Yersinia Has a Tropism for B and T Cell Zones of Lymph Nodes That Is Independent of the Type III Secretion System

Pathogenic Yersinia have a pronounced tropism for lymphatic tissues and harbor a virulence plasmid that encodes a type III secretion system, pTTSS, that transports Yops into host cells. Yops are critical virulence factors that prevent phagocytosis by macrophages and neutrophils and Yersinia mutants lacking one or more Yops are defective for survival in lymphatic tissues, liver, and gastrointestinal tract. However, here we demonstrate that Y. pseudotuberculosis (Yptb) mutants lacking the pTTSS survived as well as or better than wild-type (WT) Yptb in the mesenteric lymph nodes (MLN). Infection with pTTSS mutants caused lymphadenitis with little necrosis, whereas infection with WT Yptb provoked lymphadenitis with multiple necrotic suppurative foci. Gentamicin protection assays and microscopic examination of the MLN revealed that pTTSS mutants resided extracellularly adjacent to B and T lymphocytes in the cortex and paracortex. WT Yptb was found extracellularly adjacent to neutrophils and macrophages in necrotic areas and adjacent to B and T lymphocytes in less-inflamed areas. To determine whether lymphocytes protected pTTSS mutants from phagocytic cells, Rag1−/− mice were infected with pTTSS mutants or WT Yptb. pTTSS mutants but not WT, were impaired for survival in MLN of Rag1−/− mice, suggesting that lymphocyte-rich regions constitute a protective niche for pTTSS mutants. Finally, we show that invasin and the chromosomally encoded TTSS were not required for Yptb survival in MLN. In summary, chromosomally encoded factors are sufficient for Yptb replication in the cortex and paracortex of MLN; the pTTSS enables Yersinia to survive within phagocyte-rich areas of lymph nodes, and spread to other tissues.


Introduction
One hallmark of Yersinia infections is colonization of the lymph nodes. After subcutaneous delivery of Y. pestis by the bite of an infected flea, Y. pestis usually migrates to draining lymph nodes and causes a fulminant infection leading to severe swelling of these organs, termed buboes, and then spreads systemically [1]. Y. pestis is thought to have evolved from the enteric pathogen, Y. pseudotuberculosis (Yptb), about 1,500-20,000 y ago [2]. However, Y. pestis acquired two additional plasmids important for its life cycle [3]. Despite the high degree of genetic similarity between the two species [2], infections with Yptb in humans occur after ingestion of contaminated food or water and are generally mild and selflimiting [1]. However, a frequent clinical manifestation of Yptb infections is mesenteric lymphadenitis [1,4]. Infection in humans with the more distantly related Y. enterocolitica (Ye) [2,5] generally produces gastrointestinal pathology as well as mesenteric lymphadenitis [1,4]. Thus, although enteric Yersinia species and Y. pestis use different means of transmission and cause different syndromes, all three species have a pronounced tropism for lymphoid tissues [1,4].
Our previous work demonstrated that Yptb strains lacking one or more Yops were deficient in colonization of the gastrointestinal (GI) tract, Peyer's patches (PP), and MLN [27]. In control studies, we tested the ability of strains lacking the entire pTTSS to colonize different tissues after intragastric inoculation of mice, expecting that these strains would be severely attenuated. Surprisingly, pTTSS mutants colonized the MLN as efficiently as or better than wild-type (WT), although they were defective in colonizing the GI tract, spleen, and liver. Further investigation into the behavior of WT and pTTSS mutants revealed that both were found predominantly in the cortex and paracortex of the MLN, and both remained extracellular regardless of the presence of the pTTSS. Our results indicate that chromosomally encoded factors are necessary and sufficient for Yptb to replicate within B and T lymphocyte zones of MLN.

Results
pTTSS Mutants Colonize the MLN and PP, but Not the GI Tract or Spleen To determine whether the pTTSS was required for colonization of various tissues after intragastric inoculation, several Yptb mutants with deletions in the operons encoding the pTTSS were created. Deletions of the yscBL operon, the yscNU operon, or both operons and the virGlcrF operon, designated yscBU, were generated. A strain lacking the pIB1 virulence plasmid was also tested to determine whether other plasmid-encoded genes were important for colonization. All four pTTSS mutants, yscBL, yscNU, yscBU, and pIB1 À were used to intragastrically inoculate BALB/c mice, and their colonization levels were compared to the isogenic WT strain, YPIIIpIB1. At 6 h, 2 d, or 5 d postinoculation, the lumen contents of the ileum, cecum, and ascending colon, the PP, cecal lymph node (CLN), MLN, and spleen were dissected, and the colony forming units per gram (cfu/g) tissue determined. At 6 h post-inoculation, no difference in colonization of any tissue was observed between the yscBU mutant and WT YPIIIpIB1 (Figure 1 A-1E, and unpublished data). At 2 d post-inoculation, the pTTSS mutants were severely attenuated for colonization throughout the GI tract, PP, CLN ( Figure 1F-1J, and unpublished data). In contrast, the level of colonization of the MLN was similar between WT, and three of the four TTSS mutants, yscBL, yscNU, and pIB1 À indicating that the TTSS was not required for colonization of MLN. At 5 d postinoculation, the defect in colonization of the pTTSS mutants was evident throughout the GI tract and spleen, whereas colonization levels of all pTTSS mutants in the MLN, PP, and CLN were similar to WT or in some cases higher than WT ( Figure  1K-1O and unpublished data).
To determine whether pTTSS mutants colonized other strains of mice efficiently, C57Bl6/J mice were infected with WT YPIIIpIB1, yscBU, or pIB1 À ( Figure 1K-1O). As observed in BALB/c mice, the pTTSS mutants colonized the PP and MLN at levels comparable to WT YPIII and were deficient in colonizing the ileum, cecum, and spleen. Therefore, the pTTSS or other elements encoded on the pIB1 plasmid are not required for Yptb survival and replication in the PP or MLN at 5 d post-inoculation in BALB/c or C57Bl6/J mice.
To confirm that the ability of Yptb to colonize the MLN in the absence of pTTSS was not due to a unique property of the YPIII strain, similar studies were performed using the Yptb strain, IP2666. IP2666, unlike YPIII, has a functional copy of PhoP which enables IP2666 Yptb to replicate intracellularly in bone marrow-derived macrophages [38]. Because IP2666 is slightly more virulent than YPIII during intragastric infections of mice (L. Logsdon and J. Mecsas, unpublished data), mice were sacrificed at day 4 post-inoculation ( Figure 1K-1O). Surprisingly, the IP2666 pTTSS mutants, yscNU, and pIB1 À , colonized the MLN at levels 10-fold higher than WT IP2666 ( Figure 1K-1O). As observed for YPIII, the IP2666 pTTSS mutants were recovered in significantly fewer numbers in the ileum and spleen; however, about 10-fold more IP2666 pTTSS than YPIII pTTSS were recovered in the spleen, consistent with the increased virulence of the IP2666 strain ( Figure 1O).
Since other Gram-negative bacteria reside in the lymphoid tissues for long periods of time [39], and commensal bacteria remain alive in dendritic cells in the MLN for several days [40], a mouse-commensal Escherichia coli strain isolated from mouse feces (see Materials and Methods) was used to intragastrically

Synopsis
The pathogenic bacteria, Yersinia, synthesize an apparatus called a type III secretion system, which transports bacterial proteins, Yops, from the bacteria into important immune cells, such as macrophages and neutrophils. Normally, macrophages and neutrophils control bacterial infections by ingesting the bacteria; however, the Yops inactivate these immune cells, which in turn, enable Yersinia to replicate extracellularly and cause disease in many types of tissues. Pathogenic Yersinia are frequently found in lymph nodes of infected hosts, and the Yops are important for the bacteria to replicate and cause disease in lymph nodes since Yersinia mutants that lack Yops do not colonize lymph nodes efficiently. Surprisingly, the authors found that Yersinia pseudotuberculosis lacking the type III secretion system colonizes the mesenteric lymph nodes and survives extracellularly next to lymphocytes. However, in mice lacking lymphocytes, the type III secretion mutants did not survive although wild-type Yersinia did. The authors' findings reveal that other bacterial factors are sufficient for mesenteric lymph node (MLN) colonization of Yersinia and that lymphocytes provide a protective niche for Yersinia strains lacking the type III secretion system. Potentially, these avirulent mutant strains, which persist for at least 5 d in the mesenteric lymph nodes, could be used as live attenuated vaccines to protect against Yersinia infections, or as carriers of other antigens.
infect BALB/c mice. At 6 h post-inoculation, E. coli colonized the GI tract and lymphatic tissues at levels comparable to or higher than Yptb ( Figure 1A-1E). However, by 2 d postinoculation, E. coli were undetectable in the MLN and were almost cleared from the PP and the CLN ( Figure 1F-1J and unpublished data). These results indicate that the ability of Yptb to survive in the MLN for up to 5 d is not a general property of all Gram-negative enteric bacteria. In conclusion, Yptb colonization of the MLN occurs in the absence of the pTTSS or other pIB1-encoded genes, demonstrating that chromosomally encoded factors are sufficient for MLN colonization.

pTTSS Mutants Cause a Milder Inflammatory Response in the MLN Than WT Yersinia
To investigate the host response to infection by WT and pTTSS mutants in the MLN, infected MLN were analyzed by histology. The IP2666 strain was selected for this experiment, since the IP2666 strain was more virulent than the YPIII strain. BALB/c mice were inoculated intragastrically with 2 3 10 9 WT IP2666 or the pIB1 À mutant, and the MLN were harvested 4 d post-inoculation and prepared for histology. Gross anatomical analysis of the MLN showed that both strains caused enlargement of the MLN, i.e., hyperplasia, compared to uninfected MLN (Table 1 and Figure 2). Histological analysis of uninfected MLN showed normal follicular lymphocytic areas (Figure 2A) adjacent to a compact paracortex ( Figure 2C and 2D) with no evidence of necrotic lesions or lymphoid hyperplasia (Figure 2A-2D). In most mice infected with WT IP2666, one or more MLN had multiple necrotic foci located at the edge of the MLN and close to the germinal centers (Table 1 and Figure 2E) or in the paracortex ( Table 1). Areas of necrosis contained bacteria  Data are from 4-12 mice from at least two different experiments. All data was combined for each strain, tissue, and time point. Each square represents the log cfu/g tissue from one mouse; open squares indicate that less than 10 cfu were recovered per tissue, and bars represent the geometric mean. Asterisks (*) and black circles () indicate statistically significant differences between the WT and the mutants, calculated by the t test (*p , 0.01) or by Mann-Whitney (U , 0.05), respectively. DOI: 10.1371/journal.ppat.0020086.g001 ( Figure 2G, arrow), neutrophils, eosinophilic amorphous material, and many apoptotic nuclei ( Figure 2H). These findings contrasted notably with MLN from mice infected with the pIB1 À mutant, which had lymphoid hyperplasia and occasionally a necrotic cell but no evidence of acute necrosis (Table 1 and Figure 2I-2L). Although WT-infected MLN presented with neutrophil infiltration primarily around the necrotic foci, the pIB1 À -infected MLN generally presented with diffused infiltration of neutrophils in the paracortex. Despite the fact that IP2666 pTTSS mutants were recovered at 10 times the level of WT IP2666 ( Figure 1N), no extensive necrosis was observed, suggesting that one or more components of the pTTSS trigger an acute inflammatory response leading to necrosis, whereas pTTSS mutants persist, causing milder inflammatory response.

The pIB1 À Strain Localizes in Lymphocyte-Rich Areas
Immunohistochemistry studies were undertaken to determine the microenvironment of WT and pTTSS mutants in the MLN. WT IP2666 microcolonies were found in necrotic areas (Figure 3), in lymphocyte-rich areas of the cortex and paracortex that had no evidence of acute necrosis although apoptotic nuclei were often observed close to the bacteria ( Figure 3G-3I), and in the medulla (unpublished data). pIB1 À microcolonies were found in lymphocyte areas of the cortex and paracortex, but not in the medulla ( Figure 3J-3L and unpublished data). In the cases in which infiltration of phagocytes were observed, pIB1 À mutants were not detected in these areas (unpublished data), unlike in necrotic areas containing WT bacteria.
The cortex and paracortex are primarily composed of lymphocytes [41,42]. In order to confirm the cell types adjacent to Yptb, immunofluorescence studies were performed. Serial MLN sections from infected BALB/c mice were stained with anti-Yptb and either a mixture of antibodies that recognize lymphocytes (CD4, CD8, and B220) ( Figure 4A-4F) or an antibody that recognized macrophages and neutrophils (CD11b) ( Figure 4G-4L). Yptb pTTSS mutants were found in lymphocyte-rich areas whereas WT Yptb was occasionally observed adjacent to CD11b þ cells ( Figure 4G-4L). To determine whether Yptb had a preference for either B or T cells, sections were stained with anti-CD4 and anti-CD8 or anti-B220 and antibody against Yptb. Both WT and pTTSS mutant strains were found in close proximity to both T and B lymphocytes (unpublished data), indicating no overt preference for either type of lymphocyte.
pTTSS Mutants Are Deficient in Colonization of Lymphoid Tissues in Rag1 À/À Mice Since pTTSS mutants survive in B and T lymphocyte areas of the MLN, we hypothesized that B and T cell-rich areas might confer a protective niche from neutrophils and macrophages. To determine whether the absence of B and T cells rendered the IP2666 pIB1 À less capable of surviving in the MLN, BALB/c or congenic Rag1 À/À mice were infected with WT IP2666 or IP2666 pIB1 À . Rag1 À/À mice lack mature B and T lymphocytes, although small numbers of immature lymphocytes are present [43]. Four days post-inoculation, the MLN and ileum colonization levels were compared in the BALB/c and Rag1 À/À mice ( Figure 5). Notably, the size and weight of the MLN in Rag1 À/À mice varied depending on whether mice had been infected with WT IP2666 or pIB1 À strains ( Figure 5A). Rag1 À/À mice infected with pTTSS mutants had larger MLN than Rag1 À/À MLN infected with WT Yptb (Figure 5A), consistent with the idea that WT IP2666 can inhibit the pro-inflammatory response [22,23,25]. Therefore recovery of bacteria in the MLN was calculated both as cfu/g and cfu/MLN ( Figure 5B). Total colonization levels in the MLN were similar in WT IP2666-infected BALB/c and Rag1 À/À mice ( Figure 5B). In contrast, colonization of the MLN by IP2666 pIB1 À in Rag1 À/À mice was 16-fold lower than colonization levels in BALB/c mice ( Figure 5B). Levels of the IP2666 pIB1 À in the ileum of BALB/c and Rag1 À/À mice showed no differences ( Figure 5C), indicating that the lack of recovery in the MLN of the pTTSS mutant in Rag1 À/À mice was not secondary to a reduced ability to colonize the GI tract in Rag1 À/À mice.
Since Rag1 À/À mice lack PP, it was conceivable that the pIB1 À mutants colonize the MLN poorly because they are unable to reach the MLN due to the lack of PP, rather than to survive in the MLN. Thus, Rag1 À/À mice were infected intraperitoneally (i.p.) to bypass any requirement for trafficking from the GI tract through the PP to the MLN. Unexpectedly, after intraperitoneal inoculation with either WT or pTTSS mutants, the MLN were not detectable by eye Three to five MLN per mouse were analyzed. b Defined by tissue enlargement: 0 (not detected), 1 (mild), 2 (moderate), and 3 (severe). c Defined by neutrophil infiltration: 0 (not detected), 1 (mild), 2 (moderate), and 3 (severe). d Areas of neutrophil infiltration or necrosis: C, cortex; P, paracortex. e Defined by cellular debris: 0 (not present), 1 (mild), 2 (moderate), and 3 (severe). f Statistically significant difference between WT and pIB1 À in the MLN displaying signs of necrosis, calculated by Mann-Whitney, U , 0.05. DOI: 10.1371/journal.ppat.0020086.t001 in Rag1 À/À mice. Our inability to detect MLN could have been a result of their initial small size and lack of inflammatory response after i.p. inoculation, or alternatively, the MLN in the Rag1 À/À mice may have been destroyed by a robust immune response to infection prior to dissection.

pTTSS Mutants Localize Extracellularly in the MLN
Since it is well documented that the pTTSS allows Yersinia spp. to remain extracellular while docked to phagocytic cells [9], we investigated whether the pTTSS mutants were extracellular or intracellular in the MLN using an ''ex vivo'' gentamicin protection assay (see Materials and Methods). BALB/c mice were infected intragastrically with WT YPIII, WT IP2666, or pTTSS mutants, and 4-or 5-d post-inoculation cell suspensions of MLN were treated with gentamicin ( Figure 6A). Although the differences were not significant, more pTTSS mutants were protected from gentamicin than the isogenic WT strains. However, only 1% of the YPIII yscBL and 3.4% of the IP2666 pIB1 À mutant were protected, indicating that over 96% of the pTTSS mutants were extracellular. The observation that more WT IP2666 were protected from gentamicin than WT YPIII may reflect the ability of the IP2666 strain to survive and replicate within macrophages [38]. As a positive control, nearly 15 % of Salmonella typhimurium were intracellular in the MLN at 4 d post-inoculation. Overall, these results indicate that the majority of Yptb remain extracellular in the MLN, even in the absence of a functional pTTSS.
The pTTSS is important for Yptb to inhibit phagocytosis by macrophages, neutrophils, and epithelial cells in cell culture assays [9]; thus, it was surprising that the vast majority of the pTTSS mutants remained localized extracellularly. To confirm that the pTTSS mutants were phagocytosed more efficiently than WT, gentamicin protection assays were performed using the murine macrophage-like cell line, RAW264.7. As expected, higher level of uptake by the macrophages of the pTTSS mutants was observed compared to WT YPIII or WT IP2666 ( Figure 6B). WT IP2666 was phagocytosed more efficiently than YPIII, an observation consistent with our ''ex vivo'' gentamicin protection assays and previous results by others [38]. Yptb pTTSS mutants efficiently invade some types of non-professional phagocytes, such as HEp-2 cells, by binding to a4or a5-b1 integrins using the bacterial protein, invasin [44,45]. Since the areas of the MLN infected by pTTSS mutants are composed primarily of lymphocytes [42], and lymphocytes express a4and a5-b1 integrins [46,47], gentamicin protection assays were performed on two human T cell lines, SUP-T1 and H9, and primary B and T lymphocytes isolated from MLN after incubation with WT and pTTSS mutants. The pTTSS mutants were not efficiently internalized by either T cell lines or by primary B and T lymphocytes isolated from MLN ( Figure 6B and unpublished data). These results are consistent with the ''ex vivo'' gentamicin protection assays, indicating that pTTSS mutants are extracellular in lymphocyte rich areas.

Invasin and the cTTSS Are Not Necessary for Survival in the MLN
Since chromosomal factors must play a role in Yptb survival in the MLN, two chromosomally encoded factors, invasin and cTTSS, were tested for growth in the MLN when the pTTSS was not present. Invasin-deficient strains (inv À ) are less competent than WT in colonizing the PP following intragastric inoculation [30,48], indicating that invasin is crucial for Yptb to reach and/or survive in the PP. To evaluate the role of invasin in a pIB1 À background, BALB/c mice were infected intragastri-cally and colonization levels were assessed 5 d post-inoculation. Notably, a pIB1 À inv À strain was 100-fold more deficient in the colonization of the ileum compared to pIB1 À , indicating that invasin plays a role in bacterial survival in the ileum that is independent of pTTSS ( Figure 7A). In the PP and MLN, the pIB1 À inv À strain was more deficient than pIB1 À in colonization ( Figure 7A); however, the lower levels in lymph tissues could reflect the lower levels of bacteria in the ileum. In order to bypass the requirement for invasin in the ileum, mice were infected intraperitoneally. No difference was observed between the pIB1 À and the pIB1 À inv À strains in MLN colonization after i.p. inoculation, indicating that although invasin is required for reaching the MLN after oral inoculation, it is not required for Yptb pIB1 À survival in MLN ( Figure 7B).
To determine whether the cTTSS is important for MLN colonization, BALB/c mice were intragastrically infected with pIB1 À and pIB1 À cTTSS mutants. No differences were observed in the colonization of the lumen, PP, MLN, or spleen between pIB1 À and pIB1 À cTTSS at 5 d post-inoculation ( Figure 7C). Thus, the cTTSS is not required for MLN colonization in the absence of pIB1.

Discussion
A prominent clinical feature of infection by pathogenic Yersinia is lymphadenitis, as large buboes [1] and mesenteric lymphadenitis [4] are frequent symptoms associated with infection by Y. pestis and enteric Yersinia pathogens, respectively. Little is known about the factors involved in Yersinia survival in the lymph nodes, although deletion of several Yops reduces the ability of Yersinia to colonize PP and MLN, which suggests that these Yops are required for reaching and/or replicating in lymph nodes [27,31]. Here, we show that Yptb replicates extracellularly in the cortex and paracortex of the MLN in the absence of pTTSS and Yops, indicating that the ability to reach and replicate in lymphocyte-rich areas is driven by chromosomally encoded factors. WT Yptb infection Figure 5. B and T Lymphocytes Are Important for pIB1 À Colonization in the MLN BALB/c or Rag1 À/À mice were intragastrically inoculated with 2 3 10 9 WT IP2666 or IP2666 pIB1 À . (A) Size and weight of MLN from BALB/c or Rag1 À/À mice that were either uninfected or infected with WT IP2666 or IP2666 pIB1 À were measured. (B) and (C) Colonization of the MLN or luminal content of the ileum 4 d post-intragastric inoculation of BALB/c or isogenic Rag1 À/À mice was determined. Each square indicates the cfu from one mouse calculated as log cfu/MLN (B) or the log cfu/g of luminal content of the ileum (C); bars represent the geometric mean. Each experiment was performed with two to three mice and repeated three times. Asterisks (*) and black circles () indicate statistically significant differences between the number of pIB1 À recovered from BALB/c mice versus Rag1 À/À , calculated by the t test (*p , 0.05) or by Mann-Whitney (U , 0.05), respectively. DOI: 10.1371/journal.ppat.0020086.g005 provoked necrotic and suppurative foci whereas pTTSS mutants induced mild lymphadenitis, demonstrating that the presence of pTTSS triggers a more acute host inflammatory response. Furthermore, B and T lymphocytes appear to provide a protective niche for pTTSS mutants since their absence results in significantly lower colonization levels of the pTTSS mutants.
Strikingly, a partial complement of Yops is more detrimental to the survival of Yptb in the MLN than the complete absence of pTTSS [24,27]. Strains lacking two or more Yops colonize the MLN very inefficiently [27], and strains lacking only yopH, yopJ, or yopE have a 5-to 10-fold defect in MLN colonization [24,27]. One explanation for these observations is based on the finding that the translocation apparatus damaged cultured cells and induced elevated levels of IL-8, which were decreased by YopH, YopE, or YopJ [25]. By analogy, during infection with WT or yop mutants, components of the pTTSS may damage lymphocytes and trigger pro-inflammatory signals in the MLN, so that in the absence of one or more Yops, the bacteria are killed by recruited macrophages and neutrophils. pTTSS mutants may survive because they do not damage host cells and therefore do not induce the same pro-inflammatory response. In addition, pTTSS mutants might replicate faster during infection than strains expressing the pTTSS, as strains lacking pTTSS have a modest growth advantage compared to WT Yptb in culture media (unpublished data). Nonetheless, the pTTSS mutants clearly have a growth advantage over mouse commensal E. coli in the MLN, indicating that Yptb chromosomal factors enable the bacteria to persist and replicate in the MLN. With this in mind, it is conceivable that a common ancestor of Yersinia may have been a commensal with low pathogenic capability but with the ability to persist in the MLN or host environments similar to lymph nodes of mammals. Acquisition of the pTTSS would have enabled the non-pathogenic species to spread to other tissue environments of mammals and survive within the GI tract, enhancing its transmissibility.
Although the Yptb pTTSS mutants are rapidly internalized by cells that are generally non-phagocytic, such as epithelial cells [30], pTTSS mutants establish an extracellular niche surrounded by B and T lymphocytes in lymphoid tissues. Non-phagocytic cells expressing a4or a5-b1 integrins internalize pTTSS mutants in an invasin-dependent manner at frequencies often 100-fold greater than WT Yptb [30,49]. Although T lymphocytes express a4and a5-b1 integrins [46,47], pTTSS mutants were not internalized significantly more than WT Yptb during infection, and did not invade T or B lymphocytes isolated from MLN. Although lymphocytes are not generally considered intracellular hosts for bacterial pathogens, S. enterica is found within late endosomallysosomal compartments in B lymphocytes of the spleen after intragastric inoculation [50]. Combined, these results indicate that the cellular mechanisms exploited by Yptb to prompt its invasion into non-phagocytic cells are not as active in B and T lymphocytes. In addition, invasin may not be expressed in the MLN [51].
It is well established that the pTTSS is an essential virulence organelle that enables Yptb to remain extracellular in the presence of phagocytes through the anti-phagocytic action of Yops [9,21,52]. WT Yptb, but not the pTTSS mutants, was occasionally found associated with macrophages and neutrophils, suggesting that our inability to detect pTTSS (B) BALB/c mice were i.p. inoculated with 2 3 10 6 YPIII pIB1 or YPIII pIB1 À inv À , and their colonization levels in the MLN and spleen were determined 3 d post-inoculation. (C) BALB/c mice were inoculated intragastrically with 2 3 10 8 YPIII pIB1 À or YPIII pIB1 À cTTSS, and colonization levels of the ileum, PP, MLN, and spleen were determined. Data are from 6-11 mice from at least two different experiments. Each square represents log 10 cfu/g tissue recovered from one mouse. Bars represent the geometric mean. Open squares indicate than less than 10 cfu were recovered per tissue. Asterisks (*) and black circles () indicate statistically significant differences between the pIB1 À and either pIB1 À inv À or pIB1 À cTTSS strains, calculated by the t test (*p , 0.01) or by Mann-Whitney (U , 0.05), respectively. DOI: 10.1371/journal.ppat.0020086.g007 mutants adjacent to CD11b þ cells may be due to rapid internalization and killing of these mutants by phagocytic cells. Furthermore, in Rag1 À/À mice, WT Yptb had a selective advantage compared to the pTTSS mutants, suggesting that pTTSS mutants are compromised for survival in lymph nodes that lack B and T lymphocytes. In the absence of lymphocytes, the protective niche may be lost and the bacteria may be more exposed to phagocytes in the MLN [53]. In the PP, Yptb first encounters dendritic cells, neutrophils, and macrophages immediately after transcytosing the M cells [48]. In contrast to the MLN, the pTTSS mutants colonize the PP poorly at day 2, suggesting that many pTTSS mutants are eliminated by these cells prior to reaching the follicular area rich in B and T cells.
The inability of pTTSS mutants to survive well in Rag1 À/À mice compared to the isogenic BALB/c mice suggests that lymphocytes provide a protective niche for pTTSS. However, it is possible that the lack of PP hindered the pTTSS in reaching the MLN. If this were true, then the lack of PP was specifically detrimental to the pTTSS mutants, since the WT Yptb was found in the MLN of Rag1 À/À mice. The latter result indicated that WT Yptb does not require PP for successful dissemination to the MLN. We attempted to investigate whether the low levels of pTTSS mutants observed in the MLN of Rag1 À/À mice were replicating in the MLN or were the result of continual seeding of the MLN from the GI tract. To eliminate pTTSS mutants in the GI tract and therefore prevent reseeding, mice were given streptomycin orally as described in [54]; however, in control experiments we found that streptomycin delivered orogastrically killed Yptb colonizing the MLN after i.p. delivery, but not Yptb colonizing the spleen, which is consistent with results reported about colonization of the spleen in [54]. Thus, this experimental approach is unfeasible. Future experiments are directed at determining where the pTTSS mutants are in the MLN of Rag1 À/À mice, and the immune response to infection with WT and pTTSS in lymph nodes.
Common host-driven pathways indiscriminately traffic bacteria to lymph nodes, in part by dendritic cells binding to bacteria in the intestinal lumen and then migrating to the MLN [40]. Consistent with these observations, we found nonpathogenic E. coli in the MLN at 6 h post-inoculation. Also, others have reported Yptb and Salmonella in the blood as early as 30 min after intragastric inoculation [40,55]. It is noteworthy that at 6 h post-inoculation E. coli colonizes the lumen at much higher levels than Yptb, but their levels in the MLN are comparable to Yptb. Thus, Yptb likely use several tactics to rapidly reach the MLN, including invasin-dependent [30,48,56] and host-driven mechanisms [40,55], whereas E. coli either fails to efficiently reach the MLN or is rapidly eliminated by innate immune defenses in the MLN.
The key to the persistence of Yptb pTTSS mutants in the cortex and paracortex may lie in adherence factors that facilitate the interaction of Yersinia with lymphocytes or the underlying extracellular matrix in these areas. We have ruled out the requirement of invasin, YadA, and potential cTTSSsecreted factors in MLN colonization. Other candidate adhesion factors include the pH6 antigen [57] and glycoproteins. In other organisms, glycoproteins play a key role in conferring cellular tropism [58,59], and intriguingly, in some cases, they have been shown to be sufficient to induce a state of proliferative unresponsiveness in lymphocytes [58]. One such example is the MV glycoprotein complex of measles virus, which interferes with the propagation of the IL-2 receptor signal by blocking activation [58]. One could envision that the presence of specific glycoproteins or specific sugars on the LPS determines Yptb tropism and/or moderates inflammatory cytokines produced by B and T cells. In addition, it is tempting to speculate that preferential interaction with certain subsets of lymphocytes, such as CD4 þ CD25 þ T regulatory cells may further aid Yptb colonization. In fact, the presence of CD4 þ CD25 þ T lymphocytes has been shown to favor the establishment of Helicobacter pylori in the gastric mucosa by moderating inflammation [60].
Despite the key roles that the pTTSS and Yops play in disease [27], there is a report of a serotype 4a Yptb pIB1 À strain isolated from a 10-y-old girl diagnosed with acute mesenteric lymphadenitis [61]. In agreement with our data, characterization of that strain in mice as well as another serotype 4a pIB1 À strain demonstrated that serotype 4a strains lacking the pIB1 plasmid colonized the MLN at levels comparable to WT Yptb for up to 7 d [61]. Interestingly, the fate of Ye pTTSS mutants in the MLN seems to be dependent on Ye serotype. Ye biotype 1A pYV À strains have been isolated from symptomatic patients [62]. On the other hand, a number of experiments with other Ye biotypes, Ye O3 pYV À , Ye O9 pYV À , and Ye8081 pYV À , indicate that these strains were eliminated from the MLN of rabbits, gnonobiotic piglets, or mice within 1 to 3 d ( [63][64][65] and unpublished data). Taken together, these results indicate that both enteric Yersinia spp. reach the MLN in the absence of pTTSS, but only a subset of strains flourishes for longer periods, and suggest that Yptb has specific chromosomally encoded factors that enhance its survival. A comparative analysis of the chromosomal genomes of Yersinia strains that persist versus those that fail to persist in the MLN may reveal factors that enable replication in lymphoid tissues.
Numerous pathogens, including Mycobacterium spp., Salmonella spp., Brucella spp., viruses, and parasites such as Leishmania spp., are capable of producing lymphadenitis [39,[66][67][68][69]. These pathogens use a variety of mechanisms to survive in the MLN. For example, S. typhimurium survives intracellularly within macrophages in the MLN, where it can be isolated for up to a year post-inoculation [39]. Brucella suis survives inside macrophages by inhibiting apoptosis [70], and Leishmania spp. survive inside granulomas in the MLN [69]. We have shown that Yptb replicates in B and T lymphocytes areas in a pTTSS-independent manner. The presence of the pTTSS provokes many necrotic lesions, and allows Yptb to survive in phagocyte-rich areas cells and invade other organs. Future directions are aimed at identifying both the bacterial and host factors that permit replication of Yptb in lymphocyte rich zones of the MLN to understand the mechanisms driving the lymphotropism of Yersinia.

Materials and Methods
Bacterial strains. Two virulent serotype III Yptb strains were used: YPIII pIB1 [27], which has recently been shown to have a mutation in phoP [38], and IP2666 pIBI [38]. These strains are described as WT YPIII and WT IP2666. The YPIII pIBI À kan and IP2666 pIB1 À were generated by growing WT YPIII kan [27], or WT IP2666 on Luria Agar supplemented with 20 mM of sodium oxalate, 20 mM MgCl 2 , and 0.05 % Congo Red Dye. Loss of a functional pTTSS results in white colonies. The absence of several genes located on different regions of the plasmid was confirmed by PCR, indicating that the whole plasmid was lost. The invasin-deficient strain JM494 [30] was cured of pIB1 as PLoS Pathogens | www.plospathogens.org September 2006 | Volume 2 | Issue 9 | e86 described above. The mouse commensal E. coli strain was isolated from the feces of BALB/c mice (Taconic, Hudson, New York, United States). This strain was transformed with either pBR322 or pBR322expressing invasin, generating JMB44 or JMB45, respectively. Expression of invasin in this E. coli strain increased the uptake by 100fold in gentamicin protection assays using HeLa cells (unpublished data); however, no difference in colonization of the GI tract, PP, or MLN was observed when either E. coli strains were used to infect mice ( Figure 1 and unpublished data). S. typhimurium, SL1344, was used [39]. Additional mutant strains generated in this study were constructed in WT YPIII or WT IP2666 by allelic exchange using the suicide vector pCVD442, and methods described in [27]. Isogenic pTTSS mutants were constructed by deleting the regions between yscB to yscL, yscN to yscU, and yscB to yscU. The primer sequences used to generate the appropriate suicide plasmids were as follows: yscBL deletion, yscB1-GTGTGAGTCGACGCGGTGGCGCAA GAAAC, yscB2-GACAGTGCATGCAGTTGGCCAACGCTTG TTGCATTAG; yscL3-CCAACTGCATGCACTGTCCAGGAGCTATTCGTGAAC; and yscL4-CTGTCAGAGCTCGTCGCTATAATTGTCTCTAC; yscNU deletion, yscN1-GTGTGAGTCGACCTACTCCCTGAGATGAAC, yscN2-GA-CAGTGCATGC AGTTGGCCAAGTGGAATAAGTAATGC, yscU3-CCAACTGCATGCACTGTCTCA TCAGTGGTGGTAGCT, and yscU4-CTGTCAGAGCTCCCAATAGCCGGTGTTAATC; and yscBU deletion, yscB1, yscB2 and yscU1, yscU2. Colonies were screened for their ability to bind to Congo Red. Deletions were confirmed by PCR and then Southern blot.
The cTTSS deletion was constructed by deleting the putative structural genes spiA-ssaV. Primers used to generate the suicide plasmid were: spiA1-GTGTGAGTC GACCTGTGCCATAAAAGC-GATCC; spiA2-GACAGTGCATGCAGTTGGTAG CGGAATATCGC; ssaV3-CCAACTGCATGCACTGTCCACATTTTGCCGTT; and ssaV4-CTGTCAGAGCTCCCTGTGCAATGCGGATAG. Deletions were confirmed by PCR and then Southern blot. In addition, the pTTSS mutants were grown at 26 8C and 37 8C, using Yop-inducing, i.e., calcium-depleted. medium [71], or non-inducing conditions, i.e., 23YT supplemented with 5 mM CaCl 2 . At 26 8C, the growth of the mutants was comparable to WT. At 37 8C, the pTTSS mutants had a growth advantage compared to WT in Yop-inducing conditions, but they grew similarly when grown in non-inducing conditions. Yops were not secreted when Yptb were grown at 37 8C under Yop-inducing conditions [27].
The Institutional Animal Care and Use Committee of Tufts University approved all animal procedures.
Histology and immunohistochemistry. MLN from uninfected mice or mice infected intragastrically with 2 3 10 9 Yptb were dissected, fixed in formalin, embedded in paraffin, and cut in 10-lm sections. Sections were treated with xylene to remove paraffin prior to staining with hematoxylin-eosin or staining for immunohistochemistry. Lauren Richey, DVM, PhD, Diplomat ACVP, and members of our lab scored MLN histology sections blindly.
Immunofluorescence. MLN from uninfected mice or mice infected intragastrically with 2 3 10 9 with Yptb were harvested and snap frozen in Optimal Cutting Temperature Compound (Tissue-Tek, Fort Washington, Pennsylvania, United States). The 10-lm sections were cut, fixed with cold 100% acetone, and stored at À80 8C. For staining, sections were washed with PBS and blocked with 2% BSA for 30 min. Antibodies used were rat anti-mouse CD8 (BD Biosciences, San Jose, California, United States) and rat anti-mouse CD4 (BD Biosciences) for T lymphocytes, rat anti-mouse B220 (BD Biosciences) for B lymphocytes, rat anti-mouse CD11b (BD Biosciences) for neutrophils and macrophages, and rabbit anti-Yptb. Primary antibodies were diluted 1:200 in 2% BSA and incubated with the samples for 3 h at room temperature. Slides were washed three times with PBS-Tw for 15 min per wash and then incubated for 1 h with FITC conjugated to donkey anti-rat (Jackson ImmunoResearch, West Grove, Pennsylvania, United States) and Texas-Red-conjugated donkey anti-rabbit antibodies (Jackson ImmunoResearch). Slides were washed three times with PBS-Tw for 15 min per wash, and a cover slip was placed over the tissue before visualization using 203, 403 and 1003 objectives with a Nikon Eclipse TE2000-U microscope with fluorescence. Pictures were taken using a Roper Scientific camera (Trenton, New Jersey, United States) controlled by Open Lab software.
In vitro gentamicin protection assay. The macrophage cell line RAW264.7 (ATCC# CRL-2278), the T cell line SUP-T1 (ATCC# CRL-1942), and the T cell line H9 (ATCC# HTB-176) were grown as recommended by ATCC (Manassas, Virginia, United States). Primary B and T lymphocytes were isolated from MLN using enrichment columns (Stem cells; R&D System, Minneapolis, Minnesota, United States). Yptb strains were grown under Yop-inducing conditions [71]. A total of 5 3 10 4 or 2 3 10 5 cells were infected at a multiplicity of infection (MOI) of 10:1 for 30 min at 37 8C in the presence of 5% CO 2 . Gentamicin was added to a final concentration of 100 lg/ml for 90 min, and then cells were washed three times with PBS. T and B cells were centrifuged at 200 g for 5 min between each wash. Cells were lysed with 100 ll of 1% Triton-X-100 for 5 min followed by addition of 900-ll LB, and lysates were diluted and plated to determine the number of intracellular bacteria. Results are presented as percent uptake equal to 100 times the recovered cfu divided by the input cfu.
Ex vivo gentamicin protection assay. MLN were dissected from 7-8-wk female BALB/c mice infected with IP2666 or S. typhimurium. MLN were placed in 1-ml RPMI on ice and then pressed through a 70-lm filter (BD Labware, Palo Alto, California, United States) with the rubber end of a 1-ml syringe plunger. The sample was divided into two 0.5-ml aliquots: one aliquot was treated with 100-lg/ml gentamicin for 2 h at 37 8C to kill extracellular bacteria, and the other aliquot was untreated for 2 h at 37 8C to calculate the total cfu recovered. Cells were washed three times with PBS and lysed with 100 ll of 1% Triton-X-100 for 5 min followed by addition of 900-ll LB, and lysates were diluted and plated to determine the number of intracellular bacteria. Results are presented as number of gentamicin resistant bacteria (recovered after 2 h incubation) divided by the input bacteria times 100.