Ly49B Is Expressed on Multiple Subpopulations of Myeloid Cells¹

Since their discovery in 1983 (1), Ly49 receptors have been the subject of intense interest. It was eventually realized that they are encoded by a series of closely related genes (2) clustered at the telomeric end of the NK gene complex, a region of ∼2 Mb on chromosome 6 that contains several families of genes expressed predominantly in NK cells (3, 4, 5). The Ly49 polypeptides are type II transmembrane proteins belonging to the C-type lectin superfamily and comprise two subclasses, one of which possesses an ITIM in the cytoplasmic domain and delivers inhibitory signals following cross-linking by MHC class I ligands, while the other lacks an ITIM, has a charged residue in the transmembrane domain, and delivers activatory signals via the adaptor protein DAP12 following interaction with ligands that may include MHC class I molecules and virus-encoded class I homologs (5). Most of the Ly49 receptors that have been studied are expressed predominantly on NK cells, although the inhibitory Ly49s can also be expressed in some circumstances on T cells (6). The only member of the Ly49 family in C57 mice that has so far not been studied is Ly49B (2). The gene encoding Ly49B is separated by >0.8 Mb from the cluster containing the remaining 10 functional Ly49 genes (4). It has an intron-exon structure similar to that of other Ly49 genes but appears to lack the upstream Pro1 element responsible for initiating the stochastic expression of Ly49 receptors in NK cells (7). Ly49B cDNAs have been obtained from several tissue sources and predict a polypeptide that lacks a transmembrane charge and bears an ITIM in the cytoplasmic domain but which is significantly divergent from other Ly49s. Interestingly, Ly49B is the only member of the mouse Ly49 family that has an identifiable ortholog in the rat (8, 9), suggesting it may have a distinctive and conserved function in murine species. We have now generated a series of mAbs that specifically recognize Ly49B and report here that Ly49B is not normally expressed on NK cells but instead is expressed on multiple subpopulations of myeloid cells that include granulocytes and macrophages.

PVG female rats and C57BL/6 and BALB/c female mice were obtained from Harlan-Olac, and all procedures were approved by the U.K. licensing authorities. Except where stated, 6- to 16-wk-old C57BL/6 mice were used as cell donors. Spleen cells were obtained by teasing of spleens with forceps, bone marrow (BM)³ cells by passing material eluted from femurs and tibias through narrow gauge needles, and peritoneal cells by flushing the peritoneum with PBS. In some experiments, animals were injected i.p. with 1 ml of a 10% solution of thioglycollate broth.

The culture medium used was DMEM (52100-039; Invitrogen Life Technologies) made up in highly purified water and supplemented with 2× nonessential amino acids, 5 × 10⁻⁵ M 2-ME, and 10% FBS (F-7524; Sigma-Aldrich). Spleen and BM macrophages were obtained by culturing aliquots of respectively 50 or 5 million nucleated cells in 90-mm bacteriological grade petri dishes containing 10 ml of medium supplemented with 10% heat-inactivated horse serum and 10% supernatant from L929 cells (provided by Dr. J. H. Robinson, Newcastle University, Newcastle, U.K.) used as a source of M-CSF. BM dendritic cells (DC) were obtained by culturing aliquots of 1–2 million cells in 25-cm² culture flasks with 5 ml of medium containing 10 ng/ml mouse GM-CSF (PeproTech) or aliquots of 5 million cells with 200 ng/ml mouse flt3 ligand (FL; PeproTech). In all cases, cultures were refed after 4–5 days, and both nonadherent and adherent cells were collected, the latter with the aid of Ca- and Mg-free PBS containing 0.5 mM EDTA. The cells were then washed and aliquots of 0.5–1 million cells cultured for 1–2 days in low adherence 24-well plates (Corning) in 1 ml of medium containing LPS at 1 μg/ml, γ-IFN (PeproTech) at 300 ng/ml, or rIFNαA/D (provided by Dr. M. Brunda, National Institutes of Health, Bethesda, MD) at 10⁴ U/ml. Cultures of macrophages and GM-CSF-derived DC were also supplemented with 5% L929 supernatant or 10 ng/ml GM-CSF, respectively. Nonadherent and adherent cells were harvested as above.

Cultured lymphocyte populations were generated by incubating aliquots of 1–2 million spleen cells for 5–7 days in 24-well plates in 2 ml of medium containing 2 μg/ml LPS (L3129; Sigma-Aldrich) plus 3.5 ng/ml IL-2 to activate B cells, 2 μg/ml Con A (C7275; Sigma-Aldrich) plus 3.5 ng/ml IL-2 to activate T cells, or 350 ng/ml human IL-2 with or without 100 pg/ml mouse IL-12 (provided by Dr. M. Gately, Genetics Institute, Boston, MA) plus 50 ng/ml mouse IL-18 (R&D Systems) to activate NK cells and T cells. Immature NK cells were generated from day 14 fetal thymocytes as previously described (10), and long-term lines of cloned immature NK cells were maintained in medium containing 3.5 ng/ml IL-2. The LNK line (11) was provided by Dr. S. Anderson (National Cancer Institute, Frederick, MD). Ly49-transfected YB2/0 and RNK cells were generated as described previously (12).

mAbs against Ly49B were generated by repeatedly immunizing PVG rats with RNK cells expressing a C57 Ly49B molecule bearing a hemagglutinin (HA) tag at its C terminus. Following a final i.v. injection, spleen cells were prepared 3 days later and fused with NS0 cells according to standard protocols. Several hybridomas that secreted mAbs that stained Ly49BHA-transfected but not untransfected RNK cells were identified and recloned three times. One of these, 1A1, secreted an IgG1,k mAb, which was purified on a protein G column and conjugated to biotin and AlexaFluor 647 using kits obtained from Molecular Probes. A rat IgG1 isotype control mAb (BD Biosciences) was similarly conjugated. The N34 rat IgG2a anti-Ly49Q mAb (13) purified on protein G and a rat IgG2a isotype control mAb (BD Biosciences) were conjugated in the same way. All other mAbs and streptavidin conjugates used in flow cytometric studies were purchased from BD Biosciences, except for those against Dectin-1 (14) and 7/4 (15) generated by ourselves, 83101 anti-CCR3 from R&D Systems, and the 2E6 anti-Ly49Q mAb (16) from MBL.

Cells at 5–20 million/ml in HBSS (61200-093; Invitrogen Life Technologies) containing no bicarbonate and supplemented with 2% FBS and 0.2% sodium azide were stained at room temperature for 20 min with appropriate combinations of reagents at predetermined optimal concentrations. Before staining, cells were incubated with a saturating concentration of 2.4G2 anti-CD16/32 mAb to block FcR. Staining was analyzed on LSR2 or FACScan instruments (BD Biosciences), using forward and side scatter to gate on single viable cells. Compensation for spectral overlap of dyes was set using mixtures of unstained and single-stained cells. Data were analyzed using FCS Express V2 software (DeNovo Software). Cell sorting was performed on a FACSDiva instrument (BD Biosciences).

RT-PCR was performed as previously described (12) using primers that amplified the complete coding sequence of Ly49B (primer set 1, fwd = atactcgagcATGAGTGAGCAGGAGGTCA, rev = atatctagatcaggcgtagtcgggcacgtcgtaggggTAACTTTCATCTTCATCCCTCTGC), exons 1–3 of Ly49B (primer set 2, fwd = CCCTTCACCAGAATTACTT, rev = CACCAACACTGCCACAGTT), the complete coding sequence of Ly49Q (fwd = CTTTTCCCCCATAACTGCAG, rev = TGACCTCTTCCAGCTCTTCC), or part of β-actin (fwd = ATCCTGACCCTGAAGTACCC, rev = TACTCCTGCTTGCTGATCC). The Ly49B full-length primers are the ones used to generate the HA-tagged Ly49B construct used to transfect the immunizing RNK cells, with lowercase letters corresponding to enzyme sites and the HA sequence.

Cytospin preparations of sorted cell populations were air-dried, fixed, and stained using Diff-Quik modified Giemsa reagent (Baxter) and mounted in DPX. For histological studies, frozen sections (5 μm) were fixed for 5 min at 4°C in ethanol, then incubated with anti-Ly49B mAb or with mAbs to sialoadhesin (CD169), macrosialin (CD68), F4/80, and Dectin-1 (all obtained from Serotec) and appropriate isotype controls (BD Biosciences) at 10 μg/ml, followed by mouse absorbed biotinylated anti-rat Ig (Vector Laboratories). For CD11c, a biotinylated mAb (clone HL3) from BD Biosciences was used. mAb labeling was developed with the avidin-biotin complex-HRP kit and NovaRed substrate, followed by hematoxylin according to the manufacturer’s instructions (Vector Laboratories). Double immunolabeling was performed with anti-Ly49B mAb plus AlexaFluor 488 anti-rat Ig together with biotin anti-CD11c and AlexaFluor 568 streptavidin (Molecular Probes) or the CR-Fc reagent, a lectin specific for sulfated glycoproteins predominantly expressed by metallophilic macrophages in the spleen and subcapsular sinus macrophages in lymph nodes (17) (donated by Dr. L. Martinez-Pomares, University of Oxford, Oxford, U.K.) +AlexaFluor 568 anti-human Ig. Specimens were examined on a Leica TCS NT confocal system.

RNK cells transfected with Ly49B bearing a C-terminal FLAG tag were collected with ice-cold PBS and suspended at 2 × 10⁶ cells/75 μl in cold PBS containing 5% FBS. After addition of 25 μl of 4 mM Na₃VO₄ (pervanadate; Sigma-Aldrich), the cells were incubated at 37°C for 5 min, followed by the addition of 1 ml of 1% Nonidet P-40 lysis buffer containing 100 mM NaF and 1 mM pervanadate. The cell lysates were incubated with agarose beads conjugated with anti-FLAG Ab M2 (Upstate Biotechnology), and the immunoprecipitates were resolved on 5–20% SDS-PAGE then transferred to polyvinylidene difluoride membrane. Proteins were detected by incubating the membranes with biotinylated Abs against FLAG (Upstate Biotechnology), Src homology region 2 domain-containing phosphatase (SHP)-1, SHP-2, and SHIP (Santa Cruz Biotechnology), followed by HRP-conjugated streptavidin, and visualized by SuperSignal West Dura Extended Duration Substrate (Pierce).

These were performed as described by others (18, 19). Briefly, RNK cells transfected with wild-type Ly49B or Ly49B bearing a C-terminal HA tag were cultured in DMEM supplemented with 20% FBS and for the last day before assay also with 10⁴ IU/ml IL-2. Titrated doses of these cells were then incubated in 96-well V-bottom plates for ∼15 min at room temperature with/without control or anti-Ly49B mAbs. Cr-labeled target cells (5000) were then added, and the plates spun down and incubated at 37°C for 4–5 h before collecting and counting aliquots of supernatant. The target cells used were the mouse macrophage lines P388D1 and J774, and the mouse B cell lymphoma A20, which have been maintained in our laboratory for many years by routine subculture in DMEM containing 5% FBS.

Candidate anti-Ly49B mAbs were tested for specificity by staining cells transfected with each of the Ly49 molecules known to be expressed in C57 mice. Several mAbs specific for Ly49B were obtained. The mAb 1A1 strongly stained cells transfected with the C57 Ly49B molecule but showed no detectable staining of cells transfected with Ly49A, C, D, E, F, G, H, I, or Q derived from C57 mice (Fig. 1). It also failed to stain cells transfected with an HA-tagged BALB/c Ly49B molecule that was efficiently stained by anti-HA Ab and with several other Ly49B-specific mAbs that we generated.

Splenic (and also BM, data not shown) B and T cells, both fresh and following activation in vitro, showed no detectable surface expression of Ly49B, the histograms of cells stained with 1A1, isotype control mAb, or medium alone (this last not shown) being almost exactly superimposable (Fig. 2, A and B). Surprisingly, adult splenic and BM NK cells also showed no detectable staining for Ly49B (Fig. 2,A), even after culture for 7 days in IL-2 (Fig. 2,C) or IL-15 (data not shown), the control and Ly49B histograms again being almost exactly superimposable. The inclusion of IL-4, IL-12, IL-18, IL-21, α-IFN, γ-IFN, or LPS in the 7-day IL2-containing cultures of spleen cells failed to induce detectable Ly49B expression (data not shown). However, the inclusion of both IL-12 and IL-18 caused NK cells but not “NKT” or T cells to express low but consistently detectable levels of Ly49B (Fig. 2,D). Low but clear levels of Ly49B expression were also found on the immature NK cell line I2/22 derived from fetal thymic progenitors (Fig. 2 E).

To investigate further the expression of Ly49B in lymphocytes, RT-PCR analysis was performed on RNA extracted from highly purified cells. In the first approach, spleen cells were stained with Abs to NK1.1, CD3, F4/80, and Gr1 and NK1.1⁺CD3⁻F4/80⁻Gr1⁻ cells purified by cell sorting. Following 5 days of culture in IL-2, NK1.1⁺CD3⁻F4/80⁻Gr1⁻ cells were purified again by a second round of cell sorting and returned to culture for an additional 7 days. RNA was prepared from these cells, >99.5% of which expressed NK1.1 and which also lacked any detectable CD3⁺, F4/80⁺, or Gr1⁺ cells. RT-PCR analysis was performed using two sets of Ly49B primers: primer set 1 whose primer binding sites overlapped the translation start and stop sites and which therefore amplified the entire coding sequence, and primer set 2 whose target sites lay in exons 1 and 3. Using RNA prepared from highly purified adult NK cells, the full-length primer set 1 failed to yield detectable product, whereas the 5′ primer set 2 consistently yielded a product of the expected size (Fig. 2 F). That this was indeed a specific Ly49B product was confirmed by sequencing. Throughout our studies, the 5′ primer set was consistently a more sensitive indicator of the presence of Ly49B transcripts presumably due to the greater innate efficiency of this primer pair and/or the greater abundance of Ly49B transcripts bearing 5′ exons caused by alternate splicing or degradation.

Next, we took advantage of the fact that day-14 fetal thymocytes are a rich source of NK progenitors (20). When cultured with IL-4 plus PMA, these progenitors proliferate rapidly with little differentiation, but when subsequently transferred to medium containing IL-15, most cells acquire NK1.1 within 1 day, and after 10 days all cells express NK1.1 and many express CD94/NKG2 receptors and Ly49E (20, 21, 22). At no stage of the differentiation of these essentially pure progenitor cells could we detect Ly49B transcripts using Ly49B primer set 1 (Fig. 2,F). However, as in the case of adult NK cells, products of the expected size were reproducibly detected throughout the differentiation of these immature NK cells using primer set 2 (Fig. 2,F). Because of the high level of proliferation that occurs in these cultures and because no feeder cells were present, it is extremely unlikely that these expanding NK cells contained significant numbers of non-NK cells. The possibility that the Ly49B transcripts originated from minute numbers of contaminating non-NK cells was formally ruled out by the finding that RNA prepared from cloned immature NK cell lines such as I2/22 and 1608b, and from the long-term BALB/c fetal liver NK cell line LNK, gave not only strong signals with primer set 2 but also, for two of these three cells, clear signals with primer set 1. Surprisingly, these fetal NK cell lines also gave strong signals with primers that amplified the entire coding sequence of Ly49Q, and a weak but clear Ly49Q signal was reproducibly obtained with RNA from purified adult splenic NK cells (Fig. 2,F). Sequencing of these various products again confirmed that they were derived from Ly49B and Ly49Q transcripts. The PCR data are consistent with the surface expression of Ly49B on I2/22 (Fig. 2 E) but not 1608b (data not shown). Curiously, none of the three NK lines expressed detectable levels of Ly49Q at the cell surface (data not shown).

Finally, pure CD4 and CD8 T cells were prepared by first culturing spleen cells, depleted respectively of CD8⁺ and CD4⁺ cells, for 3 days with Con A plus IL-2. Nonadherent cells were then sorted for CD4⁺CD11b⁻Gr1⁻ or CD8⁺F4/80⁻Gr1⁻ cells, respectively, and cultured for an additional 3 days in IL-2. The resulting nonadherent cells comprised >99% CD4⁺ or CD8⁺ cells with no detectable CD11b⁺, F4/80⁺, or Gr1⁺ cells. RNA prepared from the pure CD4 T cells gave a weak signal with primer set 2 but no detectable signal with primer set 1, whereas RNA prepared from CD8 T cells gave a clear signal with Ly49B primer set 2 and occasionally a detectable signal with Ly49B primer set 1 (Fig. 2,F), although this latter was not always observed (Fig. 3). The nature of these PCR products was again directly confirmed by sequencing. RT-PCR analysis also indicated that CD4 and CD8 cells may express low levels of Ly49Q transcripts, and the thymic T cell lymphoma line SL8 appeared to express high levels of Ly49Q transcripts, although this last finding was not verified by sequencing. It should be noted that the input levels of RNA in terms of cell equivalents were four times higher for CD4 and CD8 T cells than for the other cells analyzed in Fig. 2 F and that at lower input levels the PCR signals were more difficult to detect.

To gain a clearer impression of the relative expression of Ly49B transcripts in various cells, semiquantitative PCR analysis was performed with cDNA prepared from highly purified cell populations obtained by cell sorting as described above. As shown in Fig. 3, when NK cells were cultured in IL-12 plus IL-18, there was a substantive increase in Ly49B transcripts, which is consistent with their acquisition of detectable levels of surface Ly49B (Fig. 1). The level of transcripts expressed by NK cells following exposure to IL-12 plus IL-18 was comparable to that found in cultured splenic macrophages (see below), in both cases transcripts being detectable with the Ly49B full-length primers down to 150 cell equivalents per reaction. By contrast, NK cells cultured in the absence of IL-12 plus IL-18 gave only weak signals with these primers at 5,000 cell equivalents per reaction, and CD8 T cells in these experiments gave no detectable signal even with 25,000 cell equivalents per reaction. As noted above, NK cells cultured without IL-12 plus IL-18 expressed detectable full-length Ly49Q transcripts, although the level of these was markedly lower than in splenic macrophages. Remarkably, whereas the level of Ly49B transcripts increased when NK cells were exposed to IL-12 plus IL-18, the level of Ly49Q transcripts declined. The level of Ly49G transcripts was unaffected.

The nature of the full-length Ly49B transcripts obtained from NK cells cultured with IL-12/IL-18 was confirmed by sequencing. Remarkably, this revealed the existence of transcripts in NK cells that contained a previously unknown additional exon of 99 nt located 6.3 kb downstream of exon 6 and 1.6 kb upstream of exon 7 (GenBank accession no. EF025059). Transcripts containing this sequence if translated would generate an Ly49B isoform having an extra 33 aa in the C-type lectin domain. No exon homologous to this has been reported in any other Ly49 cDNA sequence. It will be important to determine whether such transcripts do indeed give rise to expressed proteins and whether these are selectively expressed in NK cells.

Although freshly prepared B, T, and NK cells lacked detectable surface expression of Ly49B, ∼1.5% of CD45⁺ C57BL/6 spleen cells (above a background of 0.7% autofluorescent cells) showed low but clear staining with the 1A1 mAb (Fig. 4,A). Most of these Ly49B⁺ spleen cells had relatively high levels of side light scatter (SSC) and comprised three distinct subpopulations, all of which expressed high levels of CD11b together with low, intermediate, or high levels of Gr1, representing ∼0.3, 0.6, and 0.8% of all spleen cells, respectively (a fourth population of cells that appear to be CD11b^lowGr1^low constitute autofluorescent cells that are inadvertently captured in the Ly49B gate). Conversely, most if not all cells in each of these three CD11b/Gr1-defined spleen cell populations expressed Ly49B (Fig. 4 B). Using the additional myeloid-specific cell markers F4/80, Dectin-1, and 7/4 (15, 23, 24, 25, 26, 27, 28, 29), it was found that the Gr1^high cells expressed little or no F4/80 but low levels of Dectin-1 and high levels of the 7/4 Ag, whereas the Gr1^int and Gr1^low cells mostly expressed high levels of F4/80 and Dectin-1 and low levels of 7/4.

A much higher proportion of C57BL/6 BM cells, typically ∼10% of CD45⁺ cells, expressed low levels of Ly49B (Fig. 4,A). Ly49B-expressing BM cells were concentrated amongst those with intermediate (SSC^int) or high (SSC^high) levels of light scatter and were present in multiple subpopulations of CD11b/Gr1-defined cells. Among SSC^int cells, Ly49B was most prominently expressed on a small subpopulation of CD11b⁺Gr1^low cells that represented 0.7% of BM cells (gate R2, upper middle panel of Fig. 4,B) and which expressed F4/80, Dectin-1, and generally low levels of 7/4 (data not shown). Ly49B was also expressed at low levels on a second subpopulation of SSC^intCD11b⁺ cells that expressed intermediate levels of Gr1 together with mostly high levels of F4/80, Dectin-1, and 7/4 (data not shown) and represented ∼2% of BM cells (gate R3, upper middle panel of Fig. 4,B). However, it was notably absent from a third SSC^intCD11b⁺ population that was Gr1^high (gate R4, upper middle panel of Fig. 4,B). These cells were F4/80^low Dectin-1⁻ 7/4⁺ and uniformly expressed CD117 (data not shown), suggesting that they were immature myeloid cells. By contrast, among SSC^high BM cells Ly49B expression was confined to the major CD11b⁺Gr1^high subpopulation that represents 15–20% of all BM cells (gate R3, upper right panel of Fig. 4,B). These cells expressed low levels of Dectin-1 and F4/80 and intermediate levels of 7/4 (data not shown); however, only ∼50% of the cells in this population expressed Ly49B. Ly49B was more consistently expressed on the subpopulation of these cells that were CD11b^high (gate R4, upper right panel of Fig. 4 B). Collectively, these results demonstrate that in spleen and BM Ly49B is not expressed at detectable levels on any lymphocyte population but instead is found on multiple subpopulations of CD11b⁺ cells that coexpress a range of markers characteristic of monocytes, macrophages, and granulocytes.

Ly49B-expressing cells were readily detected in the peritoneum. In untreated C57BL/6 mice, ∼30% of peritoneal cells were Ly49B⁺, most of which were CD11b^high, Gr1^−/low, and F4/80^high cells (Fig. 5,A) that lacked 7/4, expressed Dectin-1, and displayed high forward light scatter and SSC (data not shown), a phenotype characteristic of macrophages. Within 3 h of injecting thioglycollate, a new population of Ly49B⁺ cells appeared that expressed low levels of Ly49B and was Gr1^highF4/80^−/low (Fig. 5,B) with high levels of 7/4 and low or no Dectin-1, a phenotype characteristic of neutrophils. Over the next 24 h, the size of this Ly49B^low population increased while the F4/80^highLy49B⁺ population virtually disappeared to be replaced by a third population of Ly49B⁺ cells that expressed low levels of Ly49B and Gr1, intermediate levels of F4/80 (Fig. 5,C), and both 7/4 and Dectin-1 (data not shown). These cells have been identified by others as infiltrating monocytes (28, 30). A fourth population of peritoneal cells that expressed Ly49B was eosinophils identified by their characteristic FSC^lowSSC^highGr1^−/low phenotype and by high expression of CCR3 (31). In normal animals, most such cells were Ly49B⁺ (Fig. 5,D), but in thioglycollate-treated animals, only a minority were (Fig. 5,E). CD117^high peritoneal mast cells were Ly49B⁻ (Fig. 5 D), as were CD117⁺ mast cells generated by culturing BM cells with IL-3 (data not shown). The eosinophilic and mast cell nature of these populations was confirmed by cell sorting and cystospin preparations (see below).

In all of the above experiments, staining was done in the presence of an excess of blocking Ab against the CD16 and CD32 FcR, and in addition, isotype control Abs gave no detectable staining, making it extremely unlikely that the staining observed by the 1A1 anti-Ly49B mAb was nonspecific. However, to formally rule out this possibility, we took advantage of the fact that the 1A1 mAb lacks detectable reactivity against the BALB/c isoform of Ly49B (Fig. 1). As shown in Fig. 6,A, whereas 1A1 stained a substantial proportion of cells in the R3 and R4 regions of SSC^highCD11b⁺Gr1⁺ BM cells from C57 mice, the corresponding populations from BALB/c mice were unstained, the histograms for 1A1 staining being completely superimposable on those for the isotype control mAb (Fig. 6,A) and for the medium control (data not shown). Similarly, whereas 1A1 stained essentially all high scatter peritoneal cells of C57 mice, it showed no detectable staining of the corresponding cells of BALB/c mice (Fig. 6 B).

The finding that Ly49B was expressed on myeloid cells suggested that its pattern of expression might be similar to that of Ly49Q, which was recently reported to be expressed at high levels on DC and at lower levels on some other myeloid populations (13, 16, 32). However, as shown in Fig. 7,A, costaining of spleen cells with mAbs to Ly49B and Ly49Q revealed that most of the cells that stained strongly for Ly49Q lacked detectable Ly49B, whereas most of the cells that expressed low levels of Ly49Q appeared to be Ly49B⁺. Conversely, the cells that expressed Ly49B showed at most only weak staining for Ly49Q. Similarly, in the BM, those cells that stained clearly for Ly49Q lacked detectable Ly49B whereas those that stained clearly for Ly49B expressed little or no detectable Ly49Q (Fig. 7 B).

The largely distinct nature of most Ly49B⁺ and Ly49Q^high cells was confirmed in several ways. First, in both spleen and BM, Ly49B⁺Q⁻ cells and Ly49B⁻Q^high cells had different light scatter characteristics (Fig. 7, A and B). Second, whereas most Ly49Q^high cells in the BM expressed CD11c and B220, most Ly49B⁺ cells lacked these markers, although ∼30% of the SSC^intLy49B⁺ cells were CD11c⁺ (Fig. 7,C). Third, most BM Ly49Q^high cells expressed CD4 and many also expressed CD8, but very few Ly49B⁺ cells expressed CD4 or CD8 (Fig. 7,C). Finally, direct gating on CD11c⁺ cells in BM showed that among those that were Gr1⁻, very few expressed Ly49B but most expressed Ly49Q; conversely, most of the CD11c⁺ cells that expressed low levels of Gr1 expressed Ly49B but little or no Ly49Q (Fig. 7 D). In the spleen, Ly49B was rarely expressed on CD11c^high cells, whereas CD11c^low cells frequently expressed Ly49B (data not shown).

To examine in more detail, the expression of Ly49B on myeloid cells, highly enriched populations of such cells were obtained by culture with appropriate cytokines. As shown in Fig. 8 A, splenic macrophages derived by culture with M-CSF constitutively expressed substantial levels of both Ly49B and Ly49Q. By contrast, although BM macrophages derived by culture in M-CSF expressed Ly49Q, they lacked detectable expression of Ly49B. Exposure of these cells to LPS, γ-IFN, or α-IFN, however, induced expression of Ly49B but left Ly49Q expression largely unchanged. DC derived from BM cells by culture in GM-CSF or FL displayed a broad and largely monotonic expression of Ly49Q but expressed Ly49B on a distinct subpopulation of cells. The expression of Ly49B and Ly49Q on these cells was differentially affected by LPS and IFN. Thus, both γ-IFN and α-IFN markedly increased the expression of Ly49B, especially on the Ly49B-expressing subpopulation of FL-derived cells where it reached levels higher than on any other cell type studied. However, whereas α-IFN increased the expression of Ly49Q on GM and FL DC, γ-IFN generally did not. Similarly, although LPS increased the expression of Ly49B on FL-derived DC, it reduced the expression of Ly49B on GM-derived DC and generally reduced the expression of Ly49Q on both cells.

Analysis at the RNA level confirmed and extended these findings (Fig. 8,B). Thus, cultured splenic macrophages clearly contained transcripts for Ly49B and Ly49Q. Unstimulated BM macrophages, despite their lack of detectable surface expression of Ly49B, also contained full-length Ly49B transcripts, most easily detected using the 5′ primer set. The level of full-length transcripts increased markedly following stimulation with LPS or γ-IFN. The macrophage lines J774, RAW264, and P388D1, all of BALB/c origin, were also examined. None of these cells showed detectable surface staining with the 1A1 anti-Ly49B mAb (data not shown) but, as exemplified by the results for J774 and RAW264 shown in Fig. 8 B, contained full-length Ly49B transcripts (again most readily detected with the 5′ primers) whose levels sometimes increased and sometimes decreased in response to LPS or γ-IFN. Ly49Q transcripts were also present in these cells.

To better understand the diverse patterns of expression of Ly49B and Q on cultured BM-derived cells, costaining was performed with a range of macrophage and DC markers. Cultured BM macrophages uniformly expressed CD11b, Gr1, F4/80, and Dectin-1, with little or no 7/4, and no detectable B220, MHC class II (cII), or CD11c (data not shown). GM-CSF DC also lacked expression of B220 but showed heterogeneous expression of cII, CD11c, and F4/80 (Fig. 8,C). Despite the complexity of GM-CSF-derived cells, Ly49B and Ly49Q were expressed in an essentially identical manner, both molecules being largely absent from cII⁺ cells but present on the majority of CD11c⁺ and F4/80⁺ cells. Indeed, the expression of Ly49B and Ly49Q on GM-CSF-derived DC was essentially confined to cells that were B220⁻, cII⁻, CD11c⁺, and F4/80⁺. By contrast, FL-derived DC differed in several ways: they were smaller, a higher proportion expressed CD11c and cII (Fig. 8,D), a lower proportion expressed Gr1 (data not shown), and many expressed B220 (Fig. 8 D). Remarkably, whereas most Ly49B⁺ FL DC resided within a B220⁻cII⁺CD11c⁻F4/80⁺ population, most Ly49Q⁺ cells resided within a B220⁻cII⁺CD11c⁺F4/80⁻ population, and a substantial minority resided in a second subpopulation that expressed high levels of B220.

Taken together, the data presented in this section demonstrate that Ly49B and Q are expressed on multiple subpopulations of in vitro-derived myeloid cells, that these two molecules can be expressed independently of each other, and that they are subject to differential regulation by bacterially derived and endogenous immunoregulators.

Fig. 9,A shows the typical morphology of cells present in various cell populations that expressed or lacked Ly49B. More than 80% of splenic CD11b⁺Gr1^high cells that uniformly expressed Ly49B had strikingly segmented nuclei with pale featureless cytoplasm, many with the classical morphology of neutrophils (Fig. 9,A-a). By contrast, ∼70% of splenic CD11b⁺Gr1^low/intLy49B⁺ cells were macrophage-like cells with generally oval or indented nuclei and substantial vacuolar cytoplasm (Fig. 9,A-b). Approximately 15% were a smaller version of this with less cytoplasm (most likely monocytes), whereas an additional ∼15% had the highly segmented nuclei and pink cytoplasmic granules characteristic of eosinophils (lower right cell in Fig. 9,A-b). Cells in the SSC^highCD11b⁺Gr1^high population of BM cells, ∼50% of which expressed Ly49B, had a variable neutrophilic appearance (Fig. 9,A-c), most having less heavily segmented nuclei than the subpopulation of these cells that expressed the highest level of CD11b (gate R4 upper right panel in Fig. 3), which had a similar mature neutrophilic appearance to the corresponding population in spleen (Fig. 9,A-a). By contrast, cells in the SSC^intCD11b⁺Gr1^high population that expressed CD117 but lacked Ly49B were noticeably smaller and frequently had nonsegment nuclei with a central hole (Fig. 9,A-d). These findings indicate that as granulocytes mature, increasing in size and granularity, losing CD117, and gaining increasingly fragmented nuclei they also gain expression of Ly49B. By contrast, the SSC^intCD11b⁺Gr1^low/int BM cells that expressed Ly49B had a monocyte/macrophage morphology (Fig. 9,A-e). The CD11b⁺Gr1⁻F4/80^highLy49B⁺ cells in normal peritoneal washings were large heavily vacuolated macrophages (Fig. 9,A-f). The CD11b⁻Ly49B⁻ population of normal peritoneal cells also included macrophage-like cells, but these were generally smaller, less vacuolated, and with less contorted nuclei (Fig. 9,A-g). Approximately 90% of the Gr1^highLy49B^low cells from the peritoneum of thioglycollate-treated mice were typical neutrophils (Fig. 9,A-h), whereas >90% of the Gr1^intLy49B⁺ cells had a monocyte/macrophage appearance (Fig. 9,A-i). By contrast, the SSC^highFSC^lowGr1^−/low peritoneal cell population comprised predominantly eosinophils (Fig. 9 A-j).

Histological analysis revealed that in spleen and lymph node, Ly49B-expressing cells were most abundant in the red pulp and medulla, respectively, and were only occasionally found in the white pulp and paracortex (Fig. 9, B and C). At higher magnifications Ly49B⁺ cells were seen to vary in size, intensity of labeling, and nuclear morphology, consistent with Ly49B being expressed on multiple cell populations. The majority of Ly49B⁺ cells was macrophage-like, although cells with a polymorphonuclear appearance were also labeled (inset arrow). Ly49B⁺ cells were found scattered in thymus cortex and medulla but in dual-labeling experiments (data not shown) did not colocalize with CD11c, suggesting they were not DCs. Ly49B⁺ cells were prominent in the lamina propria of gut villi and also within the dome region of the Peyer’s patch where they were identified as the autofluorescing macrophages in this region and not the CD11c⁺ DCs (data not shown). Ly49B was also clearly expressed on at least some alveolar macrophages in the lung and Kupffer cells in the liver.

Remarkably, the distribution of Ly49B labeling in lymphoid tissue was different from that seen using a series of other myeloid cell markers (Fig. 9,C). In the splenic red pulp, the Ly49B⁺ cells appeared to represent a subpopulation of the more numerous F4/80⁺ macrophages, whereas within the marginal zone and T and B cell areas, a few Ly49B⁺ cells but no F4/80⁺ cells could be seen. The distribution of Ly49B was more similar to that of Dectin-1 (27), which is also expressed on a subpopulation of red pulp macrophages, but Dectin-1⁺ cells were more numerous within the marginal zone and T cell areas due to the expression of Dectin-1 on a proportion of DC and macrophages at these sites. Ly49B labeling was similar to that seen with macrosialin (CD68), a molecule generally considered to be a pan-macrophage-specific marker (29, 33), but again, fewer cells in the white pulp expressed Ly49B than macrosialin. The pattern of Ly49B labeling was very different to that of sialoadhesin, a member of the siglec family expressed on specialized macrophages within the spleen marginal zone and lymph node subcapsular sinus (34) (Fig. 9 C). It was also different to that seen with anti-CD11c, cells staining strongly for CD11c being concentrated in the marginal zone and around the central arterioles where Ly49B⁺ cells were rare.

In lymph nodes (Fig. 9,C), the anti-Ly49B mAb labeled intensely large cells within the medullary region, which appeared to represent a subpopulation of both F4/80⁺ cells and macrosialin⁺ cells. Unlike F4/80, however, some Ly49B⁺ cells encroached upon the paracortex and some were present in the B cell follicles, perhaps representing the tangible body macrophages that are also positive for macrosialin and Dectin-1. In lymph node, the patterns of Ly49B and Dectin-1 expression were more distinct than in the spleen. First, the expression of Ly49B on the medullary macrophages appeared greater than Dectin-1, and second, Dectin-1 was clearly expressed within the paracortex among the T cells and also within the B cell follicle where Ly49B expression was rare. As in the spleen, Ly49B expression in lymph node appeared very different to that of both sialoadhesin, which is largely restricted to the subcapsular sinus macrophages and a few cells in the follicles, and CD11c, which is prominently expressed throughout the paracortex with only occasional weak staining elsewhere. CR-Fc, a chimeric protein containing the cysteine-rich domain of the mannose receptor that binds to sialoadhesin and other molecules on marginal zone metallophilic macrophages and follicular macrophages (17, 35), was used in two-color staining with mAb 1A1. These experiments demonstrated that in the marginal zone of the spleen and the subcapsular sinus of the lymph node Ly49B and CR-Fc staining identified largely distinct subpopulations of cells and that cells expressing high levels of Ly49B and CD11c were largely distinct from each other in spleen, lymph node, and gut (Fig. 9 D).

To examine the association of Ly49B with intracellular signaling molecules, RNK cells were transfected with a Ly49B construct bearing a C-terminal FLAG tag, and lysates from these cells were precipitated with anti-FLAG mAb. Western blots with Abs to SHP-1, SHP-2, and SHIP-1 revealed that each of these phosphatases was associated with Ly49B (Fig. 10 A). In the case of SHP-1 and SHP-2, this association was only seen when cells were incubated with pervanadate before immunoprecipitation, implying that the association of SHP-1 and SHP-2 with Ly49B was dependent on tyrosine phosphorylation of substrate molecules. By contrast, SHIP-1 was strongly associated with Ly49B in untreated cells, and this association was lost following incubation with pervanadate.

To examine the possible functional consequences of these associations, cloned lines of RNK cells transfected with native Ly49B or Ly49B bearing a C-terminal HA tag were examined for their ability to kill FcR-bearing cells in the presence or absence of appropriate Abs. An example of the data obtained is shown in Fig. 10,B. RNK cells uniformly expressing Ly49B at levels similar to those shown in Fig. 1 efficiently killed P388D1 targets, but inclusion in the assays of the 1A1 mAb at concentrations that gave saturation binding to Ly49B caused no significant enhancement or inhibition of killing. Similar results were obtained with the FcR-bearing targets J774 and A20, with two other anti-Ly49B mAbs of a different IgG subclass (IgG2c), and when RNK cells transfected with Ly49BHA were assayed on the same targets in the presence of rabbit anti-HA Ab.

The data presented in this report have revealed for the first time that, unlike most members of the Ly49 family, Ly49B is not normally expressed at detectable levels on NK cells but instead is expressed on multiple subpopulations of myeloid cells present in both lymphoid and nonlymphoid tissues. Among the myeloid cells present in cell suspensions obtained from the lymphoid tissues of normal animals, Ly49B expression was generally quite low, requiring the use of strong fluorochromes for clear visualization. However, following culture with inflammatory cytokines or bacterial products, the expression of Ly49B frequently increased, especially when cells expanded from BM with FL were exposed to α-IFN or γ-IFN. Histological studies showed that although staining for Ly49B was generally quite modest, occasional cells in the splenic red pulp and the medullary regions of lymph nodes were stained strongly. Cells expressing high levels of Ly49B were also seen sporadically in the thymus and the lung but were much more numerous in the dome region of Peyer’s patches and the lamina propria of the gut, which is consistent with the notion that Ly49B levels are increased by interaction with bacterial products and suggesting that Ly49B expression on gut macrophages may be of considerable importance for the normal homeostasis and immunobiology of this organ.

The histological studies also revealed that the pattern of Ly49B expression on myeloid cells was distinct from that of several other myeloid-specific molecules, confirming and extending the growing view that myeloid cells, and macrophages in particular, exist in numerous subpopulations that presumably have distinct functions in the immune system. Heterogeneity was also apparent in flow cytometric analysis of cell suspensions, Ly49B being expressed on multiple distinct CD11b- and Gr1-defined subpopulations of cells in the spleen, BM, and peritoneum. These subpopulations were dominated by cells with the morphological characteristics of granulocytes and monocytes/macrophages. The expression of Ly49B in myeloid cells was confirmed by flow cytometric and PCR analysis of macrophage cell lines and of highly enriched myeloid populations derived by expansion of BM progenitors with myelopoietic cytokines. PCR analysis also revealed the presence of Ly49B transcripts in NK cells and T cells, particularly CD8⁺ T cells, suggesting that these cells may under certain circumstances express Ly49B at the cell surface. In the case of NK cells, surface expression, albeit at a low level, was found on immature NK cell lines grown in IL-2 and on mature NK cells exposed to the inflammatory cytokine combination of IL-12 and IL-18. These findings are in line with previous reports that populations containing immature NK cells expanded from BM precursors (36) or derived from embryonic stem cells (37) expressed Ly49B transcripts, although the possibility that these originated from contaminating myeloid cells would now need to be considered.

Of particular interest and importance in this study was the relationship between the expression of Ly49B and Ly49Q, which was recently reported to be expressed on myeloid cells. Our studies showed that among freshly prepared spleen and BM cells Ly49B and Ly49Q were most prominently expressed on distinct cell types, Ly49B-expressing cells mostly lacking high levels of CD11c, B220, CD4, and CD8, whereas Ly49Q⁺ cells mostly expressed these markers, which is consistent with previous work demonstrating that Ly49Q was most strongly expressed on cells characterized as plasmacytoid DC (16, 32). Some Ly49B⁺ cells present in fresh cell preparations appeared to express low levels of Ly49Q, and Ly49Q was clearly coexpressed with Ly49B on various subpopulations of cultured myeloid cells. Importantly, however, Ly49B and Ly49Q expression on these cells was found to be regulated independently. Histological data showed that Ly49B was not prominently expressed, if at all, on the classical DC pop-ulations in the T cell rich areas of splenic white pulp and lymph node paracortex. Flow cytometric analysis, however, showed that Ly49B was expressed on the CD11c^lowcells present in suspensions of fresh spleen and BM cells, and also on subpopulations of CD11c⁺ cells expanded from BM progenitors using GM-CSF or FL, some of which also expressed B220. Such CD11c⁺ cells are generally termed DC, although their relationship to DC in vivo is unclear given that they lack expression of the CD4 and CD8 markers frequently found on fresh DC.

The demonstration that Ly49B is expressed primarily on myeloid cells raises important questions concerning its function. Whereas most mouse Ly49 receptors show 60–90% amino acid identity with each other, Ly49B displays only ∼50% homology with other family members, including Ly49Q, and contains 20 additional aa at its C terminus. It also displays only ∼45% homology with most members of the rat Ly49 family but shows 63% homology with rat Ly49-i8 (which also has a C-terminal extension), suggesting that these two molecules are functional othologs and share a common and possibly unique function in murine species. Despite this, significant sequence variability exists within the mouse, the C57 (GenBank accession no. AK017140) and 129 (GenBank accession no. AF395446) strain sequences (which are identical) differing by 12 aa from the BALB/c (GenBank accession no. AF253057) and CBA (GenBank accession no. MMU10304) sequences (also identical). All but one of these differences reside in the carbohydrate recognition domain (CRD) and collectively would explain the recognition of the C57/129 isoform but not the BALB/c isoform by the 1A1 mAb. Remarkably, one of these differences involves the cysteine at position 251 that is conserved in virtually all animal C-type lectins (38, 39). In the C57/129 isoform, this is replaced by arginine, thereby precluding the formation of a conserved disulfide bond, found in all CRD crystal structures that have been solved, with the preceding conserved cysteine (C238 in Ly49B). The only other animal C-type lectins known to lack this conserved cysteine are the mouse Clr molecules (40) and their human ortholog LLT1 (41) that are encoded in the NKRP1 region of the NK complex and act as ligands for NKRP1 molecules (42, 43, 44, 45). In addition, several of the conserved core hydrophobic residues thought to be important for determining the basic structure of the CRD are altered in Ly49B. Among these are Cys¹⁵², Leu¹⁶⁶, Ser²¹⁴, and Gln²¹⁶ that are Tyr, Trp, Trp, and Trp, respectively, in all other mouse and rat Ly49s and in most other C-type lectins. These observations collectively suggest that the structure of Ly49B will be significantly different from that of other Ly49s and that, by implication, it will interact with quite different ligands.

The similarity between Ly49B and other mouse Ly49s is greatest in the cytoplasmic and transmembrane domains where it is typically ∼70%. Ly49B lacks a charged residue in the transmembrane region and contains the ITIM-like sequence VTYTTL in the cytoplasmic tail. Although this sequence is different from the ITIM sequence found in other mouse Ly49s (most commonly VTYSTV) or indeed in any other mammalian immune receptor, it conforms to the consensus ITIM sequence I/L/S/VxYxxL/V (46) and is preceded by the same five amino acids and followed by the same five amino acids as the ITIM in Ly49A and many other mouse and rat Ly49s. Therefore, it would be expected that following cross-linking by as yet unknown ligand(s), the tyrosine residue in the Ly49B ITIM becomes phosphorylated and delivers inhibitory signals via recruitment and activation of the protein tyrosine phosphatase SHP-1, as appears to be the case for several other Ly49 receptors (47, 48). Consistent with this, we found that in transfected cells Ly49B does indeed associate with SHP-1 in a phosphorylation-dependent manner. Ly49B also interacted with SHP-2, a protein tyrosine phosphatase that can associate with other Ly49 receptors (13, 49, 50) and forms an inhibitory complex with killer Ig-related receptors (51). It was also found to associate with SHIP-1, a phosphatase that is thought to play a major role in negative intracellular signaling via its degradation of phosphatidylinositol 3,4,5-trisphosphate. Remarkably, and in contrast to the situation of SHP-1 and SHP-2, the association of Ly49B with SHIP-1 appeared to be constitutive and was inhibited rather than enhanced by pervanadate-induced phosphorylation. Constitutive association of SHIP-1 with other (but not all) ITIM-containing Ly49 receptors, as well as perturbed expression of Ly49 and other NK cell receptors in SHIP-1-deficient mice, has been reported recently (52, 53). Furthermore, SHIP-1-deficient mice have been known for some time to have severely disrupted myeloid cells homeostasis, with elevated numbers of immature and mature myeloid cells, the latter being hyperresponsive to a variety of ligand-induced signals (54, 55, 56, 57).

Therefore, it is tempting to speculate that, despite the variations in the extracellular sequence of Ly49B suggestive of pathogen-driven evolution, Ly49B recognizes an endogenous ligand and via its constitutive association with SHIP-1 plays a critical role in controlling the expansion and/or survival of myeloid cells and preventing their spontaneous activation. Following encounter with inflammatory cytokines or bacterial products Ly49B is up-regulated or induced on myeloid cells and NK cells and develops new regulatory interactions with SHP-1 and SHP-2. In this context, Ly49B would join an increasingly large number of ITIM-containing receptors whose primary function appears to be to dampen the reactivity of myeloid cells. Examples of such receptors include the LIR and PIRB receptors that recognize MHC class I molecules (58), the CD33-related Siglecs that recognize sialic acid in various contexts (59), and SIRPα, which through its recognition of CD47 expressed on healthy RBC and lymphocytes prevents their phagocytosis (60), and via its interaction with surfactants may prevent nonspecific inflammatory responses in the lung (61). All of these molecules belong to the Ig superfamily. However, several ITIM-bearing members of the C-type lectin family, all encoded in the NK gene complex, such as DCIR (CLECSF6) (62), MICL (63), and Ly49Q, are widely expressed in myeloid cells. It will clearly be of considerable importance for understanding the role of myeloid cells in various disease states to formally establish whether and under what circumstances Ly49B (and these other ITIM-bearing C-type lectin molecules) inhibits myeloid cell function. Initial attempts to reveal inhibitory activity using reverse Ab-dependent cellular cytotoxicity assays were unsuccessful, but negative results in such assays have been reported previously with receptors of known inhibitory capability (see Discussion in Ref. 64), and alternative approaches will need to be explored. Additional priorities for future work include the determination of the three-dimensional structure of Ly49B and identification of its ligands.

The Ly49B transcript sequence containing a novel additional exon has been deposited in GenBank (accession no. EF025059).

We are grateful to Ian Harvey and Brian Shenton for their advice, expertise, and assistance with flow cytometry and cell sorting.

The authors have no financial conflict of interest.