Monocyte and neutrophil recruitment during oral Salmonella infection is driven by MyD88-derived chemokines



Oral Salmonella infection recruits phagocytes to Peyer's patches (PP) and MLN. The chemokines induced in infected PP and MLN, the cellular sources during infection and the TLR signaling pathways involved in vivo are not known. Here, we show that CCL2, CXCL9 and CXCL2 mRNA are up-regulated in PP and MLN coincident with the first arrival of monocytes and neutrophils. Laser capture microdissection microscopy revealed that chemokine mRNA up-regulation was differently distributed in PP. Despite this, recruited monocytes and neutrophils formed inflammatory cell clusters throughout PP. Monocytes and neutrophils purified from infected mice preferentially produced CXCL2 and small amounts of CCL2, and neutrophils from infected mice migrated towards CXCL2 and CCL3. Furthermore, phagocyte recruitment to PP and MLN was intact in mice lacking TLR4 alone and when signaling through TLR4 and TLR5 was simultaneously absent; however, recruitment was compromised in MyD88−/− and more so in MyD88−/−TLR4−/− double knockout mice. Phagocyte release into the blood, however, was only marginally reduced in MyD88−/−TLR4−/− mice. Defective phagocyte recruitment to PP and MLN of MyD88−/−TLR4−/− mice was paralleled by low chemokine induction. These data provide insight into the chemokines and TLR signaling pathways that orchestrate the early phagocyte response to oral Salmonella infection.


Monocytes and neutrophils are critical in the first line of defense against bacteria. They develop in the bone marrow, are released into the circulation and enter tissues in response to infection. Murine monocytes in the blood are a heterogeneous population and have been divided into two main subsets based on whether they are present in steady state or recruited under inflammatory conditions 1. Inflammatory Gr-1hiCCR2hiCX3CR1low monocytes, which have also been called TipDC upon entering a tissue, are a source of molecules important in controlling bacterial infection, particularly iNOS and TNF-α 2–4.

Salmonella enterica serovar Typhimurium (S. typhimurium) is a Gram-negative facultative intracellular bacterium that infects via the oral route. Orally acquired Salmonella exits the gut lumen primarily by crossing the follicle-associated epithelium overlying Peyer's patches (PP) via M cells (reviewed in 5). The bacteria are then further spread to the MLN, most likely by DC, and to the spleen and liver via the blood 5, 6. After oral infection with S. typhimurium, neutrophils and Gr-1hiCCR2hi inflammatory monocytes accumulate in PP and MLN, the first organs encountering the bacteria, beginning 2–3 days after infection 3.

Chemokines are the main mediators involved in the recruitment and migration of leukocytes to and within tissues. An array of chemokines is induced by inflammation and recruits monocytes, neutrophils and other cells to sites of infection. The chemokine receptors CCR2 and CXCR2 are required for monocytes and neutrophils, respectively, to egress from the bone marrow into the blood during steady state and infection 7–10. They are also central for monocyte and neutrophil recruitment and host survival during infections 9, 11–16.

Despite the undisputed role of chemokine receptors and their ligands in ensuring survival to infection, it is difficult to assess the role of individual chemokines in phagocyte recruitment, particularly in the complex environment of bacterial-infected tissues. For instance, different infection models can induce distinct chemokine profiles, most chemokine receptors are bound by multiple ligands and chemokine receptor/ligand pairs have overlapping roles 17, 18. Moreover, although chemokine-deficient models have shown increased bacterial burden during Salmonella infection, 19–21, it is not known which chemokines are induced in the intestinal lymphoid tissue during oral Salmonella infection or where and by what cells the chemokines are produced. Finally, although impaired recruitment of inflammatory cells in mice deficient in TLR pathways has been observed in some infection models 22, 23, the role of the different TLR and the adaptor MyD88 in cell recruitment during oral Salmonella infection is not known.

Thus, the aims of this study were to use an oral Salmonella infection model to: (i) investigate chemokine induction and the cellular sources in the PP and MLN when phagocyte influx is first observed; (ii) localize chemokine production in infected PP using laser capture microdissection microscopy; (iii) assess distribution of phagocytes recruited to infected PP by fluorescence microscopy and (iv) establish the relationship between TLR, chemokine production and phagocyte recruitment. The data provide insight into the chemokine milieu that orchestrates the early recruitment of phagocytes to combat oral Salmonella infection.


CCL2, CXCL2 and CXCL9 are rapidly up-regulated in PP and MLN of mice orally infected with Salmonella

In kinetic studies, we previously showed that monocytes and neutrophils first increased in PP and MLN 2–3 days after oral Salmonella infection 3. To gain insight into the mechanism of this early recruitment, we assessed which chemokines are induced in PP and MLN at two time points. The first is day 2, which coincides with the arrival of the first phagocytes, and the second is day 4 when cells are rapidly accumulating 3. Thus, the expression of nine chemokines reported to be involved in recruiting myeloid lineage cells was assessed in PP and MLN (Fig. 1A and B). CCL2, CXCL2 and CXCL9 mRNA were up-regulated in both organs already on day 2 post infection and increased at day 4, although CXCL2 mRNA was only modestly increased in the MLN. CCL3 and CCL4 mRNA were slightly increased at day 2 but highly up-regulated at day 4. CCL6, CCL9, CCL20 and CX3CL1 mRNA remained low in PP while these three CCLs but not CX3CL1 mRNA were up-regulated somewhat (3–7-fold) in the MLN at day 4.

Figure 1.

Chemokine expression in PP and MLN during early oral Salmonella infection. (A and B) C57BL/6 mice were orally infected and 2 or 4 days post infection, PP and MLN were removed, total RNA was isolated and cDNA was prepared. Expression of the indicated chemokines was analyzed by real-time RT-PCR. hprt was used as a reference gene and a calibrator was used to compensate for inter-run differences. From these data, a normalized ratio was calculated for every sample. Results are presented as fold up-regulation of chemokines from the normalized ratio of infected mice compared with the normalized ratio of naïve mice±SEM. Data are from two independent experiments with a total of five to eight mice/group. (C and D) Two days post infection, blood was collected to extract serum and the PP and MLN were removed and lysed. PP and MLN lysates and serum were assayed by ELISA. Circles indicate the mean value of duplicate (CCL2, CXCL2) or single (CXCL9) wells of individual mice and the horizontal bar is the mean of all samples. Data are from two to four independent experiments with a total of eight to 14 infected and four naïve mice. *p≤0.05 and **p≤0.005, for differences between infected and naïve mice. n.s., not significant.

We further investigated CCL2, CXCL9 and CXCL2 for two reasons. First, the arrival of the first myeloid cells to PP and MLN (3 and data not shown) coincides with increased expression of CCL2, CXCL9 and CXCL2 at day 2 post infection. Second, monocytes and neutrophils in the blood of Salmonella-infected mice express CCR2 and CXCR2, which are the receptors for CCL2 and CXCL2, respectively 3, and a small subset of monocytes express CXCR3, the receptor for CXCL9 24. Consistent with the results at the mRNA level (Fig. 1A and B), CCL2 and CXCL2 protein was detected in both PP and MLN at day 2 post infection compared with the very low or undetectable level in naïve mice (Fig 1C). The protein level of CXCL9 in MLN and blood showed a tendency to increase, although it was not statistically significant compared with naïve mice, while no increase of CXCL9 in PP was apparent in infected mice (Fig. 1C and D).

The follicle-associated epithelium and subepithelial dome of infected PP produce early chemokines

We next investigated which cells in PP were expressing CCL2, CXCL2, and CXCL9 2 days after infection, (prior to a large influx of leukocytes). The follicle-associated epithelium (FAE), which overlies PP, the villus-associated epithelium and the subepithelial dome (SED) are the primary sites in contact with Salmonella upon bacterial traversal from the intestinal lumen into PP 5. We thus used laser capture microdissection microscopy to isolate the FAE (Fig. 2A, before (left) and after (right) excision) as well as the SED, and villus epithelium from PP of infected mice to assess chemokine expression by RT-PCR. The B-cell follicle was used for comparison. To confirm specific excision of the FAE and SED, CCL20, which is produced by the FAE but not by the SED or villus epithelium 25, was used as a control.

Figure 2.

Differential production of chemokines by the FAE and SED of PP. C57BL/6 mice were orally infected and 2 days later, PP were collected and cryosections were prepared and stained with hematoxylin. The FAE, SED, B-cell follicle and villus epithelium (Vill epith) were microdissected and total RNA was extracted and transcribed into cDNA. (A) Before (left) and after (right) laser capture microdissection of the FAE. (B) Expression of the indicated chemokines in microdissected samples was analyzed by real-time RT-PCR. For every chemokine, a normalized ratio was calculated using hprt as a reference gene and a calibrator to compensate for inter-run differences. Data are presented as fold up-regulation of chemokines from the normalized ratio of the indicated area compared with the normalized ratio of the B follicle (indicated by the dotted line) ±SEM. Data are from four infected mice except for the SED where two infected mice were examined.

CXCL2 was expressed eight- and fivefold higher in the FAE and SED, respectively, than in the B-cell follicle (Fig. 2B). CCL2 was not increased in the FAE or SED compared with the B-cell follicle while CXCL9 was expressed only somewhat greater by the SED than the B-cell follicle or FAE. As expected 25, CCL20 was expressed >19-fold higher in the FAE than the other sites. Only low expression of the chemokines examined was detected in the villus epithelium (Fig. 2B).

Monocytes and neutrophils form inflammatory foci in PP

The expression pattern of CXCL2 in particular (Fig. 2) prompted us to investigate if chemokine production correlated with different localizations of monocytes and neutrophils in PP. There is no single marker available to distinguish monocytes from other cells 3, 26, 27. However, iNOS is expressed by about 30% of the monocytes recruited to infected PP while other myeloid cells are not iNOS+3. We thus used an anti-iNOS antibody to identify monocytes in situ. Ly6G, which is not expressed by inflammatory monocytes 3, 28, was used to identify neutrophils.

As predicted, no iNOS-positive monocytes or Ly6G-positive neutrophils, were detected in PP from naïve mice, and the tissue structure was smooth with a well-defined FAE (Fig. 3A). At day 2 post infection, some infiltration of neutrophils occurred in the interfollicular region but not elsewhere (Fig. 3C), and CD11c+ cells were present in the SED just below the FAE (Fig. 3E). At day 4 post infection, however, most PP were swollen, had a disrupted architecture and B-cell follicles were dispersed with other cells being close to the edge of the follicle distal to the SED (Fig. 3D and F). A large increase of iNOS+ monocytes and Ly6G+ neutrophils was apparent and these cells formed inflammatory foci scattered throughout the PP (Fig. 3D). Thus, despite the expression pattern of chemokines at day 2 post infection in PP (Fig. 2), recruited cells were only detected in the interfollicular region at this time. At day 4, monocytes and neutrophils form inflammatory cell clusters throughout the tissue.

Figure 3.

Monocytes and neutrophils form clusters in PP of infected mice. C57BL/6 mice were orally infected with Salmonella, PP were removed at days 2 and 4 post infection, and sections were prepared from frozen tissue. Anti-iNOS was used to identify monocytes and anti-Ly6G was used to identify neutrophils. (A, C and D) Sections were stained with Ab to iNOS (red) and Ly6G (green). (B) Sections were stained with isotype-matched control, rabbit polyclonal IgG (red) and rat IgG2 (green). (E and F) Sections from the same PP as (C) and (D) were stained with Ab to CD11c (red) and B220 (green). All panels are representative of two to four experiments with a total of 5–12 mice in each group using a total of three to six PP from each mouse. Isotype-matched control Ab were always included and were negative. Arrows indicate the FAE. The original magnification is 10×. Scale bar=200 μm.

Monocytes and neutrophils produce CCL2 and CXCL2 and neutrophils migrate towards CXCL2 and CCL3

In addition to the early production of chemokines in PP at day 2 post infection, which comes from resident cells including the intestinal epithelial layer (Fig. 2) and the earliest, albeit very few, recruited myeloid cells 3, the monocytes and/or neutrophils recruited beyond day 2 might produce chemokines and amplify their recruitment as the infection progresses. To determine the capacity of recruited monocytes and neutrophils to produce chemokines, Ly6Chi monocytes and Ly6Ghi neutrophils were sorted from naïve blood and blood, PP and MLN 4–6 days after oral infection and cultured ex vivo with or without heat-killed Salmonella. Monocytes and neutrophils recovered from blood at day 4 post infection produced low amounts of CCL2, and bacterial restimulation had little effect on the production (Fig. 4A), while both cell types produced CXCL2 after ex vivo co-culture with bacteria (Fig. 4A). Monocytes (Fig. 4B, left) and neutrophils (Fig. 4B, right) from the blood and MLN, and neutrophils from PP, of orally infected mice produced more CXCL2 than the respective cell population from the blood of naïve mice. Moreover, all cells except neutrophils from PP released more CXCL2 after restimulation with heat-killed Salmonella (Fig. 4B). However, little if any CCL2 or CXCL9 production by monocytes or neutrophils from the PP, MLN or blood was detected and co-culture with bacteria did not induce production (data not shown). Finally, CCL3 and CCL4 mRNA were also up-regulated by neutrophils, and to a lesser degree by monocytes, sorted from the MLN of infected mice compared with the respective cells in the blood of both naive and infected mice (data not shown).

Figure 4.

Monocytes and neutrophils recruited to infected tissues produce CXCL2 and neutrophils from infected mice migrate towards CXCL2 and CCL3 ex vivo. Mice were orally infected and 4–6 days later, cells from the blood (A–E) or PP/MLN (B) of five to ten infected or naïve mice were pooled and CD11b+ cells were enriched by MACS. (A and B) Cells were stained with 7AAD, CD11c, MHC-II, Ly6C, Ly6G and a cocktail of anti- NK1.1, TCR-α/β and CD19 conjugated to the same fluorochrome and sorted into Ly6ChiLy6Glow monocytes and Ly6CintLy6Ghi neutrophils. (A) A total of 2×105 CD11b+ cells from the blood of naïve mice or 2×105 sorted cells were incubated in medium supplemented with GM-CSF alone or with GM-CSF+heat-killed Salmonella (+Sal). Supernatant was collected at 24 h (1st experiment) or 36 h (2nd experiment) and CCL2 and CXCL2 was measured by ELISA. Results are expressed as mean±SEM of two different samples, each having duplicate wells. Results are representative of two independent experiments. (B) A total of 40 000 sorted cells were incubated in medium alone or in medium+heat-killed Salmonella and the supernatant was collected at 24 h. CXCL2 was measured by ELISA. Monocytes and neutrophils obtained from blood of naïve mice were compared with the respective cell type from the indicated organs of infected mice since both cell populations are absent/very few in PP and MLN of naïve mice. Results are expressed as mean±SD of two to four different samples obtained from two to three independent experiments except restimulated neutrophils from PP, which is one experiment. *p≤0.05. n.s., not significant. (C–E) CD11b+ cells were resuspended in chemotaxis buffer, rested and then allowed to migrate towards the indicated concentration of chemokine for 90 min in a transwell assay. Assays were performed in duplicate for each concentration of chemokine. Migrated cells were quantified by flow cytometry separately for each well. Cells from duplicate wells were then pooled and stained as in (A). The mean percent migrated cells in duplicate wells was calculated as the number of migrated cells minus background migration divided by the original number of cells. (C) The mean percent neutrophil migration towards the indicated concentrations of CXCL2 is shown. Data are representative of three independent experiments. (C–E) The mean percent migration of the indicated cell population towards CCL3 (D) and CCL4 (E) is shown. Data are obtained from three independent experiments with the optimized concentration of 20–200 ng/mL. Error bars represent the SD.

To address whether monocytes and neutrophils that accumulate during infection were attracted by the chemokines they produce, we examined their migration towards CXCL2, CCL2, CCL3 and CCL4 directly ex vivo. Neutrophils from the blood of Salmonella-infected mice migrated towards CXCL2 ex vivo (Fig. 4C). In addition, despite that neutrophils from MLN/PP of infected mice down-regulate the CXCL2 receptor, CXCR2 3, these cells also migrated towards CXCL2, albeit to a lesser extent than blood-derived cells (Fig. 4C). Neutrophils from the blood, but only few from PP/MLN, of infected mice migrated towards CCL3 (Fig. 4D and data not shown). In contrast, little if any migration of monocytes from the blood of infected mice towards CCL3, CCL2 or CXCL2 was apparent (Fig. 4D and data not shown). Finally, little if any migration of either blood neutrophils or monocytes towards CCL4 was observed (Fig. 4E), despite that control wells showed migration of blood/spleen-derived NK1.1+ cells towards this chemokine (data not shown). Thus, monocyte and neutrophil production of chemokines during oral Salmonella infection could contribute to further attract more inflammatory cells and facilitate the formation of inflammatory foci in PP and MLN.

Unimpeded phagocyte recruitment in the absence of TLR4, TLR5 and TLR4/5 but not MyD88

Given the importance of TLR in initiating innate immune defense against invading bacteria, we reasoned that a compromised inflammatory response in TLR-deficient mice impairs phagocyte recruitment. Due to their enhanced susceptibility to Salmonella infection and the slight delay in chemokine production in the liver of i.p.-injected mice 29–31, we first examined cell recruitment in TLR4−/− mice. However, no significant difference in accumulation of CD11b+Ly6Chi monocytes or CD11b+Ly6Ghi neutrophils (gated as in (Fig. 5A)) was apparent in orally infected TLR4−/− mice in the blood, PP or MLN at day 2 (data not shown) or day 4 post infection (Fig. 5B). This occurred despite higher bacterial burden in TLR4−/− mice (Fig. 5B). In addition, myeloid cells accumulated normally in mice lacking TLR6 (data not shown), which forms a heterodimer with TLR2 to recognize bacterial lipoproteins. Next, we tested if recruitment of phagocytes was impaired when triggering TLR5 alone was abrogated (by using flagella-deficient (ΔfljBfliC) Salmonella) as well as when both TLR4 and TLR5 engagement were simultaneously defective (by infecting TLR4−/− mice with ΔfljBfliC Salmonella). However, neither the absence of TLR5 engagement alone nor the simultaneous absence of TLR4 and TLR5 signaling altered monocyte or neutrophil recruitment (data not shown).

Figure 5.

MyD88 and MyD88/TLR4 deficiency impair monocyte and neutrophil recruitment to PP and MLN but not mobilization into the blood during Salmonella infection. Cells from the blood, PP and MLN of naïve mice or mice orally infected 4 days earlier were stained with 7AAD and anti- CD11b, Ly6G, Ly6C, CD11c, a cocktail of antibodies against NK1.1, TCR-αβ and CD19 and were analyzed by six color flow cytometry (see also 3). (A) Viable (7AAD) CD11b+ cells negative for TCR-αβ, NK1.1 and CD19 were further gated into Ly6Ghi neutrophils (neut) and Ly6GlowLy6Chi monocytes (mo). (B) Monocyte and neutrophil accumulation in the blood, PP and MLN of infected WT or TLR4−/− mice. (C) Monocyte and neutrophil accumulation in the blood, PP and MLN of infected WT or MyD88−/− mice. (D) Monocyte and neutrophil accumulation in the blood, PP and MLN of infected WT or MyD88−/−TLR4−/− double deficient mice. (B–D) The number of bacteria recovered from MLN of individual mice is shown to the right. The percent of total cells±SEM is shown. Data are from two to four independent experiments with a total of 5–12 mice/group for each organ. *p≤0.05, differences between WT and knockout mice. n.s., not significant.

To further investigate the TLR4/5-independent recruitment of phagocytes, we examined recruitment in mice lacking MyD88, the adaptor used by all TLR that recognize bacterial ligands 32. Orally infected MyD88−/− mice had somewhat compromised accumulation of monocytes and neutrophils in PP and MLN, with only a minor decrease in the accumulation of neutrophils being apparent in PP (Fig. 5C). In contrast, no reduction of monocytes and only a partial reduction of neutrophils were detected in the blood of infected MyD88−/− mice (Fig. 5C, left).

TLR4 can signal via a MyD88-independent pathway using TIR domain-containing adaptor inducing IFN- β 32. To investigate if the release of phagocytes into the blood of infected MyD88−/− mice was mediated by this TLR4-mediated MyD88-independent pathway, we monitored cell accumulation in MyD88−/−TLR4−/− mice. There was a tendency that monocytes and neutrophils increased to a lesser degree in the blood of infected MyD88−/−TLR4−/− mice compared with WT mice (Fig. 5D, left), although not statistically significant. However, a statistically significant reduction in monocyte and neutrophil accumulation in both PP and MLN was apparent in infected MyD88−/−TLR4−/− mice (Fig. 5D). Thus, the accumulation of neutrophils and monocytes is defective in the PP and MLN of orally infected MyD88−/− mice and MyD88−/−TLR4−/− mice. In contrast, accumulation is unaffected in TLR4−/− mice, in the absence of TLR5 engagement and in the simultaneous absence of both TLR4 and TLR5 engagement.

Chemokine expression is dependent on MyD88/TLR4 signaling during oral Salmonella infection

Since the accumulation of phagocytic cells in the PP and MLN was abrogated in infected MyD88−/−TLR4−/− mice, we next determined whether chemokine induction in response to Salmonella infection depended on MyD88/TLR4 signaling. To test this, RNA was purified from the PP and MLN of WT and MyD88−/−TLR4−/− mice at days 2 and 4 post infection and chemokine expression was assessed. All chemokines examined were up-regulated less than threefold in the PP (Fig. 6A) or MLN (Fig. 6B) of infected MyD88−/−TLR4−/− mice. The compromised infection-induced expression of chemokines in MyD88−/−TLR4−/− mice is consistent with defective accumulation of monocytes and neutrophils in these animals.

Figure 6.

Altered monocyte and neutrophil recruitment in Salmonella-infected MyD88−/−TLR4−/− mice is reflected in impaired chemokine expression. C57BL/6 and MyD88−/−TLR4−/− mice were orally infected and 2 or 4 days post infection, and PP and MLN were removed. cDNA was prepared and chemokine expression was analyzed as in Fig. 1A and B. Data are from two independent experiments with a total of five to eight mice/group. Error bars represent the SEM. Fold up-regulation of chemokines in PP (A) and MLN (B) from the normalized ratio of infected mice compared with the normalized ratio of naïve mice±SEM. (A and B) *p≤0.05 and **p≤0.005, for differences between WT and MyD88−/−TLR4−/− mice. Other differences were not statistically significant.


Microbe recognition by cells within and beneath the intestinal epithelium is critical to recruiting the phagocytes required to contain infection by an oral pathogen. Here, we investigated chemokine induction by intestinal epithelial cells and chemokine up-regulation in PP and MLN, the first tissues targeted by orally acquired Salmonella. CCL2, CXCL2 and CXCL9 mRNA were up-regulated in PP and MLN coincident with accumulation of monocytes and neutrophils in these tissues. Using laser capture microdissection, we show that CXCL2 was preferentially expressed by cells in the FAE and SED while localized production of CCL2 and CXCL9 in PP was not readily apparent. In addition, chemokine expression by the villus epithelium was not detected despite it expressing TLR 33; the villus epithelium may have a more limited capacity to sense bacteria relative to the FAE due to the greater physical barrier overlying the villus epithelium 5, 34.

The ligands for CCR2 and CXCR2 (CCL2 and CXCL2) were present in the blood and infected PP and MLN already on day 2 after oral Salmonella infection. This suggests that these chemokine receptors are involved in monocyte and neutrophil accumulation in infected PP and MLN. However, CCR2 and CXCR2 are important in mobilizing monocytes and neutrophils, respectively, from the bone marrow 7, 8, 35, which complicates deciphering the role of these chemokine receptors for subsequent recruitment to tissues. In support of this, Serbina and Pamer showed that CCR2 is not required for monocyte migration into infected spleen or inflamed peritoneum 7 while Tsou et al. found that early, but not late phase, monocyte recruitment to the peritoneum is CCR2-dependent 9. Thus, very refined studies are required to directly assess the role of CCR2 for migration from blood into inflamed tissues during oral Salmonella infection.

CXCL9 mRNA was also highly up-regulated in PP and MLN early during infection and some infected mice had increased CXCL9 protein in the MLN and blood. CXCL9 can mediate migration of CXCR3+ monocytes 24 and has a well-established role in recruiting NK cells and activated T cells 36. Salmonella-induced CXCL9 may be involved in recruiting CXCR3+ monocytes. It could also influence NK cells and T cells, which are recruited early during Salmonella infection and are important early producers of IFN-γ 3, 37.

Despite some localization of chemokine expression in PP 2 days post infection, particularly CXCL2, inflammatory clusters of monocytes and neutrophils were scattered throughout the organ at day 4. It is likely that recruited phagocytes produce chemotactic substances and establish a gradient to attract other cells into the cellular foci. Indeed, monocytes and neutrophils from the blood, PP and MLN of infected mice produced CXCL2, and the levels increased after restimulation with bacteria. However, only small amounts of CCL2 were detected by monocytes and neutrophils from the blood of infected mice. Furthermore, neutrophils, and to a lesser degree monocytes, in the MLN of infected mice up-regulated CCL3 and CCL4 (data not shown). Interestingly, the NK cells surrounding clusters of monocytes and neutrophils in Listeria-infected WT mice were reduced in CCR5−/− mice 38. This suggests that the production of the CCR5 ligands CCL3 and CCL4 by myeloid cells might recruit NK cells to the inflammatory foci. In addition, studies investigating the sophisticated granulomas formed during Mycobacterium tuberculosis infection show that infected cells produce elevated levels of multiple chemokines, and taking away a single chemokine receptor often does not abrogate granuloma formation 39. Overall, chemokines acting in concert, rather than a single chemokine, seem important for co-ordinating migration within an infected organ. Moreover, the formation of cellular foci, which are preferentially made to prevent bacterial spread 40, 41, is ensured by redundancy in the chemokine network. Thus, CCL2, CXCL2, CCL3 and CCL4 may all play a role in formation of cellular foci in Salmonella-infected tissues.

Two of the most important candidates for initiating an immune response to Salmonella are TLR4 and TLR5 29–31, 42–44, which recognize LPS and flagellin, respectively 32. Despite this, the absence of signaling through TLR4, TLR5 or both TLR4/5 simultaneously did not compromise phagocyte recruitment to PP and MLN. Thus, despite that TLR4−/− mice are more susceptible to Salmonella31, 42, neutrophils and monocytes accumulate in PP and MLN to a similar extent as WT mice. One explanation is that other TLR can replace TLR4 regarding cell recruitment, but not their full activation and expression of anti-microbial functions such as cytokines and iNOS. Consistent with this, fewer iNOS+ monocytes were present in infected TLR4−/− compared with WT mice with a similar bacterial load (data not shown). TLR5-independent recruitment is also consistent with greater resistance of TLR5−/− mice to oral Salmonella44.

We found that the absence of MyD88 rather than TLR4/5 compromised phagocyte recruitment. As IL-1 and IL-18 signaling also depend on MyD88, it remains possible that the abrogation of signaling via these cytokines, rather than TLR, results in defective recruitment in infected MyD88−/− mice. This seems unlikely, however, since several cytokines released upon TLR activation induce chemokine production, and TLR activation also can generate IL-1β and IL-18 directly 45. Instead, other TLR signaling via MyD88, such as TLR2, could replace TLR4 and TLR5 and allow recruitment of cells and ability to control the bacteria. Consistent with this, orally infected TLR4−/−TLR2−/− and MyD88−/− mice were much more susceptible to Salmonella than TLR4−/− or TLR2−/− single knockout mice 30. Whether this was due to altered cell recruitment was not analyzed by Weiss et al. 30. However, our analysis revealed that neutrophils accumulated to some extent in the PP of infected MyD88−/− mice. This could be due to MyD88-independent chemokines 46 produced by, for example, the TLR4-dependent MyD88-independent pathway. Alternatively, pattern recognition receptors other than TLR, which recognize intracellular bacterial ligands 32, may be involved. In support of this, some accumulation of monocytes and neutrophils was still apparent in blood and PP of infected MyD88−/−TLR4−/− mice.

In summary, CXCL2, CXCL9 and CCL2 mRNA are first induced in PP and MLN followed by CCL3 and CCL4 in response to oral Salmonella infection. Monocytes and neutrophils themselves produce some of these chemokines and can thus amplify their recruitment and direct the formation of inflammatory foci. Moreover, both MyD88-dependent and MyD88/TLR4-independent pathways are involved in phagocyte recruitment to the intestinal lymphoid tissue during oral Salmonella infection. These studies provide insight into the mechanisms underlying the orchestrated phagocyte recruitment to orally acquired Salmonella.

Materials and methods


C57BL/6 mice were purchased from Charles River Laboratories (Sulzfeld, Germany). TLR4−/−, MyD88−/− and MyD88−/−TLR4−/− mice, all backcrossed ≥7 generations on the C57BL/6 background, were bred at the Experimental Biomedicine animal facility of Göteborg University. Mice used were 8–12 wk of age and were provided with food and water ad libitium. All experiments were performed using protocols approved by the government animal ethical committee and institutional animal use and care guidelines.

Bacterial strains

S. enterica serovar Typhimurium χ8554 (SL1344ΔasdA16, rpsL,hisG) harboring the ASD+ vector pYA810 was used for all experiments. In addition, flagella-deficient S. enterica serovar Typhimurium χ8962 (SL1344ΔasdA16, ΔfljB217ΔfliC823)/pYA810 was used to perform a subset of the experiments in Fig. 5. Bacteria were grown in Lennox broth overnight at 37°C. The bacterial suspension was diluted and the optical density was measured at 600 nm. After centrifugation, the bacteria were resuspended at the appropriate concentration in sterile PBS. The actual bacterial dose administered was determined by serial plating of bacteria on Lennox agar plates.

Infection of mice

Mice were given 0.1 mL of 1% NaHCO3 intragastrically 10 min before infection. C57BL/6 and TLR4−/− mice were then infected intragastrically with 6×108–1×109 bacteria in 200 μL PBS. MyD88−/− and MyD88−/−TLR4−/− mice are more susceptible to Salmonella and were given approximately 8×106 bacteria in 200 μL PBS intragastrically. In Fig. 5, only mice with a bacterial load above 6×103 bacteria/MLN were included in data analysis. Mice were sacrificed 2–4 days post infection for all experiments except the transmigration assays (Fig. 4C and D) and the chemokine production assay (Fig 4B) where mice were taken at day 5–6. The bacterial load in PP, MLN and spleen was determined at the time of sacrifice by plating serial dilutions of single-cell suspensions on Lennox agar plates (Fig. 1C and D, 5B–D).

Measurements of chemokines by ELISA or cytokine bead array

Tissue lysates and serum were prepared as described previously 4. Concentrations of CCL2, CXCL9 and CXCL2 in PP and MLN extracts, serum samples and sorted cell populations were determined using either the Opt EIA Elisa kit (BD) or mouse cytokine bead array (CBA) kit (BD Biosciences) for CCL2. CXCL9 was quantitated using MIG (CXCL9) CBA flex set (BD Biosciences), while CXCL2 was measured using the Quantikine kit from (R&D Systems, Abingdon, UK) according to the manufacturer's recommendations.

Immunohistochemistry and laser capture microdissection microscopy

Sections of PP from naïve and infected mice were prepared for immunohistochemistry as previously described 47. Detection of surface markers was carried out at room temperature for 30–60 min using the following mAb that were biotinylated or conjugated to FITC: anti-CD11c (HL3), B220 (RA3-6B2), Ly6C (AL-21), Ly6G (1A8) or appropriate isotype controls. The biotin-conjugated antibodies were visualized after washing in PBS by adding streptavidin-conjugated Alexa488 or Alexa594 for 30–60 min at room temperature. When intracellular staining was performed, 0.5% saponin was used throughout the staining procedure. The sections were incubated for 30–60 min at room temperature with anti-iNOS Ab (M-19) (Santa Cruz Biotechnology, Santa Cruz, CA, USA) or appropriate isotype control. Staining with rabbit anti-iNOS was followed by incubation with Alexa594-conjugated anti-rabbit IgG (Santa Cruz Biotechnology) for 30 min at room temperature. Sections were washed in PBS and mounted with DakoCytomation fluorescent mounting medium and visualized using a microscope (Axioscope 2, Zeiss) fitted with an AxioCam MRc digital camera and Axiovision Rel 4.5 software. Final images were produced with ImageJ (National Institutes of Health, Bethesda, MD, USA).

For laser capture microdissection, sections were prepared as above and placed on PEN membrane-coated slides (PALM, Microlaser Technologies, Bernried, Germany). Sections were fixed in 70% acetone for 30 s followed by 100% acetone for 4 min and air dried and washed. Sections were stained with hematoxylin for 30 s and stored in an airtight box in −80°C until use. Laser capture microdissection was performed with a Zeiss microscope using a microcatapulting laser system and the cut tissues were collected in adhesive caps-tubes (PALM) and stored at −80°C until RNA was isolated.

RNA isolation and cDNA synthesis

Total RNA was isolated from PP and MLN using RNeasy Minikit (Qiagen) and from microdissected samples using RNeasy Microkit (Qiagen). RNA quantity and purity was estimated using spectrophotometry (260/280) and the integrity was visualized on an agarose gel for each sample. SuperScriptIII reverse transcriptase and oligo (dT) primers were used according to the manufacturer's protocol (Invitrogen Life Technologies) to obtain first-strand cDNA from 1–5 μg RNA. When less RNA was available (samples dissected using laser capture microdissection), SuperScriptIII reverse transcriptase kit for low amounts of RNA was used according to the manufacture's protocol (Invitrogen Life Technologies) to obtain first-strand cDNA.

Quantitative real-time RT-PCR

Primers were designed using the ProbeFinder software, Universal Probe library for Mouse (Roche Applied Science, Mannheim, Germany), and purchased from Invitrogen, UK. Probes were purchased from Universal ProbeLibrary for mouse (Roche) with the exception of the probe for hypoxanthine-guanine phosphoribosyltransferase (hprt), which was made in-house and labeled with the fluorescent reporter Texas Red and the quencher Blackhole Quencher 2. Details of the probes and primers are given in Table 1. RT-PCR was performed using the LightCycler 480 Probes Master and amplification and detection of specific products was performed using a LightCycler480 (Roche Diagnostics). cDNA standard curves for each primer set were generated by measuring amplification of twofold serial dilutions of cDNA. hprt was used as a reference gene in all experiments. A calibrator was used in every run to compensate for inter-run differences and the normalized ratio was calculated using the second derivative maximum method (LightCycler 480 system software, Roche Diagnostics). The normalized ratios of all chemokines in naïve C57BL/6 mice were set to 1. The normalized ratio of chemokines in PP and MLN of naïve MyD88−/−TLR4−/− mice (Fig. 6) was at most 1.9, and at the lowest 0.5 that of naïve C57BL/6 mice (data not shown).

Table 1. Primers and probes
TargetPrimera) or probeb)SequenceAccession no.
  • a)

    a) F=forward, R=reverse

  • b)

    b) The number refers to the specific probe used from the Universal ProbeLibrary for mouse (Roche).

Ccl2FcatccacgtgttggctcaNM 011333.2
Ccl3FtctgtcacctgctcaacatcaNM 011337.1
Ccl4FgccctctctctcctcttgctNM 013652.1
Ccl6FccttgtggctgtccttggNM 009139.1
Ccl9,MMRP2FcccactaagaagatgaagcctttU15209.1/U15209:EMBL/104533:MGI/Ccl9; MGI
Ccl20FaactgggtgaaaagggctgtNM 016960.1
Cx3cl1FcgcgttcttccatttgtgtaNM 009142.2
Cxcl2FaaaatcatccaaaagatactgaacaaNM 009140.1
Cxcl9FcttttcctcttgggcatcatNM 008599.2
Hprt1FtcctcctcagaccgcttttNM 013556.2

Cell preparation, flow cytometry and cell sorting

Cell suspensions were prepared from the spleen, MLN, PP and blood as described previously 3. Flow cytometry was performed as published earlier 3 using the additional antibody MHC-II Alexa fluor 700 (eBioscience, San Diego, CA, USA).

For cell sorting, blood or MLN/PP were pooled from naïve mice or mice infected 4–6 days earlier. CD11b- and CD11c-expressing cells were enriched using an AutoMACS (Miltenyi Biotech, Bergish Gladbach, Germany) after incubation with anti-CD11b (M1/70) and CD11c (N418) beads, (Miltenyi Biotech) according to the manufacturer's protocol. After staining for flow cytometry, cells were sorted into Ly6Chi monocytes and Ly6Ghi neutrophils using a FACS ARIA flow cytometer (BD Biosciences) and DIVA software (BD Biosciences).

Transmigration assays

Monocyte and neutrophil migration was evaluated using 24-well low adherence plates with 5.0 μm pore size membrane inserts (Corning Costar, NY, USA). CD11b-expressing cells were enriched from nine to ten infected or naïve mice per experiment using an AutoMACS. Viability was assessed using trypan blue and cells were resuspended at 5×106 cells/mL (blood) or 1×106 cells/mL (MLN/PP) in complete RPMI 1640 Glutamax-1 medium (Invitrogen Life Technologies) containing 10% FBS (PAA Laboratories, Linz, Austria). Following a 1 h incubation at 37°C, 100 μL of the cell suspension was loaded to the upper chamber of a Transwell plate where 600 μL of complete medium with or without the indicated chemokines had previously been added to the lower chamber. After 90 min at 37°C, the migrated cells were quantified using a flow cytometer (LSRII) and stained for flow cytometry analysis. The experiments were performed in duplicates or triplicates and the chemokines were titrated. The concentrations of chemokines tested (all purchased from R&D Systems) were: CCL2: 1600, 800, 160 and 16 ng/mL; CXCL2: 500, 50, 5, and 0.5 ng/mL; CCL3: 400, 200, 40, 20 and 2 ng/mL; CCL4: 1200, 600, 200, 120, 20 and 12 ng/mL. Net migration was calculated by the number of migrated cells minus background migration divided by original number of cells.

Statistical analysis

Statistical analysis was performed using the non-parametric Mann–Whitney U-test for unpaired samples (two-tailed) using SPSS. Asterisks in figure legends (*/**) represent p values of ≤0.05 and ≤0.005, respectively.


The technical assistance of Kristina Lindgren and Emilia Heimann is gratefully acknowledged as is the valuable input of Miguel A. Tam. The authors thank Knut Kotarsky (Lund University) for excellent advice about laser capture microscopy and Matthias Mack (University of Regensburg, Germany) for anti-CCR2 mAb MC21. They are also grateful to Roy Curtiss III (Arizona State University, USA) for Salmonella χ8554, χ8962 and pYA810. This work was supported by grants from the Swedish Research Council (621-2004-1378 and 621-2007-6536), the Swedish Foundation for Strategic Research Microbes and Man Program (A3 01:93/01/01) and was performed at the Mucosal Immunobiology and Vaccine Center (MIVAC) funded by the Swedish Foundation for Strategic Research. We acknowledge the Center for Cellular Imaging at Göteborg University for the use of imaging equipment and support of the staff.

Conflict of interest: The authors declare no financial or commercial conflict of interest.