Female wild-type BALB/c mice (6–8 weeks old, 17–21 g) were obtained from ENVIGO, Rossdorf, Germany or from Janvier Labs, Saint Berthevin Cedex, France. Myeloid differentiation factor (MyD) 88^−/− (33), toll-like receptor (TLR) 4^−/− (34), and recombination-activating gene (RAG) 2^−/− (35) mice on BALB/c background were bred at the local animal facility of the University Hospital Essen. All mice were kept under specific pathogen-free conditions and had access to standard rodent food and water ad libitum. All animal experiments were performed according to the ethical principles and guidelines for scientific experiments. The protocol has been approved by the local ethic committee Landesamt für Natur-, Umwelt-, und Verbraucherschutz (LANUV), North-Rhine-Westphalia.

Polymicrobial sepsis was induced through cecal ligation and puncture (CLP) as described previously (29). Briefly, after anesthesia with an intramuscular injection of ketamine (100 mg/kg) and xylazine (10 mg/kg), the mice underwent a midline laparotomy and the cecum was exposed and ligated by 50%. The cecum was punctured once with a 27-gauge needle and approximately 1 µl of cecum content was extruded. Thereafter, the cecum was replaced, 1 ml sterile saline was administered intraperitoneally (i.p.) for resuscitation, and the peritoneum was closed in two layers. CLP caused a moderate sepsis with a mortality rate of 20% within the first 2 days. Where indicated mice received the CCR2 antagonist RS102895 (5 mg/kg body weight Tocris Biosciences, Bio-Techne, Avonmouth, Bristol, UK) dissolved in DMSO i.p. 6 h before sham or CLP operation and every 6 h thereafter until the end of the experiment. This dosage maintains a constant concentration of circulating RS102895 (36). Where indicated, the functional S1P receptor antagonist FTY720 (2 mg/kg body weight; Fingolimod, LC Labs, Woburn, MA, USA) dissolved in 20% 2-hydroxypropyl-β-cyclodextrin (2-HPCD) was intravenously (i.v.) applied immediately before sham and CLP operation and 12 h thereafter. Control mice received the solvents only. In some experiments, the mice were treated i.p. with antibodies against PDCA-1/Bst2 (clone JF05-1C2.4.1, Miltenyi Biotech; 100 µg/mouse) or with the respective rat IgG2b isotype control antibodies 6 h before and 24 h after surgery.

Splenectomy was performed 3–4 weeks before sham or CLP operation. Therefore, mice were anesthetized with an intramuscular injection of ketamine (100 mg/kg) and xylazine (10 mg/kg). After a 1 cm incision of the left lateral part of the peritoneum, the spleen was exposed and removed after ligation of the blood vessels. After resuscitation with 1 ml sterile saline i.p., the peritoneum was closed in two layers. Control mice underwent laparotomy only. For analgesia, carprofen (4 mg/kg body weight Paracarp, Norbrook Laboratories, Corby, Great-Britain) was injected subcutaneously immediately after surgery and 24 h thereafter.

VLE RPMI 1640 (Biochrom, Berlin, Germany) supplemented with 10% fetal calf serum (FCS, Biochrom), 10 mM HEPES (Biochrom), 2 mM glutamine, 0.06 mg/ml penicillin, 0.02 mg/ml gentamicin, and 0.05 mM 2-Mercapthoethanol (all from Sigma-Aldrich, Taufkirchen, Germany) was used as culture medium (CM).

Whole blood was collected by cardiac puncture and transferred into EDTA tubes. For erythrocyte depletion, 200 µl of blood were lysed using 2 ml Red Blood Cell Lysing Buffer (Sigma-Aldrich).

To collect the peritoneal lavage and cells, 3 ml PBS were injected into the peritoneal cavity, aspirated, and centrifuged. The supernatant was harvested as peritoneal lavage. Thereafter, 4 ml of CM containing 5 IU/ml Heparin Ratiopharm (Ratiopharm GmbH, Ulm, Germany) were injected into the peritoneal cavity, aspirated, and pooled with the cells previously obtained from the lavage with PBS. Care was taken to avoid the injury of blood vessel.

For the isolation of total spleen cells, the spleen was removed and digested with collagenase as described previously (28). Erythrocytes were lysed with Red Blood Cell Lysing Buffer (Sigma-Aldrich). Splenic DCs were isolated using magnetic beads coupled with antibodies against CD11c (MagniSort CD11c Positive Selection Kit, Invitrogen, Carlsbad, CA, USA) as recommended by the manufacturer and were cultured as described (28).

Bone marrow cells (BMCs) were flushed out of femurs and tibiae with CM, gently dissociated, and filtered through a 30 µm filter. To generate bone marrow-derived dendritic cells (BMDCs), BMCs (2 × 10⁵/ml) were cultured in CM containing 20 ng/ml recombinant murine granulocyte/macrophage-colony-stimulating factor (GM-CSF; R&D Systems, Wiesbaden, Germany) for 7–9 days as described previously (37). Non-adherent cells were harvested and BMDC were cultured in the absence or presence of 5 µg/ml CpG (ODN1668, InvivoGen, Toulouse, France) for 18 h. The supernatant was stored at −20°C for further analyses. Alternatively, BMCs were cultured in the presence of Flt3L (R&D Systems) for 7 days in order to generate conventional and pDCs as described (38).

For the preparation of bone marrow lavage, femurs and tibiae were flushed with 2 ml PBS. After centrifugation, the supernatant was harvested and stored at −20°C for further analyses.

For surface staining, cells were incubated with diverse combinations of fluorochrome-labeled antibodies against CD11c (clone N418), MHC class II (clone M5/114.15.2), CD4 (clone RM4-5), Ly6C (clone AL-21), CD103 (clone 2E7), CD11b (clone M1/70), Bst2 (clone eBio927), B220 (clone RA3-6B2), SIRP-α (clone P84), CCR9 (clone eBioCW1.2), or Siglec-H (clone eBio440c) for 15 min in the dark at 4°C as indicated. Finally, the cells were washed with Cell Wash (BD Biosciences). Appropriate isotype controls were used to define the threshold of negative versus positive staining. All antibodies were purchased from BD Biosciences or from eBioscience (Frankfurt am Main, Germany). Data were acquired using a FACSCalibur or FACSCanto II (BD Biosciences). Data analysis was performed using CellQuest Pro, FACSDiva Software (both BD Biosciences), or NovoExpress (ACEA Biosciences, San Diego, CA, USA).

Bone marrow cells were adjusted to a density of 1 × 10⁷ cells/ml in PBS supplemented with 2% FCS and were stained with fluorescent antibodies against MHC class II, CD4, and CD11c as described above. The cells were washed twice with PBS and CD11c^hi MHCII⁺CD4⁺ cells were sorted using a FACS Aria (BD Biosciences). The positive fraction was used for cytospin preparations (see below) and for RNA extraction, the negative fraction was used as depleted BMC for cell culture. Flt3L generated DCs were sorted as CD3⁻CD19⁻CD11c⁺CD11b⁺B220⁻ conventional DCs and CD3⁻CD19⁻CD11c⁺CD11b⁻B220⁺ pDCs.

Twenty thousand cells were mounted onto a glass slide using a CytospinTM 4 (Thermo Fisher Scientific, Dreieich, Germany) for 5 min at 400 rpm and were fixed with ice-cold methanol–acetone (1:1 vol/vol). Cells were stained with Giemsa-solution for 5 min.

The content of IL-10, IL-12p70, and CCL2 was quantified using ELISA (all DuoSet, R&D Systems) or cytometric bead array (CBA, BD Biosciences) as recommended by the manufacturers.

RNA was extracted using the RNeasy micro kit (Qiagen, Hilden, Germany) and reverse transcribed using the Sensiscript Reverse Transcriptase (Qiagen) or SuperScript III Reverse Transcriptase (ThermoFisher Scientific, Waltham, MA, USA). Real-time PCR was performed with the following primers: e2-2 (fwd TGGGCTCAGGGTACGGAACT, rev CAGAGCCACGCCATCTTCAC), id2 (fwd GACAGAACCAGGCGTCCAGG, rev AGCTCAGAAGGGAATTCAGATG), il-10 (fwd CTGGACAACATACTGCTAACCGACTC, rev ATTTCTGGGCCATGCTTCTCTGC), β-actin (fwd CGCTCAGGAGGAGCAATG, rev TGACAGGATGCAGAAGGAGA), rps9 (fwd CTGGACGAGGGCAAGATGAAGC, rev TGACGTTGGCGGATGAGCACA). Wobble primers for several ifn-α subtypes were: fwd ATGGCTAGRCTCTGTGCTTTCCT and rev AGGGCTCTCCAGAYTTCTGCTCTG. PCR was run on the iQ5 iCycler (Bio-Rad) or on the CFX96 RealTime C1000 Thermo Cycler (Bio-Rad) using FastStar Taq Man Probe Master (Roche) and Mesa Green qPCR Mastermix Plus SYBR Assay with fluorescein (Thermo Fisher Scientific), respectively according to the manufacturers’ recommendations. The target gene expression was normalized on the housekeeping genes β-actin or rps9 and was calculated as 2^−ΔCt with ΔCt = Ct target-Ct housekeeping.

Data are shown as individual values with median and interquartile range or as mean ± SD or SEM. Differences between two groups were tested using Mann–Whitney U-test. Multiple comparisons were performed with the Kruskal–Wallis test followed by Dunn’s posttest or by One-way ANOVA followed by Newman–Keuls Multiple Comparison test. Normalized values were tested using the One-sample t-test. A p-value <0.05 was considered as significant. Statistical analyses and the preparation of graphs were performed using GraphPad Prism 5.0.

We have previously shown that BMC from septic mice when cultured in the presence of GM-CSF in vitro give rise to BMDC that resemble splenic DCs during sepsis in terms of increased IL-10 synthesis (29). Bone marrow cell cultures in the presence of GM-CSF mimic the differentiation of DCs under inflammatory conditions. Due to their enhanced secretion of IL-10 in response to bacterial stimuli such as CpG immunostimulatory oligonucleotides BMDC from septic mice possess immunosuppressive properties (29). To elucidate the sepsis-induced changes in the bone marrow that may result in the altered differentiation of BMDC, we investigated the CpG-induced cytokine secretion pattern of BMDC generated from bone marrow at different time points after CLP. To adjust inter-assay variations, the values obtained from BMDC of septic mice were normalized to the values received from BMDC of sham mice (set as 100%; a representative data set of absolute values is given in Figure S1 in Supplementary Material). At least up to 12 h after CLP, the bone marrow gave rise to BMDC that secreted moderately enhanced levels of IL-12 but similar amounts of IL-10 in comparison with bone marrow from sham mice. From 24 h after CLP, BMDC displayed a strong increase in IL-10 production (Figure 1A).

We next searched for changes in the cellular composition of the bone marrow that were associated with the altered differentiation of BMDC 24 h and later after sepsis induction. We identified an enlarged population of cells that highly expressed CD11c, along with MHC class II and CD4 by 36 h after CLP (Figure 1B). Strong expression of CD11c together with MHC class II is characteristic for terminally differentiated DCs. Morphologically, the CD11c^hiMHCII⁺CD4⁺ cells displayed a large nucleus and partly showed the characteristic dendritic cell shape (Figure 1C). Therefore, we termed the CD11c^hiMHCII⁺CD4⁺ cells “CD4⁺ DCs” hereafter. Few CD4⁺ DCs were also present in the bone marrow of sham mice (Figure 1D). The number of CD4⁺ DCs in the bone marrow rose 24 h after CLP and this population further expanded at least until 36 h (Figure 1D) and, thus, paralleled the increased release of IL-10 from BMDC generated from bone marrow of septic mice. By 96 h after CLP, the population of CD4⁺ DCs had largely disappeared from the bone marrow (Figure 1D). The rise in the number of CD4⁺ DCs in the bone marrow after CLP was dependent on MyD88 but not on TLR4 (Figure 1E).

In order to address the potential impact of CD4⁺ DC in the bone marrow on the differentiation of BMDCs, CD4⁺ DCs were depleted from BMC (through cell sorting) obtained 36 h after surgery. Total BMC and cells after depletion of CD4⁺ DCs were used to set up cultures for the generation of BMDC (Figure 2A). At the end of the in vitro culture, the percentage of CD11c⁺MHC class II⁺ BMDC did not differ irrespective of the origin of the bone marrow (sham or CLP) or of the presence of CD4⁺ DCs (Figure 2B). With regard to the cytokine secretion, prior depletion of CD4⁺ DCs from BMC of sham mice did not significantly change the CpG-induced release of IL-12 and IL-10 from BMDC (Figure 2C). Likewise, the absence of CD4⁺ DCs in the bone marrow of septic mice did not affect the secretion of IL-12 from generated BMDC (Figure 2C). In contrast, BMDC released significantly less IL-10 when they had been generated from BMC of CLP mice that had been depleted from CD4⁺ DCs before the onset of the culture (Figure 2C). Consequently these BMDC displayed an increased IL-12/IL-10 ratio in comparison to BMDC that differentiated from total bone marrow of CLP mice (Figure 2C). Thus, CD4⁺ DCs accumulate in the bone marrow during sepsis and modulate the cytokine secretion pattern of subsequently differentiating BMDC toward increased IL-10 production.

In general, the enlargement of a specific population may be caused by local cell proliferation or by the recruitment of cells from the periphery. The analysis of the proliferation marker Ki-67 indicated a comparable though low percentage of proliferating CD4⁺ DCs in the bone marrow of both sham and CLP mice (Figure S2 in Supplementary Material).

Sphingosine 1-phosphate contributes to the migration of DCs and lymphocytes through lymphatic and blood vessels, respectively (39, 40). Since the bone marrow is not connected to the lymphatic system the circulation is the only way to enter the bone marrow from the periphery. In order to evaluate whether S1P played a role in the accumulation of CD4⁺ DCs in the bone marrow after CLP, the mice were treated with FTY720 or with its solvent before CLP or sham operation. The sepsis-induced accumulation of CD4⁺ DCs in the bone marrow was completely blocked upon application of FTY720 (Figure 3A). In contrast, FTY720 did not affect the number of CD4⁺ DCs in the bone marrow of sham mice (Figure 3A).

The spleen is a large reservoir for conventional CD11c^hiCD4⁺ DCs (gated as shown in Figure 3B). The expansion of CD4⁺ DCs in the bone marrow 24 and 36 h after CLP was paralleled by a striking decrease of CD4⁺ DCs in the spleen (Figures 3B,C). We further investigated whether the enlarged population of CD4⁺ DCs in the bone marrow was associated with the loss of DCs in the spleen during the course of sepsis. Therefore, mice underwent splenectomy or control laparotomy 4 weeks before sham or CLP operation. Subsequent analysis showed that prior splenectomy did not prevent the sepsis-induced accumulation of CD4⁺ DCs in the bone marrow (Figure 3D).

We further examined the impact of FTY720 on the distribution of CD4⁺ T cells in the bone marrow during sepsis. Numerous studies have shown that lymphocytes undergo apoptosis-induced cell death in secondary lymphoid organs and in the blood within 24 h after induction of sepsis (41, 42). In contrast, there exists limited if not any information on the behavior of T lymphocytes in the bone marrow during sepsis. Interestingly, we observed an enlarged population of CD4⁺ T cells in the bone marrow by 36 h after CLP that was dependent on S1P receptor function (Figures 4A,B). In line with a previous study on the effect of pharmacologically induced modulation of S1P signaling during sepsis (43), the application of FTY720 caused severe lymphopenia in the blood of both groups (Figure 4C).

The comparable S1P-dependent expansion of CD4⁺ T cells and CD4⁺ DCs implied a relationship between these two cell types in the bone marrow. To examine whether the accumulation of CD4⁺ DCs in the bone marrow was linked to the increased number of CD4⁺ T cells, sepsis was induced in Rag2 knockout mice that lack CD4⁺ T cells (Figure 4D). The accumulation of CD4⁺ DCs was also observed in Rag2 deficient mice (Figure 4E). Thus, the expansion of the CD4⁺ DC population in the bone marrow after CLP is dependent on S1P but independent of CD4⁺ T cells that accumulate in parallel.

We further analyzed the expanding CD4⁺ DC population in the bone marrow after CLP for the expression of CD103, SIRP-α, and Ly6C that are associated with conventional DC lineages and/or monocyte-derived DCs, respectively (12, 44–47). The CD4⁺ DCs were negative for CD103 but expressed SIRP-α and Ly6C suggesting a monocytic origin (Figure 5A). Despite the knowledge on the importance of monocytes in inflammatory diseases, only few data exist on their behavior during CLP-induced polymicrobial sepsis. We investigated the distribution of monocytes in the bone marrow and in the peritoneal cavity as the primary site of infection. A large number of inflammatory Ly6C^hi monocytes infiltrated the peritoneal cavity between 6 and 12 h after CLP and dropped thereafter (Figures 5B,C). In parallel, the number of Ly6C^hi monocytes in the bone marrow strongly declined (Figures 5B,C). The majority of the remaining Ly6C^hi monocytes expressed CCR2, the receptor for CCL2 (Figure S3 in Supplementary Material). Within 96 h after CLP, the population of Ly6C^hi monocytes in the bone marrow was restored to homeostasis (Figure 5C). The egress of Ly6C^hi monocytes from bone marrow after CLP was preceded by the release of CCL2 in the peritoneal cavity (Figure 5D).

To investigate a potential association between monocytes and the expansion of the CD4⁺ DC population in the bone marrow, mice were treated with RS102895, a specific small molecule antagonist for CCR2 (48), before induction of sepsis. The administration of the CCR2 antagonist completely blocked the accumulation of CD4⁺ DCs in the bone marrow after CLP but did not significantly affect this population in sham mice (Figure 5E). The number of Ly6C^hi monocytes in the bone marrow of sham mice tended to be increased as a consequence of CCR2 inhibition (Figure 5F). In contrast, the application of the CCR2 antagonist did not affect the egress of monocytes from bone marrow after CLP (Figure 5F). Thus, the accumulation of CD4⁺ DCs in the bone marrow during sepsis is regulated by CCR2 but seems to be independent of monocyte mobilization.

After excluding a link between monocytes and CD4⁺ DCs in the bone marrow after CLP, we performed additional analyses of surface molecule expression. The CD4⁺ DCs in the bone marrow of septic mice did not express the myeloid marker CD11b but were positive for B220, Ly6C, Bst2, CCR9, and Siglec-H indicating that these cells represented pDCs (Figure 6A). In line with this assumption, sorted CD4⁺ DCs from septic mice expressed the transcription factor e2-2 that is characteristic for pDCs but not id2 that is expressed by conventional DCs (Figure 6B). In sham mice, the population of CD4⁺ DCs largely consisted of pDCs but additionally contained a small population of CD11b⁺ cells, most likely, conventional DCs (Figure 6A).

Recently, it has been reported that according to the expression of CD4 two maturation stages of pDCs in the bone marrow exist: CD4⁻ pDCs may develop into CD4⁺ pDCs and both populations may be released into the circulation (49). To get further insight into the expansion of CD11c^hiCD4⁺ pDCs (that we had previously termed “CD4⁺ DCs”) in the bone marrow during sepsis, we investigated the expression of CD4 and MHC class II on CD11c^hi and CD11c^int pDCs among total CD11c⁺Bst2⁺Ly6C⁺B220⁺ pDCs (for gating, see Figure 7A). The discrimination between CD11c^hi and CD11c^int was performed equivalent to Figure 1B. 16 and 36 h after CLP, the number of total pDCs in the bone marrow did not differ between sham and CLP mice (Figure 7B; Figure S4A in Supplementary Material). Among the small subpopulation of CD11c^hi pDCs, we observed an increased percentage of MHCII⁺ pDCs at the expense of MHCII^- pDCs, both for the CD4⁻ and CD4⁺ subset at the earlier time point (Figure 7C). Moreover, the MHCII⁺CD4⁺ pDC from septic mice, expressed higher levels of MHC class II per cell and a larger part expressed CD86 in comparison with sham mice indicating an activated phenotype (Figures 7D,E). The activation of CD11c^hiCD4⁺ pDCs was still apparent by 36 h after CLP but was not associated with ifn-α expression (Figures S4B–D in Supplementary Material). Among the major population of CD11c^int pDCs, a similar shift toward an increased percentage of MHC class II⁺ pDC was detected in the CD4⁻ but not in the CD4⁺ subpopulation (Figure 7C). Later on, by 36 h after CLP, the previously emerged shift toward MHCII⁺ pDCs was still apparent and additionally had established for CD11c^intCD4⁺ pDC (Figure S4B in Supplementary Material).

Substantial amounts of CCL2 were present in the bone marrow by 16 h after CLP (Figure 7F). Considering that blocking of CCR2 prevented the expansion of the CD4⁺ DC population in the bone marrow after CLP in vivo and that pDCs are known to express CCR2 (50), we asked whether there exist a direct effect of the CCR2 antagonist on pDCs in the bone marrow. To address this question, BMCs were isolated 16 h after CLP and were cultured in the absence or presence of the CCR2 antagonist RS102895 before analysis of CD86 expression. The inhibition of CCR2 led to a decreased expression of CD86 on CD11c^hiCD4⁺ pDCs (Figure 7G).

The number of pDCs that circulated in the blood of septic mice was strongly reduced irrespective of the subset (Figure 7H and Figure S4E in Supplementary Material for 16 and 36 h after surgery, respectively). Similar to pDCs in the bone marrow, circulating pDCs, especially the CD4⁺ subset, expressed higher levels of MHC class II after CLP (Figure 7I and Figure S4F in Supplementary Material for 16 and 36 h after surgery, respectively). Thus, the previously identified CD4⁺ DCs that expand in the bone marrow during sepsis are activated pDCs.

We have previously shown that 4 days after induction of sepsis, splenic DCs still maintain their bias toward increased IL-10 synthesis. At that time point, a large proportion of splenic DCs has been replaced by de novo generated DCs (37). Considering that pDCs in the bone marrow modulate the differentiation of DCs in vitro as shown above, we asked whether pDC influence the function of DCs in vivo. To address this issue, pDCs were depleted before CLP or sham surgery and splenic DCs were analyzed 4 days later. The population of the CD4⁺ pDCs was almost completely removed from bone marrow after application of the specific antibodies in sham and CLP mice (Figures 8A,B). Splenic DCs from CLP mice expressed clearly increased levels of il-10 mRNA ex vivo and released more IL-10 after stimulation with CpG than DCs from sham mice when they had received the isotype control antibodies (Figures 8C,D). In contrast, prior depletion of pDCs reduced ex vivo il-10 transcripts and CpG-induced IL-10 production in splenic DCs from septic mice but not in DCs from sham mice (Figures 8C,D). Thus, pDC favor the generation of IL-10-producing splenic DCs during sepsis.