Published online before print November 6, 2007
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Department of Veterinary Medicine, University of Cambridge, Cambridge, United Kingdom
1 Correspondence: Department of Veterinary Medicine, University of Cambridge, Madingley Road, Cambridge, CB3 0ES, UK. E-mail: bab2{at}cam.ac.uk
ABSTRACT
The interactions of Salmonella enterica subspecies I serotype Abortusovis (S. Abortusovis) with ovine afferent lymph dendritic cells (ALDCs) were investigated for their ability to deliver Maedi visna virus (MVV) GAG p25 antigens to ALDCs purified from afferent lymph. Salmonellae were found to enter ALDC populations by a process of cell invasion, as confirmed by electron and confocal microscopy. This led to phenotypical changes in ALDC populations, as defined by CD1b and CD14 expression. No differences in the clearance kinetics of intracellular aroA-negative Salmonella from CD1b+ CD14lo and CD1b+ CD14– ALDC populations were noted over 72 h. ALDCs were also shown to present MVV GAG p25 expressed by aroA-negative S. Abortusovis to CD4+ T lymphocytes. Thus, the poor immune responses that Salmonella vaccines elicited in large animal models compared with mice are neither a result of an inability of Salmonella to infect large animal DCs nor an inability of these DCs to present delivered antigens. However, the low efficiency of infection of ALDC compared with macrophages or monocyte-derived DCs may account for the poor immune responses induced in large animal models.
Key Words: Salmonella enterica subspecies I serotype Abortusovis • Salmonella Abortus ovis • sheep • cannulation • lymphatic
INTRODUCTION
The interactions of professional APCs with invading pathogens are of great interest, considering the essential role these cells play in initiating and regulating immune responses [1 , 2 ]. In particular, dendritic cells (DCs) situated at the primary interface between the body and the environment, i.e., skin and mucosal surfaces, are exposed to all manner of antigenic insults. As the only APC capable of stimulating naive T cells, DCs represent important targets for vaccination strategies. In this respect, future vaccine design is likely to take into account the ability of a vaccine to be presented by DCs.
As a result of the increased immune responses they elicit, live vaccines are generally favored over their killed or subunit counterparts [3 ]. Unfortunately, for various reasons, not all pathogens can be attenuated. Thus, the use of successfully attenuated pathogens as live delivery vehicles for recombinant antigens has been studied [4 , 5 ]. In particular, bacteria, which can have an intracellular existence in APCs, e.g., Salmonella or Listeria monocytogenes, are particularly attractive as vehicles to deliver antigen to the appropriate inductive cells of the immune system, stimulating CD4 and CD8 T cell responses. Unfortunately, the success obtained in murine models has yet to be fully transferred to humans and other larger animal models.
There are a limited number of studies suggesting that Salmonellae are taken up by DCs [6 7 8 9 10 ]. In these studies, DCs derived in vitro or a DC-like cell line were used. Although these do show morphological and phenotypical features similar to DCs derived ex vivo, the full extent to which DCs generated with nonphysiological quantities of cytokines model those in vivo is still uncertain. More recently, investigations into the interaction between murine DCs and Salmonella typhimurium have been carried out in vivo, providing useful information about the involvement of DCs in the priming of naive T cells to Salmonellain vivo [11 , 12 ]. Only one study in an in vivo large animal model is available [13 ]. With many clinical applications failing to make the transition from rodent to humans, there is a place for these studies, which offer experimental opportunities not available in in vitro murine systems [14 ].
Here, we investigate the ability of Salmonella enterica subspecies I serotype Abortusovis (S. Abortusovis), a sheep-specific strain of Salmonella, to infect ex vivo afferent lymph DCs (ALDCs). The pseudoafferent lymphatic cannulation model offers a source of physiologically generated DCs and can be considered to be representative of the DC populations that are migrating from the periphery, having encountered antigen and/or pathogens. Work with more physiologically relevant sources of DCs [15 , 16 ] is essential if attenuated Salmonella delivery vehicles are to be used successfully for clinical applications.
MATERIALS AND METHODS
Animals
Adult Finnish Landrace-crossed sheep of either sex between the ages of 4 and 6 years were purchased from the Moredun Research Institute (Edinburgh, Scotland). Animals were maintained on-site at the Department of Veterinary Medicine, University of Cambridge (Cambridge, UK).
Cell and tissue-culture medium
Unless stated otherwise, RPMI-1640 medium containing L-glutamine (Cat. No. 11875, Invitrogen, Paisley, UK), 50 µM 2-ME, and antibiotic (10 µg/ml gentamicin) where appropriate was supplemented with 10% heat-inactivated FCS (RPMI/10% FCS).
Bacterial strains
S. Abortusovis SS44wt has been described previously [17
]. SU304 is an aroA deletion mutant and was a gift from Dr. Sergio Uzzau (Department of Biomedical Science, University of Sassari, Italy) [18
]. SU304 was grown in Luria Bertani (LB) broth supplemented with para-aminobenzoic acid at 10 µg/ml. SS44wt-GFP was created by transforming SS44wt with GFP cloned downstream of a TAC promoter in the plasmid pKEN [19
]. SU304225, expressing the EV1 strain of the Maedi visna virus (MVV) p25 gag gene [20
], was generated by transforming p25 cDNA cloned into the vector pTECH2 (Medeva, London, UK) [21
] into SU304. SU3042 was SU304 transformed with the empty pTECH2 vector. pTECH2 uses the nirB promoter to express tetC fusion proteins. This allows expression of recombinant antigen in situations of low oxygen tension, i.e., within cells. Plasmid was stable within cells for up to 24 h when ampicillin selection was removed.
Flow cytometry and antibodies
Cell-surface phenotype was analyzed by flow cytometry. mAb were against ovine cell surface antigens or were mAb known to cross-react with their ovine homologue. Briefly, cells were washed in PBS containing 0.1% BSA and 0.01% sodium azide (FACS wash buffer) and incubated with 1–2 µg/ml purified primary antibody for 40 min on ice. After two washes in FACS wash buffer, cells were incubated on ice in isotype-specific, Tri-color-conjugated sheep anti-mouse antibodies and strepavidin-PE at optimal dilutions for 20 min. Isotype control antibodies were used to indicate background staining and autofluorescence and used to set gates and quadrants for analysis. mAb used were VPM65 for CD14 [22
] and biotinylated CC20 for CD1b [23
].
Production of macrophages and DCs from PBMCs
Generation of macrophages from ovine PBMCs was carried out as described previously [24
]. Ovine in vitro-derived DCs were also generated as described previously [25
].
Pseudoafferent lymphatic cannulations
Sheep had their prefemoral or popliteal lymph nodes removed under general anesthesia. A minimum of 8 weeks later, the pseudoafferent lymphatic vessel, resulting from re-anastomosis of afferent and efferent vessels, was cannulated as described previously [26
]. Lymph was collected in sterile 250 ml capacity polypropylene bottles containing 250 U sodium heparin. Bottles were changed two to three times daily, as and when necessary
Enrichment and isolation of afferent lymph for DCs by density gradient, MACS, and FACsort
Cells from afferent lymph were washed in calcium and magnesium-free HBSS with 2% FCS, 0.2 mM EDTA, at 4°C. These were then resuspended in ice-cold RPMI/10% FCS, overlaid onto a discontinuous gradient of 11.5% v/v Optiprep (Nyegaard Diagnostics, Oslo, Norway), diluted in the same medium, and centrifuged at 800 g for 20 min at 4°C. Cells at the interface were washed twice in RPMI/2% FCS before use as an enriched source of ALDCs. These cells were stained as above for CD1b (with CC20 biotin for MACS) and/or CD14 expression and used for MACS or FACsorting. For MACS, cells were stained with streptavidin microbeads, and positively stained cells were selected on a magnetized column. Cells were kept on ice for FACsorting on a Becton Dickinson FACStar.
An Optiprep density gradient of 11.5% was found to enrich for large cells at levels comparable with that of a 14.5% Metrizamide gradient [26 , 27 ], i.e., up to 80% (data not shown) with a contaminating population of small cells (lymphocytes) present. Named as Optiprep gradient-enriched cells, the morphology and phenotype of these large cells were consistent with ovine ALDCs, as defined previously by others when gates were set for a DC analysis region, i.e., cells of high forward- and side-scatter [27 ].
Bacterial internalization by cells quantified using bacteria expressing GFP
Density gradient-enriched ALDCs, monocyte-derived DCs, or macrophages were infected with bacteria at a multiplicity of infection (MOI) of 10 for 1 h. Cells were then washed three times with PBS and stained for flow cytometry as usual. As controls for determining the number of bacteria associated with but not internalized by DCs or macrophages, cells were preincubated with Cytochalasin D at a final concentration of 10 µg/ml for 60 min at 37°C prior to infection. Cytochalasin D was present throughout the duration of the infection at 10 µg/ml.
Transmission electron microscopy (TEM)
Samples containing 5 x 106 cells (infected as above) were washed in PBS, pelleted, and fixed in 2.5% glutaraldehyde. After washing with 0.13 M phosphate buffer, pH 7.4, fixed cells were postfixed in 1% (w/v) osmium tetroxide/0.13 M phosphate buffer at 4°C before dehydrating in a graded series of ethanol. The cells were then incubated in propylene oxide and embedded in 1:1 propylene oxide/epoxy resin mix followed by a 100% epoxy resin mix. Samples were polymerized (cured) by baking at 60°C until the resin hardened.
Scanning EM (SEM)
Washed samples were pipetted as a cell suspension onto an Anodisc and fixed in 2.5% (v/v) glutaraldehyde. After coating in 1% (w/v) osmium tetroxide/0.13 M phosphate buffer at 4°C for 1 h, samples were washed and dehydrated through an acetone gradient. Finally, samples were dried with the critical point drying agent Hexamethyldisilazone mounted on an aluminium stub and coated with gold.
Confocal microscopy
Confocal microscopy was performed on a Leica TCS-NT-UV confocal laser-scanning microscope. Cells were cytospun onto precleaned microscope slides and then stained with phalloidin-conjugated rhodamine (0.2 mg/ml) and mounted in a 4'-diamidino-2-phenylindole microscopy slide mountant.
Salmonella survival in phagocytic cells (gentamicin survival assay)
Cells were infected with Salmonella at a MOI of 10 for 1 h at 37°C before washing three times with PBS and incubating at 37°C with DMEM (Cat. No. 19965, Invitrogen), supplemented with 10% FCS. Initially, for the first hour, gentamicin was added to a final concentration of 50 µg/ml, after which, it was lowered to 10 µg/ml. At different time intervals, the DCs were washed with PBS and lysed with 0.1% Triton X-100, equilibrated at 37°C. The number of viable intracellular bacteria was determined by plating serial dilutions onto LB agar with the appropriate supplements and counting colonies that grew overnight.
DC presentation of recombinant MVV p25 antigen expressed by Salmonella
ALDCs and in vitro monocyte-derived DCs from sheep primed to MVV p25 were infected with SU304225 at a MOI of 50 for 4 h (cells were viable after this time, as assessed by trypan blue staining). Extracellular bacteria were removed by washing, and cells were cocultured with autologous-purified CD4 lymphocytes (1x105 cells/well of a 96-well plate) using 2 x 105 in vitro-derived DCs or 1 x 105 ALDCs in RPMI/10% FCS containing 10 µg/ml gentamicin. Cells were cultured for 5 days at 37°C with 1 µCi [3H]-thymidine, added for the last 16 h, harvested, and then counted by liquid scintillation. Wells were set up in triplicate, and proliferation was compared with T cells stimulated by APCs infected with SU3042.
CD4 T cell lines specific for MVV p25 were derived from sheep immunized with recombinant p25. PBMCs were stimulated with p25 (at an optimal concentration, 25–50 µg/ml for 5–6 days) and then recombinant human IL-2 (20 U/ml for 5–6 days). The stimulation was repeated once before cells were stored in liquid nitrogen.
RESULTS
Surface molecule expression on ex vivo ALDC prior to and following encounter with Salmonella
Murine in vitro-derived bone marrow DCs and/or DC-like cell lines have previously demonstrated the ability of Salmonella to invade these cells successfully. To assess how ex vivo DC populations from afferent lymph interact with S. Abortusovis, flow cytometry was performed on Optiprep gradient-enriched cells ± Cytochalasin D (to differentiate surface adherent and internalized bacteria), incubated with S. Abortusovis, expressing GFP, SS44wt-GFP. Setting markers so that less than 1% of cells was associated with SS44wt-GFP in the presence of Cytochalasin D, large and small cell populations were analyzed separately (Fig. 1A
1B
1C
). Intracellular SS44wt-GFP could not be detected in the small cell (lymphocyte) population (Fig. 1B)
. The large cell population (Fig. 1C)
shows high autofluorescence, and the marker set in the presence of Cytochalasin D reflects this (less than 1% positive cells within the gate); however, these cells may also represent cells with bound extracellular bacteria. However, in the absence of Cytochalasin D, SS44wt-GFP was found in 1–3% of the large cell population, leading to the identity of the large cell population containing SS44wt-GFP to be investigated further by phenotypical analysis.
 View larger version (33K): [in a new window] |
Figure 1. S. Abortusovis uptake by ovine ALDCs was assessed by incubating Optiprep gradient-enriched cells (A) with SS44wt-GFP for 1 h at a MOI of 10 in the presence and absence of Cytochalasin D. Flow cytometry was performed, and the large cells (C) and smaller lymphocytes (B) as defined by their forward- and side-scatter properties (A) were analyzed for S. Abortusovis uptake by GFP fluorescence. The solid histogram profiles represent Salmonella uptake in the presence of Cytochalasin D. The black line depicts uptake without Cytochalsin D present. (D and E) Dot-plots show the CD1b and CD14 expression profile of large cells that did not and did contain Salmonella, respectively. Changes found in the levels of CD1b and CD14 expressed on large cells when cells were mock-infected (F) and infected with Salmonella (G) at a MOI of 10 for 1 h are also shown. Figures are percentage of cells in each quadrant (total of 1000 events displayed in each of the two-dimensional contour plots; quadrants set on control-stained cells).
|
The proportion of ovine ALDCs containing intracellular Salmonella postinfection was two- to threefold lower than that seen with murine in vitro-derived DCs when incubated with Salmonella at an identical MOI. As most Salmonella-DC studies have used in vitro-derived DCs, uptake of SS44wt-GFP by ALDCs was also compared with uptake by in vitro ovine monocyte-derived DCs. These were consistently found to have intracellular SS44wt-GFP at levels that were at least twofold higher when compared with ex vivo ovine ALDCs (Fig. 2A and 2C , n=3, P<0.05). The proportion of cells containing SS44wt-GFP was also five- to sixfold lower in ALDCs when compared with those of ovine monocyte-derived macrophages (Fig. 2A and 2B , n=3, P<0.05). Repeating the above experiments with a GFP-expressing, noninvasive Escherichia coli showed no uptake of the bacteria by ALDCs and greatly reduced levels of bacterial uptake by in vitro-derived macrophages and DCs (data not shown).
 View larger version (19K): [in a new window] |
Figure 2. Comparison of the relative uptake of S. Abortusovis by ovine ALDCs (A), monocyte-derived macrophages (B), and in vitro monocyte-derived DCs (C). Cells were incubated with SS44wt-GFP in the presence (shaded histograms) and absence (lines) of Cytochalasin D. The data shown are representative of three experiments carried out with cells from two different sheep. Groups were compared using a paired, two-tailed t-test (see Results).
|
Distinguishing ALDCs from lymphocytes and macrophages by size and morphology, TEM revealed that few of the infected DCs contained intracellular S. Abortusovis. Of the 50 DCs counted, only three contained Salmonella, consistent with the low percentage of DCs found to contain Salmonella by flow cytometry. Generally, three to four Salmonella were found in each of the infected DCs (Fig. 3A ). These intracellular bacteria were not observed to escape into the cytosol and inhabited vacuoles that were surrounded by single membranes (Fig. 3B) . Occasionally, more than one bacterium was observed within these vacuoles (Fig. 3C) ; otherwise, each vacuole was inhabited by a single bacterium. As expected with the intracellular survival abilities of Salmonella, fusion of lysosomal vesicles to vacuoles containing S. Abortusovis SS44wt was not observed, nor were vacuoles containing bacteria encapsulated by multilaminar structures. The latter has been described for S. typhimurium phoP mutants in the DC line CB1 and is reminiscent of MHC class II compartments [8 ].
 View larger version (79K): [in a new window] |
Figure 3. TEM of ALDCs after infection with wild-type S. Abortusovis for 1 h. Three to four intracellular Salmonella (denoted by arrowheads) can be found within each DC following infection (A). Each Salmonella inhabits a membrane-bound vacuole (B). Occasionally, several Salmonella could be observed within these structures (C). Original bar scales denote 1 µm.
|
 View larger version (48K): [in a new window] |
Figure 4. SEM of ALDCs pre- (A) and post-incubation with wild-type S. Abortusovis for 1 h (B). After incubation with bacteria, changes in the cell surface morphology characterized by membrane extensions and cell surface ruffling are seen. Original scale bar denotes 5 µm.
|
 View larger version (8K): [in a new window] |
Figure 5. Optiprep gradient-enriched ALDCs were positively selected for CD1b on magnetic columns and incubated with GFP-expressing S. Abortusovis at a MOI of 10 and fixed after 10 min. (A) Composite picture of B + C; (B) GFP-positive S. Abortusovis; (C) polymerized actin revealed by staining with rhodamine-conjugated phalloidin. Original magnification, x100. Images were processed with Adobe Photoshop 7 software for Windows.
|
 View larger version (24K): [in a new window] |
Figure 6. Intracellular survival in vitro of S. Abortusovis SS44wt ( , black) and SU304 ( , gray) in CD1b+ CD14lo and CD1b+ CD14– ex vivo ovine ALDCs from two sheep. CD1b+ CD14lo cells (1x105) from sheep B1136 (A) and sheep B1693 (C) or 2 x 104 CD1b+ CD14– from sheep B1136 (B) and sheep B1693 (D) were infected with Salmonella at a MOI of 10 for 1 h. Viable intracellular counts were performed by determining the number of CFU at designated time-points over a 72-h period. Data are expressed as means ± SD (n=3). Purity of FACsorted cell populations was >95%.
|
DCs infected for 1 h at MOIs of 10 and 20 did not induce the proliferation of CD4 antigen-specific p25 T cell lines (data not shown). Therefore, cells were infected for an increased length of time (4 h) at a higher MOI—50. Figure 7 demonstrates that under these conditions, in vitro monocyte-derived DCs and ex vivo ALDCs were capable of presenting Salmonella-expressed MVV GAG p25 antigen to CD4 antigen-specific p25 T cell lines.
 View larger version (16K): [in a new window] |
Figure 7. Optiprep gradient-enriched ALDCs and in vitro monocyte-derived DCs were incubated with SU304225 (solid bars) or SU3042 (open bars) for 4 h at a MOI of 50 before removing extracellular bacteria and culturing with CD4+ antigen-specific T cell lines. ALDCs (1x105) and 2 x 105 in vitro monocyte-derived DCs were cultured with 1 x 105 cells from a p25-specific CD4+ T cell line. Graphs show the proliferative response of CD4+ T cells when incubated with infected DCs. Medium controls were always less than 1000 cpm, and Con A controls were all greater than 100,000 cpm. Data are representative of two experiments set up in triplicate, and mean ± SD cpm are shown.
|
To determine whether Salmonella can infect and survive in physiological populations of DCs, we have carried out investigations into the interactions between ovine ALDCs and the sheep-specific strain of S. Abortusovis. To our knowledge, this is the first large animal investigation of Salmonella DC interaction in vitro to use DCs from a physiological source.
Despite the ability of Salmonella to be internalized by ovine ALDCs, bacterial uptake was less than that seen with ovine monocyte-derived macrophages. With TEM studies confirming the low levels of intracellular S. Abortusovis residing within ALDCs, the internalization and uptake of bacteria by ovine ALDCs were less efficient than that of macrophages. Similar findings have been described when comparing the relative uptake of E. coli by bone marrow-derived mouse DCs to murine peritoneal macrophages [31
]. A likely explanation is the lower levels of phagocytosis found in maturing ALDC populations [32
], meaning that uptake is reliant on invasion alone. Indeed, our in vitro-derived DC data along with preliminary data indicating a lack of noninvasive E. coli uptake by ALDCs would suggest that uptake of Salmonella by afferent lymph cells is a result of invasion and not phagocytosis. The cell-surface modifications seen with SEM and actin polymerization observed by confocal microscopy support this, which may also explain the low levels of bacteria associated with ALDCs draining a mucosal site of infection with S. Abortusovis [13
]. This study showed that phagocytic cells, granulocytes and migratory monocytes, had higher levels of cell-associated bacteria than CD1b+ DCs; indeed, there were 
3.5-fold more Salmonella-positive monocytes than DCs.
Unlike humans and sheep, mice do not express a homologue of CD1b. Therefore, the changes in ALDC CD1b expression after incubation with S. Abortusovis were of particular interest. The overall percentage and actual number of CD1b expressing large cells increased by 20%; this was at the expense of the CD1b– cell population. As the role of CD1b is the presentation of lipid antigens [33
34
35
], it is probable that these ovine ex vivo DCs up-regulate CD1b in response to a maturation stimulus. Increased CD1 expression on ovine ALDC is also seen in vivo after antigenic challenge [26
].
The inability of S. Abortusovis to infect lymphocytes (Fig. 1 and ref. [13 ]) suggested that in addition to intestinal mucosa in sheep [36 ], macrophages (as shown in this study), and afferent lymph monocytes [13 ], S. Abortusovis is able to infect ovine ALDCs selectively in vitro (this study) and in vivo [13 ].
Salmonella containing deletions in the aro genes are impaired in their ability to reproduce [37 ]. Such Salmonella produce a self-limiting infection that is of long enough duration to deliver heterologous antigens they may be engineered to carry into APCs. FACsorting ovine ALDCs into CD1b+ CD14lo and CD1b+ CD14– cell populations did not reveal an obvious difference in the kinetics of S. Abortusovis survival between the two DC populations, in that SS44wt survived, and SU304 was cleared in both. A similar study that FACsorted CD11c+ DCs from fetal liver tyrosine kinase 3 ligand-treated mice into CD8a+ and CD8a– populations also failed to show any difference in intracellular bacterial growth and survival once these two DC populations were infected with wild-type Salmonella Dublin [38 ]. Salmonellae residing within DC vacuoles do not require macrophage virulence factors for intracellular DC survival, and the vacuoles themselves do not have lysosomal markers, unlike macrophage vacuoles in which Salmonella reside [8 , 39 ]. Wild-type S. Abortusovis was able to persist in ovine ALDCs, whereas an aroA– S. Abortusovis mutant, SU304, is cleared rapidly. In contrast, wild-type S. typhimurium has been shown to be cleared slowly from human monocyte-derived DC [9 ].
Despite the rapid clearance of SU304 from ALDCs following infection, SU304225, which expressed MVV p25 from the nirB promoter as a tetC fusion protein, presented p25 to specific CD4 T cell lines, albeit at a high MOI. Although murine studies have been carried out by others [6 , 11 ], this is the first study to demonstrate that ex vivo maturing ALDCs infected by an attenuated Salmonella carrying recombinant antigens can stimulate a CD4 antigen-specific T cell response. The high MOIs required for the ALDCs to present antigen to CD4 T cell lines were unexpected, considering the efficiency with which DCs present antigen. However, ALDCs are considered to be a maturing population, which may mean that their ability to process new antigen from invading Salmonellae is decreased. In contrast, Ruedl and colleagues [40 ] have described Langerhans cells given a maturation stimulus in vivo, continuing to have the ability to take up and process native protein (which was present throughout the T cell proliferation assays). Recent data from mice have also shown that virulent S. typhimurium reduces antigen presentation on MHC classes I and II molecules by interfering with phagosome lysosome fusion using Salmonella pathogenicity island 2-coded effector proteins [41 42 43 44 ]. The need for a high MOI to produce efficient antigen presentation may reflect the presence of similar immune evasion mechanisms in S. Abortusovis.
Overall, the present work demonstrates that Salmonella expressing recombinant antigens can infect and deliver heterologous antigens to ex vivo-obtained ovine ALDCs in vitro. This extends much of the findings from similar in vitro murine DC studies to large animals. Moreover, based on the in vitro data obtained, the poorer immune responses obtained in large animals when Salmonella are used as antigen-delivery vehicles are not a result of an inability to infect large animal DCs but may be a result of the high MOI needed to allow antigen presentation. This is difficult to achieve in larger animals.
ACKNOWLEDGEMENTS
S. S. M. C. was supported by a Biotechnology and Biological Sciences Research Council studentship. The recombinant human IL-2 was provided by the EU Program EVA/MRC Centralized Facility for AIDS Reagents, National Institute for Biological Standards and Control, UK (grant nos. QLK2-CT-199900609 and GP828102). We thank Sergio Uzzau (Department of Biomedical Science, University of Sassari, Italy) for providing S. Abortusovis SS44wt and SU304 and Mike Peacock (University of Cambridge, Cambridge, UK) for assistance with the EM. All animal experiments were performed in accordance with the Animals (Scientific Procedures) Act 1986, UK.
Received June 14, 2007; revised October 5, 2007; accepted October 6, 2007.
REFERENCES
β+ T cells Nature 372,691-694[CrossRef][Medline]
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
