LLT1 and CD161 Expression in Human Germinal Centers Promotes B Cell Activation and CXCR4 Downregulation

Germinal centers (GCs) are microanatomical structures critical for the development of high-affinity Abs and B cell memory. They are organized into two zones, light and dark, with coordinated roles, controlled by local signaling. The innate lectin-like transcript 1 (LLT1) is known to be expressed on B cells, but its functional role in the GC reaction has not been explored. In this study, we report high expression of LLT1 on GC-associated B cells, early plasmablasts, and GC-derived lymphomas. LLT1 expression was readily induced via BCR, CD40, and CpG stimulation on B cells. Unexpectedly, we found high expression of the LLT1 ligand, CD161, on follicular dendritic cells. Triggering of LLT1 supported B cell activation, CD83 upregulation, and CXCR4 downregulation. Overall, these data suggest that LLT1–CD161 interactions play a novel and important role in B cell maturation within the GC in humans.

T he germinal center (GC) reaction is critical for long-lasting protection against pathogens. GCs are the anatomical sites within secondary lymphoid organs where B cells proliferate and mutate their BCRs to be selected according to their affinity for Ag. Two distinct areas with different functions can be identified within the GC; these are the dark zone (DZ) and the light zone (LZ). In the former, B cells proliferate and hypermutate their BCRs to generate Ab variation, whereas the quality of these BCRs is assessed in the latter, ultimately leading to selection of highaffinity B cell clones (1,2). DZ B cells are characterized by being CD83 low CXCR4 high , whereas LZ B cells are CD83 high CXCR4 low (3). B cells that have successfully competed for Ag develop into clones and exit the GC expressing high-affinity Abs and long-lived memory. Thus, this process is crucial to vaccinology. At the same time, however, as a site of mutation and proliferation, aberrant reactions can lead to the development of B cell lymphomas and autoimmunity. Understanding the mechanisms that drive this process has significant implications in health care.
C-type lectin-like receptors (CLRs) are encoded in the NK gene complex (NKC) and can be expressed in a wide range of human cell types, including NK cells. They are particularly relevant in the context of innate immune responses. The CLRs lectin-like transcript 1 (LLT1) and CD161 are genetically linked physiological binding partners, located adjacent to one another within the NKC (4-7). Structurally, LLT1 shares the greatest homology with the other C-type lectins activation-induced C-type lectin and CD69 (8). Within murine models, LLT1 shows a similar expression pattern to MHC class I (9,10), whereas in humans it is limited to activated lymphocytes and monocytes (8,(11)(12)(13) and recently on respiratory syncytial virus-infected primary human bronchial epithelial cells (14), although the published literature presents some inconsistencies. In contrast, the expression of LLT1's binding partner, CD161, has been relatively well characterized, delineating a family of innate-like T lymphocytes and NK cells (15). Functional studies have described inhibitory and activating roles for both molecules (6,7,(15)(16)(17)(18)(19)(20)(21)(22)(23). These studies suggest that interactions between LLT1 and CD161 can result in bidirectional signaling and have functional consequences for both cells involved.
In this study, we show the high expression of LLT1 on human GC B cells and GC-derived B cell lymphomas, extending previous studies (6,8,(11)(12)(13)17). We also show that LLT1 expression remains on early plasmablasts, but is absent from memory B cells and plasma cells. The LLT1 ligand, CD161, was found, unexpectedly, on follicular dendritic cells (FDCs). Finally, triggering of LLT1 promoted the upregulation of CD83 on B cell and drives DZ B cells toward a LZ phenotype through the downregulation of CXCR4. Previously, LLT1 and CD161 were considered part of innate immune responses. The present study demonstrates a functional role for an innate receptor pairing at the heart of a critical adaptive immune process, the GC reaction in humans.

Tissues, cells, and cell lines
Human tonsillar tissue was obtained following routine tonsillectomy from the files of the Department of Cellular Pathology (University College London Hospital, London, U.K.); Human Tissue Resource Centre, Barts and the London National Health Service Trust, Queen Mary School of Medicine and Dentistry; and from the Ear, Nose, and Throat Department, John Radcliffe Hospital, Oxford, U.K. Normal tonsillar tissue sections were obtained from ProteoGenix (Schiltigheim, France). Tonsil-derived single cells were collected by mechanical disruption of tonsil samples or collagenase D (1 mg/ml; Boehringer Mannheim) and DNase I (0.1 mg/ml; Sigma-Aldrich, Dorset, U.K.) digestion, as stated. The lymphoma samples analyzed were in the form of 0.6-to 1-mm core tissue arrays.
PBMCs obtained from the National Blood Transfusion Service (National Health Service Blood and Transplant) were isolated on a Lymphoprep gradient (Axis-Shield, Dundee, U.K.).
Bulk B cells were isolated by negative selections from PBMCs or tonsils by magnetic isolation (Stemcell Technologies, Cambridge, U.K.) following manufacturer's protocols. 300.19-hCLEC2D cells were created by transfection of 300.19 with a vector expressing human CD161/CLEC2D cDNA and maintained under selection.
Vaccine samples were obtained from the Oxford Vaccine Group, Churchill Hospital (Oxford, U.K.) following HBV vaccinations.
Bone marrow samples were obtained from routine hip joint operations (Newcastle University, Newcastle upon Tyne, U.K.). Samples were filtered (40 mm), washed with PBS, homogenized, isolated on a Lymphoprep gradient (Axis-Shield) and aliquoted in FCS plus 10% DMSO (Sigma-Aldrich) and stored in liquid nitrogen until required.
For sorted cells, DZ B cells were FACS sorted using a MoFlo XDP (Beckman Coulter, High Wycombe, U.K.) from purified B cells and defined as singlets, alive, CD38 mid IgD 2 CXCR4 high CD83 low cells.
Informed consent was obtained from all subjects.

In vitro differentiation of plasma cells
Memory B cells were differentiated in vitro into plasmablasts and plasma cells, following the protocol described in Jourdan et al. (24). Briefly, CD27 + Memory B cells were first purified as bulk B cells by negative selection from PBMCs by magnetic isolation followed by positive selection for CD27 + cells (Stemcell Technologies) following the manufacturer's protocol. The cells were then seeded at 1.  Novel mouse anti-human CLEC2D (LLT1) mAbs were generated, including clones 359.7G7 (mIgG1) and 359.2H7 (mIgG2al). All purified Abs are dialyzed against PBS, are low in endotoxin (,2 endotoxin units/mg), and are filtered sterile.
For LLT1 crosslinking, recombinant CD161 (R&D Systems) or IgG1 isotype control (R&D Systems) were bound to a 96-well ELISA plate (Greiner Bio-One, Stonehouse, U.K.) overnight prior to the addition of B cells and BCR stimulus as described above.

Proliferation assay
Proliferation assays were performed using a CTV cell proliferation kit (Life Technologies). Purified B cells were stained with 2.5 mM CTV for 10 min at room temperature, following the manufacturer's instructions. Cells were then counted and 2 3 10 5 B cells were plated per well in a U-bottom 96-well plate with CpG (0.5 mg/ml) for 5 d.

Data acquisition and analysis
FACS analysis was performed on Miltenyi Biotec MACSQuant cytometer and analyzed with FlowJo version 9.6.2 software (Tree Star, Ashland, OR). Graphs and statistical analysis were performed using GraphPad Prism version 6.0a (GraphPad Software, San Diego, CA).
For analysis of immunohistochemical staining, images were acquired on a DSS1 Coolscope slide scanner (Nikon, Kingston upon Thames, U.K.).
For immunofluorescent microscopy, images were acquired on an Olympus Fluoview FV1000 microscope (Olympus, Southend-on-Sea, U.K.) and analyzed using Fiji (ImageJ, National Institutes of Health, Bethesda, MD). The Leica TCS-SP2 AOBS confocal laser-scanning microscope was used for the CD161 immunofluorescence, and the images were captured and processed with Leica confocal software (LCS Lite) and ImageJ.

Results
LLT1 is highly expressed on GC B cells LLT1 expression was observed within normal human tonsils and predominantly expressed within GCs (Fig. 1A). This was demonstrated by immunohistological staining using two independent Abs: a commercial goat polyclonal anti-LLT1 Ab, and a novel anti-LLT1 Ab (clone 7G7, see Materials and Methods, Supplemental Fig 1A, 1B). Furthermore, this was not restricted to tonsillar tissue, as GCs within spleen, lymph nodes, and Peyer's patches were also LLT1 + (Supplemental Fig. 1C-F).
LLT1 expression was not homogeneous throughout the GC. A clear difference in LLT1 staining intensity could be seen by immunohistology. Cross-sectional analysis through the GC revealed increased LLT1 levels within the DZ compared with the LZ (defined by the presence of CD3 + cells, stained green, Fig. 1G). Moreover, FACS analysis confirmed higher LLT1 expression on DZ B cells (CD83 low CXCR4 high ) compared with LZ B cells (CD83 high CXCR4 low ) (Fig. 1H).
LLT1 is expressed on plasmablasts but not memory B cells B cells exit the GC upon differentiation into plasma or memory B cells. Whether LLT1 expression is maintained in these GC B cell derivatives is not known; therefore, we sought to track LLT1 expression through these stages. Plasmablasts were identified by costaining tonsillar sections with LLT1 plus CD19, CD20, CD38, CD45, CD79a, CD138, IRF4 (MUM-1), and Igs ( Fig. 2A). LLT1 + cells coexpressed these markers in a pattern suggestive of a plasmablast phenotype (CD45 low , CD20 2 , CD19 low , CD79a 2/low , pan Ig + , MUM-1 +/2 , CD38 + , Ki67 +/2 , CD138 2 ), although there remained a number of LLT1 + cells that did not fit this phenotype and remain to be identified.
To confirm that LLT1 is downregulated as plasmablasts mature into plasma cells, we differentiated CD27 + memory B cells into plasmablasts and then plasma cells in vitro, as described by Jourdan et al. (24). LLT1 expression was measured by FACS analysis at days 0 (memory B cells), 4 (early plasmablasts), 7 (late plasmabalsts), and 10 (plasma cells) (Fig. 2F). Day 0 resting CD27 + memory B cells expressed very low levels of LLT1, but LLT1 was highly upregulated at day 4 on early plasmablasts, but downregulated at further stages of differentiation (Fig. 2G). Additionally, we confirmed the absence of LLT1 on plasma cells in vivo by FACS and immunohistological staining of long-lived, bone marrow-resident CD138 + cells from bone marrow samples (data not shown).
Given access to bone marrow samples, we took the opportunity to assess expression levels of LLT1 on the early developmental stages of B cells; thus, we were able to assess LLT1 expression throughout the whole life of a B cell (Supplemental Fig. 2).

Regulation of LLT1 expression on memory B cells
Next we questioned how LLT1 was regulated within B cells. Previous work has shown LLT1 can be upregulated following TLR9, BCR, and CD40 stimulation (21). We confirmed CpG-induced expression (Fig. 3A) and extended this to show that this mechanism is MyD88-dependent, as the presence of the MyD88 inhibitor, Pepinh-MYD, blocked upregulation (Fig. 3B). In addition to CpG, we explored the kinetics of both BCR and CD40 stimulation to induced LLT1 expression. BCR signaling in isolated PBMC-derived memory B cells induced a maximum expression after 5 d (Fig. 3C), whereas CD40 stimulation induced LLT1 expression within 2 d (Fig. 3D). Furthermore, B cell activation is associated with a broad range of cytokines; therefore, we assessed the ability of IL-4, IL-10, IL-21, PGE 2 , IL-15, and BAFF to upregulate LLT1. None of the cytokines tested was able to induce LLT1 expression (Fig. 3E) (32-35).
Interestingly, LLT1 expression was associated with early rounds of proliferation after CpG stimulation, increasing with each division until the third, were it plateaued (Fig. 3F). This is consistent with the observation of high LLT1 expression within DZ B cells (Fig. 1G, 1H), plasmablasts (Fig. 2), and B cell lymphomas (see Fig. 6).

CD161 is expressed on FDCs and on a subset of tonsillar T cells
Before we addressed the function of LLT1, we looked for the expression of the LLT1 ligand, CD161, within the context of the GC. CD161 expression has been described on both NK cells and T cells. Therefore, T follicular helper (Tfh) cells were an obvious candidate cell type, with a potential to interact with GC B cells. Flow cytometric analysis of tonsillar T cells revealed a loss of CD161 expression as the Tfh cells progressed to full GC phenotype, as defined by CXCR5 expression (36) (Fig. 4A, top right panel). This was confirmed at the mRNA level in sorted tonsillar T cells (Fig. 4A, bottom right panel). Although a small fraction (,10%) of CXCR5 high Tfh cells expressed CD161, they were not observed within any of the GCs assessed by immunofluorescence microscopy (Fig. 4B, Supplemental Fig. 3). Similarly, within the PBMCs, circulating Tfh cells (defined simply as CXCR5 mid ) (37) also had lower frequencies of CD161 + cells compared with the CXCR5 2 fraction (Supplemental Fig. 3).
Unexpectedly, immunofluorescence staining of human tonsils revealed the expression of CD161 on FDCs (Fig. 4C). Costaining of CD161 and a FDC marker (CNA.42 clone) showed a strong overlap, with a typical stain pattern associated with FDCs. This striking finding was confirmed by histological staining and was not specific to just one clone of anti-CD161 Ab (Supplemental Fig. 3). Costaining of PNA, LLT1, and CD161 showed interaction between LLT1 on GC B cells and CD161, expressed in a FDC pattern, within the GC (Fig. 4D).

LLT1 promotes B cell activation and downregulation of CXCR4 in GC B cells
Interactions between FDCs and B cells are complex, involving a wide range of receptors (38). Binding of BCRs to Ags, presented by the FDC, and of CD40 to CD40L delivers crucial signals required for B cells to progress to the next stage of the GC reaction and interact with Tfh cells, and subsequently go on to re-enter the DZ (10, 39-41).
As B cell-FDC interactions involve many different ligands and receptors, we explored the role of LLT1 in isolated B cells. Initially, we investigated LLT1 signaling in primary B cells, inducing LLT1 expression through stimulation with anti-BCR, as described above, with and without rCD161 protein or a control protein (IC). After 5 d of stimulation, B cell expression of CD83 and CD38 was measured by flow cytometry. Triggering of LLT1 on B cells through rCD161 resulted in increased B cell activation. Significantly higher increases in CD83 and CD38 were seen on B cells in the presence of rCD161 compared with the controls (Fig. 5A).
As both proteins are highly glycosylated and have a weak binding strength (K d of 48 mM), we addressed whether binding of naturally expressed LLT1-CD161 would reveal stronger signaling (42). To analyze this, we cultured tonsillar cells, which included both B cells and CD161 + cells, for 3 d with either anti-CD40 plus IL-4 or CpG and assessed their activation state by flow cytometry on day 3. The effect of LLT1 signaling was measured through the use of blocking Abs to CD161 (Fig. 5B, left panel). Stimulation of B cells with anti-CD40 plus IL-4 resulted in a striking upregulation of CD83 on day 3, which was significantly inhibited by the addition of the anti-CD161 Ab. This was not observed with the CpG-stimulated sample. In contrast, levels of CD38 expression were highly increased upon CpG stimulation. Once again, the addition of the anti-CD161 blocking Ab significantly reduced CD38 upregulation (Fig. 5B, right panel). Furthermore, depletion of CD161 + cells from the cellular cultures, prior to stimulation with either anti-CD40 plus IL-4 or CpG, also resulted in a significant decrease in CD83 and CD38 expression, respectively (Fig. 5C, left panel for anti-CD40 plus IL-4 stimulation and right panel for CpG stimulation).
Finally, we explored the functional consequences of LLT1 triggering on GC B cells. The presence of rCD161 or the crosslinking anti-LLT1 Ab (2H7) did not have an impact on CD83 expression levels (Fig. 5D), CD38, or survival (Supplemental Fig.  4); however, LLT1 signaling resulted in a significant reduction of CXCR4 (Fig. 5E). This was also analyzed on sorted DZ B cells, where the effect of LLT1 crosslinking resulted in the downregulation of CXCR4 at mRNA level (Fig. 5F).
These findings suggest that LLT1 is involved in the transition of DZ B cells to the LZ through the downregulation of CXCR4, while contributing to enhanced levels of CD83 expression.

LLT1 is expressed in human B cell lymphomas
As a site of mutation and proliferation, many B cell lymphomas are derived from the GC. Therefore, having established LLT1 expression on GC B cells in vivo under normal conditions, we next examined whether such linked expression was maintained in the context of neoplasia. We assessed LLT1 expression in five B cell lymphoma cell lines. FL18, KHM2, L421, and SUDHL6 were positive for LLT1, whereas Burkitt's lymphoma-derived Namalwa cells where negative (data not shown). Using a commercially available LLT1 Ab (goat polyclonal, see Materials and Methods), LLT1 expression was evaluated in a series of lymphomas ( Fig. 6A-E). LLT1 was absent on non-GC-derived lymphomas, such as B-ALL and plasma cell neoplasms. In contrast, LLT1 expression was seen on most Burkitt lymphomas (14 of 19, 73.68%), follicular lymphomas (48 of 95, 51%), and also on Hodgkin lymphomas (7 of 16, 44%). Subdivision of the follicular lymphomas revealed the highest association of LLT1 on atypical phenotypes that have been previously described: Bcl-2 2 (59%)  and Bcl-2 2 and CD10 2 (62%) (Fig. 6F, lower left panel) (43). Interestingly, LLT1 expression was seen on lymphocytepredominant Hodgkin lymphomas and not classical Hodgkin lymphomas (Fig. 6F, lower right panel). Our data suggest that LLT1 is maintained during malignant transformation in some GC-derived lymphomas and may be a useful additional diagnostic marker.

Discussion
GC were first characterized .125 years ago. In recent years, our understanding of how these dynamic and complex reactions are orchestrated has advanced rapidly. In this study, we report the expression and function of the innate receptor pairing, LLT1 and CD161, within the GC reaction in humans.
A GC can be divided into two microanatomically distinct regions, the LZ and the DZ. Competition for Ag and Tfh cell help occurs within the LZ, whereas somatic hypermutation and proliferation are restricted to the DZ (reviewed in Ref. 40).
Although LLT1 expression was marginally higher on DZ cells, it was observed on all GC B cells in all lymphoid tissues tested. LLT1 expression was not, however, restricted just to GC B cells, as LLT1 + cells were detected outside of the GC. Many of these cells expressed markers associated with a plasmablast phenotype (CD45 low , CD20 2 , CD19 low , CD79a low , Ig + , IRF + , CD38 + , Ki67 + , CD138 2 ), although this does not exclude other cell types. In vitro differentiation of memory B cells into plasmablasts and then plasma cells demonstrated that LLT1 expression was lost as cultures progressed toward terminally differentiated plasma cells at day 10 (24). This was also confirmed on staining bone marrow-derived plasma cells. It appears that LLT1 expression on plasmablasts in vivo is a marker of recent exit from the GC, although whether it can also be independently upregulated remains to be addressed.
It was noted that LLT1 was frequently associated with dividing cell populations both in vivo and in vitro. Zhou et al. (23) also observed that OCIL (LLT1) mRNA expression was at its highest during the osteoblast proliferative phase of bone marrow cultures. In contrast, we observed that LLT1 was absent from plasma cells and both early and late memory B cells, which are nondividing cells. This suggests a link between proliferation and LLT1 expression, although we have not as yet been able to find any functional involvement of LLT1 during proliferation (data not shown). However, LLT1 might be a useful marker for the identification of GC B cells with high-affinity BCRs, as B cells with high-affinity BCRs are able to more efficiently present Ag to Tfh cells and consequently undergo more rounds of division (44).
To better understand the function of LLT1 within the context of the GC, we investigated potential sites of interactions by seeking the expression of its ligand, CD161. Lanier et al. (45) originally identified the expression of CD161 on both CD4 + and CD8 + T cells, which has been broadened to include innate lymphocyte subsets (15,(46)(47)(48)(49). However, murine homologs of CD161, NKR-P1B and NKR-P1D, are not fully analogous, as they differ both in expression profiles and function (reviewed in Ref. 50).
Unexpectedly, CD161 was observed on FDCs. This was particularly surprising, as it is uncommon for NK cell receptors to be expressed on nonhematopoietic cells, although relevant examples exist, for example, KACL on keratinocytes (51) and LLT1 on osteoblasts (23).
Given the close association of CD161 expression and T cells, CD161 expression was also investigated on Tfh cells (CXCR5 mid/high ). However, CD161 expression negatively correlated with CXCR5 expression, at both the protein and mRNA levels. A potential explanation for this may lie in the transcription factor Bcl-6. This master transcription factor drives maturation of T and B cells toward a full GC phenotype. As Bcl-6 expression increases it suppresses other lineage-associated transcription factors, such as retinoic acid-related orphan receptor gT, a transcription factor associated with CD161 expression (46,(52)(53)(54). Thus, it is plausible that Bcl-6 inhibits the expression of CD161 through the suppression of retinoic acid-related orphan receptor gT.
The LZ and DZ can be identified by the higher expression of either CD83 or CXCR4, respectively (1,3,(55)(56)(57). Movement between the zones is fundamental for efficient participation within the GC reaction (3,44,58). Entry into the DZ requires Tfh interactions and the expression of CXCR4, although CXCR4 2/2 B cells are still able to transit into a centroblast phenotype and acquire mutations, albeit at a lower level compared with their wild-type counterparts (3,44,58). Migration back to the LZ is associated with lower CXCR4 expression, but the underlying mechanisms have not been identified. In this study, crosslinking of LLT1 alone resulted in a reduction in the expression of CXCR4 in DZ B cells. Therefore, these data suggest that LLT1 triggering occurs within the LZ.
Once within the LZ, B cells interact with FDCs and Tfh cells and receive a number of costimulatory signals (59)(60)(61)(62). A key receptor in these interactions expressed by the B cells is CD83 (63)(64)(65). Triggering of LLT1 alone did not alter the expression of CD83 GC B cells. However, in the context of BCR or CD40 signaling, LLT1 triggering promoted CD83 expression. This suggests that LLT1 may also help enhance B cell interactions with both FDCs and Tfh cells. Combined with the downregulation of CXCR4, these data suggest that LLT1 triggering helps drive B cells through the different GC transitional states within the LZ (57).
LLT1's high expression within the GC, and its association with proliferating cells, suggests its use as a potential aid for the clinical diagnosis of B cell lymphomas, in particular non-Hodgkin lymphomas, which represent 4-5% of all cancers diagnosed (66). Treatment selection relies on accurate diagnosis, staging of the disease, and identification of adverse prognostic factors. Expressions of Bcl-6, CD10, and IRF4 are used in immunohistochemical stains to separate diffuse large B cell lymphomas into GC and non-GC subtypes (67,68). However, although immunohistochemistry is widely used, it is not always reliable, in particular when identifying follicular lymphomas with atypical phenotypes. LLT1, therefore, provides a potential new marker for the clinical investigation of lymphomas, and it warrants larger prognostic studies to provide more data on the independent significance of positive or negative stains. LLT1 and CD161 are both CLRs encoded within the NKC, located adjacent to one another (4-7, 51, 69). Expression of other NKC proteins within the GC is limited to CD69 on both B cell and Tfh cells; however, a function for CD69 within the GC has not been described. The role of particular lectins in the modulation of BCR responses has been explored before. For instance, CD22 and CD72 have been shown to downmodulate BCR-mediated signals (70). Furthermore, the C-type lectin CD23, the low-affinity receptor for IgE, is also relevant in the context of B cell responses (71), as it acts as a ligand for CD21 within the BCR complex. Interestingly, CD23 has been found highly expressed in a proportion of FDCs. Thus, the data presented in this study are consistent with a contribution of C-type lectins to B cell responses through FDC-B cell interactions. This is particularly meaningful, as CLRs, once thought to be essentially involved in innate immunity, have now been shown to play a prominent role in the GC reaction, a hallmark of adaptive immunity.
The data presented in the present study suggest a role for innate components of the human immune system, such as LLT1 and CD161, in the GC reaction. Our data suggest that LLT1 triggering may play a key role in the GC reaction promoting B cell activation and downregulating CXCR4, thus helping to promote an LZ phenotype and facilitating both FDC and Tfh interactions.