Enhanced Virulence of Salmonella enterica serovar Enteritidis ATCC13076 under Acid Stress by Global Transcriptomics

Salmonella enterica serovar Enteritidis is a primary pathogen causing foodborne diseases and intestinal inammatory responses. Acid tolerance response (ATR), as a strategy of adaption and resistance to acid stress, may contribute to enhanced virulence. In this study, there was a moderately acid adaption (pH 5.0) for S. Enteritidis cells prior to treatment with acid stress (pH 3.0). To gure out whether S. Enteritidis up-regulated the virulence or not, a global transcriptomic analysis was carried out by high-throughout RNA-sEq. The results showed 74 differentially expressed genes (DEGs) involved in virulence were identied after acid stress, among which, 62 DEGs were up-regulated and 12 DEGs were down-regulated. Afterwards, those virulence-linked DEGs were discussed and classied into four aspects based on the steps of infection, including agellar functions, mbrial adhesins, T3SS-mediated invasion and other virulent determinants. In conclusion, S. Enteritidis seemed to exhibit a trend of virulent genes towards high-expression under acid stress, revealing risks of Salmonella in acid-containing food. To our knowledge, there were few studies on comprehensively analyzing virulent genes expression changes of Salmonella, but it’s novel to put forward pathogenicity as the highest priority under acid environment.


Introduction
Throughout the world, Salmonella enterica serovar Enteritidis and Typhimurium respectively ranking as the rst and second places of the common serovars, are linked with outbreaks of human salmonellosis (Hendriksen et al. 2011). In spite of most studies focusing on S. Typhimurium, there are few on S. Enteritidis, which is responsible for some foodborne diseases and intestinal in ammatory responses. A survey about the worldwide distribution in human cases (Ekdahl et al. 2005), suggested S. Enteritidis is the most prevalent serovar in all but Africa, India and America. Besides, there have happened a series of outbreaks triggered by S. Enteritidis (Mandilara et al. 2017;Pavlova et al. 2018;Wright et al. 2015). Since researchers realized this serotype posed great risks to food industry and public health, a great many relevant reports have been published to gure out the mechanisms of the virulence.
The agellum is a required multifunctional complex to enable the cells motility of swarming and swimming, and its functions involve attachment and adhesion, colonization, bio lm formation, as well as chemotaxis (Oguri et al. 2019). After approaching to the targets, Salmonella deploy a variety of adhesive structures in its surface at the initial step. In a review, Wagner and Hensel (2011) have systematically elucidated those structures including mbriae, agellum and other non-mbrial adhesins, participating in speci city of adhesion to the host cells and cooperating with other virulence determinants. Gene clusters of mbriae, including m, csg, pef and lpf (Ledeboer et al. 2006), contributed directly to its functions to a certain extent. Besides, lipopolysaccharide (LPS) and translocon proteins in T3SS are a second form of adhesins (Wagner and Hensel 2011).
In terms of the pathogenicity of Salmonella, the genes located on pathogenicity islands were attached great importance to. With the progress of research on pathogens, type III secretion system (T3SS) was believed to exist in some bacteria as a sophisticated machine that could attack and invade eukaryotic cells to manipulate the host cells, including Salmonella. The invasion is achieved with a needle-shaped structure, a vital component of T3SS, to deliver virulent proteins termed "effectors" (Cascales 2017). It was con rmed that Salmonella pathogenicity island 1 and 2 (SPI-1,2) encoded T3SS-1,2 that synthesized two types of effectors respectively. The effectors translocated by T3SS-1 are required for invasion into host cells and Salmonellacontaining vacuole (SCV) biogenesis, such as SopA, SopB, SopD, SopE (Fabrega and Vila 2013). While T3SS-2 effectors are linked to SCV maturation, replication and achievement of infection inside host cells, including SseL, SifA, SopD2, SopE2 and PipB2 (McGhie et al. 2009). SopA, a HECT-type E3 ligase, plays a key role in the stimulation of in ammation when an animal is infected with Salmonella Typhimurium. (Kamanova et al. 2016). SopD was found to directly bind to the InvC ATPase (Boonyom et al. 2010). SopE is a guanidyl exchange factor (GEF) for Rac1 and Cdc42, secreted by the T3SS-1 upon host cell contact and promotes entry through triggering of actin-dependent ru es (Vonaesch et al. 2014). SseL is retained predominantly in the cytoplasm of infected cells following translocation by the T3SS (Coombes et al. 2007). SifA is related to localize to Salmonella-induced laments (SIFs) and enables continuous fusion of host vesicles to SCV membranes (Knuff and Finlay 2017).
A series of stresses have been veri ed to render Salmonella to adjust its pathogenicity as survival strategies. For instance, Shah et al. (2012) performed an experiment in which acid and oxidative stresses impaired intestinal colonization and systemic dissemination in orally inoculated chickens. Meanwhile, the study showed the stress-resistance signi cantly changed the expression of some virulence genes. Especially, Salmonella have already acquired the ability to survive acid environment such as the gastrointestinal tract of their hosts. Acid tolerance response (ATR) depends on various strains (Lianou et al. 2017), types acid and its acidity (Alvarez-Ordonez et al. 2009), duration and so on. The virulent aspect of ATR remains ambiguous over S. Enteritidis, posing a great threat to the safety of acidic food and beverages. The molecular understanding of S. Enteritidis pathogenicity under acid stress is necessary for further insight and food industry.

Data processing and DEGs analysis
After ltration and processing, the results showed the basic data of RNA sequencing, which suggested they are quali ed and credible for following analysis and discussion. According to the threshold of log2(FC) foldchange ≥ 1, FDR < 0.05, 554 DEGs were identi ed under acid stress. And then, 74 DEGs involved in virulence were classi ed and analyzed referring to the description of genes, Fabrega and Vila (2013), Wagner and Hensel (2011), as well as using the website (http://www.mgc.ac.cn/cgi-bin/VFs/genus.cgi?Genus=Salmonella).
Among those selected DEGs, 62 DEGs were up-regulated and 12 DEGs were down-regulated, shown in the volcano plots diagram (Fig. 1).

GO analysis classi cation of DEGs
GO analysis (Quinteiro-Filho et al. 2012) were employed to classify the functions of identi ed DEGs. GO analysis results showed the percentage (number) of virulence-associated DEGs in various parts of cellular component, molecular function and biological process according to GO database (Gelli et al. 2017). Within cellular component, a big part of DEGs which were related to cell projection and bacterial-type agellum accounted for the rst two largest clusters. And in the category of biological process, a high percentage of DEGs existed in locomotion and cell process, as well as biological adhesion.

KEGG analysis classi cation of DEGs
In order to understand the virulent responses of acid mediated pathways in ATR. KEGG pathway enrichment analysis was also conducted. According to results of analysis, pathways including agellar assembly, bacterial secretion system, bacterial chemotaxis, lipopolysaccharide biosynthesis and bio lm formation were in signi cant up-regulation under such an acid stress.

Validation of DEGs using qRT-PCR
In order to testify if the results of RNA-seq was in accord with the qPCR, some DEGs were analyzed through qPCR. Nine genes that include 7 up-regulated and 2 down-regulated genes were selected for the qRT-PCR analysis and used to validate the RNA-seq data from the preliminary experiment. It was shown in Fig. 2 that qRT-PCR data correlated well with the RNA-seq data (R 2 = 0.9117). Totally, the qRT-PCR data were paralleled to those of RNA-seq aiming at these genes, in spite that the speci c values of fold-change were different.

Discussion
As a consequence of virulence potentials and the high incidence of Salmonella-speci c infections, we put forward virulence of S. Enteritidis to be the highest priority in this study. Generally, the infection caused by Salmonella should be attributed to several virulent structures, such as agellum, mbriae, T3SS-mediated invasion, lipopolysaccharide (LPS) and regulators. On the other hand, different structures are responsible for different processes and functions as follows.
1 Up-regulated agellar genes In Salmonella, agellum plays a crucial role in infection, especially in the initial phase, such as colonization and attachment to the intestinal epithelium (Barbosa et al. 2017;Salehi et al. 2017). The two major functions of agellum include motility and chemotaxis. The synthesis and function of the agellar and chemotaxis system requires the expression of more than 50 genes or at least 17 operons ( h, g, i, j, mot, che, tar, tsr, and aer) that constitute the large and coordinately regulated agellar regulon (Chilcott and Hughes 2000). It is demonstrated in Fig. 3 that agellum consists of three parts, lament, hook, and basal body. Main component proteins of lament and hook are FliCD and FlgE respectively, and the latter one is regulated by FliK and FlgD. While h and j didn't show a signi cant up-regulation. In our study, iC was up-regulated by 2.13 and iD by 1.48 log 2 (FC), as well as gE by 1.05. Crawford et al. (2010) reported that FliC can mediate binding to cholesterol and initiating bio lm formation, suggesting agella enhanced bacterial adhesion. The center part of basal body is type III injectisome, a sophisticated nanomachine similar to T3SS (Erhardt et al. 2010). FliI (+ 1.32 log 2 (FC)) was identi ed as an ATPase along with FliH (+ 1.24 log 2 (FC)), both involved in the apparatus of type III injectisome (Minamino 2014). Apart from the structure protein described above, proximal rod (FlgB, FlgC and FlgF), MS ring (FliF) along with C ring (FliG) distinctly had an up-regulation by 1.09 ~ 1.44 log 2 (FC). Chevance and Hughes (2008) have elaborated on the mechanism of agellar assembly: a three-tier hierarchy regulation. According to our data, in spite of the rst class genes ( hDC operon) not showing different expressions, the high-expressed FliA by 1.19 and FliZ by 1.13 log 2 (FC), both of which belonged to class 2. And FliA was identi ed as the sigma factor (σ 28 ) of agellar genes, while FliZ as the expression activator to regulate its own class genes (Tanabe et al. 2011). Actually, class 2 proteins couldn't express in absence of FlhD 4 C 2 revealing a fact that hDC operon had nished its work in earlier stage as shown in Fig. 3 (Barker et al. 2014). In addition, gANM and iT were all signi cantly up-regulated coincidently in varying levels ranging from 1.02 to 1.50 log 2 (FC). Prior to the completion of class 2 assembly (Fig. 3), the anti-σ 28 factor FlgM regulated the expression of late substrate genes under regulation of class 3 promotor and hook-associated genes ( gKL and iD) (Erhardt et al. 2010). Moreover, genes of chemotaxis (cheYZ, tsr and aer) were also respectively up-regulated by 1.24, 1.19 1.37 and 1.12 log 2 (FC) under acid stress. Whereas the motility genes (motAB) did not appear a signi cant up-regulation as expectation. Obviously, what are listed above offers the amplest evidence that agella functioned better in such extreme situation.
2 Fimbriae-linked genes were slightly up-regulated Previous to Salmonella's invasion into host cells, mbriae is a required and decisive structure to adhere to cell membrane. As shown in Table 2, S. Enteritidis up-regulated majority of genes involved in mbrial assembly, transporter, structure, regulator and chaperone/usher protein. Fimbrial adhesins are produced in following three pathways referring to Wagner and Hensel (2011): chaperone-usher pathway, the extracellular nucleation pathway and type IV pili. mD and mI, members of type I mbriae (SEF21) family, generally appeared a downregulation by 1.06 and 1.54 log 2 (FC), however, mF was up-regulated by 1.36 log 2 (FC). Another type of mbriae -SEF14-existed speci cally in S. Enteritidis. It turned out to be three genes (sefABC) in SEF14 operon. SEF14 mbriae contributes to SE adherence with mouse epithelial cells but not to human HEp-2, Caco-2, INT-407, or HeLa cells (Quan et al. 2019). SefA, a novel mbrin, owns the ability to enhance S. Enteritidis adhesion to epithelial cells and survival in macrophages and results in S. Enteritidis virulence in mice (Zhu et al. 2013). SefB and SefC shared homology with E. coli mbrial chaperone and outer membrane proteins (Clouthier et al. 1993). From our results, sefABC was up-regulated by 1.54, 1.45 and 1.90 log 2 (FC), respectively. In addition, type IV pili genes seemed to be regulated during acid stress, hofC showing a down-regulation by 1.46 and pilCN by 1.67 and 1.00 except for SEN1977. As far as long polar mbriae (Lpf) was concerned, it was rstly proposed to role in the adhesion of S. Typhimurium to cells of the Peyer's patches (Bäumler et al. 1996). In our study, lpfA acting as a precursor of long polar mbrial protein A up-regulated by 2.09 log 2 (FC) than unstressed cells. Such abundance of expression suggested the cells were likely to up-regulate the subsequent genes (lpfCD) in the next steps. On the other hand, Salmonella can speci cally assemble curli ber or thin aggregation mbriae (Ta ) via the nucleation-participation pathway like E. coli. Curli mbriae relates to adhesion, induction of proin ammatory response and bio m formation (Fabrega and Vila 2013). And CsgBA (subunits) and CsgDEFG are required for curli biogenesis, these two divergently transcribed operons are the homologues in E. coli (Römling et al. 1998). In a recent study (Newman et al. 2018), it was unraveled that regulator CsgD mediated stationary phase counter-silencing of csgBA in S. Typhimurium. Under acid stress, S. Enteritidis also up-regulated csgD by 1.84 and csgC by 1.71 log 2 (FC) in site of insigni cance (FDR ≥ 0.05), not shown in Table 2, as well as other protein with slightly up-regulation. Moreover, up-regulations of yehD encoding a mbrial subunit protein and virF encoding a mbrial operon positive regulatory protein re ected an active state of mbriae. In total, roles of mbriae in adhesion turned out to be a reasonable enhancement even though S. Enteritidis cells partly down-regulated type I mbriae and type IV pilin.

S. Enteritidis up-regulated genes for invasion and infection
T3SS-1 was con rmed to enable Salmonella to invade into host epithelial cells and establishing infection in the gastrointestinal tract. T3SS-1 would not function completely until DsbA arrived its destination and activated the apparatus, which is required for translocation and secretion of effectors (Fabrega and Vila 2013). After treated with acid stress, S. Enteritidis slightly up-regulated dsbA by 0.79 log 2 (FC). Initiation of actin cytoskeletal remodeling and induction of a proin ammatory response are attributed mainly to the effectors SopE, SopE2 that function as a guanidine exchange factor (GEF) for host cellular Rho GTPases (Hardt et al. 1998). Despite of the differences of their structures, these two proteins function in invasion into host cell identically. Our data shows S. Enteritidis up-regulated DsbA by 0.79, SopE and SopE2 by 1.48 log 2 (FC). Two additional key SPI-1 effectors, SopA and SopD also contributing to enteropathogenesis. In this study, SopA was down-regulated by 1.13 log 2 (FC), while SopD was up-regulated by 2.04 log 2 (FC). Kamanova et al. (2016) concluded that sopA encodes a HECT-type E3 ligase, contributing to the stimulation of in ammation when an animal is infected with S. Typhimurium. It was reported SopD directly bound to the InvC ATPase and was found in the host-cell cytosol not only during the early stages of infection but also later in the SCV (Boonyom et al. 2010;Brumell et al. 2003). Moreover, SopD and SopB play role cooperatively in enhancing membrane ssion and to promote micropinocytosis during S. Typhimurium invasion (Bakowski et al. 2007). In addition, T3SS encoded by SPI-2 generally functions and translocates effectors that participate in SCV maturation and establishment of infection within host cells. As shown in Table 2, there is an up-regulation of 1.19 log 2 (FC) for SseL. Such a T3SS-2 effector protein was reported to retain predominantly in the cytoplasm of infected cells following translocation by the T3SS (Coombes et al. 2007). PipB2 and SifB as another two translocated effectors of T3SS-2 were respectively up-regulated by 1.57 and 1.22 log 2 (FC). And it has been proven that PipB2 and SifB can localize to SIF and SCV membranes and result in subsequent infection (Freeman et al. 2003;Knuff and Finlay 2017). Apart from the effectors, SsaQ (type III secretion system protein), SseC (translocation machinery component), RcsC (secretion system regulator: Sensor component) and SEN3371 (pathogenicity island protein) were also signi cantly up-regulated by 1.01 ~ 1.43 log 2 (FC).
Surprisingly, inv, hilA, spt, spa, sip, iag, iac and sic that had a close connection with colonization and invasion of Salmonella referring to earlier studies (Bajaj et al. 1996;Boyd et al. 1997;Phoebe Lostroh and Lee 2001;Rahn et al. 1992), turned out to be no expression in either unstressed or stressed cells in our results. The unexpected phenomenon could attribute to the hierarchy of regulation of invasion. Gene regulation plays an extremely important role in the e cacy of the pathogenesis of Salmonella in order to coordinate all the virulence traits (Fabrega and Vila 2013). As shown in Fig. 4, this regulation is under a temporal hierarchy in which virulence elements need to be progressively expressed. In our study, S. Enteritidis mildly up-regulated HilD by 0.99 log 2 (FC), which was regarded as the most important activator of HilA in vitro and at the upper class of the hierarchy. Gene hilD was activated by FliZ and Fis, but repressed by CsrA, all the three type of regulator was up-regulated by 1.13 ~ 1.29. And the presence of Hha (Nucleoid proteins) with an up-regulation of 1.39 log 2 (FC) absolutely repressed hilA, contributing to repression of inv/spa, prg/org, sic/sip and some effectors. Furthermore, physiology of S. Enteritidis should be considered. If not in a period of invasion, it is not an economic alternative to maintain a high level of expression for these proteins. As a result, regulators such as DsbA and HilD made a greater contribution to the subsequent invasion, let alone the various effectors encoded by SPI-1 and SPI-2.
4 Other pathogenicity determinants were up-regulated 4.1 LPS LPS, located in the outer layer of membrane, not only functions as a defender of cells, but is also identi ed as a key determinant of virulence involved in colonization (Nevola et al. 1985), invasion and self-protection from bacteria being killed by macrophages (Kong et al. 2011). It consists of three components, including lipid A, core oligosaccharide (C-OS), and O-antigen polysaccharide. The assembly system of C-OS A rfaL mutant of S. Typhimurium, which lacks the entire O-antigen (O-Ag), was virtually avirulent in a Galleria mellonella infection model (Bender et al. 2013). According to our data (Table 2), not only was rfaL up-regulated by 1.37 log 2 (FC).
Besides, Kong et al. (2011) also elucidated rfaGIHJL, rfbP and rfc encoded a series of expressions that were concerned with LPS's virulent aspect. The results showed that genes in Rfa, Rfb and Rfc families were upregulated by 1.17 ~ 2.23 log 2 (FC) except for RfaF's down-regulation by 1.20 log 2 (FC).

Regulators
Importantly, CsrA, a global post-transcriptional regulator controlling various properties related to metabolism and virulence, had a signi cant high-expression by 1.23 log 2 (FC). For instance, CsrA was proved to positively regulate FlhC and FlhD (Lawhon et al. 2003), both the subunits of the primary regulator in agella expression, which didn't show an obvious up-regulation according to our data. Additionally, Sterzenbach et al. (2013) gured out that CsrA is a positive regulator of PefA expression in S. Typhimurium by binding a GGA motif in the 50-UTR of the pefACDEF transcript. Meanwhile, capability of regulating HilD leaves CsrA involved in invasion, especially in the expression of the T3SS part (Fortune et al. 2006).
In this study, marA and soxS was up-regulated by 2.37 and 1.27 log 2 (FC), respectively. MarA was proved to effect the invasion into host cells for S. Choleraesuis, when exposed to an environment of acid and bile salts (Lee et al. 2015). In E. coli, MarA and SoxS were both described as members of AraC family to regulate a great many proteins, when adapting to unfavorable conditions, including acid stress (Duval and Lister 2013).
Another study also implicated that SoxS controlling expression of distinct genes acted as a key factor in persistence of E. coli in murine pyelonephritis (Casaz et al. 2006).

Conclusion
Altogether, by using global transcriptomic, we surprisingly found that S. Enteritidis enhanced its virulence after exposed to such an acid environment. We analyzed and divided results into four essential virulence-associated aspects, including agellar functions, mbriae-mediated adhesion, invasion of T3SS, virulent roles of LPS and some regulators. Each part showing an up-regulation contributed jointly and strongly to our conclusion that Salmonella Enteritidis had evolved the ability to cope with low pH environment such as the gastrointestinal tract of their hosts during infection. This study paid attention to a threat of Salmonella existing in acid food, as an inspiration for the food industry. Furthermore, it's valuable to gure out the mechanisms of virulent aspects in Salmonella.

Material And Methods 1 Bacterial strain and gastric uid simulation
The strain used in this study was Salmonella enterica subsp. enterica serovar Enteritidis ATCC13076, stocked at -80°C in Luria-Bertani (LB) with 30% glycerol. Initially, it was activated by streaking onto LB plate with incubation at 37°C for 24 h before each experiment. Then the exponential phase cells (OD600 ≈ 1.0) were centrifuged at 5000×g for 3 min and washed twice with sterilized phosphate buffer. Subsequently, the washed culture was dissolved in LB broth at a concentration of approximately 10 6 CFU/mL (6 Log 10 culture).
Under acid stress, S. Enteritidis was pretreated with acid adaption as follows. Brie y, during acid adaption process, 6 Log10 culture was collected in the LB broth (pH 5.5) for 1 h. And immediately, the broth was changed to pH 3.0 with HCl (1 M) or incubated for another hour. For the control, treatment in LB broth (pH 7.0) for 2 h was conducted. The temperature throughout the duration of above experiments was maintained at 37°C.

RNA isolation and library construction
According to the instruction from manufacturer, the total RNA of S. Enteritidis in each sample was extracted by using the Trizol Kit (Promega, USA). There existed some DNA left in the total RNA to be degraded by RNase-free DNase I (Takara Bio, Japan) for 30 min at 37°C. Results of RNase free agarose gel electrophoresis and an Agilent 2100 Bio-analyzer (Agilent Technologies, Santa Clara, CA) can testify if RNA isolation was of good quality. Afterwards, Poly (A) mRNA isolated by oligo-dT beads (Qiagen) was fragmented in the lysis buffer.
According to the fragmented mRNA, there generated the rst-strand cDNA through random hexamer-primed reverse transcription. Subsequently, the synthesis of the second-strand cDNA took place in presence of RNase H and DNA polymerase I (Chen et al. 2016). A QIA quick PCR extraction kit was used to enable the puri cation of cDNA fragments. These puri ed fragments were subjected to EB buffer for end reparation poly (A) addition and ligated to sequencing adapters. At last, extraction, puri cation and enrichment of the ligated cDNA fragments were conducted by PCR aiming to construct cDNA library after agarose gel electrophoresis.

Sequencing and reads alignment
It was performed on the Illumina sequencing platform (Illumina HiSeq™ 2000) to sequence the cDNA library with the pairedend technology applied (Gene Denovo Co. Guangzhou, China). A Perl program was employed to remove low quality sequences that refer to more than half of entire bases with quality lower than 20 or reads with more than 5% N bases (N means bases unknown) and adaptor sequences (Chen et al. 2016). Short sequences alignment was conducted with SOAPaligner/soap2 paralleled to a previous research. These sequencing reads were subsequently mapped to reference sequence by the SOAPaligner/soap2 (Li et al. 2009). The expression statistics and visualization throughout this study was accomplished with the technical assistance of R package (http://www.r-project.org/).

Page 9/22
4 Differentially expressed genes (DEGs) and function enrichment analyses The Bioconductor package edgeR (Robinson, McCarthy, & Smyth, 2010) was employed in identifying of differential expression genes. As the false discovery rate (FDR) was to examine and lter out the genes without signi cance in differential expression, we applied a p value, a threshold of the FDR ≤ 0.05 and |log 2 Ratio|≥1 according to Chen et al. (2016).
Referring to a previous study (Zhang et al. 2013) with minor modi cation, the DEGs were then subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. Once a Pvalue was not more than 0.05, we considered that both two methods have achieved a signi cant enrichment.

Quantitative real-time PCR (qRT-PCR) validations
It was 7500 Fast Real-Time PCR System (Applied Biosystem, Foster, USA) that was employed to justify the uniformity of gene expression between RNA-seq and qRT-PCR, with 9 DEGs selected. We regarded the 16 S RNA gene of S. Enteritidis as the internal control gene. In accordance with instruction manual, total RNA extraction was conducted with Trizol reagent, and the extracted RNA was instantly utilized in cDNA synthesis by using reverse transcriptase. Quantitation of each transcript was performed with total RNA in duplicate and the starting materials and each qPCR were conducted in triplicate.    Volcano plots diagram of DEGs associated with virulence between acid stressed and unstressed S. Enteritidis.
Red spots in the right part: up-regulated genes; green spots in the left part; down-regulated genes; black spots in the middle part: genes with insigni cant changes between the stressed and unstressed. Figure 2 qRT-PCR validation. X: the log2 fold change according to RNA-seq; Y: the log2 fold change according to qRT-PCR.