**Microbial Virulence Factors**

Editor

**Jorge H. Leit˜ao**

MDPI • Basel • Beijing • Wuhan • Barcelona • Belgrade • Manchester • Tokyo • Cluj • Tianjin

*Editor* Jorge H. Leitao˜ Instituto de Biotecnologia e Bioengenharia Portugal

*Editorial Office* MDPI St. Alban-Anlage 66 4052 Basel, Switzerland

This is a reprint of articles from the Special Issue published online in the open access journal *International Journal of Molecular Sciences* (ISSN 1422-0067) (available at: https://www.mdpi.com/ journal/ijms/special issues/Virulence Factors).

For citation purposes, cite each article independently as indicated on the article page online and as indicated below:

LastName, A.A.; LastName, B.B.; LastName, C.C. Article Title. *Journal Name* **Year**, *Article Number*, Page Range.

**ISBN 978-3-03936-946-1 (Hbk) ISBN 978-3-03936-947-8 (PDF)**

c 2020 by the authors. Articles in this book are Open Access and distributed under the Creative Commons Attribution (CC BY) license, which allows users to download, copy and build upon published articles, as long as the author and publisher are properly credited, which ensures maximum dissemination and a wider impact of our publications.

The book as a whole is distributed by MDPI under the terms and conditions of the Creative Commons license CC BY-NC-ND.

## **Contents**



## **About the Editor**

**Jorge H. Leit ˜ao**, Ph.D., is associate professor at the scientific and pedagogical area of Biological Sciences of the Department of Bioengineering, Instituto Superior Tecnico, Universidade de Lisboa, ´ Portugal, and researcher at IBB- Institute for Bioengineering and Biosciences. After obtaining his Ph.D. in Biotechnology and Biosciences in 1996 by Instituto Superior Tecnico, his research work was focused on the biology and pathogeneis of bacteria of the Burkholderia cepacia complex (Bcc), in particular on the biosynthesis of exoplysaccharides. Bcc is a group of opportunistic pathogens causing life-threatening infections in patients suffering from Cystic Fibrosis, Chronic Granulomatous Disease, and immunocompromised patients. Current research interests are the post-transcription regulation of bacterial gene expression and the roles played by small non-coding regulatory RNAs and RNA chaperones on the biology and pathogenesis of bacteria of the Bcc. Other research interests include bacterial virulence, focusing on their exploitation as targets for the development of novel antimicrobial strategies; bacterial resistance to antimicrobials and the development of novel antimicrobials; and the molecular characterization of microbial populations of ecological, industrial or health interest.

## *Editorial* **Microbial Virulence Factors**

#### **Jorge H. Leitão**

IBB–Institute for Bioengineering and Biosciences, Instituto Superior Técnico, Department of Bioengineering, Universidade de Lisboa. Av. Rovisco Pais, 1049-001 Lisboa, Portugal; jorgeleitao@tecnico.ulisboa.pt; Tel.: +351-21-841-7688

Received: 21 July 2020; Accepted: 25 July 2020; Published: 27 July 2020

**Keywords:** microbial virulence factors; bacterial pathogens; fungal pathogens; pathogenicity

Microbial virulence factors encompass a wide range of molecules produced by pathogenic microorganisms, enhancing their ability to evade their host defenses and cause disease. This broad definition comprises secreted products such as toxins, enzymes, exopolysaccharides, as well as cell surface structures such as capsules, lipopolysaccharides, glyco- and lipoproteins. Intracellular changes in metabolic regulatory networks, governed by protein sensors/regulators and non-coding regulatory RNAs are also known to contribute to virulence. Furthermore, some secreted microbial products have the ability to enter the host cell and manipulate their machinery, contributing to the success of the infection. The knowledge, at the molecular level, of the biology of microbial pathogens and their virulence factors is central in the development of novel therapeutic molecules and strategies to combat microbial infections. This is of particular importance in the present days with the worldwide emergence of microbes resistant to available antimicrobials, as well as of novel pathogens such as the SARS-CoV-2 responsible for the present pandemics. Advances in recent years in molecular biology, genomics and post-genomics technologies, and bioinformatics contributed to the molecular identification and functional analyses of a wide range of microbial virulence factors. The Special Issue of IJMS focused on virulence factors and their regulatory networks from microbes such as bacteria, viruses, fungi, and parasites, as well as on the description of innovative experimental techniques to characterize microbial virulence factors. A total of 18 papers was published in this Special Issue. The collection comprises state of the art papers on virulence factors and mechanisms from a wide range of bacterial and fungal pathogens for humans, animals, and plants, thus reflecting the impact of microorganisms in health and economic human activities and the importance of the topic.

Due to their impact on human health, bacterial pathogens that cause infections in humans have received a higher attention, with *Escherichia coli* as one of the most studied bacteria. Pokharel et al. investigated the roles played by the recently described serine Protease Autotransporters (SPATE) TagB, TagC, and Sha of *E. coli* on urinary infections using a 5637 bladder epithelial cell line [1]. Members of the SPATE family owe their proteolytic activity to the serine protease catalytic triad composed of an aspartic acid, a serine, and a histidine residue. Evidence is presented showing that the three SPATE proteins are internalized by bladder epithelial cells, leading to alterations of actin cytoskeleton distribution. Results presented indicate that Sha and TagC degrade mucin and gelatin, respectively [1]. The mutation analysis of the serine catalytic site showed that secretion of the three proteins is not affected, but impaired their entry into epithelial cells, affecting their cytotoxicity and proteolytic activity [1].

The presence of genes related to virulence factors including adhesins, siderophores, protectines or invasins, and involved in allantoin metabolism were investigated among 32 non-*E. coli Enterobacterales* isolates obtained from the feces of 20 healthy adults [2]. Similar studies analyzed virulent NECE strains from patients with an ongoing infection, and not commensal NECE from healthy subjects as in the present study [2]. Isolates were taxonomically characterized by 16S RNA sequencing and MALDI-TOF MS analysis, and profiled by pulsed-field gel electrophoresis. The genus *Klebsiella* was found as the most represented, followed by *Enterobacter* and *Citrobacter* [2]. The isolates were further characterized concerning the presence in their genomes of genes encoding selected virulence factors, as well as their phenotypes related to biofilm formation and resistance to a selection of antibiotics. Results point out that the isolates do not encompass particularly virulent strains and in most of the cases were susceptible to antibiotics [2].

Yang et al. investigated the role of the *Salmonella enterica* serovar Typhimurium (ST) *pdxB*-*usgtruA*-*dedA* operon on intracellular survival using deletion mutants constructed with the λ-Red recombination technology [3]. The *Salmonella* genus comprises several facultative intracellular pathogens capable of infecting both human and animal hosts. The ST deletion mutants was investigated in J774A.1 macrophage cells. The deletion mutants Δ*pdxB*, Δ*usg*, and Δ*truA* exhibited reduced replication abilities compared to ST and the deletion mutant Δ*dedA*. The *pdxB-usg-truA-dedA* operon is shown to contribute to ST virulence in mice, and to resistance to oxidative stress [3].

*Aeromonas hydrophila* is an aquatic Gram-negative bacterium, capable of causing serious and lethal infections to a wide range of hosts, including fish, birds, amphibians, reptiles, and mammals [4]. Dong et al. described the identification and functional characterization of the LahS global regulator of *A. hydrophila* [4]. LahS was identified after the screening of a Tn5-derived library of 947 *A. hydrophila* mutants for reduced hemolytic activity. The LysR family transcriptional regulator family member LahS was found to play a role in biofilm formation, motility, antibacterial activity, resistance to oxidative stress, and proteolytic activity, as well as essential for *A. hydrophila* virulence to zebrafish [4]. The comparative proteomics analysis performed by the authors confirmed the role of the protein as a global regulator in *A. hydrophila* [4].

Bacteria of the *Dickeya* genus comprise plant pathogens that affect crops such as potatoes. In order to succeed when infecting their hosts, *Dickeya* secrete several proteins with plant cell wall degrading activities, including pectinases, cellulases, and proteases [5]. To investigate the role played by the protease Lon on *D. solani* pathogenicity towards potato, Figaj et al. used a λ-Red-derived protocol to construct a *lon* deletion mutant [5]. Results presented indicate that the Lon protein plays a role in protecting the bacterium to high ionic and temperature stresses, affecting the activity of pectate lyases, the organism motility, and delaying the onset of infection symptoms in the potato host [5].

The plant pathogen *Candidatus Phytoplasma mali* is the causal agent of apple proliferation disease, that affects apple production in Northern Italy [6]. *Phytoplasma* are biotrophic, obligate plant and insect bacterial symbionts, with a biphasic life cycle comprising reproduction in phloem-feeding insects and in plants [6]. The paper of Mittelberger et al. focused on the effector protein PME2 (Protein in Malus Expressed 2), expressed by *P. mali* when infecting apples [6]. The in silico analysis of the PME2 protein sequence performed revealed that the protein has features of effector proteins of Gram-positive bacteria, with a predicted final localization at the cytoplasm or nucleus of the host [6]. Two main protein variants, PME2ST PME2AT, were found associated in infected apple trees from Italy and Germany. Using protein variants tagged with GFP, both variants were found to translocate to the nucleus of *Nicotiana* spp. protoplasts. A better understanding of the molecular mechanisms used by *P. mali* to manipulate its host will rely on genomics analysis, since no genetic manipulation is presently available for these organisms [6].

The necrotrophic fungal pathogen *Sclerotinia sclerotiorum* (Lib.) de Bary infects a wide range of plants causing devastating agricultural losses. The organism forms a typical structure named sclerotia when vegetative hyphae gather to form a hardened multicellular structure important in its development and pathogenesis, and that under favorable conditions germinate leading to vegetative hyphae or apothecia that will initiate novel disease cycles by producing ascospores [7]. Li et al. used a proteomics approach based on 2D gels followed by spot isolation and protein identification by MALDI-TOF to identify proteins differentially expressed between a wild-type strain and a deletion mutant on the gene *SsNsd1* encoding a type IVb GATA zinc finger transcription factor [7]. Although the gene encoding SsNsd1 was found as expressed at low levels during the hyphae stage, the mutant is unable to form the compound appressoria. The authors were able to identify a total of 40 proteins as differentially expressed, 17 with predicted functions and 23 as hypothetical proteins [7]. The authors emphasize the utility of the approach used to identify important proteins involved in the SsNsd1-mediated formation of appressorium.

In addition to other factors, the success of pathogens rely on cell-cell communication. Bacterial outer membrane vesicles (OMV) are recognized as an efficient means of bacteria-bacteria and bacteria-host communication, not only intra-species, but also interspecies [8]. Despite the lack of data on a possible role played by OMVs in bacterial-yeast communication, Roszkowiak et al. investigated the role played by *Moraxella catarrhalis* OMVs on the susceptibility of selected bacterial and fungal pathogens to the cationic peptide polymyxin B, and to the serum complement [9]. Using OMVs from *M. catarrhalis* strain 6, the authors found that these OMVs conferred protection against the cationic peptide polymyxin B to the non-typeable *Haemophilus influenzae*, *Pseudomonas aeruginosa*, and *Acinetobacter baumannii*. Furthermore, OMVs also protected serum-sensitive non-typeable *H. influenza* and promoted the growth of the serum-resistant *P. aeruginosa* and *A. baumannii* against the complement [9]. In addition, the results presented also show that OMVs facilitate the formation of hyphae by the pathogenic yeast *Candida albicans*, promoting its virulence [9]. As stated by the authors, this work might pave the way to uncover additional roles played by OMVs-dependent interactions in multispecies communities [9].

The RNA chaperone Hfq is a master regulator of gene expression in bacteria, mediating the interaction of small noncoding RNAs with their mRNA targets, including those related to virulence in Gram-negative bacteria [10]. Dienstbier et al. performed an integrated Omics comparative analysis of the Hfq regulon in the *Bordetella pertussis* human pathogen, responsible for respiratory tract infections, in particular of a whooping cough [11]. Based on the use of RNAseq, and gene ontology analysis, genes significantly upregulated in the *hfq* mutant fall into categories including "Translation", "Regulation of transcription", and "Transmembrane transport", while genes downregulated fall in the categories "Transmembrane transport", "Iron–sulfur cluster assembly", "Oxido-reduction process", "Pathogenesis", and "Protein secretion by the type III secretion system" [11]. Correlations of transcriptome, proteome, and secretome datasets are also presented [11]. Results presented corroborate the central role played by Hfq on the physiology and pathogenicity of *B. pertussis* [11].

In their brief report, Maisetta et al. performed the ex vivo evaluation of the bactericidal activity of combinations of the semi synthetic antimicrobial peptide lin-SB056-1 in combination with EDTA (Ethylenediaminetetraacetic acid) against endogenous *P. aeruginosa* present in the sputum from patients suffering from primary ciliary dyskinesia (PCD) [12]. The authors observed that the peptide and EDTA were almost inactive against PCD sputum endogenous *P. aeruginosa* when used alone, but exhibited a significant synergistic killing effect with a sputum sample-dependent efficacy [12]. EDTA, but not lin-SB056-1, was found to inhibit biofilm formation and the production of virulence factors including alginate, pyocyanin, and the metalloprotease LasA [12].

Various bacterial species have evolved various strategies to invade, survive, and multiply intracellularly in host cells. The paper of Denzer et al. presents an updated review of the mechanisms used by bacteria to invade the host cell, to manipulate their biochemical and gene expression machinery, and to multiply and escape from the host cell [13]. The authors present a thorough review of mechanisms used by intracellular pathogens, including the highjack of host immune defenses to enter into the host cell. Central attention is given to the various mechanism used to manipulate gene expression, including histone modification, control of host DNA methylation patterns, sabotage of host long non-coding RNAs, interfering with the host RNA transcription and translation, as well as with host protein stability [13]. The importance of the detailed molecular knowledge of pathogenesis mechanisms to the development of strategies to combat bacterial infections is highlighted [13].

The functions of grimelysin of *Serratia grimesii* and protealysin of *Serratia proteamaculans* that use actin as a substrate and promote bacterial invasion was reviewed by Khaitlina et al. [14]. The *Serratia* genus comprises facultative pathogens able to cause nosocomial infections or infections in immunocompromised patients, but nosocomial infections by *S. grimesii* or *S. proteamaculans* are low [14]. The paper focused on the discovery, properties and substrate specificity of the two proteases,

their high specificity towards actin, and discussed their contribution to the invasiveness of *Serratia*, although further knowledge of the bacterium virulence factors and the cellular response mechanisms is required to fully understand the mechanism of *Serratia* invasion of the host cell [14].

The virulence factors that the bacteria use to cross the blood-brain barrier and cause meningitis is reviewed by Herold et al. [15]. Meningitis remains a worldwide problem often associated with fatalities and severe sequelae. After reviewing important traits of the central nervous system barriers to bacterial entrance, the authors review the various stages of the virulence processes of bacterial meningitis, including the processes of attachment and invasion, the routes used to enter the central nervous system, and the general mechanisms used to survive intracellularly [15]. The roles played by virulence factors produced by bacteria when crossing the central nervous system is also addressed, followed by the review of the specific traits of bacterial species more commonly associated with meningitis [15].

Coagulase-negative Staphylococci are a broad group of skin commensals that emerged as major nosocomial pathogens, with the species *S. epidermidis*, *S. haemolyticus*, *S. saprophyticus*, *S. capitis*, and *S. lugdunensis* as the most frequent pathogens [16]. In their paper, Argemi et al. reviewed the recent progress achieved in the pathogenomics of these species, based on published work supported by whole-genome data deposited in public databases [16]. As stated by the authors, the ever increasing amount of data available at the genomic, molecular, and clinical levels is expected to enhance the development of innovative approaches to characterize the pathogenicity of this bacterial group of pathogens [16].

Bacteria of the *Trueperella pyogenes* species are considered as belonging to the microbiota of animals skin and mucous membranes of the upper respiratory and urogenital tracts, but it is also an important opportunistic pathogen to animals, leading to important economic losses [17]. In their paper, Rzewuska et al. reviewed the taxonomy of the species, their pathogenicity to animals, and the various diseases associated, as well as their possible involvement in zoonotic infections, as well as the reservoirs and routes of transmission and infections [17]. The authors also present a thorough review of the main virulence factors used by the organism, including pyolysin, fimbriae, extracellular matrix-binding proteins, neuraminidases, and ability to form biofilms [17]. The availability of complete genome sequences and a better knowledge of *T. pyogenes* virulence factors, transmission routes, and epidemiology of infections is expected to lead to the development of effective vaccines, with particular hope deposited on DNA vaccines [17].

Candidiasis are on the rise worldwide, with *Candida albicans* and *Candida glabrata* as the more prevalent etiologic agents of these fungal diseases [18]. The paper by Galocha et al. thoroughly reviewed the distinct strategies used by the two *Candida* species to successfully cause human infections, starting by the adhesion and ability to form biofilms [18]. While *C. albicans* is dimorphic, growing as yeast or pseudohyphae, *C. glabrata* cannot undergo hyphal differentiation. As a consequence, *C. albicans* relies on the production of proteolytic enzymes and hyphal penetration to invade the host cell, while *C. glabrata* is thought to invade host cells by inducing endocytosis [18]. The authors extensively review the distinct mechanisms used by the two pathogenic to evade the host immune system, and succeed as pathogens. The detailed knowledge of the virulence mechanisms is critical to develop therapies that specifically target virulence traits of these two pathogenic yeasts [18].

Bacterial small non-coding regulatory RNAs (sRNAs) have emerged over the last decade as key regulators of post-transcriptional regulators of gene expression, being involved in a wide range of cellular processes, including bacterial virulence [19]. In their review, Pita et al. updated knowledge on sRNAs from two pathogens associated with respiratory infections and lung function decline of patients suffering from Cystic Fibrosis, *P. aeruginosa* and bacteria of the so-called *Burkholderia cepacia* complex (Bcc) [20]. As stated by the authors, the knowledge on *P. aeruginosa* sRNAs is far more extensive than from bacteria of the Bcc. After reviewing the main molecular characteristics of bacterial RNAs and their modes of action, including the role played by Hfq as a mediator of RNA-RNA interactions, the authors detail the description of the roles played by *P. aeruginosa* sRNAs known for their involvement in virulence traits of the bacterium. Despite the shorter information on Bcc sRNAs, the authors make a brief description of known sRNAs from Bcc [19]. The identification and functional characterization of additional sRNAs from these two pathogens will certainly enlighten our knowledge on their virulence traits.

The development of new tools to investigate microbial pathogenesis, at the molecular and cellular level, is of keen importance to comprehend how the microorganism can invade the host and cause infection. The paper from Hatlem et al. reviewed the basic molecular traits and applications of the SpyCatcher-SpyTag system, originally developed as a method for protein ligation [20]. The system consists of a modified domain of the SpyCatcher surface protein from *Streptococcus pyogenes* that recognizes the cognate SpyTag peptidic sequence composed of 13 amino acid residues [20]. Upon recognition, a covalent isopeptide bond is formed between a lysine side chain of the SpyCatcher and an aspartate of the SpyTag [20]. The authors describe in detail the fundamentals of the system and of related variants, emphasizing their uses in molecular studies of microbial virulence factors, surface proteins, membrane dynamics, as well as in the development of vaccines [20].

Microorganisms employ a wide array of virulence factors to successfully thrive and flourish with their hosts, leading this interaction to the development of infections that can often be fatal. The molecular knowledge of the virulence traits, associated with the recent availability of genomics data and bioinformatics tools for the more frequent human pathogens, is expected to lead in the near future of novel molecules and strategies to battle infectious diseases.

**Funding:** This research was funded by Fundação para a Ciência e a Tecnologia, through Project UIDB/04565/2020 from IBB—Institute for Bioengineering and Biosciences.

**Acknowledgments:** IBB—Institute for Bioengineering and Biosciences is acknowledged for funding.

**Conflicts of Interest:** The author declares no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

#### **References**


© 2020 by the author. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## **The Serine Protease Autotransporters TagB, TagC, and Sha from Extraintestinal Pathogenic** *Escherichia coli* **Are Internalized by Human Bladder Epithelial Cells and Cause Actin Cytoskeletal Disruption**

**Pravil Pokharel 1,2, Juan Manuel Díaz 2,3, Hicham Bessaiah 1,2, Sébastien Houle 1,2, Alma Lilián Guerrero-Barrera <sup>3</sup> and Charles M. Dozois 1,2,4,\***


Received: 1 March 2020; Accepted: 23 April 2020; Published: 26 April 2020

**Abstract:** TagB, TagC (*t*andem *a*utotransporter *g*enes *B* and *C*), and Sha (*S*erine-protease *h*emagglutinin *a*utotransporter) are recently described members of the SPATE (serine protease autotransporters of *Enterobacteriaceae*) family. These SPATEs can cause cytopathic effects on bladder cells and contribute to urinary tract infection in a mouse model. Bladder epithelial cells form an important barrier in the urinary tract. Some SPATEs produced by pathogenic *E. coli* are known to breach the bladder epithelium. The capacity of these newly described SPATEs to alter bladder epithelial cells and the role of the serine protease active site were investigated. All three SPATE proteins were internalized by bladder epithelial cells and altered the distribution of actin cytoskeleton. Sha and TagC were also shown to degrade mucin and gelatin respectively. Inactivation of the serine catalytic site in each of these SPATEs did not affect secretion of the SPATEs from bacterial cells, but abrogated entry into epithelial cells, cytotoxicity, and proteolytic activity. Thus, our results show that the serine catalytic triad of these proteins is required for internalization in host cells, actin disruption, and degradation of host substrates such as mucin and gelatin.

**Keywords:** SPATEs; UTIs; cytotoxicity; serine proteases; 5637 bladder cells; mucin; gelatin; actin

#### **1. Introduction**

Urinary tract infections (UTIs) present a broad range of symptoms and include urosepsis, pyelonephritis (or upper UTI,withinfectionin the kidney), and cystitis (orlower UTI,with bacteriainfecting the bladder) [1,2]. Uropathogenic *Escherichia coli* (UPEC) is the main cause of community-acquired UTIs (about 80–90%) [3], and the ability of UPEC to establish a UTI is due to the expression of a variety of virulence factors. These factors include type 1 and P fimbriae (pili), flagella, capsular polysaccharides, iron acquisition systems, and toxins including hemolysin, cytotoxic necrotizing factor (CNF), and serine protease autotransporters of *Enterobacteriaceae* (SPATEs) [4].

The bladder urothelium constitutes a physical barrier to ascending urinary tract infections [5]. UPEC can produce toxins that damage bladder tissue and can lead to release of host nutrients and

counter host defenses and innate immunity. A pore-forming toxin HlyA, can lyse erythrocytes and nucleated host cells [6], induce apoptosis [7], promote exfoliation of bladder epithelial cells and cause extensive uroepithelial damage [8–11]. Another UPEC toxin, cytotoxic necrotizing factor 1 (CNF1), has been reported to mediate bacterial entry into host epithelial cells [12], induce apoptotic death of bladder epithelial cells [13], and potentially promote bladder cell exfoliation [13]. SPATEs such as Sat, Pic, and Vat were also shown to affect bladder or kidney epithelial cells [14–16].

An important step to understand the role of SPATEs in UPEC pathogenesis is to elucidate molecular mechanisms underlying their effect on the bladder epithelium and during urinary tract colonization. The proteolytic activity of SPATEs is mediated by a serine protease catalytic triad of aspartic acid (D), serine (S), and histidine (H), wherein serine is the nucleophile, and aspartic acid interacts with histidine [17]. Mutations within the catalytic triad have been shown to abolish proteolytic activity in a number of SPATEs [15,17–19].

Recently, members of our group identified three new SPATEs: TagB, TagC (*t*andem *a*utotransporter *g*enes *B* and *C*), and Sha (*S*erine-protease *h*emagglutinin *a*utotransporter) in some strains of extra-intestinal pathogenic *E. coli* (ExPEC). In ExPEC strain QT598, *tagB* and *tagC* are tandemly encoded on a genomic island, and were present in 10% of UTI isolates and 4.7% of avian pathogenic *E. coli* (APEC) that we screened [20]. Further, Sha, which is encoded on a virulence plasmid in strain QT598 was present in 1% of UTI isolates and 20% of avian pathogenic *E. coli* [20]. The *tagBC* genes are also present in the genomes of sequenced UPEC strains such as multidrug-resistant CTX-M-15-producing ST131 isolate *E. coli* JJ1886 (Accession number CP006784), *E. coli* CI5 (Accession number CP011018), and multidrug-resistant uropathogenic *E. coli* strain NA114 (Accession number CP002797.2). When cloned into *E. coli* K-12, TagB, TagC, and Sha mediated autoaggregation, hemagglutination, and adherence to human HEK 293 renal and 5637 bladder cell lines, but did not contribute significantly to biofilm production [20]. Further, TagB and TagC exhibited cytopathic effects on the bladder epithelial cell line [20]. Following transurethral infection of CBA/J mice with a *tagBC* mutant or *sha* mutant, no significant difference in colonization was observed. However, the competitive fitness of a mutant derivative lacking all of the SPATEs present in QT598 was significantly lower in the kidney [20].

The purpose of this report was to more fully investigate the effects of the TagB, TagC, and Sha SPATEs on the 5637 bladder epithelial cell line focusing on the actin cytoskeleton. We also investigated potential entry of SPATE proteins within these bladder epithelial cells and whether they demonstrate mucinase or gelatinase activity.

#### **2. Results**

#### *2.1. Processing and Secretion of TagB, TagC, and Sha Is Independent of the Serine Protease Motif*

To evaluate the importance of the serine protease motif for processing and secretion of three novel SPATEs, we generated variant proteins of TagB, TagC, and Sha lacking the serine catalytic site. Plasmids expressing TagB, TagC, or Sha [20] were used as the templates for construction of site-directed mutant clones where the serine site was substituted for an alanine at residue S255, S252, and S258 respectively (Figures S1 and S2). Each of these three plasmids expressing mutant SPATEs, produced a high-molecular-weight protein (>100 kDa) in culture supernatants that corresponded to the expected size of the native protein, and also lacked breakdown products that are present in samples containing native SPATEs that exhibit some autoproteolytic activity (Figure 1A, asterisks). This demonstrated that the serine protease motif is not necessary for SPATE secretion and release from bacterial cells.

To further localize each of the SPATEs expressed from plasmids in *E. coli* BL21 on the bacterial cell surface, we used immunogold labeling and transmission electron microscopy. Polyclonal antibodies against the entire secreted Vat SPATE [20] were used as they were shown to strongly cross-react and recognize conserved epitopes of the other SPATEs. Thus, we used these polyclonal "SPATE antibodies" for the detection of other SPATEs in our experiments.

*E. coli* BL21 pBCsk+ expressing TagB, TagC, and Sha were immunogold-labeled (Figure 1B–D); demonstrating localization of these SPATEs on the bacterial surface, as well as release into culture

supernatant. The inset depicts the heavily concentrated proteins on the bacterial surface that appears to cluster around each other and produce fiber-like aggregates on the cell surface (white chevron, Figure 1B–D). *E. coli* BL21 bacteria containing only the empty plasmid vector were not labeled after immunogold labeling with anti-SPATE antibodies and secondary anti-rabbit immunoglobulin conjugated to 10-nm gold particles (Figure 1E).

**Figure 1.** (**A**) Silver stained SDS-PAGE analysis of concentrated supernatants of *E. coli* BL21 expressing SPATE proteins. Filtered supernatants from clones expressing TagB, TagC, and Sha or the variant TagB S255A, TagC S252A, and Sha S258A proteins were concentrated through Amicon filters with a 50 kDa cutoff. Samples containing 5 μg of protein were migrated and stained with silver stain. (**B**–**E**) Immunogold Electron Microscopy (EM) of SPATEs (Serine protease autotransporters of *Enterobacteriaceae*) localized to the outer membrane and extracellular medium. Immunogold-TEM micrographs of SPATEs using SPATE-specific antiserum. Bacteria were cultured to 0.6 OD600nm in Luria-Bertani medium. *E. coli* BL21 pBCsk+ expressing TagB (**B**), TagC (**C**), and Sha (**D**) labelled with immunogold particles. (**E**) *E. coli* BL21 pBcsk+ (vector only control) shows no immunogold staining. Insets represents boxed areas of higher magnification showing clustering of SPATE proteins. All images were acquired at ×17,000 magnification; scale bars represent 1 μm, and 0.5 μm (Insets).

#### *2.2. The Serine Catalytic Motif of SPATEs Is Not Required for Autoaggregation or Hemagglutination Activity*

To determine if the serine catalytic site of each of the three SPATEs is involved in autoaggregation or hemagglutination activity, we tested these phenotypes with our mutant SPATEs as described in [20]. Macroscopic analysis of autoaggregation of *E. coli* BL21 expressing TagB S255A, TagC S252A, or Sha S258A showed that bacterial cells settled at the bottom of the tube under static incubation similar to their respective native protein-expressing clones. The percentage of reduction in turbidity is given as

a percentage of the initial OD600 value and was similar for both mutant and native proteins (Figure 2). The reduction in turbidity of the negative control *E. coli* BL21 pBCsk+ was significantly lower compared to clones expressing TagB, TagC, Sha, or AIDA-1 (positive control) (Figure 2). AIDA-1 (Adhesin Involved in Diffuse Adherence) of *E. coli* is a characterized self-associating autotransporter protein which mediates bacterial cell–cell interactions and autoaggregation [21].

These results show that inactivation of the serine catalytic site in each SPATE does not affect autoaggregation. It is therefore likely that other motifs or residues present in these proteins contribute to autoaggregation of bacterial cells.

**Figure 2.** The autoaggregation phenotype is independent of the serine protease motif. Clones of *E. coli* BL21 expressing TagB, TagC, Sha, or their respective serine-site mutants were grown 18 h and adjusted to an OD600 of 1.5 and left to rest at 4 ◦C. Samples were taken at 1 cm from the top surface of the cultures after 3 h to determine the change in OD600. Assays were performed in triplicate, and the rate of autoaggregation was determined by the mean decrease in OD600nm after 3 h. *E. coli* BL21 pBCsk+ vector without insert (empty vector) was used as a negative control and the AIDA-1 autotransporter was the positive control for autoaggregation. Error bars represent standard errors of the means (\*\*\* *p* < 0.001 compared to empty vector using one-way ANOVA).

Likewise, Sha S258A showed similar hemagglutination of human blood as reported previously for the Sha native protein [20]. When cloned in the hemagglutination negative *E. coli* strain ORN172, there was no hemagglutination activity for either the native or mutant TagB or TagC proteins. In addition to autoaggregation and hemagglutination activity, the adherence capability of the serine catalytic site mutants of the SPATEs to 5637 human bladder epithelial cells was not affected (Figure S3). Hence, loss of the serine catalytic site did not affect autoaggregation, adherence, or hemagglutination phenotypes associated with each of the SPATEs compared to the native proteins.

#### *2.3. Cytopathic E*ff*ect of TagB and TagC Requires the Serine Protease Motif*

To assess the role of the serine protease motif for the cytopathic effect of SPATEs, extracts of supernatants of the different SPATEs (30 μg of protein per well) were incubated with human bladder epithelial cell line 5637 for 5h. Then the cells were fixed, stained with Giemsa stain, and observed by light microscopy. Cytopathic changes (dissolution in cytoplasm, enlargement of the nucleus with vacuoles) observed under the microscope for TagB and TagC (Figure 3A) were absent from cells treated with either TagB or TagC proteins lacking the serine protease active site. In addition, no significant morphological changes were observed with cells treated with either Sha or the mutant, Sha S258A, protein. No cytopathic effect was observed after treatment of cells with concentrated filtered supernatant from *E. coli* BL21 pBCsk+ (empty vector) or media alone (Figure 3A). To examine this cytopathic effect quantitatively, we measured lactate dehydogenase (LDH) release from epithelial cells incubated with each of SPATEs or their respective catalytic site mutant proteins. There was release of LDH after 5 h upon exposure of cells to TagB or TagC. However, the catalytic site mutant proteins did not release LDH from cells (Figure 3B). Further, no LDH release was detected from cells treated with either Sha or its catalytic site mutant variant (Figure 3B), indicating that cytotoxicity to human bladder epithelial cells by TagB and TagC was dependent on the serine protease catalytic site.

**Figure 3.** The serine catalytic site is necessary for the cytopathic effect of TagB and TagC. (**A**) Concentrated supernatants containing 30 μg of protein per well derived from *E. coli* BL21 clones expressing TagB, TagC, Sha, or their respective serine mutant variant proteins were incubated with monolayers of the 5637 human bladder epithelial cell line for 5 h at 37 ◦C. Cytopathic effects (white triangle) were absent in cells treated with the serine catalytic site mutant variants of TagB or TagC. The empty vector (pBCsk+) without insert was used as a negative control. The scale bar represents 20 μm. (**B**) Cytotoxicity measured by LDH release from 5637 human bladder cells after incubation with supernatant filtrates of different clones (30 μg of protein per well) at 37 ◦C for 5 h. Empty vector (pBCsk+) was used as a negative control and maximum LDH release (positive control) was determined by treatment with lysis solution. Data are the means of three independent experiments, and error bars represent the standard errors of the means. Significant differences between lysis caused by native and mutant SPATEs were determined using Student's *t*-test with \*\*\* *p* < 0.001.

#### *2.4. Exposure to TagB, TagC, or Sha Alters Actin Distribution in Bladder Epithelial Cells*

Based on the cellular changes seen with bladder epithelial cells after exposure to TagB and TagC, we hypothesized that TagB and TagC could alter the distribution of cytoskeletal components such as actin, with actin being one of the most abundant intracellular proteins in the eukaryotic cell. So, to examine the effect on F-actin cytoskeleton organization, 5637 bladder cells were incubated with native and mutant TagB, TagC, or Sha (30 μg of protein per well) for 5 h at 37 ◦C, stained with fluorescently labeled phalloidin, and then observed under confocal microscopy. Cells treated with the supernatant extract from the empty vector containing clone were uniform, smooth-edged, and contained clearly visible actin stress fibers (yellow triangle) and strong actin staining around the cell (Figure 4A). By contrast, bladder cells treated with TagB showed reduced actin stress fibers and less actin staining (Figure 4A). Bladder cells treated with TagC, also had a pronounced effect on the cytoskeleton as demonstrated by the absence of actin stress fibers and reduced levels of actin staining. Sha treated cells showed a loss of actin stress fibers and the presence of punctate patterns of actin within the cytoplasm of the cells (yellow arrowheads, Figure 4A). By contrast, the TagB, TagC, and Sha mutants lacking the serine protease catalytic sites demonstrated no changes in the actin cytoskeleton and had actin stress fibers similar to negative control cells, indicating that the serine protease activity of these SPATEs mediates the changes in actin distribution within bladder cells. To quantify the level of phalloidin binding, we measured the staining intensity and distribution of fluorescence of phalloidin around each cell using ImageJ software [22]. Fluorescence intensity for cells was calculated using the channel for actin staining. In comparison with the negative control (empty vector), the density of F-actin staining was significantly lower in cells treated with TagB, TagC, or Sha. Cells treated with the serine catalytic site mutant proteins, demonstrated F-actin staining that was greater when compared to cells treated with the native SPATE proteins (Figure 4B). Overall, these results demonstrate that these SPATEs alter the cytoskeleton and reduce the distribution of actin in bladder epithelial cells.

**Figure 4.** *Cont*.

**Figure 4.** *Cont*.

**Figure 4.** Effects of TagB, TagC, and Sha on the actin cytoskeleton of bladder epithelial cells is serine-protease-motif dependent. (**A**) Concentrated supernatant extracts (30 μg of protein per well) from *E. coli* BL21 clones expressing TagB, TagC, or Sha and their respective serine catalytic site mutants were incubated with monolayers of human bladder (5637) epithelial cells for 5 h at 37 ◦C. After incubation, cells were fixed and permeabilized. Actin was stained with fluorescently labeled phalloidin (green) and the nucleus was stained by DAPI (blue). Cells treated with the filtered supernatant of *E. coli* BL21 pBCsk+ without insert (empty vector) were used as a negative control. Slides were observed by confocal microscopy. Inset images from the left panels are magnified in the panels to the right. Bars represent 10 μm. (**B**) Quantitative analysis of fluorescence intensity of F-actin. Analysis of fluorescent intensity was done at the original magnification by measuring the mean gray value with ImageJ software [22] with an *n* value of at least 10 cells. Data values represent the mean ± SEM of at least three independent experiments. (\* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001 one-way ANOVA with multiple comparisons).

#### *2.5. SPATE Entry into Bladder Epithelial Cells Is Dependent on the Serine Protease Active Site*

We previously showed that TagB and TagC demonstrated cytotoxicity as measured by lactate dehydrogenase (LDH) release from epithelial cells within 5 h [20]. This toxicity could be due to the interaction of the SPATEs with targets inside host cells. So, to gain insight into the potential internalization of these SPATEs, we employed immunofluorescence labeling of proteins followed by visualization using confocal or immunogold electron microscopy. Firstly, confocal Z-sections (optical slices) of 5637 bladder cells treated with SPATEs were examined to determine if SPATEs were translocated within cells. After 5 h of incubation, TagB, TagC, and Sha (red color) were found within cells as evidenced by cell sectioning analysis (Figure 5A). By contrast, the serine active-site mutant variants were unable to enter epithelial cells and were not detected (absence of red staining) (Figure 5A), suggesting that serine protease activity is needed for the entry of SPATEs within cells. Interestingly, TagB within cells also co-localized with actin (green color) in the outer border of the cell (Figure 5A). Further, cells incubated with serine mutant variants of SPATEs did not enter cells, and these cells also produced actin stress fibers (Figure 5).

**Figure 5.** *Cont*.

**Figure 5.** Intracellular localization of TagB, TagC, and Sha determined by confocal microscopy. (**A**) Z-stack imaging showing the localization of TagB, TagC, and Sha and their respective serine active site mutant variants during interaction with 5637 bladder epithelial cells after 5 h of incubation. SPATEs were detected by Alexa Fluor 594 (white arrowheads, red fluorescence) using anti-mouse secondary antibody and actin was stained by Alexa Fluor 488- phalloidin (green fluorescence). Images are displayed in a 3D section view with large Z-sections in the X-Y direction (R), Z-projection in the X–Z direction (P), and Z-projection in the Y–Z direction (Q). The green and red lines in R indicate the orthogonal planes of the X–Z and Y–Z projection. For each selected section, the signal was gathered from a span of 5 μm. Scale bar: 10 μm (**B**) Quantitative analysis of fluorescence intensity of F-actin in the cells treated with native or mutant SPATEs. Analysis of fluorescence intensity was done in green channel by measuring the mean gray value on ImageJ. Data represent the mean ± SEM of at least three independent experiments. Significant differences between fluorescence intensity of each native and mutant SPATE treated cell was determined using Student's t-test with \*\* *p* < 0.01, \*\*\* *p* < 0.001.

Analysis of thin-sections of SPATE-treated cells using immunogold staining and transmission electron microscopy (TEM) also confirmed the intracellular localization of all three SPATEs within cells. TagB and Sha were found in the cytoplasm, whereas TagC was present in the nucleus (Figure 6). However, in multiple independent experiments, we failed to detect the presence of serine mutant variants of TagB, TagC, or Sha within cells. The serine catalytic-site mutant proteins when visualized were almost exclusively observed on the extracellular surface of cells as seen in cells treated with TagB S255A (Figure 6D).

**Figure 6.** Transmission electron micrographs of 5637 bladder cells showing internalized SPATEs, immunolabelled with 10-nm-diameter gold particles after 5 h of incubation. Gold particles are highlighted with red triangles. (**A**) TagB is principally located in the cytoplasm (CP). (**B**) For Tag C, gold particles were associated with the nucleus (N) and cytoplasm (CP). (**C**) Sha was located mainly in the cytoplasm (CP). (**D**) For the serine mutant variants of TagB, TagC, and Sha, gold particles were only localized on the extracellular surface of cells (red box). Only the TagB S255A mutant protein localization is shown. Cell membrane (CM), Cytoplasm (CP), Nuclear Membrane (NM), Nucleus (N), Nucleolus (NC) Bars, 1 μm.

#### *2.6. Sha Exhibits Serine Protease-Dependent Mucinase Activity*

Epithelial cell damage caused by SPATEs was shown to require protease activity, and some other SPATEs were previously shown to demonstrate activity against host proteins such as mucin or gelatin [18,23]. Further, we also tested for mucinase activity, since two of the novel SPATEs identified in APEC QT598, Sha and TagB [20], belong to the class 2 SPATE family whose members have been shown to demonstrate mucinolytic activity. Clones of *E. coli* BL21 expressing each of the SPATEs were grown on agar plates containing 0.5% porcine gastric mucin for 24 h at 37 ◦C, followed by amido black-staining. Plates containing clones growing on discs expressing Sha revealed clear zones of mucin lysis (Figure 7A) and the lysis zone produced by Sha was intermediate when compared to clones expressing either Tsh (positive control) or Vat. Mucin containing plates had a clearing zone with a diameter of 3.9 ± 0.1 cm after exposure to Sha expressing bacteria, which was less than following exposure to Tsh expressing cells (4.2 ± 0.1 cm), but more than following exposure to Vat expressing cells (3.7 ± 0.2 cm). By contrast, TagB and TagC were mucinase-negative as evidenced by the absence of any clearing zones (Figure 7B). Further, the critical role of the serine catalytic site of Sha for mucinase activity was demonstrated with the clone expressing Sha S258A, which did not produce a zone of mucin lysis (Figure 7B). The clone containing only the empty vector (negative control) did not grow well in the presence of mucin and also demonstrated no clearing zone. When mucin was treated with culture filtrates of SPATE proteins (Figure 7C), it was not degraded by either TagB, TagC, or in the negative control (empty vector). Sha as well as Tsh and Vat degraded mucin, whereas the serine protease mutant of Sha, Sha S258A, did not. Hence, the serine catalytic site of Sha is required for mucinase activity.

**Figure 7.** Sha demonstrates serine-protease dependent mucinase activity, but not TagB nor TagC. Mucinase activity was tested in a medium containing 1.5% agarose and 0.5% porcine gastric mucin. Filter discs inoculated with clones containing the empty vector, expressing Sha, Vat, Tsh, (**A**) TagB, TagC, or Sha S258A (**B**) were placed on the agar surface and incubated overnight at 37 ◦C. Mucin lysis zones were visualized by staining with 0.1% amido-black in 3.5 M acetic acid for 15 min, followed by destaining with 5% acetic acid and 0.5% glycerol for 6 h to overnight. (**C**) Zones of 0.5% porcine gastric mucin hydrolysis are visible in the stacking region of the SDS-PAGE gel (boxed area), concentrated supernatant extracts of SPATEs (5 μg of protein per well) were incubated at 37 ◦C for 48 h with mucin prior to migration. The gel was stained with a PAS glycoprotein staining kit.

#### *2.7. TagC Exhibits Serine Protease-Dependent Gelatinase Activity*

Some SPATEs were previously reported to degrade extracellular matrix proteins such as collagen and gelatin [23]. We previously demonstrated that TagB, TagC, and Sha could mediate increased adherence to chicken fibroblasts [20], which are cells that are associated with connective tissues and produce extracellular matrix proteins such as collagen. The hydrolyzed form of collagen—gelatin was used as a substrate to test for potential gelatinase activity from supernatant extracts containing SPATEs. Culture supernatant filtrate from *Pseudomonas aeruginosa* was used as a positive control, since it is known to demonstrate gelatinase activity [24]. Samples were incubated with 1% bovine gelatin for 48 h at 37 ◦C. Culture filtrates containing TagC as well as other SPATEs EspC, Tsh, and Vat demonstrated gelatinase activity (Figure 8A). By contrast, neither TagB nor Sha demonstrated gelatinase activity, since high-molecular-weight bands, indicating intact gelatin, remained after exposure to these SPATEs. Further, gelatinase activity from TagC was shown to be dependent on the serine protease motif, since the *E. coli* clone expressing a serine active site mutant protein, TagC S252A, did not generate a hydrolysis zone on medium containing 1% gelatin, whereas the TagC expressing clone did exhibit a hydrolysis zone (Figure 8B).

**Figure 8.** TagC demonstrates serine-protease dependent gelatinase activity, but not TagB nor Sha. (**A**) Zones of 1% bovine skin gelatin hydrolysis are visible in the stacking region of the SDS-PAGE gel (boxed area), concentrated supernatant extracts of SPATEs (5 μg of protein per well) were incubated at 37 ◦C for 48 h prior to migration. (**B**) Gelatinase activity of TagC was tested in a medium containing 1.5% agarose and 1% bovine skin gelatin. The disc inoculated with a clone expressing TagC or its serine catalytic site mutant variant, TagC S252A, were inoculated on the agar surface and were incubated for 48 h at 37 ◦C. Zones of gelatin lysis were visualized by staining with 0.1% amido-black in 3.5 M acetic acid for 15 min, followed by destaining with 5% acetic acid and 0.5% glycerol for 6 h.

#### **3. Discussion**

Colonization of the bladder is vital for UTI pathogenesis and UPEC deploys an array of virulence factors to infect and colonize the bladder, including secreted toxins [25]. Hemolysin A [8,9], UpxA (TosA) [26], cytotoxic necrotizing factor-1 (CNF-1) [27,28], and a variety of SPATEs (serine-protease autotransporters of *Enterobacteriaceae*) [29] are known toxins of host cells that are produced by some UPEC strains. The recent identification of new members of the SPATEs family present in some pathogenic *E. coli* and their cytotoxic activity on bladder cell lines [20], led us to further investigate mechanisms underlying the cytotoxic and proteolytic activity of the TagB, TagC, and Sha SPATEs on an established human urinary bladder cell line [30,31] and other properties of these virulence-associated proteins.

TagB, TagC, and Sha proteins demonstrated autoaggregating activity, and also promoted adherence of *E. coli* strain BL21 to the human HEK 293 renal and 5637 bladder human cell lines. Further, Sha also contributed to increased biofilm production [20]. SPATEs present on the bacterial surface are likely to contribute to the autoaggregation phenomenon. (Figure 1A). TagB and TagC also exhibited cytopathic effects on the bladder epithelial cell line. Further, we also previously determined that proteolytic activity of these SPATEs was strongly inhibited upon addition of serine protease inhibitor (PMSF), providing evidence for the importance of the serine protease motif in the activity of these SPATEs [20]. To further investigate the role of the serine protease activity, we generated catalytic site mutants of these three SPATEs. It is of note that the serine protease consensus motif (GDSGS) is conserved among different members of SPATEs [20,32–35]. Importantly, loss of the serine active site did not affect the processing or secretion of the SPATE proteins into the extracellular milieu (Figure 1A). Further, loss of the serine active site also eliminated any autoproteolytic activity (Figure 1A). Similarly, autoproteolytic activity has also been reported for other SPATEs including EspP, Sat, Pic [36], and for AspA autotransporter from *Neisseria meningitidis* [37]. Thus, from our results, it is clear that the processing of the passenger domain across the bacterial surface and autocatalytic activities of the TagB, TagC, and Sha is independent of the proteolytic serine site.

We investigated the role of the serine catalytic site of TagB, TagC, and Sha in autoaggregation or hemagglutination, as either SPATE protease activity on the bacterial or host cell surfaces could have possibly mediated these phenotypes. For instance, cleavage could have led to certain domains within the protein, leading to exposure of hydrophobic sites which could promote aggregation [38]. However, the serine protease site was not required for TagB, TagC, or Sha-mediated aggregation (Figure 2). These results indicate that other specific SPATE structural domains are likely to be responsible for aggregation. However, importantly, the autoaggregation phenotype is not a generalized phenotype of SPATEs, since in previous experiments, both the Tsh and Vat SPATEs did not demonstrate any autoaggregation phenotype [20]. Currently, the molecular mechanism of autoaggregation of TagB, TagC, and Sha is unknown. Unlike the three SPATEs described herein, loss of the active site serine of the Hap adhesin, a *Haemophilus influenzae* serine protease autotransporter, abrogated autoproteolytic processing leading to retention of this AT protein on the bacterial cell surface [39]. In fact, the increase in Hap present on the bacterial surface also increased aggregation, formation of microcolonies, and adherence of *H. influenzae* to host cells [40]. With regards to hemagglutination activity of the Sha protein, the serine active site was also dispensable. We found that Sha S258A hemagglutinated human blood with a similar titer to the native Sha protein. Similarly, a Tsh S259A variant protein was also able to bind to avian erythrocytes, turkey hemoglobin, collagen IV, fibronectin, and laminin [41]. Considering that the TagB S255A, TagC S252A, and Sha S258A variant SPATEs all retained the respective phenotypes present in the native SPATE proteins, this suggests that, despite lacking catalytic activity, that these variants are likely to have maintained a properly folded conformation.

In contrast to adherence or aggregation phenotypes, the presence of a serine protease motif was clearly required for cytotoxicity and entry of the SPATE proteins into bladder epithelial cells. In this study, the TagB S255A and TagC S252A mutant proteins were no longer cytopathic. Our results are similar to those described for other SPATEs [15,41,42] which have demonstrated a key role for the serine active site with regards to any native proteolytic or cytopathic activity of SPATEs on protein substrates or host cells.

Since TagC shares 60% identity/74% similarity with another SPATE, EspC, a non-LEE-encoded enterotoxin of enteropathogenic *E. coli* (EPEC) which causes cytotoxic effects and cleavage of cytoskeletal actin-associated protein [43]; we explored potential cellular targets in relation to the cytopathic effect observed in bladder epithelial cells. Following treatment with either TagB, TagC, or Sha, reorganization of the cytoskeleton and loss of actin stress fibers were seen in bladder epithelial cells (Figure 4). The effect of TagC was severe with faint staining remaining for actin compared to TagB interaction with cells. Exposure to Sha caused punctate localization of actin within the cytoplasm. Diminished actin staining and the formation of punctate actin accumulation suggests that each of these SPATEs are targeting the actin cytoskeleton or other cellular targets that lead to modifications in actin fiber formation or distribution within bladder cells. As expected, alterations in actin distribution were absent from bladder cells exposed to the serine catalytic site mutant variant proteins TagB S255A, TagC S252A, or Sha S258A, confirming the critical cytopathic role of serine protease activity.

Many pathogens exploit host actin for various stages of infection, including cellular invasion, intracellular replication, and dissemination by different mechanisms [44,45]. Specifically, during UTIs, UPEC utilizes the Rho family GTPase member Rac1 to mediate actin polymerization for *E. coli* bladder epithelial cell invasion [46]. It has been well documented that there is a relation between intracellular growth of UPEC in the bladder epithelium and the host F-actin cytoskeleton [47]. Based on the observation of actin rearrangement observed in bladder cells, it is also possible that the TagB, TagC, and Sha SPATEs might also contribute to UPEC invasion of the bladder epithelium, as these proteases may promote adhesion and loss of integrity of the protective epithelial barrier which could increase bacterial entry into epithelial cells as well as increase entry and systemic spread of the bacteria to other tissue sites during infection.

Before reaching the epithelial cell surface in the urinary tract, bacteria must cross the protective mucus layer that is coated with mucin [48]. Mucin serves as a primary antibacterial defense in the bladder and contributes to host innate defense by providing a barrier and by trapping bacteria [49]. Many pathogens can invade or reduce the viscosity of mucin by cleaving it [50–52]. Certain SPATEs, belonging to the Class 2 family, including Pic [18,53], PicC of *Citrobacter rodentium* [53], and Tsh, demonstrate mucinase activity [54]. We, therefore, tested whether any of the three novel SPATEs were mucinolytic, and only Sha was identified as a mucinase (Figure 7). The zone of mucinolytic activity of Sha was intermediate when compared to Vat and Tsh and, as has been shown for Pic [18], the serine catalytic site of Sha was required for mucinase activity. From this standpoint, it is interesting to note that in strain QT598, 3 of the 5 SPATEs (Tsh, Vat, and Sha) demonstrate mucinase activity [20] which might facilitate bacterial colonization by degrading mucus to overcome the mucous barrier at the interface of epithelial surfaces. TagC was also shown to degrade gelatin, which is the hydrolyzed form of collagen, although this activity was absent from Sha and TagB. Collagen is an abundant and ubiquitous extracellular matrix protein that forms an essential component of connective tissues [55]. From this standpoint, the TagC protease may contribute to tissue invasion and systemic spread of ExPEC by degradation of extracellular matrix proteins. As expected, the activity of TagC on gelatin was also dependent on the active serine catalytic site. Similarly, Pic [23] also demonstrated gelatinase activity that required an active serine catalytic site.

Previous reports have described different mechanisms of internalization of SPATEs and types of cytoskeletal damage in various epithelial cells in vitro. The Pet SPATE from enteroaggregative *E. coli* (EAEC) is internalized by a retrograde trafficking pathway [56] through the Pet host cell receptor, cytokeratin 8 [57]. Once internalized, Pet causes loss of actin stress fibers due to the breakdown of spectrin [58,59]. Internalization of EspC by EPEC requires the type 3 secretion system [60] and leads to cleavage of cytoskeletal proteins [43]. Sat is secreted by UPEC, enters the cell by an unknown mechanism, and localizes to the cytoskeletal fraction of fodrin/spectrin and integrin present within bladder and kidney epithelial cells [15]. In the present report, we have demonstrated that TagB, TagC, and Sha are also internalized in bladder epithelial cells by a mechanism that requires an active serine catalytic domain. We used confocal Z-sections to verify the intracellular localization of the SPATEs within human bladder epithelial cells. Of note, we observed the internalization of TagB, TagC, and Sha within bladder cells after 5 h and this was concomitant with diminished fluorescence staining of actin in the vicinity of the localized SPATEs. This observation was pronounced following exposure to TagB, and TagB was shown to be closely associated with actin. Furthermore, to confirm the internalization of SPATEs within bladder epithelial cells, we carried out immunogold TEM of cross-sections of cells to demonstrate SPATE proteins within epithelial cells. TEM demonstrated localization of TagB and Sha in the cytoplasm, whereas TagC targeted the nucleus. We speculate that, since TagC has previously been shown to promote nuclear enlargement [20], TagC may alter nuclear targets and elicit a significant increase in nuclear size. The entry of these SPATEs into host bladder cells does not require a type 3 secretion mechanism since it is absent from *E. coli* QT598 and *E. coli* BL21, and SPATE proteins from bacterial supernatants entered bladder epithelial cells directly. Future studies will elucidate the cytoplasmic or nucleo-cytoplasmic shuttling pathways that mediate the entry and trafficking of these three SPATEs. Importantly, the serine catalytic site was required for cell entry and cytotoxicity of all three SPATEs, since serine protease active site mutants were unable to enter cells or cause any cytopathic effects, further demonstrating a critical role for the serine catalytic site of these SPATEs.

Taken together, the TagB, TagC, and Sha SPATE proteins mediate multiple activities. These include adhesion, aggregation, cytopathic effects, mucinase and gelatinase activities that may collectively contribute to different stages of bacterial infection including initial colonization, invasion of host epithelia, and an increased potential for systemic infection.

#### **4. Materials and Methods**

#### *4.1. Ethics Statement*

This study was performed in accordance with the ethical standards of the University of Quebec, INRS. A protocol for obtaining biological samples from human blood donors was reviewed and approved by the ethics committee—*Comité d'éthique en recherche* (CER 19-507, approved November 19, 2019) of INRS.

#### *4.2. Bacterial Strains, Plasmids, and Growth Conditions*

*E. coli* clones expressing TagB, TagC, or Sha were described previously [20]. All DNA constructs were transformed into *E. coli* strain BL21 or the type 1 fimbriae *fim*-negative *E. coli* strain ORN172. Strains were grown at 37 ◦C on solid or liquid Luria-Bertani medium (Alpha Bioscience, Baltimore, MD, USA) supplemented with the appropriate antibiotics when required at concentrations of 100 μg/mL ampicillin, 30 μg/mL chloramphenicol, or 50 μg/mL of kanamycin. Strains, plasmids, and primers are listed in Table 1.




**Table 1.** *Cont.*

#### *4.3. Site-Directed Mutagenesis*

Site-directed mutagenesis was performed using the Q5® Site-Directed Mutagenesis kit as specified by the manufacturer. pIJ548, pIJ544, and pIJ553 were used as a template for the construction of the serine catalytic site mutants TagB S255A (pIJ554), TagC S252A (pIJ555), and Sha S258A (pIJ556) at 25 to 50 ng per reaction with 10 pmol of each of the complementary primers. Primers used to generate the single point mutation substituting alanine for serine for TagB were 5 -TCCCGGTGACGCCGGCTCTCCT-3 and 5 -GTACCGTAGGTTGAGAGTG-3 ; TagC were 5 -AGGAGGAGACGCCGGTTCCGGA-3 and 5 -GTCACTTCATTATAAAATCCACC-3 ; and Sha were 5 -GGCTGGTGATGCCGGTTCTCCGC-3 and 5 -TCACCATAGATCGGTAATAC-3 . Following mutagenesis, all constructs were verified by sequencing at the proteomics platform of the Institut de Recherche en Immunologie et en Cancérologie (IRIC) of the Université de Montréal (Montréal, QC, Canada).

#### *4.4. Recombinant Protein and Antibody Preparation*

Expression and purification of SPATE proteins from concentrated filtered culture supernatant fractions were obtained as described previously [20] and the extract was checked by silver staining before each assay. Antibodies against ~ 112 kDa Vat protein were used to generate a Vat-specific rabbit polyclonal antibody, according to a standard protocol [64] (Laboratorio de Biología Celular y Tisular, Departamento de Morfología, Universidad Autónoma de Aguascalientes (UAA), Aguascalientes, Mexico). Since SPATE proteins contain some highly conserved epitopes, anti-Vat antibodies were used to detect and label each of the SPATE proteins. The alignment of the passenger domain of Vat with TagB, TagC, and Sha share identities of 39%, 30%, 56%, respectively. Specific epitopes are not established but Vat-antibodies demonstrate multiple conserved residues (Figure S4) and strong immune cross-reactivity. Cross-reactivity of antibodies raised against other SPATEs have also been reported. Antibodies raised against Pet protein (45% identity with EspC and 60 gaps) cross-reacted with EspC [65]. Likewise in the supernatant of CFT073, anti-Pic (44% identity with Vat and 76 gaps) antibodies were used to detect Vat and PicU SPATEs [66]. Polyclonal antisera adsorption was done by incubating the filtered supernatant of *E. coli* BL21 pBCsk+ without insert with a 1:50 dilution of the Vat polyclonal antiserum for 1 h at room temperature under mild agitation followed by centrifugation at 2000× *g* for 5 min at 4 ◦C.

#### *4.5. Autoaggregation and Hemagglutination Tests*

Autoaggregation of bacterial cells was measured by a settling assay as performed previously [20]. The sedimentation of 10 mL of each culture of *E. coli* BL21 cells expressing native or serine active site mutant SPATEs were adjusted to an OD600nm 1.5 from an overnight culture grown at 37 ◦C in liquid Luria-Bertani medium. Then, they were monitored for a reduction in turbidity from the top of

the tube which was left at 4 ◦C for 3 h. The reduction of turbidity was plotted as a ratio against the initial turbidity.

For hemagglutination assays, human blood cells (RBCs) were washed and resuspended in PBS at a final concentration of 3% using a protocol adapted from [67]. The *E. coli fim*-negative K-12 strain ORN172 expressing either native or serine active site mutant SPATEs was grown overnight at 37 ◦C in Luria-Bertani medium, harvested and adjusted to an optical density (O.D.600nm) of 60. Suspensions were serially diluted in 96-well round-bottom plates containing 20 μL of PBS mixed with 20 μL of 3% red blood cells and incubated for 30 min at 4 ◦C.

#### *4.6. Epithelial Cell Culture*

The 5637 bladder epithelial cell line was routinely cultured in RPMI 1640 medium (Thermo Fisher Scientific) supplemented with 10% heat-inactivated FBS at 37 ◦C in humidified 5% CO2, and 2 <sup>×</sup> 105 cells/well were seeded into eight-well chamber slides (Thermo Fisher Scientific, Waltham, MA, USA) and allowed to grow to 75% confluence.

To determine cytopathic effects on bladder cells, a final concentration of 30 μg/mL of native SPATEs or the serine catalytic site mutants were added directly to monolayers and incubated for 5 h in RPMI 1640 medium at 37 ◦C with 5% CO2. Cells were then washed twice with PBS (phosphate-buffered saline), fixed with 70% methanol, and stained with Giemsa stain. Cell morphology was analyzed at a magnification of ×20 with standard bright-field light microscopy. For the lactate dehydrogenase assay, supernatant from cells treated with native or mutant SPATEs were collected and the release of LDH in cell culture supernatants were quantified by using the CytoTox 96® Non-Radioactive Cytotoxicity Assay kit (Promega, Madison, WI, USA). Maximum LDH release (positive control) was determined by adding lysis solution (provided in the kit) to the non-infected cells.

For fluorescence actin-staining and immunostaining assays, cells were fixed with 3.0%–4.0% formaldehyde in PBS, washed, permeabilized by addition of 0.1% Triton X-100-PBS, stained with 0.05 μg of Alexa Fluor 488-phalloidin/mL (AAT Bioquest, Sunnyvale, CA, USA) at 37 ◦C for 1 h and counterstained with ProLong Gold/DAPI antifade reagent (Invitrogen, Carlsbad, CA, USA). After image acquisition using confocal microscope, the actin staining intensity was quantified by measuring mean gray value (mean pixel intensity) in ImageJ (https://imagej.nih.gov/ij/) [22]. The cells of interest as well as background with no fluorescence were selected manually to analyze the areas integrated intensities and mean gray value. The value was then corrected and total fluorescence (CTF) was calculated as CTF = Integrated Density – (Area of selected cell X Mean fluorescence of background readings). The averaged corrected mean gray value was used to generate relative quantitative comparison of fluorescence intensity.

SPATE protein localization in bladder cells was detected by immunofluorescence. Treated cells were fixed, permeabilized, and incubated with blocking solution (PBS with 5% BSA) for 1 h at 37 ◦C. Samples were then incubated with rabbit anti-SPATE polyclonal antibodies (UAA, Mexico) for 2 h at 37 ◦C. This was followed by incubation with secondary antibody Alexa Fluor 594-labeled goat anti-rabbit IgG antibody (Thermo Fisher Scientific, Waltham, MA, USA). Samples were mounted and imaged with the 60X objective of an LSM780 confocal microscope (Carl Zeiss microscopy Gmbh, Jena, Germany). Images were processed with ZEN 2012 software (Carl Zeiss microscopy Gmbh, Jena, Germany).

#### *4.7. Electron Microscopy*

Immunogold labeling of bacteria was carried out by culturing *E. coli* BL21 expressing different SPATEs in Luria-Bertani medium supplemented with 30 μg/mL chloramphenicol for 5 h. Bacterial suspensions (50 μL) were spotted on nickel-coated TEM grids. After 15 min, liquid was wicked away with bibulous paper and blocked with drops of PBS containing 1% ovalbumin for 15 min. A blocking solution was exchanged with a drop of SPATE antiserum diluted 1:100 in PBS. After 15 min, excess fluid was wicked away with bibulous paper and exchanged for PBS containing 1% ovalbumin drops for 5 min. The wash was repeated and then incubated in suitable goat anti-rabbit IgG (H+L), Alexa

Fluor 488–10 nm colloidal gold secondary antibodies (Thermo Fisher Scientific, Waltham, MA, USA) diluted 1:250 in incubation solution. After 15 min, grids were washed twice with PBS drops and rinsed twice with distilled water. Grids were dried with bibulous paper and imaged on a Philips CM-100 transmission electron microscope.

For immunogold labeling of epithelial cell thin sections, cells were fixed in 0.1% glutaraldehyde + 4% paraformaldehyde in cocodylate buffer at pH 7.2, and post-fixed in 1.3% osmium tetroxide in collidine buffer. After dehydration by successive passages through 25, 50, 75, and 95% solutions of acetone in water (for 15–30 min each) samples were immersed for 16–18 h in SPURR: acetone (1:1). Samples were then embedded in BEEM capsules using SPURR resin with the ELR-4221 kit (Polysciences Inc, Warrington, PA, USA) followed by placing the capsules at 60–65 ◦C for 20–30 h to polymerize the resin. After resin polymerization, samples were cut using an ultramicrotome (Ultratome) and were put onto Formvar and carbon covered-copper 200-mesh grids treated with sodium metaperiodate and were blocked with 1% BSA in PBS. Grids were then incubated with primary antibodies, washed, and incubated with goat anti-rabbit IgG (H+L), Alexa Fluor 488–10 nm colloidal gold secondary antibodies (Thermo Fisher Scientific, Waltham, MA, USA). After washing, samples were contrasted with uranyl acetate and lead citrate and subsequently visualized using a Philips EM 300 transmission electron microscope.

#### *4.8. Cleavage of Protein Substrates*

For mucinase activity, cultures of *E. coli* BL21 expressing SPATEs were incubated for 24 h at 37 ◦C on a medium containing 1.5% agarose and 0.5% porcine gastric mucin (Sigma-Aldrich, St. Louis, MI, USA). Plates were subsequently stained with 0.1% amido-black in 3.5 M acetic acid for 15 min, followed by destaining with 5% acetic acid and 0.5% glycerol for 6 h to overnight. Zones of mucin lysis were visualized as discolored halos around colonies. For the Periodic Acid Schiff (PAS) assay to detect mucin degradation [53], 5 μg of each SPATE protein were incubated with 5 μg of 0.5% porcine gastric mucin (Sigma-Aldrich) in 30 μL of MOPS buffer and incubated for 48 h at 37 ◦C. Treated samples were electrophoresed on an 8% SDS-PAGE gel and the gel staining was developed using a colorimetric Pierce™ Glycoprotein Staining kit (ThermoFisher Scientific, Waltham, MA, USA).

For gelatinase activity, 5 μg of each SPATE protein were incubated with 5 μg of bovine skin gelatin (Sigma-Aldrich, St. Louis, MI, USA) in 30 μL of MOPS buffer and incubated for 48 h at 37 ◦C. Samples were then boiled with Laemmli sample buffer, were electrophoresed on an 8% SDS-PAGE gel and then resolved by Coomassie blue staining. In addition, gelatinase activity was also tested by growing the clones on agar plates containing 1.5% agarose and 1% bovine skin gelatin for 48 h at 37 ◦C. Plates were subsequently stained with 0.1% amido-black in 3.5 M acetic acid for 15 min, followed by destaining with 5% acetic acid and 0.5% glycerol for 6 h to overnight. Zones of gelatin lysis consist of discolored halos around colonies.

#### *4.9. Statistical Analysis*

Experimental data were expressed as a mean ± standard error of the mean (SEM) in each group. The means of groups were combined and analyzed by a two-tailed Student *t*-test for pairwise comparisons and analysis of variance (ANOVA) to compare means of more than two populations. A *p* value of <0.05 was considered statistically significant. All data were analyzed with the Graph Pad Prism 7 software (GraphPad Software, San Diego, CA, USA).

#### **5. Conclusions**

In conclusion, TagB, TagC, and Sha are novel SPATEs that demonstrate different proteolytic activities on different substrates as well as distinct cytopathic effects on bladder epithelial cells. Additional molecular *in vitro* and *in vivo* studies are in progress in an effort to understand the link between protease activity of the TagB, TagC, and Sha SPATEs and how these proteases disrupt or alter the actin cytoskeleton during ExPEC infections. It will be of further interest to also investigate

their potential interactions with other host cells or extracellular matrix proteins, and determine how these relatively large proteins (generally greater than 100 kDa) manage to enter host cells through serine protease activity and what specific trafficking pathways may be involved in their localization or association with specific cellular compartments.

**Supplementary Materials:** Supplementary materials can be found at http://www.mdpi.com/1422-0067/21/9/3047/s1.

**Author Contributions:** Conceptualization, Investigation, Data curation, Methodology, Validation, Writing-of the manuscript, P.P.; Methodology, Investigation, Software, Data curation, J.M.D., H.B., and S.H.; Writing—review and editing, J.M.D., H.B., S.H. A.L.G.-B. and C.M.D.; project administration, S.H., C.M.D.; Funding acquisition, Supervision, C.M.D. All authors have read and agreed to the published version of the manuscript.

**Funding:** Funding for this work was supported by NSERC Canada Discovery Grants 2014-06622 and 2019-06642 and scholarships from the CRIPA-FRQNT network Funds n◦RS-170946 to P.P. and J.M.D.

**Acknowledgments:** We thank Arnaldo Nakamura for assistance in electron microscopy and Jessy Tremblay for assistance in confocal immunofluorescence microscopy imaging.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**


#### **References**


and Necrosis/Lysis in Vitro and Necrosis/Lysis and Lung Injury in a Rat Pneumonia Model. *Am. J. Physiol. Lung Cell. Mol. Physiol.* **2005**, *289*, L207–L216. [CrossRef]


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Antibiotic Resistance, Virulence Factors, Phenotyping, and Genotyping of Non-***Escherichia coli* **Enterobacterales from the Gut Microbiota of Healthy Subjects**

**Alberto Amaretti 1,2,**†**, Lucia Righini 1,**†**, Francesco Candeliere 1, Eliana Musmeci 1, Francesca Bonvicini 3, Giovanna Angela Gentilomi 3,4, Maddalena Rossi 1,2 and Stefano Raimondi 1,\***


Received: 2 March 2020; Accepted: 5 March 2020; Published: 7 March 2020

**Abstract:** Non-*Escherichia coli* Enterobacterales (NECE) can colonize the human gut and may present virulence determinants and phenotypes that represent severe heath concerns. Most information is available for virulent NECE strains, isolated from patients with an ongoing infection, while the commensal NECE population of healthy subjects is understudied. In this study, 32 NECE strains were isolated from the feces of 20 healthy adults. 16S rRNA gene sequencing and mass spectrometry attributed the isolates to *Klebsiella pneumoniae*, *Klebsiella oxytoca*, *Enterobacter cloacae*, *Enterobacter aerogenes*, *Enterobacter kobei*, *Citrobacter freundii*, *Citrobacter amalonaticus*, *Cronobacter* sp., and *Hafnia alvei*, *Morganella morganii*, and *Serratia liquefaciens*. Multiplex PCR revealed that *K. pneumoniae* harbored virulence genes for adhesins (*mrkD*, *ycfM*, and *kpn*) and enterobactin (*entB*) and, in one case, also for yersiniabactin (*ybtS*, *irp1*, *irp2*, and *fyuA*). Virulence genes were less numerous in the other NECE species. Biofilm formation was spread across all the species, while curli and cellulose were mainly produced by *Citrobacter* and *Enterobacter*. Among the most common antibiotics, amoxicillin-clavulanic acid was the sole against which resistance was observed, only *Klebsiella* strains being susceptible. The NECE inhabiting the intestine of healthy subjects have traits that may pose a health threat, taking into account the possibility of horizontal gene transfer.

**Keywords:** Enterobacterales; *Klebsiella*; *Enterobacter*; *Citrobacter*; virulence; antibiotic resistance; biofilm

#### **1. Introduction**

Important enteric pathogens belong to Enterobacterales, a bacterial order within the phylum Proteobacteria. This order encompasses permanent colonizers of the human gut that, in healthy conditions, constitute minor bacterial components of the microbiota. Opportunistic Enterobacterales can persist as gut commensals without inducing any infections, as long as the microbiota is balanced and

the complex and dense bacterial community prevents their overgrowth. A bloom of Enterobacterales may occur as a result of disturbance of the microbiota, yielding pathogen-mediated infections and triggering inflammatory host responses.

*Escherichia coli* is the most studied among the Enterobacterales with regards to the traits that differentiate commensalism and pathogenicity. It normally colonizes the intestine but comprises both harmless commensals and different pathogenic variants that may instigate infections in the gut or in other tissues [1,2]. Virulent strains of *E. coli* isolated from infected patients attracted most research interest [3,4], but also fecal isolates from healthy subjects and environmental strains are the target of increasing attention, aiming to determine the pathogenic potential of a wider biodiversity reservoir [5,6].

Many non-*E. coli* Enterobacterales (hereinafter referred to as NECE) that can colonize the gut (e.g.,*Klebsiella*, *Enterobacter*, *Citrobacter*, and *Serratia*) also present traits that can confer them virulence and pathogenicity or phenotypes that may result in severe heath concern, such as multidrug resistance [7–10]. The greatest efforts have been carried out to describe virulent strains, generally isolated from patients with an ongoing infection, while the pathogenic potential of NECE inhabiting the gut of healthy subjects has not been thoroughly investigated with genetic and phenotypic analysis, except for some genera [9]. The research herein presented aims to fill this gap, providing a genotypic and phenotypic description of the NECE population isolated from the feces of 20 healthy adults, and to complement a previous study that described the *E. coli* population of the same cohort of subjects [5]. A set of 32 NECE strains was isolated, taxonomically classified, subjected to PFGE genotyping, and described in terms of the genotypic determinants and phenotypic traits that may confer on them potential pathogenicity or invasiveness.

Information on the genes associated to virulence is detailed especially for *Klebsiella* spp., where a number of genes associated to harmful traits were identified, such as those encoding adhesins, siderophores (e.g., enterobactin, aerobactin, yersiniabactin), protectines, or invasins (responsible for mucoid phenotype and invasiveness), and involved in allantoin metabolism [7,11]. For other NECE genera, such as *Enterobacter*, *Cronobacter*, and *Citrobacter*, the knowledge of the genetic determinants associated to virulence and invasiveness is less comprehensive and, with few exceptions (e.g., *Citrobacter koseri*), mainly acquired from better characterized pathogens [12–15].

In addition to fimbrial and afimbrial adhesins, the production of surface cellulose structures and curli favors the adhesion of Enterobacterales and can exert a significant role in enteric biofilm-related infections [16,17]. Although not directly involved in pathogenic mechanisms, the acquisition of multiple antibiotic resistances favors the success of opportunistic Enterobacterales pathogens in invasion, survival, and spread, severely complicating the containment and treatment of infections [9,18]. Therefore, the occurrence of drug resistant bacteria within a commensal population and the possibility to exchange genetic material by horizontal gene transfer may represent a major health concern.

In the present study, multiplex PCR assays were utilized to screen the NECE, isolated from the feces of healthy subjects, for the presence of 17 main virulence genes associated to those of *Klebsiella* and *E. coli* [19–21]. From the point of view of the phenotype, the isolates were characterized for the ability to form biofilm and to yield curli and surface cellulose, were screened for the susceptibility to the most common antibiotics and for the ability to act as recipients in conjugation experiments, and biochemical tests were performed to compare the metabolic profile.

#### **2. Results**

#### *2.1. Counting and Isolation of NECE*

The selective differential medium HiCrome Coliform Agar (HCCA) was utilized to count and isolate *E. coli* [5] and NECE from fecal samples of 20 healthy adults. Total counts in HCCA ranged from 4.6 <sup>×</sup> 105 to 2.2 <sup>×</sup> <sup>10</sup><sup>8</sup> cfu/g (Figure 1). Blue colonies attributed to *E. coli* overcounted the pink ones attributed to NECE in all the samples except V11 (Figure 1; Supplementary Figure S1). NECE ranged between <sup>&</sup>lt; <sup>10</sup><sup>4</sup> and 1.3 <sup>×</sup> 108 cfu/g and, except in V11, accounted for a minority of total Enterobacterales (NECE + *E. coli*), the 75th percentile being the 5.7% (Figure 1). In some cases, NECE were not recovered, being outnumbered by *E. coli*. Spearman's rank correlation analysis excluded any significant correlation between NECE and *E. coli* counts.

**Figure 1.** Counts of *Escherichia coli* and non-*Escherichia coli* Enterobacterales (NECE), enumerated onto HiCrome Coliform Agar (HCCA) plates. (**a**) Percentage of colonies attributed to *E. coli* (blue shades) and NECE (pink shades) in the feces of 20 subjects. For each subject, different shades indicate different biotypes according to enterobacterial repetitive intergenic consensus-PCR (ERIC-PCR) and random amplification of polymorphic DNA-PCR (RAPD-PCR) fingerprinting. The total count of Enterobacterales (*E. coli* + NECE) is reported in the top margin. (**b**) Distribution of the percentage of NECE colonies. The median (dashed line), the 25th and 75th percentiles (colored box), the 10th and 90th percentiles (whiskers), and outliers (\*) are indicated.

#### *2.2. Taxonomic Attribution and PFGE Genotyping*

The isolates putatively attributed to NECE were clustered in 32 different biotypes utilizing ERIC-PCR (enterobacterial repetitive intergenic consensus-PCR) and RAPD-PCR (random amplification of polymorphic DNA-PCR) fingerprinting. A representative isolate of each biotype was assigned a taxonomic designation utilizing 16S rRNA gene sequencing and MALDI-TOF MS (Supplementary Table S1). The genus *Klebsiella* was the most represented (14/32), with the species *K. pneumoniae* (10 strains) and *Klebsiella oxytoca* (4) found in seven and three fecal samples, respectively. The genus *Enterobacter* was represented by eight strains belonging to *Enterobacter cloacae* (6), *Enterobacter aerogenes*, and *Enterobacter kobei* (1 strain each). Other strains belonged to *Citrobacter* (4 to *Citrobacter freundii* and 1 to *Citrobacter amalonaticus*), *Cronobacter* sp. (2), and to *Hafnia alvei*, *Morganella morganii*, and *Serratia liquefaciens* (1 strain each).

PFGE highlighted a wide diversity of the NECE isolates, which did not cluster according to the taxonomic attribution (Figure 2).

**Figure 2.** *Xba*I-PFGE pattern of NECE strains: unweighted pair group method with arithmetic means (UPGMA) dendrogram derived from Dice's coefficients, calculated based on the band profile. Strains are colored based on their MALDI-TOF MS taxonomic attribution.

#### *2.3. Virulence Genotyping*

PCR was used to investigate 17 virulence genes encoding adhesins (*fimH1*, *mrkD*, *kpn*, and *ycfM*), siderophores (enterobactin, *entB*; aerobactin, *iutA*; yersiniabactin, *irp-1*, *irp2*, *ybtS*, *fyuA*; catechols receptor, *iroN*; and other, *kfu*), protectines or invasins (*K2*, *magA*, *rmpA*, and *traT*), and involved in allantoin metabolism (*allS*).

Most strains of *K. pneumoniae* harbored the *mrkD*, *ycfM*, and *kpn* encoding adhesins and *entB* encoding enterobactin (Figure 3). Only *K. pneumoniae* 11.55 was positive to the main virulence genes involved in the synthesis of *Yersinia* siderophore, including *ybtS* (encoding for the synthase), *irp1* and *irp2* (for regulatory proteins), and *fyuA* (for the siderophore receptor). *irp2* was also detected in most of the other strains of *K. pneumoniae* although they lacked the counterpart *ybtS.* All the *K. pneumoniae* strains were negative to the genes *K2*, *magA*, and *rmpA* associated with hypermucoid phenotype and invasivity, except for *K. pneumoniae* 01.49 that was positive to *K2*. Similarly, other virulence genes, such as *allS*, *kfu*, and *iutA* occurred only once among the tested strains.

The strains of *K. oxytoca* harbored *entB* (three out of four isolates) but were negative to most of the other virulence genes. A sole strain harbored *ytbS*. Most of *Cronobacter* and *Enterobacter* isolates were characterized by the presence of the gene *irp2* but never harbored *ybtS* or other *Yersinia* siderophore genes. A few strains were positive to *mrkD* or *entB*.

The strains ascribed to *Citrobacter*, *H. alvei*, and *M. morganii* were negative to those virulence genes whose presence could not be excluded by primer-blast search. The strain of *S. liquefaciens* was positive to *ytbS*.

**Figure 3.** PCR assay of the NECE isolates for the presence of virulence genes. Colors: red, positive amplification; green, negative amplification; grey, PCR analysis not performed since the gene was putatively absent based on a primer-blast search.

#### *2.4. Biofilm Formation and Production of Curli and Cellulose*

NECE strains were tested for biofilm formation in minimal and rich media (M9 and LBWS, respectively; Supplementary Figure S2). The vast majority of the strains (26 out of 32), belonging to all the species except *E. aerogenes*, *H. alvei*, and *M. morganii*, formed biofilm in M9 (Figure 4; Supplementary Figure S2). Biofilm formation was less frequent in LBWS, being observed only in 10 strains of *K. oxytoca*, *K. pneumoniae,* and *S. liquefaciens*. The extent of biofilm production was always less abundant in the rich medium compared to M9 (*p* < 0.05).

Extracellular cellulose was detected in most of *Citrobacter* and *Enterobacter* strains, in five strains of *K. pneumoniae* and two out of four of *K. oxytoca*, in *H. alvei* and in *S. liquefaciens*. Curli were produced by nearly all *Citrobacter*, *Cronobacter*, and *Enterobacter* strains and by one isolate belonging to *K. oxytoca*. The isolates of *H. alvei* and *M. morganii* were also positive to curli. The strains belonging to *Citrobacter*, *E. cloacae*, *H. alvei,* and *K. oxytoca* 19.49 produced both cellulose and curli.

**Figure 4.** Phenotypic characterization of NECE isolates: biofilm formation in LBWS and M9 media, curli and cellulose production, conjugation, and antibiotic resistance. Colors: red, positive; green, negative. For antibiotics: red, resistant; green, susceptible, yellow, intermediate. Antibiotics: amikacin (AMK), amoxicillin–clavulanic acid (AMC), cefotaxime (CTX), ceftazidime (CAZ), ciprofloxacin (CIP), gentamicin (GEN), piperacillin-tazobactam (PZT), and trimethoprim-sulfamethoxazole (SXT).

#### *2.5. Conjugation*

The strains were challenged as conjugation recipients for receiving pOX38: Cm plasmid from *E. coli* N4i. Only two strains of *K. pneumoniae*, and single strains of *Cronobacter* and *Citrobacter amalonaticus* succeeded in plasmid acquisition (Figure 4).

#### *2.6. Antibiotic Resistance*

Phenotypic susceptibility to amikacin, amoxicillin–clavulanic acid, cefotaxime, ceftazidime, ciprofloxacin, gentamicin, piperacillin-tazobactam, and trimethoprim-sulfamethoxazole was assayed. Amoxicillin-clavulanic acid was the sole antibiotic against which few isolates presented some resistance, with all the strains of *Enterobacter*, *Citrobacter*, *Cronobacter*, *H. alvei*, *M. morganii*, and *S. liquefaciens* being resistant. All the 14 biotypes of *Klebsiella* spp. were sensitive to the whole set of tested antibiotics, with the exception of *K. pneumoniae* 11.71 that was partially resistant to amoxicillin–clavulanic acid, presenting a minimum inhibitory concentration (MIC) intermediate between resistance and susceptibility thresholds.

#### *2.7. Biochemical Characterization*

The fermentation of substrates and some distinctive enzymatic reactions and metabolic routes were assayed utilizing the API 20 E system (Figure 5). Generally, the NECE strains were positive to β-galactosidase. Most strains were capable of utilizing citrate, glucose, mannose, inositol, sorbitol, rhamnose, sucrose, melibiose, amygdalin, and arabinose. The main exceptions were *M. morganii* that could utilize only glucose, *H. alvei* that fermented a restricted number of sugars, and some *Citrobacter*, *Cronobacter*, and *Enterobacter* strains that exhibited specific substrate preferences. The majority of the isolates produced either lysine decarboxylase (*Klebsiella*) or ornithine decarboxylase (*Cronobacter* and *Enterobacter*). Urease was characteristic of *Klebsiella*, while arginine dihydrolase was found in most *Enterobacter*, *Citrobacter*, and *Cronobacter*. Acetoin was produced by *Cronobacter*, *Enterobacter*, *Hafnia,* and *Klebsiella*. Indole was produced by *K. oxytoca* and by few other strains, H2S by two strains of *Citrobacter freundii*. Only *S. liquefaciens* was positive to gelatinase. All the strains except *M. morganii* and *S. liquefaciens* exhibited denitrifying activity, in most cases yielding nitrite. Nitrate reduction to N2 was observed in *K. pneumoniae* and few other strains.

**Figure 5.** Biochemical reaction profiles of NECE isolates in the API 20 E assay: β-galactosidase (ONPG), arginine dihydrolase (ADH), lysine decarboxylase (LDC), ornithine decarboxylase (ODC), citrate utilization (CIT), production of hydrogen sulfide (H2S), urease (URE), tryptophan deaminase (TDA), indole (Kovac's test, IND), acetoin (Voges-Proskauer test, VP), gelatinase (GEL), fermentation of glucose (GLU), mannitol (MAN), inositol (INO), sorbitol (SOR), rhamnose (RHA), sucrose (SAC), melibiose (MEL), amygdalin (AMY), arabinose (ARA), and reduction of nitrates to nitrites (N2O) or nitrogen (N2, tested only in case of negative N2O). Colors: red, positive; green, negative.

#### **3. Discussion**

Thirty-two NECE strains were isolated from the feces of 20 healthy adults that did not present any dysbiosis, and thus as members of a relatively balanced gut microbiota. The load of Enterobacterales was in the order of millions or tens of millions per gram of feces, with a sole exception where they reached the magnitude of 108. NECE represented a small population of Enterobacterales, with *E. coli* being on average 20 times more abundant, and a minor component of the whole microbiota, being less than 0.1%. The genus *Klebsiella*, which is ubiquitous in nature, colonizing humans, animals, and plants, and frequently detected in waters, sewages, and soils, was the most represented, encompassing 14 of the 32 strains. *K. pneumoniae*, the most frequently isolated species (10 strains), was present in 7 out of 20 fecal samples, in accordance to the 35% of healthy adults from which it was isolated in a pioneering study [22].

*K. pneumoniae* encompasses opportunistic pathogens that can cause human infections in lungs, urinary tract, and bloodstreams, mostly to hospitalized and/or immunocompromised patients [23,24]. Virulence of *K. pneumoniae* is associated to the presence of capsule and pili, to the production of lipopolysaccharides and siderophores, to allantoin utilization, and to iron uptake systems, efflux pumps, and type VI secretion systems [7]. Surface molecules, such as capsular polysaccharides and lipopolysaccharides, are some of the major virulence factors that *Klebsiella* use to protect itself from the host innate immune apparatus. Furthermore, the capability to compete for iron has a pivotal role to the establishment of the infection. The genes involved in iron assimilation are generally clustered in pathogenicity islands, large chromosomal regions that were likely acquired by horizontal transfer. Virulence genetic determinants can be located both in the core or in the accessory genome [7], the former also including metabolic genes required for some species to cause disease.

Among the 10 *K. pneumoniae* isolates, all but one strain harbored *entB*, encoding a common catecholate siderophore located in the core genome, and *mrkD*, involved in the synthesis of type 3 pili that promote adherence to the surfaces [19]. The genes encoding other adhesins, such as *ycfM* and *kpn* were detected in nine and five strains, respectively. Only two *K. pneumoniae* strains were positive to the yersiniabactin gene *ybtS*, a common virulence factor associated with human infections [25,26], and one of the two also harbored *irp1*. These genes are involved in the synthesis of the siderophore yersiniabactin by virulent *Yersinia* strains, which harbor them within the high-pathogenicity island (HPI). HPI is widely distributed among members of the order Enterobacterales, including *E. coli, K. pneumoniae*, *Citrobacter* spp., *Salmonella enterica*, *Serratia liquefaciens*, and *Enterobacter* spp. [27,28]. In addition, *irp2*, another marker of HPI, was detected in the majority of the strains of *K. pneumonia* and *E. cloacae*, albeit they lacked the counterpart *ybtS*.

Although Enterobacterales are normally present as a low fraction of commensal bacteria in the healthy gut, their numbers can increase in the inflamed gut, and take advantage over other commensals [29]. Siderophores are major contributors of exploitative competition, since iron is an essential nutrient present in very low amounts in the gut and may play a role in virulence. Overall, *K. pneumoniae* 11.55 was the strain equipped with the broadest set of virulence genes involved in iron metabolism, including those encoding enterobactin (*entB*), yersiniabactin (*irp1*, *irp2*, and *ybtS)*, and, the sole strain among the 32 isolates, also the *Yersinia* siderophore receptor (*fyuA*).

In agreement with their commensal behavior, none of the *K. pneumoniae* strains were positive for the two genes associated with invasive infections, i.e., the mucoviscosity-associated gene *magA* and the regulator of mucoid phenotype *rmpA* [30,31], that are associated with a hypervirulent and hypermucoid phenotype. In our isolates, the presence of other genetic determinants of virulence was sporadic or quite rare. In the gut of the healthy host, the Enterobacterales resides in the outer loose mucus layer, separated from the epithelial mono-cell barrier by an inner dense mucin coat [32]. During the infection, they penetrate the mucus layer, interact with the epithelial cells, and may breach the mucosal barrier. In case these enterobacteria strains reach other tissues, biofilm formation may act as a fitness factor concurring to pathogenesis [33]. Adhesion factors and extracellular matrix components are involved in formation of biofilms [34]. All but one strain produced biofilm in minimal medium M9, whereas this

phenotype was less frequent when growth occurred in the rich medium LBWS. The higher extent of biofilm formation in M9 is consistent with the fact that this mineral medium is more challenging for these bacteria. Extracellular cellulose structures, determined in five strains, were consistent with the capability to form biofilm on both media. Curli were found in a sole strain that presented also cellulose structures (19.58 CA) and produced biofilm on both the substrates.

Based on genetic and functional features, some *K. pneumoniae* isolates present a higher potential to cause infections, albeit they are present at low charge in the microbiota of healthy hosts. The link between colonization and infection by *K. pneumoniae* in hospitalized patients has been demonstrated, with robust evidence that their own microbiota is the main source of the infective strains [35,36]. Thus, potentially more virulent *K. pneumoniae* strains may take advantage of critical conditions, becoming responsible for nosocomial infections [37].

The isolates of *K. oxytoca*, another prominent pathogen that may be involved in diseases and life-threatening infections [38,39], generally encoded the siderophore gene *entB*, but were negative to most of the other virulence genes. A sole strain harbored the gene encoding the yersiniabactin siderophore. Biofilm production was a general feature of the *K. oxytoca* strains, regardless of the presence of curli or cellulose structures.

Similar considerations are valid for the other NECE strains herein described, belonging to genera that may have clinical relevance, such as *Enterobacter*, *Citrobacter*, and *Cronobacter*. The isolates were generally capable of forming biofilm and producing curli and cellulose and were negative for most virulence genes. However, unlike *Klebsiella* that shares many virulence genes with *E. coli*, the genetic determinants of virulence of these genera have not been fully disclosed.

In this study emerged that NECE isolates from feces of healthy subjects are still quite susceptible to most of the antibiotics. This is important, since any treatment of opportunistic outbreaks of NECE (e.g., in case nosocomial infections) requires antibiotics, but resistance developments would seriously curb the therapeutic options [40]. All the isolates of *Klebsiella* were sensitive to the whole set of tested antibiotics. Amoxicillin-clavulanic acid was the sole antibiotic against which was detected some resistance: the strains ascribed to the genus *Enterobacter* and the isolates belonging to *Citrobacter*, *Cronobacter*, *Hafnia alvei, Morganella morganii*, and *Serratia liquefaciens* were all resistant to this combination of antibiotics. Interestingly, Enterobacterales presented the highest increase in terms of relative abundance in a short-term amoxicillin-clavulanic acid treatment in healthy adults [41]. Some genera belonging to this family, such as *Enterobacter* and *Citrobacter*, are recognized as intrinsically resistant [42], and may take advantage to this antibiotic treatment. In general, the profile of resistance was independent by the subject of the fecal sample, but two clusters encompassing sensitive or resistant strains were sharply differentiated by taxonomy.

PFGE genotyping was carried out to evaluate the genetic similarity among the bacterial isolates of this study and highlighted a wide dispersion of the strains, regardless their taxonomic attribution and phylogenetic relationships. The spreading of the strains regardless of species or genera may be attributed to the presence of plasmids, the horizontal acquisition of additional genes from diverse species of Enterobacterales, and to the exchange of mobile elements that rapidly integrate and promote DNA shuffling, in agreement with the capability of some strains to accept DNA by conjugation from *E. coli* as a donor. However, the biochemical profiling mostly clustered the strains in species (data not shown), confirming that energy production and conservation, and lipid, amino acid, and nucleotide metabolism are part of the conserved reactome, despite the genome plasticity.

The Enterobacterales encountered in this study are generally recognized as opportunistic pathogens, with some potential capability to cause disease, on the basis of predicted virulence factors. Except for *K. pneumoniae* hypervirulent strains, NECE virulence seems more associated to the host features than to the strain traits. According to the similar results obtained for the *E. coli* isolated from the same cohort of healthy subjects, the absence of antibiotic resistance for most of the tested antibiotics does not pose a serious challenge for infection control. This highlights the stratification of antibiotic resistance distribution among healthy and hospitalized/diseased subjects, with NECE associated risk increasing with both illness and antibiotic therapy.

#### **4. Materials and Methods**

#### *4.1. Isolation and Enumeration of Enterobacterales*

Fresh fecal samples were collected from 20 healthy adult subjects who gave written informed consent regarding their participation in the study in accordance with the protocol approved by the local research ethics committee (reference number: 974/2019/SPER-UNIMO-ENTEROPOP; Comitato etico dell'Area Vasta Emilia Nord, Italy). The subjects—10 males and 10 females aged 35 to 45, following a western omnivore diet, and who had not been treated with prebiotics and/or probiotics for 1 month and antibiotics for 3 months—were enrolled among the employees of the University of Modena and Reggio Emilia and their relatives and were not in relationship with the researchers.

Feces were homogenized (10% *w*/*v*) in isotonic Buffered Peptone Water (Sigma, Steinheim, Germany), then serial dilutions were spread onto plates of HiCrome Coliform Agar (HCCA, Sigma) and incubated at 37 ◦C. The medium differentiates *E. coli* (blue colonies) from NECE (salmon to red). For each subject, up to 48 colonies of putative NECE were picked and clustered into biotypes with ERIC-PCR [43] and RAPD-PCR [44] fingerprint presenting Pearson's similarity > 75%.

#### *4.2. PFGE Genotyping*

PFGE was performed according to PulseNet protocol (http://www.cdc.gov/pulsenet/PDF/ecolishigella-salmonella-pfge-protocol-508c.pdf). The genomic DNA was digested with 50 U of *Xba*I at 37 ◦C for 3 h. Fragments were resolved in a CHEF-DRIII apparatus (Bio-Rad, Hercules, CA, USA) using counter-clamped homogeneous electric field electrophoresis (24 h at 6.0 V/cm; initial switch time, 2.2 s; final switch time, 54.2 s). The run was digitally captured and analyzed with GelCompare II 6.0 software (Applied Maths NV, Sint-Martens-Latem, Belgium). Dice coefficient was computed to evaluate similarity between band profiles (position tolerance, 1%; optimization, 1%) and to derive an UPGMA dendrogram (unweighted pair group method with arithmetic means). Strains were ascribed to the same pulsotype if PFGE profile possessed >85% similarity.

#### *4.3. Taxonomic Attribution*

A strain for each biotype was taxonomically characterized by partial sequencing of the 16S rRNA gene sequencing, utilizing primers targeting the V1-V3 portion. Primer sequences and PCR conditions were set up according to Raimondi et al. [45]. The sequences, obtained from a service provider (Eurofins Genomics, Ebersberg, Germany), were compared to those in SILVA SSU database utilizing SINA Aligner v1.2.11 (https://www.arb-silva.de/aligner/).

In addition, the MALDI-TOF MS-based biotyping was carried using the MALDI Biotyper 3.1 system (Bruker Daltonics, Bremen, Germany). Sample preparation for MALDI-TOF MS was performed as previously described with minor modifications [46]. Briefly, colonies of fresh overnight culture were placed on a MALDI sample slide (Bruker Daltonics) and dried at room temperature. The sample was then overlaid with 1 μL of matrix solution (α-cyano-4-hydroxycinnamic acid in 50% acetonitrile and 2.5% trifluoroacetic acid) and dried at room temperature. A MALDI-TOF MS measurement was performed with a Bruker MicroFlex MALDI-TOF MS (Bruker Daltonics) using FlexControl software and a *Escherichia coli* DH5α protein extract (Bruker Daltonics) was placed on the calibration spot of the sample slide for external calibration. Spectra collected in the positive-ion mode within a mass range of 2000–20,000 Da were analyzed using a Bruker Biotyper (Bruker Daltonics) automation control and the Bruker Biotyper 3.1 software and library (a database with 5627 entries). High confidence species identification was accepted, if the log(score) was ≥2.00, low confidence species identification log(score) values (≥1.70 and <2.00) were accepted if the three best matches showed the same species name. Any results with log(score)<1.70 were considered as an unacceptable identification.

#### *4.4. Profiling of Virulence Genes*

All the isolates were screened by multiplex-PCR for the genes associated to 17 virulence factors: *allS*, *entB*, *fimH*, *fyuA*, *iroN*, *irp1*, *irp2*, *iutA*, *K2*, *kfu*, *kpn*, *magA*, *mrkD*, *rmpA*, *traT*, *ybtS*, and *ycfM*. Primer sequences and amplification conditions were set up according to El Fertas-Aissani et al. [19], Compain et al. [20], and Johnson et al. [21]. In order to assess the possibility to obtain the amplicon in the different NECEs species, a search with the primer-blast tool (https://www.ncbi.nlm.nih.gov/ tools/primer-blast/) was performed for all the set of primers developed for *K. pneumoniae* and *E. coli*. The result of PCR amplification was reported only for the species in which the annealing and the possibility to yield an amplicon were predicted.

#### *4.5. Biofilm and Phenotype Assays*

Biofilm formation was quantified with crystal violet in the microtiter assay described in [47]. Two growth media were compared: Luria Bertani without salt (LBWS) and M9 (BD Difco, Sparks, MD, USA) containing 4 g/L glucose and 0.25 g/L yeast extract. Strains exhibiting a specific biofilm formation (i.e., the ratio between crystal violet absorbance at 570 nm and culture turbidity at 600 nm) > 1. The data herein reported are the means of three independent experiments, each carried out in triplicate.

The strains were screened for curli and cellulose production utilizing LBWS agar plates supplemented with the appropriate stain [48]. Red colonies in Congo red-supplemented plates were considered positive to curli. Colonies in calcofluor white-supplemented plates that emitted fluorescence due to UV exposure (315–400 nm) were considered positive to cellulose.

#### *4.6. Solid Mating Conjugation Experiments*

The strains were screened as recipients in conjugation experiments with the donor *E. coli* N4i pOX38:Cm (N4i: EcN immE7 Gmr; pOX38:Cm: Tra+ RepFIA+ Cmr) [5]. The donor and the recipient strains were cultured and put in contact onto LB plates under the conditions described in [5]. HCCA and HCCA with 20 μg/mL chloramphenicol were utilized to differentiate recipient, transconjugant, and donor colonies [5].

#### *4.7. Antibiotic Susceptibility*

The strains were tested for antimicrobial susceptibility with a Vitek2 semi-automated system (bioMerieux, Marcy-l'Étoile, France). Minimum inhibitory concentrations (MICs) were interpreted according to EUCAST (European Committee on Antimicrobial Susceptibility Testing—www.eucast.org) and susceptibility (S) or resistance (R) were defined based on the following thresholds (mg/L): amikacin, S < 8 and R > 16; amoxicillin/clavulanic acid, S < 8 and R > 8; cefotaxime, S < 8 and R > 2; ceftazidime, S < 8 and R > 8; ciprofloxacin, S < 0.5 and R > 1; gentamicin, S < 2 and R > 4; piperacillin + tazobactam, S < 8 and R > 16; trimethoprim/sulfamethoxazole, S < 40 and R > 80.

#### *4.8. Biochemical Characterization*

The strains were tested for distinctive enzymatic reactions and metabolic routes utilizing API 20 E test system (bioMerieux, France), according to the manufacturer's instructions.

**Supplementary Materials:** Supplementary materials can be found at http://www.mdpi.com/1422-0067/21/5/1847/s1.

**Author Contributions:** Conceptualization, G.A.G., F.B., M.R, and S.R.; data curation, S.R., F.C., and A.A.; investigation, A.A., F.C., L.R., E.M., S.R., and F.B.; methodology, S.R., L.R., F.B., M.R., and G.A.G.; project administration, M.R.; resources, M.R. and G.A.G.; supervision, M.R.; visualization, A.A.; writing—original draft, F.B., S.R., A.A. and M.R.; writing—review & editing, G.A.G. All authors have read and agreed to the published version of the manuscript.

**Funding:** Please add: This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Abbreviations**


#### **References**


© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Relation of the** *pdxB-usg***-***truA***-***dedA* **Operon and the** *truA* **Gene to the Intracellular Survival of** *Salmonella enterica* **Serovar Typhimurium**

**Xiaowen Yang 1,2,3, Jiawei Wang 1, Ziyan Feng 1, Xiangjian Zhang 4, Xiangguo Wang 1,\* and Qingmin Wu 2,\***


Received: 17 December 2018; Accepted: 15 January 2019; Published: 17 January 2019

**Abstract:** *Salmonella* is the genus of Gram-negative, facultative intracellular pathogens that have the ability to infect large numbers of animal or human hosts. The *S. enterica usg* gene is associated with intracellular survival based on ortholog screening and identification. In this study, the λ-Red recombination system was used to construct gene deletion strains and to investigate whether the identified operon was related to intracellular survival. The *pdxB*-*usg*-*truA*-*dedA* operon enhanced the intracellular survival of *S. enterica* by resisting the oxidative environment and the *usg* and *truA* gene expression was induced by H2O2. Moreover, the genes in this operon (except for *dedA*) contributed to virulence in mice. These findings indicate that the *pdxB*-*usg*-*truA*-*dedA* operon functions in resistance to oxidative environments during intracellular survival and is required for in vivo *S. enterica* virulence. This study provides insight toward a better understand of the characteristics of intracellular pathogens and explores the gene modules involved in their intracellular survival.

**Keywords:** *usg*; *truA*; *Salmonella enterica* serovar Typhimurium; oxidative stress; intracellular survival

#### **1. Introduction**

*Salmonella* is a genus of Gram-negative and facultative intracellular pathogens that consists of a large group of genetically similar organisms with the ability to infect many animal and human hosts [1]. In 2005, the *Salmonella* genus was divided into *S. bongori* and *S. enterica* species [2]. *Salmonella* spp. are the second most common causative-agents of gastrointestinal infections in humans, after *Campylobacter* spp. [3]. Some *Salmonella* serovars cause large outbreaks of gastroenteritis associated with contaminated meat and produced or processed food [4].

A previous study found that the *S. enterica usg* gene was associated with intracellular survival based on ortholog screening and identification [5]. Because the complemented STΔusg/p-usg and wild type strains showed significant variation in the infection assay, our lab suggested that the *usg* gene was located in an operon with other genes. Operon prediction for the *S. enterica* strain LT2 whole genome sequence [6] found that the *pdxB*, *usg*, *truA* and *dedA* genes were located in the same operon. The promoter of this operon was located upstream of the *pdxB* gene (http://www.softberry.com/berry. phtml?topic=bprom&group=programs&subgroup=gfindb; accessed on 2 May 2018). Based on the genome annotation, the *pdxB* gene encodes 4-phosphoerythronate dehydrogenase, which is associated with de novo vitamin B6 biosynthesis [7]; the *usg* gene encodes putative aspartate-semialdehyde dehydrogenase, *truA* encodes RNA pseudouridine (38–40) synthase, and the *dedA* gene encodes a hypothetical protein.

In this study, the λ-Red recombination system was used to construct gene deletion strains and investigate whether the identified operon was related to intracellular survival. Furthermore, this study investigated the possible functions of the genes in the operon and assessed their roles in the virulence of *S. enterica*.

#### **2. Results**

#### *2.1. The pdxB-usg-truA-dedA Operon Is Required for Intracellular Survival of S. enterica*

The predicted operon showed that the *pdxB*, *usg*, *truA* and *dedA* genes were transcribed in the same direction, indicating that these genes may have co-transcription. The primers were used to amplify the cDNA of the upstream and downstream genes and the intergenic regions, and the DNA and RNA were used as controls. The results revealed that the genes in the operon were co-transcribed. The above results indicate that the operon contains four genes: *pdxB*, *usg*, *truA* and *dedA* (Figure 1). The expression of each gene, using RT-PCR, showed that the genes in the operon had no effect on other genes in the lysogeny broth (LB) medium.

**Figure 1.** Analysis of co-transcription detection in the operon. Note: "+" means there was a fragment, "−" means there was no fragment. Lane 1 was a DNA marker, lanes 2, 5 and 8 were amplified using genome DNA; lanes 3, 6 and 9 were amplified using total RNA; lanes 4, 7 and 10 were amplified using cDNA.

The *pdxB*, *usg*, *truA* and *dedA* genes were located in the same operon and deletion strains for these genes were constructed using the λ-Red recombination system. The intracellular survival of the gene deletion strains was assessed using J774A.1 macrophage cells (Figure 2). At 12 and 24 h post-infection, the cells infected with the STΔ*pdxB*, STΔ*usg* and STΔ*truA* strains showed lower bacterial loads than the cells infected with the ST and STΔ*dedA* strains (*p* < 0.01) (Figure 2A,B). The STΔ*pdxB*, STΔ*usg* and STΔ*truA* strains had reduced replication abilities inside the J774A.1 macrophages and, therefore, exhibited reduced virulence in vitro. However, the STΔ*dedA* strain did not show reduced replication.

Subsequently, the complemented gene plasmids were electroporated into all of the gene deletion mutants. As shown in Figure 2C,D, the STΔ*pdxB* strains electroporated with the recombinant plasmids p-*pdxB* and p-*usg* showed significant differences in bacterial loads (*p* < 0.001), whereas the STΔ*pdxB* strains electroporated with the recombinant plasmid p-*truA* had almost the same bacterial loads in the

macrophages. Similar results were found for the other complemented strains. These results indicated

In summary, these results showed that all genes in this operon, except for the *dedA* gene, confirmed that *S. enterica* has the ability to survive in macrophages and that the *truA* gene played a key role in this function.

#### *2.2. The pdxB-usg-truA-dedA Operon Contributes to Virulence in Mice*

Ten mice were intraperitoneally inoculated with a dose of 105 Colony-Forming Units (CFU) of the STΔ*pdxB*, STΔ*usg*, STΔ*truA*, STΔ*dedA* and wild type strains. The animals did not survive more than seven days after intraperitoneal inoculation with the wild type strain and not more than nine days post-inoculation with STΔ*dedA*. The mice in the group inoculated with STΔ*pdxB* did not survive more than 15 days. Conversely, the survival rates of the groups inoculated with STΔ*usg* and STΔ*truA* were both 80% at 24 days post-inoculation (Figure 3A).

Five infected mice from each group were randomly selected at 6, 12, and 18 days post-inoculation. Their spleens were removed to assess the bacterial loads (Figure 3B). At six days post-inoculation, the bacterial loads were significantly lower for the gene deletion strains than for the wild type strain and a large reduction (above 2-log) in the spleen bacterial load was observed in the mice inoculated with STΔ*usg* and STΔ*truA* compared to the mice infected with the wild type strains. At 12 days post-inoculation, the spleen bacterial load increased in the mice inoculated with STΔ*pdxB*, which was similar to the load measured in the mice inoculated with the wild type strain at six days. The

bacterial loads of the mice inoculated with STΔ*usg* and STΔ*truA* reduced slowly (Figure 3B). The results showed that *usg* and *truA* contributed to virulence in mice, which was consistent with the cell infection assay results.

**Figure 3.** Survival of the gene deletion mutant and wild type strains in mice. (**A**) Survival curves of the gene deletion and wild type strains. (**B**) Bacterial loads of the spleens at six, 12, and 18 days post-inoculation with the gene deletion and wild type strains. \*\*\* *p* < 0.001.

#### *2.3. The usg and truA Expression Levels Were Higher in the Oxidative Environment*

The expression levels of genes in the operon were evaluated in the gene deletion and wild type strains in the routine growth medium and under oxidative conditions (Figure 4). As shown in Figure 3A, *pdxB*, *usg*, *truA* and *dedA* expression was induced in the wild type strain by H2O2 and was significantly higher in the samples treated with H2O2 than in the untreated controls (Figure 4A).

**Figure 4.** Gene expression levels in the bacterial strains. (**A**) Gene expression levels in the wild type strain under oxidative conditions. (**B**) Gene expression levels in wild type and gene deletion strains under oxidative conditions. \*\*\* *p* < 0.001.

Under oxidative conditions, the *usg*, *truA* and *dedA* genes were barely expressed when *pdxB* was deleted compared with the expression levels in the wild type strain. The *pdxB* gene expression levels were similar in the *usg*, *truA* and *dedA* gene deletion strains. When *dedA* was deleted, the expression levels of the other genes were not significantly different from the expression levels in the wild type strain. When *usg* was deleted, *truA* expression was induced by oxidative conditions and *truA* expression was significantly higher in these strains compared with the wild type strain. *Usg* expression also significantly increased when *truA* was deleted (Figure 4B).

#### *2.4. The pdxB-usg-truA-dedA Operon Contributed to Resistance to Oxidative Conditions*

The growth characteristics of the gene deletion and parent strains were determined in an LB medium. No significant variations were observed between the gene deletion and wild type strains (Figure 5A). These results suggested that the genes in the operon did not affect the in vitro growth of *Salmonella* spp. at normal temperatures. Under oxidative conditions, all of the strains grew slowly for the first 2 h. After 6 h, the growth of the wild type and STΔ*dedA* strains reached the plateau phase (Figure 3B). At this time, the STΔ*pdxB* strain was in the logarithmic phase, and the STΔ*usg* and STΔ*truA* strains had barely replicated. At 10 h post-infection, the STΔ*usg* and STΔ*truA* strains began to replicate, whereas the other strains were in the plateau phase (Figure 5B).

**Figure 5.** Growth characteristics of the gene deletion and complemented strains in the lysogeny broth (LB) medium and under oxidative conditions. (**A**) Growth characteristics of the gene deletion strains in the LB medium. (**B**) Growth characteristics of the gene deletion strains under oxidative conditions. (**C**) Growth characteristics of the strains complemented with the p-*truA* plasmid in the LB medium. (**D**) Growth characteristics of the strains complemented with p-*truA* plasmid under oxidative conditions.

The growth characteristics and oxidative resistance were also assessed for the STΔ*pdxB*/p-*truA*, STΔ*usg*/p-*truA* and STΔ*truA*/p-*truA* strains, which were complemented strains with the same recombinant p-*truA* plasmid (Figure 3C,D). These strains had characteristics similar to the wild type, even under oxidative conditions. All of the results suggested that the operon (except for the *dedA* gene) contributed to a resistance to oxidative conditions promoted the survival of *S. enterica* in macrophages.

#### **3. Discussion**

*S. enterica* is a common facultative intracellular pathogen. The main pathway used by macrophages to eliminate invading pathogens are endocytosis and digestion. *S. enterica* can exploit multiple aspects of host defenses to promote its replication in the host after adaptation to a variety of harsh environments, such as oxidative conditions [8]. A previous study found that the *usg* gene was related to intracellular survival and that the *pdxB*, *usg*, *truA* and *dedA* genes were located in the same operon. Another study found that *pdxB* was related to intracellular survival in *S. enterica*. This study found that a *pdxB* mutant strain was sensitive to oxidative conditions and reduced the bacterial load in macrophages. Another study found that *pdxB* in *E. coli* and *Pseudomonas aeruginosa* contained tightly bound NAD+ and/or NADH [7,9,10] and that the nucleotide-binding domains of *pdxB* were homologous to the corresponding domains of D-3-phosphoglycerate dehydrogenase (PGDHs) from *E. coli* and *Mycobacterium tuberculosis* [7]. Because orthologs usually have conserved biological structures and functions [11,12], the results of this study suggest that *pdxB* in *S. enterica* has the same function. The *usg* and *truA* genes had similar results in the cell assay, similar expression levels under oxidative conditions, and inhibited TNF-α and IL-1β expression in macrophages. One study used random insertions of TnphoA-132 and found that *truA* was one target of glyoxal [13]. Another

study showed that *truA* was associated with resistance to quinoxaline 1, 4-dioxides (QdNOs) in *E. coli*, which have been used in animals as antimicrobial agents and growth promoters for decades [14]. The results showed that *usg* and *truA* were related to intracellular survival. Although the *dedA* gene was in the same operon as the other genes, no effect on intracellular survival was observed for this gene.

There were many genes related to intracellular survival via oxidative resistance. The *sodA* gene deletion strain resulted in a slightly reduced growth rate, low SOD activity, increased susceptibility to reactive oxygen species and chicken serum, and no effect on the motility of the wild type strain [15]. One study found that three catalases (*KatE*, *KatG*, and *KatN*) and two alkyl hydroperoxide reductases (*AhpC* and *TsaA*) were related to oxidative resistance using silico genome analysis and gene deletion methods [16]. Large-scale profiling of *Salmonella* protein expression was performed under H2O2 treatment. The results showed that the abundance of 116 proteins were altered significantly among 1600 quantified proteins and that iron acquisition systems were induced to promote bacterial survival under oxidative stress [17]. Macrophages play important roles in the phagocytosis of pathogens and antigen presentation. Macrophages are immediately activated after phagocytosing pathogens, resulting in a variety of bactericidal mechanisms. These mechanisms include both oxidative and non-oxidative bactericidal mechanisms [8,18]. These poor survival environments lead *Salmonella* spp. to secrete effectors and generate a replicative compartment known as the Salmonella-containing vacuole (SCV). This study showed that genes in this operon (except for the *dedA* gene) conferred *S. enterica* with the ability to survive in macrophages and that the *truA* gene played a key role. The *usg* and *truA* expression levels were increased under oxidative treatment. All of these results suggested that this operon enhanced the intracellular survival of *S. enterica* by increasing resistance to oxidative environments and that *truA* played a key role in this function.

Operons are polycistronic clusters of genes transcribed from a promoter at the 5 end of the cluster [19]. Several operons reportedly related to virulence have been identified in *S. enterica* [20–24]. Typically, genes in the same operon have similar functions and interact with each other. The amino acid sequences *pdxB*, *usg*, *truA* and *dedA* were uploaded to the STRING database [25] to predict protein–protein interactions. *Usg* and *truA* had the highest combined association score (0.935). Co-expression of *usg* and *truA* orthologs was also found in *Acinetobacter* sp. ADP1 and *Pseudomonas aeruginosa*. The combined association score of *dedA* was lower than the association scores for the other genes and no co-expression of *dedA* orthologs had been found to date. The results of our study are similar to our predictions.

#### **4. Materials and Methods**

#### *4.1. Ethics Statement*

All animal research was approved by the Beijing Association for Science and Technology. The approval ID is SYXK (Beijing) 2015–0028 (Validity period: 22 September 2015 to 22 September 2020), and the animal research complied with the Beijing Laboratory Animal Welfare and Ethics guidelines of the Beijing Administration Committee of Laboratory Animals.

#### *4.2. Bacterial Strains and Media*

All of the bacterial strains and plasmids used in this study are listed in Table 1. The *S. typhimurium* and *E. coli* strains, including the parental strain and the derived mutants, were routinely grown or incubated in an LB medium. Antibiotics were added at the following concentrations when required: ampicillin, 100 mg/L and chloramphenicol, 34 mg/L. All bacterial strains were frozen at −80 ◦C with 15–20% (*v*/*v*) glycerol.


#### **Table 1.** Strains and plasmids used in this study.

#### *4.3. Mice*

BALB/c mice (aged 4 to 6 weeks) were purchased from the Weitong Lihua Laboratory Animal Services Center (Beijing, China), and bred in individually ventilated cage rack systems. All experiments involving animals followed the regulations of the Beijing Administration Office for Laboratory Animals.

#### *4.4. Construction of Gene Deletion and Complemented Gene Deletion Mutant Strains*

Deletion mutants and their complemented mutants were constructed for all genes in the operon. Gene deletion mutants were constructed using the λ-Red recombination system. After sequencing confirmation, the recombinant plasmids with the coding regions and their promoters were subsequently electroporated into every gene deletion mutant to complement the gene function. The complemented strains were selected from an LB medium containing ampicillin. The primers are shown in Table S1. The gene deletion and complemented gene deletion mutants were confirmed by PCR amplification and sequencing.

Total RNA was extracted from all strains using TRIzol (Invitrogen, Inc., Carlsbad, CA, USA) according to the manufacturer's instructions and treated with DNase (TaKaRa Bio, Inc., Dalian, China) before reverse transcription to remove DNA contamination. Total RNA was dissolved in diethypyrocarbonate (DEPC)-treated water, and the concentration and purity of the total RNA were estimated by reading the absorbance at 260 and 280 nm, respectively. cDNA was synthesized using the Prime ScriptTM RT Reagent Kit (TaKaRa Bio, Inc., Dalian, China), according to the manufacturer's instructions. The reverse transcription product was stored at −20 ◦C. PCR was performed with the primers shown in Table S1 to evaluate gene expression.

To determine if the genes in the operon were co-transcribed, the intergenic regions of genes were amplified by RT-PCR [27]. The RNA of the parental strain was extracted and cDNA was synthesized by reverse transcription using the method descibed before. The gene spacers were amplified using primers (Table S1) while the wild strain genomic DNA and RNA were used as controls. The amplification system and conditions were the same as before.

#### *4.5. Cell Infection Assay*

To investigate the intracellular survival of the strains, infection assays were performed using J774A.1 murine macrophages (Key Laboratory of Animal Epidemiology and Zoonosis of the Ministry of Agriculture, Beijing, China). The cells were cultured in 24-well plates and infected with each strain

at a multiplicity of infection (MOI) of 10 CFU. Then, the infected plates were centrifuged at 1000 rpm for 5 min at room temperature and incubated at 37 ◦C in an atmosphere containing 5% (*v*/*v*) CO2. After 20 min, the cells were washed three times with phosphate buffered solution (PBS) and incubated in a medium containing gentamycin (50 μg/mL) at 37 ◦C under 5% CO2 until the end of the infection period. At 12 and 24 h post infection (p.i.), the cells were washed and lysed, and the numbers of bacteria exhibiting intracellular survival were determined through serial dilution and plating on an LB medium.

The pBR322 plasmid encoding the green fluorescent protein (GFP) was transferred into the parent and gene deletion strains using the electroporation method. Recombinant clones were selected from the LB medium containing ampicillin. Then, infection assays were performed as described above. After 12 h of incubation, the macrophages were washed three times with PBS. Infection of the J774A.1 macrophages by the strains was observed under a fluorescence microscope (Olympus, Tokyo, Japan).

#### *4.6. Growth Characteristics and Oxidative Resistance Assay*

The in vitro growth analysis of the deletion mutants and complemented strains was described previously. An oxidative resistance assay was performed as follows. One colony of each strain was inoculated into 3 mL of LB or LB with ampicillin medium and cultured overnight at 37 ◦C with shaking at 200 rpm. Subsequently the cultures were adjusted to the same concentration (OD600 ≈ 1.0) and a 50 μL sample of each strain was inoculated into 5 mL of an LB or LB with ampicillin medium. Then, 30% H2O2 was add to the liquid medium (final concentration 4.4 mM) [28] to provide an oxidative environment. The cultures were incubated at 37 ◦C with shaking at 200 rpm, and the OD600 value was determined every 2 h using a BioTek microplate reader (Gene Company Limited, Hong Kong, China).

#### *4.7. Gene Expression Levels in an Oxidative Environment*

The gene expression levels under oxidative treatment were assessed by real-time PCR (RT-PCR). One colony of the gene deletion and parent strain was inoculated into 3 mL of LB medium and cultured overnight at 37 ◦C with shaking. The cultures were adjusted to the same concentration (OD600 ≈ 1.0). A 50 μL sample of each strain was inoculated into 5 mL of LB medium and LB medium with H2O2. Total RNA was extracted from all strains, and cDNAs were obtained as described above. The cDNA samples were subjected to quantitative RT-PCR using the SYBR® Premix Ex TaqTM II Kit (TaKaRa Bio, Inc., Dalian, China). Each PCR reaction consisted of 2 μL of cDNA, 0.8 μL of each primer (10 μM), 10 μL of SYBR®Premix Ex TaqTM II, and 20 μL RNase-free water. The cycling conditions were a denaturation step, at 95 ◦C for 10 min, followed by 40 cycles of 95 ◦C for 10 s and 60 ◦C for 20 s. The specificity of the RT-PCR products was confirmed using a melting curve analysis. These reactions were repeated in triplicate for every sample as technical replicates. Gene mRNA quantification was performed using the 2−ΔΔ*C*<sup>t</sup> method to analyze the expression levels. The 16S rRNA expression level in *S. enterica* was used as a reference to normalize all values. The results presented in this study represent the averages from at least three separate experiments.

#### *4.8. Virulence in BALB/c Mice*

There were two experiments. The first experiment concerned the survival of the mice. Ten mice were intraperitoneally inoculated with a dose of 105 CFU of the gene deletion and wild type strains in 100 μL of phosphate-buffered saline (PBS) [29]; the control group included five mice intraperitoneally inoculated with 100 μL of PBS. The survival of the mice was observed over the next 24 days. The second experiment was the virulence of the gene deletion strains. Based on survival time, five mice were intraperitoneally inoculated with the same dose of the gene deletion and wild type strains. Five infected mice from each group were randomly selected at 6, 12, and 18 days post-inoculation. At each time point, the spleens were removed and homogenized individually in an aseptic manner in 1 mL of PBS and then serially diluted to isolate the bacteria. The results are presented as the mean number of CFU per spleen± the standard deviation (SD) in each group.

#### *4.9. Statistical Analysis*

The statistical analyses of the data, including the data from the growth curve analysis, cell infection study, oxidative resistance assay and virulence experiments, were performed using IBM SPSS Statistics version 23 (IBM, Armonk, New York, NY, USA, https://www.ibm.com/analytics/datascience/predictive-analytics/spss-statistical-software). A *p* value < 0.05 obtained through one-way analysis of variance (ANOVA) was considered significant. All graphics were drawn with GraphPad Prism 5 (GraphPad Software, La jolla, CA, USA, https://www.graphpad.com/).

#### **5. Conclusions**

In conclusion, this study used the λ-Red recombination system to construct gene deletion strains and determine whether the identified operon was related to intracellular survival. Except for the *dedA* gene, all of the genes in this operon confirmed the ability of *S. enterica* to survive in macrophages, and the *truA* gene played a key role in resistance to oxidative conditions. Moreover, the genes in this operon (except for *dedA*) contributed to virulence in mice. These findings indicate that the *pdxB*-*usg*-*truA*-*dedA* operon functions in resistance to oxidative environments and contributes to intracellular survival; moreover, the operon is required for the virulence of *S. enterica* in vivo. In this study, clues were examined to gain a better understanding of the characteristics of intracellular pathogens and to explore the gene modules involved in the intracellular survival of intracellular pathogens.

**Supplementary Materials:** Supplementary materials can be found at http://www.mdpi.com/1422-0067/20/2/ 380/s1.

**Author Contributions:** X.Y. conceived of the experiments, interpreted the data, and supervised the research project; X.Y., Z.F., and J.W. performed the experiments and wrote the manuscript; X.Z., X.W., and Q.W. participated in the discussion and revised the manuscript; all authors approved the final draft.

**Funding:** This work was funded by the National Natural Science Foundation of China (No. 31372446 and No. 31802263), the Research of Key Technology for Prevention of Major Zoonosis in Dairy Cattle (No. 2015NZ0104), and the National Special Foundation for Science and Technology Basic Research (No. 2012FY111000).

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Discovery of** *lahS* **as a Global Regulator of Environmental Adaptation and Virulence in** *Aeromonas hydrophila*

#### **Yuhao Dong, Yao Wang, Jin Liu, Shuiyan Ma, Furqan Awan, Chengping Lu and Yongjie Liu \***

Joint International Research Laboratory of Animal Health and Food Safety, College of Veterinary Medicine, Nanjing Agricultural University, Nanjing 210095, China; 2015207019@njau.edu.cn (Y.D.); 2015107061@njau.edu.cn (Y.W.); 2017207019@njau.edu.cn (J.L.); 2017107059@njau.edu.cn (S.M.); 2015207039@njau.edu.cn (F.A.); lucp@njau.edu.cn (C.L.)

**\*** Correspondence: liuyongjie@njau.edu.cn; Tel.: 86-25-84398606; Fax: 86-25-84398606

Received: 18 August 2018; Accepted: 4 September 2018; Published: 11 September 2018

**Abstract:** *Aeromonas hydrophila* is an important aquatic microorganism that can cause fish hemorrhagic septicemia. In this study, we identified a novel LysR family transcriptional regulator (LahS) in the *A. hydrophila* Chinese epidemic strain NJ-35 from a library of 947 mutant strains. The deletion of *lahS* caused bacteria to exhibit significantly decreased hemolytic activity, motility, biofilm formation, protease production, and anti-bacterial competition ability when compared to the wild-type strain. In addition, the determination of the fifty percent lethal dose (LD50) in zebrafish demonstrated that the *lahS* deletion mutant (Δ*lahS*) was highly attenuated in virulence, with an approximately 200-fold increase in LD50 observed as compared with that of the wild-type strain. However, the Δ*lahS* strain exhibited significantly increased antioxidant activity (six-fold). Label-free quantitative proteome analysis resulted in the identification of 34 differentially expressed proteins in the Δ*lahS* strain. The differentially expressed proteins were involved in flagellum assembly, metabolism, redox reactions, and cell density induction. The data indicated that LahS might act as a global regulator to directly or indirectly regulate various biological processes in *A. hydrophila* NJ-35, contributing to a greater understanding the pathogenic mechanisms of *A. hydrophila.*

**Keywords:** *Aeromonas hydrophila*; LysR-family; Δ*lahS*; global regulator; virulence

#### **1. Introduction**

*Aeromonas hydrophila* is a gram-negative bacterium that is widely distributed in various aquatic environments, such as rivers, lakes, and swamps. In addition, *A. hydrophila* has a diverse host range, which includes fish, birds, amphibians, reptiles, and mammals [1]. Motile aeromonas septicemia (MAS) caused by this bacterium has caused serious damage to the aquaculture industry. In recent years, this bacterium has also been confirmed as an important pathogen of various human diseases, including diarrhea, sepsis, necrotizing fasciitis, meningitis, and hemolytic uremic syndrome [2,3]. *A. hydrophila* has a variety of factors that are associated with virulence, such as exotoxins, S-layers, extracellular enzymes and secretion systems [4]. Of these factors, hemolytic molecules may contribute to the reddening skin and systemic hemorrhagic septicemia; symptoms that are observed in diseased fish infected with *A. hydrophila*, and have been defined as one of the major virulence markers of this bacterium [5].

The expression of virulence factors in response to changes in environmental conditions is especially important for pathogenic bacteria, and this process is commonly governed by a complex network of regulatory elements [6]. Histone-like nucleoid structuring protein (H-NS) has been reported as a negative regulator of hemolytic activity by acting as a repressor of hemolysin gene expression in *Vibrio anguillarum* [7]. Sigma factor RpoE controls the development of hemolytic activity and virulence in *Vibrio harveyi* [8]. A transcriptional regulator of the MarR family, Eha, is required for the bacterial infection and the transcriptional regulation of important virulence factors in *Edwardsiella tarda*, including those that are associated with hemolytic activity [9]. Recently, the LysR family of transcriptional regulators (LTTRs), which is a well-characterized group of transcriptional regulators, has been shown to be global transcriptional regulators in a variety of bacteria [10,11]. LTTRs can regulate a diverse set of genes, such as those involved in metabolism, pilus synthesis, biofilm formation, antioxidant activity, acid resistance, toxin production, and drug efflux [12–17]. More often than not, LTTRs indirectly affect the expression of virulence factors by regulating virulence-related transcriptional regulators. For example, LrhA in enterohemorrhagic *Escherichia coli* positively regulates the expression of LEE genes by the regulation of their master regulators PchA and PchB [18]. The LysR family transcription factor LeuO in *Vibrio cholerae* regulates the transcription of the CarRS two-component regulatory system to control *almEFG* expression, which contributes to cationic antimicrobial peptide (CAMP) resistance by glycosylation of lipid A [19]. In *Haemophilus parasuis*, approximately 500 differentially expressed genes were identified, resulting from the deletion of *oxyR*, and these genes were involved in various biological processes [20]. Despite the importance of LTTRs, very little information is currently available regarding their regulation of virulence factors in *A. hydrophila*.

To identify the factors responsible for the hemolytic activity of *A. hydrophila*, in this study, we individually screened for reduced hemolytic ability in a Tn5-derived library of mutants that were previously generated by our group [21]. Interestingly, we identified a LysR-type transcriptional regulator (LahS) in *A. hydrophila* NJ-35. Our study showed that, in addition to hemolytic activity, LahS positively regulates a variety of virulence factors, including motility, biofilm formation, and protease activity, and it is involved in the anti-bacterial competition ability and virulence of this bacterium. This is the first study to evaluate the role of LahS in virulence and environmental adaptation in *A. hydrophila*, the results of which are of great importance in gaining a better understanding of the pathogenesis of this bacterium.

#### **2. Results**

#### *2.1. Isolation of Transposon Mutants with Reduced Hemolytic Activity*

Among 947 random EZ-Tn5 transposon mutants, a total of six mutants were identified as having reduced hemolytic activity (Figure 1). The EZ-Tn5 chromosomal insertion sites of the six mutants were analyzed by Tail-PCR, followed by BLASTn searches and sequence analyses. The data revealed that the six hemolysis-associated genes that were identified in this study included those encoding for PTS alpha-glucoside transporter subunit IIBC, a LysR family transcriptional regulator, an AraC family transcriptional regulator, and three hypothetical proteins (Table 1).


**Table 1.** Characteristics of the hemolysis-reduced mutants screened by transposon mutant library.

<sup>a</sup> Insertion sites were identified using Tail-PCR and sequence analysis. <sup>b</sup> Putative functions were obtained from the NCBI BLAST online server [22].

M705 U876\_01475 711/857 Arac family transcriptional regulator

M402 U876\_02250 330/516 hypothetical protein

**Figure 1.** Hemolytic activity of the mutants compared to the wild-type strain. A total of six mutants were identified based on their hemolytic activities. \* *p* < 0.05 or \*\* *p* < 0.01.

#### *2.2. Effect of the LysR-type Transcriptional Regulator on A. hydrophila Hemolytic Activity*

As shown in Figure 1, the M307 mutant exhibited decreased hemolytic activity. A sequence analysis of this strain revealed that the EZ-Tn5-disrupted in this strain encodes a protein belonging to LTTRs. This newly identified gene was designated as *lahS*. The open reading frame (ORF) of *lahS* is 909 bp, which encodes a hypothetical 34 kDa protein consisting of 302 amino acid (aa) residues. Secondary structure analysis showed that LahS protein has a conserved N-terminal DNA-binding helix–turn–helix (HTH) motif, which is located at amino acids 5–64 from N terminus (Figure 2A). In addition, there is a LysR substrate binding region that is located at amino acids 87–293, which has the ability to affect the binding capability of this protein. The two domains are connected via a long flexible linker helix that is involved in the oligomerization of the proteins (Figure 2B). The predicted three-dimensional (3D) structure of LahS showed that there are two α/β regulatory domains (RD1 and RD2) in the LysR substrate binding region, which are connected by two crossover regions (Figure 2B).

To verify whether the hemolytic phenotype of the M307 mutant was due to *lahS* inactivation, we constructed *lahS* mutant and complemented strains via homologous recombination. Similar to the M307 mutant, the *lahS* deletion mutant showed decreased hemolytic activity, and the hemolytic activity in the complemented strain *C*Δ*lahS* was restored to wild-type levels (Figure 3A). Through a BLAST analysis, *lahS* homologs were identified within the genomes of several *Aeromonas* species, including *Aeromonas salmonicida* strain A527 (89%), *Aeromonas veronii* strain TH0426 (83%), *Aeromonas caviae* strain 8LM (83%), and *Aeromonas schubertii* strain WL1483 (72%). Additionally, the transcription levels of both the upstream and the downstream genes, which encode an oxidoreductase and C4-dicarboxylate ABC transporter, respectively (Figure 3B), showed no significant differences between the Δ*lahS* and wild-type strains (Supplementary Figure S1). This finding indicated that the *lahS* mutation had no polar effect on the transcription of adjacent genes. Collectively, our findings indicate that LahS is a newly identified regulator of hemolytic activity in *A. hydrophila* NJ-35, and homologs of this regulator are present in a number of *Aeromonas* species.

**Figure 2.** Structure of LahS protein. (**A**) Secondary structure of LysR family transcriptional regulator (LahS) protein. The N-terminal helix–turn–helix (HTH) domain and the LysR-substrate binding region were predicted using simple modular architecture research tool (SMART) software [23]. Rectangular box, α-helix; arrowheaded box, β-strand. (**B**) Predicted three-dimensional (3D) structure of LahS protein. The predicted maps were constructed with the SWISS-MODEL software [24]. Domains were shown in the corresponding position with the same colour coding, as in Figure 2A.

**Figure 3.** The *lahS* gene was involved in hemolytic activity of *A. hydrophila.* (**A**) Hemolytic activity of the wild-type and *lahS* mutant strains. *A. hydrophila* strains were grown overnight in Luria Bertani broth (LB) medium at 28 ◦C, after which 100 μL of supernatants were added to 1% sheep blood for 1 h at 37 ◦C. Hemolytic activity was expressed as the values measured at OD540. \* *p* < 0.05. (**B**) Schematic diagram of the EZ-Tn5 transposon insertion in the *lahS* gene. The gene cluster shows the location of the *lahS* gene in the NJ-35 genome. The red rectangle shows the transposon insertion site in *lahS*.

#### *2.3. LahS Is Involved in Biofilm Formation of A. hydrophila NJ-35*

CLSM analysis of the *A. hydrophila* NJ-35 wild-type, Δ*lahS,* and *C*Δ*lahS* strains showed that wild-type strain and *C*Δ*lahS* exhibited abundant biofilm formation, while the biofilm formation phenotype of the Δ*lahS* mutant is poor when compared to that of the wild-type strain (Figure 4A). This is consistent with our crystal violet staining results at 24 h. The ability of Δ*lahS* mutant to form biofilms was significantly decreased (by 38.8%) as compared to the wild-type strain (*p <* 0.01), whereas the biofilm formation was restored to the wild-type level in the *C*Δ*lahS* strain (Figure 4B). These data indicate that the transcriptional regulator LahS may directly or indirectly regulate biofilm formation of *A. hydrophila.*

**Figure 4.** Biofilm formation of the wild-type and *lahS* mutant strains. (**A**) CLSM images of biofilms of the wild-type and *lahS* mutant strains. The viable cells exhibit green fluorescence. (**B**) Biofilm formation was measured by crystal violet staining using 96-well plates and was expressed as the values measured at OD595. The data are presented as the means ± SEM from three independent experiments. \*\* *p* < 0.01.

#### *2.4. LahS Influences Motility in A. hydrophila NJ-35*

The swimming motility of *A. hydrophila* was measured by examining the distance cells migrated from the center to the periphery of the plate. As shown in Figure 5, migration diameters of 12.4 ± 0.6 and 8.4 ± 0.5 mm were measured for the wild-type and Δ*lahS* mutant strains, respectively, indicating that the swimming motility of the Δ*lahS* mutant was significantly decreased when compared with that of the wild-type strain. The motility phenotype was partially restored in the *C*Δ*lahS* strain. This finding indicates that LahS contributes to the motility of *A. hydrophila* NJ-35.

**Figure 5.** Motility of the wild-type and *lahS* mutant strains. Swimming ability was observed after culturing strains at 28 ◦C for 48 h on 0.3% Luria Bertani broth (LB) agar plates. The migration diameters were measured to assess the motility. The results are presented as the means ± SEM from three independent replicates. \*\* *p* < 0.01.

#### *2.5. LahS Contributes to the Antibacterial Activity of A. hydrophila NJ-35*

To determine whether the transcriptional regulator LahS influences the antibacterial ability of *A. hydrophila* NJ-35, we carried out growth competition experiments by co-culturing *A. hydrophila* strains with *E. coli* BL21. The *A. hydrophila* strains were ampicillin resistant, and the *E. coli* BL21 strain contained the plasmid pET-28a (+), which confers kanamycin resistance to allow for the selection of viable *E. coli* BL21 cells. As shown in Figure 6, co-culturing of *E. coli* BL21 with Δ*lahS* mutant showed about a six-fold rise in colony-forming unit (CFU) when compared to *E. coli* that was co-cultured with the wild-type strain (*p* < 0.05). The antibacterial capacity of the Δ*lahS* mutant was restored after complementing with *lahS* gene.

**Figure 6.** Competition ability of the wild-type and *lahS* mutant strains. (**A**) The image of competition ability is shown on Luria Bertani broth (LB) agar plate. (**B**) Quantification analysis of competition ability of the wild-type and *lahS* mutant strains. The respective bacterial strains were cultured together at a ratio of 1:1. The competition capability of *A. hydrophila* strains against *E. coli* BL21 was defined as the amount of observed *E. coli* survival after antagonism. Data are presented as the means ± SEM of three independent experiments. \* *p* < 0.05.

#### *2.6. LahS Plays a Role in the Resistance of A. hydrophila NJ-35 to Oxidative Stress*

The antioxidant abilities of *A. hydrophila* strains were determined by treating each strain with H2O2. As shown in Figure 7, the Δ*lahS* mutant was hyposensitive to H2O2 as compared with the wild-type strain (*p* < 0.001). The viable cell number of the Δ*lahS* mutant was six-fold higher than that of the wild-type strain. This anti-oxidation ability was partially restored in the *C*Δ*lahS* strain. This result suggested that disruption of *lahS* enhanced the antioxidant capacity of *A. hydrophila* NJ-35.

**Figure 7.** Oxidative stress resistance of the wild-type and *lahS* mutant strains. The effect of hydrogen peroxide on cell viability was examined to investigate the role of *lahS* in the resistance of *A. hydrophila* to oxidative stress. The H2O2 resistance levels were expressed as the colony-forming unit (CFU) of the viable *A. hydrophila* after treatment with H2O2. Data are presented as the means ± SEM of three independent experiments. \*\*\* *p* < 0.001.

#### *2.7. Deletion of lahS Reduced the Protease Activity of A. hydrophila NJ-35*

The culture supernatants of *A. hydrophila* strains were used to test for the presence of protease activity. The results in Figure 8 show that the Δ*lahS* mutant was significantly reduced for the production of protease (0.295 ± 0.013) as compared with that observed in the wild-type strain (0.409 ± 0.025) (*p* < 0.05). The protease activity was partially restored in the complemented strain *C*Δ*lahS.* The result suggested that the *lahS* gene was involved in the protease activity of *A. hydrophila* NJ-35.

 **Figure 8.** Protease activity of the wild-type and *lahS* mutant strains. The protease activity was detected using azocasein as a protease substrate and was measured at OD440. Data are presented as the means ± SEM of three independent experiments. \* *p* < 0.05.

#### *2.8. LahS Is Essential for the Virulence of A. hydrophila in Zebrafish*

To determine whether the *lahS* gene affected bacterial virulence, the LD50 values of the wild type and *lahS* mutant strains were compared while using zebrafish. The LD50 of *A. hydrophila* NJ-35 was 8.84 × 102 CFU, and all the fish were dead within three days. However, the *lahS* mutant had a LD50 of more than 10<sup>5</sup> CFU (Figure 9). Most of the dying fish showed typical clinical signs of hemorrhagic septicemia. Colonies of *A. hydrophila* were recovered from all dead fish, and no evident external lesions were observed in the surviving fish.

**Figure 9.** LD50 values of the wild-type and *lahS* mutant strains in zebrafish. The zebrafish were intraperitoneally (*i.p.*) injected with 10-fold serially diluted bacterial suspensions. The control group was *i.p.* injected with sterile phosphate buffered saline (PBS) only. The results are presented as the means ± SEM from three independent replicates. \*\* *p* < 0.01.

#### *2.9. Comparative Proteomic Analysis*

When considering that LahS functions as a transcriptional regulator, we investigated the differentially expressed profiles of the wild-type and the *lahS* mutant strains while using a label-free mass spectrometry method. The comparative proteomic analysis showed that a total of 2051 proteins matched to the Universal Protein Resource (UniProt). Thirty-four proteins were differentially expressed (change of > 1.5-fold) between the *lahS* mutant and wild type strain, including 10 upregulated and 24 downregulated proteins. A complete list of the names or locus tags of the 34 proteins is shown in Table 2.

To identify the relationship between the 34 different proteins, a hierarchical clustering method based on Pearson's correlation of variances was applied while using R studio. Figure 10A shows the hierarchical clustering of the 34 identified proteins, where an increasing color intensity indicates increasing protein expression levels. To further understand the functions of these identified proteins, we classified 34 differential proteins by Gene Ontology (GO) categories, including cellular component (CC), molecular function (MF), and biological process (BP) (Figure 10B). According to the analysis of CCs, the majority of the proteins were involved in cell projection and flagellum. According to the analysis of BPs, the primary functions were cellular detoxification and antioxygenation. According to the analysis of MFs, the assayed proteins regulated by LahS could be classified into six categories, as follows: heme binding, tetrapyrole binding, antioxidant activity, catalase activity, peroxidase activity, and oxidoreductase activity. In addition, to determine the accuracy of the mass spectrometry results, 14 of these 34 genes (six of the upregulated genes and eight of the downregulated genes) were randomly selected for further verification by qRT-PCR, the results of which showed that the quantitative PCR results were consistent with the proteomic data (Figure 10C).


**Table 2.** Differentially expressed proteins in *lahS* mutant compared to the wild type strain.

**Figure 10.** Label-free quantitative proteomics analysis of differentially expressed proteins between the wild-type and *lahS* mutant strains. (**A**) Heat map of the 34 identified proteins. Up- and down-regulated proteins are indicated in shades of green (increased) and red (decreased), respectively. (**B**) Gene Ontology (GO) classification of differentially expressed proteins. The differentially expressed proteins are grouped into three hierarchically structured terms: biological process, cellular component, and molecular function. (**C**) Relative mRNA expression levels of 14 genes coding the differentially expressed proteins in *lahS* mutant and wild-type strains. The results were expressed as n-fold increases with respect to the control. Data are presented as the means ± SEM from three independent experiments. \* *p* < 0.05, \*\* *p* < 0.01 or \*\*\* *p* < 0.001.

#### **3. Discussion**

Hemolytic molecules are the major contributors to the hemorrhagic septicemia that is characteristic of *A. hydrophila* infections in fish, although this bacterium uses a variety of virulence factors [5]. However, prior to this study, little was known regarding the regulation of these hemolytic factors in this bacterium. In the present study, we identified a novel hemolysis-associated regulator (LahS) in *A. hydrophila* that belongs to the LTTRs. Interestingly, we observed that the LahS participated in

different biological activities in this pathogen, including motility, biofilm formation, environmental adaptability, and virulence.

To investigate the global impact of the *lahS* deletion on protein expression in *A. hydrophila* NJ-35, we performed a label-free quantification analysis between the Δ*lahS* and wild-type strains. From this analysis, it was observed that LahS regulated the expression of a series of proteins that were involved in a wide range of biological processes, including oxidative stress, transcriptional regulation, DNA replication, and metabolism. LTTRs have been reported to regulate motility in many bacteria, but the molecular mechanisms are different in different strains. For example, LeuO in *V. cholerae* O1 modulates motility by cooperating with the nucleoid-associated protein H-NS to repress *vieSAB* transcription [25]. In *E. coli*, LrhA controls the transcription of flagellar, motility, and chemotaxis genes by regulating the expression of the master regulator FlhDC [26]. In this study, 34 proteins were observed to be differentially expressed in the *lahS* mutant as compared to the wild-type strain. Among the downregulated proteins, four proteins are involved in flagellin formation, including FlgE, FliK, FlgD, and FliD. In addition, the expression of ORF (U876\_07335), which is 94% identical to FlrA from *A. hydrophila* AH-3 by a BLASTp analysis, was downregulated in the *lahS* mutant. FlrABC is known as the polar flagellum master regulator of a four-step hierarchical for the expression of flagellar genes in *A. hydrophila* [27]. Therefore, we speculate that LahS may control motility by positively regulating the expression of FlrABC.

Bacterial biofilm formation has been divided into three stages: the planktonic stage, the monolayer stage and the biofilm stage [28]. The transition between different stages is initiated by various environmental signals through the action of specific transcription factors. In *V. cholerae*, LeuO regulates the transition from the monolayer to the biofilm stage by the modulation of exopolysaccharide VPS gene transcription [29]. In addition to stage-specific regulatory functions, certain members of LysR family have roles in biofilm formation through a combination of direct and indirect regulation. For example, OxyR from *H. parasuis* regulates the DNA-binding transcriptional regulator FabR, which is directly involved in biofilm formation, as well as some membrane-related genes that are indirectly related to biofilm formation, such as an outer membrane assembly protein and an outer-membrane lipoprotein carrier protein [20]. In the present study, no proteins that are directly related to biofilm formation were identified among the differentially expressed proteins between the Δ*lahS* and wild-type strains, but RNA polymerase sigma 70 (σ70), which has been reported to participate in the biofilm formation process [30], was observed to be downregulated. This led to us to speculate that deletion of *lahS* might downregulate the expression of the σ<sup>70</sup> gene, which in turn, leads to a decrease in biofilm formation. In addition, a previous study confirmed that flagellar genes were actively transcribed during the planktonic stage, and motility and chemotaxis were important for the initiation of bacterial biofilm formation [31]. Therefore, there is another possibility that the reduction in motility may affect biofilm accumulation.

It is particularly noteworthy that we observed that the inactivation of *lahS* resulted in a significant downregulation of a LuxR family transcriptional regulator. LuxR regulators are key players in quorum sensing (QS), which coordinates the expression of a variety of genes, including those encoding virulence factors, motility, and biofilm formation [32]. Our previous study showed that the LuxR-type response regulator, AhyR, contributes to exoprotease and hemolysin production and the virulence of *A. hydrophila* [33]. In the present study, LahS was observed to be involved in motility, biofilm formation, interbacterial competition, hemolytic and proteolytic activities and virulence in zebrafish. All of these findings led us to speculate that the diverse role exhibited by LahS in *A. hydrophila* virulence might be achieved by upregulating the expression of a LuxR regulator. This speculation is further strengthened by a recent study that was conducted by Gao et al. [34], which indicated that a LysR-type transcriptional regulator (VqsA) is an important QS regulator in *Vibrio alginolyticus* and it plays essential roles in QS-regulated phenotypes, such as type VI secretion system 2 (T6SS2)-dependent interbacterial competition, biofilm formation, exotoxin production, and in vivo virulence. In this regard, it will be interesting to elucidate whether a correlation between LahS and QS exists in *A. hydrophila*.

Members of the LysR family have been widely reported to participate in regulating the antioxidant capacity and environmental adaptability, of which OxyR is the most thoroughly studied [35–37]. OxyR controls oxidative stress by acting as an activator [38] or a repressor [39] for defensive factors, such as catalase. In this study, the inactivation of *lahS* resulted in a substantial increase in the expression level of the catalases (U876\_01465), which might partly explain the phenomenon that antioxidant capacity of the *lahS*-deleted strain was significantly higher than that of the wild-type strain.

In conclusion, our data suggest that the hemolysis-associated regulator LahS is positively involved in motility, biofilm formation, protease production, anti-bacterial competition ability, and virulence, but it negatively regulates the antioxidant capacity in *A. hydrophila*. The comparative proteomic analysis revealed that LahS directly or indirectly controls the expression of 34 proteins involved in flagellar assembly, cellular metabolism, oxidative stress, and environmental adaptability. It will be interesting to further examine the roles of LahS in virulence expression in *A. hydrophila* and to explore whether interfering with its function is an effective way to defend against this bacterial infection.

#### **4. Materials and Methods**

#### *4.1. Strains, Plasmids and Growth Conditions*

The bacterial strains and plasmids that were used in this study are listed in Table 3. *A. hydrophila* NJ-35, which belongs to the ST251 clonal group, was isolated from diseased cultured crucian carp in the Jiangsu province of China in 2010 [40]. The genome sequence of *A. hydrophila* NJ-35 has been published in GenBank (accession number CP006870). *A. hydrophila* and *E. coli* were cultured in Luria Bertani broth (LB) at 28 and 37 ◦C, respectively. When necessary, chloramphenicol (Cm) (Sigma Louis, MO, USA), kanamycin (Kan) (Sigma), or ampicillin (Amp) (Sigma) were added to the medium.


**Table 3.** Bacterial strains and plasmids used in this study.

<sup>a</sup> Characteristics of strains or plasmids. Kan<sup>r</sup> , kanamycin resistant; Cmr , chloramphenicol resistant.

#### *4.2. Screening Transposon Insertion Mutants for Hemolytic Activity*

A library containing 947 random EZ-Tn5 transposon mutants based on *A. hydrophila* NJ-35 was previously constructed in our laboratory [21]. All the mutants were assayed for hemolytic activity, as described previously [44], with some minor modifications. Briefly, *A. hydrophila* strains were grown overnight in LB medium at 28 ◦C, diluted to an optical density of 0.2 at 600 nm (OD600), and pelleted by centrifugation at 10,000× *g* for 10 min. Next, the supernatants were filter-sterilized while using 0.22-μm (pore-size) membrane filters, and 100 μL of supernatant was double ratio diluted and dispensed into 96-well polystyrene plates. Subsequently, 100 μL of 1% sheep blood was added to each well and mixed completely. The plates were incubated for 1 h at 37 ◦C without agitation, followed by an overnight incubation at 4 ◦C. The plates were centrifuged at 1000× *g* for 10 min, after which 100-μL aliquots of supernatants were transferred into a new 96-well polystyrene plate. The OD540 was monitored while using a spectrophotometer (BIO-RAD, Hercules, CA, USA).

#### *4.3. Identification of Insertion Sites by Tail-PCR*

The insertion sites of the EZ-Tn5 in the chromosome of *A. hydrophila* NJ-35 were identified by Tail-PCR. The primers that were used in this study are listed in Supplementary Table S1. Six specific primer pairs (SP1–SP6) were designed to amplify the DNA sequence from either end of the transposon. Degenerate primers AD1 to AD6 were paired with SP1 to SP6, respectively, to amplify the flanking sequences of the inserted EZ-Tn5. The detailed operation of the Tail-PCR was performed, as previously described [21]. After Tail-PCR, the specific products were gel purified and sequenced. The BLASTn program was used to compare the DNA sequence with the genome of the reference strain NJ-35 to confirm the sequences flanking the EZ-Tn5.

#### *4.4. Construction of a lahS Mutant and Complemented Strains*

The *lahS* deletion mutant strain was constructed, as previously described [42]. Briefly, two flanking regions of the target gene were amplified and ligated by PCR using two sets of primer pairs, *lahS-1/lahS-2* and *lahS-3/lahS-4* (Table 4). Next, the fusion fragment was generated with the primer pair *lahS-1/lahS-4* and then was inserted into pYAK1 and transformed into *E. coli* SM10 [41]. The recombination vector from the donor *E. coli* SM10 strain was conjugated into the recipient *A. hydrophila* NJ-35 strain. LB agar plates containing 100 μg/mL Amp and 34 μg/mL Cm were used to select for isolates in which the plasmid had integrated into the chromosome via recombination. The positive colonies were inoculated into LB broth supplemented with 20% sucrose to induce a second crossover event, and aliquots of this culture were subsequently spread onto LB agar plates containing 20% sucrose to generate the deletion of mutants. The *lahS* mutant was confirmed by sequencing the deleted region and flanking DNA in the mutated strain.



The complemented Δ*lahS* strain was constructed using the shuttle plasmid pMMB207 [43]. The complete *lahS* gene and its putative promoter and terminator regions were amplified using the primer pair *lahS*-C-F/R and then ligated into the pMMB207 vector. The recombinant plasmid pMMB207-*lahS* was transformed into the Δ*lahS* mutant strain by bacterial conjugation to generate the complemented strain *C*Δ*lahS*, which was verified by PCR.

#### *4.5. Biofilm Formation Assay*

To compare the biofilm formation ability of strains NJ-35 and Δ*lahS*, confocal laser scanning microscopy (CLSM) was performed to analyze the three-dimensional architecture of biofilms as previously described [45]. Briefly, the stationary phase bacterial cultures were adjusted to an OD600 of 1.0 and then diluted 1:1000 in LB medium. Two milliliters of these dilutions were subsequently added into each well of 6-well plate containing pre-sterilized microscopic glass slides for biofilm growth. After incubating at 28 ◦C for 24 h to allow for biofilm development, the glass slides were carefully washed with phosphate buffered saline (PBS) to remove the planktonic cells. Next, 20 μL of fluorescein diacetate (FDA, Sigma, Louis, MO, USA) was added to each glass slide, which were then protected from light for 20 min. The slides were then observed by CLSM (Carl Zeiss LSM700, Oberkochen, Germany) to observe the biofilms using an argon laser. Seven random spots were measured for each of the three replicate glass slides. The image stacks were recorded under identical conditions (i.e., similar

area and vertical resolution). Sterilized glass slides incubated with fresh LB medium were used as a control group.

Biofilm formation was also quantitatively measured by crystal violet staining, as previously described [46]. *A. hydrophila* strains were grown to an OD600 of 0.6–0.8 in LB broth at 28 ◦C and then diluted to an OD600 of 0.1. Next, 200-μL aliquots of suspensions (1:100 dilution in fresh LB) were dispensed into 96-well polystyrene plates, which were incubated at 28 ◦C for 24 h without shaking. The medium was decanted, and the wells were washed three times with sterile PBS. Next, the bacterial cells were fixed with 200 μL of methanol for 15 min, after which they were air dried at room temperature. After drying, 200 μL of a crystal violet solution (1% *w/v*) was added to each well and the cells were stained for 10 min. The wells were then rinsed with ddH2O to remove the unbound crystal violet. The bound crystal violet was solubilized using 95% ethanol for 10 min, and the optical density was measured at OD595. The assay was performed in three independent experiments.

#### *4.6. Motility Assay*

The swimming motility assay was performed using 0.3% agar plates, as previously described [47]. *A. hydrophila* strains were grown to the log phase in LB broth at 28 ◦C, and 1 <sup>μ</sup>L of each culture (5 × 108 CFU/mL bacteria) was stabbed into motility assay agar plates. The plates were incubated at 28 ◦C for 48 h, after which motility was assessed by measuring the migration diameter of the bacterial cells. The assay was performed in three independent experiments.

#### *4.7. Anti-Bacterial Competition Assay*

The *E. coli* inhibition assay was performed as previously described with some modifications [48]. *A. hydrophila* and *E. coli* BL21 strains grown to an OD600 of 1.0 and then were concentrated 10 times. Cells were mixed at a ratio of 1:1, spotted onto 0.22-μm sterile filters on LB plates, and incubated at 28 ◦C for 3 h. *E. coli* BL21 cells that were mixed with equal volume of LB media was used as a control. Then, the spots were serially diluted in 1 mL LB medium, and the survival *E. coli* was quantified by serial dilution in LB and visualized on LB plates containing kanamycin. The assay was performed in three replicates.

#### *4.8. Oxidative Stress Resistance Test*

For oxidative stress tests, 4 mM H2O2 was freshly prepared before each experiment and was filter-sterilized using 0.22-μm (pore-size) membrane filters. Log-phase cultures were normalized to an OD600 of 0.5. Next, 400-μL aliquots were added to 100 μL H2O2 and the mixtures were incubated at 28 ◦C for 1 h. The oxidation was terminated with 2000 U of catalase for 10 min. The number of viable *A. hydrophila* was counted via serial dilution of the suspensions and plating on LB agar plates. The assay was performed in three independent experiments.

#### *4.9. Protease Activity*

*A. hydrophila* strains were grown overnight in LB medium at 28 ◦C and the OD600 values were normalized to 2.0 with fresh LB medium. Next, the cells were pelleted by centrifugation at 10,000× *g* for 10 min, and the supernatants were then filter-sterilized using 0.22-μm (pore-size) membrane filters. Subsequently, 250-μL aliquots of supernatants were added to 250 μL of 0.5% (*w/v*) azocasein in 50 mM Tris-HCl (pH 8.0) and then incubated at 37 ◦C for 2 h. After incubating, 500 μL of 10% (*w/v*) trichloroacetic acid (TCA) was added and the mixture was incubated on ice for 30 min. The cells were then pelleted by centrifugation at 10,000× *g* for 10 min, and 500 μL of the supernatants were added to 500 μL NaOH. The azodye was measured at OD440.

#### *4.10. Determination of LD50 in Zebrafish*

The animal experiment was carried out in accordance with the animal welfare standards and guidelines of the Animal Welfare Council of China and was approved by the Ethical Committee for Animal Experiments of Nanjing Agricultural University, China [permit number: SYXK (SU).2017-0007]. Zebrafish were supplied by the Pearl River Fishery Research Institute, Chinese Academic of Fishery Science. The animal-challenge experiment was performed according to a previous study [49]. Log-phase bacteria were washed three times with sterile PBS and the suspensions were serially tenfold diluted from 5 × 102 to 5 × 107 CFU/mL. Ten zebrafish per group were intraperitoneally (*i.p.*) injected with 20 μL of the suspensions in PBS. An additional 10 zebrafish were injected with 20 μL of sterile PBS as the negative control. Mortality was recorded for seven days, and the LD50 values were calculated by the method of Reed and Muench [50].

#### *4.11. Comparative Proteomic Analysis*

*A. hydrophila* strains were grown overnight in LB broth at 28 ◦C, after which the bacterial cells were washed three times with PBS and then resuspended in lysis buffer. The bacterial suspensions were then ruptured by sonication at 4 ◦C and centrifuged (12,000× *g*, 30 min, 4 ◦C) to collect the supernatants. Next, 10% (*w/v*) TCA was added to the supernatants, which were incubated in ice-cold water for 30 min. The supernatants were centrifuged at 12,000× *g* for 10 min and then washed three times with ice-cold acetone. The proteins were harvested by centrifugation at 12,000× *g* for 10 min and analyzed using a label free mass spectrometry method. For quantitative proteomics analysis, 200 μg samples were digested with trypsin (1:50 *w/w*; Sigma Louis, MO, USA) at 37 ◦C for 24 h. Peptide mixtures were fractionated by nano-liquid chromatography and analyzed by mass spectrometry (MS). MS data were searched against uniprot\_Aeromonas\_hydrophila\_27500\_20170605.fasta (27,500 total entries, downloaded 5 June 2017). The identified proteins were analyzed by GO categories, KEGG enrichment, and clustering analyses.

#### *4.12. Quantitative Reverse Transcription-PCR (qRT-PCR)*

To validate the proteomic results, we used qRT-PCR to measure the transcription levels of randomly selected genes. The primer pairs used in this assay are shown in Supplementary Table S2. Total RNA was isolated using an E.Z.N.A. bacterial RNA isolation kit (Omega, Beijing, China). cDNA synthesis was performed while using a PrimeScript RT reagent kit (TaKaRa, Dalian, China). qRT-PCR was performed to quantify each target transcript using a QuantiTect SYBR green PCR kit (Qiagen, Valencia, CA, USA). The constitutively expressed *recA* gene was chosen as a reference gene for qRT-PCR, and the 2−ΔΔ*C*<sup>t</sup> method was used as previously described [51].

#### *4.13. Statistical Analysis*

Statistical analyses were performed while using GraphPad Prism 6. Student's *t* test was used for examining the differences between the wild-type and mutant strains. *p* values of <0.05 were considered significant.

**Supplementary Materials:** Supplementary materials can be found at http://www.mdpi.com/1422-0067/19/9/ 2709/s1. Figure S1: Fold difference in mRNA level; Table S1: Primers used for Tail-PCR; Table S2: Primers used for qRT-PCR.

**Author Contributions:** Y.L. and Y.D. conceived the study and drafted the paper. Y.D. and Y.W. performed most of the experiments described in the manuscript. J.L., S.M. and F.A. helped with the experiments. C.L. provided valuable suggestions of the manuscript. All authors read and approved the final manuscript.

**Funding:** This work was supported by the National Natural Science Foundation of China (31372454), the Three New Aquatic Projects in Jiangsu Province (D2017-3-1), the Independent Innovation Fund of Agricultural Science and Technology in Jiangsu Province (CX (17) 2027) and Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD).

**Acknowledgments:** We would like to thank the Shanghai Applied Protein Technology Co., Ltd., to provide technical support in our proteomics analysis.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
