BILL-Cadherin/Cadherin-17 Contributes to the Survival of Memory B Cells

Memory B cells (MBCs) and long-lived plasma cells (LLPCs) are responsible for immunological “memory”, which can last for many years. The long-term survival niche for LLPCs in the bone marrow is well characterized; however, the corresponding niche for MBCs is unclear. BILL-cadherin/cadherin-17 (CDH17) is the only member of the cadherin superfamily that is expressed on mouse B lymphocytes in a spatiotemporally regulated manner. Here, we show that half of all MBCs regain expression of CDH17 during the later stage of development. The maintenance of high affinity antigen-specific serum antibodies was impaired in CDH17-/- mice and the number of antigen-specific MBCs was reduced as compared to wild-type mice (WT). Also, specific responses to secondary antigens were ablated in CDH17-/- mice, whereas primary antibody responses were the same as those in WT mice. Cell cycle analysis revealed a decline in the proliferation of CDH17- MBCs as compared to CDH17+ MBCs. In addition, we identified a subpopulation of splenic stromal cells, MAdCAM-1+ blood endothelial cells (BEC), which was CDH17+. Taken together, these results suggest that CDH17 plays a role in the long-term survival of MBCs, presumably via an “MBC niche” comprising, at least in part, BEC in the spleen.


Introduction
BILL-cadherin/cadherin-17 (CDH17) is a cell adhesion molecule that belongs to the cadherin superfamily, a large group (more than 100 members) of cell adhesion molecules with properties similar to those of integrins and selectins. Cadherins are Ca 2+ -dependent adhesion molecules characterized by their unique extracellular domains, which primarily comprise multiple cadherin-repeats. Cadherins primarily mediate homotypic (cell to cell) adhesion; therefore, they play important roles in intercellular recognition during embryogenesis and morphogenesis [1,2].
CDH17 contains seven cadherin domains and has no catenin-binding region within its cytoplasmic domain; the latter feature means that CDH17 is classified as a non-classical cadherin [3,4]. CDH17 requires Ca 2+ for homotypic adhesion [3,5]; however, heterotypic adhesion to E-cadherin has been reported [6]. In mice, CDH17 is expressed in the spleen, bone marrow, and intestine [3,7], whereas in rats it is also expressed in the liver [4].
We previously showed that precursor B cells express CDH17 during early development in the bone marrow [8]. T cells, however, do not express CDH17 [3,8]. CDH17 is expressed during the pro-B/pre-B-I stages before being downregulated during the pre-B-II stage; it is then upregulated again on immature B cells [3]. CDH17-deficient mice have an increased number of pro-B cells and a reduced number of immature B cells, indicating that CDH17 plays a role(s) in early B cell development (i.e., during transition from the pro/pre-B-I stage to the pre-B-II stage) [8]. Also, the size and the number of germinal centers (GC) in non-immunized CDH17 -/mice is reduced, and the antibody response to a T-independent antigen is decreased as compared to WT mice [8]. These observations suggest that CDH17 might also play a role in late B cell development.
The aim of the present study was to compare T cell-dependent antigen-specific antibody responses to nitrophenylated chicken gammaglobulin (NP-CGG) in wild-type (WT) mice with those in CDH17 -/mice. The results showed that CDH17 contributes to the long-term survival of memory B cells. Furthermore, we identified a population of MAdCAM-1 + blood endothelial cells (BEC) that is CDH17 + . Taken together, these results suggest that CDH17 is involved in the long-term survival of MBCs, and that CDH17 + BEC are a candidate for the elusive "MBC niche". The findings of the present study provide crucial clues that will improve our understanding of the mechanisms underlying long-term MBC survival.

Materials and Methods
Mice and ethics statements CDH17 knock-out mice (BT262) were generated as previously described [8]. The KO mice were backcrossed onto a C57BL/6 background for ten generations. CDH17 +/+ and CDH17 -/homozygous littermates were used for all experiments. All mice were bred and maintained in a specific-pathogen-free (SPF) facility. All animal experiments were performed according to institutional guidelines and with the approval of the National Institute of Infectious Diseases Animal Care and Use Committee (Permit Number: 213045-2). Mice were housed under a 12 hour light/dark cycle, and provided with food and water ad libitum. All efforts were made to minimize suffering. Mice were immunized intraperitoneally with antigen in a volume of less than 200 μL containing 50% Alum adjuvant. Blood samples were drawn from the tail vein and less than 100 μL was collected each time. Mice were euthanized by carbon dioxide inhalation and the spleens were explanted.

ELISA
Flat-bottomed 96-well plates (Immuno-MaxiSorp; Nunc 442404) were coated with 5 μg/mL of NP 1.6 -BSA (to detect high affinity anti-NP antibodies) or with 5 μg/mL NP 4 -BSA or NP 16 -BSA (to detect total affinity anti-NP antibodies) and incubated at room temperature for 2 h or at 4°C overnight, followed by incubation with 1% BSA in PBS/0.05% Tween 20 (Sigma, P-1379). Sera collected from immunized littermates were serially diluted, added to the plates, and incubated for 1 h at room temperature, followed by incubation with HRP-anti-mouse IgM and HRP-antimouse IgG1 antibodies. Hydrogen peroxide in citrate buffer and O-phenylenediamine were used as the chromogen. Optical density was measured at 490 nm. NP-specific IgM and IgG1 monoclonal antibodies (established in-house) were used as standards to calculate the relative antibody titers and affinities.

Confocal microscopy
Frozen spleen sections (6 μm thick) were treated with the Catalyzed Signal Amplification (CSA) System (Dako, K1500) to block endogenous peroxidase activity, followed by treatment with the Endogenous Biotin-Blocking Kit (Invitrogen, E21390) to block endogenous biotin and 2.4G2 to block Fcγ receptors. Sections were then stained with biotin-BD1B. After washing, the tissue sections were treated with a streptavidin biotin complex (CSA system) followed by the Individual Indirect Tyramide Reagent (PerkinElmer, SAT700001EA) to amplify the biotin-BD1B signal. Sections were then stained with Qdot 655-streptavidin, Alexa Fluor 647-antimouse CD169, Brilliant Violet 421 anti-mouse IgG1, and Alexa Fluor 488-anti-mouse MAd-CAM-1. Immunostained tissue sections were visualized under a Zeiss LSM710 confocal microscope. Data were analyzed using ZEN software.

Statistical analysis
Microsoft Excel was used for statistical analysis. The error bars on the figures represent the standard deviation from the mean. Serum antibody titers, the differential between the preboost and post-boost titers, the percentage of MBCs, and the MBC cell cycle indices were tested using the Mann-Whitney U-test. All other data were tested using Student's t-test. In all cases, differences were considered significant at p<0.05 ( Ã p<0.05, ÃÃ p<0.01, and ÃÃÃ p<0.001).
The maintenance of serum antibody affinity is impaired in CDH17 -/mice To elucidate the role of CDH17 in affinity maturation and MBC formation, we compared the antibody responses in WT and KO mice. Mice were immunized with NP-CGG, and antigenspecific serum IgM ( Fig. 2A) and IgG1 (Fig. 2C) antibody titers were measured. Antibody affinity for NP 1.6 , NP 4 , and NP 16 -BSA was examined in an ELISA. There were no significant differences in the total NP-specific serum IgM and IgG1 titers between WT and KO mice at 45 weeks after the primary immunization. However, when we examined WT and CDH17 -/mice for the presence of high affinity NP-specific serum antibodies, we found that the affinity of NP-specific serum IgM was lower in CDH17 -/mice than in WT mice at 4-6 weeks post-immunization (Fig. 2B), and that the affinity of IgG1 was lower in CDH17 -/mice from week 31 post-immunization (Fig. 2D). Because there were no significant differences in the anti-NP 16 total antibody titer between WT and CDH17 -/mice, the lower affinity of IgM and IgG1 in CDH17 -/mice must be due to a reduction in the high affinity anti-NP 1.6 and anti-NP 4 antibody titers. Since no significant differences were detected between WT and CDH17 -/mice during the early stages of immunization, the results also suggest that CDH17 becomes important during the later stages of an immune response, e.g., by promoting long-term maintenance of high affinity antigen-specific antibody titers.
The secondary antibody response is markedly impaired in CDH17 -/mice High affinity antibody titers are thought to be maintained by MBCs and LLPCs. Because we found PCs to be CDH17and about half of MBCs to be CDH17 + (Fig. 1), one possible explanation for the loss of high affinity antibodies could be impaired maintenance of high affinity-NPspecific MBCs in CDH17 -/mice. To test this hypothesis, we examined secondary antibody responses to NP-CGG. WT and CDH17 -/mice were boost immunized with NP-CGG in the absence of alum adjuvant at around 50 weeks post-priming. The increases in NP-specific serum antibody titers at Day 5 post-boost were then compared with those before the boost. We found that the NP-specific serum IgM and IgG1 titers in WT mice increased markedly at Day 5 postboosting. On the other hand, there was only a small increase in boosted CDH17 -/mice (Fig. 3). The difference was most evident when changes in antibody titers were compared between individual mice. Antibody titers increased significantly in WT mice, whereas only a small increase, or even a decrease, was observed in individual CDH17 -/mice (Fig. 3A).
Because the NP-specific IgG1 titers at the time of the boost immunization were different between WT and CDH17 -/mice, we next compared the net difference in NP-specific serum IgG1 antibody titers before and after the secondary immunization (Fig. 3B). The differences in NP-specific serum IgM (excluding total NP 16 -specific antibodies) and IgG1 titers were significantly higher in WT mice 5 days after the secondary immunization, but not in CDH17 -/mice. This bias was more evident for IgG1 antibodies (Fig. 3C). Because there were no significant differences in the NP-specific serum antibody titer between WT and CDH17 -/mice after the primary immunization, it appears that the defect in CDH17 -/mice is specific for the secondary antibody response.
In conclusion, these results showed that the secondary antibody response is functional, but impaired, in CDH17 -/mice, indicating that CDH17 affects the production of antigen-specific antibodies during the secondary immune response.

CDH17 contributes to the long-term maintenance of MBCs
Because CDH17 -/mice show an impaired secondary antibody response, we next examined whether antigen-specific MBCs were less common in CDH17 -/mice than in WT mice. WT and CDH17 -/mice were immunized with NP-CGG and changes in the percentage of NPspecific IgG1 + MBCs (Lin -B220 + IgD -IgM -CD38 + IgG1 + NIP + ) were monitored by flow cytometry over a long time period (Fig. 4). There was no significant difference in the percentage of NP-specific MBCs between WT and CDH17 -/mice during the first 2-6 weeks postimmunization (Fig. 4B), indicating that the initial rate of MBC development is not affected by CDH17 deficiency. Moreover, the number of progenitor cells, including mature B cells, T1 cells, T2 cells, antigen-specific GC B cells, and the total number of IgG1 + MBCs, was almost the same in WT and CDH17 -/mice (S2 and S3 Figs.). However, the number of NP-specific MBCs in CDH17 -/mice was significantly reduced (to about one-third of that in WT mice) after longer periods of time (50-67 weeks) post-immunization (Fig. 4B). These results indicate that CDH17 plays an important role in the long-term maintenance of antigen-specific MBCs, which becomes clear after 50 weeks post-antigen exposure.
The "in vivo" proliferation of CDH17 -IgG1 + MBCs is significantly retarded To elucidate the role of CDH17 in the long-term maintenance of MBCs, we next examined the influence of CDH17-deficiency on the rate of cell cycle entry, i.e., cell proliferation. Because the number of antigen-specific IgG1 + MBCs was very small at 50 weeks post-primary immunization, we performed cell cycle analyses to examine switched IgG1 + MBCs rather than antigenspecific IgG1 + MBCs. WT and CDH17 -/mice (17 ± 3-months-old) were injected intraperitoneally with BrdU and sacrificed 1 hour later. The cell cycle status of switched IgG1 + MBCs (B220 + IgD -IgM -CD38 + IgG1 + ) was then examined by flow cytometry (Fig. 5C). We analyzed the cell cycle status of switched IgG1 + MBCs in CDH17 + and CDH17populations separately, and found that CDH17 -IgG1 + MBCs from WT mice showed the same cell cycle distribution as those from CDH17 -/mice. On the other hand, the cell cycle progression of CDH17 + switched IgG1 + MBCs was higher in WT mice (Fig. 5D). These results suggest that CDH17 plays a role in the long-term maintenance of MBCs by increasing the rate at which MBCs enter the cell cycle.
A subpopulation of spleen BEC are CDH17 + Cells that are retained for long periods of time, such as hematopoietic stem cells, require a specialized niche for survival [12,13]. MBCs also require such a survival niche to support their long-term maintenance; however, no such niche has been identified. In general, it is thought that appropriate adhesion molecules are required for cells to reside within a specific niche. Therefore, we hypothesized that CDH17 holds MBCs within a specific survival niche via homotypic adhesion. We next examined CDH17 + stromal cells as a possible constituent of the MBC survival niche. Mouse spleens were treated with dispase and collagenase IV, and the expression of CDH17 by various stromal cells was examined by flow cytometry (Fig. 6A). We found that a fraction of MAdCAM-1 + BEC expressed significant levels of CDH17 (Fig. 6B). However, we noticed that the endothelial cell surface was damaged by dispase/collagenase IV treatment. Although this may cause relatively high background signals, resulting in a lower signal intensity for CDH17, we found that BEC expressed statistically significant levels of CDH17. In this context, we found that fibroblastic reticular cells (FRCs) were also CDH17 + ; however, in this case, the level of specific staining was not statistically significant due to the large standard deviation (Fig. 6B). Thus, FRCs may be vulnerable to the harsh conditions caused by dispase/ collagenase IV. Taken together, these results suggest that a CDH17-expressing subpopulation of MAdCAM-1 + BEC is a candidate constituent of the MBC survival niche.
In the spleen, MAdCAM-1 is expressed on cells lining the marginal sinuses, which partly overlap with the MOMA-1 + marginal zone [14]. CDH17-expressing cells were localized in the region overlapping these MOMA-1 + cells and MAdCAM-1 + cells; also, MAdCAM-1 + CDH17 + cells were present in this region (although they were infrequent) (Fig. 6C). Overall, about 6% of cells in the MZ sinus (MAdCAM-1 + ) were MAdCAM-1 + Cdh17 + BECs (Fig. 6E). Class- switched IgG1 + B cells (B220 + IgM -IgD -IgG1 + ) comprise about 0.01% of the splenic B220 + B cell population, and about 70% of IgG1 + B cells show a CD38 + MBC phenotype (B220 + IgG1 + IgM -IgD -CD38 + ). It is not easy to identify MBCs in the spleen by histology; however, small numbers of IgG1 + B cells were consistently observed in association with white pulp, red pulp, or the MZ. Indeed, the number of IgG1 + B cells found in the MZ was associated with cells those expressing MAdCAM-1 and CDH17 (Fig. 6C). The memory-like IgG1 + B cells associated with MAdCAM-1 + CDH17 + BEC were CDH17 + (Fig. 6C, bottom panel), suggesting that CDH17 mediates adhesion. Other memory-like IgG1 + B cells found on histological examination were also CDH17 + . The number of IgG1 + cells associated with the MZ area was significantly lower in KO mice, whereas IgG1 + cells on the red pulp or white pulp were not affected by CDH17-deficiency (Fig. 6D). These observations further support the idea that a CDH17expressing subpopulation of MAdCAM-1 + BEC located near marginal sinus constitutes a possible "MBC niche" in the spleen.

Discussion
We showed that almost half of MBCs express CDH17. We examined the expression of chemokine/cytokine receptors and other adhesion molecules (which are reported to be expressed on MBCs) on CDH17 + and CDH17 -MBCs [11,15]: CXCR3, CCR6, integrin-β2, integrin-αL,  Cadherin-17 Contributes to the B Cell Memory VE-cadherin, CD80, and CD273 (S1 Fig.). Of these, integrin-αL, integrin-β2, and CD273 were expressed at slightly higher levels on CDH17 + MBCs than on CDH17 -MBCs. Integrin-α L β 2 (LFA-1) is involved in B cell localization to the MZ [16], suggesting that CDH17 + MBCs are more likely to localize in the MZ than CDH17 -MBCs. Also, MBCs are classified into five subsets according to their differential expression of CD273, CD73, and CD80; CD80 hi and/or CD273 hi cells are considered to be more "memory-like" [15]. The data presented herein suggest that CD273 hi CDH17 + MBCs are more memory-like than CDH17 -MBCs. [17] CDH17-deficient mice showed a reduction in the titer of high affinity anti-NP antibodies over time, without any obvious difference in the total antibody titer. Fig. 2C shows that overall anti-NP16 IgG1 antibody titers changed: titers were almost equal in WT and KO from Week 2 to Week 31, but were slightly lower in KO mice at week 45; however, the difference was not statistically significant. The titers of high affinity IgG (NP1.6 and NP4) antibodies also changed: they were almost equal in WT and KO mice at Weeks 2-6, but were slightly lower in KO mice at Weeks 17-45. One possible explanation for this is that CDH17 is involved in the maintenance of MBCs but not in the formation of MBCs or PCs. We found that during the early phase of the antibody response (Weeks 2-6), there was no difference in antibody titers between WT and KO because PCs developed equally well in both. During the later phase (Weeks 17-45), high affinity antibodies produced by PCs would be replenished; however, these antibodies would be different from those produced by maintained high affinity MBCs. The maintenance of high affinity MBCs was impaired in CDH17 -KO mice, and the differentiation of PCs from MBCs decreased; therefore, the replenishment of high affinity antibodies also decreased. LLPCs, which developed during the early phase, continued to produce low affinity antibodies for a long time; thus the amount of antibodies produced by LLPCs in WT and KO was the same because PCs do not express CDH17. Hence, after a long time post-antigen immunization, the high affinity antibody ratios in KO mice decreased and the total affinity antibody titer also decreased (slightly), even though the differences were not statistically significant.
Our data showed that the cell cycle turnover of CDH17 + IgG1 + MBCs was faster than that of CDH17 -MBCs. Recent reports examining CDH17 + cancer cells reveal that cell cycle progression is slowed after CDH17 knockdown [18,19]. The Wnt signaling pathway forms part of the cadherin signaling pathway [20], which regulates the expression of cyclin D1. Immunoprecipitation experiments show that CDH17 associates with β-catenin, although CDH17 has no cateninbinding domain in its intracellular region [21]. These findings suggest that CDH17 accelerates cell cycle progression in CDH17 + switched IgG1 + MBCs via the Wnt signaling pathway.
In humans, CDH17 is expressed in the fetal liver and gastrointestinal tract [22] but is downregulated in the adult liver and gastrointestinal tract; however, it is upregulated in gastric cancer, esophageal cancer, pancreatic cancer, and hepatocarcinoma [18,23,24]. CDH17 is also expressed on human B cell lines [25] and CD19 + peripheral blood lymphocytes [26], suggesting a role for CDH17 in human B cells. However, the exact role of CDH17 in human MBCs remains unclear.
It is thought that MBCs and memory T cells require a specific survival niche for long-term maintenance [27][28][29][30]. The survival niches for memory CD4 + T cells and LLPCs are well documented [31][32][33]. Memory CD4 + T cells and LLPCs are maintained by IL-7 + stromal cells and by CXCL12 + stromal cells, respectively, in the bone marrow. The number of maintained memory cells is limited by the available space within the survival niche. The methods used to create space for newly emerged memory cells within the survival niche have been reported [34]. However, it is still unclear whether stromal cells are required for MBC survival and, if so, what kind of stromal cells were required [27,35,36]. Here, we showed that a fraction of MAdCAM-1 + BEC express CDH17. According Kraal et al., MAdCAM-1 + vascular cells in the spleen are marginal sinus endothelial cells [14]. MBCs reside within the MZ in rodents and humans [37][38][39].
We previously showed that the expression of CDH17 is especially high in the MZ [3] (see also Fig. 6). Thus, we consider that the CDH17 + BECs identified in the present study comprise a fraction of the marginal sinus that offers an area suitable for CDH17-mediated homophilicadhesion to MBC. Interestingly, recent studies show that endothelial cells within sinusoids comprise part of the stromal cell population that constitutes the hematopoietic stem cell niche [16,27,40]. However, we cannot exclude the possibility that other CDH17 + sessile-type cells, such as FRCs or other as-yet-unidentified cells, contribute to the MBC survival niche. The frequency of MAdCAM-1 + CDH17 + BECs in the MZ sinus (MAdCAM-1 + ) was about 6% (Fig. 6E). We do not think this is too low to support MBCs because a recent report shows that FRCs comprise about 30% of total stromal (CD45 -) cells and govern B cell homeostasis in mouse lymph nodes [41]. The memory-like IgG1 + B cells shown in Fig. 6C are CDH17 + . Collectively, these data suggest that a fraction of MAdCAM-1 + BEC, which expresses CDH17, comprises the MBC survival niche and that MBCs are anchored within this niche by homotypic adhesion mediated by CDH17.
Recent reports on MBC localization in the mouse spleen identify the periphery of contractile GC as the region in which MBCs are maintained [28,29,39]. As for the splenic microarchitecture around the GC and MZ, the inner layer of MOMA-1 + metallophilic macrophages occasionally associates with the outer layer of GC in which CDH17 is also expressed. The difference between their experiments and ours is the timing of MBC examination: 8-25 weeks after the primary immunization in their case and after around 60 weeks in ours. The respective difference in the age of the MBCs would be reflected in the quality of the MBC population. In this context, it is noteworthy that we examined two populations of MBCs: CDH17 + and CDH17 -. Our current hypothesis is that these two types of MBC are generated by asymmetric cell division ("self-renewal" and "differentiation-oriented") on CDH17 + BEC, as observed for hematopoietic stem cells [42]. We speculate that the fate of CDH17 + and CDH17 -MBCs is as follows: whereas CDH17 + MBCs remain within the CDH17 + BEC niche and are maintained by homeostatic proliferation (which is presumably induced by catenin-mediated signaling via cell surface CDH17), CDH17 -MBCs may leave the niche and differentiate into PCs, which partly replenish long-term antigen-specific antibodies. The differentiation of CDH17 -MBCs into PCs requires antigen. We speculate that, after asymmetric cell division, CDH17 -MBCs leave the MBC niche and circulate throughout the body until they encounter an antigen. If CDH17 -MBCs do encounter an antigen, the cell is activated and differentiates into PCs to produce high affinity antibodies (if it does not encounter an antigen, the cell may die). It was reported that antigenspecific memory follicular T cells (Tfh) are in the draining regional lymph node [43]. We speculate that such cells may play a role in the activation of circulating CDH17 -MBC. It would also be possible that circulating CDH17 -MBCs are activated by polyclonal stimulation such as bystander T cell help and CpG DNA [44]. In KO mice, CDH17 -MBCs localize out of the MBC niche in the spleen by adhering to molecules other than CDH17; hence, the cells lack a homeostatic proliferation signal. [16] [45,46][6][21] [11]In conclusion, we showed that CDH17 plays an important role in the long-term survival of antigen-specific MBCs, thereby maintaining the capability for rapid antibody production during a secondary immune response. We found that a fraction of MAdCAM-1 + BEC may represent a candidate constituent of the MBC survival niche. Further elucidation of the molecular mechanisms underlying the role of CDH17 in MBC maintenance may make it possible to establish improved vaccination protocols. (A) IgG1 + MBCs (Lin -B220 + IgD -IgM -CD38 + IgG1 + ) obtained from CDH17 -/mice and their WT littermates at 52 weeks after primary immunization with NP-CGG in alum were analyzed by flow cytometry. Numbers represent the percentage (%) of the indicated cell populations in the respective parental gates (shown on top of the panels). The same experiments described in Fig. 4. (B) The percentages of IgG1 + MBCs are plotted on a bar graph. The y-axis shows the percentage of IgG1 + MBCs (Lin -B220 + IgD -IgM -CD38 + IgG1 + ) in the respective B220 + parental gate. The number of weeks post-immunization is shown for each bin. (TIF)