The early responses of follicular (Fo) and marginal zone (MZ) B cells to T cell-dependent Ag were compared using anti-hen egg lysozyme (HEL+) B cells capable of class switch recombination and somatic hypermutation (SHM). Purified CD21/35intCD23high Fo and CD21/35highCD23low MZ splenic B cells from SWHEL Ig-transgenic mice were transferred into wild-type recipients and challenged with HEL-sheep RBC. Responding HEL+ B cells from both populations switched efficiently to IgG1, generated syndecan-1+ Ab-secreting cells, and exhibited equivalent rates of proliferation. However, the expansion of HEL+ MZ B cells lagged significantly behind that of HEL+ Fo B cells due to less efficient homing to the outer periarteriolar lymphatic sheath and reduced recruitment into the proliferative response. Despite the equivalent rates of class switch recombination, the onset of SHM was delayed in the MZ subset, indicating that these two activation-induced cytidine deaminase-dependent events are uncoupled in the early response of MZ B cells. Migration of HEL+ B cells into germinal centers coincided with the onset of SHM, occurring more rapidly with Fo vs MZ responders. These results are consistent with the concept that Fo and MZ B cells have evolved to specialize in T cell-dependent and T-independent responses respectively.
Mature long-lived B cells in the spleen comprise two distinct subsets that differ significantly in their localization, cell surface phenotype, and functional properties. B cells resident in the splenic marginal zone (MZ)3 are IgMhigh, IgDlow, CD21/35high, CD23low, and express high levels of the nonclassical MHC molecule CD1d. In contrast, B cells that populate the primary follicle (Fo) are IgMlow-high, IgDhigh, CD21/35int, CD23high, and express lower levels of CD1d (reviewed in Ref. 1). MZ B cells are also larger than Fo B cells and display higher levels of class II MHC and the costimulatory molecules B7.1 (CD80) and B7.2 (CD86) (2). The in vitro responses of MZ B cells to BCR cross-linking, T cell-independent (LPS), and T cell-derived (CD40L plus IL-4) stimuli are greater than those of Fo B cells (2, 3, 4). Furthermore, MZ B cells activated either in vitro or in vivo are more effective than equivalently activated Fo B cells at costimulating naive CD4+ T cells (2, 5).
Several pieces of evidence indicate that MZ B cells play a particularly important role in mediating responses to thymus-independent type 2 (TI-2) Ags in vivo. Their localization in the splenic MZ where blood is directly filtered means that they are ideally positioned to survey the periphery for blood-borne pathogens, such as encapsulated bacteria expressing strong TI-2 Ags (6). Furthermore, the rapid response of MZ B cells to TI stimuli anticipates their prompt response to and rapid elimination of such pathogens. Consistent with this, anti-phosphorylcholine B cells enriched in the MZ of M167 Ig-transgenic (Ig-Tg) mice produce a vigorous plasmablast response in the bridging channels of the spleen after in vivo stimulation with TI-2 Ags, which is 2- to 3-fold greater than that elicited by equivalent numbers of anti-phosphorylcholine Fo B cells (7). However, the idea that MZ B cells are specialized only for TI-2 responses is not consistent with the fact that they are also superior to Fo B cells in their ability to respond to BCR cross-linking (2, 4, 8) and T cell-derived signals (2, 4) and to induce T cell proliferation (2, 5).
In an attempt to compare the relative abilities of Fo and MZ B cells to mount T cell-dependent (TD) responses in vivo, a recent study was performed in which purified Fo or MZ B cells were challenged with TD Ag following transfer into lymphocyte-deficient scid/scid mice (9). In this system, MZ B cells were the major source of primary Ab secreting cells (ASCs), formed germinal centers (GCs) albeit with delayed kinetics, and underwent somatic hypermutation (SHM). However, during TD responses, responding cells are required to navigate secondary lymphoid tissues under the control of highly regulated chemotactic cues for effective T cell-B cell (T-B) collaboration and subsequent GC formation (10). Because the scid/scid mice lack intact secondary lymphoid structure and do not have the normal competition for physiological space from excess nonresponding lymphocytes, the study does not allow definitive conclusions to be reached about the relative contributions made by Fo and MZ B cells to normal TD responses. To avoid the potential complications associated with using immunodeficient recipients, therefore, it is imperative to track the response of B cells in an adoptive transfer system within the context of normal secondary lymphoid tissues. One way of achieving this goal is to use donor B cells purified from Ig-Tg mice encoding a defined specificity identifiable by flow cytometry and immunohistology (7, 11, 12). In the current study we have used SWHEL Ig-Tg donor mice to compare the TD responses of Fo and MZ B cells that have identical specificity for hen egg lysozyme (HEL) (13) within the immune system of unmanipulated recipients.
SWHEL mice offer several advantages for this type of study. First, the anti-HEL IgH variable region gene in SWHEL mice has been targeted to the endogenous IgH locus, allowing anti-HEL (HEL+) B cells to undergo the normal processes of Ig class switch recombination (CSR) and SHM that are hallmarks of TD responses. Second, in contrast to transgene expressing B cells from many other Ig-Tg models (14, 15), the development of HEL+ B cells in SWHEL mice is not biased toward either the Fo or the MZ B cell fate (13). Finally, the combination of staining for BCR specificity (HEL-binding) and the congenic CD45 marker allows small numbers of transferred SWHEL HEL+ B cells to be tracked throughout their response, thereby providing a unique window into the early events that may distinguish between the responses of Fo and MZ B cells.
The studies in the SWHEL model reported here showed that both HEL+ peripheral B cell subsets proliferated, isotype switched, and differentiated into ASCs equivalently in response to TD challenge with HEL conjugated to the multivalent carrier sheep RBC (HEL-SRBC). However, expansion of HEL+ MZ B cells lagged behind that of HEL+ Fo B cells due to a delay in their recruitment into the immune response. This was associated with less efficient migration to the T-B border of the follicle coincident with their relative inability to access T cell help. Moreover, despite the equivalent rates of CSR, the onset of SHM was delayed in HEL+ MZ B cells indicating that these two activation-induced cytidine deaminase (AID)-dependent processes are uncoupled in the early responses of MZ B cells. The delayed SHM in responding MZ B cells correlated with their slower entry into GCs compared with responding Fo B cells. Taken together these data indicate that, despite the greater intrinsic responsiveness of MZ B cells to T cell-derived stimuli, Fo B cells are more effective at mediating TD responses in vivo.
Materials and Methods
BALB/c, BALB/c nu/nu (athymic), C57BL/6 (CD45.2+), and congenic C57BL/6-SJL.Ptpca (CD45.1+) mice were purchased from the Animal Resources Centre (Canning Vale, Western Australia). SWHEL mice (13) were generated and maintained on the original C57BL/6 background at the Centenary Institute Animal Facility. For adoptive transfers into C57BL/6 mice, SWHEL mice were crossed onto a homozygous CD45.1+ congenic background. For adoptive transfer into nu/nu mice, SWHEL mice were backcrossed with BALB/c mice for six generations. SWHEL mice were crossed with cd40−/− mice on C57BL/6 background (16) purchased from The Jackson Laboratory. The il-4−/− mice (17) on a C57BL/6 background were a kind gift of Dr. B. Fazekas de St. Groth (Centenary Institute, Sydney, Australia).
Flow cytometric analysis and FACS sorting
Four-color flow cytometry was performed on a dual laser FACSCalibur flow cytometer (BD Biosciences) and analyzed with FlowJo software (Tree Star). Single cell suspensions of splenocytes were prepared and aliquots of 106 cells were stained in 96-well round-bottom plates after blocking with purified anti-CD16/32 mAb (2.4G2) as described (13). HEL-binding (HEL+) BCRs were detected with 50 ng/ml HEL (Sigma-Aldrich) and revealed with purified HyHEL5 hybridoma supernatant conjugated to Alexa Fluor 647 (Molecular Probes). Donor B cells were tracked either by congenic markers anti-CD45.1-FITC or -PE (A20), or anti-CD45.2-FITC (104). In some experiments donor B cells were labeled with the intracellular and membrane dye CFSE (Molecular Probes) as described (13). Activated B cells were identified by staining with anti-CD95-biotin (Jo2), anti-T and B cell activation Ag (GL7) and peanut agglutinin (PNA)-biotin (Biomeda). Plasma cells were identified by anti-syndecan-1- (anti-CD138)-PE (281-2). Ig isotype-switched HEL+ B cells were detected with anti-IgG1-PE (A85-1) and then blocked with 10% mouse serum before staining for HEL-specific BCR with HEL followed by HyHEL5-Alexa Fluor 647. Data files consisting of 1–20 × 106 events in the live lymphocyte gate were collected for analysis on the basis of forward light scatter (FSC) and side light scatter while autofluorescent events (predominantly granular NK cells) were dumped on an empty fluorescent channel (FL3-H). To obtain sufficient data points for analysis early in the response, multiple data files containing 20 × 106 events each were collected and concatenated using FlowJo software.
MZ and Fo B cells were purified from spleens of naive SWHEL mice as follows. RBC were lysed in hypotonic NH4Cl and adherent cells depleted by incubation on a plastic tissue culture plate. T cells were depleted by immunomagnetic bead separation using anti-Thy1.2 Dynabeads (Dynal Biotech) and the remaining B cells were stained with anti-CD21/35-FITC (7G6) and anti-CD23-PE (B3B4) and propidium iodide (PI; Molecular Probes) for FACS sorting on a FACSVantage flow cytometer (BD Biosciences). CD21/35highCD23low MZ and CD21/35intCD23high Fo B cells were obtained with >98% purity and the proportion of HEL+ B cells in the sorted population determined before adoptive transfer by staining with anti-B220-PerCp (RA3-6B2) and HEL directly conjugated to Alexa Fluor 647. For ELISPOTs, spleens from recipient mice were prepared on day 5 postimmunization and sorted into live CD45.2intsyndecan-1+ and CD45.2highsyndecan-1− donor cells to >99% purity. For mutation analysis, HEL+ Fo and MZ B cells from naive (unimmunized) SWHEL mice and HEL+ CD45.2+ donor B cells from day 6 and day 7 immunized mice were sorted to >99% purity and DNA extracted from >105 cells for analysis of the IgH variable gene. All mAbs were purchased from BD Biosciences.
In vitro cultures
Sorted Fo and MZ B cells were labeled with 5 μM CFSE and stimulated in vitro with 5 μg/ml agonistic anti-CD40 mAb (HM40-3; BD Biosciences) and 10 ng/ml recombinant murine IL-4 (Sigma-Aldrich) for 4 days as described (13). Cells were harvested and analyzed by flow cytometry to determine the cell division profile.
Adoptive transfers and immunizations
HEL was covalently conjugated to SRBC with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (Sigma-Aldrich) as described (18). Either 1 or 2 × 104 CD45.2+ HEL+ B cells were adoptively transferred into 8- to 16-wk-old nonirradiated wild-type (WT) CD45.1+ congenic mice together with an inoculum of 2 or 5 × 108 HEL-SRBCs or mock-conjugated SRBC control. Sorted Fo and MZ B cells were also labeled with 5 μM CFSE to track cell proliferation in vivo (19). For these experiments, CD45.1+ SWHEL B cells were used to transfer into CD45.2+ WT C57BL/6 mice. Mice were sacrificed at the indicated time points, sera were analyzed by ELISA, and spleens were analyzed by flow cytometry, immunofluorescence microscopy, and ELISPOT assay.
Splenic sections (5–7 μm) were air-dried and fixed in acetone and stained as previously (13). HEL-specific BCRs were detected with 200 ng/ml HEL followed by polyclonal rabbit anti-HEL antisera and visualized with sheep anti-rabbit IgG-FITC (Silenus Labs). Sections were stained with anti-CD4 (GK1.5; BD Biosciences) and goat anti-rat IgG-Texas Red (Caltag Laboratories) to reveal the T cell area. Sections were then blocked with 5% rat serum before staining with rat anti-B220-biotin (RA3-6B2; BD Biosciences) and SA-FluoroBlue (Biomeda) to reveal the B cell area. PNA was biotinylated and used to detect GCs with SA-FluoroBlue, in which case B cells were stained with purified rat anti-B220 mAb and revealed with goat anti-rat IgG-Texas Red.
ELISA and ELISPOT
Serum anti-HEL Igs were measured by indirect ELISA using HyHEL10 anti-HEL Ig standards as previously described (13). Biotinylated anti-Igκ (187.1), anti-IgM (R6-60.2), and anti-IgG1 were purchased from BD Biosciences. ELISPOT for the detection of anti-HEL ASCs was based on the method described previously (20). Briefly, 96-well filter separation plates containing mixed cellulose ester membranes (Millipore) were coated overnight at 4°C with either 1 μg/ml HEL in carbonate buffer or carbonate buffer alone (background control). Plates were washed in PBS and blocked with 5% skim milk powder. Sorted HEL+CD45.2intsyndecan-1+ and HEL+CD45.2highsyndecan-1− cells were then serially diluted across the plate in RPMI 1640 medium supplemented with 10% heat-inactivated FCS (Invitrogen Life Technologies) and incubated at 37°C for 6–8 h. Plates were washed with PBS followed by 0.05% Tween 20/PBS and blocked with 1% BSA/0.05% Tween 20/PBS before overnight incubation at 4°C with biotinylated anti-IgG1 mAbs. Plates were washed in 0.05% Tween 20/PBS, and bound mAb was revealed by SA-alkaline phosphatase (Roche Diagnostics). Positive ELISPOTs were visualized with the substrate 5-bromo-4-chloro-3-imidolyl phosphate/NBT (Sigma-Aldrich) and counted by indirect light microscopy.
The VH10 IgH variable region gene was amplified from genomic DNA from >105 CD45.2+HEL+ B cells FACS sorted from mice 6 and 7 days after adoptive transfer and challenge of Fo and MZ SWHEL B cells. Isolated DNA was amplified for 33 cycles at 68°C with the polymerase Elongase (Invitrogen Life Technologies) using the following primers: GTTAAGTCTTCTGTACCTGTCGACAGC and CCCAACTTCTCTCAGCCG. The resulting fragment was purified with the QIAEX II Gel Purification kit (Qiagen), digested with SalI and NsiI, and ligated into pBluescript (Invitrogen Life Technologies). Recombinant plasmid clones were sequenced at the Australian Genome Research Facility (Brisbane, Queensland). The VH10 IgH variable gene was also amplified and sequenced from >105 sorted naive (unimmunized) Fo and MZ HEL+ B cells to control for the baseline mutation rate. Finally, the VH10 IgH variable gene was amplified from genomic brain DNA from a SWHEL mouse and sequenced to control for the background PCR error rate. Translated protein sequences were aligned with the HyHEL10 sequences numbered according to Ref. 21 .
SWHEL B cells make a robust anti-HEL IgG1 response to HEL-SRBC
To study TD responses, SWHEL splenocytes (CD45.2+) were transferred i.v. into WT CD45.1+ congenic recipients together with 2 × 108 HEL-SRBCs. SRBC was used as a carrier to provide a strong source of primary T cell help (18, 22, 23). To ensure that HEL+ B cells were not in vast excess over the available SRBC-specific T cell help, the number of transferred cells was limited so as to contain only 1 × 104 HEL+ B cells. FACS analysis of splenocytes from recipient mice allowed precise tracking of the donor cells during the response by staining for the Ag-specific BCR (HEL-binding) in conjunction with the CD45 congenic marker (CD45.2).
Seven days after transfer, extensive expansion of CD45.2+HEL+ donor B cells was evident in recipients given HEL-SRBC (Fig. 1,A). This expansion was Ag-specific because it did not occur in recipients of SWHEL splenocytes immunized with mock-conjugated SRBC (Fig. 1,A). At this point of the response, 90–100% of HEL+CD45.2+ B cells expressed the activation markers CD95, GL7, and PNA (our unpublished data), ∼60% had undergone CSR to IgG1 (Fig. 1,A), and up to 30% had differentiated into syndecan-1+ cells (Fig. 1,B). Expression of syndecan-1 was associated with increased FSC and down-regulation of surface BCR and CD45.2 expression (Fig. 1, A and B, and our unpublished data). FACS sorting followed by ELISPOT analysis showed that ASCs producing anti-HEL IgG1 resided in the CD45.2int syndecan-1+ but not in the CD45.2highsyndecan-1− compartment (Fig. 1,B). On day 7, SWHEL B cells transferred together with HEL-SRBC made a robust anti-HEL Ab response consisting predominantly of the IgG1 isotype (Fig. 1, C—E, left panel and our unpublished data). Production of anti-HEL Abs was again Ag-specific because none were detected in recipients of SWHEL splenocytes challenged with mock-conjugated SRBCs (Fig. 1, C—E, and our unpublished data). In addition, no anti-HEL Abs were detected in mice inoculated with HEL-SRBC alone indicating that they were produced entirely by the transferred HEL+ B cells (our unpublished data). This is presumably due to the small numbers and relatively low affinity of endogenous anti-HEL B cells relative to anti-SRBC B cells in the absence of transferred SWHEL B cells.
Response of SWHEL HEL+ B cells to HEL-SRBC is TD
Ab responses to SRBC are classically considered to be TD (22, 23). To determine whether this was also true for the response of SWHEL HEL+ B cells to HEL-SRBC in the current adoptive transfer system, SWHEL× BALB/c spleen cells were cotransferred with HEL-SRBC into T cell-deficient nu/nu (athymic) mice or WT syngeneic (BALB/c) controls. T cell dependency of the response was confirmed by the 100-fold lower levels of serum anti-HEL IgG1 and 105-fold fewer splenic IgG1+ HEL+ cells observed in nu/nu recipients (Fig. 1,C). The response was also shown to depend on CD40L-CD40-induced proliferation and survival because the transfer of SWHEL HEL+ B cells lacking CD40 expression into syngeneic (C57BL/6) WT recipients resulted in 16-fold lower levels of serum anti-HEL IgG1 and 610-fold fewer splenic CD45.2+ HEL+ cells (Fig. 1,D). Further, IL-4 was found to be particularly important in class switching to IgG1; thus, serum anti-HEL IgG1 levels were reduced 5-fold and the proportion of IgG1+ responding B cells was decreased 10-fold from 50% to 5% in il-4−/− recipients (Fig. 1,E). However, the proportion of responding cells that differentiated into ASCs was not significantly affected by IL-4 deficiency (Fig. 1 E).
Purification of Fo and MZ B cells by FACS sorting
To compare the relative abilities of Fo and MZ B cell subsets to mount a TD immune response, CD21/35intCD23high Fo and CD21/35highCD23low MZ B cells were purified from SWHEL mice by FACS sorting (Fig. 2,A). Using this gating strategy we predict that the Fo population would contain <5% CD21/35low transitional T2 and T3 B cells and that these cells would be absent from the MZ population (24). The purity of the sorted populations was reproducibly >98% and their phenotypes conformed to those typical of Fo and MZ B cells in terms of cell size (FSClow vs FSChigh, respectively, Fig. 2,A) and CD1d expression (CD1dlow vs CD1dhigh, respectively, our unpublished data). Purified Fo and MZ B cells were then transferred together with HEL-SRBC into nonirradiated WT CD45.1 congenic recipients such that each recipient was given 1 × 104 HEL+ B cells. Because donor B cells were deliberately not sorted on the basis of BCR specificity, recipients coincidentally received 4–9 × 104 HEL− B cells due to the fact that SWHEL mice contain a population of HEL− B cells generated by VH gene replacement (13). Thus, HEL+ B cells typically comprised 10–20% of both the Fo and MZ compartments (Fig. 2 A).
Fo and MZ B cells both make efficient anti-HEL IgG1 Ab responses
Analysis of serum Abs in recipients of the two subsets revealed that Fo and MZ SWHEL B cells produced anti-HEL IgM and IgG1 with similar kinetics over a 10-day period (Fig. 2,B). In each case, only anti-HEL IgM was detectable on day 3 (Fig. 2,B, left panel) whereas anti-HEL IgG1 predominated by day 5 and increased further until day 10 of the response (Fig. 2,B, right panel). The magnitude of the response in each case was similar with the exception of day 5, where the production of both IgM (Fig. 2,B, left panel) and IgG1 (Fig. 2 B, right panel) by MZ B cells was lower than that by Fo B cells.
MZ B cells do not expand as rapidly as Fo B cells in response to HEL-SRBC
The slight lag observed in the anti-HEL Ab response of MZ SWHEL B cells was accompanied by an even more dramatic delay in their cellular expansion in the spleens of recipient mice. Thus the number of CD45.2+ HEL+ B cells recovered from recipients of MZ vs Fo B cells was lower at day 3 (14-fold), day 5 (7-fold), and day 7 (2-fold), and did not reach parity until after the peak of the response on day 10 (Fig. 2 C).
MZ B cells switch to IgG1 and differentiate into ASCs as well as Fo B cells in response to HEL-SRBC
Despite the delayed expansion of the responding HEL+ MZ B cells, their ability to switch to IgG1 was not compromised compared with Fo B cells. Thus the proportion of IgG1+ cells within the responding CD45.2+ HEL+ B cell populations was similar at all time points regardless of the Fo or MZ origin of the donor B cells (Fig. 2,C). In contrast, the responding MZ B cells contained a higher proportion of syndecan-1+ cells than that of their Fo counterparts at days 5 and 7 of the response (Fig. 2,C). Because this population contains the splenic ASCs (Fig. 1,B), it is likely that the more efficient generation of ASCs by the responding MZ B cells explains why their Ab secreting potential is similar to that of Fo B cells (Fig. 2,B) even though their expansion lags significantly behind (Fig. 2 C).
MZ B cells are inefficiently recruited into the TD response to HEL-SRBC
To investigate the basis of the delayed expansion of MZ vs Fo HEL+ B cells, we compared the relative abilities of the two subpopulations to home to the spleen and proliferate in response to HEL-SRBC. To do this, Fo and MZ B cells were purified from the spleen of SWHEL (CD45.1+ congenic) mice, labeled with the dye CFSE (19) and transferred into WT C57BL/6 (CD45.2+) mice with either HEL-SRBC or mock-conjugated SRBC. Splenocytes were harvested from recipient mice 64 h after transfer, stained for CD45.1 and HEL binding, and analyzed by flow cytometry.
We first tested whether HEL+ B cells from the Fo and MZ compartments might home less efficiently to the spleen compared with HEL− B cells by transferring CFSE-labeled B cells into recipients given mock-conjugated SRBC. In this case the transferred CD45.1+ B cells did not proliferate, as indicated by the maintenance of high CFSE fluorescence in both the HEL+ and HEL− B cell populations (Fig. 3, A and B, left panels). Significantly, the proportion of HEL+ B cells within the transferred CFSE+ B cell populations at 64 h was the same as that within the input cell inoculum for this experiment (i.e., 12–13%; Fig. 3, A and B, left panels and our unpublished data) indicating that the homing capacity of Fo and MZ HEL+ B cells in each case reflected that of the donor population as a whole.
Despite the equivalent migration of HEL+ and HEL− B cells from each donor population, analysis of HEL-SRBC-challenged recipient mice indicated that MZ B cells migrated to the spleen less efficiently than Fo B cells. Thus the frequency of CFSE+ HEL− B cells in recipients of MZ donor B cells was 2.8-fold lower than in mice that received Fo donor B cells (Fig. 3, A and B, right panels). This disparity in homing was nevertheless insufficient on its own to explain the lesser expansion of HEL+ MZ B cells, because these were 14-fold less frequent than HEL+ Fo B cells at the same time point (Fig. 3, A and B, right panels). Thus, the overall expansion of MZ HEL+ B cells in response to HEL-SRBC was impaired up to 5-fold relative to that of Fo HEL+ B cells.
To determine the nature of this impairment, the CFSE cell division profiles of HEL+ Fo and MZ B cells were directly compared. This revealed that the HEL+ B cells recruited into the proliferative response had in fact divided to the same extent regardless of whether they originated from the Fo or MZ compartment (Fig. 3,C). Therefore, the differential expansion of the two populations must be explained by more efficient recruitment of Fo vs MZ B cells into the proliferative response. Consistent with such a conclusion, the fraction of HEL+ B cells remaining undivided at this time point was much greater in MZ than Fo HEL+ B cells (26% vs 7%, Fig. 3 C).
Purified MZ B cells remain hyperresponsive to T cell-derived stimuli in vitro
The relatively poor recruitment of purified MZ B cells into the anti-HEL response was surprising in light of previous reports that they in fact show greater responsiveness than Fo B cells to T cell-derived stimuli in vitro (2, 4). To test whether our purification strategy may have compromised the responsiveness of MZ B cells to such signals, Fo and MZ B cells were purified as detailed above, labeled with CFSE, and stimulated with anti-CD40 mAb plus IL-4 in vitro. The in vitro recruitment and subsequent proliferation of MZ B cells in response to this stimulus was greater than that of Fo B cells (Fig. 3 D) in agreement with previous studies (2, 4). In other words, the poor recruitment of MZ B cells into the TD response to HEL-SRBC in vivo was not due to an intrinsic or induced hyporesponsiveness to T cell-derived signals. On the contrary, this finding suggests that the ability of MZ B cells to access T cell help in response to HEL-SRBC in vivo is limited by physiological factors operant within an intact immune system.
MZ B cells do not home efficiently to the T-B border following activation with HEL-SRBC
After Ag engagement, unfractionated B cells have been shown to migrate to the border between the periarteriolar lymphatic sheath (PALS) and primary follicle (outer PALS) to seek T cell help and initiate TD immune responses (10, 11, 25, 26). Therefore, it is possible that the poor recruitment of MZ B cells into the anti-HEL response was due to their inefficient homing to the outer PALS. To test this hypothesis, we examined the splenic localization of responding HEL+ Fo and MZ B cells 24 h after transfer and activation with HEL-SRBC. HEL+ B cells were not detected by histological analysis in the spleens of control mice given HEL-SRBC without SWHEL donor B cells (Fig. 4,A, upper left panel). Therefore, HEL+ B cells were specifically detected only in the spleens of mice that received Fo or MZ donor SWHEL B cells as well as HEL-SRBC (Fig. 4,A, lower left panel). In the case of Fo HEL+ donor B cells, the majority (>90%) localized to the T-B border after 24 h (Fig. 4,A, lower left panel). In contrast, only a minority (<20%) of MZ HEL+ donor B cells localized to the T-B border at this point, the majority being spread throughout the red pulp (Fig. 4 A, upper and lower right panels). Thus the immediate homing of MZ B cells to the outer PALS in response to HEL-SRBC was significantly less efficient than that of Fo B cells.
Entry of activated MZ B cells into GCs is delayed compared with Fo B cells
Differential migration of mature peripheral B cell subsets was also observed later during the response to HEL-SRBC. By day 3, HEL+ B cells remained undetectable by histological analysis in the spleens of control mice given HEL-SRBC but no SWHEL B cells (Fig. 4,B, upper left panel). At this point, the frequency of HEL+ Fo donor B cells had increased, and many had migrated away from the outer PALS and into the inner regions of the primary follicle (Fig. 4,B, lower left panel). In contrast, HEL+ MZ B cells were still widely scattered around the spleen and could be found in the red pulp (Fig. 4,B, upper right panel), bridging channel and MZ as well as the T-B border (Fig. 4,B, lower right panel). By day 5, multiple GCs were established in spleens of mice given Fo B cells and these were heavily colonized with HEL+ B cells (Fig. 5,A, left panel). At the same time HEL+ MZ B cells were found mainly in the bridging channels with only a few having entered the follicle and GCs, which were much less numerous (Fig. 5,A, right panel). By day 7, clusters of ASCs with intense cytoplasmic staining for HEL were seen in recipients of both Fo (Fig. 5,B, left panel) and MZ B cells (Fig. 5,B, right panel). It was not until day 7 that significant numbers of HEL+ GC B cells were seen in mice given MZ B cells (Fig. 5,B, right panel). Hence, MZ B cells were capable of generating a GC response but this response lagged significantly behind that of Fo B cells. Interestingly the expression of classical GC-associated markers such as GL7, PNA binding, and Fas did not differ between responding Fo and MZ B cells on day 5 despite their differential localization in GCs (Fig. 5 C and data not shown). Therefore, these markers may be more representative of activated rather than GC B cells, consistent with their ability to be induced on activated B cells in vitro (27, 28).
Naive Fo and MZ HEL+ B cells from unimmunized SWHEL mice exhibit limited somatic mutation of their H chain variable region gene
We next sought to compare the ability of MZ and Fo B cells to undergo SHM. To do this, the targeted VH10 H chain variable region gene was PCR amplified and individual amplicons were cloned, sequenced, and compared with the original VH10 variable gene in the targeting construct used to generate SWHEL mice. The rate of PCR-introduced errors was established using DNA isolated from the brain of a SWHEL mouse and revealed a background rate of two mutations per 104 bp (Fig. 6,A) that was significantly lower than the mutation rates observed in resting peripheral HEL+ Fo and MZ B cells (eight and six mutations per 104 bp, respectively; Fig. 6 B). Of the 16 mutations identified in these naive HEL+ B cells, 62.5% were in hypermutation hotspots (in or adjoining an RGYW motif (29)) and 62.5% were silent mutations (our unpublished data). Thus it was possible that these mutations were not Ag-selected but had arisen via the recently identified mechanism of T cell-independent SHM that occurs during immature B cell development (30). From our limited analysis, therefore, there was no indication that this mechanism is deployed differentially during the development of Fo vs MZ B cells.
TD SHM is delayed in the response of MZ vs Fo B cells
To examine SHM induced during a TD response, CD45.2+HEL+ responding B cells were purified by electronic cell sorting from recipients of Fo and MZ SWHEL B cells on days 6 and 7 following challenge with HEL-SRBC. On day 6, the rates of SHM in the H chain variable gene from both Fo and MZ responding B cells had increased significantly above those present in naive B cells (Fig. 6, B and C), but were twice as prevalent in the responding Fo vs MZ population (58 vs 29 per 104 bp; Fig. 6,C). Moreover, 52% of the Fo clones contained multiple (>1) mutations compared with only 21% of the MZ clones (Fig. 6,C). By day 7, the mutation rate of clones derived from HEL+ Fo and MZ B cells were similar and both contained clones that had acquired multiple mutations (Fig. 6 D). Thus, in contrast to the rapid isotype switching of the response to IgG1, the onset of SHM was delayed in MZ compared with Fo B cells, in parallel with their delayed entry into GCs.
MZ and Fo B cells have different patterns of SHM
In addition to the slower onset of SHM during the response of MZ B cells, it was apparent that the mutations that accumulated in responding MZ B cells differed from those in the Fo B cells. First, the early (day 6) mutations present in the Fo but not MZ responders were heavily biased toward CDR2, in particular the contact residues S52 and S56 (Fig. 7,A). In addition, by day 7, the MZ but not the Fo responders had accumulated multiple mutations in the Y47 and especially the E46 residues within framework region (FR)2 (Fig. 7,B). All the E46 mutations observed in the day 7 MZ response occurred in independent clones because the E46 codon was mutated in four alternative ways in the six clones identified and each clone carried independent mutations in other variable region residues (Fig. 8). Thus, we not only observed delayed onset of SHM in the MZ B cells but also an alteration to the pattern of SHM.
The experimental system reported here has several advantages over previously published models used for analyzing TD B cell responses in vivo. First, the responses take place in nonirradiated WT mice, thereby allowing the normal navigation of responding lymphocytes through intact secondary lymphoid structures. Importantly, this approach avoids the potential problems associated with the use of lymphopenic recipients such as scid/scid and rag-1−/− mice to visualize donor B cells (9, 31). Moreover, it means that the B cell-specific functions of candidate immunoregulatory molecules can be interrogated by using B cells from SWHEL Ig-Tg mice crossed onto different genetically manipulated backgrounds. This is demonstrated in the experiment shown in Fig. 1,D where the requirement for B cell expression of CD40 in TD responses was confirmed with B cells derived from SWHEL× cd40−/− donor mice. Second, the response is generated from only a small number (1–2 × 104) of Ag-specific B cells, a situation that more closely reflects the frequency of such cells in the normal B cell repertoire and results in a large proportion of HEL+ B cells being recruited into the response (Fig. 3). Thus the potential artifacts associated with having a heterogeneous population of responding and nonresponding Ag-specific B cells in recipient mice are avoided. Such a situation typically occurs when large numbers (105–107) of HEL+ B cells are used (our unpublished data) or in other models in which the numbers of Ag-specific donor B cells greatly exceed the available Ag and T cell help (5, 11, 12). Third, the fact that the SWHEL B cells are capable of both CSR and SHM means that the regulation of these critical components of TD responses can be analyzed while still maintaining the advantages of using an Ig-Tg system for tracking Ag-specific cells in vivo. Indeed, because the variable gene sequences of the primary responding (HEL+) B cells are known, the process of SHM can be monitored with precision during the early phase of the response (Fig. 6) in contrast to other experimental systems (9, 32). Finally, simultaneous staining for BCR specificity (HEL-binding) and donor CD45 allotype in this model provides a unique window into the earliest stages of the B cell response by using a combination of flow cytometry (Fig. 3) and histological analysis (Fig. 4).
The many advantages of using SWHEL B cells in the adoptive transfer system described here are accompanied by some caveats which need to be considered when interpreting the data. First of all, the HyHEL10 BCR expressed by SWHEL B cells binds HEL with high affinity (18). Thus, although Fo and MZ B cells clearly make different TD responses to high affinity Ag, the same results may not be obtained in response to lower affinity Ag. Future studies using mutated forms of HEL will help to resolve this issue. Second MZ B cells, unlike Fo B cells, do not recirculate and remain largely resident in the splenic MZ (33). Indeed SWHEL MZ B cells transferred in the absence of Ag did not home efficiently to the MZ (data not shown). Therefore, while it is necessary to purify and transfer Fo and MZ B cells to compare their responses in vivo, it is possible that MZ B cells activated in this way may respond differently to those that contact Ag in situ. Another factor that requires consideration is that all HEL+ B cells from SWHEL mice express high levels of surface IgM, presumably due to the lack of cross-reactivity of their BCR with endogenous Ags (13). This phenotype is typical for MZ B cells but also for a subset of Fo B cells, because surface IgM expression on the Fo B cell compartment as a whole can vary from high to low (4). Indeed the surface IgM levels on Fo HEL+ and HEL− B cells from SWHEL mice overlap considerably (Ref. 13 and data not shown). Importantly, HEL+ SWHEL B cells are not self-reactive and express only transgene-encoded H and L chains (13), meaning that they lack the dual receptor expression that can occur on self-reactive MZ B cells (34).
Comparison of the responses of CD21/35intCD23high Fo and CD21/35highCD23low MZ B cells to HEL-SRBC confirmed several findings observed in a recent study in which TD responses were analyzed in scid/scid recipient mice (9). Thus both Fo and MZ B cells secreted IgM and IgG Abs (Fig. 2,B), participated in GC reactions (Fig. 5,B), and underwent SHM (Fig. 7). In the study by Song and Cerny (9), however, MZ but not Fo B cells produced a large amount of specific IgM Ab and rapidly generated proliferative foci early in the response, a difference we did not observe here (Fig. 2,B). Although this disparity between the two systems can be explained by differences in the Ag used and/or its route of administration, it may also have arisen due to the fact that B cells transferred into a lymphopenic environment do not face the physiological obstacles to locating T cell help associated with intact secondary lymphoid tissues. Because our data indicate that MZ B cells cannot efficiently locate and respond to T cell help under such circumstances, it is possible that the TD response of MZ B cells in a lymphopenic environment may simply reflect their known hyperresponsiveness to T cell-derived signals (as shown in Fig. 3 D and Refs. 2 and 4) rather than their ability to mediate a TD response within a normal immune system.
To explain the poor recruitment of MZ B cells into the TD response against HEL-SRBC we examined the relative performance of the Fo and MZ HEL+ B cells during the earliest stages of the response. After 24 h, before any proliferation of either subset of HEL+ B cells had occurred (our unpublished data), the migration of MZ B cells to the outer PALS was found to be less efficient than that of Fo B cells (Fig. 4 A). Migration to the outer PALS at this early stage of the response is thought to be primarily dependent on BCR-mediated up-regulation of CCR7 (10). In our model, however, no evidence for differential expression of CCR7 between Fo and MZ B cells was demonstrated early in the response (our unpublished data). Therefore, CCR7 may deliver quantitatively or qualitatively distinct signals to MZ vs Fo B cells. Alternatively, Fo and MZ B cells may migrate differently due to differential expression of other homing receptors such as sphingosine-1 phosphate receptor 1 (35). Irrespective of the precise mechanism, the impaired ability of MZ B cells to migrate to the outer PALS is likely to explain their limited recruitment into TD responses in intact secondary lymphoid tissues.
Because GC formation is not seen at day 3 of the response of either Fo or MZ B cells (Fig. 4), the early B cell proliferation and IgM secretion that occurs before this point (Figs. 2,B and 3) is not GC dependent. The fact that by day 5 responding MZ B cells have not yet entered GCs (Fig. 5,A) but have undergone CSR to IgG1 as efficiently as Fo B cells (Fig. 2,C) indicates that CSR is also GC-independent in this case. Strikingly, however, the onset of SHM was significantly delayed in MZ relative to Fo B cells (Fig. 6). Because both CSR and SHM are dependent on the inducible modifier AID (36, 37), it is unlikely that AID expression is differentially induced in responding Fo and MZ B cells. Rather, the uncoupling of these two AID-dependent events in MZ B cells suggests that some additional signal required for SHM but not CSR may be delayed in responding MZ B cells. The fact that the entry of MZ B cells into GCs lags behind Fo B cells (Fig. 5) and coincides with the onset of SHM in these cells (Figs. 6 and 7), points to a requirement for a GC-specific signal for the onset of SHM but not CSR. Such a scenario is consistent with the recent data implying that the association of AID with replication protein A is required for SHM but not CSR (38).
Overall, MZ B cells proved to be significantly less efficient than Fo B cells at mediating a TD response in vivo despite their inherent hyperresponsiveness to T cell-derived stimuli (Fig. 3,D and Refs. 2 and 4). This result is consistent with the concept of a “division of labor”, in the sense that Fo B cells are specialized to deal with TD Ag and MZ B cells with TI-2 Ag (7). Interestingly, we observed that responding MZ B cells exhibited a distinct pattern of SHM compared with Fo B cells (Fig. 7). In particular, clones derived from MZ B cells mutated E46 and Y47 in FR2 just proximal to CDR2 much more frequently than Fo B cells. This unconventional mutation pattern may conceivably diversify the BCR in such a way as to generate alternative specificities to those induced by conventional targeting of CDRs. Such a difference may arise either due to differential recruitment of cofactors to the AID-mutator complex (39, 40) or to differential antigenic selection of these cells by their milieu (41). Either way, these data point to the possibility that TD responses from MZ B cells may contribute specificities complementary to those generated from the predominant Fo response. Thus although Fo B cells may be programmed to respond more efficiently to TD Ag, there may be circumstances when their response is inadequate to deal with the pathogen. In such a situation, a delayed synergistic response by MZ B cells generated via an alternative pathway may prove useful in eliminating that pathogen.
The authors have no financial conflict of interest.
We thank Dr. Jenny Kingham, Rachel Nowell, and the Centenary Institute Animal Facility for animal husbandry, Chris Brownlee for mouse screening, and Tara McDonald for FACS sorting. We also thank Dr. Barbara Fazekas de St. Groth for generously providing the il-4−/− mice and Drs. Stuart Tangye, Didrik Paus, Adrian Grech, and Cindy Ma for their review of the manuscript.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by a program grant from the National Health and Medical Research Council of Australia. T.G.P. received a National Health and Medical Research Council postgraduate research scholarship.
Abbreviations used in this paper: MZ, marginal zone; AID, activation-induced cytidine deaminase; ASC, Ab secreting cell; CSR, class switch recombination; Fo, follicular; FR, framework region; FSC, forward light scatter; GC, germinal center; HEL, hen egg lysozyme; PALS, periarteriolar lymphatic sheath; PI, propidium iodide; PNA, peanut agglutinin; SHM, somatic hypermutation; SRBC, sheep RBC; TD, T cell-dependent; Tg, transgenic, TI-2; thymus-independent type 2; WT, wild type.