Germinal centers (GCs) are complex, multicell-type, transient structures that form in secondary lymphatic tissues in response to T cell-dependent stimulation. This process is crucial to the adaptive immune response because it is the source of affinity maturation and long-lived B cell memory. Our previous studies showed that the growth of murine splenic GCs is nonsynchronized, involving broad-volume distributions of individual GCs at any time. This raises the question whether such a thing as a typical GC exists. To address this matter, we acquired large-scale confocal data on GCs throughout the course of the 2-phenyl-5-oxazolone chicken serum albumin-driven primary immune response in BALB/c mice. Semiautomated image analysis of 3457 GC sections revealed that, although there is no typical GC in terms of size, GCs have a typical cellular composition in that the cell ratios of resident T cells, macrophages, proliferating cells, and apoptotic nuclei are maintained during the established phase of the response. Moreover, our data provide evidence that the dark zone (DZ) and light zone (LZ) compartments of GCs are about the same size and led us to estimate that the minimal cell loss rate in GCs is 3% per hour. Furthermore, we found that the population of GC macrophages is larger and more heterogeneous than previously thought, and that despite enrichment of T cells in the LZ, the DZ of murine splenic GCs is not poor in T cells. DZ and LZ differ in the T cell-to-macrophage ratio rather than in the density of T cells.

Antibody affinity maturation is a hallmark of the adaptive immune response that requires formation of germinal centers (GCs) in secondary lymphoid organs such as spleen and lymph nodes. GCs are crucial sites because they fine-tune the B cell response with regard to amplitude and specificity, and thereby lead to the generation of (long-lived) high-affinity plasma cells and memory B cells (1, 2). Affinity maturation is accomplished through a microevolutionary process during which GC B cells are subject to random somatic hypermutation (SHM) of their BCR genes and subsequent affinity-based selection (1).

GCs are complex, multicell-type, transient structures that form in response to antigenic stimulation. The primary GC response in the spleen has been previously shown to exhibit clearly defined kinetics with inductive, established, and dissociative phases (35). Although B cells make up the majority of their cell population, GCs also encompass T cells, macrophages, and stromal follicular dendritic cells (FDCs) (1). Other cell types, such as dendritic cells (6) and accessory CD4+ CD3 cells (7), might be present at times. The fine processes of FDCs form a network that histologically divides the GC into two compartments referred to as the dark zone (DZ) and the light zone (LZ). As distinguished from the DZ, the LZ shows a compact FDC network wherein FDCs serve as Ag deposits that build upon trapping of immune complexes through complement and FcRs (810). Because competition for Ag binding has been proposed to drive B cell selection (11), the concentration of Ag on the surface of FDCs in the LZ is commonly taken as evidence for spatial confinement of selection to this zone. The capability of FDCs to promote B cell survival through switching off the apoptotic machinery in adhering cells (12) has further led to the concept of selection as a result of BCR cross-linking by FDC-trapped Ag (13, 14). Alternatively, other models presume a role of follicular Th cells in selection (2, 15, 16). The general idea is that BCR-bound Ag is first internalized and then presented as peptide-MHC for follicular T cell recognition. The possible outcomes of such an interaction with follicular Th cells, a subset believed to accumulate in the LZ of GCs, are survival and differentiation of GC B cells.

SHM of BCRs generates both B cell variants with advantageous and deleterious mutations. Therefore, deleterious mutations need to be eliminated from the GC B cell population to achieve enrichment of high-affinity B cells. As a corollary of this condition, GCs appear as sites of extensive proliferation on the one hand and massive programmed cell death on the other hand (17). Classically, proliferation is thought to occur in the DZ of GCs (1), a notion strongly supported by a recent multiphoton microscopy study by Victora et al. (18). Cell death by apoptosis is described to be most extensive among LZ B cells in rodents; however, in chronically inflamed tonsillar GCs, apoptosis was reported to be most evident at the interface between DZ and LZ, a region also referred to as basal LZ (17). Apoptotic GC B cells are strong candidates as sources of autoantigens, and their impaired clearance by macrophages has been linked to the pathogenesis of autoimmune diseases (19). GC-resident macrophages constitute a unique subset known as tingible body macrophages (TBM) (20), which, besides fulfilling an important scavenger function, is believed to play a role in regulating the magnitude of the GC reaction (21). Although TBM express MHC class II molecules, they are believed neither to function as APCs in GCs nor to be required for GC formation (21).

Because of their crucial role in establishing an effective humoral immune response, GC formation and function continue to be important subjects of research for immunologists. Assessing the role of a designated protein in affinity maturation, for instance, often includes comparison of GC growth in transgenic mice or mice deficient in said protein and wild-type animals; a few past examples of such proteins include CD19 (22), BAFF (23), and Bcl-XL (24). The results reported in this article, together with our previous findings (25), however, alert us to caution when such a comparison is made by cross-sectional profiling of GCs. At any time point, the cross-sectional profile of an ensemble of GCs shows a broad size distribution of individual GCs (25). Although this is certainly due in large part to a sectioning bias (i.e., the section plane does not necessarily pass through the center of GCs), we have previously demonstrated that such a profile also reflects a real-size distribution of GCs in three dimensions (25). Simply put, the variety of cross-sectional GCs in a tissue section is not the mere result of cutting these GCs differently but also of these GCs being different. Accordingly, the sampling of GCs for scoring is a main source of bias in the data. Both the number of scored GCs and the criteria according to which they are selected strongly influence the results and, in the worst case, might even distort statistical inferences. Therefore, cross-sectional scoring should ideally include all GCs, that is, the whole ensemble and not a subjectively chosen subset thereof.

Because there are no such things as typical size (25) and typical clonal diversity (26) of a GC at a given time point, the question arises whether typicality exists at all for GCs. To address this matter, this study acquired large-scale confocal data on cellular composition of GCs during the primary immune response to 2-phenyl-5-oxazolone (phOx)-chicken serum albumin (CSA) in BALB/c mice. We evaluated the recorded images of GCs and their DZ and LZ compartments quantitatively according to cross-sectional size and counts of proliferating cells, T cells, macrophages, and apoptotic nuclei. In particular, we show that during the established phase of the phOx-CSA–driven GC response, GCs maintain a typical cellular composition. This composition is independent of GC size and the time elapsed since immunization.

Six- to 8-wk-old BALB/c mice were immunized with a single i.p. injection of 100 μg phOx coupled to CSA at a ratio of 10:1 and precipitated onto alum as described previously (27). BALB/c mice were purchased from the Bundesinstitut für Risikobewertung, Berlin, Germany and were housed under specific pathogen-free conditions at the animal facility of the Deutsches Rheuma-Forschungszentrum Berlin (DRFZ), Berlin, Germany. All animal experiments were performed in accordance with institutional, state, and federal guidelines.

The following Abs and other reagents were used to visualize splenic architecture and to detect GC cell populations: unconjugated rat IgG2a to mouse Ki-67 (clone Tec-3; Dako, Glostrup, Denmark); biotin-labeled Ab to mouse FDC (clone FDC-M2; ImmunoK, Abingdon, U.K.); Alexa 488-labeled anti-mouse CD3 (clone KT3; AbD Serotec, Düsseldorf, Germany); Alexa 488-labeled anti-mouse CD68 (clone FA-11; AbD Serotec); Alexa 647-labeled anti-rat IgG (Invitrogen, Karlsruhe, Germany); biotin-labeled anti-mouse B220 (clone RA3.6B2, in-house conjugate, DRFZ); Cy5-labeled anti-mouse CD4 (clone GK1.5, in-house conjugate, DRFZ); streptavidin-Alexa 546 and streptavidin-Alexa 647 (Invitrogen); and rhodamine-labeled peanut agglutinin (PNA; Vector Laboratories, Burlingame, CA).

For cross-sectional evaluation of GC cell population kinetics, cohorts of two to four immunized mice were killed on days 4, 6, 8, 10, 12, 14, 16, 18, and 21 postimmunization. Spleens were removed, bisected, and snap frozen, and longitudinal cryostat spleen sections of 10 μm thickness were prepared as described previously (25). Cryostat sections of NZB and NZB/W spleens were kindly provided by Prof. Rudolf Manz (Institute for Systemic Inflammation Research, University of Lübeck, Lübeck, Germany). Spleen sections were treated before staining by fixation in ice-cold 1% PFA for 30 min and permeabilization in ice-cold 1% sodium citrate containing 1% Triton X-100 (Promega, Mannheim, Germany) for 2 min. After blocking in PBS containing 3% BSA for 30 min, sections were triple stained with anti–Ki-67, anti–FDC-M2, and either anti-CD3 (staining 1) or anti-CD68 (staining 2). Bound Ki-67 and biotinylated FDC-M2 Abs were detected using Alexa 647 anti-rat IgG and streptavidin-Alexa 546, respectively. In some experiments, sections were subjected to TUNEL assays (DeadEnd Colorimetric TUNEL System; Promega). In these cases, biotinylated nucleotides were detected using streptavidin-Alexa 647, and sections were colabeled with PNA and anti-CD68 (staining 3). An overview of stainings 1, 2, and 3 is given in Fig. 1.

FIGURE 1.

Overview of histological image data sets and their quantitative evaluation. This study is based on a database of 3457 histological images of murine splenic GCs recorded during the primary response of BALB/c mice to phOx-CSA. The database includes three GC image series derived from different triple-immunofluorescence stainings: Ki-67, FDC-M2, and CD3 (staining 1, number of images [Σ] = 1130); Ki-67, FDC-M2, and CD68 (staining 2, Σ = 1,289); PNA, TUNEL, and CD68 (staining 3, Σ = 1038). For each image, ROI boundaries outlining the GC and its LZ (if present) were drawn and saved; DZ ROIs were obtained by subtraction. ROI areas were measured, and cell numbers within ROIs were either counted manually (CD3+, CD3+Ki-67+, CD3+Ki-67, CD68+) or automatically using the Nucleus Counter plugin of the ImageJ image analysis software (Ki-67+, TUNEL+). Symbols in parentheses are the same as in Table II. Scale bar, 100 μm.

FIGURE 1.

Overview of histological image data sets and their quantitative evaluation. This study is based on a database of 3457 histological images of murine splenic GCs recorded during the primary response of BALB/c mice to phOx-CSA. The database includes three GC image series derived from different triple-immunofluorescence stainings: Ki-67, FDC-M2, and CD3 (staining 1, number of images [Σ] = 1130); Ki-67, FDC-M2, and CD68 (staining 2, Σ = 1,289); PNA, TUNEL, and CD68 (staining 3, Σ = 1038). For each image, ROI boundaries outlining the GC and its LZ (if present) were drawn and saved; DZ ROIs were obtained by subtraction. ROI areas were measured, and cell numbers within ROIs were either counted manually (CD3+, CD3+Ki-67+, CD3+Ki-67, CD68+) or automatically using the Nucleus Counter plugin of the ImageJ image analysis software (Ki-67+, TUNEL+). Symbols in parentheses are the same as in Table II. Scale bar, 100 μm.

Close modal

Slides were viewed under a Leica DM Ire2 confocal laser-scanning microscope, and digital images of GCs were acquired using a ×40 objective and Leica LCS software (Leica, Wetzlar, Germany). In stainings 1 and 2, GCs were identified as dense clusters of Ki-67+ proliferating cells in the right anatomical context (white pulp, proximity to the periarterial lymphatic sheath, immediate vicinity to a FDC network). GC LZ and DZ were distinguished by FDC polarity (2830). The outer boundary of a GC was defined by the stretch of the Ki-67+ cell cluster; its inner DZ-LZ border was delineated based on the absence (DZ) and presence of a dense reticular network of FDC processes (LZ; Fig. 1). In case of staining 3, GCs were identified by PNA reactivity, and outer GC boundaries were defined by the stretch of the PNA staining (Fig. 1). Outer GC and, where relevant, LZ boundaries were assigned manually to each GC, and the areas included were saved as regions of interest (ROIs); GC DZ ROIs were obtained by subtraction of GC LZ ROIs from GC ROIs. ROI areas were measured using ImageJ image analysis software (31). The numbers of Ki-67+ cells and TUNEL+ nuclei within ROIs were determined automatically, applying an adapted version of the Nucleus Counter plugin for the ImageJ image analysis software; CD68+ macrophages and CD3+ T cells were counted manually using an adapted version of the Cell Counter plugin. For each mouse and each staining, two independent spleen sections (S1 and S2, distance ≥ 400 μm) were analyzed.

To make our data accessible to the public, we set up an online database including all recorded confocal images and their associated data sheets. For reasons of portability and fast access, we decided to use plain HTML pages to represent this information. The scripting language Perl (version 5.10.0) was used to generate HTML pages that compile images and tables from ∼30,000 source files. Access to the database is organized in a hierarchical fashion by grouping data according to the experimental setup, which makes each of the roughly 10,000 Web pages accessible with a few mouse clicks. The system was developed in collaboration with MicroDiscovery GmbH and can be accessed at: http://sysimmtools.eu.

Quantitative data were tested for normality using the D’Agostino–Pearson omnibus test for normality. Association between variables in the data sets was assessed by Spearman correlation (ρ) and linear regression statistics. Significant differences among and within groups of mice or days were determined by Kruskal–Wallis ANOVA with Dunn’s posttest for n > 2 and the two-tailed Mann–Whitney U test for n = 2. One-way ANOVA with Tukey’s post test was used to test for differences among groups of means. DZ and LZ data were compared using the two-tailed Mann–Whitney U test. Significance levels were set at *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, and ****p ≤ 0.0001. Slope estimates and R2 values of linear regression curves (y = my,x × x) fitted to each individual day’s GC composition parameters and theoretical slope estimates (m˜) of GC composition parameters that were not recorded from the same staining are summarized in Supplemental Table I. All statistical analyses were performed with GraphPad Prism 5.0 software (GraphPad Software, La Jolla, CA).

This study is based on a database of 3457 histological images of murine splenic GCs recorded during the primary response of BALB/c mice to phOx-CSA. The database includes three datasets of GC image series derived from different triple-immunofluorescence stainings with: 1) Ki-67, FDC-M2, and CD3 (Σ = 1130); 2) Ki-67, FDC-M2, and CD68 (Σ = 1289); and 3) PNA, TUNEL, and CD68 (Σ = 1038). Summaries of composition and evaluation of the database are given in Table I and Fig. 1; the recorded GC composition parameters are summarized in Table II. An online version of the database, GCImagePresenter, including all images and their associated data evaluation sheets is available at: http://sysimmtools.eu.

Table I.
Overview of the histological image database
Staining, Time (d)M1, S1M1, S2M2, S1M2, S2M3, S1M3, S2M4, S1M4, S2Σ
1, 4  10 17   11   42 
1, 6  11 16 19 15 16 35 15 11 138 
1, 8  18 30   35 45 27 37 192 
1, 10 27  10 19 31 41   128 
1, 12 22 16   37 52   127 
1, 14 27 25 44 12     108 
1, 16 34 26   50 38 26 25 199 
1, 18     21 20 30 39 110 
1, 21 17 22   19 28   86 
2, 4 15 18     44 
2, 6 11 18 13 14 23 15 12 114 
2, 8 25 30   31 51 31 36 204 
2, 10 23 29 16 32 45   153 
2, 12 18 19 13 32 41 55   178 
2, 14 23 21 41 29 21 12   147 
2, 16 26 27   46 38 28 28 193 
2, 18   14 47 23 23 25 46 178 
2, 21 15 20   16 27   78 
3, 4 13 12     37 
3, 6 11 17   11 17 16 80 
3, 8 24 36   34 48 30 42 214 
3, 10 28 31   31 37   127 
3, 12 26 22   20 38 57 170 
3, 14 21 18       39 
3, 16 24 29   50 22 18 29 172 
3, 18     20 23 34 36 113 
3, 21 20 17   21 28   86 
Staining, Time (d)M1, S1M1, S2M2, S1M2, S2M3, S1M3, S2M4, S1M4, S2Σ
1, 4  10 17   11   42 
1, 6  11 16 19 15 16 35 15 11 138 
1, 8  18 30   35 45 27 37 192 
1, 10 27  10 19 31 41   128 
1, 12 22 16   37 52   127 
1, 14 27 25 44 12     108 
1, 16 34 26   50 38 26 25 199 
1, 18     21 20 30 39 110 
1, 21 17 22   19 28   86 
2, 4 15 18     44 
2, 6 11 18 13 14 23 15 12 114 
2, 8 25 30   31 51 31 36 204 
2, 10 23 29 16 32 45   153 
2, 12 18 19 13 32 41 55   178 
2, 14 23 21 41 29 21 12   147 
2, 16 26 27   46 38 28 28 193 
2, 18   14 47 23 23 25 46 178 
2, 21 15 20   16 27   78 
3, 4 13 12     37 
3, 6 11 17   11 17 16 80 
3, 8 24 36   34 48 30 42 214 
3, 10 28 31   31 37   127 
3, 12 26 22   20 38 57 170 
3, 14 21 18       39 
3, 16 24 29   50 22 18 29 172 
3, 18     20 23 34 36 113 
3, 21 20 17   21 28   86 

Indicated are the numbers of GCs evaluated per staining and time point. Two to four mice (M1–M4) were analyzed per time point, and two independent spleen sections (S1 and S2, distance > 400 μm) were evaluated per individual. Staining 1: Ki-67, FDC-M2, CD3; staining 2: Ki-67, FDC-M2, CD68; and staining 3: PNA, TUNEL, CD68.

Table II.
Summary of recorded GC composition parameters
ParameterSymbolDadaStaining
Area GC, LZ DZ = GC − LZ 1, 2, 3b 
Ki-67+ cells Ki GC, LZ DZ = GC − LZ 1, 2 
CD3+ T cells DZ, LZ GC = DZ + LZ 
CD3+Ki-67+ T cells TKi DZ, LZ GC = DZ + LZ 
CD68+ macrophages DZ, LZ GC = DZ + LZ 
TUNEL+ nuclei GC  
ParameterSymbolDadaStaining
Area GC, LZ DZ = GC − LZ 1, 2, 3b 
Ki-67+ cells Ki GC, LZ DZ = GC − LZ 1, 2 
CD3+ T cells DZ, LZ GC = DZ + LZ 
CD3+Ki-67+ T cells TKi DZ, LZ GC = DZ + LZ 
CD68+ macrophages DZ, LZ GC = DZ + LZ 
TUNEL+ nuclei GC  
a

Listed are the directly measured (D) and derived (d) parameters (obtained by either subtraction or addition). Staining 1: Ki-67, FDC-M2, CD3; staining 2: Ki-67, FDC-M2, CD68; and staining 3: PNA, TUNEL, CD68.

b

In the case of staining 3, only the total GC area was recorded.

Regarding individual mice, all recorded parameters showed broad, nonnormal size and count distributions; however, distributions were found to be robust for mice analyzed at the same time point, with differences usually being statistically insignificant (Supplemental Fig. 1A). For the purposes of graphical presentation and statistical analysis, data were therefore aggregated by day postimmunization. The raw data for all recorded parameters are displayed in Supplemental Fig. 1B. Slope estimates and R2 values of linear regression curves fitted to our data are summarized in Supplemental Table I.

We investigated the dynamics of GC zoning after phOx-CSA immunization in BALB/c mice by immunohistology of spleens on days 4, 6, 8, 10, 12, 14, 16, 18, and 21 postimmunization. In the absence of immunization, the background level of environmental Ag-induced GCs was minimal, and the sizes of B cell zones decreased significantly compared with immunized mice (Supplemental Fig. 2D, 2E), which indicates that the examined GCs were induced by immunization. PhOx-CSA–induced GCs first became detectable on day 4, with the majority (54%) exhibiting only an LZ but no DZ (Fig. 2A). Thereafter, the proportion of GCs having both DZ and LZ increased, reaching a plateau on day 8 (79%) that was maintained until day 21. Thus, starting at day 8, the established phase of the response extends to at least day 21. As we have reported previously for whole GCs (25), DZs and LZs also showed broad area distributions that underwent dynamic changes over time (Fig. 2B). Notably, although the variability in the relative sizes of DZ and LZ among GCs turned out to be considerable at all sampled time points (coefficient of variation [CV] > 67%), the cumulative frequency curves of relative size are statistically indistinguishable between days 6 and 21 (p ≥ 0.05, Kruskal–Wallis ANOVA; Fig. 2B). Likewise, LZ and DZ share similar average growth kinetics with peaks on days 8 (10,983 ± 2718 μm2) and 10 (11,290 ± 2325 μm2), respectively (Fig. 2C).

FIGURE 2.

Dynamics of GC zoning. Mice were immunized with phOx-CSA and sacrificed 4–21 d later. Spleen sections were prepared, stained, and evaluated for sizes of GCs as described in 1Materials and Methods. Two sections at least 400 μm apart were scored for each spleen, and two to four mice were evaluated per time point (Table I). A, Proportions of GCs showing only a DZ, only an LZ, or both DZ and LZ at different points in time after immunization. GCs were identified as Ki-67+ cell clusters and by anatomical location in triple-immunofluorescence stainings of proliferating cells (Ki-67, blue), T cells (CD3, green), and FDC networks (FDC-M2, red). GC LZ and DZ were distinguished by the presence of FDCs in the LZ. Photomicrographs are representative of day 18 GCs. Scale bar, 100 μm. B, Size distributions and size ratio of LZ and DZ plotted as cumulative frequency curves. Dotted vertical lines indicate medians. C, Average size kinetics of GCs (filled area), DZ (filled squares), and LZ (open squares), expressed as mean and SD of two to four mice. D, The numbers of Ki-67+ proliferating cells in DZ and LZ are positively correlated to the size of DZ and LZ, respectively (Spearman ρ > 0.95; p < 0.0001). Lines represent linear fits of compiled data for each day (see Supplemental Table I for individual slopes m and R2 values); mean slopes (days 8–21) are indicated by m¯. E, Density of Ki-67+ proliferating cells in DZ (squares) and LZ (circles). Each symbol represents a single DZ or LZ, and the data are compilations of all calculated values for a given day; bars show mean and SD. Statistically significant differences between DZ and LZ are indicated: *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001, Mann–Whitney U test. F, Average kinetics of the densities of Ki-67+ proliferating cells in GCs (filled area), DZ (filled squares), and LZ (open squares). Data are expressed as mean and SD of two to four mice.

FIGURE 2.

Dynamics of GC zoning. Mice were immunized with phOx-CSA and sacrificed 4–21 d later. Spleen sections were prepared, stained, and evaluated for sizes of GCs as described in 1Materials and Methods. Two sections at least 400 μm apart were scored for each spleen, and two to four mice were evaluated per time point (Table I). A, Proportions of GCs showing only a DZ, only an LZ, or both DZ and LZ at different points in time after immunization. GCs were identified as Ki-67+ cell clusters and by anatomical location in triple-immunofluorescence stainings of proliferating cells (Ki-67, blue), T cells (CD3, green), and FDC networks (FDC-M2, red). GC LZ and DZ were distinguished by the presence of FDCs in the LZ. Photomicrographs are representative of day 18 GCs. Scale bar, 100 μm. B, Size distributions and size ratio of LZ and DZ plotted as cumulative frequency curves. Dotted vertical lines indicate medians. C, Average size kinetics of GCs (filled area), DZ (filled squares), and LZ (open squares), expressed as mean and SD of two to four mice. D, The numbers of Ki-67+ proliferating cells in DZ and LZ are positively correlated to the size of DZ and LZ, respectively (Spearman ρ > 0.95; p < 0.0001). Lines represent linear fits of compiled data for each day (see Supplemental Table I for individual slopes m and R2 values); mean slopes (days 8–21) are indicated by m¯. E, Density of Ki-67+ proliferating cells in DZ (squares) and LZ (circles). Each symbol represents a single DZ or LZ, and the data are compilations of all calculated values for a given day; bars show mean and SD. Statistically significant differences between DZ and LZ are indicated: *p ≤ 0.05, **p ≤ 0.01, ****p ≤ 0.0001, Mann–Whitney U test. F, Average kinetics of the densities of Ki-67+ proliferating cells in GCs (filled area), DZ (filled squares), and LZ (open squares). Data are expressed as mean and SD of two to four mice.

Close modal

Throughout the response, the numbers of Ki-67+ proliferating cells in DZs and LZs were positively correlated to the size of the respective compartment (Spearman ρ > 0.95; p < 0.0001; Fig. 2D). Interestingly, the slope estimates of linear regression curves fitted to each day have a CV <5% (Table III, Supplemental Table I), indicating that the association between size and number of Ki-67+ proliferating cells is time invariant. Accordingly, the average densities of Ki-67+ cells in the LZ and the DZ remained constant over time (Fig. 2F). However, the density of Ki-67+ cells was significantly higher in the DZ than in the LZ at all sampled time points (p ≤ 0.05; Mann–Whitney U test; Fig. 2E).

Table III.
Mean slope estimates for GC composition parameters
GC
DZ
LZ
EstimateMean ± SDCV (%)Mean ± SDCV (%)Mean ± SDCV (%)
m¯Ki,a  0.01182 ± 0.00047 4.0  0.01293 ± 0.00057 4.4  0.01085 ± 0.00054 5.0 
m¯T,a  0.00151 ± 0.00023 15.2  0.00127 ± 0.00022 17.6  0.00169 ± 0.00028 16.3 
m¯M,a  0.00101 ± 0.00013 13.0  0.00112 ± 0.00012 10.9  0.00085 ± 0.00014 16.8 
m¯A,a  0.00231 ± 0.00027 11.6       
m¯T,Ki  0.12446 ± 0.01527 12.3  0.09644 ± 0.01419 14.7  0.15131 ± 0.02055 13.6 
m¯M,Ki  0.08245 ± 0.01231 14.9  0.08904 ± 0.00991 11.1  0.07235 ± 0.01559 21.6 
m˜¯M,T  0.67083 ± 0.07219 10.8  0.89747 ± 0.14098 15.7  0.50851 ± 0.04627 9.1 
m˜¯A,M  2.32895 ± 0.37440 16.1       
m˜¯A,T  1.55889 ± 0.28183 18.1       
m˜¯A,Ki  0.19578 ± 0.02094 10.7       
GC
DZ
LZ
EstimateMean ± SDCV (%)Mean ± SDCV (%)Mean ± SDCV (%)
m¯Ki,a  0.01182 ± 0.00047 4.0  0.01293 ± 0.00057 4.4  0.01085 ± 0.00054 5.0 
m¯T,a  0.00151 ± 0.00023 15.2  0.00127 ± 0.00022 17.6  0.00169 ± 0.00028 16.3 
m¯M,a  0.00101 ± 0.00013 13.0  0.00112 ± 0.00012 10.9  0.00085 ± 0.00014 16.8 
m¯A,a  0.00231 ± 0.00027 11.6       
m¯T,Ki  0.12446 ± 0.01527 12.3  0.09644 ± 0.01419 14.7  0.15131 ± 0.02055 13.6 
m¯M,Ki  0.08245 ± 0.01231 14.9  0.08904 ± 0.00991 11.1  0.07235 ± 0.01559 21.6 
m˜¯M,T  0.67083 ± 0.07219 10.8  0.89747 ± 0.14098 15.7  0.50851 ± 0.04627 9.1 
m˜¯A,M  2.32895 ± 0.37440 16.1       
m˜¯A,T  1.55889 ± 0.28183 18.1       
m˜¯A,Ki  0.19578 ± 0.02094 10.7       

Mean slope estimates (m¯) of linear regression curves (y = my,x × x) fitted to each individual day’s GC composition parameters. Further listed are the theoretical mean slope estimates (m˜¯) of GC composition parameters that were not recorded from the same staining. Indicated means were calculated taking into account data from days 8 to 21 postimmunization. Symbols are the same as in Table II.

The dynamics of T cell accumulation within splenic GCs were assessed in BALB/c mice after immunization with phOx-CSA (Table I). As exemplified in Fig. 1, GC T cells were identified and enumerated as CD3+ cells within manually assigned ROI boundaries including total GC area, DZ area, and LZ area. At all sampled time points, there were marked differences in the numbers of CD3+ T cells among GCs and their DZ and LZ compartments (CV > 71%; Fig. 3B). Such broad distributions of CD3+ T cell numbers were subject to change over time; however, between days 10 and 21, the cumulative frequency curves of the numbers of CD3+ GC T cells were statistically indistinguishable (p > 0.05 for GC, DZ, and LZ; Kruskal–Wallis ANOVA). CD3+ T cells, although low in number, were already detectable in nascent GCs (Fig. 3A, day 4), and their average numbers increased progressively, reaching 27 ± 1 T cells per GC on day 8 (average calculated across mice). The numbers of CD3+ T cells in DZ and LZ followed this kinetics, with peak numbers of 13 ± 2 and 16 ± 2 CD3+ T cells (Fig. 3C). Correlation analysis further revealed a positive association between the number of CD3+ T cells and GC or compartment size (Spearman ρGC > 0.84, ρDZ > 0.71, and ρLZ > 0.77; p < 0.0001; Fig. 3D). Likewise, there was a positive correlation between the numbers of CD3+ T cells and Ki-67+ proliferating cells (Spearman ρGC > 0.82, ρDZ > 0.64, and ρLZ > 0.79; p < 0.0001). The slope estimates of linear regressions between T cell numbers and GC compartment size were very similar for all time points (CV < 19%; Fig. 3D, Table III, Supplemental Table I), but the average densities of CD3+ T cells in the DZ and the LZ showed a slight increase over time (Fig. 3F). However, when the groups of means were compared statistically, the changes were insignificant (p > 0.05, one-way ANOVA). Most importantly, from day 10 onward (1151 ± 222 CD3+ cells/mm2 DZ and 1438 ± 75 CD3+ cells/mm2 LZ), the density of CD3+ T cells was significantly higher in the LZ than in the DZ (p ≤ 0.05, Mann–Whitney U tests; Fig. 3E).

FIGURE 3.

Dynamics of CD3+ GC T cells. Mice were immunized with phOx-CSA and sacrificed 4–21 d later. Spleen sections were prepared, stained, and scored as described in the legend for Fig. 1 and in 1Materials and Methods. A, Photomicrographs representative of GCs recorded on days 4, 6, 10, 14, and 21 after immunization. GCs were identified as Ki-67+ cell clusters and by anatomical location in triple-immunofluorescence stainings of proliferating cells (Ki-67, blue), T cells (CD3, green), and FDC networks (FDC-M2, red). LZ and DZ were distinguished by the presence of FDCs in the LZ. Scale bar, 100 μm. B, Time-dependent changes in numbers of CD3+ T cells in GCs, DZ, and LZ plotted as cumulative frequency curves. C, Average kinetics of CD3+ T cells present in GCs (filled area), DZ (filled squares), and LZ (open squares) expressed as mean and SD of two to four mice. D, The numbers of CD3+ T cells in GCs, DZ, and LZ are positively correlated to the size of the respective compartment (Spearman ρ > 0.75; p < 0.0001). Lines represent linear fits of compiled data for each day (see Supplemental Table I for individual slopes m and R2 values); mean slopes (days 8–21) are indicated by m¯. E, Density of CD3+ T cells in DZ (squares) and LZ (circles). Each symbol represents a single DZ or LZ, and the data are compilations of all calculated values for a given day; bars show mean and SD. Statistically significant differences between DZ and LZ are indicated: **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, Mann–Whitney U test. F, Average kinetics of the densities of CD3+ T cells in GCs (filled area), DZ (filled squares), and LZ (open squares). Data are expressed as mean and SD of two to four mice.

FIGURE 3.

Dynamics of CD3+ GC T cells. Mice were immunized with phOx-CSA and sacrificed 4–21 d later. Spleen sections were prepared, stained, and scored as described in the legend for Fig. 1 and in 1Materials and Methods. A, Photomicrographs representative of GCs recorded on days 4, 6, 10, 14, and 21 after immunization. GCs were identified as Ki-67+ cell clusters and by anatomical location in triple-immunofluorescence stainings of proliferating cells (Ki-67, blue), T cells (CD3, green), and FDC networks (FDC-M2, red). LZ and DZ were distinguished by the presence of FDCs in the LZ. Scale bar, 100 μm. B, Time-dependent changes in numbers of CD3+ T cells in GCs, DZ, and LZ plotted as cumulative frequency curves. C, Average kinetics of CD3+ T cells present in GCs (filled area), DZ (filled squares), and LZ (open squares) expressed as mean and SD of two to four mice. D, The numbers of CD3+ T cells in GCs, DZ, and LZ are positively correlated to the size of the respective compartment (Spearman ρ > 0.75; p < 0.0001). Lines represent linear fits of compiled data for each day (see Supplemental Table I for individual slopes m and R2 values); mean slopes (days 8–21) are indicated by m¯. E, Density of CD3+ T cells in DZ (squares) and LZ (circles). Each symbol represents a single DZ or LZ, and the data are compilations of all calculated values for a given day; bars show mean and SD. Statistically significant differences between DZ and LZ are indicated: **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, Mann–Whitney U test. F, Average kinetics of the densities of CD3+ T cells in GCs (filled area), DZ (filled squares), and LZ (open squares). Data are expressed as mean and SD of two to four mice.

Close modal

We also investigated GC T cell proliferation by counting Ki-67+ cells among CD3+ T cells. The number of proliferating GC T cells per GC was generally low, ranging from 0 to 15, and 79% of all GCs showed ≤3 proliferating GC T cells (Supplemental Fig. 3A). At the peak on day 8, GCs contained 3.2 ± 0.9 proliferating T cells, on average, but the highest density of proliferating T cells was already observed on day 6 (248 ± 50 CD3+Ki-67+ cells/mm2; Supplemental Fig. 3A). Also, the frequency of proliferating per total GC T cells peaked on day 6 (17 ± 2%; Supplemental Fig. 3A).

Accumulation of macrophages within de novo–induced GCs was assessed in BALB/c mice 4–21 d after primary immunization with phOx-CSA (Table I). Distributions of GC macrophages, identified as CD68+ cells within ROI boundaries (Fig. 1), were broad in terms of the numbers of CD68+ cells among GCs, LZs, and DZs, at all sampled time points (CV > 62%; Fig. 4B). These distributions were each found to be statistically indistinguishable between days 8 and 12 and days 16 and 21 (p > 0.05 for GC, DZ, and LZ; Kruskal–Wallis ANOVA). Macrophages were already detected in considerable numbers in nascent GCs on day 4 (Fig. 4A, 4B), and their average numbers more than doubled until day 10, reaching 18 ± 7 CD68+ cells at the peak; thereafter, their numbers gradually declined to 14 ± 2 CD68+ cells on day 21 (Fig. 4C). The numbers of macrophages in DZs and LZs followed the same kinetics, with peak numbers of 12 ± 4 and 7 ± 2 CD68+ cells on day 10 (Fig. 4C). Correlation analysis revealed a significant positive relationship between CD68+ cell numbers and GC or compartment size (Spearman ρGC > 0.81, ρDZ > 0.83, and ρLZ > 0.76; p < 0.0001; Fig. 4D). Furthermore, CD68+ cell numbers were positively correlated to Ki-67+ proliferating cell numbers (Spearman ρGC > 0.74, ρDZ > 0.79, and ρLZ > 0.77; p < 0.0001). The slope estimates of linear regressions between CD68+ cell numbers and GC compartment size showed only small variation in the course of the immune response (CV < 17%; Fig. 4D, Table III, Supplemental Table I), and the average densities of macrophages in DZs and LZs proved to be constant (e.g., 1124 ± 170 cells/mm2 DZ and 840 ± 20 cells/mm2 LZ on day 10; Fig. 4F). As of day 8, the density of CD68+ macrophages in the DZ was significantly increased compared with that of the LZ (p ≤ 0.05, Mann–Whitney U tests; Fig. 4E).

FIGURE 4.

Dynamics of CD68+ GC macrophages. Mice were immunized with phOx-CSA and sacrificed 4–21 d later. Spleen sections were prepared, stained, and scored as described in the legend for Fig. 1 and in 1Materials and Methods. A, Photomicrographs representative of GCs recorded on days 4, 6, 10, 14, and 21 postimmunization. GCs were identified as Ki-67+ cell clusters and by anatomical location in triple-immunofluorescence stainings of proliferating cells (Ki-67, blue), macrophages (CD68, green), and FDC networks (FDC-M2, red). LZ and DZ were distinguished by the presence of FDCs in the LZ. Scale bar, 100 μm. B, Time-dependent changes in numbers of CD68+ macrophages in GCs, DZ, and LZ plotted as cumulative frequency curves. C, Average kinetics of CD68+ macrophages present in GCs (filled area), DZ (filled squares), and LZ (open squares), expressed as mean and SD of two to four mice. D, The numbers of CD68+ macrophages in GCs, DZ, and LZ are positively correlated to the size of the respective compartment (Spearman ρ > 0.78; p < 0.0001). Lines represent linear fits of compiled data for each day (see Supplemental Table I for individual slopes m and R2 values); mean slopes (days 8–21) are indicated by m¯. E, Density of CD68+ macrophages in DZ (squares) and LZ (circles). Each symbol represents a single DZ or LZ, and the data are compilations of all calculated values for a given day; bars show mean and SD. Statistically significant differences between DZ and LZ are indicated: **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, Mann–Whitney U test. F, Average kinetics of the densities of CD68+ macrophages in GCs (filled area), DZ (filled squares), and LZ (open squares). Data are expressed as mean and SD of two to four mice.

FIGURE 4.

Dynamics of CD68+ GC macrophages. Mice were immunized with phOx-CSA and sacrificed 4–21 d later. Spleen sections were prepared, stained, and scored as described in the legend for Fig. 1 and in 1Materials and Methods. A, Photomicrographs representative of GCs recorded on days 4, 6, 10, 14, and 21 postimmunization. GCs were identified as Ki-67+ cell clusters and by anatomical location in triple-immunofluorescence stainings of proliferating cells (Ki-67, blue), macrophages (CD68, green), and FDC networks (FDC-M2, red). LZ and DZ were distinguished by the presence of FDCs in the LZ. Scale bar, 100 μm. B, Time-dependent changes in numbers of CD68+ macrophages in GCs, DZ, and LZ plotted as cumulative frequency curves. C, Average kinetics of CD68+ macrophages present in GCs (filled area), DZ (filled squares), and LZ (open squares), expressed as mean and SD of two to four mice. D, The numbers of CD68+ macrophages in GCs, DZ, and LZ are positively correlated to the size of the respective compartment (Spearman ρ > 0.78; p < 0.0001). Lines represent linear fits of compiled data for each day (see Supplemental Table I for individual slopes m and R2 values); mean slopes (days 8–21) are indicated by m¯. E, Density of CD68+ macrophages in DZ (squares) and LZ (circles). Each symbol represents a single DZ or LZ, and the data are compilations of all calculated values for a given day; bars show mean and SD. Statistically significant differences between DZ and LZ are indicated: **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, Mann–Whitney U test. F, Average kinetics of the densities of CD68+ macrophages in GCs (filled area), DZ (filled squares), and LZ (open squares). Data are expressed as mean and SD of two to four mice.

Close modal

Whereas B cell zones of naive mice were sparse in CD3+ T cells, they already contained numerous small CD68+ macrophages (Supplemental Fig. 2A, 2B). Indeed, we found that the density of CD68+ cells in B cell zones was even slightly higher in naive than in immunized mice (1443 ± 213 versus 1130 ± 60; p < 0.0001, Mann–Whitney U test; Supplemental Fig. 2D, 2F). Thus, macrophages are present in B cell follicles before antigenic stimulation, that is, when GCs are absent.

To examine whether the stable macrophage densities are tied to our model system and immunization protocol, we additionally analyzed spleen sections from systemic lupus erythematosus-prone mouse strains (NZB, NZB/W) that spontaneously develop GCs in the absence of either purposeful immunization or infection (32). Importantly, the mean macrophage densities in B cell zones of young NZB (1150 ± 149 cells/mm2) and NZB/W mice (1387 ± 130 cells/mm2) turned out to be very similar to those of immunized (1130 ± 60 cells/mm2) or naive BALB/c mice (1443 ± 213 cells/mm2; Supplemental Fig. 2G). However, disease progression in NZB/W mice involved a decrease in macrophage density of more than one third (845 ± 59 cells/mm2; Supplemental Fig. 2G).

GCs are sites of significant cell death, at which apoptosis is thought to reflect negative selection of lower affinity and autoreactive B cells. To monitor the dynamics of cell death within GCs, we applied the TUNEL assay to spleen tissue derived from BALB/c mice 4–21 d after phOx-CSA immunization (Fig. 1, Table I). In accordance with previous studies, TUNEL+ signals in GCs were confined exclusively to macrophages (33), with the number of apoptotic nuclei engulfed by a macrophage defining its size (Fig. 5A, 5D). Next to the highly loaded large macrophages commonly referred to as TBM, we observed considerable numbers of small macrophages at all time points (Fig. 5A). Although large and small macrophages were detected consistently in both DZs and LZs, the TBM-like macrophages preferentially resided at the interface between DZ and LZ (Fig. 4A).

FIGURE 5.

Dynamics of cell death within GCs. Mice were immunized with phOx-CSA and sacrificed 4–21 d later. Spleen sections were prepared, stained, and scored as described in the legend for Fig. 1 and in 1Materials and Methods. A, Photomicrographs representative of GCs recorded on days 4, 6, 10, 16, and 21 after immunization. GCs were identified by PNA reactivity and anatomical location in triple-immunofluorescence stainings of GC B cells (PNA, blue), macrophages (CD68, green), and apoptotic nuclei (TUNEL assay, red). Scale bar, 100 μm. B, Time-dependent changes in numbers of TUNEL+ nuclei in GCs plotted as cumulative frequency curves. C, Average kinetics of TUNEL+ nuclei present in GCs (filled area), expressed as mean and SD of two to four mice. D, GCs of the same cross-sectional size harbor arbitrary numbers of apoptotic nuclei (TUNEL assay, red) and differ in the size of GC macrophages (CD68, green). Scale bar, 100 μm. Images are representative of three GCs on day 10 postimmunization. E, The number of TUNEL+ nuclei in GCs is positively correlated to GC size (Spearman ρ > 0.84; p < 0.0001). Lines represent linear fits of compiled data for each day (see Supplemental Table I for individual slopes m and R2 values); mean slopes (days 8–21) are indicated by m¯. F, Density of TUNEL+ nuclei in GCs. Each symbol represents a single GC, and the data are compilations of all calculated values for a given day; bars show mean and SD. G, Average kinetics of the densities of TUNEL+ nuclei in GCs. Data are expressed as mean and SD of two to four mice.

FIGURE 5.

Dynamics of cell death within GCs. Mice were immunized with phOx-CSA and sacrificed 4–21 d later. Spleen sections were prepared, stained, and scored as described in the legend for Fig. 1 and in 1Materials and Methods. A, Photomicrographs representative of GCs recorded on days 4, 6, 10, 16, and 21 after immunization. GCs were identified by PNA reactivity and anatomical location in triple-immunofluorescence stainings of GC B cells (PNA, blue), macrophages (CD68, green), and apoptotic nuclei (TUNEL assay, red). Scale bar, 100 μm. B, Time-dependent changes in numbers of TUNEL+ nuclei in GCs plotted as cumulative frequency curves. C, Average kinetics of TUNEL+ nuclei present in GCs (filled area), expressed as mean and SD of two to four mice. D, GCs of the same cross-sectional size harbor arbitrary numbers of apoptotic nuclei (TUNEL assay, red) and differ in the size of GC macrophages (CD68, green). Scale bar, 100 μm. Images are representative of three GCs on day 10 postimmunization. E, The number of TUNEL+ nuclei in GCs is positively correlated to GC size (Spearman ρ > 0.84; p < 0.0001). Lines represent linear fits of compiled data for each day (see Supplemental Table I for individual slopes m and R2 values); mean slopes (days 8–21) are indicated by m¯. F, Density of TUNEL+ nuclei in GCs. Each symbol represents a single GC, and the data are compilations of all calculated values for a given day; bars show mean and SD. G, Average kinetics of the densities of TUNEL+ nuclei in GCs. Data are expressed as mean and SD of two to four mice.

Close modal

Automated quantification of cell death (Fig. 1) showed that the distributions of the numbers of TUNEL+ nuclei among GCs were broad at all sampled time points (CV > 55%) and changed with time (Fig. 5B). Between days 8 and 10 and days 12 and 21, however, these changes were found to be statistically insignificant (p > 0.05, Kruskal–Wallis ANOVA). Correlation analysis further demonstrated a positive relationship between TUNEL+ nuclei and GC size (Spearman ρGC > 0.84; p < 0.0001; Fig. 5E), with very similar slope estimates for linear regression curves from days 8 to 21 (CV < 12%; Table III, Supplemental Table I). Despite the established correlation, we occasionally observed GCs of the same size in a tissue section showing strikingly arbitrary numbers of TUNEL+ nuclei. Visually, these GCs can additionally be disting–uished by the size of their macrophages (Fig. 5D). The levels of cell death in B cell zones of naive mice and in nascent GCs of immunized mice were low and indistinguishable from other regions of the spleen (Fig. 5A, Supplemental Fig. 2C). A steep increase in the average numbers of TUNEL+ nuclei by a factor of 3 was first observed between days 6 and 8 (Fig. 5C), coextensive with the believed onset of SHM. Over the same period, the density of TUNEL+ nuclei also doubled significantly (p ≤ 0.001, Mann–Whitney U test; Fig. 5F, 5G). Whereas the average number of TUNEL+ nuclei per GCs showed a peak on day 10 (64 ± 12 TUNEL+ nuclei; Fig. 5D), the average density of TUNEL+ nuclei in GCs remained constant between days 8 and 21 (e.g., 2357 ± 167 TUNEL+ nuclei/mm2 on day 10; Fig. 5G).

Although cross-sectional profiling does not justify specifying a typical GC in terms of absolute size and cell numbers, it allows delineation of the typical cellular composition of GCs. In this study, the recorded numbers of T cells, macrophages, and apoptotic nuclei per GC all feature large dispersion from the mean of the ensemble (CV > 55%) and by no means follow a Gaussian distribution. We therefore believe that definite quantitative statements, for instance, that the typical GC on day 10 has a size of 0.019 mm2 and shows ∼22 T cells, on average, are questionable and not particularly meaningful. Just as an example, on day 10, <20% of phOx-CSA–induced GCs have a size in the range of 0.019 mm2 ± 25%. However, common to all recorded cell types is the strong correlation of their cell numbers with GC size. These correlations are remarkably stable during weeks 2 and 3 postimmunization, as indicated by small SDs (4–22%) of the mean slope estimates of linear regression curves fitted to each day’s GC composition parameters (Table III). This means that the quantitative relations between the cellular players of GCs are maintained. Thus, during the established phase of the phOx-CSA–induced GC response (days 8–21), the cellular composition of a GC is always roughly the same (summarized in Table IV). Moreover, because we have previously shown that the cross-sectional area distributions of spleen sections reflect broad real-size distributions of GCs, it is valid to argue that the typical cellular composition is independent of GC size.

Because there are no systemic quantitative data on the cellular composition of GCs, this study was designed as a baseline study. Nevertheless, the question whether the presented findings are specific to the hapten-carrier–induced response or equally valid for other model systems is certainly relevant. The literature provides instances of differences in the immune response as to strain of mouse and nature of Ag; however, we believe that the parameters reported in this study look unlikely to vary significantly for two reasons. First, the laboratory of T. J. Waldschmidt previously demonstrated that the GC response shows signs of a high degree of regulation that are strain and Ag independent (3, 4, 34). In particular, they found a steady ratio of nonswitched to switched GC B cells that is maintained throughout the response. This holds equally true for immunization with SRBCs, hapten-carrier conjugates, PE, and mouse-adapted influenza A virus (3). Second, we could show that the remarkably stable macrophage density in B cell zones is strain independent and not tied to our model system (Supplemental Fig. 2G). Importantly, the mean macrophage densities in B cell zones of systemic lupus erythematosus-prone NZB and NZB/W mice, which spontaneously develop GCs in the absence of either purposeful immunization or infection (32), are very similar to those of immunized or naive BALB/c mice (Supplemental Fig. 2G). However, disease progression in NZB/W mice appears to involve a severe decrease in the density of macrophages (Supplemental Fig. 2G), underscoring the importance of maintenance of cell ratios during the response.

All Ab markers, or combinations thereof, used in this study were chosen by two main criteria: 1) unambiguity of their staining pattern, and 2) suitability for semiautomated cell counting. Although PNA is the most commonly used marker for detection of GC by immunohistology, we opted for Ki-67 staining in two of the three staining series. PNA staining is sticky and the separation of PNA-high GC B cells from surrounding PNA-low resting B cells is distinctly “blurred” (a circumstance well reflected by the GC images of staining series 3 in our database). These characteristics make for difficulties in assigning “tight/accurate” outer GC boundaries to GCs visualized by PNA. In terms of measuring GC size, Ki-67 outdoes PNA for two reasons: 1) clear separation of GC B cells (Ki-67+) from surrounding resting B cells (Ki-67), and 2) applicability for automated cell counting. Finally, because it is a nuclear stain, Ki-67 facilitates counting of other surface-stained cells (in our case, CD3+ T cells and CD68+ macrophages in staining series 1 and 2).

The Ki-67 Ag is a well-known proliferation marker that stains the growth fraction of cell populations, that is, cells that are in any phase of the cell cycle other than G0 (35). However, although not all B cells in a GC are actively proliferating, almost all GC B cells stain positive for Ki-67. Although this statement may appear a contradiction, it is experimentally validated: in 2003, Rahman et al. (23) showed that all splenic GCs of SRBC-immunized A/J and C57BL/6 mice stain extensively for Ki-67. In agreement with this, Wang et al. (22) and Linterman et al. (36) report proportions of Ki-67–expressing cells among GC B cells of 90–100% in SRBC-immunized 129/Sv or C57BL/6 mice. Most importantly, the immunohistological results published by Wang et al. (22) reveal that anti–Ki-67 and PNA stain and cover identical GC areas. Altogether, this proves Ki-67 as an eligible marker for immunohistological identification of GCs.

FDC-M2 is a well-established and widely used FDC marker that is also lowly expressed on white pulp reticular cells and some uncharacterized perivascular cells (37). However, the latter does not interfere with detection of FDC networks or other GC cell populations. This and the fact that FDC-M2 expression is constant and independent of whether FDCs are involved in GC reactions (38) made it the marker of choice for our study.

Recent live-imaging studies using a photoactivatable fluorescent reporter emphasize once more that GCs are highly polarized with respect to function, whereas the DZ specializes in cell division, Ag-driven selection takes places in the LZ (18). Such segregation suggests a DZ-LZ interdependence model for GC action wherein the two zones are connected by B cell migration (3941). In line with this, we find evidence for a dependency between DZ and LZ in terms of their growth kinetics. First, DZs and LZs grow and decay in parallel during the established phase of the response (Fig. 2B, 2C). Moreover, the overall ratio of B cells in the DZ and the LZ is fixed over time, as indicated by stable DZ/LZ area curves over days (Fig. 2B). Interestingly, a great proportion of cross-sectional GCs show a DZ/LZ size ratio of ∼1 (Fig. 2B), which led us to conjecture that the volumes occupied by the DZ and the LZ are about the same. Indeed, when we determined the DZ-to-LZ volume ratios of 70 three-dimensional–reconstructed day 10 GCs (25), the values ranged from 0.5 to 1.9 with a mean of 1.0 ± 0.4 for GCs with a size ≥2.5 mm3 (Supplemental Fig. 3B). GC formation starts in the LZ (Fig. 2A) (22) and occurs over an extended period (25). Therefore, small GCs tend to show DZ/LZ ratios <1 (Supplemental Fig. 3B).

The GC reaction depends critically on T cell help because the latter not only affects initiation and maintenance of GCs but also differentiation of high-affinity GC B cells into memory B cells or Ab-forming cells (16, 42). Within the population of follicular-homing T cells that is commonly distinguished by expression of the CXCR5, different subsets of follicular T cells have been identified (reviewed in Ref. 16); however, in this study, we did not consider subsets but evaluated the entirety of T cells within the confines of GCs. To visualize follicular/GC T cells, we chose CD3 instead of CD4 as a marker because it has previously been demonstrated that many CD4-immunoreactive cells in B cell follicles are not T cells but belong to a CD4+CD3 population of accessory cells (7).

The observation that CD3+ T cells are absent or only sporadically present in B cell zones of naive mice suggests that follicular localization of T cells depends on immunization. This runs counter to previous research that identifies substantial numbers of T cells in B cell follicles before immunization (43). The discrepancy is probably due to the different markers used to visualize follicular T cells (e.g., like CD4, Thy1 has been reported to be expressed by follicular non-T CD4+CD3 accessory cells). Colonization of nascent GCs by T cells was found to be followed by an early phase of T cell proliferation reaching its peak by the end of the first week. This early proliferative response of T cells in GCs may not be related to cognate B–T interaction and provision of B cell help because T cell accumulation in follicles is not directed by B cells but by DCs (43), and the early phase of GC formation takes place in the absence of T cells (44, 45). It is more likely that proliferation of T cells during GC formation is linked to regular T cell differentiation pathways including clonal expansion and formation of CD4 memory (4649).

Interactions between GC B and T cells are required for positive selection of high-affinity GC B cells and maintenance of GCs (18, 45). Recent insights obtained by real-time imaging of GCs support a model in which competition among GC B cells for cognate T cell help is one aspect of selection (50). Because T cells are believed to be most abundant in the LZ, these events have been attributed especially to this GC compartment. However, although enrichment of T cells in the LZ is certainly real (and statistically significant), another reality is that the DZ of murine GCs is anything but poor in T cells. In stark contrast with chronically inflamed human tonsils, where T cell numbers in the DZ are vanishingly small (51), the T cell density in the DZ of murine GCs is more than two thirds of that in the LZ (Fig. 3F). DZ and LZ differ in the T cell-to-macrophage ratio rather than in the density of T cells: in the LZ, two T cells must share one macrophage, whereas in the DZ, every T cell has its own (Table IV). The former might be of importance because dying GC B cells have been reported to release cytoplasmic blebs that can be picked up by macrophages and T cells alike. Based on the latter observation, it was suggested that apoptotic B cell blebs influence the availability of T cell help to live GC B cells, thereby driving selection (50). However, because both T cells and macrophages pick up blebs, it seems possible that the fraction of macrophages also has an impact on the available T cell help, and that the T cell-to-macrophage ratio constitutes a selective force.

Table IV.
The typical GC
RatioGCaDZLZ
T:Ki 12.5:100 9.6:100 15.1:100 
M:Ki 8.3:100 8.9:100 7.2:100 
A:Ki 19.6:100   
M:T 6.7:10 9.0:10 5.1:10 
T:A 6.4:10   
M:A 4.3:10   
DZ:LZ (volume to volume)b 1:1   
RatioGCaDZLZ
T:Ki 12.5:100 9.6:100 15.1:100 
M:Ki 8.3:100 8.9:100 7.2:100 
A:Ki 19.6:100   
M:T 6.7:10 9.0:10 5.1:10 
T:A 6.4:10   
M:A 4.3:10   
DZ:LZ (volume to volume)b 1:1   
a

Cellular composition of the typical GC and its DZ and LZ. Indicated are the ratios of the different cellular players as derived from the mean slope estimates for GC composition parameters. Symbols are the same as in Table II.

b

Estimated from 70 previously three-dimensional–reconstructed day 10 GCs (Supplemental Fig. 3B).

The presence of large phagocytic cells in GCs has been recognized for 125 y (52). Early light and electron microscopic studies revealed that these characteristic phagocytes contain many tingible bodies of lymphocyte origin in their cytoplasm, which led to their naming as TBM (53, 54). The use of CD68 Abs to investigate GC macrophages in this study brought out results that conflict with previous research using Abs specific to the Mac-2 Ag (21). Most importantly, CD68 reveals considerable numbers of macrophages in both primary follicles and GCs that are missed by Mac-2.

Regarding GCs, macrophage populations are larger and more heterogeneous than previously thought. This finding manifests itself in an increased macrophage-to-B cell ratio (8:100 versus 1:350) (21) and in the abundance of small macrophages at all time points. Hence TBM as defined earlier make up only a subpopulation of GC macrophages. Furthermore, in contrast with current concepts (20, 21), macrophages are present in follicles before immunologic stimulation and may, after all, play a role in the induction of de novo GC formation, for instance, by functioning as APCs. Unlike the large TBM-like macrophages (20–30 μm in major axis length) stained by anti–Mac-2 Ab (21), the Mac-2 macrophages in primary follicles are typically of small size (10–15 μm in major axis length) and have not yet taken up apoptotic cells. Given the differential expression of Mac-2 during macrophage differentiation (55) and its variation with strength of inflammatory stimuli (56), it is tempting to speculate that follicular macrophages upregulate Mac-2 in response to uptake of dying cells.

The copresence of both Mac-2+ and Mac-2 macrophages in GCs may be significant because shifts in phenotypes of macrophages during tumor growth have been shown to be associated with immunoregulation (57, 58). In this context, Mac-2+ macrophages produce large amounts of PGE2, an arachidonic acid metabolite capable of suppressing B cell proliferation (59, 60). Interestingly, Mac-2 macrophages counteract this suppression (57, 58). As we have shown, the apoptotic load of GCs increases with their size. At the same time, increasing numbers of apoptotic cells lead to accumulation of large macrophages. That is, if uptake of apoptotic cells in GCs induces a phenotypic shift from small (nonsuppressive) Mac-2 to large (suppressive) Mac-2+ macrophages, this could provide a regulatory framework for GC B cell homeostasis. Indeed, two lines of evidence support the presence of a feedback relation between GC B cells and macrophages: first, Mac-2+ TBM have been identified as a rich source of PGs (61); second, the macrophage-to-B cell ratio is remarkably constant throughout the GC response, emphasizing the importance of stable macrophage–B cell interactions.

Where, when, and how fast do GC B cells die? The answers to these questions are of importance because they may provide information on the spatiotemporal regulation of GC selection. Marked apoptosis beyond basal levels is first detected by the end of the first week postimmunization (Fig. 5C) and thus coincides with the onset of SHM and selection (62). Thereafter, the apoptotic (TUNEL+)-to-proliferating (Ki-67+) B cell ratio remains fixed at ∼1:5 (Table III), an observation consistent with either stable average susceptibility of GC B cells to apoptosis or steady selection stringency. As to the where, the traces of cell death are spread across GCs. Although we did not quantify the numbers of TUNEL+ nuclei in different GC compartments in this study, we qualitatively observed two regions of elevated cell death. Both the interface between DZ and LZ (also referred to as the basal LZ) and the interface between DZ and the periarterial lymphatic sheath are often rich in large, TBM-like macrophages (staining series 2) that colocalize with large clusters of TUNEL+ nuclei (staining series 3). However, the finding that TUNEL+ nuclei of dying GC B cells are only present in association with macrophage scavengers but not “freely” has two substantial implications for the interpretation of the data: 1) the clearance time of apoptotic GC B cells is very likely to be short, and 2) the spatial distribution of TUNEL+ signals in GCs need not necessarily reflect where GC B cells are programmed to die. Provided that dying GC B cells are rapidly engulfed and degraded by macrophages, they ought to leave the site of death almost immediately, especially if cell intermixture is quick, as has been shown for GCs (63). Thus, TUNEL detection alone gives insufficient evidence as to where the initial selective signals that eventually lead to apoptosis of GC B cells are (or, for that matter, are not) delivered.

Indeed, where calculated, clearance times of cells undergoing apoptosis are surprisingly short, ranging from 1 to 3 h for different tissues (64). Given an incidence, i, of TUNEL+ nuclei among GC B cells of 20% (1:5) and assuming a clearance time, d, of 3 h and a correction factor, f = 0.5, to account for the formation of more than one TUNEL+ nucleus from an apoptotic cell, the minimal cell loss rate, r (65), in GCs by apoptosis is given by r=i*fd3%/h.

In our previous work we found that the growth of de novo–induced GCs is nonsynchronized and further characterized by broad GC volume distributions at anytime (25). Given this heterogeneity in size, and probably also age, the stable cellular composition of GCs we report in this article is remarkable. GCs can be viewed as open systems (30) that continuously deal with factors such as high proliferative and apoptotic activities, internal cell migration, and cell influx and efflux. The observed maintenance of quantitative relations between the cellular players during the life span of GCs therefore implies the existence of a tightly regulated network of cellular interactions and communication. It will be particularly interesting to explore how this network achieves its robustness in the highly dynamic environment of a GC. We believe that integrative mathematical models, combining quantitative data on cellular composition, cell migration, clonal diversity, and overall GC growth kinetics, will be essential for understanding the complex regulation of GC cell dynamics and its linkage with GC function. Our study makes a contribution toward such models because it provides quantitative time-resolved data on the cellular composition of GCs. Setting up elaborate models, however, requires further data. These are, in particular, the cell influx and efflux rates of GCs. Aside from the future challenges of modeling GC dynamics, such as assessment of aberrations from the “perfect” GC reaction, we suggest that monitoring changes in the ratios of GC cells to each other is valuable and more sensitive than following average values.

We especially thank the DRFZ, a Leibnitz Institute, in Berlin for providing laboratory space, equipment, and generous support during the project. We also thank Prof. Rudolf Manz and Dr. Katrin Moser for providing cryostat sections of NZB and NZB/W spleens.

This work was supported by the Volkswagen Foundation and the Bundesministerium für Bildung und Forschung (Germany) (Grant 0315005B).

The online version of this article contains supplemental material.

Abbreviations used in this article:

CSA

chicken serum albumin

CV

coefficient of variation

DRFZ

Deutsches Rheuma-Forschungszentrum Berlin

DZ

dark zone

FDC

follicular dendritic cell

GC

germinal center

LZ

light zone

phOx

2-phenyl-5-oxazolone

PNA

peanut agglutinin

ROI

region of interest

SHM

somatic hypermutation

TBM

tingible body macrophage.

1
MacLennan
I. C.
1994
.
Germinal centers.
Annu. Rev. Immunol.
12
:
117
139
.
2
Tarlinton
D. M.
2008
.
Evolution in miniature: selection, survival and distribution of antigen reactive cells in the germinal centre.
Immunol. Cell Biol.
86
:
133
138
.
3
Alexander
C. M.
,
Tygrett
L. T.
,
Boyden
A. W.
,
Wolniak
K. L.
,
Legge
K. L.
,
Waldschmidt
T. J.
.
2011
.
T regulatory cells participate in the control of germinal centre reactions.
Immunology
133
:
452
468
.
4
Shinall
S. M.
,
Gonzalez-Fernandez
M.
,
Noelle
R. J.
,
Waldschmidt
T. J.
.
2000
.
Identification of murine germinal center B cell subsets defined by the expression of surface isotypes and differentiation antigens.
J. Immunol.
164
:
5729
5738
.
5
Wolniak
K. L.
,
Shinall
S. M.
,
Waldschmidt
T. J.
.
2004
.
The germinal center response.
Crit. Rev. Immunol.
24
:
39
65
.
6
Grouard
G.
,
Durand
I.
,
Filgueira
L.
,
Banchereau
J.
,
Liu
Y. J.
.
1996
.
Dendritic cells capable of stimulating T cells in germinal centres.
Nature
384
:
364
367
.
7
Kim
M. Y.
,
Gaspal
F. M.
,
Wiggett
H. E.
,
McConnell
F. M.
,
Gulbranson-Judge
A.
,
Raykundalia
C.
,
Walker
L. S.
,
Goodall
M. D.
,
Lane
P. J.
.
2003
.
CD4(+)CD3(-) accessory cells costimulate primed CD4 T cells through OX40 and CD30 at sites where T cells collaborate with B cells.
Immunity
18
:
643
654
.
8
Carroll
M. C.
1998
.
The role of complement and complement receptors in induction and regulation of immunity.
Annu. Rev. Immunol.
16
:
545
568
.
9
Liu
Y. J.
,
Xu
J.
,
de Bouteiller
O.
,
Parham
C. L.
,
Grouard
G.
,
Djossou
O.
,
de Saint-Vis
B.
,
Lebecque
S.
,
Banchereau
J.
,
Moore
K. W.
.
1997
.
Follicular dendritic cells specifically express the long CR2/CD21 isoform.
J. Exp. Med.
185
:
165
170
.
10
Qin
D.
,
Wu
J.
,
Vora
K. A.
,
Ravetch
J. V.
,
Szakal
A. K.
,
Manser
T.
,
Tew
J. G.
.
2000
.
Fc gamma receptor IIB on follicular dendritic cells regulates the B cell recall response.
J. Immunol.
164
:
6268
6275
.
11
Liu
Y. J.
,
Johnson
G. D.
,
Gordon
J.
,
MacLennan
I. C.
.
1992
.
Germinal centres in T-cell-dependent antibody responses.
Immunol. Today
13
:
17
21
.
12
van Eijk
M.
,
Medema
J. P.
,
de Groot
C.
.
2001
.
Cutting edge: cellular Fas-associated death domain-like IL-1-converting enzyme-inhibitory protein protects germinal center B cells from apoptosis during germinal center reactions.
J. Immunol.
166
:
6473
6476
.
13
Kosco
M. H.
,
Pflugfelder
E.
,
Gray
D.
.
1992
.
Follicular dendritic cell-dependent adhesion and proliferation of B cells in vitro.
J. Immunol.
148
:
2331
2339
.
14
Liu
Y. J.
,
de Bouteiller
O.
,
Fugier-Vivier
I.
.
1997
.
Mechanisms of selection and differentiation in germinal centers.
Curr. Opin. Immunol.
9
:
256
262
.
15
Meyer-Hermann
M. E.
,
Maini
P. K.
,
Iber
D.
.
2006
.
An analysis of B cell selection mechanisms in germinal centers.
Math. Med. Biol.
23
:
255
277
.
16
Vinuesa
C. G.
,
Tangye
S. G.
,
Moser
B.
,
Mackay
C. R.
.
2005
.
Follicular B helper T cells in antibody responses and autoimmunity.
Nat. Rev. Immunol.
5
:
853
865
.
17
Hardie
D. L.
,
Johnson
G. D.
,
Khan
M.
,
MacLennan
I. C.
.
1993
.
Quantitative analysis of molecules which distinguish functional compartments within germinal centers.
Eur. J. Immunol.
23
:
997
1004
.
18
Victora
G. D.
,
Schwickert
T. A.
,
Fooksman
D. R.
,
Kamphorst
A. O.
,
Meyer-Hermann
M.
,
Dustin
M. L.
,
Nussenzweig
M. C.
.
2010
.
Germinal center dynamics revealed by multiphoton microscopy with a photoactivatable fluorescent reporter.
Cell
143
:
592
605
.
19
Gaipl
U. S.
,
Voll
R. E.
,
Sheriff
A.
,
Franz
S.
,
Kalden
J. R.
,
Herrmann
M.
.
2005
.
Impaired clearance of dying cells in systemic lupus erythematosus.
Autoimmun. Rev.
4
:
189
194
.
20
Smith
J. P.
,
Kosco
M. H.
,
Tew
J. G.
,
Szakal
A. K.
.
1988
.
Thy-1 positive tingible body macrophages (TBM) in mouse lymph nodes.
Anat. Rec.
222
:
380
390
.
21
Smith
J. P.
,
Lister
A. M.
,
Tew
J. G.
,
Szakal
A. K.
.
1991
.
Kinetics of the tingible body macrophage response in mouse germinal center development and its depression with age.
Anat. Rec.
229
:
511
520
.
22
Wang
Y.
,
Carter
R. H.
.
2005
.
CD19 regulates B cell maturation, proliferation, and positive selection in the FDC zone of murine splenic germinal centers.
Immunity
22
:
749
761
.
23
Rahman
Z. S.
,
Rao
S. P.
,
Kalled
S. L.
,
Manser
T.
.
2003
.
Normal induction but attenuated progression of germinal center responses in BAFF and BAFF-R signaling-deficient mice.
J. Exp. Med.
198
:
1157
1169
.
24
Takahashi
Y.
,
Cerasoli
D. M.
,
Dal Porto
J. M.
,
Shimoda
M.
,
Freund
R.
,
Fang
W.
,
Telander
D. G.
,
Malvey
E. N.
,
Mueller
D. L.
,
Behrens
T. W.
,
Kelsoe
G.
.
1999
.
Relaxed negative selection in germinal centers and impaired affinity maturation in bcl-xL transgenic mice.
J. Exp. Med.
190
:
399
410
.
25
Wittenbrink
N.
,
Weber
T. S.
,
Klein
A.
,
Weiser
A. A.
,
Zuschratter
W.
,
Sibila
M.
,
Schuchhardt
J.
,
Or-Guil
M.
.
2010
.
Broad volume distributions indicate nonsynchronized growth and suggest sudden collapses of germinal center B cell populations.
J. Immunol.
184
:
1339
1347
.
26
Faro
J.
,
OrGuil
M.
.
2011
.
Reassessing germinal center reaction concepts
. In
Mathematical Models in Immune Cell Biology
.
Lythe
G.
,
Molina-París
C.
, eds.
Springer Press
,
New York
, p.
424
.
27
Berek
C.
,
Griffiths
G. M.
,
Milstein
C.
.
1985
.
Molecular events during maturation of the immune response to oxazolone.
Nature
316
:
412
418
.
28
Allen
C. D.
,
Ansel
K. M.
,
Low
C.
,
Lesley
R.
,
Tamamura
H.
,
Fujii
N.
,
Cyster
J. G.
.
2004
.
Germinal center dark and light zone organization is mediated by CXCR4 and CXCR5.
Nat. Immunol.
5
:
943
952
.
29
Hauser
A. E.
,
Junt
T.
,
Mempel
T. R.
,
Sneddon
M. W.
,
Kleinstein
S. H.
,
Henrickson
S. E.
,
von Andrian
U. H.
,
Shlomchik
M. J.
,
Haberman
A. M.
.
2007
.
Definition of germinal-center B cell migration in vivo reveals predominant intrazonal circulation patterns.
Immunity
26
:
655
667
.
30
Schwickert
T. A.
,
Lindquist
R. L.
,
Shakhar
G.
,
Livshits
G.
,
Skokos
D.
,
Kosco-Vilbois
M. H.
,
Dustin
M. L.
,
Nussenzweig
M. C.
.
2007
.
In vivo imaging of germinal centres reveals a dynamic open structure.
Nature
446
:
83
87
.
31
Abramoff
M. D.
,
Magelhaes
P. J.
,
Ram
S. J.
.
2004
.
Image processing with ImageJ.
Biophotonics International
11
:
36
42
.
32
Luzina
I. G.
,
Atamas
S. P.
,
Storrer
C. E.
,
daSilva
L. C.
,
Kelsoe
G.
,
Papadimitriou
J. C.
,
Handwerger
B. S.
.
2001
.
Spontaneous formation of germinal centers in autoimmune mice.
J. Leukoc. Biol.
70
:
578
584
.
33
Ninomiya
E.
,
Ito
Y.
,
Shibata
M. A.
,
Kawashima
K.
,
Sakamoto
T.
,
Maruyama
E.
,
Doi
H.
,
Tokitsu
K.
,
Otsuki
Y.
.
2003
.
The activation of caspase-3 and DNA fragmentation in B cells phagocytosed by macrophages.
Med. Electron Microsc.
36
:
87
93
.
34
Wolniak
K. L.
,
Noelle
R. J.
,
Waldschmidt
T. J.
.
2006
.
Characterization of (4-hydroxy-3-nitrophenyl)acetyl (NP)-specific germinal center B cells and antigen-binding B220- cells after primary NP challenge in mice.
J. Immunol.
177
:
2072
2079
.
35
Scholzen
T.
,
Gerdes
J.
.
2000
.
The Ki-67 protein: from the known and the unknown.
J. Cell. Physiol.
182
:
311
322
.
36
Linterman
M. A.
,
Beaton
L.
,
Yu
D.
,
Ramiscal
R. R.
,
Srivastava
M.
,
Hogan
J. J.
,
Verma
N. K.
,
Smyth
M. J.
,
Rigby
R. J.
,
Vinuesa
C. G.
.
2010
.
IL-21 acts directly on B cells to regulate Bcl-6 expression and germinal center responses.
J. Exp. Med.
207
:
353
363
.
37
Taylor
P. R.
,
Pickering
M. C.
,
Kosco-Vilbois
M. H.
,
Walport
M. J.
,
Botto
M.
,
Gordon
S.
,
Martinez-Pomares
L.
.
2002
.
The follicular dendritic cell restricted epitope, FDC-M2, is complement C4; localization of immune complexes in mouse tissues.
Eur. J. Immunol.
32
:
1888
1896
.
38
Camacho
S. A.
,
Kosco-Vilbois
M. H.
,
Berek
C.
.
1998
.
The dynamic structure of the germinal center.
Immunol. Today
19
:
511
514
.
39
Kepler
T. B.
,
Perelson
A. S.
.
1993
.
Cyclic re-entry of germinal center B cells and the efficiency of affinity maturation.
Immunol. Today
14
:
412
415
.
40
MacLennan
I. C.
,
Johnson
G. D.
,
Liu
Y. J.
,
Gordon
J.
.
1991
.
The heterogeneity of follicular reactions.
Res. Immunol.
142
:
253
257
.
41
Oprea
M.
,
Perelson
A. S.
.
1997
.
Somatic mutation leads to efficient affinity maturation when centrocytes recycle back to centroblasts.
J. Immunol.
158
:
5155
5162
.
42
Han
S.
,
Hathcock
K.
,
Zheng
B.
,
Kepler
T. B.
,
Hodes
R.
,
Kelsoe
G.
.
1995
.
Cellular interaction in germinal centers. Roles of CD40 ligand and B7-2 in established germinal centers.
J. Immunol.
155
:
556
567
.
43
Fillatreau
S.
,
Gray
D.
.
2003
.
T cell accumulation in B cell follicles is regulated by dendritic cells and is independent of B cell activation.
J. Exp. Med.
197
:
195
206
.
44
Lentz
V. M.
,
Manser
T.
.
2001
.
Cutting edge: germinal centers can be induced in the absence of T cells.
J. Immunol.
167
:
15
20
.
45
de Vinuesa
C. G.
,
Cook
M. C.
,
Ball
J.
,
Drew
M.
,
Sunners
Y.
,
Cascalho
M.
,
Wabl
M.
,
Klaus
G. G.
,
MacLennan
I. C.
.
2000
.
Germinal centers without T cells.
J. Exp. Med.
191
:
485
494
.
46
Gulbranson-Judge
A.
,
MacLennan
I.
.
1996
.
Sequential antigen-specific growth of T cells in the T zones and follicles in response to pigeon cytochrome c.
Eur. J. Immunol.
26
:
1830
1837
.
47
Linton
P. J.
,
Harbertson
J.
,
Bradley
L. M.
.
2000
.
A critical role for B cells in the development of memory CD4 cells.
J. Immunol.
165
:
5558
5565
.
48
van Essen
D.
,
Dullforce
P.
,
Brocker
T.
,
Gray
D.
.
2000
.
Cellular interactions involved in Th cell memory.
J. Immunol.
165
:
3640
3646
.
49
Zheng
B.
,
Han
S.
,
Zhu
Q.
,
Goldsby
R.
,
Kelsoe
G.
.
1996
.
Alternative pathways for the selection of antigen-specific peripheral T cells.
Nature
384
:
263
266
.
50
Allen
C. D.
,
Okada
T.
,
Cyster
J. G.
.
2007
.
Germinal-center organization and cellular dynamics.
Immunity
27
:
190
202
.
51
Brachtel
E. F.
,
Washiyama
M.
,
Johnson
G. D.
,
Tenner-Racz
K.
,
Racz
P.
,
MacLennan
I. C.
.
1996
.
Differences in the germinal centres of palatine tonsils and lymph nodes.
Scand. J. Immunol.
43
:
239
247
.
52
Flemming
W.
1885
.
Studien über Regeneration der Gewebe.
Arch Mikrosk Anat
24
:
355
.
53
Congdon, C. C., and J. W. Goodman. 1961. Changes in lymphatic tissues during foreign tissue transplantation. In International Symposium on Tissue Transplantation. A. P. Christoffanini and G. Hoecker, eds. Universidad de Chile, Santiago. p. 181.
54
Swartzendruber
D. C.
,
Congdon
C. C.
.
1963
.
Electron microscope observations on tingible body macrophages in mouse spleen.
J. Cell Biol.
19
:
641
646
.
55
Hughes
R. C.
1994
.
Mac-2: a versatile galactose-binding protein of mammalian tissues.
Glycobiology
4
:
5
12
.
56
Ho
M. K.
,
Springer
T. A.
.
1982
.
Mac-1 antigen: quantitative expression in macrophage populations and tissues, and immunofluorescent localization in spleen.
J. Immunol.
128
:
2281
2286
.
57
Garner
R. E.
,
Malick
A. P.
,
Yurochko
A. D.
,
Elgert
K. D.
.
1987
.
Shifts in macrophage (M phi) surface phenotypes during tumor growth: association of Mac-2+ and Mac-3+ M phi with immunosuppressive activity.
Cell. Immunol.
108
:
255
268
.
58
Malick
A. P.
,
Elgert
K. D.
,
Garner
R. E.
,
Adkinson
N. F.
 Jr.
1987
.
Prostaglandin E2 production by Mac-2+ macrophages: tumor-induced population shift.
J. Leukoc. Biol.
42
:
673
681
.
59
Murn
J.
,
Alibert
O.
,
Wu
N.
,
Tendil
S.
,
Gidrol
X.
.
2008
.
Prostaglandin E2 regulates B cell proliferation through a candidate tumor suppressor, Ptger4.
J. Exp. Med.
205
:
3091
3103
.
60
Simkin
N. J.
,
Jelinek
D. F.
,
Lipsky
P. E.
.
1987
.
Inhibition of human B cell responsiveness by prostaglandin E2.
J. Immunol.
138
:
1074
1081
.
61
Smith
J. P.
,
Burton
G. F.
,
Tew
J. G.
,
Szakal
A. K.
.
1998
.
Tingible body macrophages in regulation of germinal center reactions.
Dev. Immunol.
6
:
285
294
.
62
Jacob
J.
,
Przylepa
J.
,
Miller
C.
,
Kelsoe
G.
.
1993
.
In situ studies of the primary immune response to (4-hydroxy-3-nitrophenyl)acetyl. III. The kinetics of V region mutation and selection in germinal center B cells.
J. Exp. Med.
178
:
1293
1307
.
63
Figge
M. T.
,
Garin
A.
,
Gunzer
M.
,
Kosco-Vilbois
M.
,
Toellner
K. M.
,
Meyer-Hermann
M.
.
2008
.
Deriving a germinal center lymphocyte migration model from two-photon data.
J. Exp. Med.
205
:
3019
3029
.
64
Coles
H. S.
,
Burne
J. F.
,
Raff
M. C.
.
1993
.
Large-scale normal cell death in the developing rat kidney and its reduction by epidermal growth factor.
Development
118
:
777
784
.
65
Bursch
W.
,
Paffe
S.
,
Putz
B.
,
Barthel
G.
,
Schulte-Hermann
R.
.
1990
.
Determination of the length of the histological stages of apoptosis in normal liver and in altered hepatic foci of rats.
Carcinogenesis
11
:
847
853
.

The authors have no financial conflicts of interest.