The presence of professional phagocytes dictates the number of host cells targeted for Yop translocation during infection

Abstract

Type III secretion systems deliver effector proteins from gram-negative bacterial pathogens into host cells, where they disarm host defenses, allowing the pathogens to establish infection. Although Yersinia pseudotuberculosis delivers its effector proteins, called Yops, into numerous cell types grown in culture, we show that during infection Y. pseudotuberculosis selectively targets Yops to professional phagocytes in Peyer’s patches, mesenteric lymph nodes and spleen, although it co-localizes with B and T cells as well as professional phagocytes. Strikingly, in the absence of neutrophils, the number of cells with translocated Yops was significantly reduced although the bacterial loads were similar, indicating that Y. pseudotuberculosis did not arbitrarily deliver Yops to the available cells. Using isolated splenocytes, selective binding and selective targeting to professional phagocytes when bacteria were limiting was also observed, indicating that tissue architecture was not required for the tropism for professional phagocytes. In isolated splenocytes, YadA and Invasin increased the number of all cells types with translocated Yops, but professional phagocytes were still preferentially translocated with Yops in the absence of these adhesins. Together these results indicate that Y. pseudotuberculosis discriminates among cells it encounters during infection and selectively delivers Yops to phagocytes while refraining from translocation to other cell types.

Introduction

Type III secretion systems (TTSS) are multi-component export machines found in gram-negative bacterial pathogens that deliver effector proteins from the bacterial cytosol into host cells (Gophna et al., 2003). Once inside host cells, effector proteins modulate cellular functions, thus enabling the bacteria to inactivate host defenses and establish replication niches (Mattoo et al., 2007). A plasmid-encoded TTSS is found in three pathogenic Yersinia species, which include the two enteric Yersinia pathogens, Y. pseudotuberculosis (Yptb) and Y. enterocolitica (Ye), and the causative agent of bubonic and pneumonic plague, Y. pestis (Gemski et al., 1980; Portnoy and Falkow, 1981). These highly homologous secretion systems deliver at least 6 effector proteins, called Yops, into cells and are required to cause significant disease in humans and other mammals (Cornelis, 2002). Yersinia spp colonize many tissues in mice including the Peyer’s patches (PP), mesenteric lymph nodes (MLN), spleen, liver and lungs, depending on the route of infection (Felek and Krukonis, 2009; Logsdon and Mecsas, 2006; Pepe and Miller, 1993). During the initial steps of infection, Yersinia are intracellular as the enteric pathogens pass through M cells lining the intestinal wall (Clark et al., 1998; Marra and Isberg, 1997) and Y. pestis initially resides in macrophages after subcutaneous inoculation (Meyer, 1950). However, in general, the vast majority of yersiniae is found extracellularly during tissue infection (Balada-Llasat and Mecsas, 2006; Bergman et al., 2009; Simonet et al., 1990). As extracellular pathogens, yersiniae could potentially deliver Yops into all of the host cells they encounter. In fact, Yersinia does effectively deliver Yops into many different cell types infected in culture including epithelial cells, macrophages, dendritic cells, T cells and neutrophils (Brodsky and Medzhitov, 2008; Davis and Mecsas, 2007; Grosdent et al., 2002; Sory et al., 1995; Yao et al., 1999)

Yop delivery requires that yersiniae bind to mammalian cells (Grosdent et al., 2002; Mejia et al., 2008). Several bacterial proteins, including Invasin and YadA, mediate Yptb binding and Yop delivery to host cells (Mejia et al., 2008). These two adhesins are not expressed in Y. pestis (Parkhill et al., 2001). Invasin binds to β₁-containing integrins, causing activation of Src kinases, which in turn enhances Yop translocation into cells (Mejia et al., 2008). The interaction of YadA with mammalian cells is mediated indirectly through β1 integrins, by YadA binding to collagen or fibronectin which then bind β1-containing integrins (Tertti et al., 1992). In addition, opsonization of Yersinia by complement or antibody is sufficient to permit binding to macrophages and neutrophils and Yop delivery (Fallman et al., 1995; Grosdent et al., 2002). It is clear that bacterial-encoded adhesins are important for infection in animals (Heise and Dersch, 2006; Pepe and Miller, 1993); however, it is unknown whether bacterial ligands, Fc and/or complement play a role in host cell binding and Yop delivery during animal infection. The fact that the receptors for Invasin and YadA, β1-containing integrins, are found on a large number of cells, including epithelial cells, professional phagocytes and B and T cells (Brakebusch and Fassler, 2005) explains, in part, why Yops are translocated into numerous cell types grown in culture and could mean that Yptb targets many cell types for Yop translocation during infection of tissues.

Most effector Yops have profound effects on cells grown in culture (Brodsky and Medzhitov, 2008; Davis and Mecsas, 2007; Grosdent et al., 2002; Sory et al., 1995; Yao et al., 1999) and each Yop is important for tissue colonization and/or full virulence in mouse infections (Leung et al., 1990; Logsdon and Mecsas, 2003; Monack et al., 1998). However, it is not clear whether the effects observed in cultured cells correspond to their role in tissues during animal infections. Identifying host cells that are targeted for Yop translocation during infection is one approach towards understanding how each Yop participates in dismantling host defenses. Cells with translocated Yops after intravenous infection with Y. pestis or Ye were identified, using fusions of the N terminus of Yops with TEM (Koberle et al., 2009; Marketon et al., 2005). TEM is a truncated version of β-lactamase, which has enzymatic activity that can be detected by cleavage of a fluorescence substrate (Zlokarnik et al., 1998). Y. pestis and Ye delivered Yops to macrophages, neutrophils and dendritic cells in the spleen; in addition, Ye also delivered Yops to B cells (Koberle et al., 2009; Marketon et al., 2005). However, when Y. pestis or Ye were incubated with isolated splenocytes this specificity was not observed. These observations raise the question of whether the specific cell-type targeting during infection was due to Yersinia selectively co-localizing with these cells.

Using the TEM technology to identify host cells that are targeted by Yptb after oral infection, we find that Yptb had a pronounced specificity for Yop delivery into phagocytic cells, particularly neutrophils, in the PP, MLN and spleen. This specificity was not due solely to co-localization with Yptb during infection because phagocyte-specific targeting was retained in the absence of tissue architecture. Furthermore, in the absence of neutrophils, overall levels of translocation were severely reduced. Therefore, inherent properties of phagocytic cells interacting with Yptb promote Yop delivery during infection.

Results

Translocated YopH is enriched in neutrophils, macrophages and dendritic cell populations in PP, MLN and spleen after oral infection

To understand how the Yptb Yops circumvent host defenses, it is essential to identify the immune cell and/or other cells targeted by Yptb for Yop translocation during infection. To identify immune cells targeted by Yptb during infection, we generated a Yptb strain expressing a chimeric protein, HTEM, which contains the N-terminal secretion and translocation domain of the effector protein YopH fused to the reporter protein TEM. Chimeric proteins containing the first 100 amino acids of a Yop fused to another protein will generally be translocated into host cells (Marketon et al., 2005; Sory et al., 1995). The strain, WT-HTEM, expressed, exported, and translocated both WT YopH and HTEM, and was as virulent as Yptb (Fig S1 and unpublished data). TEM, a β-lactamase, cleaves the membrane permeable dye CCF2-AM and changes its fluorescence from green to blue (Gao et al., 2003), thus enabling identification of mammalian cells containing HTEM by their blue fluorescence. In contrast, when cells were infected with a translocation defective mutant expressing HTEM, ΔyopB-HTEM, a mutant lacking a critical component of the translocon, cells remain green because the chimeric protein is not translocated into cells (Fig S2A).

To identify immune cells into which HTEM has been translocated, mice were orogastrically infected with WT-HTEM and infection was allowed to proceed for 5 days. An oral-gastric route of infection was used because Yptb is a natural enteric pathogen, and its ability to disseminate from the GI tract to the PP, MLN and spleen permitted analysis of multiple tissues after oral infection (Barnes et al., 2006). At day 5 post-infection, mice were sacrificed, the PP, MLN and spleen were harvested, and single-cell suspensions were generated in the presence of antibiotics to halt any Yop translocation post-harvest (see experimental procedures). Cells were incubated with CCF2-AM, which freely diffuses through mammalian plasma membranes and is then modified by esterases, trapping CCF2-AM inside the cell (Zlokarnik et al., 1998). Cells were then fluorescently labeled with the following antibodies to identify the indicated cell type: α-GR1 and α-CD11b to distinguish neutrophils (GR1⁺CD11b⁺) from macrophages (CD11b⁺GR1⁻), α-CD11c (primarily dendritic cells), α-B220 (primarily B cells), α-CD4 (primarily T helper cells), or α-CD8 (primarily cytotoxic T cells). Five days post-infection, the total number of neutrophils (GR1⁺CD11b⁺) and macrophages (GR1+CD11b+) increased 4–12 fold in the PP, MLN, and spleen compared to their levels in uninfected tissues (Table 1), while the level of B, T, and dendritic cells remained relatively constant (Table 1). Two to four percent of the cells in the PP and 1–3% of the cells in the MLN and spleen were blue as determined by flow cytometry following infection with WT-HTEM (Fig 1A & Table S1). As observed with Marketon et al, a fraction of the tissue cell suspensions were not green (Fig 1A) indicating that they are dead because CCF2 is retained only in live cells (Marketon et al., 2005). In general, 10–30% of the neutrophils, macrophages and dendritic cells in the PP, MLN and spleen were blue, demonstrating that HTEM was readily translocated into these cell types (Fig 1B, D & F). In contrast, less than 5% of the B and T cells contained HTEM.

Table 1.

% Cell Type found in uninfected and infected tissues^a

	Log CFU	PMNsb	Macrophagec	Dendritic cellsd	B cellse	T helperf	Cytotoxic T cellsg
PP		0.38±0.09	0.86±0.14	1.17±0.23	48.5±16.2	11±1.4	7.00±4.24
Infected PP	5.6±0.3	4.63±3.01	7.0±0.6	1.49±0.76	38.7±11.4	7.8±2.5	4.40±3.17
MLN		0.13±0.02	0.69±0.11	1.38±0.82	27.2±3.3	18.00 h	12.96 h
Infected MLN	5.0±0.6	4.84±1.0	6.32±3.0	3.0±0.54	38.6±3.45	12.9±2.15	3.0±2.0
Spleen		0.54±0.16	1.05±0.18	1.92±0.78	34.1±2.7	19.5±5.3	9.36±2.42
Infected Spleen	4.8±0.4	2.0±.5	4.0±2.0	1.0±0.5	39.5±5.0	12.3±3.0	10.44±3.75

^aAll data were collected from 4 independent experiments unless indicated

^bDefined as GR1⁺CD11b⁺

^cDefined as CD11b⁺GR1⁻

^dDefined as CD11c⁺

^eDefined as B220⁺

^fDefined as CD4⁺

^gDefined as CD8⁺

^hNumbers are from 1 experiment

Fig. 1. Neutrophils, macrophages, and dendritic cells are preferentially translocated with YopH during Yptb infection.

Mice were orogastically infected with 2x10⁹ CFU of WT-HTEM. Day 5 post-infection tissues were harvested and filtered to generate a single cell suspension. Cells were incubated with CCF2-AM and labeled with antibodies to the indicated cell surface marker(s). Fluorescence intensity was analyzed by flow cytometry. (A) Detection of green and blue cells by flow cytometry in PP of uninfected mice or mice infected with WT or WT-HTEM. The blue⁺ cells are gated (R1) with the percentage of blue⁺ cells indicated in the gate. (B, D, F) The percentage of blue⁺ cells in the cell type indicated on the x-axis in (B) PP, (D) MLN, and (F) spleen. (C, E, G) The percentage of a cell type present in the entire organ (white bars) and the percentage of specific cell type present in the blue⁺ population (black bars) in the (C) PP, (E) MLN, and (G) spleen. The fold enrichment of each cell type in the blue⁺ population compared to total of that cell type in the whole organ is indicated above each set of bars. The experiment was repeated 4 times and the bars are the average + SEM from all the experiments. The asterisk indicates that there was a significant difference in the number of the indicated cell type in the blue⁺ population compared to that same cell type in the whole organ based on a unpaired, two tailed, t test (P<0.05). Note the Y-axis in (C) differs from that of (E) and (G).

We next determined whether Yptb targeted specific cell types more frequently than others. To do this, the percentage of each specific cell type in a tissue (Fig 1C, E and G, left bars) was compared to the percentage of that cell type in the HTEM-containing (blue cells) population (Fig 1C, E, & G, right bars). These percentages were compared to determine whether a type of cell was over or under-represented in the blue cell population compared to its representation in the tissue. Neutrophils were significantly enriched in the blue cell population of each organ (14-, 7-, and 8-fold in the PP, MLN, and spleen, respectively) compared their percentage in these organs. In addition, dendritic cells were significantly enriched in the blue cell population in the PP and spleen, while macrophages were significantly enriched in the MLN (Fig 1C, E, & G). In contrast to translocation of professional phagocytes, HTEM-positive B cells were significantly underrepresented in the PP and MLN and HTEM-positive CD4-T cells were significantly underrepresented in the PP (Fig 1C & E).

To determine whether the apparent enhanced targeting to professional phagocytes and the apparent reduced targeting to B and T cells was due to killing of B and T cells by Yptb, we determined whether any specific cell types were preferentially killed during Yptb infection. Suspended splenocytes were infected at an MOI of 20 for 1, 2, 4 or 18 hours, after infection cells were stained with fluorescent antibodies to specific cell type markers and propidium iodine to identify dead cells. No enhanced cell death was observed among any particular cell type between 1–4 hours indicating that Yptb did not kill any specific cell types during this time period (Fig S3). As expected (Bergsbaken and Cookson, 2007), 60–70% of the macrophages and neutrophils were dying 18 hours after infection. This data supports the idea that the selectivity of translocation towards professional phagocytes was not due to killing of other targeted cells. In fact, since Yptb causes death of macrophages in the MLN and spleen after infection (Bergsbaken and Cookson, 2007), our observed enrichment of targeting of macrophages could be an underestimate of the actual amount of Yptb targeting.

YopE is translocated into the same neutrophils as HTEM during infection

During infection of mice, it is unknown whether Yptb injects more than one Yop into the same cell or distributes Yops to different cells. We investigated whether other Yops were injected into the same cells that had been translocated with HTEM. Mice were infected orally with WT-HTEM and five days post-infection, spleens were harvested and filtered to generate a single-cell suspension. Splenocytes were incubated with CCF2-AM and labeled with α-Gr1. Three populations of GR1⁺ cells, blue^neg, blue^lo, and blue^hi were gated based on their levels of blue and green fluorescence and sorted by flow cytometry (Fig 2A). Equal numbers of Gr1⁺ cells from each population were assessed for both HTEM and YopE translocation by Western blot. As expected, HTEM was found in the blue cell population. YopE was also found in the blue cells, indicating that at least two Yops are translocated into the same cells during infection (Fig 2B). The YopE detected in the western is unlikely to come from lysed bacteria because the buffer used to lyse cells does not release YopE from bacteria (Fig S4) and ((Davis and Mecsas, 2007). In summary, these results demonstrate that both HTEM and YopE are translocated into the same neutrophils and suggest that other Yops may be delivered within the same cells during infection.

Fig. 2. HTEM and YopE are translocated into the same cells in the spleen during infection.

Four mice were infected orogastrically with WT-HTEM and five days post-infection, spleens were harvested, pooled, and labeled with CCF2-AM and antibody to GR1. (A) Gr1+ cells were sorted based on their blue/green fluorescence into blue^neg/green⁺ (R1), blue^low/green⁺ (R2), and blue^high/green⁺ (R3) populations. B. Western blot of 2x10⁵ GR1⁺ cells per gate. The lanes were loaded as follows, R1 blue^neg (uncleaved CCF2-AM); R2 blue^low (some CCF2-AM conversion); R3 blue^high. Blots were probed with antibody to TEM, YopE and actin, which was used as a loading control. The blot is representative of three independent experiments.

A comparable amount of HTEM and YopE were found in both the blue^hi and blue^lo cell populations. Since neutrophils, macrophages, and dendritic cells were significantly overrepresented in the blue^hi population (Fig 1C, E & G), the blue^lo cell population was analyzed to determine whether these cells were similarly enriched. Neutrophils, macrophages and dendritic cells were also overrepresented in the blue^lo population, while B and T cells remained underrepresented (Fig S5). Previous work has shown that Yptb can be found close to B and T cells in the MLN, suggesting that its close proximity to B and T cells does not result in Yop translocation (Balada-Llasat and Mecsas, 2006).

Yptb co-localizes with neutrophils, macrophages, B and T cells in lymph nodes

The strong preference for translocation of Yops into neutrophils, macrophages, and dendritic cells by Yptb could be due to several factors, including preferred co-localization of Yptb with these cells types. To investigate whether Yptb were localized predominantly in areas rich in professional phagocytes, immunohistochemical analyses of PP, MLN, and spleens from infected mice were performed. Tissues were stained with an α-Yptb antibody to detect Yptb and were counterstained with hematoxylin. The location of Yptb with respect to areas of inflammation, which included neutrophils, B and/or T cell rich areas, or boundary zones between germinal centers and areas of inflammatory was determined (Fig 3). Histological analysis of the PP indicates that the majority (over 60%) of Yptb microcolonies were found in areas of inflammation (Fig 3A–D, O), which included many neutrophils and macrophages as well as cellular debris. A significant minority (35%) of the microcolonies were found in B and T cell areas despite the observation that very few B and T cells were translocated with HTEM (Fig 1B–C). Likewise in the MLN, a majority of Yptb microcolonies were found either in areas of inflammation or boundary zones between germinal centers and areas of inflammation (Fig 3E–G, P); however, 30% of the microcolonies were found in B and T cell rich areas. In contrast, almost 90% of the microcolonies in the spleen were found in areas of inflammation and the boundary zones adjacent to areas of inflammation. Thus, in the spleen, the location of bacteria correlated more closely with the cell types that contain HTEM. The fact that fewer numbers of B and T cells were injected with HTEM (Fig 1) despite their close proximity to Yptb (Fig 3) suggests that Yptb was less proficient in translocating Yops into B and T cells than into professional phagocytes in tissue infection.

Fig. 3. Localization of Yptb in the PP, MLN and spleen.

Immunohistochemical analysis of (A–D) PP, (E–G) MLN, and (H–N) spleens after 5 days infection with 2x10⁹ CFU WT-HTEM. Sections were stained with α-Yptb and hemotoxylin to identify Yptb and distinguish B and T cells from areas of inflammation. The location of Yptb was detected by 3,3′-diaminobenzidine which produces brown staining (indicated by white arrows). Black arrows indicate areas of inflammation. (A, E, H) Uninfected tissues (10X); (B–C, F–G, I–J, L–M) areas of inflammation in infected tissues (10X and 60X); (C, G arrow 3, M) colonies detected in areas of infiltration (60X); (D, G arrow 1, N) colonies detected in B and T cell areas (60X); (G arrow 2, L) colonies detected at boundary zone (60X). (O–P) Quantification of microcolonies is areas of inflammation, B and T cell zones, and boundary zones in (O) PP, (P) MLN, and (Q) spleens from 5 mice; 1–2 sections/mouse were analyzed and scored by 3 independent viewers.

Neutrophil depletion and suppression of inflammation reduces the total amount of translocated Yops

To investigate the hypothesis that Yptb selectively translocates Yops into neutrophils rather than B and T cells in lymph tissues, we tested whether the number of cells targeted for Yop translocation changed when neutrophils were depleted. One day prior to infection with WT-HTEM and two days post-infection, mice were injected intraperitoneally with an isotype control antibody or the monoclonal antibody RB6-8C5, which binds to Gr1⁺ cells and causes their depletion. Three days post-infection, the levels of Gr1⁺ cells in the PP were reduced by 75–95% compared to the levels in mice treated with the isotype control antibody as ascertained by flow cytometry (Fig 4A–C). Strikingly, the percentage of total cells containing HTEM was also reduced by 80–90% in neutropenic mice compared to mice receiving the isotype control antibody (Fig 4C). The colonization levels were similar in both cohorts of mice (Fig 4D), indicating that the reduction in translocation was not due to a decrease in bacterial load in the RB6-8C5 treated mice. Further analysis of the specific cell types targeted after treatment with RB6-8C5 indicated that most of the blue cell population was comprised of the remaining neutrophils (data not shown). Only the PP were analyzed because the bacterial loads in spleens were not comparable as the R6B-8C5 treated mice had detectable bacteria while the control mice were not colonized. The R6B-8C5-treated and infected mice did not survive when the infection was allowed to proceed longer than 3 days hindering any analysis at day 5. In summary, when fewer neutrophils were present in the PP, fewer numbers of cells were targeted overall despite similar bacterial loads indicating that Yptb refrained from translocating Yops.

Fig. 4. Neutrophil depletion and suppression of inflammation reduces the total amount of translocated Yops.

(A–D) Mice were injected intraperitoneally with an isotype control (α-IgG2bκ) or the RB6-85C (α-GR1) rat monoclonal antibodies 1 day prior to and 2 days after oral infection with WT-HTEM. Three days post-infection, mice were sacrificed, the PP harvested and the number of blue⁺ and GR1⁺ cells were quantified. (A) Percent GR1+ cells in PP of an isotype control treated and an RB6-8C5 treated mouse by flow cytometry. The percentage of GR1+ cells (gated in R3) is indicated on the plots. (B) Percent blue⁺ cells from PP (gated in R1) of isotype control treated mice and RB6-8C5 treated mice. (C) The percentage of blue⁺ cells (x-axis) were plotted versus the percentage of GR1⁺ cells (y-axis) in the PP from mice treated with the isotype control antibody (squares) and mice treated with α-GR1 (triangles). The experiment was repeated twice and data from all the mice are shown. Significant differences between both the GR1+ cells and the number of blue⁺ cells from the two groups of mice were determined by an unpaired, two tailed t test. Both the percentage of GR1+ cells and the percentage of blue⁺ cells in each group of mice were statistically different (unpaired, two tailed, t test; * indicates P<0.01) (D) Bacterial colonization in PP of the isotope control or RB6-8C5-treated mice shown as CFU/gm PP. Each dot represents one mouse; the bar represents the geometric mean. There was no statistical difference in the bacterial load between the two populations of mice by an unpaired, two tailed, t test, P=0.54. (E–F) Mice were intragastically infected with 2x10⁹ YPIII-HTEM or YPIIIyopER144A-HTEM and the number of blue⁺ and Gr1⁺ cells were quantified by flow cytometry. 5 days post-infection PP were harvested. (E) The percentage of Gr1⁺CD11b⁺ cells and blue⁺ cells in the PP were plotted from mice infected with YPIII-HTEM (squares) or YPIIIyopER144A (triangles). The experiment was repeated three times and data from all mice are shown. Significant differences between the GR1+ and the blue⁺ cells between the two groups of mice were determined by an unpaired, two tailed t test. (F) Bacterial colonization in mice infected with YPIII or YPIIIyopER144A-HTEM shown as CFU/gm PP. There was no statistical difference in bacterial load by an unpaired two tailed t test P=0.59

To further test the hypothesis that Yptb selectively targets professional phagocytes during infection, we exploited our previous observation that infection with a ΔyopE mutant fails to recruit neutrophils to the PP at day 5 post-infection despite colonizing PP at levels equivalent to WT bacteria (Logsdon and Mecsas, 2006) (Fig 4F). This experiment allows us another means to functionally reduce neutrophil levels in Yptb-infected tissues. YPIII WT-HTEM or an Yptb strain expressing a catalytically inactive mutant of YopE and HTEM, YPIII yopER144A-HTEM, were used to infect mice. Five days post-infection, PP were harvested and assessed for the number of neutrophils, the number of blue cells, and the number of bacteria in the tissues (Fig 4E–F). Mice infected with YPIII yopER144A-HTEM had lower levels of neutrophils and fewer HTEM-containing cells than mice infected with YPIII WT-HTEM, but comparable bacterial loads. Together, this experiment and the neutrophil depletion experiments suggest that Yptb does not indiscriminately translocate Yops into whatever cells are present, but rather selectively translocates Yops into neutrophils.

Professional phagocytes in spleen cell suspensions are preferentially translocated by Yptb at low multiplicity of infection

We next analyzed whether the tissue microenvironment played a role in target specificity or whether there are inherent properties of professional phagocytes that permit Yptb to selectively translocate Yops into these cells. Spleens were harvested from uninfected mice and filtered to disrupt all tissue architecture and generate a single-cell suspension. Splenocytes were incubated with CCF2-AM, infected at either a high MOI (20:1) or a low MOI (1:1) (without spinning the bacteria with the cells), labeled with different antibodies, and then analyzed by flow cytometry. A high MOI was used to determine whether all cell types tested could be translocated with HTEM while a low MOI was used to determine whether specific cells were preferentially targeted for HTEM translocation when bacteria were limited. At a high MOI, all cell types tested were translocated with HTEM (Fig 5A), indicating that Yop translocation can occur in all splenocytes when the number of bacteria-host cell interactions were frequent, which is consistent with previous results (Marketon et al., 2005). Furthermore, no translocation specificity was observed as all cells were susceptible to translocation (Fig 5B). In contrast, at a low MOI, neutrophils, macrophages and dendritic cells were preferentially targeted by Yptb for HTEM translocation by 25-, 3-, and 5-fold, respectively, when compared to the total amount of these cell types in the spleen (Fig 5C–D). Furthermore, B cells were selectively excluded from translocation by Yptb. These results with splenocyte cell suspensions mirror those obtained during infection of mice (Fig 1C, E & G) and show that intrinsic properties of neutrophils, macrophages and dendritic cells can dictate selective targeting of Yop translocation into these cells. Furthermore, they support the idea that co-localization during tissue infection is insufficient for cell-type specific targeting.

Fig. 5. WT-HTEM preferentially targets macrophages, neutrophils and dendritic cells from splenocytes at a low, but not high MOI.

Splenocytes were infected with WT-HTEM (A, B) at a MOI of 20:1 or (C, D) at a MOI 1:1, incubated with CCF2-AM, and then distinguished by cell type by antibody and flow cytometry. The number of blue cells of a particular cell type was quantified and the number of each cell type in the spleen was quantified. (A, C) The percentage of blue cells (y axis) in the cell type indicated on the x-axis. (B, D) The percentage of each cell type present in the entire organ (white bars, y-axis on left) compared to the percentage of each cell type present in the blue⁺ population (black bars, y-axis on right). The fold enrichment of each cell type in the blue⁺ population when compared to the same cell type in the organ is denoted on top of each pair of bars. The experiment was repeated 4 times and the average + SEM are plotted. The asterisk indicates that there was a significant difference in the percentage of a specific cell type in the blue⁺ cells compared to the entire organ (unpaired, two tailed t test P<0.05).

Yptb binds preferentially to neutrophils and macrophages using YadA

Translocation of Yops is dependant upon Yptb binding to cells (Mejia et al., 2008). Preferential binding of Yptb to neutrophils, macrophages and dendritic cells could determine the specificity of translocation to these cells and so we tested whether Yptb preferentially bound to these cells. Yptb expressing GFP was incubated with splenocytes, and cells were then labeled with different fluorescent antibodies to identify the cell types associated with GFP-Yptb by flow imaging and flow cytometry (Fig 6A&B-white bars). To rule out the possibility that Yptb was associating with professional phagocytes because professional phagocytes chemotaxed towards and/or phagocytosed Yptb, cytochalasin D was added to the splenocytes to inhibit these activities. Yptb preferentially bound to neutrophils, macrophages and dendritic cells compared to B and T cells (Fig 6A&B, white bars) in the absence of chemotaxis and/or phagocytosis. To test whether more professional phagocytes associated with Yptb when their ability to chemotax and phagocytose bacteria was present, the numbers of different splenocytes that associated with Yptb was determined by flow cytometry (Fig 6B, black bars). An increase in binding of GFP-Yptb to neutrophils, macrophages and dendritic cells was observed in the absence of the inhibitor (Fig 6B, black bars), indicating that chemotaxis and/or phagocytosis enhanced the association of these cell types with Yptb. Nonetheless, even in the presence of cytochalasin D, Yptb preferentially attached to neutrophils, macrophages and dendritic cells.

Fig. 6. Yptb binds preferentially to professional phagocytes cells at low MOI.

(A) A single cell suspension of splenocytes was treated with 2 μM of cytochalasin D and then infected with WT IP2666 expressing GFP at a MOI of 1:1 for 1 hour. Splenocytes were labeled with antibodies to different cell markers and the binding of Yptb to specific cell types was analyzed by ImageStream. The experiment was done twice and the average number of a particular cell type associated with Yptb is shown. (B) A single cell suspension of splenocytes was treated with 2 μM of cytochalasin D (white bars) or left untreated (black bars) and then infected with WT IP2666 expressing GFP at a MOI of 1:1 for 1 hour. Splenocytes were labeled with antibodies to different cell markers and the binding of Yptb to specific cell types was analyzed by flow cytometry. The experiment was repeated 3 times and the average + SEM are graphed. The asterisk indicates significant differences between cytochalasin D treated and untreated cells as determined by t test (P<0.05). (C–D) Cells were treated with 2μM of cytochalasin D and infected with WT, Δinv, or ΔyadA expressing GFP at MOI 1:1 for 1 hour. (C) The percent cells bound to GFP-expressing bacteria was determined by fluorescence intensity in the FITC channel of the total splenocyte population. (D) The percentage of specific cell types in spleens bound by WT, Δinv and ΔyadA. The experiment was repeated 4 times and the average + SEM are graphed. The asterisk indicates significant differences in the association of the indicated cell type by WT versus ΔyadA infected cells as determined by t test (P<0.05).

We next analyzed the role of two bacterial adhesins in directing specific binding to neutrophils and macrophages. Binding by Invasin and YadA to cells containing β1-integrins, which include professional phagocytes, as well as B cells and T cells, facilitates Yop delivery (Mejia et al., 2008). The presence of YadA and Invasin in Yptb grown at 37°C was confirmed by western blot analysis (Fig S6). The binding of ΔyadA and Δinv expressing GFP to splenocytes was tested in the presence of cytochalasin D by flow cytometry. No differences in the total amount of binding to splenocytes were detected among WT, Δinv or ΔyadA (Fig 6C). However, when specific cell types were analyzed for their ability to bind to GFP-Δinv or GFP-ΔyadA (Fig 6D), the ΔyadA mutant bound significantly less to neutrophils than GFP-Yptb, while binding to B cells and dendritic cells remained unchanged (Fig. 6D). Since neutrophils are only approximately 0.54% of all splenocytes (Table 1), the 4-fold reduction in binding to neutrophils was not detectable in the total splenocytes (Fig 6C). In the absence of Invasin, no difference was observed in binding to any cell type analyzed (Fig 6D). These results demonstrate that YadA plays a critical role in selective association of Yptb with neutrophils whereas Invasin does not.

yadA and invasin mutants have reduced translocation into splenocytes compared to WT, but still target professional phagocytes for translocation

Yop translocation depends on both the binding of Yptb to host cells and activation of signal-transduction cascades within these cells, the latter of which is triggered by ligand-host cell receptor binding (Mejia et al., 2008). The ΔyadA mutant bound inefficiently to neutrophils (Fig. 6) and binding of invasin triggers signal transduction cascades that enhance translocation (Mejia et al., 2008). We tested whether ΔyadA-HTEM, Δinv-HTEM and/or ΔinvΔyadA-HTEM targeted fewer splenocytes for translocation using flow cytometry (Fig 7A–B). Fewer splenocytes contained HTEM when infected with ΔyopB-HTEM, Δinv-HTEM, ΔyadA-HTEM or ΔinvΔyadA-HTEM compared to WT-HTEM (Fig 7A–B and Fig S2A). These data suggest that the individual adhesins play an important role in facilitating interactions with splenocytes and increase the number of cells targeted for translocation.

Fig. 7. ΔyadA and Δinv translocate HTEM into fewer numbers of splenocytes.

Splenocytes were left uninfected or infected with WT-HTEM, ΔyopB-HTEM, Δinv-HTEM, or ΔyadA-HTEM at an MOI of 1:1, incubated with CCF2-AM and antibodies to distinguish particular cell types. (A) The percentage of blue⁺ cells was determined by flow cytometry. (B) The relative percent of blue+ cells by setting WT-HTEM to 100% and normalizing the percentage of blue cells of the HTEM mutant strain to WT-HTEM. Experiments with the Δinv-HTEM, ΔyadA-HTEM and ΔinvΔyadA-HTEM were repeated 9, 13 and 6 times, respectively. Differences were determined by using a paired t test with the WT-HTEM sample. (C) The percentage of blue⁺ cells in the indicated cell type after infection with Δinv-HTEM, ΔyadA-HTEM or ΔinvΔyadA-HTEM was compared to infection with WT-HTEM. (D) The percentage of a cell marker in the blue cell population after infection with Δinv-HTEM, ΔyadA-HTEM or Δinv ΔyadA-HTEM was compared to infection with WT-HTEM. (C–D) The bars represent the average + SEM of at least 8, 12 and 6 experiments for Δinv-HTEM, ΔyadA-HTEM and ΔinvΔyadA-HTEM analyzed with the indicated markers. Asterisk indicates significant differences between the number of blue cells in the WT-HTEM infected population versus the adhesin mutant population as determined by paired, t test (P<0.05).

We next assessed whether specific cell types were targeted less frequently by specific adhesin mutants or if all cell types were reduced for targeting after infection with the different adhesin mutants. The number of neutrophils, macrophages, dendritic cells and B cells targeted for translocation by the single adhesin mutants, Δinv-HTEM and ΔyadA-HTEM was significantly reduced compared to WT-HTEM (Fig 7C). While the translocation into these cells was not significantly reduced after infection with the ΔinvΔyadA double mutant, this is likely due to the observation that 2 out of 6 experiments had very high levels of translocation after infection with the Δinv ΔyadA while 4/6 had lower levels than WT. These results indicate that YadA and Invasin interact with a variety of different cell types to promote translocation. Furthermore, these results combined with our previous results (Fig 6C) indicate that YadA specifically promotes binding to neutrophils and plays an additional role to facilitate translocation when Yptb associates with other cells. When an MOI of 40 was used, all types of splenocytes were translocated with Yops; however again more professional phagocytes were targeted than B and T cells (Fig S7).

To determine whether the adhesion mutants targeted a different spectrum of cell types compared to WT, the percentage of each cell type in the translocated population was analyzed. This analysis should indicate whether either of these adhesins is critical for Yop translocation into a specific cell type. Surprisingly, neutrophils, macrophages and dendritic cells had enhanced translocation in the absence of either Invasin or YadA compared to WT (Fig 7D). This result indicates that while the overall numbers of professional phagocytes with translocated Yops was reduced (Fig 7B&C), Yptb still interacted preferentially with professional phagocytes compared to other cell types in splenocytes.

In these translocation assays, phagocytosis could not be prevented because disruption of actin polymerization also blocks translocation (Mejia et al., 2008) and our unpublished data. Thus, it is possible that translocation of HTEM from Yptb may be occurring from within the phagosomes of professional phagocytes after Yptb has been internalized. In fact, recent studies support the idea that some Yop translocation can occur after phagocytosis of Yptb (Zhang et al., 2008). To investigate this possibility, WT-HTEM was grown at 26°C or 37°C, conditions which promote or block invasion of phagocytes, respectively, and RAW264.7 macrophages were infected at an MOI of 10. At 26°C, Yptb expresses adhesins, but not the TTSS, so more bacteria are internalized whereas at 37°C both adhesins and TTSS are expressed, so the Yops are rapidly delivered to phagocytes preventing phagocytosis. After 20 minutes, gentamicin was added to kill extracellular bacteria. Infection was allowed to proceed for an additional hour to permit the internalized Yptb to potentially express the TTSS and translocate HTEM. Afterwards, one cohort of cells used to measure HTEM translocation and the second was analyzed to determine the numbers of internalized bacteria. Importantly, 10 fold more macrophages were blue than contained internalized Yptb after infection with Yptb grown at 37°C (Fig S2B). These results indicate that HTEM was most frequently translocated from extracellular bacteria when the bacteria were grown at 37°C. In contrast, more cells contained Yptb than were blue after infection with Yptb grown at 26°C. Although the presence of Yptb in the blue cells was not tested, these results are consistent with the idea that some internalized Yptb may translocate Yops.

ΔyadA and Δinv mutants colonize the Peyer’s patches poorly after oral infection, but Δinv translocate Yops into professional phagocytes

To determine whether YadA and/or Invasin are critical for targeting neutrophils or macrophages during infection, mice were infected orogastrically with WT-HTEM, ΔyadA-HTEM, Δinv-HTEM or ΔinvΔyadA-HTEM and the numbers of bacteria, blue cells, and neutrophils in the PP were determined at three and five days post-infection (Fig 8A–C). We found significantly fewer bacteria, fewer blue cells and fewer neutrophils in the PP of mice infected with ΔyadA-HTEM, Δinv-HTEM and ΔyadAΔinv-HTEM compared to WT-HTEM at both days. Therefore, it was impossible to conclude whether the reduced number of translocated neutrophils were due to the reduced bacterial load, the reduced numbers of neutrophils or an inability of the mutants to target these cells for translocation. However, fortuitously at day 5 post-infection, the PP of three mice infected with Δinv-HTEM had bacterial levels that were within the lower range of mice infected with WT-HTEM (Fig 8A). The number of blue cells and the types of cells targeted for Yop translocation in these PP were compared to three mice infected with WT-HTEM that had similar bacterial loads (Fig 8D–G). The 3 mice in each cohort had similar numbers of blue cells and similar numbers of neutrophils (Fig 8D). In addition, the number of neutrophils and macrophages translocated with Yops were comparable (Fig 8F & G). These preliminary results suggest that Invasin is not solely responsible for the targeted translocation of HTEM into professional phagocytes during infection of PP.

Fig. 8. YadA and Invasin are critical for colonization of Peyer’s patches.

Seven to nine week old female BALB/c mice were infected orally with 2x10⁹ CFU of WT-HTEM, ΔyadA-HTEM, Δinv-HTEM or ΔinvΔyadA-HTEM and sacrificed at 3 or 5 days post infection. Each dot represents one mouse; the bars represent the average. (A) At each time point PP were harvested, a cell suspension was generated and plated for CFU. (B) Cells were incubated with CCF2-AM and the percentage of blue cells in PP was determined by flow cytometry at each time point (C) Cells were labeled with α-Gr1-PeCy5 and α-Cd11b-PeCy7 to determine the percentage of neutrophils by flow cytometry. Each dot represents data from an individual mouse. Bars indicate the geometric mean in (A) and average in (B) and (C). The experiment was repeated three-five times and all the data was combined and analyzed using ANOVA with Tukey-Kamer multiple comparison. * indicates P value < 0.01 and ** indicates P value < 0.001. (D) The percentage of blue⁺ cells (x-axis) were plotted versus the percentage of GR1⁺CD11b⁺ cells (y-axis) in the PP from mice colonized with comparable numbers of either WT-HTEM (squares) or Δinv-HTEM (triangles). (E) Bacterial colonization in PP of the WT-HTEM and Δinv-HTEM mice used in panels D, F and G. Each dot represents one mouse; the bar represents the geometric mean. There were no significant differences. (F–G) The percentage of Gr1+CD11b+ (F) or GR1+CD11b− (G) cells in the blue cell population in the PP after infection with either WT or Δinv-HTEM. There were no significant differences.

Since it had previously been reported that a yadA mutant can reach the PP during early stages of infection but does not survive (Heise and Dersch, 2006; Marra and Isberg, 1997), mice were infected with WT-HTEM and ΔyadA-HTEM for 6 hours, 1 day or 2 days. The number of bacteria in the PP, the number of blue cells and the number of neutrophils were counted to determine whether at earlier time points comparable numbers of WT-HTEM and ΔyadA-HTEM were detected (Fig S8). Consistent with previous results (Heise and Dersch, 2006; Marra and Isberg, 1997), comparable bacterial loads were detected in the PP at 6 hours post-infection, but by day 1, most mice infected with WT-HTEM had higher bacterial loads than mice infected with ΔyadA-HTEM. Unfortunately, the number of blue cells at 6 hours were too low to evaluate whether the distribution of cells targeted by WT-HTEM versus ΔyadA-HTEM were different (Fig S8). These results demonstrate that YadA plays a critical role in the GI tract during infection, but whether YadA facilitates targeting of Yops to neutrophils or other cell during tissue infection cannot be evaluated due to low levels of overall colonization and translocation.

Discussion

After oral infection, pathogenic Yptb replicates primarily extracellularly in many organs and must counteract bactericidal actions of resident and incoming cells (Balada-Llasat and Mecsas, 2006; Bergman et al., 2009; Simonet et al., 1990). During infection of cultured cells, Yersinia translocates Yops into many different cell types including epithelial cells, macrophages, B cells, T cells, and dendritic cells (Brodsky and Medzhitov, 2008; Davis and Mecsas, 2007; Grosdent et al., 2002; Sory et al., 1995; Yao et al., 1999). Therefore, it seemed plausible that Yptb translocates Yops into all cells found in infected tissues, especially given that these cells express receptors capable of binding to Yptb adhesins (Eitel and Dersch, 2002; Grosdent et al., 2002; Leong et al., 1990; Pettersson et al., 1996). In contrast, here we demonstrate that after oral infection Yptb selectively targets neutrophils, and to a lesser extent macrophages and dendritic cells in the PP, MLN and spleen for Yop delivery. Moreover, in lymph nodes Yptb discriminates against B cells for translocation. Finally, in the absence of the preferred cellular targets, overall levels of Yop translocation were significantly reduced demonstrating that interactions between Yptb and specific cell types during infection must determine when and whether Yops are translocated.

Several lines of evidence indicate that Yptb specifically targets professional phagocytes because of inherent properties between these cells and Yptb rather than targeting phagocytes because phagocytes are their closest neighboring cells during tissue infection. First, in neutrophil-depleted mice or under tissue infection conditions where fewer neutrophils migrated to lymph nodes, Yptb targeted significantly fewer cells for translocation in the PP despite the fact that the bacterial load was similar. Consistent with this observation, mice lacking TNFR displayed increased levels of professional phagocytes as well as an increase in the level of total blue cells during infection with Ye (Koberle et al., 2009). Second, very few B and T cells were translocated with Yops in the lymph nodes, although Yptb co-localized with B and T cells in lymph nodes (Balada-Llasat and Mecsas, 2006). Finally, the marked preference for translocation of Yops into professional phagocytes and the discrimination against B cells was recapitulated when the tissue architecture was disrupted and splenic cell homogenates were infected at low MOI. In contrast at high MOI, all cell types in the spleen were targeted by Yptb, suggesting that all cell types are capable of being translocated with Yops. Together, these results show that the inherent properties of the interaction between Yptb with certain host cells cause preferential translocation into those cells during infection, rather than the specificity of Yop translocation being driven merely by proximity.

Several features of Yptb and professional phagocytes could result in the selective targeting of professional phagocytes by Yptb during infection and in splenocytes suspensions. These features could include binding, the ability to chemotax, activation state of the bound cell, activation of specific signal-transduction cascades, and/or plasma membrane domains that favor insertion of the translocon. Since translocation requires binding to cells (Mejia et al., 2008), the receptors on innate immune cells may recognize Yptb better than receptors on B and T cells. For instance, during the course of infection Yptb may become coated with complement or Fc and thus Yptb interactions with cells containing complement receptor and/or Fc receptor may be favored. Previous work with cultured cells has demonstrated that coating Yersinia with complement or Fc is sufficient to induce Yop translocation into cells with complement receptor or Fc receptor (Fallman et al., 1995; Grosdent et al., 2002). Alternatively, Yptb-specific ligands may dictate the bacteria’s association with neutrophils, macrophages and dendritic cells. In fact, our observation that ΔyadA binds to significantly fewer neutrophils from isolated splenocytes than WT Yptb supports the idea that some Yptb ligands can promote interactions with particular host cells. While analysis of the few mice that were colonized by a Δinv mutant indicated that the invasin mutant retained the ability to target neutrophils and macrophages as well as WT Yptb in the PP, Yptb may use YadA and/or other bacterial ligands to promote translocation to neutrophils, macrophages and/or dendritic cells during infection.

Several studies have examined the role of YadA in promoting Yptb dissemination from the GI tract to PP, MLN and spleen (Heise and Dersch, 2006; Marra and Isberg, 1997). In a ligated loop model of infection, a ΔyadA mutant rapidly penetrated the PPs; however, 48 hours post-oral inoculation ΔyadA was attenuated for growth in the PP (Heise and Dersch, 2006; Marra and Isberg, 1997). These results indicate that YadA is not needed for penetration and initial colonization of the PP, but YadA is required for survival within the PP. Our data and others is consistent with the idea that the ΔyadA mutant cannot target resident neutrophils and therefore is eliminated (Heise and Dersch, 2006; Marra and Isberg, 1997). Although invasin did not appear to influence binding to specific cell types, it is important to note that the IP2666 strain of Yptb expresses very little invasin compared to many other Yptb strains (Simonet and Falkow, 1992) and unpublished data. Thus a deletion in invasin in IP2666 may have less of an impact in cell binding and translocation than in other Yptb strains.

The activation state of the cells or the nature of the cell surface components may contribute to the efficiency of Yop translocation. Recently, Mejia et al showed that an increase in the activation of Src kinases after integrin stimulation correlated with an increase of Yop translocation in HeLa cells (Mejia et al., 2008). Perhaps cells that have recently migrated to tissues or have become activated are consequently more primed for translocation by Yptb. It remains to be determined whether specific subsets of dendritic cells, macrophages and T and B cells are targeted by Yptb. In fact, Ye targeted Yop predominantly to a subset of B cells, follicular B cells (Koberle et al., 2009). Another potential means of regulating translocation is that different cell membranes may be more or less conducive towards insertion of the translocon. Work in Shigella has indicated that translocon components are preferentially secreted in the presence of membranes rich with sphingomyelin and cholesterol and bind in regions rich in cholesterol (Epler et al., 2009; Hayward et al., 2005). While a similar phenotype has not been demonstrated with Yptb, it is possible that the translocon can more easily insert into the membranes of professional phagoctes than into B and T cells.

An intriguing question is whether or not Yop translocation occurs after phagocytosis by professional phagocytes during infection. Over 95% of Yptb is extracellular in the MLN and spleen after infection (Balada-Llasat and Mecsas, 2006; Bergman et al., 2009) and we showed that when Yptb were grown at 37°C, 10 fold more macrophages were detected that had translocated Yops than that contained internalized Yptb. Combined these results suggest that at least some macrophages in tissues are likely to be translocated with Yops from extracellular Yptb. However, we cannot exclude the possibility that during infection some Yptb is first internalized by professional phagocytes and then translocates HTEM or that some Yptb translocate Yops while they are being engulfed by phagocytes, with the result that a professional phagocyte has both internalized Yptb and translocated Yops. Distinguishing between these two possibilities during tissue infection is technically challenging. Nonetheless, we think it plausible that some fraction of professional phagocytes with translocated Yops may contain Yptb. Understanding the fate of these two populations will provide insights into whether internalized Yptb alters host defenses and enhances bacterial growth during infection.

Previous studies have examined the splenic cell types targeted by Y. pestis and Ye after intravenous infection (Koberle et al., 2009; Marketon et al., 2005). Interestingly, both the enteric Yersinia and, Y. pestis targeted professional phagocytes, despite the fact that they do not share many of the same adhesins. Specifically, Y. pestis lacks YadA and invasin, and expresses some unique adhesins and some which are shared among Yersinia spp (Felek and Krukonis, 2009; Forman et al., 2008). Given the similar cell tropism for translocation, it is possible that some of the shared adhesins function during infection to direct Yop translocation into professional phagocytes. Alternatively, these three species may have functionally redundant adhesins. Yptb, Y. pestis and Ye targeted all splenocyte suspensions when infected at high MOI but as no studies were done with a lower MOI with Y. pestis or Ye, it is unclear if this selectivity is apparent under conditions where bacteria are limiting (Koberle et al., 2009; Marketon et al., 2005). One notable difference between the enterics and Y. pestis was that B cells were not discriminated against translocation in the spleens by the enteric Yersinia spp, but were discriminated by Y. pestis. This difference could be due to the difference in species, the location of the bacteria, and/or rate cell death of specific cell types. An important difference between the two studies with the enteric Yersinia spp is that after oral infection with Ye not enough bacteria were found in the PP to detect blue cells (Koberle et al., 2009). In contrast, we found that PP had the highest levels of colonization and greatest number of blue cells after oral infection with Yptb. The molecular basis for this difference could be that the Yptb strains, IP2666 and YPIII, used in our studies might deliver Yops more efficiently into the targeted cells during infection. Supporting this idea is that observation that the level of colonization of the spleen after oral infection with Yptb was 100x less than that of Ye (Koberle et al., 2009) suggesting that Yops are delivered more efficiently in our model. Additional differences between our model and that with Ye were the strain of mice used (C57Bl/6 versus BALB/c) and that mice infected with Ye were also given the immunosuppressant, desferrioxamine (Koberle et al., 2009). The immunosuppressant may have altered the physiology of the cells such that they were less receptive to Yop delivery. Nonetheless, it is striking that despite the fact that the overall levels of bacteria were 100x higher in the Ye and Y. pestis infected mice, the cellular tropism for Yop delivery was similar among all species.

This study is the first to examine the different types of cells with translocated Yops in multiple tissues after oral infection. Interestingly, there were differences in the cell types that were both enriched and/or underrepresented among each tissue. For example, CD4 T cells were underrepresented for targeting in the PP, but were not underrepresented in the MLN and spleens while B cells were underrepresented in both the PP and MLN but not the spleen. On the other hand, dendritic cells were enriched for targeting in the PP and spleen, while macrophages were enriched for targeting in the MLN, but not the PP. As of yet, the reasons for these different tropisms is unclear, but further investigation into the subset of dendritic, macrophages, B and T cells might reveal that the differences are due to subsets of cells found in each tissue. For instance, the ability of specific subsets cells to chemotax towards bacteria in different tissues may explain, in part, both the tropism for professional phagocytes and the differences in the cells targeted in different tissues.

Using the TEM technology to identify intracellular niches and regulation of effector proteins, Salmonella was found to reside inside neutrophils after intraperitoneal infection (Geddes et al., 2007). Furthermore, several Salmonella effector proteins were translocated at different times in infection demonstrating that there was a hierarchy of effector protein translocation during infection. While YopH and YopE were both translocated into the same neutrophils in the spleen at day 5 post-infection, it remains to be determined whether all effector Yops are translocated continually throughout the course of infection and/or whether all Yops enter each individual cell.

In conclusion, despite the fact that Yptb readily translocates Yops into a variety of cell types in culture, Yptb demonstrates selective translocation into specific cell types during infection. Future experiments determining what cellular and bacterial factors are critical for the tropism to neutrophils, macrophages and dendritic cells, and how the Yops have altered the physiology of those cells during infection will be essential to understanding how Yersinia dismantles the immune response.

Experimental Procedures

Construction of strains and plasmids

Plasmids, strains and strain constructions are described in supplemental materials. The absence of adhesins, YadA and Inv, in the ΔyadA and Δinv strains were confirmed by Western blot as described in supplemental materials (Fig S6).

CCF2-AM conversion assays after murine infections

Mice were infected as previously described (Logsdon and Mecsas, 2003) with the following modifications. BALB/c mice (NCI) were infected intragastrically with 2x10⁹ CFU of WT-HTEM unless otherwise is indicated. Infections were allowed to proceed for 5 days, except in the experiments involving neutrophil depletion, where infections were allowed to proceed for 3 days.

To generate a single cell suspension from cells in the PP, MLN and/or spleen, these tissues were harvested aseptically into 5 ml HBSS containing MgCl₂ and CaCl₂ (Cellgro). All tissues were pressed through a 70μm cell strainer (Falcon) and 100ug/ml gentamicin was added to kill bacteria and preventing Yop translocation for all subsequent steps. Cells were transferred to a 15 ml tube and spun at 340xg for 5 minutes. Prior to straining the spleens, spleens (but not PP or MLN) were perfused with 80U/ml collagenase (Roche) and incubated for 30 minutes at 37°C to liberate dendritic cells and eliminate auto-fluorescence. Collagenase activity was halted by the addition of 2ml HBSS lacking MgCl₂ and CaCl₂ (Cellgro) and supplemented with 1mM EDTA. Cells from spleens were resuspended in 10 mL BD Pharm Lyse^™ solution (Pharmagen) to lyse erythrocytes and immediately spun at 340xg for 5 min. Cells from PP, MLN and spleens were washed twice with PBS and then MLN and spleens were resuspended in 3 ml RPMI+10% FBS and PP were resuspended in 1mL of RPMI+10% FBS.

To label single cell suspensions, the cells were incubated with 1μg/ml CCF2-AM compound (Invitrogen) for 2 hours at 30°C in the presence of 1.5mM probenecid (Sigma) and 100μg/ml gentamicin (Sigma). 100 μL of cells were aliquoted into a 96-well plate and blocked with 50μl of a 1:200 dilution of Mouse BD Fc Block^™ (BD) for 10 minutes at 4°C. Cells were incubated in 50μl of FACS buffer (PBS+1% FBS+0.02% NaAzide) containing fluorescent antibodies to GR-1-PE-Cy5 (eBiocience), CD11b-PE-Cy5 (eBiocience), CD11b-PE-Cy7 (eBiocience) CD11c-PECy5 (eBiocience), B220-PeCy5 (BD), CD4-PECy5 (BD) and/or CD8-PECy5 (BD) at dilutions of 1:75 for 30 minutes at 4°C. Samples were washed twice in FACS buffer, centrifuged at 340xg, resuspended in 100μl in FACS buffer and analyzed on an LSRII (Becton Dickson) FACS machine. 2X10⁵ cells were acquired per sample and data were analyzed using Summit v4.3 software. Cells from tissues that not were incubated with CCF2-AM and/or antibodies as well as cells from uninfected tissues or tissues infected with WT were used as negative controls

The Institutional Animal Care and Use Committee of Tufts University approved all animal procedures.

CCF2-AM conversion assays in splenocyte suspensions

Spleens from uninfected, 6–8 week old BALB/c mice were harvested in RPMI+10%FBS in a 6-well plate, treated with collagenase and a cell suspension was generated as described above. Bacteria were grown overnight in LB and diluted 1:40 in 2xYT supplemented with 20mM sodium oxalate and 20mM MgCl₂. The bacteria were grown with aeration at 26°C for 2 hours and then shifted to 37°C for 2 hours prior to infection of splenocytes. Cells were infected with WT-HTEM or mutant strains at the indicated MOI for 1hr at 37°C. Infected splenocytes were labeled with CCF2-AM and prepared for flow cytometry with antibodies as described above.

Yptb adherence assays to splenocyte suspensions

A 1mL aliquot splenocyte cell suspension from an uninfected spleen was infected with Yptb strains expressing GFP for an hour in a 24-well plate at 37°C with no spinning. Yptb expressing GFP were grown as described above and the media was supplemented with 10 μg/ml chloramphenicol. Cells were labeled with antibodies and analyzed on ImageStream (Amnis) with IDEAS analytical software or an LSRII (Becton Dickson) FACS with Summit v4.3 software as described above.

Immunohistochemistry

The location of Yptb was determined as described (Balada-Llasat and Mecsas, 2006) with the following modifications. PP, MLN, and spleen, from uninfected mice or mice infected intragastrically with 2×10⁹ Yptb were processed, embedded in paraffin, cut in 8 μm sections and stained as described (Balada-Llasat and Mecsas, 2006). Samples were scored blindly by at least two investigators using a Nikon Eclipse TE2000-U microscope.

YopE and HTEM Translocation

Splenocytes from infected mice were treated with CCF2-AM and labeled with GR1-PE-Cy5 antibody (eBiocience 15-5931-81). Cells were sorted in the MoFLo FACS sorter (Cytomation, Fort Collins, CO,) and 2x10⁵ blue^hi-GR1+, 2x10⁵ blue^lo-GR1+ and 2x10⁵ blue^neg-GR1+ cells were collected. Cells from each population were lysed with 50 μl of eukaryotic lysis buffer (10 mM HEPES pH 7.4, 150 mM NaCl, 5 mM EDTA, 0.1% NP-40, 1 mM phenylmethylsulfonyl fluoride, and 5 μg/ml each of aprotinin, leupeptin, and pepstatin) for 20 min with gentle rocking at 4°C and processed as described (Davis and Mecsas, 2007). These lysis conditions have been shown to lyse the plasma cell membrane, but not bacteria membranes as shown in Fig S4 and (Davis and Mecsas, 2007).

Granulocyte Depletion

Granulocyte depletion experiments were performed as described (Logsdon and Mecsas, 2006) with the following modifications. Mice were injected with the antibodies 1 day prior to and 2 days after oral inoculation with 8x10⁸ Yptb YopHTEM. Day 3 post infection PP were harvested and single cell suspension was generated. 10μL of the 5ml cell suspension was plated on L plates containing kanamycin (50 μg/mL) to determine the bacterial burden in the tissue. Cells were labeled with CCF2-AM and with GR-1-PE-Cy5 antibody as described above. Fluorescent intensities of labeled cells were detected by flow cytometry using BD LSR II System. Data were analyzed using Summit v4.3 software. The experiment was performed twice and data from both experiments are shown.