The expression of mouse CLEC‐2 on leucocyte subsets varies according to their anatomical location and inflammatory state

Expression of mouse C‐type lectin‐like receptor 2 (CLEC‐2) has been reported on circulating CD11bhigh Gr‐1high myeloid cells and dendritic cells (DCs) under basal conditions, as well as on a variety of leucocyte subsets following inflammatory stimuli or in vitro cell culture. However, previous studies assessing CLEC‐2 expression failed to use CLEC‐2‐deficient mice as negative controls and instead relied heavily on single antibody clones. Here, we generated CLEC‐2‐deficient adult mice using two independent approaches and employed two anti‐mouse CLEC‐2 antibody clones to investigate surface expression on hematopoietic cells from peripheral blood and secondary lymphoid organs. We rule out constitutive CLEC‐2 expression on resting DCs and show that CLEC‐2 is upregulated in response to LPS‐induced systemic inflammation in a small subset of activated DCs isolated from the mesenteric lymph nodes but not the spleen. Moreover, we demonstrate for the first time that peripheral blood B lymphocytes present exogenously derived CLEC‐2 and suggest that both circulating B lymphocytes and CD11bhigh Gr‐1high myeloid cells lose CLEC‐2 following entry into secondary lymphoid organs. These results have significant implications for our understanding of CLEC‐2 physiological functions


Introduction
Many physiological functions are critically regulated by finetuned interactions between diverse subsets of hematopoietic and Correspondence: Dr. Guillaume E. Desanti e-mail: g.desanti@bham.ac.uk nonhematopoietic cells within primary and secondary lymphoid organs (SLOs) as well as in the circulation. These interactions are mediated by surface receptors or secreted molecules that display complex cellular, spatial, and temporal expression patterns. A deep understanding of these patterns is required for a full knowledge of their physiological significance and for effective therapeutic intervention. For example, expression of receptor activator of NFκB ligand by bone osteoblasts regulates bone density while its expression by developing thymocytes regulates medullary thymic epithelial cell maturation (for review; [1]).
CLEC-2 surface expression on platelets was first demonstrated in humans [16] and soon after in mouse [17] and chicken [18]. The expression of CLEC-2 and its RNA transcript -encoded by the C-type lectin domain family 1, member B (Clec1b) genehas also been studied in leucocytes isolated from different species leading to a rather confusing mosaic of results. While CLEC-2 is absent from chicken leucocytes [18] and restricted to liverresident Küppfer cells in human [19][20][21][22], a much broader expression profile of CLEC-2/Clec1b has been reported in rodent leucocytes, particularly in mice.
While one report claims that mouse CLEC-2 surface expression by leucocytes is restricted to monocytes and liver-resident Küppfer cells [20], other studies using a different antibody clone (17D9), or the fusion protein PDPN-Fc, reported that CLEC-2 is constitutively expressed by CD11b high Gr-1 high cells isolated from bone marrow (BM) and whole blood, splenic B lymphocytes, a small subset of splenic natural killer (NK) cells, splenic plasmacytoid dendritic cells (pDCs), splenic conventional DCs (cDCs), GM-CSF stimulated BM-derived DCs (BMDCs), Flt3L BMDCs, as well as peripheral LN DCs [19,23,24]. With the exception of NKT cells and T lymphocytes, in vivo LPS challenge has been reported to upregulate CLEC-2 expression in almost all splenic leucocyte subsets as well as peripheral LN DCs [23,24]. In a thioglycolate-induced peritoneal inflammation model, CLEC-2 expression was observed in F4/80 + macrophages but not in CD11b high Gr-1 high cells [19,23]. Notably, CLEC-2-deficient negative control cells were not included in most of these studies [19,23]. Our study aimed to clarify these contradictory findings and improve our understanding of CLEC-2 expression on mouse leucocytes. These results have important physiological consequences that will be discussed below.

Peripheral blood B lymphocytes and CD11b high Gr-1 high cells present CLEC-2 on their surface
Previous studies that investigated the temporal, spatial, and proinflammatory expression of CLEC-2 in the murine adult hematopoietic system have been hampered by the high neonatal mortality rate (>95%) of Clec1b −/− mice [10,20], impeding the inclusion of appropriate Clec1b −/− negative control cells in previous studies aiming to define the temporal, spatial, and postinflammatory expression of CLEC-2 in vivo [19,23,24].
In parallel, we investigated CLEC-2 expression on hematopoietic cells isolated from lethally irradiated wild-type (WT) adult mice reconstituted with foetal liver (FL) cells from E14.5 Clec1b +/+ or Clec1b −/− embryos [25]. This second experimental strategy was used to rule out potential side effects of tamoxifen on CLEC-2 expression. It is known that sex steroid hormones and their synthetic derivatives (such as tamoxifen) affect hematopoiesis due to the presence of estrogen receptors on most immune cells [26,27]. Moreover, tamoxifen has anti-inflammatory effects that could counteract LPS-mediated proinflammatory challenges [28][29][30]. In addition, we used two different antibody clones, 17D9 [19,23] and INU1 [31], reported to bind to mouse CLEC-2.
Initially, CLEC-2 expression was measured on circulating platelets, T lymphocytes, B lymphocytes, and CD11b high Gr-1 high cells from Clec1b fl/fl xRosa26 +/creERT2 mice and Clec1b fl/fl littermates by flow cytometry using the two antibody clones 17D9 and INU1 ( Fig. 1A and Supporting Information Fig. 2). Following tamoxifen treatment, Clec1b fl/fl xRosa26 +/creERT2 platelets showed full abrogation of CLEC-2 expression compared to Clec1b fl/fl littermates using both 17D9 and INU1 (Fig. 1A), confirming the efficiency of our inducible genetic mouse model.
On platelets, INU1 staining was found to be weaker than 17D9 staining in both control animals (Clec1b fl/fl mice treated with tamoxifen and Clec1b fl/fl xRosa26 +/creERT2 mice fed with normal diet). Furthermore, the geometric mean fluorescence intensity associated with 17D9 binding to leucocytes was on average threefold lower than that observed on platelets (Fig. 1A), while INU1 discrimination power was too weak for detecting CLEC-2 on leucocytes (Fig. 1A). As a result, we solely used the 17D9 clone to further investigate CLEC-2 expression on leucocytes.
In both the tamoxifen-inducible and radiation chimeric CLEC-2-deficient mouse models, the levels of 17D9 binding to circulating B lymphocytes were significantly reduced compared to controls ( Fig. 1A and B), suggesting that peripheral blood B lymphocytes constitutively express CLEC-2. Although CLEC-2 appeared to be downregulated on circulating B lymphocytes following LPS treatment in chimeric mice, this was not statistically significant ( Fig.  1B), indicating that activated Clec1b +/+ B lymphocytes remain positive for CLEC-2 when compared to their activated Clec1b −/− B lymphocyte counterparts.
In the absence of LPS-induced inflammation, CD11b high Gr-1 high cells (which includes a mix of monocytes, granulocytes/neutrophils, and a small subset of NK cells [32,33]  Sixteen to eighteen hours post LPS injection, blood was harvested by full exsanguination and processed as described above. Each symbol represents a sample from an individual mouse. Bars represent the means. The graphs summarize one to three independent experiments pooled together. Statistical significance was measured by a Mann-Whitney test with a 95% confidence interval, where: *p < 0.05; **p < 0.005; ***p < 0.0005; N.S., not significant. isolated from Clec1b −/− reconstituted animals were negative for CLEC-2 compared to Clec1b +/+ littermates (Fig. 1B). This demonstrates that circulating CD11b high Gr-1 high cells constitutively express CLEC-2 in line with previous reports [19,23]. However, treatment with tamoxifen led to the loss of specific CLEC-2 staining, since no significant difference was observed between Clec1b fl/fl xRosa26 +/creERT2 mice and their controls (Fig.  1A). Tamoxifen is known to inhibit B lymphocyte and DC maturation by altering the surface membrane expression of molecules such as CD22 on B lymphocytes or MHC-II and CD86 on BMDCs [34,35]. Additional uncharacterized phenotypic changes leading to the appearance of new CLEC-2-independent binding sites might explain the nonspecific binding of 17D9 to Clec1b fl/fl xRosa26 +/creERT2 B lymphocytes and CD11b high Gr-1 high cells as the WT Clec1b DNA was undetectable in these cells ( Similarly, after LPS challenge, 17D9 acquired the ability to bind to Clec1b −/− CD11b high Gr-1 high cells leading to a 3.7-fold higher geometric mean fluorescence intensity than in equivalent unstimulated cells (Fig. 1B). This suggests that 17D9 binds to CD11b high Gr-1 high cells in a CLEC-2-independent manner following LPS activation. In T lymphocytes, there was no evidence for 17D9 binding in any of our experimental conditions (Supporting Information Fig. 3).
From these findings, we suggest that mouse peripheral blood B lymphocytes and CD11b high Gr-1 high cells present CLEC-2 on their surface at a much lower level than platelets. These data contrast with the lack of evidence for CLEC-2 expression in peripheral blood leucocytes in chickens or humans [18,21] and with the absence of significant Clec1b transcripts in human leucocyte subsets according to microarray analyses on the BioGPS database [21,36]. This indicates that, while CLEC-2 expression by platelets is conserved through species, the presence of CLEC-2 on B lymphocytes and CD11b high Gr-1 high cells is specific to mice. Whether the presence of CLEC-2 on these cells provides any particular features to the mouse immunological system remains unknown. Our data add to previous reports establishing important differences between the mouse and human immune systems [37]. Each symbol represents a sample from an individual mouse and bars represent means. The graphs summarize three independent experiments pooled together. (E) Relative expression of Clec1b transcript in leucocytes isolated from the spleen and the MLN. C57BL/6 mice were injected i.p. with PBS (white bars) or 25 μg LPS (black bars). Sixteen and eighteen hours later the spleen and MLN were harvested, the erythrocytes were lysed and the remaining leucocytes stained. Leucocyte populations were isolated by FACS based on the following phenotypes: T cells (T): DAPI neg CD41 neg F4/80 neg CD11c neg CD19 neg B220 neg CD3ε pos CD8α/CD4 pos ; B cells (B): DAPI neg CD41 neg F4/80 neg CD11c neg CD8α neg CD3ε neg B220 pos CD19 pos ; NK cells (NK): DAPI neg CD41 neg CD19 neg CD8α neg CD4 neg CD3ε neg NK1.1/NKp46 pos ; CD11b pos F4/80 neg cells: DAPI neg CD41 neg CD19 neg CD3ε neg F4/80 neg CD11b pos ; F4/80 pos CD11b int cells: DAPI neg CD41 neg CD19 neg CD3ε neg CD11b int F4/80 pos ; cDC CD11b neg/int cells: DAPI neg CD41 neg CD19 neg CD3ε neg B220 neg CD11c pos CD11b neg/int ; cDC CD11b high cells: DAPI neg CD41 neg CD19 neg CD3ε neg B220 neg CD11c pos CD11b high . After mRNA isolation and cDNA preamplification for the genes of interest, the samples were analyzed by quantitative PCR. The signal for Clec1b was normalized against the house-keeping gene Actnb (β-actin). Total BM isolated from PBS-injected mice was used as positive control while the T cells coming from these animals were used as reference to set the arbitrary unit. Each population was isolated from three to five independent cell sorting experiments including one LPS-injected and one PBS-injected mouse for each cell sorting, with the exception of the F4/80 pos CD11b int control and LPS-stimulated cells for which n = 2 experiments. Data are shown as mean + SEM. Statistical significance was measured by a Mann-Whitney test with a 95% confidence interval where: *p < 0.05; **p < 0.005; ***p < 0.0005; N.S., not significant.

Most SLO-resident leucocytes do not express CLEC-2/Clec1b at steady state
At steady state, we observed comparable 17D9 binding to spleen and mesenteric LN (MLN) B lymphocytes, CD11b high Gr-1 high cells, pDCs, CD11b neg/int cDCs, and CD11b high cDCs isolated from Clec1b −/− radiation chimeras or their WT counterparts (Fig. 2,  Supporting Information Figs. 4 and 5). These results, that demonstrate a lack of CLEC-2 expression on these leucocytes, contradict previous observations made using the 17D9 antibody clone and PDPN-Fc recombinant protein that suggested that mouse CLEC-2 was constitutively expressed on all these cell types in spleen and on peripheral LNs' cDCs [23,24]. Moreover, we could not detect significant Clec1b transcript levels in most of the leucocyte populations isolated from the spleen and the MLN (Fig. 2E). The absence of Clec1b transcripts in most resting leucocytes is supported by two independent microarray analyses performed by the ImmGen [38] and BioGPS [36] consortia (Supporting Information Fig. 6).
Interestingly, in agreement with the ImmGen database, we did observe that CD11b int F4/80 pos red pulp splenic macrophages express high levels of Clec1b transcripts (Fig. 2E). However, we were unable to detect surface CLEC-2 on the surface of these cells when comparing our FL-reconstituted animals (data not shown). Given that F4/80 pos red pulp splenic macrophages, which express PDPN, play a key physiological role in the clearance of senescent blood erythrocytes and platelets by phagocytosis [6,39,40], the Clec1b transcripts detected in these cells are likely to derive from engulfed platelets.
These results suggest that CLEC-2 surface expression by peripheral blood B lymphocytes and CD11b high Gr-1 high cells (Fig. 1) is lost upon entry into SLOs (Fig. 2). The majority of mouse B lymphocytes and pre-cDC monocytes migrating from peripheral blood toward LNs enter via high endothelial venules (HEVs), arriving in the T-cell zone, which is rich in PDPN pos fibroblastic reticular cells (FRCs) [41]. Downregulation or shedding of CLEC-2 by B lymphocytes and CD11b high Gr-1 high cells during their entry through the HEVs might represent a functional mechanism to prevent inappropriate activation of PDPN pos FRCs in the absence of infection. In this context, recent studies demonstrated that close interactions between antigen-activated CLEC-2 pos DCs and LN PDPN pos FRCs are required for mounting an effective immune response by favoring DC recruitment, FRCs activation and LN swelling [14,15,24].

Most SLO-resident leucocytes remain CLEC-2-negative following LPS stimulation
CLEC-2 was not upregulated on splenic and MLN-resident B lymphocytes, CD11b high Gr-1 high cells, pDCs, or CD11b neg/int cDCs in response to intraperitoneal (i.p.) LPS challenge of Clec1b −/− and Clec1b +/+ reconstituted animals ( Fig. 2A-C and Supporting Information Fig. 5). It has been shown that B lymphocyte stimulation via LPS/TLR4 favors their emigration from the blood into SLOs [42]. Interestingly, we could not observe any CLEC-2 pos B lymphocytes in the SLOs of LPS-stimulated mice (Fig. 2), supporting the observation that circulating activated B lymphocytes downregulate CLEC-2 before entering into SLOs (Fig. 1B). Kerrigan and collaborators have suggested the same CLEC-2 downregulation process by circulating CD11b high Gr-1 high cells upon reaching inflammatory sites [19]. As entry to both SLOs and inflammatory sites requires leucocyte rolling, arrest, and transendothelial migration [41], it is tempting to suggest that CLEC-2 downregulation or shedding by these leucocytes facilitates the completion of this three-step mechanism. Indeed, it has been shown that the shedding of transmembrane molecules is essential for leucocyte transendothelial migration [43][44][45][46]. We hypothesize that CLEC-2 could be lost via the same mechanisms.
CD11b high Gr-1 high cells, pDCs, and CD11b neg/int cDCs isolated from LPS-stimulated Clec1b −/− mice exhibited a higher level of 17D9 binding than Clec1b −/− nonchallenged counterparts. Once again, these results indicate that the 17D9 antibody clone has the capacity to bind LPS-stimulated leucocytes in a CLEC-2independent manner. The lack of Clec1b transcript upregulation in LPS-stimulated leucocytes (Fig. 2E) provides further evidence to support this conclusion and challenges previously described LPSinduced CLEC-2 upregulation by most splenic resident leucocytes and peripheral LN cDCs [23,24].

MLN but not splenic CD11b high cDCs acquire CLEC-2 following LPS stimulation
In agreement with the results detailed above, CLEC-2 expression was not detected on splenic CD11b high cDCs. However, a modest but significant increase in staining of MLN-derived Clec1b +/+ CD11b high cDCs compared to Clec1b −/− controls was observed after LPS injection (Fig. 2D), indicating that LPS-stimulated MLN CD11b high cDCs have the capacity to upregulate CLEC-2 (Fig. 2D). However, we could not correlate the appearance of CLEC-2 on the membrane with a higher relative amount of Clec1b transcripts in the stimulated CD11b high cDCs (Fig. 2E), suggesting that LPS stimulation regulates CLEC-2 expression in MLN CD11b high cDCs via posttranscriptional mechanisms.
Taken together, our results confirmed high levels of CLEC-2 expression on splenic platelets (data not shown), while no significant expression of CLEC-2 was observed on most leucocyte populations investigated, both at steady state and after LPS injection. However, we did observe an increase in CLEC-2 expression on activated CD11b high cDCs isolated from the MLN. This increase was absent on splenic-activated cDCs.
The existence of cell-specific and tissue-specific regulation of CLEC-2 expression has previously been observed in the context of human rheumatoid arthritis, a chronic inflammatory disease where CLEC-2 expression was found to be restricted to tissue infiltrating platelets, while absent from activated DCs [22]. In contrast, FITC skin painting, FITC footpad immunization, or OVA/CFA subcutaneous injections in mice were found to contribute to the generation of an immune response that relies on CLEC-2 expression by activated DCs and their interaction with PDPN pos lymphatic endothelial cells and PDPN pos FRCs [14,15,24]. These studies www.eji-journal.eu demonstrate that CLEC-2 upregulation is a characteristic of locally activated DCs migrating toward the draining LN [14,15,24] and not the systemic feature of an activated immune system [23]. In accordance with this, we only observed CLEC-2 upregulation by activated DCs in the MLN, a SLO close to the LPS site of administration (i.e. the peritoneal cavity), but not in a remote SLO such as the spleen. This suggests that the MLN CLEC-2 pos CD11b high cDCs we observed may have recently migrated from the surrounding mesenteric tissue following LPS administration.

Normal lymphocyte homeostasis in B cell Clec1b −/− mice as B cells do not produce CLEC-2
To gain insight into the potential physiological roles of CLEC-2 in B lymphocytes, we first visualized the LNs from B lymphocytedeficient Jh −/− κ −/− mice and saw no evidence of erythrocyte infiltration in the LNs, similar to WT animals (Fig. 3A), indicating that CLEC-2 on B lymphocytes does not play an important role in blood-lymph separation or the maintenance of HEVs.
In order to investigate if CLEC-2 on peripheral blood B lymphocytes contributes to lymphocyte homeostasis, we generated mice with a Clec1b −/− deficiency restricted to the B-cell lineage by mixing Clec1b −/− FL and Jh −/− κ −/− BM cell suspensions (at a 1:9 ratio) that we injected into lethally irradiated C57BL/6 recipients (Supporting Information Fig. 7A) [47]. The absolute numbers of lymphocytes in the blood, spleen, and MLNs were monitored ( Fig.  3B and C, Supporting Information Fig. 7B and C). Both in the blood and the SLOs, the absolute numbers of B and T lymphocytes were normal. In the spleen, the numbers of follicular, nonfollicular, and marginal zone Clec1b −/− B lymphocytes were comparable to the controls. Contrary to the MLNs, a small increase in CD4 + T lymphocytes was observed in the spleen of Clec1b −/− animals. However, no-significant increase in naïve (CD62L hi CD44 int ) or activated (CD44 hi CD62L − ) CD4 + T lymphocytes was noted in these animals. Similarly, the CD4 − T lymphocytes showed the same level of activation between Clec1b −/− and control animals. These results indicate that deletion of the Clec1b gene in the B-cell lineage has no effect on lymphocyte homeostasis.
Despite the absence of a functional Clec1b gene, peripheral blood Clec1b −/− B lymphocytes were stained by 17D9 at the same level as Clec1b +/+ B lymphocytes (Fig. 3D). In B lymphocyte Clec1b-deficient mice, 71% of the platelets were CLEC-2 pos on average (Fig. 3D), while the level of CLEC-2 expression by CD11b high Gr-1 high cells was comparable to that found in controls (Supporting Information Fig. 7D). We compared the amount of Clec1b transcripts by quantitative PCR in B lymphocytes isolated from the peripheral blood, MLNs, and spleen. In all three populations, the amount of Clec1b was extremely low and at a comparable level (Fig. 3E). This last result corroborates a study on rat B lymphocytes showing that Clec1b transcripts are hardly detectable in these cells [48]. From these observations, we conclude that peripheral blood B lymphocytes do not intrinsically express CLEC-2. Instead we propose that the CLEC-2 molecules detected on the surface of circulating B lymphocytes are derived from MHC-II antigen presentation, trogocytosis, or exosomes/microparticles attached to the B-cell membrane [49].

Concluding remarks
Our study confirms that CLEC-2 is constitutively expressed by mouse platelets and circulating CD11b high Gr-1 high myeloid cells and shows for the first time that CLEC-2 is present on the surface of peripheral blood B lymphocytes. These B cells do not produce CLEC-2 but likely acquire CLEC-2 molecules from other yet uncharacterized cell types. Our data suggest that both circulating B lymphocytes and CD11b high Gr-1 high myeloid cells lose CLEC-2 when entering SLOs. This loss of CLEC-2 might depend on the same mechanisms that are selectively shedding CD23 and CD62L from leucocytes during transendothelial migration [43,45,46]. As CLEC-2 stimulates PDPN pos FRCs in SLOs in order to mount a proper immune response [14,15,24], we propose that the loss of CLEC-2 by naïve B lymphocytes and CD11b high Gr1 high myeloid cells entering in the SLOs might be a prerequisite to prevent untimely FRCs activation in absence of antigenic challenge.
The use of animals reconstituted with Clec1b −/− FL and the measurements of Clec1b transcripts allowed us to rule out any constitutive CLEC-2 expression by most of the leucocyte subpopulations isolated from SLOs. We also demonstrated that isotype controls are not adequate when working with the 17D9 antibody clone. Finally, we showed that LPS peritoneal injection induces CLEC-2 acquisition to the unique MLN-activated DCs leucocyte population. Taken together with other studies, our findings emphasize the notion that the expression of CLEC-2 is not only restricted to specific subsets of resting leucocytes and platelets but is determined both by the activation state and anatomical site where immune responses take place.

Mice and diets
Clec1b +/− [25], Clec1b fl/fl [25], Rosa26 +/ERT2cre (Jackson Laboratory, ME) [50], Clec1b fl/fl x Rosa26 +/ERT2cre , Jh −/− κ −/− [51,52], and BoyJ mice were maintained in the Biomedical Services Unit, University of Birmingham. C57BL/6 mice were purchased from Harlan, UK. Animals were fed with FormulaLab Diet 5008 (Lab-Diet, St-Louis, MO). When required, 6-to 8-week-old Clec1b fl/fl × Rosa26 +/ERT2cre and their Clec1b fl/fl x Rosa26 +/+ control littermates were continuously fed with tamoxifen-supplemented diet TAM 400 (Harlan, UK). For isolation of embryonic FL, the morning of vaginal plug detection was designated day 0.5 of gestation. Animal experiments were performed in accordance with UK Home Office legislation.  Seven to nine weeks later, the blood, spleen, and MLNs were isolated and analyzed by flow cytometry to determine (B and C) leucocyte numbers and (D) CLEC-2 expression was assessed using the 17D9 antibody compared to its respective isotype control. Each symbol represents an individual mouse and bars represent means. The graphs summarize two independent experiments pooled together. (E) Platelets and B lymphocytes were isolated from the blood, MLN, and the spleen of C57BL/6 mice by FACS, based on the following phenotypes: platelets: FSC low CD41 pos ; B cells: FSC hi CD41 neg DAPI neg CD3ε neg CD19 pos . After mRNA isolation and cDNA synthesis, the relative expression of Clec1b transcripts was analyzed by quantitative PCR. The signals for Clec1b was normalized against the house-keeping gene Actnb (β-actin). Total BM was used as positive control and reference to set the arbitrary unit. Each population was isolated from three independent cell sorting experiments, including one to two mice for each cell sorting, and one quantitative PCR was ran from each cell sorting. The graphs summarize these three independent quantitative PCRs pooled together and data are shown as mean + SEM. Statistical significance was measured by a Mann-Whitney test with a 95% confidence interval where: *p < 0.05; **p < 0.005; ***p < 0.0005; N.S, not significant. www.eji-journal.eu Mouse hematopoietic system reconstitution C57BL/6 or BoyJ mice (8-10 weeks old) were given Baytril in the drinking water for 7 days prior to irradiations with two doses of 450rad, 3 h apart. One hour after the last irradiation Clec1b +/+ or Clec1b −/− E14.5 FL cells were injected intravenously (i.v.). For the generation of B-cell Clec1b-deficient recipients, 2 × 10 5 Clec1b +/+ or Clec1b −/− E14.5 FL cells were mixed with 18 × 10 5 Jh −/− κ −/− BM cells. Mice were left for 6-to 8-weeks postinjection before analysis or further challenged by i.p. injection of 25 μg of LPS (Chondrex) diluted in PBS or PBS only. Mice were analyzed 16-18 h post LPS or PBS injection. Successful LPS injections were confirmed by ࣙ5% weight loss over this period.

Tissue sampling and preparation
In all cases, cell centrifugation was performed at the average force of 275 g for 4 min. Blood was sampled from the tail vein of Clec1b fl/fl ; Rosa26 +/ERT2cre ; or Clec1b fl/fl × Rosa26 +/ERT2cre mice into 20 mM EDTA/PBS 1× solution. After centrifugation, the supernatant was removed and the pellet resuspended in red blood cell lysis buffer (Sigma-Aldrich) at room temperature for 5 min. Samples were centrifuged and resuspended in cold PBS 1×, 2% foetal calf serum (FCS), 2 mM EDTA solution and stained for flow cytometry analysis or processed for genomic DNA extraction.
Whole blood from FL reconstituted animals was drawn into acid citrate dextrose solution (9:1 v/v) from the inferior vena cava under isofluorane anesthesia and mixed to 20 mM EDTA/PBS 1× solution. An aliquot of blood was centrifuged and processed as described above.
The spleen and the MLN were harvested into cold 2% FCS RPMI solution (Sigma-Aldrich). For cell sorting of T cells, B cells, and NK cells, part of the spleen was mechanically dissociated on a 100 mm mesh (Greiner Bio-one). In all the other cases, spleen or MLN were teased apart with dissection forceps in a 2% FCS RPMI solution containing 2.5 mg/mL of Collagenase D (Roche) and 2 mg/mL of DnaseI (Sigma-Aldrich). Cell suspensions were kept under magnetic stirring at 37°C for 45 min before centrifugation. Pellets were resuspended in a 2% FCS RPMI solution containing 2.5 mg/mL of Collagenase/Dispase (Roche) and 2 mg/mL of Dna-seI (Sigma-Aldrich) and kept under magnetic stirring at 37°C for 30 min. Cell suspensions were adjusted to a final EDTA concentration of 5 mM by the addition of a 0.5 M EDTA solution and kept under magnetic stirring at 37°C for 5 min. Cell suspensions were centrifuged and the pellets resuspended in red blood cell lysis buffer (Sigma-Aldrich) as described above before processing for FACS staining.

Antibodies, FACS analysis, and cell sorting
The full list of antibodies used is provided in Supporting Information Table 1. Anti-mouse CLEC-2-FITC 17D9 clone was mainly obtained from a commercial provider (AbD Serotec, 17D9) and compared to rat IgG2b-FITC (AbD Serotec, MCA1125FT). For some experiments, purified 17D9 (a kind gift from Caetano Reis e Sousa, Cancer Research UK, London) and purified rat IgG2b (R&D Systems) were used. Purified anti-mouse CLEC-2 INU1 clone (a kind gift from Bernhard Nieswandt, University of Würzburg, Germany) was used in conjunction with purified rat IgG1k (Biolegend, 400402). All purified antibodies were conjugated to AlexaFluor R 488 using a monoclonal antibody labeling kit (Invitrogen).
Cells were stained with antibodies at 4°C in the dark in PBS 1×, 2% FCS, 2 mM EDTA solution. Cells were washed and centrifuged twice before being resuspended in a cold a PBS 1×, 2% FCS, 2 mM EDTA, 1 mg/mL DAPI solution. Flow cytometry acquisitions were performed on a three laser (405, 488, 633 nm) Cyan (Beckman Coulter) using Summit v4.3 software (Beckman Coulter). Flow cytometry cell sorting was performed on a MoFlo highspeed cell sorter and Astrios cell sorter (Beckman Coulter) using Summit software (Beckman Coulter). FACS data were analyzed with FlowJo software 8.7 (Tristar).

Statistical analysis
All statistical analyses were performed on Prism v4.0 (GraphPad Software, CA) using two-tailed Mann-Whitney tests with 95% confidence interval.