particularly important in intracellular pathogens such as Y. ruckeri. Interestingly, this study revealed no significant differences in protein profile, using liquid chromatography-mass spectrometry analyses, between motile and non-motile Y. ruckeri strains, corresponding with biotype 1 and biotype 2, when they were cultured under iron-limited conditions, indicating that iron availability it is not involved in the differential virulence behavior of the two biotypes (Kumar et al., 2016).

A similar study was conducted in order to establish the differential protein profile between virulent and avirulent strains of Y. ruckeri grown under standard conditions (Kumar et al., 2017). Interestingly, in virulent strains, a total of 16 proteins were shown to be upregulated, including the haemolysis expression regulator HtrA, DegQ proteases, an anti-sigma regulator factor, different transcriptional regulators belonging to the LuxR, AsnC, and PhoP families, RNA-binding protein Hfq, invasion protein Inv, Cu-Zn superoxide dismutase, among others. All of these have been previously described as virulence-related proteins in different bacterial pathogens, which strongly suggests a role for them in the infection process of Y. ruckeri.

Invasins play an essential role in the initial colonization and attachment of Y. enterocolitica and Y. pseudotuberculosis (Chauhan et al., 2016). In Y. ruckeri, invasin (yrInv) and invasinlike molecule (yrllm) are two putative inverse autotransporter genes that, according to qRT-PCR analysis, were upregulated in a medium that mimics the environmental conditions present in the fish host (Wrobel et al., 2018), so they are likely to be involved in the infection process of Y. ruckeri although there are no conclusive data.

from Cascales et al., 2017).

Whole genome analysis of Y. ruckeri strains provides a new approach to identifying putative virulence factors. Indeed, genome analysis of the Y. ruckeri SC09 strain isolated from catfish (Ictalurus punctatus) showed that it harbors a ysa locus containing all the genetic components of a type III secretion system (T3SS) quite similar in gene sequence and genetic structure to the Salmonella enterica pathogenicity island 1 and chromosomally encoded T3SS of Y. enterocolitica 1B (Liu et al., 2016). It should be remembered that T3SS are present only in virulent strains of human pathogenic Yersinia (Chen et al., 2010). For this reason, it is possible that the T3SS of Y. ruckeri contributes to the survival of the bacterium in fish macrophages and helps in the invasion of gill and gut epithelia. Other putative systems in Y. ruckeri SC09 that, according to the genomic analysis carried out by Liu et al. (2016), could be involved in the development of the infection in fish are: a type II secretion system, different two-component signal transduction systems and a phage shock protein system (Psp), which is distributed among enterobacteria and is implicated in the virulence of Salmonella, Shigella and Yersinia species (Huvet et al., 2011). Interestingly, a type IV secretion system described as a putative virulence factor by Liu et al. (2016) is unique to Y. ruckeri SC09 strain and it is not present in the other four Y. ruckeri strains analyzed by Cascales et al. (2017).

The comparative analysis of the genome sequences of five Y. ruckeri strains showed that, although they share approximately 75% of their genes, significant genetic differences were detected depending on the serotype or virulence of these strains (Cascales et al., 2017). Thus, a cluster of 18 genes related to the biosynthesis of legionaminic acid, the major component of the LPS, was exclusively present in serotype O1 strains. This cluster was absent in serotype O2 strains as well as in other Yersinia species, but present in different aquatic pathogenic bacteria such as Vibrio vulnificus, Aeromonas salmonicida, Vibrio fischeri (Cascales et al., 2017) and also in Campylobacter jejuni, where it is involved in virulence (Zebian et al., 2016). In addition to this cluster, serotype O1 strains share a total of 268 genes, among which are the invasin and the type IV secretion system corresponding with the traHIJKCLMN operon (Méndez et al., 2009).

These genes shared by serotype O1 strains could adapt the bacterium to a particular host, since the three serotype O1 strains analyzed (150, CSF007-82, and ATCC29473) were isolated from rainbow trout. In the same way, a serotype O2 strain (Big Creek 74) together with Y. ruckeri SC09, a putative O2 strain, which were isolated from chinook salmon and catfish respectively, share 122 genes which are absent in serotype O1 strains. These include a cluster involved in fimbriae biosynthesis similar to the Stf cluster of S. Typhimurium, associated with virulence (Emmerth et al., 1999) and an insecticidal toxin complex similar to the one found in Vibrio parahaemolyticus, involved in hepatopancreatic necrosis disease in penaeid shrimps (Tang and Lightner, 2014).

In the study of Cascales et al. (2017) it is proposed that the characteristic attenuation in virulence of the ATCC29473 type strain is a consequence of the lack of 21 genes that were present in the other virulent strains. It should be noted that 17 out of these 21 genes are clustered in all four genomes analyzed (**Figure 3**). This region includes genes encoding a Crp-Fnr family transcriptional regulator, enzymes related to enterobactin-like siderophore and gene clusters involved in iron transport, hexose phosphate uptake and citrate metabolism. Most of these genes are related to virulence in different bacteria (Gray et al., 2006; Urbany and Neuhaus, 2008; Moisi et al., 2013) so the absence of this region could be responsible for the attenuation of the type strain.

#### CONCLUSIONS AND FURTHER STUDIES

The results lead to the conclusion that there exist at least three different routes of infection by Y. ruckeri. The most important pathway for the development of ERM has still to be elucidated, but it seems clear that, although the primary bacteremia is generated by the entry of the bacterium into the fish through the gills, the very significant presence of the bacterium in the intestine and also the signs of the disease, strongly suggest that the digestive route must play a relevant role. The virulence of Y. ruckeri is multifactorial and it depends not only on the presence of certain genes in each strain but also on certain environmental factors, particularly, temperature. In fact, a temperature below the optimal for bacterial growth and close to the one at which outbreaks occur seems to be important in the expression of some of the genes encoding virulence factors. This appears to be one of the areas for future research, since presently unknown regulation systems could be involved in this temperature-dependent response. Although a lot of genes involved in the pathogenesis of the bacterium have recently been identified, there are still many points that need to be addressed, such us the temporal and spatial expression pattern of each virulence factor and their specific function or the identification of the regulatory mechanisms which modulate the virulence gene expression. The new techniques and methodologies developed in recent years, including CLIQ-BID, a method allowing quantification of damage in eukaryotic cells infected by bacteria (Wallez et al., 2018), the new RNA-seq techniques such as Term-seq (Dar et al., 2016) and Dual RNA-seq (Westermann et al., 2016), useful for the analysis of pathogens under complex environmental conditions and Single Cell RNAseq (Avraham et al., 2015; Saliba et al., 2017) can help to develop different approaches. Moreover, analysis of the data provided by the genetic and proteomic studies may be a good starting point for elucidating the role of new genes in the infectious process of Y. ruckeri. In fact, two specific gene clusters should be considered for their potential role in virulence. This is the case of a cluster involved in legionaminic acid biosynthesis that could be important in the pathogenesis of serotype O1 strains, and the 21 genes absent from the virulence-attenuated Y. ruckeri type strain, some of them encoding putative virulence factors.

#### AUTHOR CONTRIBUTIONS

JG and JM drafted the first version of the manuscript. DC, JM, AG-T, JG contributed to bibliography analysis and manuscript improvement. All authors have reviewed the final version of the manuscript.

### FUNDING

This research was supported by the AGL2015-66018-R grant from the Ministerio de Economia y Competitividad of Spain.

### REFERENCES


#### ACKNOWLEDGMENTS

DC and AG-T were the recipients of FPU and FPI fellowships, respectively.


a utilization system for carbon and phosphate sources. J. Bacteriol. 195, 1800–1808. doi: 10.1128/JB.01952-12


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Guijarro, García-Torrico, Cascales and Méndez. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# pYR4 From a Norwegian Isolate of *Yersinia ruckeri* Is a Putative Virulence Plasmid Encoding Both a Type IV Pilus and a Type IV Secretion System

#### Agnieszka Wrobel <sup>1</sup> , Claudio Ottoni <sup>2</sup> , Jack C. Leo<sup>1</sup> and Dirk Linke<sup>1</sup> \*

<sup>1</sup> Department of Biosciences, University of Oslo, Oslo, Norway, <sup>2</sup> Centre for Ecological and Evolutionary Synthesis, University of Oslo, Oslo, Norway

#### *Edited by:*

Victoria Auerbuch, University of California, Santa Cruz, United States

#### *Reviewed by:*

Hanh N. Lam, University of California, Santa Cruz, United States Gokhlesh Kumar, Veterinärmedizinische Universität Wien, Austria

> *\*Correspondence:* Dirk Linke dirk.linke@ibv.uio.no

#### *Specialty section:*

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

> *Received:* 06 July 2018 *Accepted:* 04 October 2018 *Published:* 30 October 2018

#### *Citation:*

Wrobel A, Ottoni C, Leo JC and Linke D (2018) pYR4 From a Norwegian Isolate of Yersinia ruckeri Is a Putative Virulence Plasmid Encoding Both a Type IV Pilus and a Type IV Secretion System. Front. Cell. Infect. Microbiol. 8:373. doi: 10.3389/fcimb.2018.00373 Enteric redmouth disease caused by the pathogen Yersinia ruckeri is a significant problem for fish farming around the world. Despite its importance, only a few virulence factors of Y. ruckeri have been identified and studied in detail. Here, we report and analyze the complete DNA sequence of pYR4, a plasmid from a highly pathogenic Norwegian Y. ruckeri isolate, sequenced using PacBio SMRT technology. Like the well-known pYV plasmid of human pathogenic Yersiniae, pYR4 is a member of the IncFII family. Thirty-one percent of the pYR4 sequence is unique compared to other Y. ruckeri plasmids. The unique regions contain, among others genes, a large number of mobile genetic elements and two partitioning systems. The G+C content of pYR4 is higher than that of the Y. ruckeri NVH\_3758 genome, indicating its relatively recent horizontal acquisition. pYR4, as well as the related plasmid pYR3, comprises operons that encode for type IV pili and for a conjugation system (tra). In contrast to other Yersinia plasmids, pYR4 cannot be cured at elevated temperatures. Our study highlights the power of PacBio sequencing technology for identifying mis-assembled segments of genomic sequences. Comparative analysis of pYR4 and other Y. ruckeri plasmids and genomes, which were sequenced by second and the third generation sequencing technologies, showed errors in second generation sequencing assemblies. Specifically, in the Y. ruckeri 150 and Y. ruckeri ATCC29473 genome assemblies, we mapped the entire pYR3 plasmid sequence. Placing plasmid sequences on the chromosome can result in erroneous biological conclusions. Thus, PacBio sequencing or similar long-read methods should always be preferred for de novo genome sequencing. As the tra operons of pYR3, although misplaced on the chromosome during the genome assembly process, were demonstrated to have an effect on virulence, and type IV pili are virulence factors in many bacteria, we suggest that pYR4 directly contributes to Y. ruckeri virulence.

Keywords: *tra* operon, *pil* operon, conjugative plasmid, *Yersinia ruckeri*, type IV secretion system

## INTRODUCTION

The genus Yersinia consists of 17 different species (Reuter et al., 2014; Savin et al., 2014). Although the human pathogens within the genus are closely related to each other, they cause diverse diseases. Y. pestis, the causative agent of bubonic and pneumonic plague, is one of the most virulent organisms known (Chauhan et al., 2016). In addition, this genus includes Y. enterocolitica and Y. pseudotuberculosis, well-known human enteropathogens. Y. pestis spreads through fleabites or aerosols, whereas Y. enterocolitica and Y. pseudotuberculosis are transmitted via ingestion of contaminated food or water (Bottone, 1997; Perry and Fetherston, 1997; Jalava et al., 2006). Y. enterocolitica and Y. pseudotuberculosis are responsible for a broad range of diseases ranging from mild gastroenteritis to life-threatening septicemia (Bottone, 1997).

Y. ruckeri is a fish pathogen causing enteric redmouth disease (ERM), mainly in salmonids (Bullock et al., 1978; Busch, 1978). This bacterium contributes to enormous economic losses in aquaculture throughout the world. Y. ruckeri is mostly transmitted through contact with carrier fish (Busch, 1978; Stevenson and Airdrie, 1984). Despite the availability of vaccines, yersiniosis outbreaks still occur in fish farms (Ormsby et al., 2016). The majority of the ERM outbreaks are caused by the highly pathogenic Y. ruckeri serotype 1 belonging to biotype 1, characterized as motile with phospholipase activity (Romalde and Toranzo, 1993). For a long time, ERM has played a minor role in Norway, with only a few outbreaks per year (Hjeltnes et al., 2017). The first report of a disease outbreak caused by Y. ruckeri among Atlantic salmon was described in Norway in 1985, and this was successfully treated with antibiotics (Sparboe et al., 1986). In recent years, the number of outbreaks in the farmed Atlantic salmon population has substantially increased. The reasons for the most recent outbreaks remain unclear, and Y. ruckeri infections are nowadays a major challenge facing the Norwegian aquaculture industry, similar to other countries such as Australia (Barnes et al., 2016), Chile (Avendaño-Herrera et al., 2017), and Scotland (Ormsby et al., 2016).

Each of the human Yersinia pathogens harbors chromosomally and plasmid-encoded virulence determinants (Chauhan et al., 2016). Y. pestis usually carries two speciesspecific plasmids, pPCP1 and pMT1, and one highly conserved plasmid shared among the three human pathogenic Yersiniae, pYV (also called pCD1) (Ben-Gurion and Shafferman, 1981; Ferber and Brubaker, 1981; Haiko et al., 2009). This large 70-kb plasmid carries a type III secretion system (T3SS), Ysc. T3SS system encodes structural proteins, chaperones as well as effector proteins called Yops (Yersinia outer proteins) required for Yersinia extracellular survival. The effector proteins and the machinery for their delivery are required for infection and manipulation of host responses to overcome the action of phagocytes (Cornelis et al., 1998). Moreover, the plasmid encodes a major virulence factor, the Yersinia adhesin A (YadA) (Mühlenkamp et al., 2015).

Despite the economic importance, the pathogenicity of Y. ruckeri has not been studied in detail. Only few virulence factors are known, and to date all of these are encoded on the chromosome. These include bacterial adhesins important in establishing a successful colonization (Romalde and Toranzo, 1993). In particular, the chromosomally encoded adhesins YrInv and YrIlm might play a role in virulence (Wrobel et al., 2017). They belong to the intimin-invasin family of adhesins, which includes also InvA, the adhesin responsible for the initial bacterial attachment and colonization of host tissues in Y. enterocolitica and Y. pseudotuberculosis (Isberg and Leong, 1990; Wrobel et al., 2017). Other virulence factors described in Y. ruckeri include cytotoxins and haemolysins (Romalde and Toranzo, 1993), the metalloprotease Yrp1 (Secades and Guijarro, 1999), the haemolysin/cytolysin YhlA (Fernández et al., 2007), the iron uptake system ruckerbactin (Fernández et al., 2004), and a chromosomal T3SS (Liu et al., 2016). Recently, a large proteomic study of Y. ruckeri strains was performed under standard (Kumar et al., 2017) and iron-limited conditions (Kumar et al., 2016). In total, 1395 proteins were identified in the whole cell lysate of Y. ruckeri under standard culture conditions. Among them, several proteins were predicted to be virulence factors, including, among others, HtrA protease, TolB, the lipoprotein NlpD and a LuxR family transcriptional regulator. This global proteomic analysis will help in understanding the biology of the pathogen, as well as in development of new effective treatments against the ERM disease (Kumar et al., 2017).

Plasmid-borne virulence factors have been found in other fish pathogens, including Vibrio anguillarium (Crosa, 1980) and Edwardsiella tarda (Yu et al., 2012), but not in Y. ruckeri. Plasmids in Y. ruckeri strains were studied previously due to their possible involvement in virulence in analogy to the human pathogenic Yersiniae (De Grandis and Stevenson, 1982). Many authors expected to find the same virulence traits as those described for the human pathogens, such as the plasmid-encoded T3SS. However, none of the plasmid-associated virulence factors of the human-enteropathogenic Yersiniae were found in these plasmids. In general, Y. ruckeri plasmids have not yet been properly characterized and further research is required to understand their role in bacterial virulence. A study including 183 Y. ruckeri strains from different geographical locations reported 8 different plasmid profiles (Garcia et al., 1998). In this study, the most virulent sorbitol-negative Y. ruckeri strains of serotype O1 contained a large 75 MDa plasmid (∼113 kb), in agreement with earlier studies (Guilvout et al., 1988; Romalde et al., 1993). In addition, smaller plasmids (12.7 MDa; ∼19 kb) have been found in most of the strains (Garcia et al., 1998).

More recent studies showed that multidrug resistance plasmids in Y. ruckeri strains are a serious aquaculture concern (Toranzo et al., 1983; De Grandis and Stevenson, 1985; Carattoli et al., 2012; Huang et al., 2014). Welch et al. (2007) showed that Y. ruckeri strain YR71 carries a multidrug resistance plasmid called pYR1, which has 99% nucleotide identity with the IncA/C (incompatibility A/C) plasmid backbone of the Y. pestis isolate IP275, plasmid pIP1202. The IncA/C group comprises a large, low-copy number, multidrug resistance plasmid family within Enterobacteriaceae such as Escherichia coli, Salmonella enterica, Y. pestis, and Klebsiella pneumoniae, as well as more distantly related species such as Vibrio cholerae. Plasmids of this family are unique with regard to their structure and gene content. They contain putative transfer regions [type IV secretion system (T4SS)], regions involved in integration of mobile genetic elements, as well as regions involved in transcription (Johnson and Lang, 2012). T4SSs are widely distributed in prokaryotes as well as in some archaea. T4SSs are large macromolecular complexes typically composed of a cell-envelope spanning mating channel and an extracellular pilus structure. T4SSs are classified into two major groups type IVA (T4ASSs) and type IVB (T4BSSs). T4ASS resemble the VirB/VirD system of Agrobacterium tumefaciens while T4BSSs are related to the conjugation system of IncI plasmids. Typical examples of T4ASSs are found on conjugative plasmids, such as F, RP4 and pKM101, as well as the prototypical VirB system of A. tumefaciens. These T4ASSs export nucleoprotein complexes during conjugation. T4BSS is represented by the Legionella pneumophila icm/dot system involved in protein translocation into host cells thus allowing the pathogen to replicate intracellularly (Wallden et al., 2010).

In this work, we sequenced a plasmid—which we named pYR4—from the highly pathogenic Norwegian Y. ruckeri isolate NVH\_3758 from the 1987 outbreak and performed a comparative bioinformatics analysis of the available Y. ruckeri plasmid sequences to evaluate their role in virulence.

### MATERIALS AND METHODS

#### Plasmid DNA Sequencing Technology

Genomic DNA as well as plasmid DNA was extracted from a locally important, highly pathogenic Norwegian Y. ruckeri isolate NVH\_3758 (biotype 1, serotype 1) recovered from an outbreak of clinical yersiniosis in farmed Atlantic salmon, kindly provided by Prof. Duncan Colquhoun at the Norwegian Veterinary Institute in Oslo, Norway (Gulla et al., 2018). Whole genome sequencing of Y. ruckeri NVH\_3758 was performed by the Norwegian Sequencing Centre (Oslo, Norway) using the Single Molecule Real Time (SMRT) sequencing technology of Pacific Biosciences. Sample preparation, reads assembly and consensus polishing were done as previously described (Wrobel et al., 2017). The final assembly yielded two contigs of circularized length of ∼3.8 Mb, representing the chromosomal genome (Wrobel et al., 2017), and ∼81 kb, corresponding to a new plasmid that we named pYR4. The DNA sequence of pYR4 has been deposited in the National Centre for Biotechnology Information (NCBI) database under the accession number CP032236.

#### Plasmid Annotation

The FASTA consensus of pYR4 from Y. ruckeri NVH\_3758 was uploaded to RAST (Rapid Annotation using Subsystem Technology) for automatic annotation (Aziz et al., 2008; Seemann, 2014). After initial annotations, all open reading frames (ORFs) with initial annotations were checked using the interactive server HHpred available at the Max Planck Institute for Developmental Biology Toolkit (Söding et al., 2005) against two databases, the PDB and PFAM (Sonnhammer et al., 1998; Sussman et al., 1999). The functional annotations obtained from the HHpred server and RAST were compared and in some cases were corrected manually. Many uncharacterized proteins which were previously labeled as hypothetical by RAST were annotated based on similarity to characterized proteins. The protein sequences were uploaded into Geneious (Kearse et al., 2012). A circular representation of pYR4 showing the annotated features, the GC content and the GC skew within 50 bp-long genomic regions, was generated with Circos (Krzywinski et al., 2009). The identification of the promoter sequences of the pil and tra operons was performed with the online server BPROM (Solovyev and Salamov, 2011) as well as bTSSfinder (Shahmuradov et al., 2017) (see **Table 1** and **Supplementary Figure 1**). The mfold Web server was used for RNA secondary structure prediction (Zuker, 2003).

#### RepA Phylogeny

The RepA protein sequence of pYR4 was annotated by RAST as "hypothetical." After the initial annotation, the protein was identified as RepA using HHpred. The protein sequence was then subjected to a search using BLASTP (Altschul et al., 1997). The BLASTP search returned 100 hits, from which the first 29 RepA protein sequences were selected, after excluding sequences of hypothetical proteins and multispecies proteins, and aligned using MUSCLE (Edgar, 2004). In the final alignment, we included RepA protein sequences from pYR1, pYR3, pYR4 in addition to eight RepA protein sequences belonging to the IncA/C plasmid family. The final alignment was then used to construct the phylogenetic tree using MEGA X software by applying the Maximum Likelihood method on the Poisson correction model (Zuckerkandl and Pauling, 1965; Felsenstein, 1985; Kumar et al., 2018) (see **Supplementary Figure 2**).

#### Plasmid Comparative Analysis

We compared the nucleotide sequence of Y. ruckeri NVH\_3758 plasmid pYR4 with the nucleotide sequences of the plasmids of Y. ruckeri strains YR71 (pYR1), CSF007-82 (pYR2, pYR3) and SC09 (pLT, pWKY) (**Table 2**) deposited in the NCBI database. To keep the same annotation system as for the pYR4 plasmid, the nucleotide sequences of the Y. ruckeri plasmids were reannotated with RAST (see also **Supplementary Figure 3** for pYR4, **Supplementary Figure 4**). Details of the annotation can be found in **Supplementary Tables 1** and **2**.

Plasmid comparisons were also done with sequences of other species containing the tra and pil operons described in literature, including Erwina amylovora (pEL60, pEA68, pEA72, pEA78), Serratia entomophila (pADAP), Citrobacter freundii (pCTX-M3), and Salmonella enterica subsp. enterica serovar Typhimurium (R64) (see **Supplementary Figure 5**). The nucleotide sequence of Y. ruckeri NVH\_3758 pYR4 plasmid was also compared to PacBio-sequenced genomic data of Y. ruckeri CSF007-82, Big Creek 74, QMA0440, SC09, and Illumina-sequenced genomes of Y. ruckeri ATCC29473 and YRB (see **Supplementary Figure 5**). In the comparative survey, we also included the ∼57 kb-long scaffold 20 of the Y. ruckeri 150 assembly, which contains the tra and pil operons.

Finally, pYR4 was compared to the Illumina-sequenced genomes of human pathogens Y. pestis CO92, Y. pseudotuberculosis YPIII, and Y. enterocolitica 8081 (see **Supplementary Figure 5**). Pairwise comparisons were


TABLE 1 | pYR4 promoter predictions by bTSSfinder (Shahmuradov et al., 2017) and BPROM (Solovyev and Salamov, 2011) used in the present study.

<sup>a</sup>A score of 0.81 or higher is considered significant for bTSSfinder, and a score of 0.2 or higher is considered significant for BPROM (Solovyev and Salamov, 2011; Shahmuradov et al., 2017). Two score values provided in the table were predicted for BPROM which correspond to the Pribnow box at the −10 position and at the −35 position, respectively.

TABLE 2 | Y. ruckeri plasmid sequences deposited in GenBank used in the present study.


performed with Progressive Mauve (Darling et al., 2010) using default options and the "seed family" option to increase sensitivity. The output backbone file was then used to plot the Locally Collinear Blocks (LCB) in a circular representation with Circos.

#### RESULTS

### pYR4 Is a Novel Plasmid Isolated From *Y. ruckeri* NHV3758

In order to define the relationship between the plasmid sequenced in this study and those described in literature, we performed a comparative analysis of pYR4 with plasmid sequences generated previously by Illumina and PacBio sequencing technologies. The comparative survey with Mauve indicated no obvious similarity between pYR4 and the Y. ruckeri plasmids pYR1, pYR2, pLT, and pWKY, as no LCBs (locally collinear blocks) were detected (data not shown). On the other hand, a ∼55 kb-long LCB (sequence identity >99%) that included the pil and the tra operons was present in pYR4 (from nucleotide position 9,100 to 64,005) and in the PacBio-sequenced plasmid pYR3 (**Figures 1**, **3**, **Table 2**). By re-annotating pYR3 and comparing it with the higher-resolution annotation of

pYR4 obtained through HHPred, we could provide a more in depth characterization of the plasmids under analysis (Nelson et al., 2015) (see Materials and Methods section, **Supplementary Figure 3** for pYR4 and **Supplementary Figure 4** for pYR3).

The remaining portion of the plasmid sequence (>25 kb, 31% of the sequence length) appears to be unique, as no LCB was found in any of the plasmids of Y. ruckeri deposited in GenBank so far. This region contains mostly hypothetical proteins (n = 10) and mobile genetic elements (n = 10). In addition, we found a partial toxin-antitoxin system, a restriction system including both a type I restriction enzyme and a corresponding ArdA-like anti-restriction protein, and a small cluster of genes coding for two alcohol dehydrogenase enzymes and a transcription factor with high similarity to FrmR from Salmonella, a formaldehyde-sensitive regulator (**Supplementary Figure 3**). pYR4 contains two potential partitioning systems (ParAG and StbAB) (**Figure 1**). These two partitioning systems are represented by two operons oriented in opposite directions. No obvious sequence similarity with the partitioning system of pYR3 was found. Interestingly, the presence of alternative partitioning systems have already been described before for pYV from Yersinia species (Pilla and Tang, 2018). The high abundance of mobile elements may suggest that this plasmid is likely subject to structural rearrangements and that the unique ∼25 kb region of pYR4 may be the result of recent horizontal gene transfer. This is also supported by the difference in G+C content between the ∼55 kb region (50.4%) and the remaining portion of the plasmid (47.1%). It is worth noting that in pYR3, this complete region is replaced by a different ∼45 kb region (**Figure 1**). These major differences suggest that Y. ruckeri NHV3758 contains a plasmid with significant differences to pYR3, which we named pYR4.

#### Sequence Analysis of pYR4

The nucleotide sequence of the circular pYR4 plasmid contains 80,842 base pairs (∼53 MDa). The G+C content of pYR4 is 49.4%, which is around 2% less than the G+C content of the Y. ruckeri NVH\_3758 genome (47.6%), suggesting acquisition by horizontal gene transfer (**Figure 2**) (Nishida, 2012; Hayek, 2013). The annotation of pYR4 with RAST showed 92 putative coding sequences along the entire plasmid sequence. The RAST server could annotate functions for 52 ORFs and we were able to expand this list to 71 ORFs manually, using the HHpred server (Söding et al., 2005) (see **Supplementary Figure 3**), leaving 21 ORFs without putative function. Fifty-five genes are encoded on the positive strand while the remaining 37 are encoded on the negative strand. The entire plasmid sequence can be divided into several gene clusters, including clusters for partitioning (parA, parG, and stbAB), a T4SS (tra), and a type IV pilus (TFP) gene cluster (pil).

green.

In order to understand the pYR4 plasmid physiology as well as to follow its evolution and spread, we classified pYR4 into a plasmid family. A classical method of plasmid classification is based on incompatibility (Inc) groups. In general, plasmids with the same replication system are incompatible while plasmids with different replication system are compatible. In other words, two plasmids of the same Inc group cannot be propagated in the same bacterial cell (Couturier et al., 1988). A set of 30 RepA protein sequences from IncFII plasmid family, together with 9 RepA protein sequences previously characterized as belonging to IncA/C, were aligned in order to classify pYR4. Evaluation of the RepA phylogeny showed that the pYR4 RepA protein is closely related to IncFII plasmids found in other Yersiniae species (see **Supplementary Figure 2**). A BLASTP search of pYR4 RepA protein returned over 100 hits of homologs found in different species. The closest RepA homologs were found in Illumina-sequenced genomes of Y. frederiksenii and Y. enterocolitica with 91 and 90% similarity over the whole protein sequence, respectively. These RepA homologs presumably are part of unnamed plasmids that were incorrectly assigned to chromosomes, since they share only 19% similarity to RepA of the well-described virulence plasmid pYV from Y. enterocolitica 8081 (**Table 3**). Thus, pYR4 was classified as a member of the IncFII plasmid family, in contrast to pYR1 which belongs to the IncA/C family and is represented as an outgroup in **Supplementary Figure 2** (Carattoli, 2009). The IncFII plasmid family includes low-copy number plasmids mostly related to virulence, such as pYV, as well as to the dissemination of antimicrobial resistance determinants. Plasmids from this family usually carry the FII replicon alone or in association with extra replicons such as repFIA and repFIB (Carattoli, 2009; Yang et al., 2015) and are common in Yersiniae (Villa et al., 2010).

#### Comparative Analysis of *Y. ruckeri* Plasmid Sequences Demonstrates Errors in Assemblies of Second Generation Genome Sequencing

Our comparative survey showed that no significant LCBs were found between pYR4 and the chromosomal genomes of Y. ruckeri CSF007-82, Big Creek 74, QMA0440, SC09, or YRB (data not shown). However, we found that the ∼100 kb-long scaffold 2 of the Y. ruckeri ATCC29473 Illumina assembly matched pYR3, except for a mobile element of pYR3 (**Figure 3** and **Supplementary Figures 4–6**). Furthermore, the higher quality of PacBio sequenced plasmids (pYR3 of CSF007-82 and pYR4 of NHV-3758) made it possible in our comparative survey to place scaffolds of previous Y. ruckeri assemblies into plasmid locations. In fact, the ∼57 kb-long Y. ruckeri 150 scaffold 20 that contained the tra and the pil operons could be mapped entirely to pYR3 (see **Supplementary Figure 7**). This scaffold included the ∼55 kb-long region containing the pil and the tra operons detected in pYR4 and pYR3. Furthermore, by aligning other unplaced scaffolds of the Y. ruckeri 150 assembly, we found that four more scaffolds (23, 31, 32, 34) could be placed within pYR3 (**Figure 3** and **Supplementary Figures 5**, **6**). (An unplaced scaffold is a sequence found in an assembly that is not associated with any chromosome). Taken together, the evidence presented here suggests that the pil and tra operons are localized on plasmids pYR3 and pYR4 and that Y. ruckeri 150 and Y. ruckeri ATCC29473 contain the plasmid pYR3. In Y. ruckeri 150, the presence of plasmid- and chromosomally-borne tra clusters has been suggested based on Southern blot hybridization evidence (Méndez et al., 2009). When searching for tra genes in the assembly, we could not find copies of tra genes other than those matching pYR3 in the scaffold 20 of Y. ruckeri 150. Based on our data, chromosomal localization of the tra cluster seems very unlikely. However, resequencing or a higher quality assembly of the genome of Y. ruckeri 150 could clarify this unambiguously in the future.

Finally, no significant LCBs were found when comparing pYR4 or pYR3 to Yersinia human pathogenic species, including the well-studied pYV plasmid from Y. enterocolitica. This suggests a very different strategy for host infection in Y. ruckeri, as the pYV plasmid is essential for virulence in Y. enterocolitica.

#### Sequence Analysis of *pil* Operon and Its Potential Involvement in Virulence

Analysis of the pYR4 nucleotide sequence of a putative pil operon showed that this region spans a 12.6 kb locus containing 17 ORFs that encode a TFP. TFPs, not to be confused with T4SSs, are surface appendages expressed by many Gramnegative bacterial species. TFPs span both bacterial membranes and they are evolutionary and structurally related to type II secretion systems. TFPs are involved in bacterial adhesion, biofilm formation, horizontal gene transfer, and pathogenesis, and in addition they mediate cell movement such as gliding motility in Myxococcus xanthus and twitching motility in Pseudomonas and Neisseria species (Shi and Sun, 2002). In the enteropathogenic Y. pseudotuberculosis, the TFP gene cluster is composed of 11 open reading frames and contributes to Y. pseudotuberculosis pathogenicity (Collyn et al., 2002). The arrangement of the pYR4 pil cluster (**Figure 2**; see also **Table 1** and **Supplementary Figure 1**) resembles the pil cluster from the plasmid pADAP, which was described in S. entomophila (Hurst et al., 2011), and in plasmid R64 from S. enterica serovar Typhimurium (Kim and Komano, 1997). By analogy with the pil operon from S. entomophila, we adopted the same names for the putative proteins as described there, and designated them as PilLMNNNOOPQRSTDUV [(with the exception of the first two hypothetical proteins, designated as HP (H- for hypothetical and P- for proteins)]. It is worth mentioning that the pil operon in S. enterica includes only 14 genes (pilIJKLMNOPQRSTUV) in contrast to pYR4 (17 genes). The number of genes for pil clusters can vary, as described by Zhang et al. for S. enterica serovar Typhi, where the pil operon lacks the pilI, pilJ, and pilK genes (Zhang et al., 2000) of S. entomophila. The overall G+C content of the pil cluster is 51.7%, and thus higher than that of the Y. ruckeri NVH\_3758 genome (47.6%) and the average of pYR4 (49.4%). The G+C content of the individual genes in the pil locus varies between 55.1% (pilP) and 46.8% (pilN). The ORFs of the pil cluster are encoded on the same strand and the length of the intercistronic region between each ORF ranges from 20 to 205 bp. In the region up to 333 bp upstream of the first hypothetical protein of the pil cluster, we could identify binding sites for three sigma factors (σ70, σ24, and σ38) using BPROM and bTSSfinder, indicating the presence of putative promoters sequences in that region. In fact, no putative promoter sequence was identified between the pil genes suggesting that this region may function as an operon (**Figure 2**) (see **Table 1** and **Supplementary Figure 1**). Additionally, analysis of the downstream region of pilV showed the presence of a palindromic sequence (5′**CTAGACAGAAT**AGCCTAGTC AATATTATCTATGGCATTAAG**ATTCTGTCAG**-3′ ) that could serve as a transcription terminator. The analysis of the secondary structure of this region showed a steam-loop like fold with the 1G = −15 kcal/mol using the mfold Web server (Zuker, 2003).

The comparison of the protein sequences encoded by the pil operon, for example PilO and PilT, of pYR4 with database sequences using BLASTP showed from 64 to 86% identity to PilO and PilT proteins found in the IncI1 plasmid family from Serratia species. Accession numbers for the proteins, together with their functions, are given in **Table 4**.

The biogenesis of TFPs involves a number of proteins. These are all present in the pil operon of pYR4, suggesting that the locus is intact and functional. The pYR4 PilS protein encodes a major pilin, which is synthesized as a prePilS. In the prePilS protein sequence we identified a hydrophilic signal peptide comprising 15 residues. A predicted cleavage site lies between the 15th (glycine) and 16th (tryptophan) residues of prePilS, which is recognized by the specific peptidase PilD (Kim and Komano, 1997). The mature PilS contains an N-terminal hydrophobic region (first 23 residues in the mature protein sequence), while


the C-terminal region is rich in cysteine residues, a common feature of TFP pilins (Hurst et al., 2011). Beside the prepilins, we identified two copies of PilO and three copies of PilN, which seems to be a unique feature among plasmids from the same family. PilO and PilN are integral membrane proteins and usually exist only in one copy. An ATPase required for the assembly (PilQ) and an inner membrane protein (PilR) that we identified are also necessary for the assembly of the pili on the bacterial surface.

#### Sequence Analysis of the *tra* Regions in pYR4

Annotations of pYR4 by HHpred (Söding et al., 2005) showed the presence of the tra region that we presume encodes a T4SS and is involved in conjugation. In fact, the presence of a chromosomally-borne tra clusters in Y. ruckeri 150 was previously described (Méndez et al., 2009). In our analysis, in addition to the tra cluster identified by Méndez et al. (tra1) that is probably also plasmid-borne (see above), we could identify another two tra clusters, which we named tra2 and tra3. The tra2 cluster comprises 10 genes with the gene order TraNOPQRTUWXY and an average G+C content of 53.5%, while the tra3 cluster is composed of 3 genes with a G+C content of 52.6%. These two tra clusters are preceded by two putative promoter sequences with one located upstream from the traN gene while another one is located upstream from the trbA gene. The identification of the putative promoters sequences were based on the prediction of sigma factor binding sites using BPROM and bTSSfinder (see **Table 1** and **Supplementary Figure 1**). The presence of the two identified putative promoter sequences and the small intercistronic region between the genes suggests that these genes might function as two operons, in addition to the tra1 operon (**Figures 1**, **2**). The genetic organization of the tra2 and tra3 operons resembles the gene order of the tra operon found in the pADAP plasmid of S. entomophila, the R64 plasmid of S. enterica serovar Typhimurium, pCTX-M3 of C. freundii, pEL60, pEA68, pEA72, and pEA78 of E. amylovora (see **Supplementary Figure 5**), as previously suggested for the tra1 operon (Méndez et al., 2009).

The G+C content of the tra2 and tra3 operons (around 53%) differs from the G+C content of the chromosomes of Y. ruckeri NVH\_3758 (47.6%), Y. ruckeri Big Creek 74 (47.6%), and Y. ruckeri CSF007-82 (47.5%). Additionally, the tra region was is not present in the chromosome of the Y. ruckeri strains mentioned above, indicating that the tra region may originate from another species. Méndez et al. suggested that S. entomophila, the causative agent of amber disease of the New Zealand grass grub, could be the source of the tra1 region. The G+C content of the S. entomophila pADAP tra region (tra1, tra2, tra3) is around 52%, which is close to the G+C content of pYR4. In addition, the gene order of that region is very similar. We suggest that the whole tra region encompassing tra1, tra2, and tra3 could have been acquired from this or a closely related Serratia species.

The amino acid sequences of TraH, TraI, TraJ, and TraK showed 29–34% similarity to the L. pneumophila T4BSS proteins such as DotD (for "defect in organelle trafficking"), DotC, DotB, and IcmT (for "intracellular multiplication") (Wallden et al., 2010). The icm/dot genes are required for virulence, including intracellular growth and host cell killing (Sadosky et al., 1993; Swanson and Isberg, 1996). The pYR4 Tra proteins and their Tra homologs from L. pneumophila are similar in size, ranging from 87 residues (for TraK) to 385 residues (for TraJ). Interestingly, it has been shown before that a traI mutant strain of Y. ruckeri 150 was attenuated in an in vivo assay in rainbow trout and showed difficulty growing inside the fish (Méndez et al., 2009).

#### Plasmid Curing Study of pYR4 in *Yersinia ruckeri* NVH\_3758

In order to understand the function of pYR4 and its involvement in pathogenesis, it is desirable to obtain a plasmid-cured strain. There is a wide number of plasmid curing procedures which have been successfully used to remove plasmids in Yersiniae. They for example include treatments with high temperatures or introduction of an incompatible plasmid (Sheridan et al., 1998; Ni et al., 2008). The genetic stability of the pYR4 plasmid was tested by treating Y. ruckeri cells with high temperature. Y. ruckeri NVH\_3758 was grown in LB medium for 10 consecutive days at 37◦C with dilutions each day (Trevors, 1986). After 10 days, colony PCR was performed on 10 clones using pil

primers that could only bind to the plasmid sequence. All tested clones were PCR-positive for the pYR4 plasmid (data not shown). Based on our high quality genomic assembly, the pil operon is found only on the pYR4 plasmid and is absent from the chromosome. The results obtained from this analysis indicate that pYR4 is a very stable plasmid. The conjugative ability of the plasmid may maintain it in the population and determine its stability, so that even if individual cells lose it, they get it back from their neighbors. This suggests that the plasmid cannot be cured easily in a short time frame such, as the one that we tested. As we were unable to cure pYR4 within a reasonable amount of time, this precluded performing virulence assays to check the involvement of the plasmid in pathogenesis.

### DISCUSSION

#### Higher Quality of pYR4 Assembly and Identification of Unplaced Scaffolds

Comparative analysis of the pil region of pYR4 plasmid with other Y. ruckeri plasmids revealed that the pil region is not a common feature among Y. ruckeri strains. Even though we could identify different pil regions in some of the Y. ruckeri plasmids such as pWKY, we did not detect any significant sequence similarity with the pil region described here. Literature suggests that the pil operon can be encoded both on plasmids (pADAP of S. entomophila) (Hurst et al., 2011) and the chromosome (in Y. pseudotuberculosis 32777) (Collyn et al., 2002). Our data shows that in Y. ruckeri NVH\_3758, the


TABLE 4 | The main characteristic features of in pil operon identified in the pYR4 plasmid from Y. ruckeri NVH\_3758.

TFP gene cluster is plasmid-encoded and no other TFP gene cluster was detected on the chromosome. In fact, our genomic comparative survey indicates that previously deposited Illuminasequenced genomic assemblies containing the tra and the pil operons—five scaffolds from a Y. ruckeri 150 assembly and a scaffold from Y. ruckeri ATCC29743—correspond to the plasmid pYR3. In particular for the 150 strain, failure in placing the genomic scaffolds may have occurred due to the presence of repetitive DNA sequences and of mobile elements in the flanking regions, e.g., the repetitive mobile element at positions 37,633– 38,328 and 41,885–42,850 of pYR3 (**Figure 3**). This might also explain the presence of sequence gaps in the scaffolds 20 and 23 of ATCC29743 (**Figure 3**). These results show the power of the PacBio SMRT technology in producing higher-quality genomic assemblies, thanks to the longer average read lengths available.

Our study highlights the problem related to incorrect assemblies when using second generation sequencing (SGS) technologies. Short-read sequencing is often not enough to properly assemble plasmid sequences. Plasmids often contain many mobile repeat structures whose DNA length exceeds that provided by limitations of SGS technology (ranging 100–600 bp), thus generating unplaced scaffolds and misassemblies. The longer average read length provided by the PacBio SMRT sequencing can address some of the limitations of the SGS technologies, making it possible to correctly place genomic scaffolds even when containing repetitive regions and obtain higher quality assemblies. Sequencing the NVH\_3758 genome using the PacBio technology yielded two contigs representing the chromosomal genome of ∼3.8 Mb and the ∼81 kb plasmid pYR4. Thanks to the PacBio platform, we could correctly determine the plasmid location of both the tra and the pil operons. The presence of the tra operon on the pYR4 plasmid, not on the chromosome, is reasonable as the tra operon encodes genes involved in bacterial conjugation and DNA transfer. We re-emphasize that these two systems, despite having similar names, are very distinct structurally and mechanistically. The T4SS is a secretion system that translocates nucleoprotein complexes or effector proteins into target cells, whereas TFP are contractile appendages mainly mediating adhesion and certain types of motility (Shi and Sun, 2002).

#### TFPs in pYR4 And Their Potential Role in Virulence and Conjugation

In the human pathogenic Yersiniae, pathogenicity is mainly related to the presence of the 70-kb virulence plasmid pYV. This plasmid encodes the Yop proteins and T3SS, which enable the bacteria to survive and multiply in the host tissues (Viboud and Bliska, 2005). Plasmids described so far in Y. ruckeri have recently gained more and more attention due to their potential association with virulence, although in-depth knowledge regarding their function is lacking.

Annotations of pYR4 showed 92 open reading frames. We could identify three different functional regions responsible for plasmid partitioning, a T4SS and a TFP. The present study strongly suggests that the pYR4 pil cluster belongs to the TFP family. In addition to attachment and motility, TFPs can be involved in DNA uptake as shown in N. gonorrhoeae (Wolfgang et al., 1998). Interestingly, there are examples of TFPs being involved in bacterial conjugation. For example, the PAPI-1 pathogenicity island of P. aeruginosa can be transferred to a recipient strain lacking this island. The mobilization of PAPI-1 was dependent on a TFP (Carter et al., 2010). The fact that the pil operon clusters together with the tra operon suggests that the two are functionally coupled and that the TFPs are involved in conjugative transfer of pYR4. However, it is also possible that the TFP is a virulence factor. We speculate that the pil proteins in pYR4, apart from being involved in virulence, can also be responsible for thin pilus formation required for liquid mating (Kim and Komano, 1997). Experiments carried out by Collyn et al. showed that the TFP gene cluster present in Y. pseudotuberculosis is not only involved in synthesis of TFPs, but also contributes to its virulence (Collyn et al., 2002). Based on these findings, we speculate that the pil operon present in Y. ruckeri could play a similarly important role in fish disease.

The difference between the G+C content of the pil operon (51.7%) and the average G+C content of the pYR4 plasmid (49.4%) suggests that the pil operon could have been acquired relatively recently by horizontal gene transfer. Likewise, the difference in G+C content compared to the Y. ruckeri NVH\_3758 chromosome (47.6%) suggests that the plasmid has been acquired from a different species. Previous studies show that for a plasmid to be horizontally transferred, the difference in the G+C composition between the genome and the plasmid should be in the range from 1 to 5% (Hurst et al., 2011). The G+C content of the plasmid pADAP of S. entomophila is 53%. Most of the pil proteins from pYR4 (PilL, PilN, PilO, PilP, PilQ, PilR, PilS, PilT, PilD) are highly similar to pil proteins found in TFPs in Serratia species, whereas others (PilL, PilU, PilN) are more similar to those found in other Yersiniae. These findings, in addition to the same gene organization of the pil cluster, suggest that the TFP locus in pYR4 may have been acquired from S. entomophila, which occupies a similar aquatic environment as Y. ruckeri (Grimont et al., 1988).

#### T4SSs and the *tra* Operon

Analysis of the pYR4 protein encoding sequences allowed us to identify a complete T4SS, which we named tra. The corresponding coding region consists of three operons. The genetic organization of the tra operons (tra1, tra2, tra3) is very similar to that found in pADAP of S. entomophila, pCTX-M3 of C. freundii and pEL60, pEA67, pEA72, and pEA78 of E. amylovora, as previously described for tra1 (Méndez et al., 2009). Comparative analysis of the pYR4 tra operons to the pYR3 tra operons from Y. ruckeri CSF007-82 showed 99.9% nucleotide sequence similarity, in contrast to tra operons from other Y. ruckeri strains, where we did not detect any strong sequence similarity. In 2007, Welch et al. showed the presence of multidrug resistance of the plasmid pYR1 in Y. ruckeri YR71. pYR1 also contains a T4SS for conjugative transfer. However, the T4SS of pYR1 does not show high levels of sequence similarity—and no LCB was found in the Mauve alignment—to the tra operons described here (Welch et al., 2007).

Interestingly, similarities between the pYR4 Tra and L. pneumophila Icm/Dot proteins implicate a role in Y. ruckeri virulence. L. pneumophila is the causative agent of Legionnaires' disease. As an intracellular pathogen, L. pneumophila is able to grow and multiply within human macrophages, leading to their killing. In L. pneumophila, many virulence genes located in icm ("for intracellular multiplication") and dot ("defect in organelle trafficking") locus have been identified. Some of the Dot/Icm proteins are homologous to the Tra proteins found in plasmid R64 (Komano et al., 2000). pYR4 Tra proteins display amino acid sequence similarity to Icm/Dot proteins ranging from 29 to 34%. They are similar in size and their predicted functions are similar. However, as we could not identify any genes encoding putative effector proteins in pYR4 of NVH\_3758, it is possible that the tra locus is purely conjugative. Nevertheless, we cannot conclusively rule out a role for the tra locus in virulence. Interestingly, Méndez et al. generated the traI mutant strain of Y. ruckeri 150 in which the traI gene was disrupted. In the in vivo competition assays in rainbow trout, the virulence of the traI mutant strain was reported to be attenuated when compared to the WT strain, suggesting that TraI is involved in virulence of this bacterium.

### CONCLUSIONS

In conclusion, the results presented here suggest that pYR4 from the 1987 outbreak strain Y. ruckeri NVH\_3758 is a conjugative plasmid that encodes a T4SS and a TFP that might contribute to Y. ruckeri virulence. The 55 kDa plasmid backbone is identical to that of pYR3 and has presumably been acquired by horizontal gene transfer through conjugation from a Serratia species that occupies the same biological niche, as previously suggested (Méndez et al., 2009). In addition, pYR4 contains a a previously undescribed ∼25 kDa region with a partitioning system completely different from that of pYR3. This region contains a set of mobile elements, several hydrogenase enzymes with a corresponding transcription factor, and additional hypothetical genes that need further investigation.

The plasmid could contribute to the dissemination of Y. ruckeri virulence by spreading the TFP-encoding pil locus among non-virulent strains. Further experiments are required to elucidate the function of the pil and tra regions and to clarify their role in virulence. However, we would like to highlight that based on 100% sequence similarity between TraI of pYR4 and TraI of Y. ruckeri 150 that we now know is located on pYR3 based on our re-assembly of the Y. ruckeri 150 genome, we assume that at least the tra region is directly involved in virulence. traI deletion in Y. ruckeri 150 lead to a significant decrease in virulence (Méndez et al., 2009). Additionally, our study demonstrates the power of PacBio SMRT sequencing technology in producing assemblies of high quality and accuracy, compared to sequencing technologies based on shorter read lengths such as Illumina and 454. The latter methods failed to show the plasmid localization of the pil and tra regions in multiple cases.

#### DATA AVAILABILITY

The sequence of pYR4 described herein has been deposited in GenBank with the accession number CP032236.

#### AUTHOR CONTRIBUTIONS

AW and CO conceived, planned, and carried out the experiments. AW, CO, JL, and DL contributed to the discussion and interpretation of the results, and wrote the manuscript. All authors provided critical feedback and contributed to the final shape of the manuscript.

#### FUNDING

This work by supported by Departmental funds of the Department of Biosciences, University of Oslo (COMPI), by Research Council of Norway Young Investigator grant 249793

#### REFERENCES


(to JL), and by Research Council of Norway FriMedBio grant 240483 (to DL).

#### ACKNOWLEDGMENTS

We would like to thank the sequencing service provided by the Norwegian Sequencing Centre (www.sequencing.uio. no), a national technology platform hosted by the University of Oslo and supported by the Functional Genomics and Infrastructure programs of the Research Council of Norway and the Southeastern Regional Health Authorities. We express gratitude to Prof. Duncan Colquhoun (Norwegian Veterinary Institute) for providing Y. ruckeri NVH\_3758.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00373/full#supplementary-material

cluster that contributes to pathogenicity. Infect. Immun. 70, 6196–6205. doi: 10.1128/IAI.70.11.6196-6205.2002


a "Y. ruckeri Invasin-like molecule", (yrIlm) sheds light on the evolution of adhesive capacities of a fish pathogen. J. Struct. Biol. 201, 171–183. doi: 10.1016/j.jsb.2017.08.008


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer HL and handling Editor declared their shared affiliation.

Copyright © 2018 Wrobel, Ottoni, Leo and Linke. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Insecticidal Toxicity of *Yersinia frederiksenii* Involves the Novel Enterotoxin YacT

Katharina Springer <sup>1</sup> , Philipp-Albert Sänger <sup>2</sup> , Christian Moritz <sup>1</sup> , Angela Felsl <sup>1</sup> , Thomas Rattei <sup>3</sup> and Thilo M. Fuchs 1,2 \*

<sup>1</sup> Lehrstuhl für Mikrobielle Ökologie, Fakultät für Grundlagen der Biowissenschaften, Wissenschaftszentrum Weihenstephan, Technische Universität München, Freising, Germany, <sup>2</sup> Friedrich-Loeffler-Institut, Institut für Molekulare Pathogenese, Jena, Germany, <sup>3</sup> Department of Computational Systems Biology, University of Vienna, Vienna, Austria

The genus Yersinia comprises 19 species of which three are known as human and animal pathogens. Some species display toxicity toward invertebrates using the so-called toxin complex (TC) and/or determinants that are not yet known. Recent studies showed a remarkable variability of insecticidal activities when representatives of different Yersinia species (spp.) were subcutaneously injected into the greater wax moth, Galleria mellonella. Here, we demonstrate that Y. intermedia and Y. frederiksenii are highly toxic to this insect. A member of Y. Enterocolitica phylogroup 1B killed G. mellonella larvae with injection doses of approximately 38 cells only, thus resembling the insecticidal activity of Photorhabdus luminescens. The pathogenicity Yersinia spp. displays toward the larvae was higher at 15◦C than at 30◦C and independent of the TC. However, upon subtraction of all genes of the low-pathogenic Y. enterocolitica strain W22703 from the genomes of Y. intermedia and Y. frederiksenii, we identified a set of genes that may be responsible for the toxicity of these two species. Indeed, a mutant of Y. frederiksenii lacking yacT, a gene that encodes a protein similar to the heat-stable cytotonic enterotoxin (Ast) of Aeromonas hydrophila, exhibited a reduced pathogenicity toward G. mellonella larvae and altered the morphology of hemocytes. The data suggests that the repertoire of virulence determinants present in environmental Yersinia species remains to be elucidated.

Keywords: *Yersinia*, *Galleria mellonella*, insecticidal activity, enterotoxin, YacT

## INTRODUCTION

The genus Yersinia so far consists of three human pathogens (Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica), and at least 16 species are considered mostly harmless to humans, namely Y. aldovae, Y. bercovieri, Y. frederiksenii, Y. intermedia, Y. kristensenii, Y. mollaretii, Y. rohdei, and the more recently described Y. ruckeri, a fish pathogen (Sulakvelidze, 2000), Y. aleksiciae (Sprague and Neubauer, 2005), Y. similis (Sprague et al., 2008), Y. massiliensis (Merhej et al., 2008), Y. nurmii (Murros-Kontiainen et al., 2010a), Y. pekkanenii (Murros-Kontiainen et al., 2010b), Y. wautersii (Savin et al., 2014), and Y. entomophaga (Hurst et al., 2010). These species that are non-pathogenic for humans have been isolated from water, soil, food, domestic and wild animals, and human beings in which they do not cause any clinical infections.

While the virulence properties of the pathogenic species have been characterized during last few decades, much less is known about the determinants that allow Yersinia species to survive in the

#### *Edited by:*

Matthew S. Francis, Umeå University, Sweden

#### *Reviewed by:*

Mark Robin Holmes Hurst, AgResearch, New Zealand Louis S. Tisa, University of New Hampshire, United States

> *\*Correspondence:* Thilo M. Fuchs thilom.fuchs@fli.de

#### *Specialty section:*

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 20 April 2018 *Accepted:* 18 October 2018 *Published:* 14 November 2018

#### *Citation:*

Springer K, Sänger P-A, Moritz C, Felsl A, Rattei T and Fuchs TM (2018) Insecticidal Toxicity of Yersinia frederiksenii Involves the Novel Enterotoxin YacT. Front. Cell. Infect. Microbiol. 8:392. doi: 10.3389/fcimb.2018.00392 environment. We have recently demonstrated that the Y. enterocolitica strain W22703 (biotype 2, serotype O: 9) is toxic to nematodes and larvae of the tobacco hornworm, Manduca sexta, upon oral infection or oral toxin application, and that this insecticidal activity correlates with the presence of the pathogenicity island (TC-PAIYe) (Bresolin et al., 2006; Spanier et al., 2010). This 20-kb fragment is present in the genome of Y. pestis, Y. pseudotuberculosis, and Y. enterocolitica biotype 2– 5 strains, but is absent in the genomes of the highly—pathogenic biotype 1B strains, including 8081 and of the most biotype 1A strains, which are considered to be non-pathogenic to humans. The TC-PAIYe carries the toxin complex (TC) genes with high identity to the tc genes of entomophagous Photorhabdus luminescens, and it might be speculated that the TC is required to penetrate the epithelial cell barrier of the insect gut to allow Y. enterocolitica cells to enter the hemocoel. In Y. entomophaga MH96, an insecticidal pathogenicity island termed PAIYe96 was characterized, which is distinct from TC-PAIYe in terms of gene homology and genetic organization (Hurst et al., 2011). A unique feature of this island is that it encodes, besides the type ABC genes, two chitinases that are associated with the mature TC (Busby et al., 2012).

The transcription of the tc genes in Y. enterocolitica is subject to a strict temperature-dependent regulation as they are completely silenced at 37◦C, but strongly upregulated at lower temperatures with a maximal transcription at 10–15◦C approximately (Bresolin et al., 2006; Starke et al., 2013; Starke and Fuchs, 2014). Thus, the TC-dependent activity of Y. enterocolitica to invertebrates is reciprocally regulated in comparison with that of many Yersinia virulence factors directed against humans (Marceau, 2005). Notably, strains lacking the insecticidal genes, including Y. enterocolitica, Y. mollaretii, Y. bercovieri, Y. ruckeri, and Y. aldovae, are still toxic when subcutaneously injected into G. mellonella, indicating the presence of yet unknown insecticidal determinants in these species (Fuchs et al., 2008). More recently, it was demonstrated that Y. enterocolitica strains of phylogroup 1 exhibit a strong virulence against G. mellonella larvae at temperatures of 25◦C and higher upon intrahemocoelic injection, independently of the presence of virulence plasmid pYV (Alenizi et al., 2016). These findings resemble functional redundancy in P. luminescens that carries a set of insecticidal factors besides the TC (ffrench-Constant et al., 2007), including the makes caterpillars floppy (MCF) toxins (Daborn et al., 2002), the Pir toxins (Waterfield et al., 2005), the protease PrtA (Bishop, 2014), Txp40 (Brown et al., 2006), the XaxAB-like binary toxins (Zhang et al., 2014), and the Photorhabdus virulence cassettes (Yang et al., 2006). Interestingly, the (partial) loss of some of these insecticidal genes does not result in a lack of toxicity to invertebrates (Wilkinson et al., 2009).

The insecticidal determinants of some Yersinia spp. remain to be investigated. Here, we used an established infection assay with G. mellonella larvae to monitor the phenotype of the host intrahemocoelically infected with Y. frederiksenii, Y. intermedia, P. luminescens, and Y. enterocolitica 8081 and W22703 cells. Dose- and temperature-dependent characteristics of their insecticidal activity were determined, and genome comparison as well as toxin injection were applied to gain further insights into the entomopathogenic repertoire of environmental Yersinia strains.

### RESULTS

Dose-dependent toxicity of Y. frederiksenii and Y. intermedia to G. mellonella larvae. When larvae of the first-instar neonates of M. sexta were challenged orally with Y. frederiksenii and Y. intermedia in preliminary experiments, the toxicity was calculated as 71 and 19%, respectively (**Table 1**). As both strains used here lack the TC-PAIYe, we switched from this oral infection model to a subcutaneous infection model and injected 5 µl of a 1:100 dilution of Y. frederiksenii and Y. intermedia overnight culture into G. mellonella larvae, and observed a 99–100% lethality after 5 days of incubation at 15◦C (**Table 1**). Thus, the insecticidal activity of Y. frederiksenii and Y. intermedia is higher than that of all other Yersinia strains tested recently in the same infection model, including Y. enterocolitica strain W22703 (biotype 2, serotype O:9) (Fuchs et al., 2008). To determine the toxicity of Y. intermedia and Y. frederiksenii toward G. mellonella larvae in more detail, we used defined infection aliquots of approximately 103–10<sup>4</sup> colony forming units (CFU), and monitored the fate of intrahemocoelically infected G. mellonella larvae over 5 days at room temperature (20◦C) (**Figure 1A**). The timecourse revealed that all larvae survived for 24 h, but most of them died within the next 2 days. The survival rates of the animals infected with two Yersinia species did not significantly (p > 0.05) differ under the conditions applied here. In parallel, we homogenized two larvae each day, and monitored the replication of Y. frederiksenii within the larvae. The CFU of Y. frederiksenii increased from 1.15 × 10<sup>3</sup> to 3.21 × 10<sup>3</sup> (day one) and to 2.79 × 10<sup>9</sup> (day two) directly after the infection, and remained constant for next 3 days (1.88–2.62 × 10<sup>9</sup> ). This unimpeded bacterial growth resembles the mortality of the larvae that starts only when the Y. frederiksenii reaches its stationary phase. This finding suggests that a high cell number of this insect pathogen in the larvae of Galleria is a prerequisite for its toxicity and/or that the pathogen has incapacitated the host in the initial phase of intrahemocoelical infection.

We further reduced the infection dose to approximately 100–150 CFU and observed a slightly higher toxicity of Y. frederiksenii in comparison with Y. intermedia, which, however, started to kill larvae a day earlier (**Figure 1B**). In contrast, 145 CFU of Y. enterocolitica W22703 were not sufficient to kill any larva. Altogether, these data demonstrate a high, dose-dependent toxicity of Y. frederiksenii and Y. intermedia toward larvae of G. mellonella.

#### Entomopathogenicity of *Y. enterocolitica* Strain 8081 Resembles That of *P. luminescens*

We compared the injectable insecticidal activity of P. luminescens, Y. frederiksenii, Y. intermedia, and Y. enterocolitica strain 8081 (biotype 1B, serotype O:8), which carries a socalled "high-pathogenicity island" encoding the siderophore yersiniabactin (Carniel et al., 1996). In each experiment, the


TABLE 1 | Oral infection of M. sexta and intrahemocoelic infection of G. mellonella for 5 days.

<sup>a</sup>The average mortality of at least three independently performed experiments with a minimum of six larvae each are shown <sup>b</sup>according to the genome sequence <sup>C</sup>(Fuchs et al., 2008).

strain were performed, with three groups composed of (A) 25, 29, and 30 larvae (Y. intermedia) and 20, 22, and 23 larvae (Y. frederiksenii), or (B) 15 larvae each. The larvae were incubated at 20◦C and monitored daily. Error bars represent the standard error of the mean of three experiments.

lethality of the larvae decreased with lower numbers of CFU (**Figure 2**). Interestingly, the survival assays demonstrated that Y. enterocolitica 8081 is nearly as toxic as P. luminescens toward G. mellonella larvae upon intrahemocoelic infection, and is more virulent than Y. frederiksenii and Y. intermedia. Approximately 38 CFU of Y. enterocolitica 8081 were revealed to be sufficient to kill nearly all larvae, after an infection period of 5 days. For comparison, larvae were intrahemocoelically infected with Y. enterocolitica W22703 and DH5α, demonstrating the high insect-pathogenicity of Y. frederiksenii, Y. intermedia, and Y. enterocolitica strain 8081 despite the lack of TC-PAIYe. Altogether, these data

show that the lethality toward G. mellonella larvae is strictly dose-dependent.

### The Lethality of *Yersinia* Strains Is Temperature-Dependent

Low temperature-dependent toxicity of Y. enterocolitica W22703 toward M. sexta and C. elegans, and of representative strains of Y. enterocolitica phylogroups 1–5 against G. mellonella has been reported previously (Bresolin et al., 2006; Fuchs et al., 2008; Spanier et al., 2010; Alenizi et al., 2016). Therefore, we tested whether the injectable insecticidal activity described earlier is higher at lower temperature. G. mellonella larvae were infected with varying cell numbers of Y. frederiksenii, Y. intermedia, Y. enterocolitica 8081, and Y. enterocolitica W22703 and incubated at 15◦C and at 30◦C. However, upon infection with eight or 95 Y. frederiksenii CFU, the larvae showed a higher survival rate at 30◦C than at 15◦C (**Figure 3A**). A temperature-dependent pathogenicity toward G. mellonella was also observed for Y. intermedia. Although this species was found to be slightly low pathogenic at 15◦C than Y. frederiksenii, we observed that at 30◦C, 120 Y. intermedia CFU killed more larvae (40% survival rate) in comparison with 95 Y. frederiksenii CFU (80% survival rate) (**Figure 3B**). Y. enterocolitica 8081 exhibited a higher toxicity against the larvae at lower temperature as well. Only 57 CFU of this pathogen killed nearly all larvae at 15◦C, but 40 CFU killed only 4% at 30◦C (**Figure 3C**). Independent of the infection dose, Y. enterocolitica W22703 did not exhibit a significant temperature-dependent toxicity in this model (**Figure 3D**). A high infection dose of 9 × 10<sup>5</sup> CFU quickly killed all larvae at 15◦C and at 30◦C, and a low infection dose of 145 CFU killed zero or only 7% of all larvae at these temperatures. Thus, a pronounced dose-dependent insecticidal activity was observed in these experiments with Y. frederiksenii, Y. intermedia, and both Y. enterocolitica strains.

#### Phenotypes of Infected Larvae

A healthy G. mellonella larva rapidly moves forward and back upon touch, and its exoskeleton is light colored. During the pathogenicity assays described earlier, we observed distinct phenotypes of the larvae at both 15◦C and 30◦C (**Figure 4A**). The insects infected with 95 Y. frederiksenii CFU were more agile at 30◦C, possibly due to the lower toxicity of the pathogen at this temperature. At this temperature, injuries by combats and thus the release of hemolymph is visible from day 3 post infectionem (p. i.) due to the high density of insects. From day 4 to 6 p. i., the number of insects in the pupal stage as well as cocoon production increased. At 15◦C, all larvae remained undamaged. However, their agility decreased from day 1 p. i. until the larvae moved only their heads or died. They also exhibited a stronger exoskeleton coloring from day 4 p. i. that strengthened until day 6 p. i. Furthermore, while non-infected larvae are sturdy, their body volume decreased upon the loss of liquid as visible in **Figure 4B**, left, on the larvae's surface.

Another interesting observation was the differential pigmentation of the larvae. Melanization is a defense mechanism of G. mellonella larvae to encapsulate pathogens, and the intensity of pigment formation correlates with the number of injected cells (Thomaz et al., 2013). Following infection with Y.

FIGURE 3 | Temperature-dependent pathogenicity of Yersinia strains against G. mellonella larvae. (A) Larvae were each infected with eight and 95 Y. frederiksenii CFU, respectively, and incubated at 15◦C and 30◦C; the experiment at 15◦C with an infection dose of eight was performed with 30 ungrouped larvae. (B) Infection was done with 120 Y. intermedia CFU and the larvae incubated at 15◦C and 30◦C. Infection with E. coli DH5α was used here as a control for all experiments. (C) 57 Y. enterocolitica 8081 CFU were used to infect G. mellonella larvae, which were incubated at 15◦C; in a further assay, larvae infected with 40 CFU were incubated at 30◦C; the experiment with an infection dose of 57 CFU was done with 3 × 10 larvae. (D) Infection assays were performed with 9 × 10<sup>5</sup> or with 145 Y. enterocolitica W22703 CFU at 15◦C and 30◦C. In all experiments, three groups of 15 larvae each were independently infected with the exceptions mentioned earlier. Error bars represent the standard deviations. The larvae survival rate was plotted against day's p. i.

frederiksenii and Y. intermedia, the larvae colored gray-brown to black before they died (**Figure 4B**); many larvae also exhibited a punctiform pigmentation that resembled that at the injection site. Larvae infected with P. luminescens, however, did not show such a pigmentation, but colored red, similar to the effect of red anthraquinones produced by P. luminescens following infection (Richardson et al., 1988). These observations point out to different factors that are involved in the injectable insecticidal activity of the three pathogens.

#### A Genome Comparison Approach Identifies Potential Virulence Genes Present in Highly and Absent in Weakly Insecticidal Strains

Although their genomes lack the tc genes or their homologs, Y. intermedia and Y. frederiksenii are much more toxic against G. mellonella larvae than Y. enterocolitica W22703. This finding suggests the presence of yet unknown genetic determinants that contribute to the insecticidal activity of yersinia. Therefore, we performed a genome comparison that identified 329 genes that are common for Y. intermedia strain ATCC 29909 and Y. frederiksenii strain ATCC 33641, but absent in Y. enterocolitica W22703 (**Table S1**). This set comprises a large number of genes whose (putative) products belong to categories such as lipoproteins and other membrane proteins (10 + 14), sensing, signaling, and regulation (36), metabolism (28), resistance toward toxic substances (17), transport and secretion (16 + 12), stress response (2), and iron uptake and storage (9). With respect to genetic determinants potentially involved in pathogenicity, the bioinformatics approach identified putative adhesins, toxins, hemolysins, and secretory systems (**Table 2**). For example, Y. intermedia and Y. frederiksenii carry a type VI secretion system (T6SS) that, among other functions, contributes to virulence (Filloux, 2013) and is present in all Yersinia spp. and in P. luminescens, but not in Y. enterocolitica W22703. The two species harbor an ATP-binding protein possibly involved in uptake of heme, which is absent in all other species of the Yersinia genus, but closely related to a protein in Klebsiella pneumoniae.

(R.T.).

## Attenuated Insecticidal Phenotype of *Y. frederiksenii* 1*yacT*

In this genome comparison approach, we identified yacT (accession numbers EEQ13070 and WP\_004712324) encoding a protein whose amino acid sequence exhibits a significant homology (e-value 0.0, identity 54%; **Supplementary Figure S1**) to the heat-stable cytotonic enterotoxin (Ast) of Aeromonas hydrophila. We termed this protein, with a molecular weight of 71.46 kDa, Yersinia Ast-like cytotonic toxin (YacT), and the corresponding gene yacT. Homologs or orthologs of YacT are also encoded by P. luminescens, P. asymbiotica, and many Yersinia spp., but neither by Y. pestis or Y. pseudotuberculosis, nor by Y. enterocolitica strains W22703 and 8081. We generated a deletion mutant of yacT termed Y. frederiksenii 1yacT, which was also complemented with pACYC-yacT carrying the toxin gene. On performing the G. mellonella infection assay at 15◦C, we observed a strongly reduced virulence of Y. frederiksenii 1yacT (time in days for 50% of the larvae to die, TD<sup>50</sup> = 5.4 ± 0.22) and of Y. frederiksenii 1yacT/pACYC184 (TD<sup>50</sup> = 5.59 ± 0.01) in comparison to strain Y. frederiksenii/pACYC184 (TD<sup>50</sup> = 3.33 ± 0.18) (**Table 3**, **Figure 5A**). When the mutant harbored gene yacT in trans via plasmid pACYC-yacT, its phenotype reverted to that of the parental strain showing a TD<sup>50</sup> = 3.79 ± 0.46. These data clearly demonstrated that yacT is required for the high virulence of Y. frederiksenii toward the larvae at 15◦C.

To understand better the role of the novel toxin during infection, the number of viable Y. frederiksenii cells within infected larvae incubated at 15◦C was determined daily over a duration of 4 days (**Figure 5B**). We observed a strong growth of Y. frederiksenii within 4 days by more than six orders of magnitude. In comparison, a mutant Y. frederiksenii 1yacT

TABLE 2 | Putative virulence factors of Y. frederiksenii and Y. intermedia absent in Y. enterocolitica W22703.


<sup>a</sup>Gene code was taken from the PEDANT 3 database (Walter et al., 2009).

TABLE 3 | Infection doses and TD50 values testing yacT.


Sd, standard deviation; \*, time in days for 50% of the larvae to die; \*\*, more than 50% of the larvae survived.

exhibited a retarded proliferation at day 2 p. i., followed by growth stagnation for 1 day. However, at day 4 p. i., the mutant reached approximately the same cell density as the parental strain. These data confirm that the yersiniae cell numbers increase before the larvae start to die and that YacT contributes to proliferation of Y. frederiksenii within the insect host.

#### Effect of YacT on Hemocytes

YacT was purified from E. coli Bl21 (DE3)/pBAD-HisA(tet)-650. Six microliter of a toxin solution with a concentration of 1.4 µg/µl or of phosphate-buffered saline (PBS) as control were injected into 20 G. mellonella larvae. The larvae of the toxin group showed paralysis of the half retral abdomen immediately after injection. In addition, some caterpillars of this group displayed a constriction of the head-thorax area and did not react to touching. In comparison, the control group showed none of these symptoms. One day p.i., animals of both groups that were kept at 30◦C maintained vigor and formed fine webs. After web removal, 2 or 3 days p. i., the caterpillars of the toxin-treated group showed punctate- to strokelike black discolorations at the dorsal–abdominal areas that we did not observe in the control group (**Supplementary Figure S2**).

boxes: Y. frederiksenii, black boxes: Y. frederiksenii 1yacT. Standard deviations of three replicates are shown.

To study the effect of YacT on hemocytes after 1 day, the hemolymph of larvae was prepared from the aorta and streaked out on microscope slides for staining. Injection of PBS (10 mM phosphate buffer, pH 7.4; 2.7 mM KCl; 137 mM NaCl) served as a control. Upon microscopic analysis, we observed repeatedly several distinct cell phenotypes: hemocytes from animals treated with the toxin showed a round-shaped morphology and they began to form aggregates in comparison to the controls and several cells also enlarged and showed a reduction of chromatin, possibly indicating the beginning of early stages of the cell death (**Figure 6A**).

As a complementary experiment, G. mellonella larvae were infected orally with Y. frederiksenii and its yacT deletion mutant. Again, the morphology and chromatin density of hemocytes were modified in the presence of Y. frederiksenii, but not of Y. frederiksenii 1yacT (**Figure 6B**). In both cases, yersinial cells were visible in the preparations, indicating that penetrating the gut epithelial barrier occurs independent of YacT.

## DISCUSSION

Members of the genus Yersinia are fascinating organisms, as they are able to adapt to the environmental life cycle stage as well as to mammals (Fuchs et al., 2011). During a transition, they encounter a broad spectrum of hostile conditions, and a major clue to overcome these challenges is the temperature-dependent production of host-specific virulence factors. Therefore, the interaction of yersiniae with invertebrates may have been a precursor to human pathogenicity during evolution (Waterfield et al., 2004). In this study, we tested the entomopathogenic potential of a set of Yersinia spp. toward larvae of G. mellonella. The larvae are considered to be a natural host of yersiniae and other pathogens and, therefore, serve as an indicator of yersinial virulence activities against insects. We identified Y. enterocolitica 8081, a representative of the highly pathogenic biovar 1B group, to be the most virulent Yersinia strain tested so far against G. mellonella larvae, resembling the high insecticidal activity of P. luminescens. Data on strain 8081 as the least pathogenic strain among several Y. enterocolitica strains tested against G. Mellonella are not in contradiction with our findings, because Alenizi et al. performed the infection experiments at 25◦C and missed the high toxicity at the environmental temperature of 15◦C (Alenizi et al., 2016). Y. enterocolitica strain 5303, which belongs to the biovar 1A group and is considered to be apathogenic toward mammals, showed an even higher toxicity toward the Galleria larvae, since only ten CFU were sufficient

magnification.

to kill 50% of the larvae within 5 days (Alenizi et al., 2016). Interestingly, also Y. intermedia and Y. frederiksenii are more virulent to Galleria larvae than other Yersinia spp., including Y. mollaretii, Y. bercovieri, Y. ruckeri, Y. aldovae, and Y. kristensenii, as tested recently (Fuchs et al., 2008). Y. frederiksenii and Y. intermedia occupy related ecological niches and exhibit very similar phenotypes (Martin et al., 2009). Y. intermedia, which is isolated mainly from the environment, animals, food, and (rarely) human beings, received its name due to its genetic and phenotypic properties that are an intermediate between those of Y. pseudotuberculosis and Y. enterocolitica (Martin et al., 2009). Y. intermedia also shares several O antigens with Y. enterocolitica (Wauters et al., 1972), of which O:4 and O:17 are probably the prevailing serotypes (Ursing et al., 1980). Y. frederiksenii was differentiated from Y. enterocolitica in 1980 (Ursing et al., 1980). The high insecticidal potential might point out to yet overlooked natural habitats of these strains.

Temperature is an important signal in the regulation of yersinia virulence factors of that are predominantly produced at 37◦C and repressed at temperatures lower than body temperature, or vice versa, as exemplified by the insecticidal tc genes in Y. enterocolitica W22703. Temperature-dependent mortality of G. mellonella upon oral infection, but not upon intrahemocoelic injection, was observed recently for Y. entomophaga (Hurst et al., 2015). Therefore, it is not a surprising outcome of this study that the toxicity of Y. frederiksenii, Y. intermedia, and Y. enterocolitica 8081 increases with lower temperature, thus pointing out to a relevant ecological niche of these strains. Irrespective of the fact that G. mellonella has been chosen here as an infection model rather than as a natural host for yersinial infection, the two temperatures applied here correspond to the lifestyle of G. mellonella larvae that grow best between 29◦C and 35◦C, and also develop at 15◦C, but not at 10◦C or less.

The pronounced contrast between the insecticidal potential of Y. frederiksenii, Y. intermedia, and Y. enterocolitica 8081, on the one hand, and the TC-PAIYe-harboring Y. enterocolitica W22703, on the other hand, at least with respect to the G.mellonella model used here, prompted us to perform a genome comparison. This approach aimed to identify the determinants that confer the high insecticidal activity of these strains. **Table 2**, which probably still lacks several factors involved in infection, points out to a broad spectrum of yersinial factors whose role in pathogenicity as well as their host specificity remains to be investigated. One of them is YacT that is highly homologous to the astencoded heat-stable, cytotonic enterotoxin of A. hydrophila (Chopra et al., 1994) that was associated with gastroenteritis Springer et al. Insecticidal Activity of Yersinia Species

and non-bloody diarrhea in children and shown to contribute to the fluid secretory response in a murine model (Sha et al., 2002). Cell lysates of E. coli cells carrying ast elongated Chinese hamster ovary cells, which is a typical response to enterotoxins (Chopra et al., 1994). Besides Yersinia strains, YacT orthologs were identified also in P. luminescens ssp. laumondii (e-value = 10−177) and in the human pathogen, P. asymbiotica (evalue = 10−179), demonstrating that this factor is not unique to A. hydrophila as assumed previously (Sha et al., 2002). The prevalence of the Ast and YacT homologs confirms the strong functional relatedness between Photorhabdus spp. and Yersinia spp. with respect to their invertebrate and vertebrate association (Heermann and Fuchs, 2008). It is important to note that YacT is distinct from the heat-stable enterotoxin Yst of yersiniae, for which a homolog is missing in Y. frederiksenii ATCC 33641 and Y. intermedia (Singh and Virdi, 2004). Our data demonstrate that YacT is required for full pathogenicity toward G. mellonella. Moreover, the finding that YacT injection affects the morphology of hemocytes suggests that the immune response of G. mellonella controls better the proliferation of Y. frederiksenii 1yacT during the first 3 days p. i. as compared with that of Y. frederiksenii. The list of determinants in **Table 3** and the variation of Yersinia spp. in pathogenesis toward Galleria larvae suggest that the yersinial toxicity toward insects upon intrahemocoelic infection is a multifactorial process due to the presence of several cytotoxic determinants. In the light of this assumption, the virulence attenuation upon deletion of yacT in Y. frederiksenii is remarkably high. Therefore, YacT is a candidate to explain the high toxicity of Y. frederiksenii against G. mellonella.

### CONCLUSION

A major implication of this study is that the yersinial toxicity toward insects not only depends on the TC, but also on a broader set of insecticidal toxins than known so far. We identified a novel yersinial entomopathogenic factor, whose activity might be associated with the hemocoel rather than with the insect gut as indicated by the distinct oral and intrahemocoelic toxicity of Y. intermedia and Y. frederiksenii. The findings of this study and other studies suggest that yersiniae strains, regardless of being human pathogens or not, acquire a substantial selection advantage by entering invertebrates. By overcoming infection barriers such as the gut epithelium or the innate immune response of insect larvae or nematodes, they might bioconvert their host, thus getting easy access to energy- and nitrogenrich nutrients. The resulting proliferation increases the chance of Yersinia strains to be transmitted to other hosts including mammals.

### MATERIALS AND METHODS

#### Bacterial Strains and Growth Conditions

The strains used in this study were Y. intermedia (Collection Institut Pasteur [CIP] 80.28; ATCC 29909), Y. frederiksenii (CIP 80.29; ATCC 33641), Y. enterocolitica 8081 (Virginia Miller, St. Louis, USA), Y. enterocolitica W22703 (Cornelis and Colson, 1975), and P. luminescens ssp. laumondii strain TT01 BX470251 (Fischer-Le Saux et al., 1999). All cultures were grown in lysogeny broth (LB) (10 g l−<sup>1</sup> tryptone, 5 g l−<sup>1</sup> yeast extract, and 5 g l−<sup>1</sup> NaCl) or on lysogeny broth (LB) agar (LB broth supplemented with 1.5 % w/v agar). Escherichia coli were grown at 37◦C and P. luminescens and Yersinia strains at 30◦C. If appropriate, the media were supplemented with the following antibiotics: 50 µg ml−<sup>1</sup> streptomycin, 12 µg ml−<sup>1</sup> tetracycline, 50 µg ml−<sup>1</sup> kanamycin, 20 µg ml−<sup>1</sup> chloramphenicol, and 20 µg ml−<sup>1</sup> nalidixic acid.

#### General Molecular Techniques

The DNA manipulation was performed according to standard procedures (Sambrook and Russell, 2001). To isolate the chromosomal DNA, 1.5 ml of a bacterial culture was centrifuged, and the sediment was re-suspended in 400 µl of lysis buffer (100 mM Tris pH 8.0, 5 mM EDTA, 200 mM NaCl). After incubation for 15 min on ice, 10 µl of 10% SDS and 5 µl of proteinase K (10 mg/ml) were added, and the sample was incubated overnight at 55◦C. The chromosomal DNA was precipitated with 500 µl of isopropanol, washed in ethanol, dried, and dissolved in 500 µl of TE buffer (10 mM Tris-HCl, 1 mM Na2EDTA, pH 7.4) containing 1 µl of RNase (10 mg/ml). Polymerase chain reactions (PCR) were carried out with Taq polymerase (Fermentas, Vilnius, and Lithunia) and the following programme: one cycle at 95◦C for 2 min; 30 cycles at 95◦C for 10 s, at the appropriate annealing temperature for 30 s, at 72◦C for 45 s to 180 s depending on the expected fragment length; one cycle at 72◦C for 10 min. Four Microliter of chromosomal DNA (100 ng ml−<sup>1</sup> ) was used as a template for PCR amplification, and the GeneRuler DNA mix (Fermentas) served as a DNA ladder.

#### Genome Comparison

The sequences of genome used for the comparison were that of Y. enterocolitica 8081 (accession numbers AM286415 for the chromosome and AM286416 for the plasmid), Y. intermedia (genome draft: GCA\_000168035.1), and Y. frederiksenii (genome draft: GCA\_000754805.1). Homology searches of predicted proteins were performed by basic local alignment search tool analysis (Altschul et al., 1997). The PEDANT software system [http://pedant.gsf.de; (Walter et al., 2009)] was used for automatic genome sequence analysis and annotation (Frishman et al., 2001). Genomes were recorded and homology searches of predicted proteins were performed by SIMAP (Arnold et al., 2014). The genome comparisons were calculated by using a custom Perl script, which formatted bidirectional-best sequence hits between all predicted proteins (E ≤ 0.0001).

#### Insecticidal Bioassays

M. sexta were reared as described (Schachtner et al., 2004). For oral bioassays, bacteria were grown at 15◦C (Yersinia strains) or 37◦C (DH5α) until stationary phase. About 50 µl of a culture was applied to 4 mm<sup>3</sup> disks of an agar-based artificial diet (David and Gardiner, 1965). The liquid was allowed to soak into the agar block, which was dried under a laminar flow. First-instar M. sexta neonate larvae were placed on the disk and incubated at 22◦C. The application of bacterial culture aliquots was repeated after 3 days, and the larvae mortality was recorded after 5 days.

Larvae of G. mellonella were obtained from the Zoo-Fachmarkt (München, Germany) and stored for less than 1 week at room temperature. Bacterial strains were grown to stationary phase (optical density at 600 nm [OD600] ∼1–5 × 10<sup>9</sup> cfu/ml) at temperatures between 15◦C and 30◦C (Yersinia spp.), at 30◦C (P. luminescens), or at 37◦C (DH5α), and 10-fold serial diluted. Larvae of 2–3 cm length and of 110–130 mg weight were used. A 5 µl of the bacterial culture or an appropriate dilution thereof were orally applied or injected by a sterilized microsyringe (Hamilton 1702 RN, 25 µl) into the hemocoel through the last left proleg. The aperture reseals after the removal of the syringe, thus preventing the loss of inoculum (Kavanagh and Reeves, 2004).

Infection doses were determined by plating serial dilutions of the cultures used for injection. Control assays had demonstrated that neither the medium nor the wounding by the syringe contributes to the mortality rate of the insects (Fuchs et al., 2008). Infected larvae were incubated for at least 5 days in the dark at the temperature indicated and the number of killed and alive larvae were enumerated each day. Larvae were considered dead if they failed to respond to touch. The TD<sup>50</sup> was calculated using the dose-response curve (drc) package of the R software. To recover bacteria from the larvae, the larvae were surface sterilized with 70% ethanol, washed in H2O, and cut into small pieces. The homogenous mass was suspended into 1 ml LB, rigorously shaken for 5 min with a vortex, and centrifuged at 1,000 rpm for 2 min. Serial dilutions were plated on agar plates with LB or with Yersinia selective medium (Schiemann CIN medium, Oxoid, Wesel, Germany).

#### Deletion Mutants and Complementing Plasmid

In-frame deletion of yacT from Y. frederiksenii was performed by the one-step method based on the phage λ Red recombinase (Datsenko and Wanner, 2000). In short, PCR products comprising the kanamycin resistance cassette of plasmid pKD4, including the flanking FRT sites, were generated using pairs of 70-nucleotide-long primers that included 20 nucleotides priming sequences for pKD4 as template DNA. Homology extensions of 50 bp overlapped 18 nucleotides of the 5′ -end and 36 nucleotides of the 3′ -end of the target gene (Link et al., 1997). About 500– 1,000 ng of fragment DNA were transferred into Y. frederiksenii cells harboring plasmid pKD119. Allelic replacement of the target gene by the kanamycin resistance cassette was controlled by PCR, and nonpolar deletion mutants were obtained via transformation of pCP20. The deletion was confirmed by PCR and sequencing.

Gene yacT including 220 bp upstream and 100 bp downstream of the coding sequence was amplified with the oligonucleotides 5′ -CGATGAATTCAGTGACCGTCTGTG GGTCTG-3′ and 5′ -CGGCCATGGGGGGGCAGCATCGTG GATTC-3′ and ligated into the chloramphenicol resistance cassette of plasmid pACYC184 via NcoI und EcoRI, resulting in pACYC-yacT. The recombinant plasmid was validated by PCR and sequencing.

#### Overproduction and Purification of YacT

Gene yacT of Y. frederiksenii was cloned into plasmid pBAD-HisA(tet) (Starke et al., 2013) via SacI and PstI using the oligonucleotides 5′ -CGATGAGCTCATGCAGAAAATCATACC GAG-3′ and 5′ -AACTGCAGTTATTGGGTGCTAGCCACAG-3 ′ . An overnight culture of E. coli Bl21 (DE3)/pBAD-HisA(tet)-650 was diluted 1:100 into 800 ml of LB medium supplemented with 12µg/ml tetracycline and incubated at 37◦C with rotation at 180 rpm. At an OD<sup>600</sup> of 0.6, protein production was induced by adding 0.2% of arabinose. After incubation for an additional 4 h at 37◦C and 180 rpm, the cells were harvested by centrifugation at 4◦C and 7,500 rpm for 20 min. The pellets were each re-suspended in 5 ml of native lysis buffer (50 mM NaH2PO4, 300 mM NaCl, and 10 mM imidazole at pH 8.0) in the presence of 1 mM protease inhibitor Pefabloc SC (Sigma-Aldrich, Taufkirchen, Germany) and lysed by 4 passages through a French press (SLM Aminca Instruments, Rochester, NY, USA) at 900 psi; residual cell debris was removed thrice by centrifugation at 4◦C and 9,000 rpm for 15 min. Following the filtration, YacT was isolated using the Ni-NTA Fast Start Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. For imidazole removal, proteins were dialyzed against 50 mM phosphate buffer plus 0.5 mM MgSO4, 0.5 mM ZnSO4, and 0.5 mM CaCl<sup>2</sup> and protein extracts were concentrated down to 1 ml with Amicon ultracentrifugal filter units (Millipore). The protein concentration was determined using Roti-Quant solution (Carl, Roth GmbH, Karlsruhe, Germany) according to the Bradford method (Bradford, 1976). The purity of the eluted fractions was analyzed by the separation on a 12.5% sodium dodecyl sulfate (SDS)-PAA gel (**Supplementary Figure S3**).

### AUTHOR CONTRIBUTIONS

KS, P-AS, and CM performed infection assays and analyzed the results, AF constructed the recombinant strains, TR was responsible for the genome comparison, and TF analyzed the data, conceived the study, and wrote the manuscript. All authors drafted and revised the work and approved of the final version.

#### ACKNOWLEDGMENTS

We thank Siegfried Scherer for supporting this study, Luise Ernst for technical assistance, Elisabeth Liebler-Tenorio for support in microscopical analysis, and Henry Derschum, Virginia Miller, Guy Cornelis, and Ralf Heermann for the gift of strains. This study was supported by a grant from the Deutsche Forschungsgemeinschaft (DFG) to TF (FU375/4-2).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00392/full#supplementary-material

#### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Springer, Sänger, Moritz, Felsl, Rattei and Fuchs. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# An Experimental Pipeline for Initial Characterization of Bacterial Type III Secretion System Inhibitor Mode of Action Using Enteropathogenic Yersinia

#### Edited by:

*Matthew C. Wolfgang, University of North Carolina at Chapel Hill, United States*

#### Reviewed by:

*Deborah Anderson, University of Missouri, United States Stephanie Rochelle Shames, Kansas State University, United States*

> \*Correspondence: *Victoria Auerbuch vastone@ucsc.edu*

#### †Present Address:

*Sina Mohammadi, Merck Exploratory Science Center, Cambridge, MA, United States*

#### Specialty section:

*This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology*

Received: *30 July 2018* Accepted: *26 October 2018* Published: *22 November 2018*

#### Citation:

*Morgan JM, Lam HN, Delgado J, Luu J, Mohammadi S, Isberg RR, Wang H and Auerbuch V (2018) An Experimental Pipeline for Initial Characterization of Bacterial Type III Secretion System Inhibitor Mode of Action Using Enteropathogenic Yersinia.*

*Front. Cell. Infect. Microbiol. 8:404. doi: 10.3389/fcimb.2018.00404* Jessica M. Morgan<sup>1</sup> , Hanh N. Lam<sup>2</sup> , Jocelyn Delgado<sup>2</sup> , Justin Luu<sup>2</sup> , Sina Mohammadi 3† , Ralph R. Isberg<sup>3</sup> , Helen Wang<sup>4</sup> and Victoria Auerbuch<sup>2</sup> \*

*<sup>1</sup> Department of Chemistry and Biochemistry, University of California, Santa Cruz, Santa Cruz, CA, United States, <sup>2</sup> Department of Microbiology and Environmental Toxicology, University of California, Santa Cruz, Santa Cruz, CA, United States, <sup>3</sup> Department of Molecular Biology and Microbiology, Tufts University School of Medicine, Boston, MA, United States, <sup>4</sup> Department of Medical Biochemistry and Microbiology, Uppsala University, Uppsala, Sweden*

Dozens of Gram negative pathogens use one or more type III secretion systems (T3SS) to disarm host defenses or occupy a beneficial niche during infection of a host organism. While the T3SS represents an attractive drug target and dozens of compounds with T3SS inhibitory activity have been identified, few T3SS inhibitors have been validated and mode of action determined. One issue is the lack of standardized orthogonal assays following high throughput screening. Using a training set of commercially available compounds previously shown to possess T3SS inhibitory activity, we demonstrate the utility of an experiment pipeline comprised of six distinct assays to assess the stages of type III secretion impacted: T3SS gene copy number, T3SS gene expression, T3SS basal body and needle assembly, secretion of cargo through the T3SS, and translocation of T3SS effector proteins into host cells. We used enteropathogenic *Yersinia* as the workhorse T3SS-expressing model organisms for this experimental pipeline, as *Yersinia* is sensitive to all T3SS inhibitors we tested, including those active against other T3SS-expressing pathogens. We find that this experimental pipeline is capable of rapidly distinguishing between T3SS inhibitors that interrupt the process of type III secretion at different points in T3SS assembly and function. For example, our data suggests that Compound 3, a malic diamide, blocks either activity of the assembled T3SS or alters the structure of the T3SS in a way that blocks T3SS cargo secretion but not antibody recognition of the T3SS needle. In contrast, our data predicts that Compound 4, a haloid-containing sulfonamidobenzamide, disrupts T3SS needle subunit secretion or assembly. Furthermore, we suggest that misregulation of copy number control of the pYV virulence plasmid, which encodes the *Yersinia* T3SS, should be considered as a possible mode of action for compounds with T3SS inhibitory activity against *Yersinia*.

Keywords: type III secretion system, T3SS, T3SS inhibitor, Yersinia, pYV

## INTRODUCTION

The type III secretion system (T3SS) is a macromolecular nanosyringe used by dozens of Gram negative pathogens, including Yersinia, Shigella, Salmonella, Chlamydia, and Pseudomonas, to inject effector proteins into target host cells (Deng et al., 2017). The core structure of the T3SS consists of a basal body that anchors the entire complex in the bacterial membrane and assembles first. The basal body is composed of three proteins, which oligomerize into rings. SctD and SctJ form two rings in the bacterial inner membrane (IM) and SctC, a member of the secretin family of proteins, forms a ring in the bacterial outer membrane (OM) (Bergeron et al., 2013). These three rings are connected in the center by a hollow inner rod and the IM rings are closely associated with the SctN ATPase complex whose integrity is essential to the T3SS (Diepold et al., 2010). In addition to ATP, the proton motive force is also important for T3SS activity (Wilharm et al., 2004), although how the T3SS harnesses the proton motive force remains unclear. Once active, the basal body secretes the SctF needle protein, which polymerizes into a straight, hollow tube protruding into the extracellular space. In Yersinia species pathogenic to mammals, the fully assembled Ysc T3SS needle is composed of ∼140 SctF subunits, is 65 nm in length, and harbors a tip complex composed of a pentamer of the hydrophilic LcrV translocator protein (Broz et al., 2007). Upon host cell contact, two additional hydrophobic translocator proteins, YopD and YopB, are secreted through the Ysc needle to form a translocon complex that leads to pore formation in the host membrane, facilitating the translocation of effector proteins to the host cytoplasm (Büttner and Bonas, 2002).

The Yersinia Ysc T3SS is highly regulated at the transcriptional, translational, and post-translational levels (Francis et al., 2002; Heroven et al., 2012). The transcription factor LcrF directs transcription of genes encoding the T3SS structural, regulatory, and effector proteins, all of which are encoded on the 70 kb pYV virulence plasmid (Schwiesow et al., 2015). Several factors govern regulation of LcrF expression, including temperature and the transcription factor IscR (Schwiesow et al., 2015). Importantly, pYV copy number increases during active type III secretion, and this is important for Yersinia virulence (Wang et al., 2016). In addition, the T3SS functions on a positive feedback loop in which active secretion leads to upregulated transcription of T3SS genes (Cornelis et al., 1987; Francis et al., 2002), although the mechanism behind this remains unclear.

A number of pathogens require one or more T3SSs for virulence, as genetic ablation causes attenuation in animal models and clinical isolates harbor plasmids or pathogenicity islands that encode T3SS genes (Coburn et al., 2007). An antibody against the Pseudomonas aeruginosa T3SS needle tip protein PcrV is part of a current Phase II clinical trial to treat nosocomial ventilator-associated pneumonia (NCT02696902), indicating that antibodies targeting the T3SS may be used as therapeutics. However, antibodies have low oral bioavailability and must be administered by injection; small molecules with high oral bioavailability are more attractive as therapeutic agents. A number of putative small molecule T3SS inhibitors have been identified in the past 15 years (Duncan et al., 2012; Marshall and Finlay, 2014; Anantharajah et al., 2017), yet only one class of compounds can be considered validated. Many published T3SS inhibitors have off target effects that may underlie their T3SS disruption. For example, the best studied class of T3SS inhibitors, the salicylidene acylhydrazides, are thought to cause deregulation of T3SS genes through an unknown mechanism, yet the activity of some salicylidene acylhydrazides is dependent on iron chelation (Beckham and Roe, 2014). The phenoxyacetamides represent the only class of compounds that inhibit the T3SS in a physiologically relevant cellular context, protect against a bacterial infection (P. aeruginosa abscess formation in mice), and have a validated molecular target, the SctF needle subunit (Bowlin et al., 2014; Berube et al., 2017).

We have developed an experimental pipeline that can be employed to determine initial mode of action for compounds with T3SS inhibitory activity. We chose to use the enteropathogens Yersinia pseudotuberculosis and Yersinia enterocolitica as the workhorses for this assay pipeline because Yersinia is susceptible to the majority of T3SS inhibitors described and because of the wealth of genetic and biochemical tools available. In addition, Yersinia are extracellular pathogens that use their T3SS to prevent phagocytosis, negating the need for a T3SS inhibitor to cross the mammalian cell membrane to inhibit the T3SS-host cell interaction. Importantly, the assays selected for the experimental pipeline had to be amenable to miniaturization, as compound availability is often limiting. Since interfering with T3SS gene expression, basal body assembly, needle assembly, and host cell effector protein translocation could all lead to inhibition of T3SS activity, we designed our pipeline to consist of distinct assays that each measure a specific stage of type III secretion. Inhibitors of LcrF and its homolog ExsA from P. aeruginosa have been described (Marsden et al., 2015). In addition, it is possible that a small molecule with T3SS inhibitory activity could exert its effect by inhibiting secretion-associated increase of pYV copy number, leading to a decrease in T3SS gene expression. Therefore, two pipeline assays measure pYV copy number and T3SS gene expression. Once T3SS genes are expressed, the T3SS basal body assembles followed by the T3SS needle. Two additional pipeline assays measure SctD (YscD in Yersinia) localization to the IM and SctF (YscF) needle assembly. The last two pipeline assays monitor efficiency of T3SS effector protein (Yop) secretion in vitro or Yop translocation into target host cells. Each individual assay on its own provides a limited snapshot of a compound's T3SS inhibitory activity. However, when performed as a pipeline of assays, the resulting data can be used to predict the mode of action of a T3SS inhibitor. Furthermore, the training set of T3SS inhibitors we use to validate this pipeline are commercially available, and therefore can serve as controls to compare the activity of other T3SS inhibitors.

### MATERIALS AND METHODS

#### Compounds and Antibodies

Compound 3 (CAS# 443329-02-0), compound 4 (CAS# 138323-28-1), and INP0007 (CAS# 300668-15-9) were obtained from Chembridge. INP0010 (CAS# 68639-26-9) was obtained from ChemDiv. Compound 20 (CAS# 489402- 27-9) was obtained from TimTek. The anti-YscF antibody was raised in rabbits against the Y. pseudotuberculosis YscF peptide KDKPDNPALLADLQH (Morgan et al., 2017). DMSO concentration did not exceed 0.2%, except where indicated.

#### Bacterial Strains and Growth Conditions

Bacterial strains used in this paper are listed in **Table 1**. Y. pseudotuberculosis and Y. enterocolitica were grown, unless otherwise specified, in 2xYT (yeast extract-tryptone) at 26◦C shaking overnight. In order to induce the T3SS, overnight cultures were diluted into low calcium medium (2xYT plus 20 mM sodium oxalate and 20 mM MgCl2) to an optical density (OD600) of 0.2 and grown for 1.5 h at 26◦C shaking followed by 2–3 h at 37◦C to induce Yop synthesis, depending on the assay, as previously described (Auerbuch et al., 2009).

#### Construction of YopH Transcriptional Reporter

An expression cassette containing the yopH promoter and FLAG (5′ terminus)- and ssrA (3′ terminus)-tagged mCherry was generated using SOEing PCR and cloned into pMMB67EH (Horton et al., 1990; Pettersson et al., 1996; Karzai et al., 2000). The yopH promoter was amplified from pHYopT (gift from J. Bliska) using oligonucleotides SMP425 and SMP426 (**Table 2**). mCherry (gift from R. Tsien) was amplified using oligonucleotides SMP427 and SMP431. FLAG and ssrA sequences were incorporated into oligonucleotides SMP426/SMP427 and 431, respectively. Oligonucleotides SMP425 and SMP431 were used to generate the pyopH-FLAGmCherry-ssrA cassette. PCR product was digested with BamHI and EcoRI and cloned into similarly digested pMMB67EH. Site directed mutagenesis using oligonucleotides SMP437 and SMP438 was used to generate the AAV variant of the ssrA tag.

### Type III Secretion Assay

Visualization of Yersinia T3SS cargo secreted in broth culture by Coomassie staining of SDS-PAGE separated proteins was performed as previously described with some modifications (Auerbuch et al., 2009). After low calcium medium cultures were grown for 1.5 h at 26◦C, compounds or DMSO were added and the cultures shifted to 37◦C for another 2 h. Postincubation cultures were spun down at 15,000 × g for 10 min at room temperature. Supernatants were transferred to a new eppendorf tube. Ten percent final trichloroacetic acid (TCA) was added and the mixture was vortexed vigorously. Samples were incubated on ice for 20 min and then spun down at 15,000 × g for 15 min at 4◦C. The pellet was resuspended in final sample buffer (FSB) with 20% DTT. Samples were boiled for 15 min prior to running on a 12.5% SDS-PAGE gel. Sample loading was normalized for bacterial culture density (OD600) measured prior to centrifugation. Densitometric quantification of the bands was done using Image Lab software (Bio-Rad), setting the relevant DMSO-treated YopE band to 1.00.

The miniaturized secretion assay was done by using Y. pseudotuberculosis expressing a YopM-β-lactamase (YopM-Bla) reporter and the chromogenic β-lactamase substrate nitrocefin to detect secreted YopM-Bla (O'Callaghan et al., 1972; Lee et al., 2007; Green et al., 2016). Y. pseudotuberculosis 1yop6 pYopM-Bla and 1yscNU pYopM-Bla (negative control) were grown overnight at 26◦C with 250 rpm shaking in 2xYT supplemented with chloramphenicol (25µg/ml). The bacterial culture was diluted in low calcium media to OD<sup>600</sup> 0.2, incubated at 26◦C with 250 rpm shaking for 1.5 h, and 25 µl was distributed into a 384-well plate (NuncTM, Thermo Fisher). Fifty micromolars of compounds or equivalent volume of DMSO was added to each well and the plate incubated at 37◦C with 250 rpm shaking for 2 h. The bacterial plate was spun down for 5 min and 10 µl of supernatant transferred to a new plate containing the same volume of nitrocefin (500µg/ml). Absorbance at 490 nm was measured 30 min after addition of nitrocefin. Three technical replicates for each sample were carried out for each independent experiment.

### Quantitative Real-Time PCR

Indicated concentrations of compounds, or the equivalent volumes of DMSO, were added to 26◦C-grown low calcium medium cultures and treated cultures were shifted to 37◦C for 3 h. RNA was isolated using an RNeasy Plus Micro Kit (Qiagen) according to the manufacturer's instructions and 2 µg RNA was used to make cDNA, as previously described (Auerbuch et al., 2009; Miller et al., 2014). SYBR Green PCR master mix (Applied Biosystems) was used for qPCR reactions according to the manufacturer's instructions and a 60◦C annealing temperature, using the 16s rRNA gene as a reference for each sample. Three technical replicates were averaged for each sample/primer pair per independent experiment. Primers used are listed in **Table 3**. Results were analyzed using the Bio-Rad CFX software.

### YopH Transcriptional Reporter Assay

Indicated concentrations of compounds, or the equivalent volumes of DMSO, were added to 26◦C-grown low calcium medium cultures and treated cultures were shifted to 37◦C for 3 h. Two-hundred microliters of cultures were spun down at 3,000 × g for 5 min, resuspended in 200 µl 1X PBS, and mCherry fluorescence and optical density measured in black, clear bottom 96 well plates (Costar <sup>R</sup> , Corning Inc.) on a Perkin Elmer Victor X3 plate reader. Two technical replicates were averaged for each sample per independent experiment.

### Immunofluorescence Staining of YscF

Quantification of YscF staining on the bacterial surface was carried out as described previously (Morgan et al., 2017). Indicated concentrations of compounds, or the equivalent volumes of DMSO, were added to 26◦C-grown low calcium medium cultures and treated cultures were shifted to 37◦C for 3 h. Bacteria were fixed by the addition of a mix of 800 µl 4% paraformaldehyde, 1 µl 25% glutaraldehyde, and 40 µl 0.5 M NaPO<sup>4</sup> pH 7.4 to 500 µl of bacterial culture for 15 min at room temperature followed by 30 min or longer on ice. Fixed bacteria were gently sedimented (5,000 × g for 5 min), washed four times with PBS and stored in 250 mM glucose, 10 mM Tris-HCl pH 7.5, and 1 mM EDTA. Eight microliters of the fixed cells were added to coverslips and allowed to set until just dried. Coverslips

#### TABLE 1 | Bacterial strains used in this study.


TABLE 2 | Oligonucleotides used to generate pYopH FLAG mCherry.


were blocked overnight with PBST with 3% BSA (PBST/BSA) at 4◦C. Blocking solution was removed and anti-YscF primary antibody was added at 1:10,000 in PBST/BSA and rocked at 4 ◦C for 4 h. Coverslips were carefully rinsed in ice cold PBST with 0.1% Tween-20 several times and incubated with Alexa fluor 594 or 488 anti-rabbit secondary antibody (Invitrogen) at 1:10,000 in PBST/BSA and rocked at 4◦C for 3 h followed by rinsing again in ice cold PBST with 0.1% Tween-20 several times. Coverslips were then stained for nuclear material with Hoechst 33342 (Thermo Scientific) at 1:10,000 in PBST/BSA and left in the dark at room temperature for 30 min. Coverslips were washed in ice cold PBST with 0.1% Tween-20 several times, allowed to dry briefly, mounted onto glass coverslips with Prolong Gold (Thermo Scientific), and sealed with clear nail polish. Images were taken with the Zeiss Axioimager Z2 widefield microscope under 63X/1.4 oil immersion using Zen software, pseudocolored, and merged in FIJI. YscF puncta were counted using IMARIS software.

#### YscD Analysis

Y. enterocolitica expressing EGFP-YscD were grown overnight in brain heart infusion (BHI) broth containing nalidixic acid (35µg/ml) and diaminopimelic acid (80µg/ml) (Diepold et al., 2010). Overnight cultures were diluted to an OD<sup>600</sup> of 0.2 in M9 minimal medium supplemented with casamino acids. Cultures were grown for 1.5 h at 26◦C. Compounds were added



at indicated concentrations or the equivalent volumes of DMSO were added and the cultures switched to 37◦C for another 3 h. Two microliters of culture were layered on a patch of 1% agarose in water (Skinner et al., 2013) supplemented with 80µg/ml diaminopimelic acid, 5 mM EDTA, 10 mM MgCl2, and either compounds at indicated concentrations or the equivalent volume of DMSO (Diepold et al., 2010). Images were taken with a Zeiss Axioimager Z2 widefield microscope under 63X/1.4 oil immersion using Zen software, pseudocolored, and merged in FIJI. YscD puncta were counted using IMARIS software.

#### pYV Copy Number

Y. pseudotuberculosis YpIII/pIBXN=<sup>1</sup> and YpIII/pIBX were grown overnight in 2xYT media containing kanamycin (30µg/ml). Overnight cultures were diluted to an OD<sup>600</sup> of 0.2 in M9 supplemented with casamino acids. Cultures were grown for 1.5 h at 26◦C. Compounds were added at indicated concentrations or the equivalent volumes of DMSO were added and the cultures switched to 37◦C for another 3 h. Two hundred microliters of each culture was added to clear bottom, white 96 well plates (Costar <sup>R</sup> , Corning Inc.) and optical density and luminescence were measured using a Perkin Elmer Victor X3 plate reader. Two technical replicates were averaged for each sample per independent experiment.

#### YopM Translocation Assay

Measuring translocation of the YopM-β-lactamase (YopM-Bla) reporter effector protein was carried out as previously described (Duncan et al., 2014). A total of 6 × 10 <sup>3</sup> CHO-K1 cells were plated in each well of a 384-well plate (CorningTM FalconTM) in 50 µl of F-12K medium plus 10% FBS and 1% glutamine, and incubated overnight. The following day, indicated concentrations of compounds, or the equivalent volumes of DMSO, were added to 26◦C-grown low calcium medium cultures of Y. pseudotuberculosis expressing YopM-Bla and treated cultures were shifted to 37◦C for 3 h. Immediately prior to infection of CHO-K1 cells, the F-12K media was removed and replaced with RPMI plus 10% FBS and compounds or DMSO were added. Y. pseudotuberculosis was added to the CHO-K1 plate at an MOI of 7. Five minutes after this transfer, the plate was centrifuged at 290 × g for 5 min to initiate bacterium-host cell contact and incubated for 1 h at 37◦C and 5% CO2. Thirty minutes prior to the end of the infection, CCF2-AM (Invitrogen) was added to each well, and the plate covered in foil and incubated at room temperature. At the end of the infection, the medium was aspirated and fresh 4% paraformaldehyde added to each well for 20 min to fix the cells. The paraformaldehyde was then aspirated and the DNA dye DRAQ5 (Cell Signaling Technology) in PBS was added to each well. The monolayers were incubated at room temperature for 10 min, washed once with PBS, and visualized using an ImageXpressMICRO automated microscope and MetaXpress analysis software (Molecular Devices). The number of YopM-Bla-positive cells was calculated by dividing the number of blue (CCF2-cleaved) cells by the number of green (total CCF2) cells. Three technical replicates were averaged for each sample per independent experiment.

#### Growth Curves

Overnight cultures of Y. pseudotuberculosis were diluted to an OD<sup>600</sup> of 0.1 in 2xYT and 200 µl was added to each well of a 96-well plate (CorningTM FalconTM). Compounds were added at 50µM and kanamycin at 50µg/ml, plates incubated at 26◦C, and the OD<sup>600</sup> of the cultures measured every hour for 13 h using a VersaMax Tunable Microplate Reader (Molecular Devices). The 96-well plates were intermittently shaken throughout the experiment. One technical replicate was used for each sample per independent experiment.

#### Motility Assay

Following overnight growth, 1 µl of Y. pseudotuberculosis overnight culture was spotted onto motility medium containing either 1% tryptone/0.25% agar or 1% tryptone/0.25% agar supplemented with 5 mM EGTA and 20 mM MgCl<sup>2</sup> in sixwell plates. Each well contained either 0.3% DMSO or 50µM INP0007 or INP0010. The plates were incubated at 26◦C for 24 h before the diameter of swimming motility was measured. One technical replicate was used for each sample per independent experiment.

### RESULTS

#### T3SS Inhibitor Training Set for Assay Pipeline

We selected a training set of commercially-available compounds to validate our assay pipeline (**Table 4**). All compounds selected TABLE 4 | Compounds used in this study and their reported activity.


met two strict requirements: demonstrated inhibitory activity on the Yersinia T3SS and absence of bactericidal activity within 6 h to accommodate the timeframe of the pipeline assays (**Figures 1A–C**).

Compound 4 (C4), a haloid-containing sulfonamidobenzamide, was identified through a luciferasebased T3SS gene promoter fusion screen as a potent inhibitor of T3SS gene expression (Kauppi et al., 2003). We chose this compound because it was shown, albeit modestly, to inhibit expression of the LcrF master regulator of the T3SS in Yersinia (Kauppi et al., 2003) and we therefore expected that it would have a broad impact on all stages of type III secretion downstream of T3SS expression in our assay pipeline. From the same high throughput screening strategy that identified C4, two members of the salicylidene acylhydrazide class of inhibitors, INP0007 (Kauppi et al., 2003) and INP0010 (Nordfelth et al., 2005), were identified. Despite a large number of studies, the mechanism of action of these compounds remains unclear and there is evidence of off target effects on global virulence gene expression by at least some salicylidene acylhydrazides (Tree et al., 2009). In fact, INP0010 decreased Yersinia growth starting 7 h after initiation of treatment (**Figure 1D**), consistent with previous reports (Veenendaal et al., 2009). Compound 3 (C3), a malic diamide, was shown to inhibit ExoS secretion in P. aeruginosa and YopE secretion in Yersinia pestis (Aiello et al., 2010). This compound is structurally related to another class of T3SS inhibitors, the phenoxyacetamides, which were proposed to target the PscF needle subunit in Pseudomonas (Bowlin et al., 2014). Based on this structural relatedness, C3 was expected to disrupt T3SS needle assembly without impacting T3SS gene expression. Compound 20 (C20) was identified in a T3SS effector protein-β-lactamase reporter translocation screen, but had no significant effect on the ability of the bacteria to secrete T3SS effectors in the absence of host cells, pointing to the possibility that this compound specifically blocks the bacteria–host cell interaction (Harmon et al., 2010). Therefore, C20 was expected to inhibit the translocation of effector proteins into host cells but not impact T3SS gene expression and assembly.

An advantage of using Ysc-expressing Yersinia for our assay pipeline is that removing calcium from the culture medium enables T3SS effector proteins to be secreted into the supernatant in the absence host cells (Yother and Goguen, 1985; Perry et al.,

FIGURE 1 | Efficiency of T3SS effector protein secretion and bacterial growth in the presence of T3SS inhibitors. (A,B) The relative efficiency of effector protein secretion into the culture supernatant was analyzed following bacterial growth for 2 h under T3SS-inducing conditions in the presence of either 50µM compound or equivalent volume of DMSO. (A) The secretome of *Y. pseudotuberculosis* IP2666 and *Y. enterocolitica* 8081 was precipitated with trichloroacetic acid, separated by SDS-PAGE, and visualized by staining with Coomassie blue. Samples were normalized to culture optical density. (1) DMSO, (2) Compound 3, (3) Compound 4, (4) INP0007, (5) INP0010, (6) Compound 20. (B) Quantification of the YopE protein band by densitometry relative to the DMSO control. The average of 3 (*Y. pseudotuberculosis*) or 4 (*Y. enterocolitica*) biological replicates ± standard deviation is shown. (C,D) *Y. pseudotuberculosis* IP2666 growth at 26◦C in the presence of 50µM compound or DMSO was tracked by measuring optical density. The average of three biological replicates ± standard deviation is shown and statistical significance is represented comparing compounds relative to the DMSO control. \**P* < 0.03; \*\**P* < 0.004; \*\*\*\**P* < 0.0001; \*\*\**P* < 0.0007 (one way ANOVA with Dunnett's *post-hoc* test).

1986; Straley and Bowmer, 1986; Sample et al., 1987; Forsberg and Wolf-Watz, 1988; Mehigh et al., 1989), providing a useful method of monitoring efficiency of T3SS inhibition in order to validate use of a given inhibitor in our assay pipeline. As expected, all test compounds except C20 exhibited significant inhibition of effector Yop secretion (**Figures 1A,B**). Importantly, this secretion assay was used routinely to ensure the efficacy of each batch of compound purchased and after 1 month in storage. Therefore, even when a compound had no effect in a particular assay, we could be confident that it was still active as a T3SS inhibitor.

#### Assessment of T3SS Gene Expression

We reasoned that inhibition of the pYV copy number upregulation that occurs during normal induction of type III secretion in Yersinia may impact T3SS gene expression, and therefore affect overall T3SS activity. To test this, we employed a Y. pseudotuberculosis strain with a luciferase reporter gene cluster integrated onto the pYV virulence plasmid (YpIII/pIBX) (Fahlgren et al., 2014). As a negative control, we used a strain in which a replication deficient copy of the virulence plasmid as well as the luciferase gene cluster was integrated into the chromosome (referred to as YpIII/pIBX<sup>N</sup> <sup>=</sup> <sup>1</sup>) (Wang et al., 2016). In this YpIII/pIBX<sup>N</sup> <sup>=</sup> <sup>1</sup> strain, the ratio of T3SS and luciferase reporter genes vs. chromosomal genes remains constant even upon induction of type III secretion (Wang et al., 2016). As expected, expression of lcrF, the effector protein yopE, the integral IM ring protein yscD, and the needle subunit yscF were significantly decreased in the YpIII/pIBX<sup>N</sup> <sup>=</sup> <sup>1</sup> strain compared to the YpIII/pIBX strain (**Figure 2A**). While these data demonstrate that inhibition of pYV gene copy number could impact pYV gene expression, four of the five compounds in our training set did not significantly affect virulence plasmid copy number as measured by luciferase activity, while INP0007 significantly increased copy number (p = 0.0054; **Figure 2B**) for reasons that are unclear but that may reflect off target effects. In contrast, luciferase expression could be inhibited by blocking general transcription or translation using rifampicin or chloramphenicol (**Figure 2B**). Therefore, while inhibiting pYV copy number in Yersinia causes a decrease in T3SS activity, we conclude that none of the compounds in our training set act by inhibiting upregulation of pYV copy number.

In order to determine if the training set compounds negatively impacted T3SS gene expression, we used a Yersinia reporter strain containing an unstable YopH-mCherry-AAV transcriptional reporter (Andersen et al., 1998). C4, INP0007, and INP0010 inhibited YopH-mCherry-AAV expression when T3SS activity was induced by incubation in low calcium medium at 37◦C (**Figure 3A**). In order to validate and extend these results, we measured transcript levels of T3SS and non-T3SS genes by qPCR. Consistent with the YopH-mCherry-AAV data, C3 and C20 had no significant impact on gene expression (**Figure 4**). However, C4 inhibited expression of all T3SS genes tested, but did not impact expression of two non-T3SS-associated genes: the iron sulfur cluster loading protein erpA and the small ribosomal subunit L9. Interestingly, INP0007 and INP0010 significantly inhibited transcript levels of lcrF, the needle tip subunit lcrV, and effector proteins yopH, yopE, and yopK, but did not inhibit expression of the ATPase yscN, yscF, and yscD under these conditions (**Figure 4**). To test whether inhibition of secretion via the positive feedback loop indirectly inhibits T3SS gene expression, we measured YopH-mCherry fluorescence and yscD mRNA levels via qPCR during growth at 37◦C in high calcium (**Figures 3B,C**). Under these conditions, Yersinia can assemble the T3SS (Diepold et al., 2010), but no secretion of Yop effectors occurs (Forsberg and Wolf-Watz, 1988); therefore, the positive feedback loop is not active in high calcium. While overall expression levels of YscD and YopH are lower in high calcium compared to low calcium conditions, C4, INP0007, and INP0010 did not significantly decrease YopH-mCherry-AAV fluorescence in high calcium conditions (**Figure 3B**), suggesting an indirect effect of these compounds on T3SS gene expression through inhibition of secretion in low calcium conditions. Likewise, yscD levels were not decreased by C4 in high calcium (**Figure 3C**). Importantly, yscD expression was significantly lower in the 1lcrF mutant compared to WT under high calcium conditions in the absence of compound. These data indicate that C4 is not likely to target LcrF activity, as yscD is predicted to be under LcrF control (Schwiesow et al., 2015).

#### Assessment of T3SS Assembly

In order to monitor T3SS assembly, we used fluorescence microscopy to quantify YscD localization in Y. enterocolitica and YscF puncta formation in Y. pseudotuberculosis as a proxy for T3SS basal body and T3SS needle assembly, respectively, as previously described (Davis and Mecsas, 2007; Diepold et al., 2010; Morgan et al., 2017). Only INP0007 significantly impacted YscD-EGFP puncta formation (**Figure 5**). Importantly, the level of diffuse fluorescence in INP0007-treated Yersinia was greater than that of the non-T3SS-inducing condition (26◦Cgrown Yersinia; **Figure 5A**, inset), indicating that while YscD is expressed at 37◦C in the presence of INP0007, it is not assembled into basal bodies in the presence of this compound. In contrast, while C3 and C20 had no significant effect on YscF puncta formation as measured with an anti-YscF antibody, C4, INP0007, and INP0010 significantly inhibited needle assembly (**Figure 6**).

The bacterial flagellar apparatus is composed of a basal body that is structurally related to the injectisome T3SS basal body and mediates secretion of the flagellar hook and filament proteins (Macnab, 2004). Therefore, it is possible that compounds with the ability to inhibit injectisome T3SS basal body assembly may inhibit flagellar assembly and therefore flagellar motility. INP0007 has been shown to inhibit motility in Yersinia (Kauppi et al., 2003), but neither INP0007 nor INP0010 inhibited motility in Salmonella (Negrea et al., 2007). In our standard Yersinia motility agar (Morgan et al., 2017), INP0007 significantly decreased flagellar motility in Y. pseudotuberculosis, while INP0010 decreased motility weakly albeit significantly (**Figure 7**). Addition of the chelating agent EGTA in the motility agar, as used in the Kauppi et al study (Kauppi et al., 2003), led to an even greater inhibitory effect on motility by INP0010 (**Figure 7**). The Y. pseudotuberculosis flhDCYpestis non-motile strain served as the negative control for the motility assay. Taken

conditions in 50µM compound or equivalent volume of DMSO and T3SS gene expression evaluated by qPCR. GOI, gene of interest. Expression levels were normalized to 16s rRNA and then the fold change compared to DMSO calculated [(GOIcompound/16scompound)/(GOIDMSO/16sDMSO)]. The average of four biological replicates ± standard deviation is shown. \*\*\*\**P* < 0.0001, \*\**P* < 0.008 (Student *T*-test). (B) *Y. pseudotuberculosis* YpIII/pIBX and YPIII/(cIBX)*n*=<sup>1</sup> were grown under T3SS-inducing conditions for 3 h in 50µM compound or equivalent volume of DMSO and luminescence measured as a readout of pYV gene copy number. The average of four biological replicates ± standard deviation is shown and statistical significance is represented comparing compounds relative to the DMSO control. \*\*\*\**P* < 0.0001; \*\**P* = 0.008 (one way ANOVA with Dunnett's *post-hoc* test).

FIGURE 4 | T3SS gene mRNA levels under low and high calcium conditions. Wildtype *Y. pseudotuberculosis* was grown under high or low calcium conditions in the presence of 50µM compound or equivalent volume of DMSO and mRNA levels of T3SS (*lcrF*, *yscN*, *yscD*, *yscF*, *lcrV*, *yopE*, *yopK*, *yopH*) and non-T3SS genes (L9, *erpA*) assessed by qPCR. The average of four biological replicates ± standard deviation is shown and statistical significance is represented comparing compounds relative to the DMSO control for each gene. \*\*\*\**P* < 0.0001, \*\*\**P* < 0.0005, \*\**P* < 0.003; \**P* < 0.022 (one way ANOVA with Dunnett's *post-hoc* test). ns, not significant.

together, these data indicate that INP0007 and INP0010 impact both the flagellar and injectisome T3SS.

### Assessment of Effector Protein Secretion in vitro

While the Yop in vitro secretion assay shown in **Figure 1** is critical for assessing the impact of a compound on overall Yop

secretion, the method used (Coomassie staining of SDS-PAGE separated proteins precipitated from the culture supernatant) is not amenable to a high throughput format. In order to assess Yop secretion in vitro in microtiter plates, Y. pseudotuberculosis expressing a YopM-β-lactamase (YopM-Bla) reporter was used in conjunction with the chromogenic β-lactamase substrate nitrocefin (O'Callaghan et al., 1972; Lee et al., 2007; Green et al., 2016). Consistent with the block in Yop secretion observed in **Figure 1**, C3, C4, INP0007, and INP0010 significantly blocked YopM-Bla secretion into the culture supernatant while C20 did not (**Figure 8**). Surprisingly, INP0010 blocked YopM-Bla secretion by only 32% as measured using nitrocefin (**Figure 8**), yet appeared to have a much more significant effect on native YopM secretion as measured using Coomassie staining (**Figure 1**). The reason for this discrepancy is unclear; however, we note that INP0010 has cytotoxic effects on bacteria and eukaryotic cells (see below) and is likely to have the greatest off target effects of the compounds in our training set.

#### Assessment of Effector Protein Translocation

One of the most important requirements of a T3SS inhibitor is its ability to prevent translocation of effector proteins into target host cells. We employed a different type of YopM-Bla reporter assay to assess translocation of effector proteins into CHO K1 cells loaded with CCF2, a fluorescent β-lactamase substrate that can enter eukaryotic cells (Dewoody et al., 2011). While C3, C4, C20, INP0007, and INP0010 significantly blocked YopM-Bla

translocation at 50µM, INP0010 appeared to be cytotoxic at this concentration (**Figure 9**).

#### DISCUSSION

A methodical approach to primary characterization of inhibitors for the T3SS has been lacking in the field. Here, we describe a pipeline of miniaturized assays, using enteropathogenic Yersinia as the workhorse organism, that enable rapid, initial characterization of the stage of T3SS expression, assembly, or function targeted by compounds with T3SS inhibitory activity. Furthermore, we used a training set of compounds with previously-identified T3SS inhibitory activity to test the utility of this pipeline (**Table 5**). INP0007 blocked T3SS basal body and needle assembly and therefore inhibited Yop secretion and translocation. C4 and INP0010 did not prevent basal body assembly but blocked needle assembly, secretion, and translocation. However, INP0010 exhibited cytotoxicity, complicating interpretation of some assay results. C3 allowed T3SS assembly but prevented Yop secretion and translocation. Lastly, C20 allowed T3SS assembly and Yop secretion in vitro, but blocked translocation of Yops into target host cells. These results demonstrate the ability of our assay pipeline both to validate T3SS inhibitors and provide testable hypotheses on their mode of action.

C3 blocked the ability of the Yersinia T3SS to secrete T3SS effector proteins into low calcium culture supernatant and to translocate effector proteins into target host cells. Yet C3 did not block T3SS basal body and needle assembly, as determined by imaging YscD basal body and YscF needle puncta formation. C3 showed a modest inhibition of YopE secretion (28%), but notably inhibited secretion of the translocator protein YopD (**Figure 1A**). These data provide a possible explanation for why C3 strongly inhibited Yop translocation into host cells (40%)



*Dark gray boxes represent statistically significant inhibitory activity at a distinct stage of T3SS deployment. Light gray boxes represent significant inhibitory activity on T3SS gene expression only under low calcium conditions, when the positive feedback loop is active and strong inhibition of secretion leads to a decrease in T3SS gene expression. <sup>a</sup>Compared to DMSO control, as measured by pYV-encoded luciferase activity.*

*<sup>b</sup>Compared to DMSO control under high calcium conditions, as measured by qPCR or monitoring YopH-mCherry expression.*

*<sup>c</sup>Under low calcium conditions.*

*<sup>d</sup>Compared to DMSO control, as measured by nitrocefin color change resulting from YopM-Bla secreted into the culture supernatant under low calcium conditions.*

*<sup>e</sup>Percent of cells with* β*-lactamase substrate CCF2 blue fluorescence in CHO cells.*

despite less potent inhibition of YopE secretion, as YopD is required for Yop delivery into host cells (Rosqvist et al., 1994) C3 has structural similarities with the phenoxyacetamides, which target the SctF needle subunit (Aiello et al., 2010; Bowlin et al., 2014). However, our data indicate that C3 blocks a specific activity of the assembled T3SS. Alternatively, the T3SS needle structure may be altered in the presence of C3, still allowing recognition by our anti-YscF antibody but impeding cargo egress.

In contrast to C3, C20 specifically blocked translocation of a T3SS effector protein into host cells without blocking any other stage of type III secretion, including T3SS effector protein secretion into low calcium culture. As C20 was previously shown to inhibit only translocation of T3SS effector proteins into target host cells (Harmon et al., 2010), our verification of this finding further demonstrates the robustness of our experimental pipeline. Harmon et al. hypothesized that C20 inhibited the host cell-bacterial interaction, as C20 strongly inhibited adherence of Y. pseudotuberculosis to HEp-2 cells. Possible modes of action of compounds that inhibit Yop translocation but not any other stage of type III secretion include interruption of bacterial adhesins, host integrin receptors, or YopBD-mediated pore formation on the host membrane.

C4 was shown to inhibit promoter activity of lcrF, the Ysc T3SS master regulator (Kauppi et al., 2003). However, while we observed a C4-dependent decrease in mRNA levels of all T3SS genes tested under low calcium T3SS-inducing conditions, C4 treatment did not impact T3SS gene expression under high calcium conditions when T3SSs are assembled but no Yop secretion occurs. As a 1lcrF mutant had significantly less T3SS gene expression under high calcium conditions compared to wildtype Yersinia, this argues against C4 inhibiting LcrF activity. C4 did not alter the number of YscD puncta observed but did decrease the number of YscF puncta per cell by greater than half. These data suggest that C4 does not impact YscD assembly in the plasma membrane but disrupts overall T3SS basal body formation in such a way that needle formation is compromised or interferes with YscF secretion or assembly.

The salicylidene acylhydrazides INP0007 and INP0010 inhibited expression of lcrF, lcrV, and yopKEH, but did not decrease expression of yscN, yscD, or yscF. While INP0007 and INP0010 are related structurally, INP0010 prevents Yersinia growth if the bacteria are exposed longer than 6 h, while INP0007 does not affect bacterial growth even up to 13 h of exposure. Surprisingly, while INP0010 did not inhibit YscD puncta formation, INP0007 caused a 10-fold decrease in YscD puncta. Both compounds significantly inhibited YscF puncta formation, Yop secretion into low calcium culture, and Yop translocation into host cells. However, measuring Yop translocation into host cells in the presence of INP0010 is complicated by the obvious toxic effects of the compound on host cells. Therefore, utilization of the Yop translocation assay of our experimental pipeline is limited to compounds without toxic effects on host cells. However, this also shows the benefit of using microscopy to observe CCF2 green-to-blue fluorescence conversion, as it allows observation of host cell morphology. The salicylidene acylhydrazide class of compounds have been suggested to have a broad impact on expression of horizontally acquired genes (Tree et al., 2009). However, we only observe a defect in T3SS gene expression in low calcium, but not high calcium medium, for INP0007. In addition, we observe a significant defect in flagellar motility by INP0007, suggesting a broader effect of the compound either on the flagellar T3SS, ATP synthesis, or the proton motive force. In total, our results argue that INP0007 inhibits formation of the injectisome T3SS and, either through a similar mechanism or by targeting multiple pathways, also disrupts flagellar T3SS activity. Therefore, we suggest that for T3SS inhibitors that block T3SS basal body assembly, a flagellar motility assay be performed to assess their breadth of action.

Given the previously-demonstrated positive feedback exerted by active type III secretion on T3SS gene transcription (Cornelis et al., 1987), we expected that C3 would inhibit T3SS gene

expression in low calcium medium. C4 and INP0007 inhibited YopE secretion by >70–80% and this correlated with the ability to affect T3SS gene expression under low calcium conditions. C3 inhibited YopE secretion <50% and this was insufficient to impact the feedback loop on T3SS gene expression. Similarly, the T3SS inhibitor piericidin A1 inhibits YopE secretion by ∼40–45% and also does not repress transcription of T3SS genes (Morgan et al., 2017). These data suggest that only compounds that potently inhibit Yop secretion will affect the T3SS gene expression feedback loop. This indicates that screening strategies based on Yop gene expression as the readout for T3SS inhibition may miss less robust compounds that could be improved by structure-activity relationship analysis.

While none of the inhibitors we tested affected pYV copy number, our data shows that prevention of pYV plasmid copy number upregulation during active type III secretion leads to dramatically reduced T3SS gene mRNA steady-state levels. Therefore, when using Yersinia as a model organism for assessing T3SS inhibitor mechanism of action, it is important to consider pYV copy number. Furthermore, as chloramphenicol and rifampicin significantly inhibited luminescence as a readout of pYV copy number, this assay also sheds light on whether a compound affects general transcription or translocation. One caveat is that compounds that act as luminescence quenchers would decrease luminescence in this assay. In this case, analyzing expression of non-T3SS genes, such as the L9 and ErpA genes shown here, via qPCR can serve as a way to test this possibility.

In summary, we propose that the experimental pipeline described here can be used to rapidly bin T3SS inhibitors into categories depending on the stage of type III secretion they inhibit, providing testable hypotheses on mode of action. The commercially available training set shown here can be used to establish these assays in other labs and perhaps enable a more standardized approach to T3SS inhibitor research.

## AUTHOR CONTRIBUTIONS

JM: experimental design, performing experiments, writing paper; HL: experimental design, performing experiments; JD and JL: performing experiments; SM, RI, HW: providing reagents; VA: experimental design, writing paper.

## ACKNOWLEDGMENTS

We thank Benjamin Abrams of the UCSC Life Sciences Microscopy Center and Walter Bray of the UCSC Chemical Screening Center for technical support. We thank Andreas Diepold for providing the fluorescent YscD Y. enterocolitica strain. Research reported in this publication was supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under Award Numbers R01AI106930 and R01AI119082. JM was supported by the National Human Genome Research Institute of the National Institutes of Health under Award Number 4R25HG006836- 04. RI was supported by NIAID R01AI110684. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.


on type III secretion in Escherichia coli o157:H7. Infect. Immun. 77, 4209–4220. doi: 10.1128/IAI.00562-09


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Morgan, Lam, Delgado, Luu, Mohammadi, Isberg, Wang and Auerbuch. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Discovering RNA-Based Regulatory Systems for Yersinia Virulence

Vanessa Knittel † , Ines Vollmer † , Marcel Volk † and Petra Dersch\*

Department of Molecular Infection Biology, Helmholtz Centre for Infection Research, Braunschweig, Germany

The genus Yersinia includes three human pathogenic species, Yersinia pestis, the causative agent of the bubonic and pneumonic plague, and enteric pathogens Y. enterocolitica and Y. pseudotuberculosis that cause a number of gut-associated diseases. Over the past years a large repertoire of RNA-based regulatory systems has been discovered in these pathogens using different RNA-seq based approaches. Among them are several conserved or species-specific RNA-binding proteins, regulatory and sensory RNAs as well as various RNA-degrading enzymes. Many of them were shown to control the expression of important virulence-relevant factors and have a very strong impact on Yersinia virulence. The precise targets, the molecular mechanism and their role for Yersinia pathogenicity is only known for a small subset of identified genus- or species-specific RNA-based control elements. However, the ongoing development of new RNA-seq based methods and data analysis methods to investigate the synthesis, composition, translation, decay, and modification of RNAs in the bacterial cell will help us to generate a more comprehensive view of Yersinia RNA biology in the near future.

#### Edited by:

Matthew S. Francis, Umeå University, Sweden

#### Reviewed by:

Kai Papenfort, Ludwig-Maximilians-Universität München, Germany Satish Raina, Gdansk University of Technology, Poland Erik Holmqvist, Uppsala University, Sweden

#### \*Correspondence:

Petra Dersch petra.dersch@helmholtz-hzi.de

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Molecular Bacterial Pathogenesis, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 31 July 2018 Accepted: 05 October 2018 Published: 25 October 2018

#### Citation:

Knittel V, Vollmer I, Volk M and Dersch P (2018) Discovering RNA-Based Regulatory Systems for Yersinia Virulence. Front. Cell. Infect. Microbiol. 8:378. doi: 10.3389/fcimb.2018.00378 Keywords: RNA thermometer, RNA stability, RNA processing, Csr/Rsm system, virulence, gene regulation, small regulatory RNAs

#### INTRODUCTION

For a long time it was thought that the RNA make-up and biology of bacteria is rather simple and constitutes mainly of mRNAs, tRNAs, and rRNAs and a few specific RNAs such as the tmRNA/SsrA of the ribosome rescue system. However, over the past 10 years, research on bacterial RNA molecules has proven this to be incorrect, since bacteria possess a huge variety of RNAs. A large repertoire of cis- and trans-acting non-coding RNAs has been identified and a plethora of different RNA-based regulatory mechanisms and functions has been unveiled mainly in Enterobacteriaceae, but lately also in many other prokaryotes. This showed that RNA-based gene expression control is highly complex and global, and almost as diverse and multifarious as transcriptional control. Multiple structured RNA elements have been identified in the 5′ -untranslated regions of mRNAs sensing temperatures (RNA thermometers) or metabolites (riboswitches) in order to modulate mRNA translation and/or stability (Kortmann and Narberhaus, 2012; Sherwood and Henkin, 2016; McCown et al., 2017). Moreover, a remarkably large number of small regulatory RNAs (sRNAs) exist in bacteria (i.e., 200–300 in Enterobacteriaceae, Barquist and Vogel, 2015; Nuss et al., 2015), which control mRNA expression and decay often in concert with RNA-binding proteins. Among the most prominent RNA-binding proteins are the RNA chaperone Hfq and the carbon storage regulator (Csr) protein CsrA, as well as RNases. Clever variations of different global RNA-sequencing (RNA-seq)-based techniques (**Figure 1**) used to define the overall RNA-protein interactome and the sRNA-target networks demonstrated that almost 50% of the bacterial mRNAs are subjected to sRNA-mediated regulation (Melamed et al., 2016; Hör and Vogel, 2017; Waters et al., 2017).

The global analysis of RNA-based control strategies of virulence genes was started with a relatively small number of model bacteria. This included human pathogenic Yersinia species, such as Y. pestis, the causative agent of bubonic plague and the two enteric pathogens Y. pseudotuberculosis and Y. enterocolitica causing gut-associated diseases (Yersiniosis), i.e., ranging from enteritis, watery diarrhea, mesenteric lymphadenitis to post-infectious extraintestinal sequelae (Bottone, 1997; Koornhof et al., 1999; Smego et al., 1999; Valentin-Weigand et al., 2014). System level RNA-seq has the advantage that the entire transcriptome of an organism with all included RNA functions can be studied in a cellular and particular environmental (e.g., infection, virulence conditions) context. This allowed a single-nucleotide resolution and a dynamic view of the pathogen's gene expression, which are summarized and discussed in the present review. The establishment of comprehensive RNA maps of Y. pseudotuberculosis highlighted not only the breadth and large variety of RNA species, it also revealed new species-specific and conserved principles of RNA-based, post-transcriptional control strategies that contribute to the dynamic, spatial, and fine-tuned control of bacterial pathogenicity.

#### GLOBAL TRANSCRIPTIONAL ANALYSIS TO ELUCIDATE FITNESS-RELEVANT TRAITS OF YERSINIA

In order to understand how the functional status of Yersinia is modulated in response to environmental conditions sensed outside or inside their hosts, detailed knowledge about the bacterial gene expression profile in different surroundings is required. RNA-seq is extremely well-suited for this task due to the high sensibility, high resolution and the possibility for highthroughput analysis of multiple in vitro and in vivo conditions (Kröger et al., 2013; Nuss et al., 2015, 2017a; **Figures 1**, **2**). In particular transition from exponential to stationary phase and shifts from moderate temperatures to 37◦C, mimicking host entry, were found to induce reprogramming of a large set of virulence genes. This includes colonization factors and immune defense mechanisms, encoded on the chromosome and the virulence plasmid, but also many catabolic/energy production genes. For instance, this uncovered the existence of a thermoregulated "acetate switch," which seems to prime the bacteria for growth in the digestive tract. This physiological switch occurs when bacteria shift from rapid growth in which they produce acetate from acetogenic carbon sources, e.g., glucose, to a metabolic program of slower growth, which is facilitated by the import and utilization of acetate (Nuss et al., 2015). This is accompanied by a thermo-induced up-regulation of multiple transport and catabolic genes for simple sugars. As the mammalian intestine is rich in short-chain fatty acids, in particular acetate, produced by the intestinal microbiota through consumption of available polysaccharides from diet and hostderived mucus (Cummings and Macfarlane, 1997), flipping the switch may facilitate ultilization of these microbiota-derived degradation products.

The RNA-seq approach also allowed us to elucidate the regulatory architecture linking nutritional status to virulence. We identified a massive remodeling of the CRP-controlled network in response to temperature and discovered CRP as a transcriptional master regulator of numerous conserved and newly identified non-coding RNAs which participate in this process (Nuss et al., 2015). This finding highlighted a new level of complexity of the regulatory network in Yersinia. The concerted action of several transcriptional regulators and multiple non-coding RNAs controlled by CRP adjusts Yersinia fitness and virulence to the requirements of their environmental and virulent life-styles. This discovery showed that CRP is an integral component of a regulatory network that controls life-style switching and highlighted the power of comparative RNA-seq analysis.

Parallel analysis of the transcriptomes of Y. pseudotuberculosis during mouse infection (Tissue Dual RNA-seq, **Figure 2**) has further allowed us to study host-pathogen interaction. It enabled us to correlate gene expression changes of the pathogen with those occurring in the host and revealed genes and regulatory RNAs that are predominantly expressed or repressed during the infection (Avican et al., 2015; Nuss et al., 2015, 2017a). Numerous alterations of host transcripts associated with IL-6-triggered inflammatory and acute phase responses (Mmp8 induction), coagulative activities, metal ion sequestration, tissue repair and damage highlighted that the immune response in the gut-associated lymphoid follicles (Peyer's patches) upon a Y. pseudotuberculosis infection is dominated by infiltration of neutrophils and dendritic cells, and elicits a mixed TH1/TH17 response (Nuss et al., 2017a; Heine et al., 2018). In response, Yersinia increases the gene and expression dose of the virulence plasmid-encoded Ysc type III secretion system (T3SS) and the anti-phagocytic Yop effector proteins (Yops) to prevent the phagocytic attack. In addition, the bacteria induce highaffine ion (Fe2<sup>+</sup> and Zn2+) sequestration systems, diverse stress response, and nutrient uptake systems to counteract ion deprivation, radical/oxidative stress, anti-microbial peptides, and nutrient restraints (Nuss et al., 2017a). Tissue Dual RNA-seq further showed that the outcome of the observed pathogenhost interactions could vary significantly when expression of certain pathogenicity factors is reduced or abrogated. The cytotoxic necrotizing factor Y (CNFY) expressed by some Y. pseudotuberculosis strains, e.g., YPIII and IP2666, is one of these factors. Suppression of CNF<sup>Y</sup> triggers interferonγ mediated responses fostering non-inflammatory bactericidal activities. Moreover, it promotes Ido1- and Arg1-mediated Tregs activation and proliferation and other immune suppressing and tolerogenic mechanisms that avoid systemic inflammation, prevent tissue destruction and drive this Y. pseudotuberculosis strain into persistency (Heine et al., 2018). This process is accompanied by a preterm reprogramming of the pathogen's transcriptome. Anaerobic and multiple acid and oxidative stress resistance genes (e.g., arcA, fnr, frdA, and wrbA) are already induced in the acute phase. It is assumed that this priming gives the bacteria a fitness edge against the host immune defense and facilitates establishment of a persistent microbiota-type life style (Heine et al., 2018). In parallel, the expression profile of many virulence genes changes toward a pattern seen at

moderate temperature under in vitro growth conditions. Upregulation of flagella and adhesion/invasion genes (e.g., invA) was observed during the establishment of persistence, whereas genes encoding important components of the T3SS and the Yops were down-regulated. This is accompanied by a change in the expression of the transcription factors of the Crp-CsrA-RovM-RovA regulatory cascade, indicating that these control factors determine the expression switch (Avican et al., 2015; Heine et al., 2018). Our findings suggest a life-style model for Y. pseudotuberculosis in which the bacteria change their gene expression profile from a virulent to an adapted phenotype, capable of persisting and spreading by fecal shedding. This further illustrates that the developed Tissue Dual RNA-seq approach is well suited to decipher the complexity of hostpathogen interactions during different stages of the infection. The comparative analysis between different strain isolates or mutants of potential virulence-relevant factors and regulators will help to decipher the complex interplay between of Y. pseudotuberculosis and the host response within distinct host niches.

#### ANNOTATION AND CHARACTERIZATION OF YERSINIA TRANSCRIPTS

Since RNA-seq provided a single-nucleotide resolution transcriptional landscape of Y. pseudotuberculosis, it was used to map all genes of the chromosome and virulence plasmid (Nuss et al., 2015, 2017a; **Figure 2**). This led to the identification of new, so far not annotated and previously overlooked open reading frames (ORFs) and revealed a large set of antisense RNAs and small cis- and trans-encoded regulatory RNAs (Koo et al., 2011; Nuss et al., 2015, 2017a). Some of the identified antisense RNAs are long and cover whole genes or multiple genes of the virulence plasmid (Nuss et al., 2015). However, their individual impact on gene expression is still unclear. Moreover, certain variations of the RNA-seq protocols (e.g., treatment of the RNA with the terminator 5′ -phosphate-dependent exonuclease short TEX), led to an enrichment of primary transcripts with 5′ -triphosphate ends. Comparative analysis between the transcriptomes obtained with or without TEX treatment, called differential RNA-seq (dRNA-seq, Sharma and Vogel, 2014, **Figures 1**, **2**), allowed global mapping of the transcription start sites of all transcripts. This led to the identification of promoter elements and the definition of operon structures, and improved the annotation of the Yersinia genome and virulence plasmid (Nuss et al., 2015).

### THE YERSINIA RNA STRUCTUROME

In contrast to double-stranded DNA, RNA is a single-stranded molecule. However, it can undergo significant intramolecular

base-pairing to take in an individual three-dimensional structure. It is known for quite some time that the intrinsic structure of an RNA impacts many cellular processes, most frequently by modulating its translation and/or decay. Protein synthesis is mainly affected by RNA structures, which influence the accessibility of the ribosome binding site (RBS). Based on the whole-genome transcriptomics study of Y. pseudotuberculosis several potential cis-regulatory RNA elements have been identified in long 5′ -untranslated regions (5′ -UTRs; >200 nt) of virulence, stress adaptation and metabolic genes in Y. pseudotuberculosis (Nuss et al., 2015). A bioinformatics approach identified 19 riboswitch-like elements (RLEs) among the long 5′ -UTRs. Riboswitches are complex RNA structures that undergo structural changes in response to binding of small molecules, typically metabolites or ions, altering the expression of their downstream gene on the transcriptional or post-transcriptional level (Breaker, 2011; Serganov and Nudler, 2013). These features make them ideal to sense small molecule concentrations, e.g., nutrient and ion availability. Five of these RLEs showed features homologous to the threonine operon leader, the Moco (Molybdenum cofactor biosynthesis), FMN and the cobalamin riboswitch of E. coli and a cation-responsive riboswitch upstream of an Mg2<sup>+</sup> transporter gene (corA) similar to mgtA of Salmonella (Nuss et al., 2015). However, the function of the majority of the riboswitch-like elements remains uncovered.

Structural alteration of a sensory RNA segment can also be triggered by temperature shifts. For instance, a thermocontrolled melting process of a double-stranded region, like a molecular zipper, is used to modulate gene expression of several bacterial heat shock and virulence-relevant genes (Kortmann and Narberhaus, 2012). One of these thermosensitive RNA elements (RNA thermometer) was identified upstream of lcrF (**Table 1**), the virulence plasmid-encoding gene for the master regulator of the T3SS/Yop machinery. The lcrF RNA thermometer is highly conserved between the human pathogenic Yersinia species and consists of a two stem-loop structure of which the second hairpin sequesters and hides the ribosome binding site (RBS) within its stem region at moderate temperatures (20–30◦C). Elevated temperatures (37◦C) lead to partial unfolding of the stem-loop, which renders the RBS accessible for the small subunit of the ribosome, leading to translation initiation (Hoe and Goguen, 1993; Böhme et al., 2012).

To investigate whether additional RNA thermometers or thermo-sensitive RNA elements are encoded on the Y. pseudotuberculosis genome, a structure RNA profile was recently generated of an entire transcriptome (**Figure 2**). For this purpose total RNA of the bacteria were isolated, incubated at 25, 37, and 42◦C, and subsequently treated with singlestrand and double-strand specific RNases. RNA-seq analysis of the resulting RNA fragments (PARS: parallel analysis of RNA structures) from 1,750 transcripts identified numerous temperature-sensitive RNAs (Rhigetti et al., 2016). This revealed that the RNA structurome of this pathogen is highly dynamic and encodes many more RNA thermometers. In total, 16 new RNA thermometers were discovered in 5′ -UTRs of metabolic, heat stress, and virulence-relevant genes. Among them are many oxidative stress genes (sodA, sodB, sodC, and katA), and two very important virulence genes, encoding the cell adhesion and invasion factor Ail, contributing to host cell colonization and host resistance, and the CNF<sup>Y</sup> toxin, triggering host tissue inflammation and damage (Rhigetti et al., 2016; **Table 1**). Besides the putative RNA thermometer, several small regulatory RNAs and tRNAs were also shown to be thermo-sensitive in this study. It will be interesting to investigate whether they promote and/or participate in the thermo-regulation of their target mRNA(s), and whether they are exploited by the pathogen to modify bacterial metabolism, physiology or pathogenicity upon entering a warm-blooded host, similar to the RNA thermometers.

### CONTROL BY REGULATORY RNAs

Small regulatory RNAs (sRNAs) are central regulators that play an essential role in various physiological and virulence associated processes in bacteria (Papenfort and Vogel, 2010, 2014; Heroven et al., 2017; Quereda and Cossart, 2017; Westermann, 2018). sRNAs are involved in the fine-tuning of these processes, allowing a rapid adaptation of pathogens to new environments. They have unique regulatory characteristics and control target gene expression post-transcriptionally through base-pairing with their transcripts. Such interaction can alter translation initiation or mRNA stability (Fröhlich and Vogel, 2009; Storz et al., 2011; Gorski et al., 2017). Functional analysis of sRNAs in different microorganisms revealed that often several sRNAs are part of the same regulatory cascade and sRNA-target network. This network typically includes multiple feedback loops and mixed regulatory circuits, illustrating the true complexity of RNA-based control system (Nitzan et al., 2017; Hör et al., 2018).

A myriad of sRNAs have been identified in the human pathogenic Yersinia species in independent high-throughput RNA-seq studies and are mainly referred to as Ysr (Yersinia small RNAs) (Koo et al., 2011; Koo and Lathem, 2012; Qu et al., 2012; Beauregard et al., 2013; Yan et al., 2013; Nuss et al., 2015, 2017a). The Ysrs are presumed Yersinia-specific sRNAs that were identified in an sRNA search in Y. pseudotuberculosis (Koo et al., 2011). A detailed list and the characteristics of these sRNAs are summarized in two recent reviews (Martínez-Chavarría and Vadyvaloo, 2015; Nuss et al., 2017b) (**Figure 2**, **Table 1**). Two aspects that are particularly striking are that (i) many sRNAs are directly regulated by the cAMP receptor protein (Crp) in response to growth phase and temperature, and that (ii) the majority of sRNAs are specific for Yersinia. This indicates that they are important players for the biological fitness and adaptation in their individual environmental and host niches, and reservoirs. A drawback of the latter feature is that most bioinformatic algorithms to identify potential targets of sRNAs rely on comparative genomics, i.e., functional and network analysis depending on sequence conservation between species (Wright et al., 2013), and are thus not suitable for the analysis of most Ysr sRNAs. Consequently, very little information is available about their molecular function and role for the pathogen. Nonetheless, the RNA chaperone Hfq, which mediates annealing of many sRNAs to their target mRNAs and protects them from ribonuclease cleavage or facilitates sRNA turnover or activity, was found to be required for virulence (**Table 1**). An hfq mutant of Y. pestis is impaired in its ability to resist phagocytosis and survive within macrophages at the initial stage of infection, and is highly attenuated in mice after subcutaneous or intravenous injection (Geng et al., 2009). Moreover, Hfq was found to regulate biofilm gut blockage that facilitates flea-borne transmission of Y. pestis (Rempe et al., 2012). Hfq of Y. pseudotuberculosis and Y. enterocolitica modulates the expression of multiple virulence-relevant cell surface structures such as adhesins/invasins (e.g., Ail, OmpX, MyfA pili, YadA, InvA) and expression of the T3SS/yop genes (Schiano et al., 2010, 2014; Schiano and Lathem, 2012; Kakoschke et al., 2014, 2016). Moreover, several Hfq-dependent alterations in the lipid A structure, motility, and biofilm formation defects, and changes in outer membrane vesicle synthesis were identified, which also impact virulence (Eddy et al., 2014; Leskinen et al., 2017). This strongly indicates that Hfq acts by controlling the expression of virulence-associated genes, probably in conjunction with small non-coding RNAs. However, up to now only very little is known about virulence-associated sRNAs and whether they are Hfqdependent.

An initial analysis with a few Ysr sRNAs showed that Ysr29, Ysr35, and Ysr141 seem to play a role in pathogenesis (Koo et al., 2011; Schiano et al., 2014; **Table 1**). The specific target of Ysr35 remains unknown, but expression of multiple proteins was found to be affected in an ysr35 mutant. Among them are several general stress response factors such as the chaperones GroEL and DnaK, the peroxidase AhpC, the translational factors S1 (RpsA) and the ribosome recycling factor Rrf (Koo et al., 2011). The virulence plasmid-encoded sRNA Ysr141 was found to enhance the synthesis of LcrF and many effector proteins (YopJ, YopE, YopK, YpkA), whereby only a direct influence of Ysr141 on YopJ translation was documented (Schiano et al., 2014). Recent work has also identified multiple sRNAs of Y. pestis, which are upregulated within macrophages. Among them is Ysr170, which was shown to be important for intracellular replication (Li et al., 2016). Much more is known about conserved sRNAs and their impact on pathogenicity of Yersinia. The most prominent influence on pathogenesis has been shown for the global carbon storage regulator system (Csr) (Heroven et al., 2012; Romeo et al., 2013; Vakulskas et al., 2015; Kusmierek and Dersch, 2017; Nuss et al., 2017b). The Csr system of Yersinia consists of the small RNA-binding protein CsrA, which is highly conserved among many bacterial species, and the two noncoding sRNAs CsrB and CsrC. The CsrA protein typically affects the translation and/or stability of target mRNAs by binding to GGA motifs within the mRNA. Activity of CsrA is controlled and antagonized by the sRNAs CsrB and CsrC as these are able to sequester multiple CsrA molecules, which prevents binding of target transcripts (Romeo et al., 1993, 2013; Liu and Romeo, 1997; Liu et al., 1997; Romeo, 1998; Babitzke and Romeo,

#### TABLE 1 | RNA regulators of Yersinia important for virulence.


#### TABLE 1 | Continued


(Continued)


2007; Romeo and Babitzke, 2018). In Yersinia, similar to other Enterobacteriaceae, the Csr system constitutes an important link between metabolic, stress, and virulence gene regulation by sRNAs, and plays a crucial role in virulence (Heroven et al., 2012; Kusmierek and Dersch, 2017; Nuss et al., 2017b). It has been found to control expression of the flagella master regulator FlhDC, the colonization factors InvA and PsaA, and multiple fitness and virulence-relevant cellular processes. These are e.g., motility/chemotaxis, biofilm formation, resistance against various stresses, including antibiotics and oxygen radicals, and adaptation of the central carbon metabolism by regulating the pyruvate-acetyl-CoA-tricarboxylate acid cycle flux (Heroven et al., 2008; Bücker et al., 2014; LeGrand et al., 2015; Willias et al., 2015; Kusmierek and Dersch, 2017). A very recent study of our group further demonstrated that the abundance of both CsrB and CsrC sRNAs in Y. pseudotuberculosis is reduced during the colonization of Peyer's patches compared to growth under different in vitro conditions (Nuss et al., 2017a). This indicates that more "free" CsrA is required during the infection process. In fact, loss of CsrA was shown to reduce the synthesis and completely abolish the secretion of the Yop effector proteins in Y. pseudotuberculosis (Nuss et al., 2017a). Most interestingly, CsrA also influences Yop secretion in Y. enterocolitica (Ozturk et al., 2017), although the overall outcome is quite different. This might be explained by the different growth conditions used for the Yop secretion assay or differences between the species.

Furthermore, RNA-seq approaches used to profile gene expression of Y. pseudotuberculosis and Y. pestis during host colonization revealed several other conserved sRNAs that are induced during infection. Among these sRNAs are RyhB1 and RyhB2—two sRNAs that are induced during iron starvation experienced in host tissue, SgrS—an sRNA upregulated under glucose-phosphate stress, and GlmY involved in cell wall synthesis (**Table 1**) (Massé and Gottesman, 2002; Vanderpool and Gottesman, 2004; Reichenbach et al., 2008; Urban and Vogel, 2008; Wadler and Vanderpool, 2009; Deng et al., 2012; Papenfort et al., 2012, 2013; Yan et al., 2013; Bobrovskyy and Vanderpool, 2014, 2016; Nuss et al., 2017a). However, lack of these sRNAs had only a mild effect on host colonization by the pathogens, indicating that they are implicated in the fine-tuning of virulencerelevant fitness processes.

Of the 80 antisense RNAs (asRNAs) that have been discovered to be part of the Y. pseudotuberculosis transcriptome, multiple are encoded on the virulence plasmid. Several were complementary to transcripts of the T3SS/Yop genes, although their influence on the expression of these virulence genes, and the influence on Yersinia virulence has not yet been investigated. However, one asRNA, CopA/incRNA, has been shown to be extremely important for the virulence process. CopA is implicated in the RNA-based control of the copy number of the virulence plasmid. The copy number of the Yersinia virulence plasmid, an IncFII plasmid, is controlled by the replicase RepA, which is tightly controlled and negatively regulated at the transcriptional and the post-transcriptional level. Under non-secretion conditions (e.g., in the absence of host cells), synthesis of RepA is repressed by (i) CopB, a transcriptional repressor that interacts with the repA promoter, and (ii) the asRNA CopA, that binds to the 5′ end of the longer of two repA transcripts preventing translation of a short leader peptide whose translation is coupled with and required for repA translation (Blomberg et al., 1990, 1992; Nordström, 2006; Wang et al., 2016; Pilla and Tang, 2018). Despite other IncFII plasmids, transcription ofcopA and copB of the Yersinia virulence plasmid are not constitutive. They are temperature-regulated and repressed during infection of the Peyer's patches, leading to a reduction of the ratio of the CopA asRNA and the repA transcript level, which results in an increase from 1–4 to up to 4–12 plasmid copies per bacterial cell during host tissue colonization (Wang et al., 2016). It could be demonstrated that up-regulation of the virulence plasmid copy number during the infection is crucial for virulence and that one of the secreted T3SS substrates (YopD) is involved. However, the molecular mechanism coupling copy number control to host cell binding still needs to be elucidated.

In summary, the identification of regulatory sRNAs and asRNAs using RNA-seq in the context of an infection process highlights that this approach is a powerful tool for the discovery of new infection-relevant regulatory processes. Application of newly developed RNA-seq technologies such as pulse-chaseexpression global target searches (Massé et al., 2005; Papenfort et al., 2006; Westermann et al., 2016), coupling of RNA-seq with ribosome profiling (Guo et al., 2014; Wang et al., 2015), M2-tagged sRNA affinity purification coupled with RNA-seq (Lalaouna et al., 2015; Tomasini et al., 2017), RNA interaction by ligation and sequencing—short RIL-Seq (Melamed et al., 2016), or global small non-coding RNA target identification by ligation and sequencing—short GRIL-seq (Han et al., 2016) could help to identify direct targets of the regulatory RNAs and their role in Yersinia biology.

### TRANSLATIONAL CONTROL

During translation mRNA is decoded for protein synthesis in a multi-step process consisting of initiation, elongation, termination, and recycling (Melnikov et al., 2012). This process is performed by the ribosome and several other translation associated components like tRNAs, initiation factors (IF), elongation factors (EF), and recycling factors (RF and RRF) (Melnikov et al., 2012). In addition to the complex transcriptional control strategies, gene expression of Yersinia is extensively regulated on the level of translation. Regulation of translation initiation occurs by changing the accessibility of the RBS for the small ribosomal subunit (30S, small subunit). As previously described, this can be achieved by masking the RBS with an RNA binding protein such as CsrA (Heroven et al., 2008, 2012) or an RNA thermo-loop (Baba et al., 1991; Böhme et al., 2012).

Other important riboregulators can modulate sRNA-mRNA interaction and influence the function of the translation machinery. For instance, the trans-translation control system or ribosome rescue system is composed of a small RNA-binding protein SsrB which transfers the polypeptide chain of ribosomes that are stalled on damaged or incomplete transcripts to the small, stable sRNA SsrA/tmRNA (Okan et al., 2006). The fact that SsrA/tmRNA is strongly upregulated in Y. pestis during the colonization of the lung and spleen and an ssrA mutant strain is strongly attenuated for virulence in a mouse infection model, strongly indicated that ribosome rescue is crucial for virulence (Okan et al., 2010; Yan et al., 2013). Reduced stress resistance and influence of the expression of the major virulence regulator LcrF in an ssrA/ssrB mutant strain could be the reason for this drastic phenotype (Okan et al., 2006, 2010).

One additional and very special translational control pathway of Yersinia virulence is mediated by the T3SS protein YopD, which, together with YopB, builds up the translocon pore in the eukaryotic membrane. Under non-secretion conditions, YopD is present in the bacterial cell and interacts with its chaperone LcrH, which prevents its degradation (Francis and Wolf-Watz, 1998; Edgren et al., 2012; Dewoody et al., 2013). Alone or bound to its chaperone LcrH, YopD seems to use sequences within the AUrich regions in the proximity of the RBS of multiple yop-encoding mRNAs. This hinders translation initiation and reduces Yop protein expression (Williams and Straley, 1998; Anderson et al., 2002; Chen and Anderson, 2011). However, other control factors seem to contribute to YopD-mediated translational repression, as the identified 5′ -UTR regions are important, but do not seem to be sufficient to promote this process. In fact, the secreted factor LcrQ (LcrM1 and LcrM2 in Y. enterocolitca) was found to contribute to post-transcriptional repression by YopD (Cambronne and Schneewind, 2002). Moreover, YopD wasfound to interact with the small 30S subunits of the ribosome. This indicates the presence of a different population of ribosomes (ribosome heterogeneity, Byrgazov et al., 2013), which negatively influences translation initiation complex formation (Kopaskie et al., 2013). How and to what extent these interactions affect translation of virulence-encoded genes is still unknown. YopDbinding could block the anti-RBS in the 16S rRNA from binding with the RBS. However, this alone would not explain why in particular translation of yop mRNAs is abrogated.

## RNA PROCESSING AND DECAY

Fast adaptation to changing surroundings is crucial for bacterial pathogens, such as Yersinia cycling between various environmental reservoirs, mammals, and humans in order to efficiently colonize their respective host and thrive in an occupied niche. Essential for this process is a controlled degradation of transcripts, which is mediated by a complex set of ribonucleases (RNases). The amount and type of RNases, their essentiality and collaboration in RNA metabolism is strongly depending on the investigated pathogen (e.g., Yersinia pseudotuberculosis > 15 RNases) (Arraiano et al., 2010; Lawal et al., 2011; Hui et al., 2014). The different RNases vary to a great extend in their mode of action, specificity for certain RNA features, and the nature of their targets in general (Arraiano et al., 2010; Lawal et al., 2011; Hui et al., 2014). Most bacteria, including all human pathogenic Yersinia species, contain low-specificity single-stranded endonucleases (e.g., RNase E), double-stranded endonucleases (e.g., RNAse III) and 3′ exonucleases (e.g., PNPase) (Mohanty and Kushner, 2016). RNA degradation occurs typically by consecutive endonucleolytic cleavage followed by exonucleolytic degradation from the 3′ end. The 5′ -end is generally more stabilized due to the triphosphate end, but also the 3′ -end can be rendered more stable toward RNases, e.g., by the presence of hairpin structures, RNA modification/polyadenylation or coverage by translating ribosomes. In addition to maturation and activation of certain transcripts (e.g., rRNA, tRNA), several conserved RNases were found to play a crucial role in global and/or targeted mRNA and sRNA turnover and also virulence in many pathogens (Clements et al., 2002; Ygberg et al., 2006; Lawal et al., 2011; Rosenzweig and Chopra, 2013).

In order to determine global mRNA decay rates, the bacterial culture is generally treated with the RNA polymerase inhibitor rifampicin prior to isolation of total RNA. To identify RNase targets and decay pathways, RNA-seq approaches with a transiently inactivated endoribonuclease followed by RNA-seq short TIER-seq (Chao et al., 2017, **Figure 1**) or with RNasedeficient mutants were recently used to quantify transcript steady-state levels in Leskinen et al. (2015) and Chen et al. (2016).

In Yersinia, especially the RNases PNPase and RNase E were found to occupy well-established roles in the regulation of virulence factors and growth under several infection-relevant conditions (oxidative and cold stress) (Rosenzweig et al., 2007; Yang et al., 2008; Rosenzweig and Chopra, 2013). PNPase forms together with the RNase E, the helicase RhlB and the glycolytic enzyme enolase a multi protein complex, called the degradosome (Rice and Vanderpool, 2011). RNase E represents the scaffolding enzyme for this complex hyperstructure and while some of the affected virulence-relevant properties were associated with an intact buildup of the degradosome, others only require RNase E or PNPase activity and RNA binding ability (Henry et al., 2012; Rosenzweig and Chopra, 2013). A Yersinia pnp mutant is significantly less virulent in the mouse compared to the isogenic wild-type strain. It was shown that PNPase is required for optimal functioning of the T3SS. Yet, this did not seem to depend on its ribonuclease activity, but requires its S1 domain. As the pnp mutant strain also expressed similar or even enhanced levels of T3SS-encoded transcripts, it is most likely that its virulence attenuation is due to a limiting T3SS activity (Rosenzweig et al., 2005, 2007). Besides the RNases of the degradosome, there is an increasing amount of information that additional RNases contribute to virulence. For instance, the single-strand specific RNase YbeY, which is important for the processing of the 3′ end of the 16S rRNAs, was recently shown to influence the expression of many virulence genes (colonization factors, T3SS, and the amount of both Csr regulatory RNAs) (Leskinen et al., 2015). How precisely YbeY influences all these processes is still unclear, but the large number of differentially regulated fitness and virulence-relevant genes in the YbeY-deficient mutant correlates with the severe phenotype and illustrates a crucial role in cell homeostasis and virulence. Besides these RNA processing and decay pathways, other RNA degrading enzymes were found to be differentially regulated in response to various virulencerelated growth conditions in vitro and during infection in vivo, indicating that more RNases may play a role in mRNA stability regulation. The exploitation of advanced RNA-seq combined with new target search approaches by coupling a UV crosslinking reaction to covalently link the RNase of interest to its target RNA will elucidate the role of RNases in various cellular processes (Waters et al., 2017). Determination of the cleavage sites and profiling of cleavage products of RNases, as established for other pathogens (Linder et al., 2014; Pobre and Arraiano, 2015; Redko et al., 2016; Chao et al., 2017; Le Rhun et al., 2017), will enable us to obtain genome-wide cleavage maps of different RNases of yersiniae under virulence-relevant conditions and will facilitate unraveling their complex RNA degradation network.

#### OUTLOOK: RNA INTERACTOME AND REGULATORY NETWORKS IN YERSINIA

The discovery of novel RNA-based regulatory strategies anticipates many exciting new insights into the complex posttranscriptional control network of virulence gene expression. Yet, determining the individual targets and interaction partners (protein and RNA) of the multitude of mRNAs, asRNAs, and sRNAs identified in Y. pseudotuberculosis during recent RNA-seq approaches is a time-consuming and very laborintensive task. However, new approaches have been successfully established to generate a snapshot of the interactome of RNAs that bind to a particular RNA-binding protein, e.g., CsrA, Hfq, or YopD. Such RNA interactomes can be obtained by RNA immunoprecipitation followed by RNA-seq (RIP-seq) (**Figure 2**), by covalently linked protein-RNA interactions followed by RNA-seq (CLIP-seq) (**Figure 2**), RNA interaction by ligation and sequencing (RIL-seq), or cross-linking, ligation and sequencing of hybrids (CLASH) (Bilusic et al., 2014; Holmqvist et al., 2016; Melamed et al., 2016; Waters et al., 2017). These techniques cannot only be used to profile the individual targets and entire binding sites (positions and sequences) of an RNAbinding protein, they will also allow us to improve sRNA target prediction, and facilitate the discovery of new sRNA control circuits and regulatory pathways (Dugar et al., 2016; Heidrich et al., 2017; Michaux et al., 2017). Similarly, RNA-RNA hybrids can be enriched or cross-linked and then sequenced to obtain a genome-wide Yersinia RNA-RNA network as done for E. coli (Liu et al., 2017).

Other open questions concerning the RNA-based control network of Yersinia virulence factors are: How are they conserved or remodeled during evolution between different strains and species, and how do acquired variations change virulence gene expression and pathogenicity? Numerous species- and strainspecific sensory and regulatory RNAs and many intra-species variations of certain sRNAs have been discovered in human pathogenic Yersiniae over the past years. For instance the PhoP regulator, controlling expression of the CsrC sRNA is not expressed in Y. pseudotuberculosis strain YPIII, but in IP32953 (Pisano et al., 2014). Moreover, the stability of the CsrC RNAs of strain YPIII and IP32953 differs significantly due to a 20 nt insertion in CsrC of IP32953, which renders the transcript more susceptible to degradation (Nuss et al., 2014). Based on the fact that the Csr system is crucial for T3SS/Yop expression and virulence (Bücker et al., 2014; Nuss et al., 2017a), it seems likely that even small variations (patho-specific alterations) between closely related strains could have a significant influence on the pathogen's potential to readjust and adapt to different hosts, host niches and reservoirs.

Another future task will be to elucidate the precise molecular mechanism how the RNA-binding translocator protein YopD controls translation of yop mRNAs. One technique to investigate ribosome-associated regulation, is ribosome profiling (Ribo-seq, **Figure 1**). With this method, all in vivo positions of extracted ribosomes on mRNAs (polysomes) can be identified enabling a dynamic view of ribosome movement along all mRNAs of a bacterial cell at a certain growth condition (Li et al., 2012; Ingolia, 2014). The function of ribosome subpopulations, such as those bound by YopD could be investigated by crosslinking of the ribosomes to the mRNA followed by affinity purification of the resulting YopD-ribosomal complex.

Ongoing characterization of the identified RNA-based control mechanisms of Yersinia will also have to address how the different riboregulators and RNAs contribute to the colonization of different host niches and whether and how they are implicated in the reprogramming of the pathogen, e.g., Y. pseudotuberculosis during the switch from the acute to the persistent infection mode. As regulatory RNAs, such as CsrB and CsrC, are implicated in the control of the regulator RovA, which forms a bistable switch, leading to high- and low-invasive subpopulations (Nuss et al.,

#### REFERENCES


2016), it is important to investigate the impact of riboregulators and non-coding RNAs on the formation of heterogeneous subpopulation with different virulence-relevant features. A major advance for this endeavor would be the development of singlecell RNA-seq approaches, in which the transcriptome of multiple individual bacterial cells could be profiled in different host niches and hosts during the entire course of an infection. So far single-cell RNA-seq has mainly been used for eukaryotic cells, however when applied to single bacteria this approach still suffers from technical limitations, which will have to be overcome in the future. This includes the establishment of more robust analyses and a more effective single-cell lysis and cDNA synthesis (Zhang et al., 2018). The next challenge will then be to analyze and compare the huge amount of generated data in an optimized and standardized integrative manner. This demands the establishment of (i) high quality, and free accessible standardized data-bases and (ii) the development of innovative bioinformatics protocols for data analysis and interpretation that allow integration of different technical and experimental approaches (e.g., different RNA-seq techniques, different OMICs, strains, tissues, and time-points of infection). Once achieved, this will give a comprehensive view on the complex RNA biology of Yersinia and will open up a new level of our understanding of virulence control of bacterial pathogens.

#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

#### FUNDING

Work on this topic was supported by Deutsche Forschungsgemeinschaft (DE616/6, DE616/7), and the Helmholtz Society. PD is member and supported by the Germany Center for Infection Research (DZIF) under grant number DZIF-TTU 06.801.

Opin. Microbiol. 10, 156–163. doi: 10.1016/j.mib.2007. 03.007


a single G-U wobble pair. Proc. Natl. Acad. Sci. U.S.A. 109, E757–E764. doi: 10.1073/pnas.1119414109


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Knittel, Vollmer, Volk and Dersch. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# All *Yersinia* Are Not Created Equal: Phenotypic Adaptation to Distinct Niches Within Mammalian Tissues

#### Kimberly M. Davis\*

W. Harry Feinstone Department of Molecular Microbiology and Immunology, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, United States

Yersinia pseudotuberculosis replicates within mammalian tissues to form clustered bacterial replication centers, called microcolonies. A subset of bacterial cells within microcolonies interact directly with host immune cells, and other subsets of bacteria only interact with other bacteria. This establishes a system where subsets of Yersinia have distinct gene expression profiles, which are driven by their unique microenvironments and cellular interactions. When this leads to alterations in virulence gene expression, small subsets of bacteria can play a critical role in supporting the replication of the bacterial population, and can drive the overall disease outcome. Based on the pathology of infections with each of the three Yersinia species that are pathogenic to humans, it is likely that this specialization of bacterial subsets occurs during all Yersiniae infections. This review will describe the pathology that occurs during infection with each of the three human pathogenic Yersinia, in terms of the structure of bacterial replication centers and the specific immune cell subsets that bacteria interact with, and will also describe the outcome these interactions have or may have on bacterial gene expression.

#### *Edited by:*

Matthew S. Francis, Umeå University, Sweden

#### *Reviewed by:*

Igor Brodsky, University of Pennsylvania, United States Petra Dersch, Helmholtz Center for Infection Research, Germany Matthew B. Lawrenz, University of Louisville, United States

> *\*Correspondence:* Kimberly M. Davis kdavi140@jhu.edu

*Received:* 07 May 2018 *Accepted:* 13 July 2018 *Published:* 03 August 2018

#### *Citation:*

Davis KM (2018) All Yersinia Are Not Created Equal: Phenotypic Adaptation to Distinct Niches Within Mammalian Tissues. Front. Cell. Infect. Microbiol. 8:261. doi: 10.3389/fcimb.2018.00261 Keywords: *Yersinia* infections, inflammation, phagocytes, gene expression, heterogeneity

## INTRODUCTION

The genus Yersinia contains three species that are pathogenic to humans: Yersinia pestis, Yersinia pseudotuberculosis, and Yersinia enterocolitica. Y. pseudotuberculosis and Y. enterocolitica are intestinal pathogens that typically cause self-limiting gastroenteritis and mesenteric lymphadenitis (Hubbert et al., 1971; Paff et al., 1976). Y. pestis causes bubonic, septicemic, and pneumonic plague, and has acquired a distinct inoculation route over the course of its divergence from an ancestral Y. pseudotuberculosis, despite relatively few genetic changes (Achtman et al., 1999). Y. pestis is inoculated intradermally into a mammalian host through injection during a flea bite, and Y. pestis can also be directly inhaled to cause primary pneumonic plague. Despite distinctions in inoculation route, all three Yersinia species have a lymphotropism, meaning they quickly traffic to lymph nodes (LNs) or lymphoid tissues and preferentially colonize these tissues. All Yersinia can also spread systemically by accessing the bloodstream and colonizing deep tissue sites, such as the spleen and liver.

Several well-characterized virulence factors play an important role in the disease progression of Yersinia and the ability of the bacteria to survive and proliferate within multiple host tissues. The human pathogenic Yersinia all contain a virulence plasmid termed either pCD1 or pYV, which contains genes encoding a central virulence factor for Yersinia: the type-III secretion system (T3SS) and its associated effector proteins, Yops (Gemski et al., 1980; Portnoy et al., 1981). The Yersinia T3SS is a needle-like structure that injects Yops directly into the host cytosol. Yops collectively function to modulate host signaling pathways to either inhibit or promote inflammation, to inhibit phagocytosis and motility of host cells, and can also inhibit the production and release of antimicrobial substances, such as reactive oxygen species (ROS) (Schesser et al., 1998; Viboud and Bliska, 2005; Songsungthong et al., 2010, Zhang and Bliska, 2010.)

Despite many genetic similarities, there are several welldefined genetic differences between Yersinia species. Y. pestis has acquired two additional virulence plasmids and 32 additional chromosomal genes since its divergence from Y. pseudotuberculosis, but also lost the functionality of as many as 13% of Y. pseudotuberculosis gene products (Achtman et al., 1999; Parkhill et al., 2001; Deng et al., 2002; Chain et al., 2004). Y. pestis pseudogenes include adhesins, and the lack of these genes may impact the ability of the bacterium to adhere to host cells and inject the T3SS (Simonet et al., 1996; Chain et al., 2004). Multiple adhesins: invasin, Ail, YadA, plasminogen activator protease (Pla, expressed from one of the Y. pestis-specific virulence plasmids), and pH 6 antigen, have been shown to play a critical role in adherence to host cells and T3SS injection (Yang and Isberg, 1993; Marra and Isberg, 1997; Felek and Krukonis, 2009; Durand et al., 2010; Felek et al., 2010; Maldonado-Arocho et al., 2013). Y. enterocolitica and Y. pseudotuberculosis utilize invasin, Ail and YadA (Yang and Isberg, 1993; Marra and Isberg, 1997; Hudson and Bouton, 2006; Durand et al., 2010; Maldonado-Arocho et al., 2013; Paczosa et al., 2014; Mühlenkamp et al., 2015). Y. pestis has acquired inactivating mutations in both yadA and invasin, and instead uses Ail, Pla, and pH 6 antigen to promote host cell interactions (Bartra et al., 2008; Felek and Krukonis, 2009; Felek et al., 2010). Ail also plays a major role in promoting serum resistance and preventing complement deposition in all three Yersinia, which allows bacteria to survive within the bloodstream and resist neutrophil-mediated killing (Miller and Falkow, 1988; Bliska and Falkow, 1992; Pierson and Falkow, 1993; Kolodziejek et al., 2007; Bartra et al., 2008).

Yersinia replicate extracellularly within host tissue sites to form clonal bacterial clusters (Simonet et al., 1990; Logsdon and Mecsas, 2006; Oellerich et al., 2007; Crimmins et al., 2012; Davis et al., 2015). The sites of Y. pseudotuberculosis replication have been termed lesions, microcolonies, and recently, pyogranulomas, but each term refers to the same structure, described in **Figure 1** (Davis et al., 2015; Peterson et al., 2017; Zhang et al., 2018). This review will first describe disease progression in mammalian hosts, to introduce the different tissues Yersinia replicates within, and then discuss similarities and differences in the inflammatory lesions that form during Yersinia infection, with a focus on how interactions with distinct immune cell subsets may alter the gene expression profile of individual Yersinia cells within a replicating population.

FIGURE 1 | Yersinia replicate to form inflammatory lesions that contain several different types of immune cells. Left: Adapted from Davis et al. (2015). Neutrophils (Ly6G, red) and iNOS<sup>+</sup> cells (iNOS, green) are recruited to replicating centers of Y. pseudotuberculosis (Yptb, blue). Right: Diagram of inflammatory lesions. (1) Neutrophils (red) form an inner layer surrounded by (2) Apoptotic cells (yellow; monocytes, some neutrophils). (3) Viable, NO producing cells (green; macrophages, dendritic cells, some neutrophils) form an outer layer of phagocytes. Peripheral bacterial cells (white rods) respond to host cell contact and diffusible antimicrobials, and allow interior bacteria (blue rods) to replicate. White numbers in the micrograph correspond with the numbered regions in the diagram.

### *YERSINIA* HUMAN DISEASE PROGRESSION

### Enteropathogenic *Yersinia* Disease Progression

Y. pseudotuberculosis in an enteric pathogen that infects the terminal ileum of the small intestine and also colonizes associated lymphoid tissues, such as Peyer's patches and mesenteric lymph nodes (LNs) (Hubbert et al., 1971; Paff et al., 1976). In healthy individuals, robust neutrophil recruitment is typically sufficient to limit bacterial growth, and can result in clearance. However, in immunocompromised individuals, individuals with hemochromatosis (iron-overload disorder), or in a mouse model of infection, bacteria access the bloodstream and spread to deep tissue sites, such as the spleen and liver (Capron et al., 1981; Abbott et al., 1986; Barnes et al., 2006; Miller et al., 2016). Bacteria replicate within deep tissues to high numbers, which can cause a lethal infection in the mouse model. Replication within intestinal tissues also leads to spread to the cecum, where bacteria associate with cecal lymphoid follicles and can establish a long term, persistent infection in the mouse (Fahlgren et al., 2014; Avican et al., 2015).

Like Y. pseudotuberculosis, Y. enterocolitica is an intestinal pathogen that initially colonizes the intestinal lumen and associated lymphoid tissues, such as Peyer's patches and mesenteric LNs. Y. enterocolitica can also spread to deep tissues following replication at intestinal sites (Heesemann et al., 1993; Viboud and Bliska, 2005; Oellerich et al., 2007). The disease progression for enteric Yersinia appears to be very similar, however, Y. enterocolitica is a relatively common cause of gastroenteritis, whereas patients rarely present with Y. pseudotuberculosis infection. This may be because Y. enterocolitica commonly colonizes many different species of livestock and wild game animals, which could lead to increased exposure rates within human populations, or because Y. enterocolitica may elicit a heightened inflammatory response resulting in more severe disease (Dube et al., 2001; Chlebicz and Slizewska, 2018; Syczylo et al., 2018).

#### *Y. pestis* Disease Progression

Y. pestis is inoculated into the dermis of mammalian hosts after replication within the midgut of the flea. Within the flea, bacteria replicate within dense aggregates that partially block the proventricular valve of the flea, thus promoting dissemination into the mammalian host during a flea bloodmeal (Hinnebusch et al., 2017). Y. pestis travels from the injection site rapidly into local draining LNs (dLNs), where colonization is established within the mammalian host. In mouse models of infection, bacteria have been shown to traffic to dLNs intracellularly within dendritic cells and monocytes, and also extracellularly within the lymphatic system (Shannon et al., 2013, 2015; St. John et al., 2014; Gonzalez et al., 2015). The exact route depends on the inoculation (intradermal or subcutaneous) and also the virulence of the bacterial strain. Following initial trafficking, bacteria appear to occupy an extracellular niche. Bacteria initially replicate within dLNs with limited inflammation, however as bacteria continue to replicate, high levels of inflammatory immune cell recruitment and tissue damage occurs. This forms the characteristic inflamed LNs, or buboes, that define bubonic plague (Butler, 1994). Tissue damage within LNs promotes bacterial bloodstream access, which initiates septicemic plague, and leads to colonization of deep tissues such as the spleen and liver (Perry and Fetherston, 1997). Septicemic plague can also occur immediately following flea-borne transmission, if the flea bite occurs deep within dermal tissues (Sebbane et al., 2006a). High levels of bacteria within the bloodstream can promote the development of a secondary pneumonic plague. Following bacterial replication in the lung, infected individuals can expel aerosols containing Y. pestis, which disseminate and cause primary pneumonic plague in new hosts. Septicemic plague also contributes to dissemination when a flea takes a bloodmeal from an infected host, which then restarts the Y. pestis life cycle in the flea vector (Perry and Fetherston, 1997; Hinnebusch et al., 2017).

#### Host Response to *Yersinia*, and *Yersinia* Responses to the Host

Yersinia encounter many different tissue sites during infection of mammalian hosts, which forces bacteria to adapt to multiple different environments within a single host organism. Recruited inflammatory cells attempt to kill Yersinia through phagocytosis or by releasing antimicrobial substances to eliminate extracellular bacteria. Yersinia, in turn, can inhibit phagocytosis and detoxify many of these antimicrobials. The genes required to battle the host response have been largely defined by transcriptional data and mutational analyses, that indicate a given gene or pathway contributes to virulence at the population level. Interestingly, distinct subsets of the bacterial population respond differently, depending on their interactions with, and proximity to, different host cell subsets (Marteyn et al., 2010; Burton et al., 2014; Davis et al., 2015). These studies have described host-driven heterogeneity during Salmonella enterica serovar Typhimurium and Shigella flexineri infections, in addition to Yersinia, which may indicate this is a widespread phenomenon within pathogenic bacterial populations. Although much is still unknown concerning Yersinia heterogeneity within different tissue sites, we will describe what has been shown thus far, and which pathways may also be utilized in other tissues based on similarities in host cell interactions.

#### Lymph Nodes

It has been hypothesized that the ability of Y. pestis to disseminate from dermal tissues and replicate within draining lymph nodes (dLNs) is due to the presence of Pla protease, which was acquired on an additional virulence plasmid after Y. pestis diverged from Y. pseudotuberculosis (Welkos et al., 1997; Lathem et al., 2007). However, Y. pseudotuberculosis can also effectively replicate within the dermis and traffic to dLNs following intradermal inoculation (Guinet et al., 2008, 2015). The major pathological difference during infection with the two species was the ability of the immune response to contain bacterial replication. Y. pseudotuberculosis elicits robust neutrophil recruitment that surrounds and contains bacteria, whereas Y. pestis exists as individual bacteria and reaches heightened bacterial loads (Guinet et al., 2008). Y. pestis has acquired the ability to produce LPS with poor stimulatory activity and also produces an anti-phagocytic capsule, both of which could limit neutrophil interactions (Montminy et al., 2006; Dudte et al., 2017). There were also many examples of Y. pestis clustered together within the same infected tissues, but it was unclear if clustering promotes bacterial growth or clearance by the host. In the absence of clustering, Y. pseudotuberculosis replication within mesenteric LNs causes a lethal infection in the mouse, suggesting clustering promotes host survival (Peterson et al., 2017). Y. pseudotuberculosis was contained by a layer of apoptotic monocytes (**Figure 1**), which may also contain Y. pseudotuberculosis within dLNs. These studies suggest that the formation of microcolony, or pyogranuloma, structure contains infection and protects the host, however this structure also allows bacteria to cooperate and promote their own growth (Davis et al., 2015).

Y. pestis replication within dLNs is characterized by high bacterial loads and high levels of tissue destruction. Replication of Y. pestis within dLNs initially induces very little host cytokine expression, which is actively inhibited by components of the virulence plasmid including the T3SS (Comer et al., 2010). At later timepoints, neutrophils and additional phagocytes are recruited to infected LNs, and IL-17 is produced (Comer et al., 2010). In response, Y. pestis increases expression of reactive nitrogen species (RNS) detoxifying genes and iron scavenging genes, indicating Y. pestis is being exposed to the antimicrobial diffusible gas, nitric oxide (NO), and has limited access to iron during replication within the dLN (Sebbane et al., 2006b). NO was produced by recruited neutrophils, and likely other phagocytes, within the rat bubo, and the detoxifying gene, hmp, played an important role in promoting Y. pestis virulence (Sebbane et al., 2006b). It is likely that bacteria in the closest proximity to NO-producing phagocytes may specifically respond to this stress, and detoxify NO to protect other members of the population (Davis et al., 2015), although this hasn't yet been investigated. It will also be interesting to determine if iron limitation is uniformly experienced across LNs, or if this is specific to a subset of cells. Interestingly there was little evidence for a bacterial response to reactive oxygen species (ROS) in dLNs, and only one superoxide dismutase increased in transcript levels. A mutant strain lacking this superoxide dismutase (sodA) was fully virulent, which indicates Y. pestis effectively inhibits this response within LNs (Sebbane et al., 2006b).

#### Deep Tissues

Multiple immune cell types are recruited to deep tissues during infection, and Yersinia effectively responds to create a hospitable environment for bacterial replication. Neutrophils directly interact with clustered centers of extracellular Y. pseudotuberculosis, and additional phagocytes are maintained at a distance relative to the bacterial centers (**Figure 1**). The layer of NO-producing cells likely contains a mixed population of monocytes, macrophages, and dendritic cells, and is adjacent to an apoptotic cell layer likely composed of monocytes (Davis et al., 2015; Peterson et al., 2017; Zhang et al., 2018). In the spleen, only the bacteria at the periphery of replicating centers respond to NO, and detoxify NO to prevent further diffusion into the microcolony, thus protecting interior bacteria (Davis et al., 2015). NO was detoxified by Hmp, which was expressed in response to the Nos2 gene product, inducible nitric oxide synthase, indicating hmp expression was a specific response to NO. This example of cooperative behavior likely also exists within mesenteric LNs, based on the recruitment of NO-producing cells, and the similarities in these microcolony or pyogranuloma structures (Zhang et al., 2018). Peripheral Y. pseudotuberculosis cells also respond to neutrophil contact in the spleen by expressing heightened levels of the T3SS (Pettersson et al., 1996; Davis et al., 2015). It remains unclear if heightened expression signifies a T3SS translocation event, or if this is due to prolonged host cell contact. Heightened T3SS expression could occur in multiple tissue sites and response to many host cell types, as it is clear that Yersinia can inject T3SS effector proteins into many different immune cell types during infection (Durand et al., 2010).

Y. pestis and Y. enterocolitica actively inhibit immune cell recruitment to the spleen by limiting CCR2 signaling and promoting Gal-1 expression, respectively (Kerschen et al., 2004; Ye et al., 2009, 2011; Davicino et al., 2017). In the liver, inflammatory monocytes and dendritic cells are recruited to sites of Y. pestis infection, and YopM promotes cell death of neutrophils and macrophages leading to the formation of an apoptotic layer of host cells around replicating bacterial centers (Ye et al., 2014). This mirrors the layer of apoptotic cells observed within Y. pseudotuberculosis-infected mesenteric LNs (**Figure 1**) (Peterson et al., 2017). iNOS<sup>+</sup> cells are also recruited to Y. pestis and Y. enterocolitica lesions, suggesting that subsets of peripheral cells may also express hmp to protect other individuals within the population. This also suggests that there may be quite a few similarities in the host response to Yersinia within deep tissues, unlike the pathological distinctions that are observed within infected LNs. This also indicates that all Yersinia may respond to similar stresses within deep tissues, and supports the idea that the cooperative behavior described for Y. pseudotuberculosis may also occur within other Yersinia infections, potentially even by expressing the same stress response pathways in distinct subsets of bacteria.

Neutrophils are a major producer of reactive oxygen species (ROS), but production can be inhibited by the T3SS, so it remains unclear if ROS restricts Yersinia growth in tissues. Mutant Y. enterocolitica lacking superoxide dismutase (sodA) were attenuated within the spleen and liver, but this could be due to sensitivity to endogeneous ROS generated during bacterial aerobic respiration, or host-derived ROS (Roggenkamp et al., 1997). Transcriptional responses to ROS have not been observed with Y. pestis or Y. pseudotuberculosis, and a Y. pestis sodA mutant was not attenuated for growth in buboes or systemic infection (Sebbane et al., 2006b; Davis et al., 2015). However, a ROSsensitive Y. pseudotuberculosis dusB-fis mutant was attenuated in immunocompetent mice, and rescued in the absence of ROS production, suggesting ROS was produced during infection (Green et al., 2016). Fis is a nucleoid-associated protein that likely has pleiotropic effects on the bacterial response to ROS, which may explain why the dusB-fis mutant uncovered an impact of ROS on Y. pseudotuberculosis growth. WT Y. pseudotuberculosis growth was not impacted by the presence or absence of hostderived ROS, indicating the WT strain can effectively defend itself from ROS without significant transcriptional changes (Davis et al., 2015; Green et al., 2016).

#### Intestinal Tissue and Peyer's Patches

Following intestinal colonization, enteropathogenic Yersinia traverse M cells to form micrcolonies within Peyer's patches, which are specialized lymphoid tissues associated with the intestinal epithelium. The immune response to Y. pseudotuberculosis and Y. enterocolitica in Peyer's patches is characterized by neutrophil influx and a mixed TH17/TH1 response, in contrast to the TH17-specific response to Y. pestis within dLNs (Comer et al., 2010; Davicino et al., 2017; Nuss et al., 2017). Y. pseudotuberculosis responds by expressing genes to prevent phagocytosis and counteract iron deprivation and RNS, similar to Y. pestis within dLNs (Sebbane et al., 2006b; Nuss et al., 2017). Y. enterocolitica also actively inhibits neutrophil recruitment by a YopH-dependent mechanism that reduces CXCR2 surface expression on neutrophils (Dave et al., 2016). Microcolonies within the lamina propria contain CD4<sup>+</sup> T cells and inflammatory macrophage and/or dendritic cell populations, in addition to CD8<sup>+</sup> T cells. Interaction with CD8<sup>+</sup> tissueresident memory T cells within the lamina propria can effectively contain Y. pseudotuberculosis replication within the terminal ileum (Bergsbaken and Bevan, 2015).

Intestinal infection with sublethal doses of Y. pseudotuberculosis can establish a persistent, asymptomatic infection with sustained fecal shedding of bacteria, which may promote dissemination to additional hosts (Fahlgren et al., 2014; Avican et al., 2015). Persistence led to a dramatic shift in bacterial gene expression, which resembled growth at room temperature in bacteriological media, and was characterized by heightened expression of flagella and invasin, and downregulation of the T3SS (Avican et al., 2015; Heine et al., 2018). RovA was involved in these changes, and is known to positively regulate invasin and pH6 antigen expression. RovA expression is bistable in bacteriological media between 30 and 34◦C, indicating bacteria utilize a bet-hedging approach to prepare a subset of the population to invade host cells, potentially as bacteria are moving through the host environment and the temperature is increasing (Quade et al., 2012; Nuss et al., 2016). RovA expression was also detected within a subset of the bacterial population replicating within the cecum, suggesting that RovA expression could be another example of cooperative behavior (Nuss et al., 2016).

#### Lungs

Very little inflammation is observed early during primary pneumonic Y. pestis infection, similar to early timepoints within LNs during bubonic plague (Lathem et al., 2005; Guinet et al., 2008; Comer et al., 2010). However, by 48 h postinfection, inflammatory cells are localized around centers of replicating bacteria and bacteria are responding to the host environment (Lathem et al., 2005). Within the lung, Y. pestis upregulates expression of the T3SS, iron acquisition genes, and RNS detoxifying genes, including hmp (Lathem et al., 2005). In contrast, expression of pH 6 antigen, Pla, and ROS detoxifying genes were downregulated, indicating bacteria were not responding to ROS, and these adhesins was not required to replicate within the lungs (Lathem et al., 2005). Similar results were seen in a Y. pseudotuberculosis lung infection model, where Ail and YadA contribute to replication and dissemination to deep tissues, but pH 6 antigen was dispensible (Paczosa et al., 2014). It will be interesting to determine if heightened gene expression occurs within all the individual Yersinia cells replicating within the lung, or if distinct subsets of cells respond in different ways. Based on experiments with Y. pseudotuberculosis, individual Y. pestis cells may be expressing heightened levels of the T3SS or heightened levels of hmp depending on their spatial location relative to immune cell subsets.

#### Concluding Remarks

As Yersinia replicate within tissues, they form microcolonies where some individual cells interact directly with host immune cells, and other cells only interact with other bacteria. Yersinia species elicit similar inflammatory responses, characterized by recruitment of neutrophils, monocytes, macrophages, and dendritic cells, and all Yersinia appear to react similarly by upregulating expression of anti-phagocytic factors, RNS detoxifying genes, and iron scavenging genes. Several key questions remain: Are microcolonies protective for the host, or do they promote bacterial replication? Can we alter the host response to better contain bacterial growth? Limiting the host response impacts the containment of Yersinia within microcolonies and increases virulence, suggesting microcolonies protect the host (Peterson et al., 2017). However, disruption of microcolonies by attenuating Yersinia can promote clearance of bacteria, suggesting there is also a protective effect for bacteria (Davis et al., 2015). More future studies will be needed to determine if the host immune response can be manipulated to better contain and clear infection.

We are only beginning to understand how host cell interactions are driving differences in bacterial gene expression, and how small subsets of the bacterial population may be producing virulence genes that are critical for promoting disease. Better understanding and characterization of the heterogeneity in gene expression across bacterial populations, specifically in virulence gene expression, will have important implications for identifying potential drug targets for novel therapeutics. We believe the heterogeneity within Yersinia populations is likely also present within populations of other bacterial pathogens, and may be a widespread phenomenon that needs to be taken into account when designing antimicrobial therapeutics.

### AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and approved it for publication.

### ACKNOWLEDGMENTS

The author of this manuscript declares no conflict of interest. This work was supported by the NIAID Career Transition award, 1K22AI123465-01.

#### REFERENCES


pathology, and bacterial transcriptional activity. Proc. Natl. Acad. Sci. U.S.A. 102, 17786–17791. doi: 10.1073/pnas.0506840102


pseudotuberculosis pyogranulomas and dictate adaptive responses at the expense of innate immunity during oral infection. Infect. Immun. 86, e00782– e00717. doi: 10.1128/IAI.00782-17

**Conflict of Interest Statement:** The author declares that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Davis. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

digital media

of impactful research

article's readership