Germinal center centroblasts transition to a centrocyte phenotype according to a timed program and depend on the dark zone for effective selection

Abstract

Germinal center (GC) B cells cycle between the dark zone (DZ) and light zone (LZ) during antibody affinity maturation. Whether this movement is necessary for GC function has not been tested. Here we show that CXCR4-deficient GC B cells, which are restricted to the LZ, are gradually outcompeted by WT cells indicating an essential role for DZ access. Remarkably, the transition between DZ centroblast and LZ centrocyte phenotypes occurred independently of positioning. However, CXCR4-deficient cells carried fewer mutations and were overrepresented in the CD73(+) memory compartment. These findings are consistent with a model where GC B cells change from DZ to LZ phenotype according to a timed cellular program but suggest that spatial separation of DZ cells facilitates more effective rounds of mutation and selection. Finally, we identify a network of DZ CXCL12-expressing reticular cells that likely support DZ functions.

Graphical abstract

Figure 1

CXCR4 Is Necessary for DZ Access and Continued Participation in Influenza and Gut Associated GCs (A) Mixed BM chimeric mice with a majority of CD45.1⁺ and a minority of CD45.2⁺Mb1-Cre⁺Cxcr4^fl/− or CD45.2⁺Mb1-Cre⁺Cxcr4^+/+ control B cells were infected with HKx31 influenza, and medLNs were harvested for IHC analysis on day 13 p.i. (B) Gating scheme for assessing the participation of CXCR4-deficient B cells within the follicular and GC compartments, showing example from day 28 p.i. (C) Data from multiple experiments were plotted as percentage of CD45.2⁺ GC B cells/% within the concurrent follicular compartment, at various time points. (D–F) Similar IHC (D) and FACS (E and F) analysis was performed for PPs and mesLNs. Dots in (C), (E), and (F) represent single animals, and error bars indicate means. Comparisons use a nonpaired two-tailed Student’s t test. Scale bars represent 200 um. See also Figure S1.

Figure 2

CXCR4 Expression Is Not Required for Effective Competition in Nonpolarized GCs BM chimeric mice were generated in which the majority (75%–95%) of B cells were CXCR4-deficient (or control), and infected with influenza virus. (A) GC polarization was determined by IHC staining of medLNs on day 12. (B) Participation of CD45.2⁺Mb1-Cre⁺Cxcr4^fl/− or control cells within follicular B cell and GC compartments at 4 weeks p.i., with each dot representing a single mouse, and error bars indicating means. Comparisons were with a paired two-tailed Student’s t test. Scale bars represent 200 um.

Figure 3

Aspects of GC B Cell Behavior Are Regulated Independently of Cues within a Particular Zone (A) IgD^loCD95⁺GL7⁺ WT GC B cells from the medLNs of influenza infected mice were costained for surface CXCR4 expression and for the LZ markers CD83 and CD86 at ∼4 wks p.i. (B) Representative plots of CD83 and CD86 staining (mean %s) on Cxcr4^fl/− and control GC B cells from mixed chimeric mice at 4 weeks p.i. (chimerism as in Figure 1 and Figure S1). Data were pooled from several experiments, with each dot in (C) representing a single mouse. (D) qPCR measurements of mRNA levels in FACS sorted IgD^loCD95⁺GL7⁺Cxcr4^fl/− and WT GC B cells from mixed BM chimeras. Data in (D) are pooled from two to three mice with error bars indicating SD. Comparisons use a paired two-tailed Student’s t test. See also Figure S2.

Figure 4

DZ Access Is Not Required for Cell-Cycle Progression or Mitosis (A and B) DNA content in Cxcr4^fl/− and control GC B cells from mixed BM chimeras (as in Figure 1) was assessed at ∼4 weeks after influenza infection (A). Data from multiple experiments are represented in (B). (C) Proliferation was determined after a single BrdU injection 30 min prior to euthanasia. Lines in (B) and (C) join CD45.1⁺ WT and CD45.2⁺Cxcr4^fl/− or control cells from the same mouse. (D) GC B cells from influenza-infected WT mice were assessed by FACS for p.H3, DNA content, and CXCR4 expression. Red box highlights p.H3 high cells. (E) MedLNs from influenza infected mixed BM chimeras were stained at 4 weeks p.i. for p.H3, CD35, IgD, and CD45.2. Arrows indicate cells in the LZ that are double positive for CD45.2 and p.H3. (F and G) p.H3 staining in WT medLN (F) and PP GCs (G). Symbols and red boxes identify the high magnification regions shown below. Comparisons are with a paired two-tailed Student’s t test. Scale bars represent 50 um.

Figure 5

Reduced Mutation Load in CXCR4-Deficient GC B Cells Follicular (A) and GC (B) B cells from medLNs of CXCR4-deficient and control mixed BM chimeric mice were FACS isolated on days 23–25 of influenza infection. The frequency of mismatch errors in 470 bp of the intron downstream of rearranged VDJ regions was determined by single molecule sequencing. CD45.2⁺Cxcr4^fl/− and control cells were compared to WT CD45.1⁺ cells from the same mice. n = number of pooled mice. (C) Similar analysis for PPs. Approximately three mice were pooled for each PP experiment, and data are pooled from indicated number of experiments. (D) Intracellular AID protein staining of medLNs from mixed BM chimeras at day 15 p.i. Aicda^−/− cells and WT CD45.1⁺ cells were mixed to provide staining controls. The GC gating scheme is shown above. Representative of six mice from three experiments. (E) RT-PCR measurements of genes associated with SHM for Cxcr4^fl/– and WT cells from mixed BM chimeras. Data are pooled from two to three mice with error bars indicating SD. See also Figure S3.

Figure 6

Normal PC Development but Increased Memory Cell Output when GC B Cells Are Restricted to the LZ (A) Representative FACS plots showing memory cell and PC gating schemes. (B) The frequency of CD45.2⁺Cxcr4^fl/− and control cells in mixed BM chimeric mice (as in Figure 1) was determined for the GC, memory, and PC compartments and plotted as fold difference relative to the concurrent follicular population. Mice were analyzed at 11 days and ∼4 wks p.i. Dots represent individual animals, and error bars show means. (C) Influenza-infected Cxcr4^fl/− and control mixed BM chimeric mice were given BrdU for 3 days at 4 weeks p.i. The proportions of follicular, GC, and BrdU⁺ memory cells that were CD45.2⁺ are connected for each mouse. (D) Memory cell proliferation was assessed by transferring splenocytes from 3–4 weeks influenza-infected mice into infection-matched recipients and giving recipients BrdU from 24 hr after transfer until analysis 3 days later. BrdU incorporation by transferred and endogenous IgD^lo GL7^lo CD73⁺ CD38⁺ memory cells was compared. Each dot represents a single animal, and error bars show means.

Figure 7

The GC DZ Contains a Network of CXCL12-Expressing Reticular Cells (A and B) Confocal microscopy of MedLNs (A) and PPs (B) from Cxcl12-gfp mice at day 11–12 after influenza infection. Note that the right micrograph in (A) is of higher magnification. (C) MedLNs and PPs from Cxcl12-gfp mice were stained for FDC markers at day 11 p.i. (D) Ubi-GFP mice were reconstituted with WT nonflorescent BM and infected with influenza. MedLNs were harvested at day 10 p.i. (E) High-powered micrographs of CXCL12-GFP⁺ stroma in the medLN GC DZ, medullary region, and T zone on day 12 p.i. (F) MedLNs from Cxcl12-gfp mice were stained with the indicated antibodies at 2 weeks p.i. (G and H) Peripheral skin-draining LN primary follicles from uninfected mice were stained with the indicated antibodies. Arrows indicate higher magnification images of boxed areas. Scale bar represents 10 μM (E), 30 μM (G and H) 60 μM (A, C, D, and F) and 100 μM (B). See also Figures S4 and S5 and Movie S1.