Changes in Tonsil B Cell Phenotypes and EBV Receptor Expression in Children Under 5‐Years‐Old

Palatine tonsils are principally B cell organs that are the initial line of defense against many oral pathogens, as well as the site of infection for others. While the size of palatine tonsils changes greatly in the first five years of life, the cellular changes during this period are not well studied. Epstein Barr virus (EBV) is a common orally transmitted virus that infects tonsillar B cells. Naïve B cells are thought to be the target of primary infection with EBV in vivo, suggesting that they are targeted by the virus. EBV enters B cells through CD21, but studies of older children and adults have not shown differences in surface CD21 between naïve B cells and other tonsil B cell populations.

germinal centers as well as extrafollicular T cell regions (3). Palatine tonsils change morphologically with age, the largest size being seen in children between 3-and 5years-old, and subsequently declining in size through life [reviewed in (4)]. As secondary lymphoid organs, palatine tonsils are also the sites of the final stages of B cell development, which involves trafficking of transitional B cells, antigen exposure, as well as affinity maturation and somatic hypermutation (2,5). B cells are the principal cellular component of palatine tonsils, comprising between 50 and 90% of subepithelial lymphocytes (3,6). B cells and their terminally differentiated form, plasma cells (PC), are the critical cell types of humoral immunity; with immunoglobulin G (IgG)-producing B cells responsible for binding blood-borne pathogens and IgAproducing B cells responsible for mucosal pathogens (7). Tonsils are known to be an important site for the development of antibody responses during exposure to oral antigens (8). The physical location of palatine tonsils in the oropharynx puts immune cells in close contact with pathogens in the oral cavity (3), and gives tonsils their important role in the immune defense against oral pathogens (8).
In addition to their role in the protection against oral pathogens, tonsils are also the sites of infection for some oral pathogens (9,10). One oral pathogen that is associated with significant morbidity and mortality is Epstein Barr virus (EBV) (11). The most common manifestation of EBV-associated disease is infectious mononucleosis, which is a self-limiting febrile illness of EBV-na€ ıve adolescents in developed countries (12). EBV is also strongly associated with Burkitt lymphoma, one of the most common pediatric malignancies in equatorial Africa (13). Children in equatorial Africa are frequently infected with EBV early in life, with earlier age of infection correlating with increased risk for Burkitt lymphoma (14). In the US, early age of EBV infection was found to be associated with socioeconomic status (15). During primary infection, EBV infects na€ ıve B cells in the tonsils through its receptor, CD21 (16,17). While EBV can infect all B cell types in vitro (18), na€ ıve B cells more readily immortalize into proliferating lymphoblasts (19), and are the only cells to display the growth program classic for primary infection in vivo, suggesting that these cells are preferentially infected by the virus (16). While it is known that blocking CD21 severely inhibits the entry of EBV into B cells (20), studies in patients of undefined ages have not found differential CD21 surface expression on na€ ıve tonsil B cells relative to other B cell types (18,19). Whether CD21 expression differs among tonsil B cells during early years of life is yet to be defined.
In clinical medicine, lymphocytes and other white blood cells are commonly enumerated in peripheral blood as an indicator of the function of the immune system as a whole (21). It has been well documented that peripheral blood lymphocytes differ with age and sex (22,23), but whether these differences apply for lymphocytes in secondary lymphoid tissues of young children is not as well described. One early study using two-color flow cytometry described age-related changes in tonsil B cell markers with age (24), but B cell subpopulations could not be identified at that time. Since then, the proportions of different B cell subsets in human palatine tonsils have been documented for adults, adolescents, and older children (25), but to our knowledge no study has measured the changes in B cell subsets of human tonsils in early childhood (< 4 years of age). And despite interest in the cell types and function of tonsil B cells (26)(27)(28), studies have not focused on children under 5 years of age. This age range is critical for understanding the unique immune interface of the palatine tonsils when they are most highly developed (4) and during the ages when palatine tonsils play their most important physiological role in defense against oral pathogens.
Palatine tonsillectomy with or without adenoidectomy is a routine procedure performed in children and adults to remove hypertrophied tonsil tissue to improve breathing (29). Since palatine tonsils grow in size during the first several years of life, they can cause airway obstruction, leading some parents and children to elect to undergo tonsillectomy. Today almost 80% of tonsillectomies are performed for upper airway obstruction, in the setting of a baseline immune status at the time of removal, rather than infection (29). Tonsils are also a rich source of B cells and have been used for multiple immune phenotyping and microbiology studies (2,5,8,25,30,31). Unfortunately these studies frequently use different antibody combinations to identify different cell populations, resulting in an impaired ability to share findings among groups. Here we propose a standardized immune phenotyping strategy based largely on a commonly used B cell phenotyping panel for peripheral blood lymphocytes (32).
This study set out to answer several unanswered questions. First, what are the reference ranges of B cell subsets in tonsil tissue for young children between 1 and 5 years of age? Second, do B cell subsets differ with age, sex, and EBV infection in these children? Third, what is the tonsillar B cell expression pattern of CD21, the EBV receptor, during the early years of life? Here we define reference ranges of tonsil B cell subsets in young children based on a standardized immunophenotyping panel, and have uploaded our flow cytometry files to an open source repository for comparison with future study populations. We also report that mature na€ ıve B cells are increased in tonsil mononuclear cells (TMC) early in life and, for the first time, that they express increased levels of CD21, suggesting a mechanism by which these cells may be preferentially infected by EBV in vivo.

Study Population
Patients between 1 and 5 years of age undergoing routine tonsillectomy for tonsillar hypertrophy at a large academic medical center were enrolled in the study. Consent was obtained from parents of study participants during a preoperative visit. A total of 37 children were enrolled in the study. Children were excluded for clinical evidence of acute infection, or if tonsils were not processed within 4 h from removal. The study was approved by the institutional review board at SUNY Upstate Medical University, Syracuse, New York, and was carried out according to the Declaration of Helsinki.

Cell Preparation
Tonsils were surgically removed and transferred into physiological saline for transport to gross pathology. Upon receiving a gross pathological diagnosis of benign hypertrophy, tonsils were placed in 25 mL of RPMI 1640 medium supplemented with 10% fetal bovine serum, supplemented with penicillin/streptomycin and L-glutamine. Tonsils were transported to the laboratory for preparation within 4 h of surgery. Upon arrival tonsils were weighed, manually homogenized, and filtered twice over nylon to remove connective tissue and cell aggregates. Mononuclear cells were isolated over Ficoll-Hypaque and washed 3-4 times in complete RPMI 1640 medium. Mononuclear cells were frozen in complete RPMI 1640 with 20% FBS and 10% DMSO in liquid nitrogen prior to flow cytometric analysis.

Flow Cytometry
Cells were thawed and washed with complete RPMI 1640 medium. Two million viable cells per tonsil were prepared for flow cytometry. Cells were stained for viability using Fixable Aqua viability dye (Thermo Fisher Scientific, Waltham, MA) as per the manufacturer's instructions. Cells were Fc blocked for 20 min with 20 lL Fc blocking reagent (eBioscience, San Diego, CA) in 30 lL flow buffer containing phosphate buffered saline, 1% BSA and 0.1% sodium azide. Cells were then stained with an antibody cocktail containing antibodies against CD3, CD19, CD10, CD27, IgD, IgM, CD24, CD38, CXCR4, and CD21 in a staining volume of 100 lL (Table  1). Cells were fixed in 2% paraformaldehyde in PBS and run on a BD LSR Fortessa flow cytometer equipped with FACS Diva software (BD Biosciences, San Jose, CA).
Flow files were analyzed using FlowJo version 9.5.2 (Tree Star, Ashland, OR). Before analysis, all flow files were biexponentially transformed to better visualize separation of subsets.

FACS Isolation of B Cell Subsets
Tonsil cells were thawed and washed in complete RPMI medium. Cells were stained with Zombie Aqua viability dye (BioLegend, San Diego, CA) as per the manufacturer's instructions then suspended in blocking buffer (PBS, 5% bovine serum albumin, and purified antihuman Fc block) for 30 min prior to staining with antibodies. 5 3 10 7 cells were stained with antibodies against CD19, CD24, CD38, CD27, IgD, and CXCR4 for 30 min. Fluorescent-activated cell sorting (FACS) was used to isolate purified na€ ıve B cells, total memory B cells (non-class-switched, classic, and atypical), centroblast, and centrocytes, gating on populations as shown in Fig. 1. Postsort analyses were performed to determine purity of each isolated B cell population. Sorts were performed on a FACS Aria II flow cytometer (BD Biosciences).

EBV Detection by Quantitative PCR from Tonsil Cells
DNA was extracted from 1 3 10 7 TMC or isolated B cell subsets using a Qiagen DNeasy kit by the blood and tissue protocol as per the manufacturer's instructions (Qiagen, Valencia, CA). DNA was eluted in 200 lL H 2 O and the concentration was measured on a NanoDrop 2000 spectrophotometer (Thermo Scientific). A total of 120 ng of template DNA, corresponding to approximately 20,040 human cells (167 diploid human cells/ng) were run per PCR reaction. The qPCR protocol was as follows: 10 min at 958C, 45 cycles of 15 s at 958C and 1 min at 608C, followed by a melt curve with a 0.58C step down. The PCR was performed on an iCycler iQ thermocycler equipped with an optical module (BioRad Laboratories, Hercules, CA). iQ supermix was used for all reactions (BioRad Laboratories, Hercules, CA). Previously designed primers were used to detect a 70 base pair region of EBV BALF-5 (33). Of lymphocytes, doublets were excluded using FSC-W/H and SSC-W/H sequentially. Of singlets, live cells were selected as viability dye negative. Viable cells were then plotted by CD19 versus CD3. B cells were gated as CD19 1 , CD3 -, and T cells were gated as CD3 1 , CD19 -. Next, B cells were plotted by CD38 versus CD24 differentiation markers. Transitional B cells were identified as CD38 hi , CD24 hi ; mature B cells as CD24 1 , CD38 -; germinal center B cells as CD38 1 , CD24 -; and PC as CD38 hi , CD24 -, and CD27 hi . Of mature B cells, IgD versus CD27 allowed discrimination between mature na€ ıve (IgD 1 , CD27 -), non-switched memory (IgD 1 , CD27 1 ), classical memory (IgD -, CD27 1 ), and atypical memory (IgD -, CD27 -). Of germinal center B cells, centroblasts were identified as CXCR4 1 and centrocytes were gated as CXCR4 -.

Statistical Analysis
All statistical analyses were performed using GraphPad Prism, version 5.0 b (Graphpad Software, La Jolla, CA). Age-related changes in cell frequencies were assessed by linear regression. Cell subset frequencies were compared by sex and EBV positivity using student's t test. CD21 MFI fold change was compared between B cell types using One-way ANOVA. When appropriate, such as comparing CD21 MFI for different B cell populations by EBV status, Two-way ANOVA was used. Statistical significance with ANOVA was determined using Bonferroni's post-test. A P value of 0.05 was considered statistically significant.

Tonsil Characteristics
A total of 33 tonsils were analyzed in this study. There were 23 male and 10 female study participants. The mean age of study participants was 37 months, with a range of 17-59 months. The mean weight of tonsils obtained was 4.34 grams (g), with a standard deviation of 2.01 g. The mean number of mononuclear cells per gram of tonsil tissue was 3.27 3 10 8 6 1.16 3 10 8 . There was no difference in the number of mononuclear cells per gram of tissue obtained from male versus female tonsils (P 5 0.36). There was also no relationship between the number of mononuclear cells/g and patient age (P 5 0.31, R 2 5 0.034).

Phenotypic Characterization of Tonsil B Cell Subpopulations
To determine the B cell subpopulation frequencies in human tonsils, TMC were phenotyped by flow cytometry according to standardized panels for peripheral blood phenotyping (32,34), with some modification (Table 1). Figure 1 represents the gating strategy used to identify B cell subpopulations after exclusion of cellular debris by forward and side scatter area, clumps of cells by width and height of forward and side scatter, dead cells (viability dye positive), and T cells (CD3 1 , CD19 -). As a control, a subset of tonsils were analyzed for B cell subsets before and after freezing, the result of which had no notable effects on any cell phenotype studied (data not shown). B cells were gated as CD19 1 , CD3 -, and subpopulations were determined on the basis of CD38, CD24, IgD, IgM, CD27, CD10, and CXCR4. The relative fluorescence intensities for different markers are shown in Table 2. Transitional B cells (Tr) were identified as CD38 hi , CD24 hi , CD27 -, and were also IgD 1 , IgM 1 , and CD10 1 . Mature B cells (M) were identified as CD38 1 , CD24 1 . This group was further subdivided based on expression of IgD and CD27 as mature na€ ıve (MN: IgD 1 , CD27 -), non-class-switched memory (NSM: IgD 1 , CD27 1 ), classic memory (CM: IgD -, CD27 1 ), and atypical memory, which are seen in a variety of chronic infections (35,36) (AM: IgD -, CD27 -) (37). As CM B cells are class-switched, the CM population was also gated to exclude IgM-cells. Other mature B cell subsets were further subdivided based on expression of IgM, with MN and NSM as largely IgM 1 and AM as largely IgM -. Germinal center B cells were identified as CD24 -, CD38 1 , and were subdivided based on the expression of CXCR4 as centroblasts (CXCR4 1 ) and centrocytes (CXCR4 -). PC were identified on the basis of CD24 -, CD38 hi , and CD27 hi , and were larger (increased FSC-A) than GC B cells. This is consistent with previous reports that GC B cells express low CD27, while PCs express high CD27 and are larger in size (38,39). Within the PC population, a group of IgM 1 PC was also enumerated.

Reference Ranges for Tonsil B Cell Subsets in Young Children
To provide reference ranges for future studies of tonsil B cells in young children, frequencies of B cell subpopulations were determined. The means with the 25th and 75th percentiles are shown in Table 3. Means for all samples tested are shown in Fig. 2 Mature B cells were further broken down on the basis of IgD and CD27 expression as described in Figure 1 and Table 3  Tr n.d. 5 Not defined in this study. a Indicates majority of cells.
Bold symbols represent core markers. PC were also subdivided based on the expression of IgM, which are phenotypically and functionally distinct from IgG PC and may be important for mucosal immunity (41). Immunoglobulin M positive PC were 3.9% of all PC (95% CI 2.6-5.2).

Changes in Tonsil B Cell Populations with Age and Sex
Next we sought to identify changes in B cell subpopulations with age over the first five years of life. Tonsil lymphocyte populations were plotted against age of study participants (Fig. 3). We identified no age-related changes in the proportions of T or B cells of all live mononuclear cells in human tonsils (R 2 5 0.003, P 5 0.78; R 2 5 0.017, P 5 0.47, respectively). Of mature B cell subsets, mature na€ ıve B cells decreased with age (R 2 5 0.159, P 5 0.02) and atypical memory B cells increased with age (R 2 5 0.199, P 5 0.01). No other age-related changes were observed for B cell populations including Transitional, Mature, Germinal Center, or PC (Tr-R 2 <0.001, P 5 0.96; M-R 2 5 0.008, P 5 0.62; GC-R 2 5 0.007, P 5 0.64; PC-R 2 5 0.002, P 5 0.83). Non-class-switched memory B cells and classical memory B cells did not change in frequency with age (R 2 5 0.022, P 5 0.42; R 2 5 0.074, P 5 0.13, respectively). The frequency of centroblasts and centrocytes of all GC B cells did not change significantly with age (R 2 5 0.032, P 5 0.33; R 2 5 0.035, P 5 0.31, respectively).
The frequencies of lymphocyte subsets were also compared by sex (data not shown). Neither T cells nor B cells differed by sex (P 5 0.99, P 5 0.98, respectively). The frequencies of major B cell subsets including transitional, mature, germinal center, and PC did not differ by sex (P 5 0.77, P 5 0.13, P 5 0.11, P 5 0.95, respectively). Of mature B cells, no differences were observed for MN, CM, or AM by sex (P 5 0.31, P 5 0.54, P 5 0.58, respectively). However, NSM B cells were found at higher frequency in boys than girls (t 5 2.08, P 5 0.05). Germinal center B cell subsets CB and CC did not differ in frequency by sex (P 5 0.70, P 5 0.72, respectively). The frequency of IgM 1 PC of all PC also did not differ significantly by sex (P 5 0.45).

EBV Infection and Tonsil B Cell Phenotypes
To determine whether EBV infection has an effect on tonsil B cell phenotypes, we separated the EBV-positive tonsils from EBV-negative tonsils and compared B cell subpopulation frequencies. EBV was detected in 19 (57.6%) of the 33 tonsils studied. There was no difference in frequency EBV positive tonsils by sex (P 5 0.85). Over the age range studied, there was no association between EBV positivity and age (R 2 5 0.004, P 5 0.73). Frequencies of lymphocyte subsets were compared between EBV-and EBV1 tonsils. The mean CD3 1 T cell frequency observed in EBV1 tonsils was higher than in EBV-tonsils (23.0% and 17.0%, respectively, P 5 0.02. However, the proportion of B cells between  We also examined EBV status against sex and EBV load to determine if these measures affect B cell phenotypes and EBV infection. EBV positive samples did not differ significantly by sex ( Supporting Information Fig.  S1A). Of EBV positive tonsils, the mean EBV load per 10 6 TMC was 3.2 3 10 6 copies (SEM 6 1.8 3 10 6 ) and did not differ by sex (P 5 0.41). The frequencies of lymphocyte populations were next plotted against EBV load for EBV-positive samples. There were no significant differences in any lymphocyte subset frequencies by EBV  Fig. S1) Collectively, a high frequency of tonsils from young children were positive for EBV with a variably high EBV copy number, EBV infection was associated with a reduced T cell frequency, but minimal changes in the tonsillar B cell compartment, and EBV viral load was also not associated with significant perturbations in tonsillar B cells.

EBV Receptor Expression on Tonsil B Cell Subsets
To determine whether different tonsil B cell subsets in young children express different amounts of surface EBV receptor CD21, we compared CD21 mean fluorescence intensity (MFI) among CD19 1 B cell subpopulations. A summary of tonsil B cell CD21 expression is shown in Figure 4. The frequency of CD21 positive B cells was determined for the major B cell subsets analyzed. CD21 1 B cells represented at least 97% of all B cell populations studied. As a control, the frequency of CD21 positive T cells was < 3% of all CD3 1 T cells. Although the vast majority of B cells studied expressed surface CD21, we asked whether B cell subsets expressed different amounts of surface CD21, which could explain why in vivo, EBV preferentially infects na€ ıve B cells (16). The MFI of CD21 was quantified for the B cell subpopulations studied. To control for donorto-donor differences in CD21 expression, CD21 MFI of B cell subsets was converted to the fold change (FC) from overall B cell CD21 MFI for that donor, which was expressed as 1. We also determined the difference in mean CD21 surface expression between EBV1 and EBV-samples to confirm that variations in CD21 were not due to EBV infection (Supporting Information Fig. S2). When cell subsets were broken down by CD21 surface expression for EBV1 and EBV-tonsil samples, no major differences were found with populations including T, M, and GC B cells, but PC surface CD21 was significantly reduced in the EBV1 samples (1.08 vs. 0.90 FC). Of GC B cells, there were no differences in CD21 expression between CB and CC by EBV status. Of the PC subset, IgM 1 PCs from EBV-tonsils had significantly higher surface CD21 than from EBV1 tonsils (1.59 vs. 0.94 FC), whereas there was no difference in surface CD21 expression in IgM -PCs. Of mature B cells, no B cell subset had different mean CD21 expression in EBV1 versus EBV-samples.
Finally, we sorted tonsil B cell populations in an EBV positive tonsil sample that had the highest EBV load to determine the site of EBV carriage. We sorted B cells into four subsets (mature na€ ıve, total memory, centroblast and centrocyte). Each subset was confirmed to have a purity of > 96% after sorting. Using quantitative PCR, we found that EBV was primarily detected in the memory B cell subset in these asymptomatic EBV-infected children.

DISCUSSION
This study sought to determine the following characteristics of TMC: reference values for B cell subsets in tonsils of children under 5-years-old, how these subsets change with age, sex, or EBV status, and to determine the expression pattern of the EBV receptor CD21 on B cell subsets. We have provided the first detailed report of B cell subsets in children < 5 years of age, with the aim of benefiting future studies on other human populations. We also observed a high frequency of tonsils from young children were positive for EBV with a variably high EBV copy number, EBV infection was associated with a reduced T cell frequency but minimal changes in the tonsillar B cell compartment, and EBV viral load was also not associated with significant perturbations in tonsillar B cells.
Our results show that mature na€ ıve B cells are the largest component of the tonsil B cell compartment in children under 5 years old, making up about 40% of all tonsil B cells in healthy children. Interestingly, we observed a proportion of B cells that were IgD1, IgM-, which has previously been described in germinal center and peripheral blood B cells (42). We hypothesize that these cells may represent a pre-GC phenotype of na€ ıve B cells or an autoreactive population, consistent with previous reports of these populations (42,43). We also observed that the proportion of mature tonsil B cell subsets change with age over the first 5 years of life. Mature na€ ıve B cells decreased in frequency with age, while atypical and classical memory B cells increased, consistent with age-dependent antigen exposure. These results add to a previous study of older children (> 4 years of age) and adults showing increased germinal center B cells, and reduced memory B cells in children compared to adolescents and adults (25). We did not observe major changes in B cell subsets by sex, and though NSM B were found more frequently in boys than in girls, this may be incidental, and the physiological importance of this finding is not known.
We observed a relatively high frequency of EBV infection during the first five years of life (60%). The EBV viral load was much higher per cell (3.2 3 10 6 copies/ 10 6 TMC) than has been reported in peripheral blood mononuclear cells (44,45). This observation is not surprising considering that tonsils are the site of EBV entry and persistence in the lymphocyte compartment (46). The variable but high viral load we observed indicates that tonsils are a reservoir of viral persistence. Accordingly, in sorted B cell subsets, EBV was overwhelmingly found in the memory B cell compartment, consistent with the model that EBV infects na€ ıve B cells during primary infection and is carried in memory B cells during latency through the life of the host (47,48). EBV infection was associated with an increase in the proportion of CD3 1 T cells of TMC. This is consistent with the role of cytotoxic T cells in controlling primary EBV infection. The length of time between EBV infection and tonsillectomy was not possible to determine for this study, but the fact that children enrolled in this study were not acutely ill suggests that EBV infection may be associated with subacute to chronic alterations in tonsil T cell frequencies. Importantly, EBV infection was not associated with major changes in the B cell compartment, suggesting both that tonsils were not sampled during primary infection when B cell proliferation is high, and that EBV infection does not alter the B cell subset distribution during persistent infection.
We found large differences in the surface expression of CD21 on tonsil B cell subsets, with mature na€ ıve B cells expressing significantly higher CD21 than classical and atypical memory B cell subsets in children under 5years-old. These results support the well-documented observation that na€ ıve B cells are more susceptible to EBV-induced transformation (19) and are more susceptible to primary EBV infection in vitro and in vivo (16,19). This contrasts with limited data from patients of undocumented ages showing that the expression pattern of CD21 was not different in tonsil B cell subpopulations (18,19). While we did not observe changes in CD21 surface expression in B cell subsets with age over the first five years of life (data not shown), it is possible that as children age, CD21 expression normalizes across different B cell subsets. It is also possible that previous studies that did not show differences in CD21 expression on B cell subsets (18,19) were underpowered to detect differences. Our data provide an important measure of susceptibility of B cell subsets to EBV infection during the first years of life. Studies on children of this age range are important for understanding primary EBV infection dynamics in general, as well as how EBV infection is modulated in the setting of sub-Saharan Africa, leading to increased risk of eBL. Furthermore, studies involving primary tonsil tissue provide much more physiologically relevant information for EBV infection than those involving peripheral blood.
Overall, our results provide reference ranges of B cell subsets that can be used in future studies of tonsil tissue of different patient groups or other experimental conditions. We also show that mature na€ ıve B cells decrease in frequency in human tonsils with age, and for the first time, that these cells express increased surface expression of the EBV receptor CD21, suggesting a mechanism for the observation that these cells are more susceptible to EBV infection in vivo (16). Our results also support the hypothesis that EBV infection early in life alters viral pathogenesis by increasing the pool of cells that can be infected by the virus. This enhanced infection, alongside other factors such as malaria that are known to profoundly affect EBV infection and the immune system as a whole, could help explain the extremely high burden of Burkitt lymphoma in the sub-Saharan African population.

ACKNOWLEDGMENTS
The authors first thank the patients who contributed their tissue to this study. They also thank Kelly Winchell for obtaining and organizing tonsil consent forms. Thanks to Ignacio Sanz for his help in optimizing our B cell phenotyping panel. The authors all declare that they have no financial or other conflicts of interest.