Salmonella Exhibit Altered Cellular Localization in the Presence of HLA-B27 and Codistribute with Endo-Reticular Membrane

Salmonella enteritica (S. enteritica) induce and require unfolded protein response (UPR) pathways for intracellular replication. Salmonella infections can lead to reactive arthritis (ReA), which can exhibit associations with Human Leucocyte Antigen (HLA)-B∗27 : 05. S. enteritica normally reside in a juxtanuclear position to the Golgi apparatus, representing the formation and residence within the Salmonella-containing vacuole (SCV). Changes in cellular localization of infecting Salmonella can alter their ability to replicate. We therefore used isogenic epithelial cell lines expressing physiological levels of HLA-B∗27 : 05 heavy chain (HC) and a control HLA-B allele, HLA-B∗35 : 01.HC to determine any changes in Salmonella localization within epithelial cells. Expression of HLA-B∗27 : 05 but not HLA-B∗35 : 01 was associated with a quantifiable change in S. enteritica cellular distribution away from the Golgi apparatus. Furthermore, the Salmonella requirements for UPR induction and the consequences of the concomitant endoplasmic reticulum (ER) membrane expansion were determined. Using confocal imaging, S. enteritica bacteria exhibited a significant and quantifiable codistribution with endo-reticular membrane as determined by ER tracker staining. Isogenic S. enterica Typhimurium mutant strains, which can infect but exhibit impaired intracellular growth, demonstrated that the activation of the UPR was dependent on an integral intracellular niche. Therefore, these data identify cellular changes accompanying Salmonella induction of the UPR and in the presence of HLA-B27.


Introduction
Misfolding of the major histocompatibility complex (MHC) class I heavy chain (HC) protein encoded by the human leucocyte antigen B27 (HLA-B27) allele can induce endo-plasmic reticulum (ER) stress responses [1,2]. The enhanced propensity to misfold has been proposed to explain why HLA-B27 drives the inflammatory arthritic disorder ankylosing spondylitis (AS) [3][4][5][6]. HLA-B27 misfolding can disrupt ER homeostasis and trigger the unfolded protein response (UPR) as determined by activation of the XBP-1 [2] and ATF6 pathways [1] as well as in AS patients, by the upregulation of UPR effector genes such as HRD1 [7]. In addition, HLA-B27 can lower the threshold at which the UPR can become activated [8]. UPR induction by HLA-B27 may contribute to arthritic disease by enhanced production of proinflammatory cytokines such as IL-23 and IL-17 [6,9,10] and/or affect function of specific immune cell subsets [11].
S. enterica is a facultative gastrointestinal (GI) pathogen which occupies an intracellular niche termed the Salmonella-containing vacuole (SCV) in intestinal epithelial or phagocytic cells [21,22]. Salmonella effector proteins are required for SCV formation and maintenance. Maturation of the SCV and bacterial cellular localization can determine the survival and replicative capability of Salmonella [23,24]. During the early stages of infection, the SCV migrates to a juxtanuclear location associated with the microtubule organizing center and the Golgi apparatus in epithelial cells (reviewed in [21,22]). SCVs follow an endosomal maturation route ,and after the onset of intracellular replication, large tubular membrane structures called Salmonella induced filaments (Sifs) grow out from microcolonies [21,[25][26][27]. The formation of the SCV though appears to interact with distinct organelles which not only involve lysosomal compartments but also the endoplasmic reticulum (ER) and the secretory pathway [28][29][30].
The mechanistic link between ReA and HLA-B27 remains poorly defined. However, recently, we demonstrated that S. enteritica Typhimurium can activate the XBP-1 and ATF6 UPR pathways, which were required for enhanced lipid metabolism and endo-reticular membrane biosynthesis [8]. In addition to mammalian cell lines expressing HLA-B27 hosting higher numbers of Salmonella [31][32][33], we further demonstrated that ER stress environments can lead to enhanced levels of Salmonella within epithelial cells, therefore, providing a potential mechanistic link between HLA-B27 expression, Salmonella infectivity, and ReA [8].
To date, the relationship between HLA-B27 expression and ER stress has not been determined with respect to the cellular localization of infecting Salmonella. We therefore sought to determine the cellular location of Salmonella within HLA-B27 expressing epithelial cells. The SCV appears to be central for optimal bacterial replication and survival. We wanted to establish the requirement for UPR induction when formation and integrity of the SCV were compromised. Our findings suggest that Salmonella exhibit changes in cellular location in epithelial cells expressing HLA-B27, and induction of the UPR is determined by the structural integrity of the SCV.

Materials and Methods
2.1. Cells Lines and Antibodies. Cells were maintained in DMEM with 10% FBS (Globepharm), maintained in a 5% CO 2 37°C incubator. HeLa isogenic cell lines generated using the Invitrogen Flp-In™ system, containing either two copies of the HLA-B * 27 : 05 and -B * 35 : 01 HCs and an empty vector (referred to as empty (E) 84), were used as previously described [34]. Monoclonal antibody Giantin (PRB-114C) was from Covance. Glibenclamide BODIPY-FL (green, Invitrogen) was used for quantification of ER membrane localization.

ER Stress Induction.
To induce ER stress responses cells were treated with tunicamycin (TUN) or thapsigargin (TPG), controls were treated with appropriate vehicle (DMSO) control.

Flow Cytometry and Microscopy.
For fluorescenceactivated cell sorting (FACS), cells were trypsinized, washed, and fixed for 10 mins with 3.8% paraformaldehyde (PFA) pH 7.4. Infected cells were analyzed on a LSR2 or LSR Fortessa (BD Biosciences). Flow cytometry data was analyzed in FlowJo 8.7.3. In each experiment, three replicate wells were analyzed for each condition tested. For microscopic analysis, coverslips containing infected cells were washed with 1× DPBS, fixed for 10 mins with 3.8% PFA (pH 7.4), and washed twice with 1× DPBS. Coverslips were stored at 4°C in the dark until analysis.
2.6. Quantification of Salmonella Cellular Distribution. Cells were fixed with 3.8% PFA 24 hrs p.i. and the cis/medial-Golgi compartment visualized using anti-giantin and antirabbit Alexa Fluor-555 (Invitrogen). Cell nuclei were stained with DAPI. Images were collected using a Leica microscope and analyzed using Metamorph software (molecular devices). The distance between Salmonella microcolonies to the medial/cis Golgi compartment was calculated using HeLa cells were grown either on glass coverslips or in glass bottomed 96 microwell plates and infected with mCherry expressing S. enterica Typhimurium and stained with glibenclamide green. Cells were fixed, washed, and counterstained with DAPI and glibenclamide green and then visualized by fluorescence microscopy or automated confocal analysis.
Microwell images were acquired by an Opera LX (Perki-nElmer) plate reader, equipped with a confocal microscope, using NA = 0:6, 40× air objective. 100 ms exposure times was applied at the DAPI channel (365 nm) for nuclear detection, 2000 ms with 3330 μW laser power at the ER channel (488 nm), and 2000 ms with 1600 μW laser power at the Salmonella channel (561 nm). The camera pixels were binned by 2 resulting pixel size of 0:323 × 0:323 μm. In total, 4,800 images were acquired per 96-well plate (50 images per well) that was processed in one batch using the same image analysis pipeline, algorithms, and parameters.
2.8. Fluorescent Image Processing. Image processing was done on a Tyan FT48-B8812 high-performance server grade barebone computer equipped with 4 twelve-core AMD Opteron CPUs and 256 GB RAM. The system is capable to run 48 parallel threads that allowed us the parallelization of computations resulting in a massively increased computation speed running under a 64 bit version Ubuntu Linux 10.10.
Image processing was performed using ImageJ [37] version 1.45s and Java 1.6.0_20. Images were acquired as Opera LX FLEX files and were converted into 16 bit TIF format. The colocalization measurement pipeline started with noise reduction. The maximum intensity of sample out-of-cell bacteria was measured in the ER channel, and all ER channel intensities below a value of 270 were considered as noise. The Salmonella channel image stack was segmented as a batch of 4800 images of a plate using the Otsu method with stack histogram as the default. The threshold was manually adjusted to increase the segmentation precision with some plates. An ImageJ macro was developed to measure the ER area colocalized with the area of bacteria. ER pixels under the segmented bacteria mask were assumed to be clustered into two classes: (a) colocalized (brighter) pixels and (b) non-colocalized (darker) pixels. The Otsu algorithm was used to find the threshold to classify those two classes of ER pixels superimposed with the bacteria pixels. After classification, the colocalized area was measured and was divided by the total area covered by bacteria to calculate the colocalized ratio for each image as ColocRatio = a coloc /a total [34].

Altered Cellular Localization of Salmonella in HLA-B27
Heavy Chain Expressing Cells. Previously, we demonstrated that expression of HLA-B27 in epithelial lines can lead to enhanced bacterial recovery and altered intracellular localization [8]. The ability of Salmonella to replicate can correlate with their intracellular localization [24]. We therefore infected HLA-B * 27 : 05.HC, HLA-B * 35 : 01.HC and E84 isogenic HeLa lines and used confocal microscopic analysis to identify and quantitate St.GFP location.
Following infection, cells were stained for the Golgispecific marker giantin (red) and the nucleus with DAPI (blue) (Figure 1(a)). We detected Salmonella concentrated in juxtaposition to the Golgi apparatus (which reflects Salmonella within the SCV) in cell lines where no enhanced bacterial recovery was observed, i.e., E84 and HLA-B * 35 : 01.HC (Figure 1

Activation of the UPR by Replicating Salmonella Enterica Typhimurium Requires Effector Translocation and
Localization within the SCV. S. enterica Typhimurium can induce the UPR predominantly at late time points post infection (p.i.), suggesting that ER stress activation occurs during the growth phase and not as a result of the invasion process [8]. To determine whether ER stress activation was associated with replication, we infected HeLa cells with ST.mCherry, which were transiently transfected with ΔDBDXBP-1v, followed by flow cytometry analysis. Cells were gated according to levels of mCherry fluorescence, i.e., low (G1), medium (G2), and high (G3) (Figure 2(a), left panel). For each group, XBP-1v activation was then assessed (Figure 2(a), right panel). HeLa cells containing high mCherry signals (G3) and thus replicating S. enterica Typhimurium exhibited significant activation of ΔDBDXBP-1v, which was comparable to ER stress induction with tunicamycin (TUN) (Figure 2(a), right panel).
We next examined whether UPR induction was associated with the infection process by using isogenic S. enterica Typhimurium mutant strains which can infect but are impaired in their intracellular growth. We employed S. enterica Typhimurium strains which expressed deletions in (a) ssaV, which encodes an inner membrane component of the Salmonella pathogenicity island-2 (SPI-2) secreton,   ANOVA was performed on mCherry MFI values (P < 0:0001) with Tukey's multiple comparison post-test to determine significant differences between individual groups P < 0:05 ( * ) and P < 0:001 ( * * * ). 5 Journal of Immunology Research essential for type III secretion system (TTSS-)-mediated protein translocation [38] and required for replication in HeLa cells and macrophages [39]; (b) aroCpurD, an auxotrophic mutant which exhibits no intracellular growth [39]; and (c) sifA, which is required for the formation of Sif tubular structures in epithelial cells and SCV integrity [23,40,41]. We also employed a ΔsopB mutant which is partially impaired in intracellular replication due to defects in SCV maturation but can form an SCV [42,43].
HeLa cells, transfected with ΔDBDXBP-1v, were infected with mCherry expressing S. enterica Typhimurium ΔssaV, ΔaroC purD, or ΔsifA strains, harvested 4 and 24 hrs p.i., and analyzed by flow cytometry. ΔssaV and ΔaroCpurD strains failed to induce any activation of ΔDBDXBP-1v. The ΔsifA mutant, of which the majority escape the SCV and replicate within the HeLa cell cytosol to equivalent or higher levels than S. enterica Typhimurium wild-type strains [23], also failed to activate ΔDBDXBP-1v, even when gating on highly infected cells (Figure 2(b)). Interestingly, the ΔsopB mutant can indeed activate XBP-1, suggesting that growth of S. enterica Typhimurium within the SCV microenvironment is necessary for XBP-1 activation (Figure 2(b)).
To determine if the increase in intracellular bacteria from ER stressed cells was dependent on intracellular replication, HeLa cells were treated with DMSO or 200 nM thapsigargin (TPG) 16 hrs prior to infection with ST.mCherry or the ΔsifA and ΔsopB mutant strains. S. enterica Typhimurium 12023 and the ΔsopB mutant strains which can replicate within the SCV, exhibit similar but significantly different increases in intracellular bacteria after TPG treatment (Figure 2(c)). However, ER stress induced cells infected with the ΔsifA mutant, which can escape and replicate within the host cell cytosol, demonstrated no increase in bacterial numbers in the TPG-treated cells (Figure 2(c)). Thus, intra-cellular localization within the SCV appears to be necessary for ER stress-mediated increases in bacterial replication.

Salmonella Codistribute with Endo-Reticular Membrane.
Previously, we demonstrated that UPR induction and increased bacterial replication required de novo lipid synthesis [8]. Furthermore, cells infected with intracellular Salmonella appear to enhance endo-reticular membrane synthesis, and recent observations suggest that ER membrane and/or associated proteins could well interact with infecting Salmonella [28][29][30]44].
We therefore wanted to infect cells and determine whether during infection we could detect any putative associations and/or codistribution between endo-reticular membrane and Salmonella. HeLa cells were therefore infected with either wild-type or ΔsifA S. enterica Typhimurium expressing mCherry and at 4 and 16 hrs p.i. were stained with ER tracker green and analyzed by fluorescence microscopy. At 4 and 16 hrs p.i., the ER tracker green staining was found to codistribute with wild-type (Figure 3(a), top panel) but not ΔsifA S. enterica Typhimurium (Figure 3(a), bottom panel). Quantification of colocalization of Salmonella with ER tracker showed that ER tracker-positive membrane exhibits enhanced codistribution with wild type but not ΔsifA deficient bacteria which escape from the SCV (Figure 3(b)).

Discussion
Here, we demonstrate that changes in the survival of Salmonella in the presence of HLA-B * 27 : 05 are associated with changes in the cellular localization of the bacteria ( Figure 1). As to how this apparent altered localization occurs, it is possible that bacteria fail to form an SCV or may exit the SCV more rapidly in the presence of misfolding  Journal of Immunology Research HLA-B * 27 : 05. In epithelial cells, Salmonella ΔSifA mutants have been shown to escape the SCV and exhibit increased replication rates within the cell cytoplasm [23]. At later infection time points, a proportion of wild-type S. typhimurium SCVs (~30%) can move to the periphery of the cell through a microtubule based, kinesin-dependent, Salmonella pathogenicity island-2 (SPI-2) effector-mediated process [45]. Our observations suggest that an altered cellular distribution of Salmonella could explain the enhanced bacterial persistence in the presence of HLA-B27 [31,33], thus contributing to dissemination of bacteria from their original infection site and the development of ReA. Recently, Salmonella enterica serovars Typhi and Paratyphi A infections were shown to be strongly associated with HLA-B * 27 : 05 [46]. Thus, it is intriguing to suggest that the expression of misfolding HLA-B * 27 : 05 could provide a permissive environment for intracellular bacterial growth [8,46].
An integral SCV appears to be necessary for UPR induction, supporting our earlier observations that bacterial strains which lacked the ability to form stable and intact SCV structures did not lead to significant ER membrane expansion [8]. Here, we report that indeed an intact SCV and bacteria capable of intracellular survival are required for activating XBP-1 ( Figure 2). Intriguingly, a strain lacking the phosphoinositol phosphatase SopB, though capable of forming an SCV but lacking recruitment of proteins to the SCV structure [43,47], could indeed induce the UPR. SopB is known to activate the serine/threonine kinase, Akt1, which can intersect the UPR [48] particularly by interacting with the PERK pathway [49]. Activation of the UPR by the sopB mutant suggests that UPR activation does not proceed via an Akt pathway, further supporting our previous observations that during the growth phase there is no engagement with the PERK arm of the UPR response [8].
Our data suggest that during Salmonella induced endoreticular membrane expansion, there is increasing codistribution between bacteria and membrane ( Figure 3). Such codistribution between Salmonella and ER tracker-positive membrane could well be associated with SCV integrity and maintenance. During SCV generation, these vacuolar structures can transiently acquire and lose early endosome associated proteins such as the GTPase Rab5, EEA1 and transferrin receptor, followed by acquisition of late endosome proteins, for example Rab7, LAMP1, and the vacuolar ATPase [26,27,50]. However, our data indicate that Salmonella can interact and/or codistribute with endo-reticular membranes. Interactions of the SCV with the secretory pathway have been documented [27]. Post Golgi vesicles and markers have been reported to be recruited to the vicinity of the SCV [51,52], while ER membrane bound markers such as calnexin can cosediment with SCVs following sucrose gradient isolation [53,54]. Furthermore, analysis of Salmonella either associated or not associated with the microtubule-associated protein 1A/1B-liight chain 3 (LC3, a marker for autophagy) revealed that up to 20% of intracellular bacteria were positive for calnexin and protein disulfide isomerase (PDI) [54]. Recently, a more in-depth proteome analysis did reveal ER membrane, and associated proteins contribute to both the SCV and Salmonella-host protein-protein interactions [29,30]. These observations indicate that different membrane compartments may contribute and/or interact with the SCV/Sif superstructure. In addition, intracellular bacteria other than Salmonella such as Brucella and Chlamydia interact with ER membrane [27,[55][56][57]. Intriguingly, such intracellular bacteria have been reported to be associated with ReA [12]. Thus, our data would further support that endo-reticular membrane and/or associated proteins can either be recruited to the SCV and/or interact with Salmonella during the growth phase and with UPR induction possibly being a shared mechanism contributing to intracellular bacterial induced ReA.

Data Availability
The figures and image data used to support the findings of this study are included within the article.

Conflicts of Interest
The authors declare that they have no conflicts of interest.

Authors' Contributions
Antony N. Antoniou contributed to the planning, designing and performing of experiments, interpretation of data, and writing of the manuscript. Janos Kriston-Vizi contributed to the planning, designing, performing of experiments, writing and designing the image processing software, interpretation of data, and writing of the manuscript. Izabela Lenart contributed, planned, and performed the biochemical analysis of the respective cell lines employed throughout the study and contributed to the writing of the manuscript. Takao Iwawaki designed and generated constructs employed in the study. Simon J Powis contributed to the data interpretation, experimental design, and writing of the manuscript. Keith Gould contributed to the generation of the constructs and cell lines used throughout the study, data interpretation, and writing of the manuscript. Darren Nesbeth contributed to the generation of the constructs and cell lines used throughout the study, data interpretation, experimental design, and writing of the manuscript.