Article
A Temporal Switch in the Germinal Center
Determines Differential Output of Memory B and
Plasma Cells
Highlights
d
Long-lived memory B cells are made predominantly in the
early germinal center (GC)
d
Long-lived bone-marrow-resident plasma cells are made
very late in the GC response
d
d
Authors
Florian J. Weisel,
Griselda V. Zuccarino-Catania,
Maria Chikina, Mark J. Shlomchik
Correspondence
Subsets of memory B cells are also produced in a temporal
order
[email protected]
A substantial fraction of long-lived IgM memory B cells are
made prior to GC onset
Generation of memory B and plasma cells
is critical for protective immunity.
Shlomchik and colleagues demonstrate
that the B cell response shifts
dramatically from generation of memory
B cells at early time points to that of longlived, bone-marrow-resident plasma cells
at late time points.
In Brief
Accession Numbers
GSE76502
Weisel et al., 2016, Immunity 44, 116–130
January 19, 2016 ª2016 Elsevier Inc.
http://dx.doi.org/10.1016/j.immuni.2015.12.004
Immunity
Article
A Temporal Switch in the Germinal Center
Determines Differential Output
of Memory B and Plasma Cells
Florian J. Weisel,1,2 Griselda V. Zuccarino-Catania,1 Maria Chikina,3 and Mark J. Shlomchik1,2,4,*
1Department
of Laboratory Medicine, Yale University School of Medicine, New Haven, CT 06520, USA
of Immunology, University of Pittsburgh School of Medicine, Pittsburgh, PA 15261, USA
3Department of Computational and Systems Biology, University of Pittsburgh School of Medicine, Pittsburgh, PA 15260, USA
4Department of Immunobiology, Yale University School of Medicine, New Haven, CT 06520, USA
*Correspondence: [email protected]
http://dx.doi.org/10.1016/j.immuni.2015.12.004
2Department
SUMMARY
There is little insight into or agreement about the
signals that control differentiation of memory B cells
(MBCs) and long-lived plasma cells (LLPCs). By performing BrdU pulse-labeling studies, we found that
MBC formation preceded the formation of LLPCs
in an adoptive transfer immunization system, which
allowed for a synchronized Ag-specific response
with homogeneous Ag-receptor, yet at natural precursor frequencies. We confirmed these observations in wild-type (WT) mice and extended them
with germinal center (GC) disruption experiments
and variable region gene sequencing. We thus
show that the GC response undergoes a temporal
switch in its output as it matures, revealing that
the reaction engenders both MBC subsets with
different immune effector function and, ultimately,
LLPCs at largely separate points in time. These
data demonstrate the kinetics of the formation of
the cells that provide stable humoral immunity and
therefore have implications for autoimmunity, for
vaccine development, and for understanding longterm pathogen resistance.
INTRODUCTION
Adaptive T-cell-dependent (TD) immune responses against
foreign antigens (Ag) first generate plasmablasts (PBs), followed by a germinal center (GC) response that engenders
both memory B cells (MBCs) and long-lived plasma cells
(LLPCs) (De Silva and Klein, 2015; Nutt et al., 2015; Shlomchik
and Weisel, 2012; Victora and Nussenzweig, 2012; Zotos and
Tarlinton, 2012). During the extrafollicular response, activated
B cells move to the splenic red pulp, proliferate, and differentiate, which leads to an early wave of short-lived PBs and
mainly unmutated and unswitched Ag-experienced B cells
that resemble MBCs, though whether these cells truly join the
long-lived compartments has not been resolved (Blink et al.,
2005; Inamine et al., 2005; Kaji et al., 2012; Obukhanych and
116 Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc.
Nussenzweig, 2006; Taylor et al., 2012). Concurrently, some
activated B cells undergo productive interaction with cognate
T cells at the splenic T cell-B cell border or the interfollicular region of the lymph node, express the transcriptional repressor
Bcl6, and migrate into the follicle to form a nascent GC (Kitano
et al., 2011; Schwickert et al., 2011). GCs, initiating 3–4 days
after immunization (Kerfoot et al., 2011; and see below), are
sites where B cells undergo proliferation, V region gene somatic
hypermutation, and class switch recombination (Jacob et al.,
1991). In the GC, Ag-driven affinity maturation preferentially selects for the expansion of B cells with higher affinity for the
cognate Ag (Le et al., 2008; Shih et al., 2002) possibly by
competition for T cell signals (Gitlin et al., 2015; Liu et al.,
2015), ultimately resulting in differentiation of mutated, higheraffinity MBCs and LLPCs.
Although it is widely accepted that bone marrow (BM) LLPCs
are derived from GCs (Chan and Brink, 2012; Good-Jacobson
and Shlomchik, 2010; Oracki et al., 2010; Takahashi et al.,
1998), MBCs can also form in Bcl6/ animals (Toyama et al.,
2002), which lack GCs, or in response to T-cell-independent
Ag (Obukhanych and Nussenzweig, 2006). Further, not all
MBCs are of switched isotype (Dogan et al., 2009; ZuccarinoCatania et al., 2014) or display somatically mutated immunoglobulin (Ig) genes (Takahashi et al., 2001; Zuccarino-Catania
et al., 2014; Tomayko et al., 2010). Although later-appearing
isotype-switched MBCs are products of the GC, early-appearing cells with a MBC phenotype were posited to originate independently of GCs (Kaji et al., 2012; Pape et al., 2011; Takemori
et al., 2014; Taylor et al., 2012). The contribution of each
pathway to the total long-lived MBC compartment remains
enigmatic.
It is unclear how differentiation of GC B cells (GCBCs) into
MBCs and LLPCs is controlled (Allen et al., 2007a; De Silva
and Klein, 2015; Nutt et al., 2015; Shlomchik and Weisel, 2012;
Victora and Nussenzweig, 2012; Zotos and Tarlinton, 2012).
Based largely on in vitro data, it has been proposed that
CD40-mediated signaling and/or cytokine signals could control
this decision, but there has not been agreement on whether
such signals promote MBC versus LLPC formation. It was
originally thought that differentiation could be controlled via
affinity-based instructive B cell receptor (BCR) signals (Paus
et al., 2006; Phan et al., 2005), but subsequently the same
group elegantly showed that higher affinity increased overall
A
B
C
D
E
F
G
H
I
J
K
L
(legend on next page)
Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc. 117
proliferation—not antibody-forming cell (AFC) differentiation—of
Ag-reactive cells, which in turn resulted in increased AFC
numbers (Chan et al., 2009). Alternatively, differentiation might
be a stochastic process, possibly metered by the number of
cell divisions and/or signaling encounters (Hasbold et al.,
2004). Given the many theories the resolution of this question
remains a major unresolved question.
Some clues to the in vivo control of this process come from
antibody (Ab) inhibition and genetic deletion studies. Blocking
GCs with antibodies directed against CD40L or ICOSL results
in a decrease of LLPCs (Takahashi et al., 1998) and deletion of
CR1 and CR2 (Gatto et al., 2005), interleukin 21 receptor (IL21R) (Linterman et al., 2010; Zotos et al., 2010), PD-1, PD-L1,
and PD-L2 (Good-Jacobson et al., 2010), and CD80 (GoodJacobson et al., 2012) allow GC initiation but affect proper GC
maturation or progression. In all these cases, the loss of the
late GC is correlated with diminished LLPC numbers, whereas
MBC populations are mainly unaffected or even increased (reviewed in Good-Jacobson and Shlomchik, 2010). It could be
that each of these signals differentially promotes LLPC versus
MBC formation. Alternatively, it might be the case that these signals allow the GC reaction to reach a certain maturation point
that favors LLPC generation.
To determine whether MBCs and LLPCs are generated at
different time points during the response, here we used BrdU
pulse labeling, an approach already successfully used to analyze
the half-life of Ag-specific PCs (Manz et al., 1997) and the lifespan of MBCs (Schittek and Rajewsky, 1990). We observed
that long-lived immune progeny are generated in a sequential
order: unswitched MBCs very early in the response, followed
by switched MBCs and finally by a delayed appearance of isotype-switched BM LLPCs. We corroborated these findings using
a combination of anti-CD40L Ab to destroy the GC at a key time
point, as well as V region gene sequencing to match the content
of early GCs with MBCs and late GCs with LLPCs. Based on
these findings, we infer that less-committed humoral immune
effector cells mainly derive from pre- or early GC reactions
whereas cells of higher maturation stage are formed during
late GC responses and propose a model that selection of proliferating GCBCs into the long-lived immune compartment is
controlled by developmental stages within the GC reaction, resulting in a switch of output over time.
RESULTS
Kinetics of the Formation of Long-Lived Immune
Effector Cells in a Synchronized Response
To determine when long-lived MBCs and LLPCs are stably
formed, we induced a TD immune response in a transfer-immunization system (Figure 1A), which allowed for a synchronized
response of a timed cohort of B cells with controlled BCR composition, and performed BrdU pulse labeling at different stages of
the immune response. To this end, we transferred limiting
numbers of 4-hydroxy-3-nitrophenyl acetyl (NP)-reactive naive
B cells from B1-8i+/ BALB/cJ genetically targeted mice (henceforth referred to as B1-8 mice) into AM14 transgenic Vk8R genetically targeted BALB/cJ mice (henceforth referred to as transfer
recipients), which have a fixed BCR that does not recognize
NP-chicken gamma globulin (CGG) (Figure 1A). Upon immunization of recipient mice, WT-like numbers of 12,500 splenic NPreactive B cells (Figure S1; Le et al., 2008; Pape et al., 2011; Tomayko et al., 2010) participated in the response quickly, without
continuous input of naive B cells into the primary response
(Schwickert et al., 2009), thereby synchronizing the GC reaction.
Transfer recipients were immunized with NP-CGG and divided
into 14 groups, each of which received three BrdU i.p. injections
per day, with each group being injected for a different sequential
3-day period (Figure 1A); we refer to these periods as ‘‘labeling
windows.’’ All mice were then analyzed at 8 weeks after immunization. This approach would label long-lived cells at the time
point of their generation, because only cells that were in S-phase
during the BrdU labeling window will have integrated BrdU into
the nascent DNA within the first 1–2 hr after BrdU injection and
only cells that stopped cycling during or immediately after the
period of bioavailable BrdU would remain detectably BrdU
labeled (Figure 1B) for at least 20 weeks (Anderson et al.,
2006, and see below). 8 weeks after immunization, we enumerated splenic MBCs, MBC subsets, and BM LLPCs by multicolor
flow cytometry (Figure S2); we posited that BrdU+ MBCs or
LLPCs identified in mice from a particular labeling window
were stably formed during that time window of labeling (further
discussed below). Thus, the percentage of BrdU+ cells among
either MBCs, MBC subsets, or LLPCs in a given window (Figures
1A, 1B, and S2) indicates the fraction of the total compartment
that was formed during that window.
Figure 1. Kinetics of the Formation of Long-Lived Immune Effector Cells in a Synchronized Response
(A and B) Experimental outline (A) and possible labeling scenarios (B).
(C, E, G, and I–L) Transfer recipients were injected with BrdU at indicated times after NP-CGG immunization, as depicted in (A).
(C, E, and G) The frequencies of BrdU+ of live (=EMA) IgM+ MBCs (C), IgG1+ MBCs (E), and BM IgG1+ LLPCs (G) were analyzed 8 weeks after immunization by
flow cytometry (Figures S2A and S2B). Each dot represents one mouse; combined data are from 3–6 experiments.
(I) Flow cytometric analysis of CD80 and PD-L2 distribution among EMABrdU+NP+CD19+ splenic MBCs at week 8 (n = 6–19, symbols are mean ± SEM;
combined data of three experiments; Figure S2C).
(J) Frequency of CD73+ of EMANP+CD19+ cells in transfer recipients 8 weeks after immunization.
(K) Frequency of CD73+ of EMABrdU+NP+CD19+IgG1+ cells for each BrdU labeling window.
(L) Frequency of BrdU+ (IgM+ squares, IgG1+ circles) of EMACD73+NP+CD19+ cells for each BrdU labeling window.
n = 6–14, symbols are mean ± SEM in (K) and (L).
(D, F, and H) Recipients were treated as depicted in (A) and sacrificed 1 hr after the last BrdU injection at the indicated day. Symbols are mean ± SEM of 3–4 mice
per group (combined data of two experiments are shown).
(D) Phenotype of B cells labeled during each BrdU labeling window.
(F) Frequencies of EMANP+CD19+ cells with GC (CD38CD95+, circles) or non-GC (CD38+CD95, squares; CD38+CD95IgG1+, triangles) phenotype over the
time course of the immune response.
(H) Kinetics of the AFC response. IgG1+ AFCs from spleen (circles) and BM (squares) of mice depicted in (C) and (E) were measured by ELISpot assay.
118 Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc.
Overall, we found that Ag-specific cells that were labeled during the early part of the response and even beyond the peak of
the GC were predominantly MBCs, with very few LLPCs that
were destined to remain in the BM until at least week 8, whereas
cells labeled during the late, post-peak of the GC response were
overwhelmingly LLPCs. There was a pronounced pre-GC wave
of IgM+ MBC formation (Figures 1C and S2A), with an initial
peak during d0–d2 and 40% of IgM+ MBCs formed by d5.
Peak formation of IgG1+ MBCs was between d6 and d8 after immunization, with 22% of all MBCs generated in this 3-day window (Figures 1E and S2A) and about half of all IgG1+ MBC were
generated by d11, which was before the peak of the total GC
response (d14, Figure 1F). Notably, 9% of IgG1+ MBCs were
formed by d5, indicative of a pre-GC origin (Figures 1E and 1F)
and consistent with the early appearance of IgG1+ B cells with
a memory-like phenotype very early after immunization (Figure 1E; Inamine et al., 2005). These data demonstrate that
such IgG1+ MBCs that formed very early in the response persist
for at least 8 weeks (Figure 1E).
To better understand whether MBCs that formed at different
time points emanated from a GC or extra-GC source, we
measured the fraction of NP+IgG1+ B cells labeled during each
BrdU window by sacrificing transfer-immunized recipient mice
1 hr after the last BrdU injection instead of 8 weeks later. From
d8 to d38, the vast majority of labeled cells had a GC phenotype
(Figure 1D), implying that long-lived BrdU+ cells formed after d6
originated mainly from the GC. In contrast, from d0 to d2, there
were no detectable labeled NP+IgG1+ cells of a GC phenotype
(Figure 1D), implying that resting cells formed during that time
window and most likely a substantial fraction formed in the
next time window were indeed GC-independent long-lived
MBCs (Figures 1C–1F and further discussed below). The percentage of BrdU+NP+ GCBCs was not different among individual
labeling windows, indicating equivalent labeling throughout the
whole GC reaction (not depicted).
In contrast to the early formation of MBCs, only 7.5% of the
IgG1+ BM LLPC compartment was formed by d14 (Figures 1G
and S2B). Instead, the majority was generated much later—between week 2 and week 5 after immunization (Figure 1G).
Consistent with these conclusions, the delayed development
of LLPCs was reflected in the kinetics of the appearance of total
AFC numbers in the BM (Figure 1H). Indeed, even though labeling windows extended to 6 weeks after immunization, we could
account for only 75% of the total LLPCs in summing up all of
the labeled cells, indicative of substantial LLPC generation
even 5 weeks after immunization (Figure 1G), whereas we could
similarly account for 95% of MBCs, indicating little if any further
generation beyond week 6 (Figures 1C and 1E). Furthermore, labeling of 95% of MBCs of both isotypes throughout the course
shows that BrdU pulse labeling was highly efficient. Additionally,
we did not find differences in the frequency of NP-specific B cells
in BrdU- versus PBS-treated mice at week 8 (not depicted),
which confirms that there was no toxic effect of BrdU.
Heterogeneity among MBCs Is Reflected in the Kinetics
of Their Generation
We previously showed that the NP-specific MBC compartment
is comprised of at least three phenotypically and functionally
distinct subsets, which can be distinguished by expression of
CD80 and PD-L2 (Tomayko et al., 2010; Zuccarino-Catania
et al., 2014). To assess whether there are distinct kinetics of formation of these distinct MBC subsets, we included staining for
these two markers in our Ag-specific BrdU analysis of MBCs.
Up to d5, mainly CD80PDL2 MBCs were produced. Between
d6 and d11, the most abundantly produced MBC subset was of
CD80PDL2+ phenotype, whereas among the MBCs formed
from d12 onward, most were CD80+PDL2+ (Figures 1I and
S2C). These data imply that qualitatively distinct effector functions (Zuccarino-Catania et al., 2014) of MBCs are reflected in
the kinetics of their generation. This result is in line with findings
from Kaji et al. (2012), who reported that the secondary immune
response of adoptively transferred d40 MBCs led to 5–6 times
more AFCs in recipient mice compared to adoptively transferred
d7 MBCs.
CD73 is expressed on a subset of MBCs (Figure 1J; Tomayko
et al., 2010), and it was suggested that CD73 expression marks
GC-derived MBCs (Taylor et al., 2012). BrdU pulse labeling revealed, however, that among the IgG1+ MBCs formed between
d0–d2 and d3–d5, about 60% were CD73+ (Figure 1K), indicating
that MBCs formed prior to the GC reaction often express CD73.
Similarly, among CD73+ MBCs, approximately 37% of IgM+ cells
and 6% of IgG+ cells formed by d5, a time point prior to substantial GC formation (Figures 1F and 1L). Thus, CD73 expression is
neither exclusive to (Figure 1K) nor specific for (Figure 1L) GCderived MBCs, even though somatically mutated MBCs are
enriched in CD73+ subset (Kaji et al., 2012).
Validation of Experimental Design
The adoptive transfer system provided sufficient numbers of
cells to enable us to validate the assumptions underlying the
short-term BrdU ‘‘window-labeling’’ approach (Figure 1A) we
used to address long-term precursor-product relationships (Figure 1B). To validate the assumption that BrdU labeling of MBCs
would become undetectable after further division, we injected
transfer recipient mice from d6–d8 after immunization with
BrdU 3 times daily, and 6 weeks later harvested and labeled
splenocytes with VPD450. We cultured these cells with
ODN1826 to stimulate division, then used VPD dilution to correlate BrdU staining and cell division by flow cytometry (Figure 2A).
Most (65%) cells had lost BrdU label by the first division and by
the third division 92% of previously labeled cells had lost detectable BrdU (Figure 2B). Because GCBCs divide 3–4 times per
day, even if it took 3 divisions to lose labeling, this still gave a resolution of a single day or less for the labeling windows, which is
as postulated in the experimental design (Figures 1A and 1B). We
further performed BrdU and 5-ethynyl-20 -deoxyuridine (EdU)
in vivo double labeling experiments to measure how many divisions a GCBC underwent in vivo from its last division in which
it was BrdU labeled as a GCBC until achieving a resting MBC
state. To evaluate this, we injected transfer recipient mice with
BrdU for 1 day and then 1–4 days later with a single dose of
EdU and harvested spleens 30 min later (Figure 2C). We found
that by d1 only 4% and by d2 only 1% of BrdU+ cells with
MBC phenotype were labeled with EdU (Figures 2D and 2E).
Taking into account that 1/4–1/3 of cycling GCBCs are in
S phase at a given time (Allen et al., 2007b), this indicates that
90% of labeled cells exited the cell cycle within 1 day
and >95% within 2 days (Figure 2E). Considering that as many
Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc. 119
as 65% (Figure 2B) of dividing cells could also have lost their
BrdU, one might correct this further by a factor of three; however,
such cells would have been captured in the non-GC BrdU and
EdU+ gate (Figure 2D, upper left plot, lower right quadrant),
which contains an even fewer number of cells. Hence, the
maximal correction that could be applied would indicate that
80% of labeled cells exited cycle within 1 day and >90% within
2 days. In addition to validating our experimental design, this
observation provides insight into the nature of GCBC commitment to memory.
Micro-anatomic Locations of the Formation of
Early MBCs
Given that MBCs are formed very early in the immune response
(Figures 1C and 1E) and that these undoubtedly stem from proliferating precursors responding to Ag stimulation (Figure S2A), we
assessed the micro-anatomic location where active proliferation
of Ag-specific B cells takes place at early time points. We injected transfer recipients at d2–d6 after immunization three
times with EdU to label proliferating cells. Mice were sacrificed
30 min after the last EdU injection and at each time point we
found that most of the lambda light chain (Igl) positive (and
thus Ag-specific) B cells were proliferating inside the T cell zones
and at the T cell-B cell border. Compared to T cell zones and the
T cell-B cell border, fewer Igl+ B cells were proliferating within
B cell follicles at d2 of the response (Figures 3A and 3C).
Because 20% of the IgM+ MBCs are formed by d2 (Figure 1C),
and the majority of proliferating Ag-specific B cells—containing
presumptive MBC precursors—are located in the T cell zones,
it is very likely that most early MBCs were generated inside
T cell areas. Proliferation of Ag-specific cells within T cell zones
was found throughout all analyzed time points, suggesting potential continued generation of extra-GC MBCs (Figure 3C) up
to at least d6. On d3, small GCs were detectable, indicating
that from this point on the GC could contribute to long-lived compartments, although that the majority of proliferating cells at d3
was outside the GC. From d4 onward, the majority of proliferating Igl+ B cells was found within GCs (Figures 3B, 3C, and
S3; Le et al., 2008).
ated with a peak at around d18–d20 (Figures 4A and S4A). About
3.5% of IgG1+ and 6.5% of IgM+ MBCs were stably formed by d5
(Figure 4), which demonstrated that previously observed earlyappearing MBC-phenotype cells (Takemori et al., 2014) indeed
enter the long-lived immune compartment in a polyclonal situation. Furthermore, IgG1+ LLPCs in the BM appeared substantially later than MBCs (Figures 4A and S4B), with only 15% of
the total LLPCs being formed by d20 whereas 40% of MBCs
had formed by this point (Figure 4B). It took fully 2 weeks longer
for LLPC formation to reach this level.
Although the key findings in adoptive transfer recipients (Figures 1C, 1E, and 1G) were confirmed in WT mice (Figure 4A),
generation of all long-lived immune compartments was shifted
toward later time points in the intact system and peaks were
less sharp compared to the transfer-immunization system.
As noted, these differences in kinetics are expected, because
effects observed in WT mice, although substantial, would be
minimal estimates of cell-intrinsic time-dependent alterations
in B cell differentiation during the response. This is because in
intact mice, responses are asynchronous, being replenished
with new cohorts of activated cells that join the pre-existing
response (Schwickert et al., 2009). The lack of synchrony obscures the ability to observe time-dependent differences in
MBC versus LLPC formation. Further, the polyclonal immune
system with a wide range of BCR affinities is homeostatic and
selection for different-affinity Ag-receptors would confound the
ability to discern inherent effects at the single-cell level, because
higher-affinity cells probably proliferate more (Gitlin et al., 2015;
Liu et al., 2015). Additionally, transferred B1-8 B cells did not
need to reach a minimum BCR affinity for NP in order to be
detectable as Ag-binding cells in flow-cytometric analysis, and
therefore the frequency of Ag-reactive cells is not underestimated, whereas in a polyclonal system, low-affinity but relevant
B cells would not bind sufficiently to fluorescent Ag for their
detection (Figures S1A and S4A), e.g., NP-reactive B cells using
the V23 heavy chain paired with l1 light chain (Dal Porto et al.,
2002). Nonetheless, even without the ability to control these variables, clear separation in the order of generation of MBCs and
LLPCs with time can be observed in the polyclonal system.
MBCs and LLPCs Are Formed with Different Kinetics in
WT Mice
Using the insights gained from validating and characterizing the
cell-transfer BrdU labeling system, we sought to track the
kinetics of B cell differentiation into MBCs and LLPCs in a polyclonal response. There are, however, limitations to this experiment: a polyclonal response is not synchronized and in addition
the affinities of stimulated cells and hence their proliferation rates
will vary over a range. We therefore would not expect the same
sharp delineations seen when these two important factors are
controlled as in the transfer system. Nonetheless, we felt it was
of interest to apply the BrdU ‘‘window labeling’’ approach to
NP-CGG-immunized BALB/cJ WT mice, similar to the protocol
used to evaluate adoptive transfer recipients (Figure 1A).
We found that even in this less temporally and affinitycontrolled setting, IgM+ MBCs were formed early during the immune response, with peak formation of 9.1% of the total
compartment between d6 and d8 followed by a steady decline
in the rate of formation. IgG1+ MBCs were subsequently gener-
Disruption of Peak GC Reactions Diminishes LLPCs but
Not MBCs
If our interpretation of the labeling data is correct, then if we were
to abrogate the GC reaction after most of the MBCs were predicted to be formed, we expect this would have a differential
and disproportional effect on LLPCs versus MBCs. To experimentally test this, we used anti-CD40L Ab to disrupt the peak
GC reaction in the transfer-immunization system at d12–d14
(Figure 1F), by which time BrdU data indicate that most MBCs
should have formed but most LLPCs would have yet to be
formed (Figures 1C, 1E, and 1G). As predicted by the BrdU
data, disruption of the GC at d12–d14 led to a significant reduction in numbers of NP-specific IgG1+ BM AFCs at 8 weeks after
immunization (Figure 5A), with a marked—almost 90%—loss of
higher-affinity AFCs (Figure 5B). In contrast, but also as predicted by the labeling data, total numbers of residual NP-specific
B cells, which represent the MBC compartment, were not significantly affected (Figure 5C). Hence, as determined independently by experimental abrogation of the immune response,
120 Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc.
B
A
C
D
E
(legend on next page)
Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc. 121
most MBCs are formed by d12–d14 whereas most LLPCs had
yet to form. In concert with our finding that CD80+PDL2+
MBCs are mainly generated late in the immune response (Figure 1I), we found these cells to be the only subset whose frequency was reduced by anti-CD40L Ab treatment (Figure 5F).
In contrast, the frequencies of CD80PDL2 (Figure 5D) and
CD80PDL2+ (Figure 5E) subsets were slightly but significantly
elevated, which was an expected result because the overall
number of MBCs (Figure 5C) remained unaltered.
Mutational Content Analysis Correlates GCBCs with
Their Long-Lived Progeny over Time
Mutations accumulate in the V gene regions of GCBCs as a function of time or cell divisions (Takahashi et al., 2001), and the
extent of mutations can be used as a footprint to compare populations of B cells that have undergone mutation. In particular,
non-mutating resting populations including both MBCs and
LLPCs should have numbers of mutations similar to those in
the GCBCs that were their direct precursors at the time that
those GCBCs differentiated. We used this as a third approach
to connect GC time windows to the types of long-lived B cell
products that were created by them. To this end, we compared
the V region mutational content of Ag-specific IgG1+ GC (and
non-GC) cells in B1-8 mice at d6–d8 and d18–d20 after NPCGG immunization to the mutational content of IgG1+ MBCs
and LLPCs formed at those time periods. We marked the latter
cells during their formation in analogous fashion to the experiments in Figures 1 and 4, except we used EdU because this
was compatible with sequencing of the labeled sorted cells (Figure 6A; Gitlin et al., 2015).
We sorted NP-specific IgG1+ GC and non-GC cells directly at
d7 (the midpoint of the d6–d8 labeling window) and d19 after
immunization and sorted NP+IgG1+EdU+ splenic MBCs and
BM LLPCs at week 8 from mice that had been given EdU
at either d6–d8 or d18–d20 (Figure 6A). The frequencies of
resulting EdU+ MBCs and BM LLPCs were comparable to frequencies obtained with BrdU labeling, indicating similar labeling efficiencies of EdU and BrdU. We then cloned and
sequenced Vl1 rearrangements from all of these sort-purified
cell populations. The Vl1 mutation frequency of EdU+ MBCs
labeled during d6–d8 matched that of d6–d8 GCBCs,
but neither that of non-GC B cells isolated at d7 nor that of
GCBCs harvested at d18–d20 after immunization (Figure 6B).
Conversely, the mutation frequency of EdU+ LLPCs formed
during d18–d20 matched only that of d19 GCBCs. Furthermore, the mutational content of d7 GCBCs and MBCs derived
at this time was indistinguishable from that of total MBCs at
week 8, most likely because the majority of MBCs were
made near this time window (Figures 1 and 4). These data
thus independently support an earlier origin for most MBCs
compared to the majority of LLPCs.
Gene Expression Analysis of Early and Late GCBCs
To gain insight in the mechanisms underlying this temporal switch
in output of GC reactions, we compared RNA expression profiles
of d8 and d18 GCBCs sorted from adoptive transfer recipients.
These data revealed consistent differences with a fold change
of 1.7 or greater in 100 genes, with q value of <0.05 (Figure 7).
Notably, a number of these are annotated as signaling molecules
and/or receptors that are known to be important in immune responses. In addition, there were several transcription factors
and chromatin modifiers that were differentially expressed in
early versus late GCs. Although extensive testing will be needed
to determine the functions of these molecules in determining the
different outputs of early versus late GCs, these data establish
clearly that the two types of GCs are not just functionally but
are indeed also transcriptionally distinct (Figure 7 and below).
DISCUSSION
Our results provide insight into the longstanding question of how
the GC generates both MBCs and LLPCs, revealing that there is
a time-dependent developmental switch in the output of the
GCs, such that it first dedicates itself to MBC generation and
later—long after its peak in size—switches to mainly producing
LLPCs. These conclusions were reached by a combination of
three strategies: pulse chase in vivo BrdU labeling, V region
sequencing, and selective elimination of the GC response at
d14 that resulted in marked reduction of LLPCs but not MBCs.
It had not been evident that the ‘‘mature’’ GC changes functionally in such a profound way over time. Previous studies might
have missed this conclusion because their underlying hypotheses posited that there would be discrete instructional signaling
that would govern the concurrent production of MBCs and
LLPCs at the local cell level. Nonetheless, consistent with our results, various mutations in GC-related molecules result in a
Figure 2. Validation of Experimental Design
(A and B) BrdU positivity is quickly lost in B cells undergoing further cell divisions in the absence of bioavailable BrdU. BrdU or PBS was i.p. injected in transfer
recipients (Figure S1) three times a day at d6–d8 after NP-CGG immunization and splenocytes were harvested 6 weeks later. Cells were labeled with the violet
proliferation dye (VPD450) and cultured in the presence of CpG ODN1826 to induce polyclonal in vitro proliferation. At d3 and d4, cultured cells were harvested
and BrdU fluorescence of EMANP+ B cells was correlated with their VPD450 fluorescence indicating distinct cell divisions.
(A) Strategy to gate live B220+NP+ B cells (left) to further analyze their BrdU and VPD450 fluorescence at d4 after in vitro activation (right).
(B) Quantification of flow cytometry data depicted in right panel of (A). The percentage of EMA NP-reactive B cells, which retained BrdU positivity, is shown for
each generation. The results were normalized to 100% for the generation 0 (symbols are mean ± SEM; combined data of d3 and d4 after in vitro activation).
(C–E) The majority of GCBCs do not undergo further cell division from their initial commitment to the MBC pool until reaching a final resting state.
(C) Experimental outline for (D) and (E). BrdU was i.p. injected three times into transfer recipients exclusively at d7 after NP-CGG immunization. Groups of five
mice received one injection of 1 mg EdU i.p. 1, 2, 3, or 4 days later to mark previously BrdU-labeled cells that were still in S phase of the cell cycle.
(D and E) Validation and analysis of EdU and BrdU double detection assay (D). Mice were sacrificed 30 min after EdU injection on d+1 and single-cell suspensions
of red-blood-cell-depleted splenocytes were subjected to flow-cytometric double detection of EdU and BrdU. EMA NP-reactive GC (CD38CD95+) and nonGC (CD38+CD95) B cells were identified (top) to assess their BrdU and EdU fluorescence (middle). Mice given PBS instead of BrdU and/or EdU served as
controls. The frequency of EdU+ of BrdU+NP+IgG1+B220+CD38+CD95 cells was assessed (bottom) and is quantified in (E). Arrows indicate subsequent gating of
populations and numbers next to outlined areas indicate percent gated population.
122 Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc.
A
B
C
Figure 3. Immunohistological Analysis of Early MBC Formation
(A and B) Transfer recipients were injected three times with EdU at d2 (A) and d6 (B) after NP-CGG immunization to label proliferating cells. Spleens were
harvested 30 min later and sections were stained for B cells (B220, green), T cells (CD4, light blue), proliferation (EdU uptake, red nuclei), and NP specificity (Igl,
dark blue). Areas of active proliferation of Ag-specific B cells are marked with color-coded rectangles in the overview of representative areas (left) and are
magnified (right) without depicting B220 and CD4 staining.
(C) Quantification of micro-anatomic location of proliferating Igl+ B cells from three individual whole reconstructed splenic sections of two mice per time point
(Figure S3A). Error bars represent ± SEM. GC phenotype was confirmed by PNA positivity on consecutive sections (Figure S3B). No significant expansion of
EdU+Igl+ B cells was observed in mice not given EdU injection, NP-CGG immunization, or without B1-8 B cell transfer (Figure S3C). Scale bars represent 200 mm.
significant loss of LLPCs but not MBCs (reviewed in GoodJacobson and Shlomchik, 2010), including CR1 and CR2 (Gatto
et al., 2005), IL-21R (Linterman et al., 2010; Zotos et al., 2010),
PD-1, PD-L1, and PD-L2 (Good-Jacobson et al., 2010), and
CD80 (Good-Jacobson et al., 2012). Because these mutants
also all share the phenotype of premature termination of the
GC response, we favor the notion that multiple signals are
required to allow the GC to mature to a certain point at which
the milieu becomes favorable for LLPC generation (Shlomchik
and Weisel, 2012). Notably, CD40 signals, Ag affinity, and certain
cytokine signals have all been investigated, without clear results
to implicate them directly in an instructive program for either
MBC or LLPC differentiation (Chan and Brink, 2012). Of note
with respect to other controversies surrounding GC function,
because formation of MBCs and LLPCs is greatly separated in
time, it seems unlikely that MBCs and LLPCs are created alternatively via asymmetric division, as has been suggested to occur
(Barnett et al., 2012).
Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc. 123
A
B
Figure 4. Kinetics of the Formation of Long-Lived Immune Compartments in WT Mice
(A) BALB/cJ WT mice were immunized with NP-CGG and i.p. injected with BrdU or PBS at indicated time points as depicted in Figure 1A. The frequencies of
BrdU+ of live splenic NP+CD38+CD95IgM+ MBCs (blue dashed line, squares) or splenic NP+CD38+CD95IgG1+ MBCs (red solid line, circles) and BM
B220loCD138+NPsurface- NPintracellular+IgG1+ LLPCs (black solid line, triangles) were analyzed 8 weeks later by flow cytometry (Figure S4). Symbols are mean ±
SEM of 15–21 mice per group.
(B) Accumulation of long-lived MBCs and BM LLPCs over time. Plotted values are based on experimental data presented in (A) and missing data (*) were
interpolated with the spline function in the R statistical package. Dotted ledger lines indicate the time after immunization when 40% of each immune compartment
is formed.
Our results modify our view of how the GC evolves and in this
way raise a new set of questions. How do changes in the GC
milieu support this time-dependent shift in the generation of
long-lived B cell progeny? Evolution in the tendency of the
GC to produce MBCs toward LLPCs could be B cell intrinsic,
with cells counting divisions in some way and this in turn influencing in a stochastic fashion the likelihood to differentiate into
a MBC versus a LLPC (Hasbold et al., 2004). This is in accord
with concepts elaborated by Hodgkin and colleagues based
on in vitro data, including recent elegant single-cell studies of
short-term B cell cultures (Duffy et al., 2012). In this regard,
mRNAs encoding several key transcription factors and co-factors related to GC, MBC, and LLPC identity are differentially expressed in early and late GCBCs, including Irf8, Egr1, Mef2b,
Sh3bgrl2 (whose protein is an interacting partner with Irf8
[Yoon et al., 2014]), and Zbtb20 (also known as Zfp288), which
has been recently implicated in the development of LLPCs
(Chevrier and Corcoran, 2014; Nutt et al., 2015; Wang and
Bhattacharya, 2014).
Previously, BCR signaling in the GC was shown to be highly
attenuated, with signaling observed only in the G2-M phase
(Khalil et al., 2012), but this has so far been studied only at the
peak of the GC reaction. Our transcriptional analysis shows
that a number of key molecules affecting signaling are differentially regulated in early and late GCs, including ITIM-containing
CD72, and also Lck and Sh2d2a (which encodes a T-cell-specific adaptor protein)—both thought to be important for T cell
receptor signaling (Granum et al., 2008) but which are evidently
differentially regulated in GCBCs. Further studies are needed
to address outcomes of BCR-mediated Ag-binding in late
GCBCs.
There could also be temporal evolution of signals delivered by
other cells to B cells in the GC. Notably, the total number of
T cells per GC as well as GC T cell density declines significantly
at d16 after immunization (Wollenberg et al., 2011), a time when
the GC reaction is switching its output. In addition, T follicular
124 Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc.
regulatory cells accumulate with time in the GC (Ramiscal and
Vinuesa, 2013; Wollenberg et al., 2011). Therefore, T-cellderived signals might change with GC maturation due to timedependent changes in quantity or quality. Commensurate with
this is the downregulation of Stat4 and Il10ra in late GCBCs,
which could affect responsiveness to key Tfh cytokines such
as IL-21 and IL-10 (De Silva and Klein, 2015; Linterman et al.,
2010; Ramiscal and Vinuesa, 2013).
Regardless of the source of evolving external or B-cellintrinsic signals that might shift in time, a common denominator
in the B cell could be NF-kB signaling, because it was recently
shown that specific subunits are required in B cells at different
stages during the GC response (Heise et al., 2014). Another
factor could be lymphocyte metabolic pathways, which can
differ as a function of differentiation and activation (Caro-Maldonado et al., 2012). The GC reaction is dynamic in size and
cell composition with the majority of GCs getting smaller at
d14 (Rao et al., 2002). It is therefore conceivable that energy
supply and oxygen availability might vary as a function of
time, perhaps providing cues for selective differentiation.
Notably, of the 100 differentially regulated genes, 20 of
them are annotated as affecting metabolism (Figure 7). All of
the potential modifiers mentioned above are mutually nonexclusive and any or several of them could facilitate the general
shift of the GC output over time. The current findings now help
to refocus the research questions on evaluating these possibilities, which should ultimately lead to a better understanding of
the origins of both MBCs and LLPCs and the differentiation
process in the GC.
Several groups have noted that MBC-phenotype cells can be
observed very early in the immune response. MBCs can form in
mice lacking Bcl6 and therefore GCs (Kaji et al., 2012; Toyama
et al., 2002), although it is possible this wouldn’t occur if Bcl6
were present. Further supporting the notion, putatively GC-independent IgG1+ cells with a surface phenotype similar to
that of MBCs were detected by d10 (Inamine et al., 2005)
A
D
B
C
E
F
Figure 5. Disruption of Peak GC Reaction Diminishes LLPCs but Not MBCs
(A–C) Transfer recipients were injected i.p. with 350 mg anti-CD40L Ab or hamster control IgG at d12, d13, and d14 after NP-CGG immunization. Each symbol
represents one mouse and lines are means. Numbers of low-affinity (NP16-BSA) (A) and high-affinity (NP2-BSA) (B) BM AFCs per 106 BM cells were measured by
ELISpot and numbers of live NP+CD38+CD95CD19+ splenic B cells were quantified by flow cytometric analysis (C) 8 weeks later.
(D–F) Frequency distributions of MBC subsets, as distinguished by their expression of CD80 and PD-L2, were analyzed by flow cytometry.
Shown are representative results of one out of two independent experiments.
and IgM+ B cells with memory phenotype were found by d5
(Pape et al., 2011). It is now apparent that GCs can form as
early as d3–d4, though not earlier than this (Kerfoot et al.,
2011, and see above). Our findings demonstrating formation
via stable BrdU labeling before the GC reaction has started
definitively support the existence of GC-independent MBC formation in normal, Bcl6-intact animals; EdU labeling studies
suggest that most form in the T cell zones and at the T cell-B
cell border. In addition, our data shed light on the fate and utility
of these cells. We show that these cells join the long-lived MBC
compartment: a substantial fraction of IgM+ MBCs (and some
IgG1+ MBCs) are made during d0–d2 and remain at least for
8 weeks. The surface phenotype of these MBCs generated
very early in the response is also of interest. The great majority
of these early MBCs lack expression of CD80 and PD-L2, yet
most of them express CD73. These results, coupled with our
recent observations on the functions of MBC subsets (Zuccarino-Catania et al., 2014), indicate that these early, mainly unmutated and unswitched MBCs are specialized to efficiently
adapt to antigenic variants upon re-exposure, as they can
undergo secondary GC reactions (Zuccarino-Catania et al.,
2014). Furthermore, the expression of CD73 on these preGC-derived MBCs clearly shows that CD73 is not a suitable
marker to identify GC-derived MBCs, in contrast to what was
previously thought (Taylor et al., 2012).
The evident design of the system, with generation of early
MBCs followed by late LLPC generation, presents a seemingly
rational adaptation. Early generation of MBCs both within and
outside of the GC could be a successful strategy for reinforcing
the response to pathogens that persist beyond the earliest adaptive immune responses. Such quickly formed MBCs could presumably rejoin the response should the pathogen persist, though
this has yet to be directly demonstrated. Early differentiation of
long-lived MBCs, with the resultant paucity of mutations in V
gene regions, probably prevents the humoral memory compartment in total from becoming overly committed to the initial immunogen, potentially a more optimal strategy for later responses to
genetically variant pathogens.
Conversely, generation of LLPCs occurs at a time when a
prolonged selective process in the GC has allowed for a finetuning and commitment to the best possible affinity for the original Ag. This design is appropriate for cells that are terminally
differentiated. The delay in making effective LLPCs might also
explain why optimally protective long-term Ab responses to
certain vaccines (e.g., hepatitis B) take so long to appear. It
is important to point out that our results examine the BM PC
compartment only at week 8 and do not bear on AFCs that
might have been present at earlier time points. It has been
postulated that BM-resident AFCs are generated early in the
response, followed by selective loss of low-affinity variants
from d14 to d28 whereas higher-affinity cells persisted from
d14 to d35 (Smith et al., 1997). Because we found that 7%–
9% of LLPCs resident in the BM at week 8 were generated
by d14, our data are not in conflict with the concept of early
Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc. 125
A
B
Figure 6. Mutational Content of MBCs Matches Overall Mutational Content of Early GCs whereas Mutational Content of LLPCs Is Reflected
in Late GCs
(A) Experimental outline. B1-8 mice were injected with 0.75 mg EdU three times a day at indicated time points after NP-CGG immunization, then rested for
8 weeks, after which EdU+ or total NP+IgG1+ splenic MBCs and BM LLPCs were sorted. Similarly prepared mice were directly sacrificed to sort splenic
EMANP+IgG1+CD19+ cells of GC (CD38CD95+) or non-GC (CD38+CD95) phenotype.
(B) Mutational content analysis. DNA from 200–5,000 target cells was used for Vl1 gene sequencing. Each dot represents a single in-frame sequence from up to
16 sequences from 3 individual mice per population.
BM AFC generation (Smith et al., 1997) and it seems likely
that long-lived, higher-affinity LLPCs that are generated later
in the response replace or outnumber the initial lower-affinity
cohorts.
126 Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc.
Finally, our results might also be informative for designing vaccine strategies. Some vaccines demonstrate only short-lived Ab
responses and protection. If the goal is to generate Ab-mediated
long-lived protection, then adjuvant and Ag designs—as well as
(legend on next page)
Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc. 127
immunization schedules—should be adjusted to prolong late GC
responses with resultant production of more and better LLPCs.
In this regard, Kasturi et al. (2011) have demonstrated that immunization with immobilized Ag on synthetic nanoparticles together
with TLR7 and TLR4 ligands induces a synergistic increase in
numbers of LLPCs and long-lived Ab titers, which correlates
with enhanced and sustained late GC responses. Our studies
probably elucidate the underlying connection between these
two observations.
The converse might be true in the case of influenza, because
mutating MBCs contribute to early PB responses to antigenically
drifted viruses (Wrammert et al., 2008), whereas Ab produced by
LLPCs will be of little long-term value as quickly as the next flu
season. In this situation, vaccines that promoted enlarged early
GC responses, coupled with approaches to forestall GC maturation, would be more optimal, because this should yield relatively
more MBCs. As appreciated also by others (De Silva and Klein,
2015), further elucidation of molecular changes that could modulate GC quality, kinetics, and subsequent output could empower
us to manipulate the GC reaction and impact efforts to prevent
infectious diseases.
EXPERIMENTAL PROCEDURES
Mice, Immunization, and In Vivo Treatment
Use of B1-8i+/ genetically targeted BALB/cJ mice as B cell donors and AM14
Transgenic (Tg) 3 Vk8R genetically targeted BALB/cJ mice as recipient strain
was recently described (Zuccarino-Catania et al., 2014). All mice were maintained under specific-pathogen-free conditions and all animal experiments
were approved by the Yale or University of Pittsburgh Institutional Animal
Care and Use Committee. 6- to 12-week-old mice were i.p. immunized with
50 mg of NP-CGG precipitated in alum. To disrupt GC reactions, mice were
i.p. injected once a day at d12–d14 with 350 mg purified anti-CD40L Ab
(MR1) in PBS or Armenian hamster IgG (Innovative Research) as control. To
prevent GCs from re-appearing, mice were continuously i.p. injected with
125 mg Ab once a week until the end of the experiment. 5-bromo-20 -deoxyuridine (BrdU, Sigma) and 5-ethynyl-20 -deoxyuridine (EdU, Invitrogen) were dissolved in sterile PBS and 0.75 mg to 1 mg in a volume of 200 ml were injected i.p.
Adoptive B Cell Transfer
Splenocytes from B1-8i+/ genetically targeted BALB/cJ mice were depleted
of T cells by incubation with supernatants of rat IgM anti-mouse CD4 and rat
IgM anti- mouse CD8 Ab-producing cell lines and the addition of Low-Tox rabbit complement (Cedarlane) followed by Percoll (GE Healthcare) gradient purification. Flow cytometric analysis was performed to determine the frequency
of NP-binding B cells. An equivalent of 2 3 105 untouched NP-reactive B cells
was then transferred into tail veins of AM14 Tg 3 Vk8R genetically targeted
BALB/cJ mice.
Flow Cytometric Analysis, Cell Sorting, and ELISpot Assay
Flow cytometric analysis and ELISpot assays were performed essentially as
described (Good-Jacobson et al., 2010; Zuccarino-Catania et al., 2014).
EdU detection was performed with the Click-iT EdU Alexa Fluor 647 or 488
Flow Cytometry Assay Kit (Invitrogen) and subsequent EdU+ cells were sorted
on a FACSAria (BD Immunocytometry Systems). BrdU detection was performed as described (Khalil et al., 2012) and is detailed in the Supplemental
Experimental Procedures together with protocols used for BrdU and EdU double-detection by multicolor-flow cytometry.
In Vitro B Cell Stimulation
B cell purification was performed with the EasySep Mouse B Cell Enrichment
Kit (StemCell Technologies). 1 3 107 cells/ml were incubated in PBS with 0.5%
(v/v) fetal calf serum (FCS, HyClone) and 4 mM of BDHorizon Violet Cell Proliferation Dye 450 (VPD450, BD Biosciences) for 5 min at 37 C and the reaction
was quenched with 3 volumes of FCS for 1 min at 37 C and cells were washed
twice in RPMI 1640 medium. Cells were cultured in the presence of 5 mg/ml
CpG ODN 1826 (50 -tccatgacgttcctgacgtt-30 ; Invivogen) for 3–4 days at 37 C
and then subjected to flow cytometric staining.
Sequencing and Immunohistological Analysis
Vl1 sequencing was performed essentially as described (Anderson et al.,
2007). Immunohistological analysis was performed essentially as described
(Kerfoot et al., 2011). EdU was detected with the Click-iT EdU Alexa Fluor
555 Imaging Kit (Invitrogen). 403 tiled images of whole splenic sections
were acquired with an IX83 fluorescent microscope (Olympus) and image analysis was performed with cellSens Dimension software (Olympus) with the
count and measure module.
Statistics
The Mann-Whitney nonparametric, two-tailed test was used for statistical analyses. Significance was determined at the 95% confidence level and is
defined as follows: ***p < 0.001; **p = 0.001–0.01; *p R 0.01–0.05; n.s.,
p > 0.05. Interpolation of missing data (Figure 4B) was performed with the
spline function in the R statistical package (http://www.R-project.org).
ACCESSION NUMBERS
The microarray data have been deposited in NCBI GEO under accession number GEO: GSE76502.
SUPPLEMENTAL INFORMATION
Supplemental Information includes four figures and Supplemental Experimental Procedures (including Ag, Ab, and detection reagents) and can
be found with this article online at http://dx.doi.org/10.1016/j.immuni.2015.
12.004.
AUTHOR CONTRIBUTIONS
F.J.W. and M.J.S. designed research; G.V.Z.-C. performed experiments and
gave conceptual advice; M.C. performed bioinformatical analysis; and
F.J.W. implemented and analyzed all experiments and wrote the manuscript
together with M.J.S.
ACKNOWLEDGMENTS
We thank M. Tomayko, T. Winkler, and M. Cancro for critical reading the manuscript and discussions; the Yale Cell Sorter Facility for expert cell sorting; the
Yale Animal Resource Center and the Division of Laboratory Animal Resources
of the University of Pittsburgh, in particular A. Durso and K. Faidley, for expert
animal care; the Keck Facility at Yale University for DNA sequencing; L. Conter
and P. Baevova for supporting experimental procedures; and W. Humphries IV
for help with imaging and analysis. Supported by the NIH (R01-Al46303 to
M.J.S.), DFG Research Fellowship (WE 4752/1-1 to F.J.W.), and NSF Graduate
Research Fellowship (G.V.Z.-C).
Received: April 12, 2015
Revised: August 15, 2015
Accepted: September 24, 2015
Published: January 12, 2016
Figure 7. Gene Expression Profiling of Early and Late GCBCs
On d8 and d18 after NP-CGG immunization of transfer recipients, 3 3 105 live CD19+NP+ kappa light chainCD38CD95+ splenic GCBCs were sorted per sample
and subjected to gene expression profiling using Illumina MouseWG-6 v2.0 Expression BeadChip arrays. Heatmap of differentially expressed genes of early (d8)
and late (d18) NP reactive late GCBCs. Each column represents an independent replicate. Genes with a statistically significant (false-discovery rate q < 0.05)
change in expression of R1.7 are displayed and were grouped by function.
128 Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc.
REFERENCES
Allen, C.D.C., Okada, T., and Cyster, J.G. (2007a). Germinal-center organization and cellular dynamics. Immunity 27, 190–202.
Allen, C.D.C., Okada, T., Tang, H.L., and Cyster, J.G. (2007b). Imaging of
germinal center selection events during affinity maturation. Science 315,
528–531.
Anderson, S.M., Hannum, L.G., and Shlomchik, M.J. (2006). Memory B cell
survival and function in the absence of secreted antibody and immune complexes on follicular dendritic cells. J. Immunol. 176, 4515–4519.
Anderson, S.M., Tomayko, M.M., Ahuja, A., Haberman, A.M., and Shlomchik,
M.J. (2007). New markers for murine memory B cells that define mutated and
unmutated subsets. J. Exp. Med. 204, 2103–2114.
Barnett, B.E., Ciocca, M.L., Goenka, R., Barnett, L.G., Wu, J., Laufer, T.M.,
Burkhardt, J.K., Cancro, M.P., and Reiner, S.L. (2012). Asymmetric B cell division in the germinal center reaction. Science 335, 342–344.
Blink, E.J., Light, A., Kallies, A., Nutt, S.L., Hodgkin, P.D., and Tarlinton, D.M.
(2005). Early appearance of germinal center-derived memory B cells and
plasma cells in blood after primary immunization. J. Exp. Med. 201, 545–554.
Caro-Maldonado, A., Gerriets, V.A., and Rathmell, J.C. (2012). Matched and
mismatched metabolic fuels in lymphocyte function. Semin. Immunol. 24,
405–413.
Chan, T.D., and Brink, R. (2012). Affinity-based selection and the germinal center response. Immunol. Rev. 247, 11–23.
Chan, T.D., Gatto, D., Wood, K., Camidge, T., Basten, A., and Brink, R. (2009).
Antigen affinity controls rapid T-dependent antibody production by driving the
expansion rather than the differentiation or extrafollicular migration of early
plasmablasts. J. Immunol. 183, 3139–3149.
Chevrier, S., and Corcoran, L.M. (2014). BTB-ZF transcription factors, a
growing family of regulators of early and late B-cell development. Immunol.
Cell Biol. 92, 481–488.
Dal Porto, J.M., Haberman, A.M., Kelsoe, G., and Shlomchik, M.J. (2002). Very
low affinity B cells form germinal centers, become memory B cells, and participate in secondary immune responses when higher affinity competition is
reduced. J. Exp. Med. 195, 1215–1221.
De Silva, N.S., and Klein, U. (2015). Dynamics of B cells in germinal centres.
Nat. Rev. Immunol. 15, 137–148.
Dogan, I., Bertocci, B., Vilmont, V., Delbos, F., Mégret, J., Storck, S., Reynaud,
C.A., and Weill, J.C. (2009). Multiple layers of B cell memory with different
effector functions. Nat. Immunol. 10, 1292–1299.
Duffy, K.R., Wellard, C.J., Markham, J.F., Zhou, J.H.S., Holmberg, R.,
Hawkins, E.D., Hasbold, J., Dowling, M.R., and Hodgkin, P.D. (2012).
Activation-induced B cell fates are selected by intracellular stochastic competition. Science 335, 338–341.
Gatto, D., Pfister, T., Jegerlehner, A., Martin, S.W., Kopf, M., and Bachmann,
M.F. (2005). Complement receptors regulate differentiation of bone marrow
plasma cell precursors expressing transcription factors Blimp-1 and XBP-1.
J. Exp. Med. 201, 993–1005.
Gitlin, A.D., Mayer, C.T., Oliveira, T.Y., Shulman, Z., Jones, M.J.K., Koren, A.,
and Nussenzweig, M.C. (2015). HUMORAL IMMUNITY. T cell help controls the
speed of the cell cycle in germinal center B cells. Science 349, 643–646.
Good-Jacobson, K.L., and Shlomchik, M.J. (2010). Plasticity and heterogeneity in the generation of memory B cells and long-lived plasma cells: the influence of germinal center interactions and dynamics. J. Immunol. 185, 3117–
3125.
Good-Jacobson, K.L., Szumilas, C.G., Chen, L., Sharpe, A.H., Tomayko,
M.M., and Shlomchik, M.J. (2010). PD-1 regulates germinal center B cell survival and the formation and affinity of long-lived plasma cells. Nat. Immunol.
11, 535–542.
Good-Jacobson, K.L., Song, E., Anderson, S., Sharpe, A.H., and Shlomchik,
M.J. (2012). CD80 expression on B cells regulates murine T follicular helper
development, germinal center B cell survival, and plasma cell generation.
J. Immunol. 188, 4217–4225.
Granum, S., Andersen, T.C.B., Sørlie, M., Jørgensen, M., Koll, L., Berge, T.,
Lea, T., Fleckenstein, B., Spurkland, A., and Sundvold-Gjerstad, V. (2008).
Modulation of Lck function through multisite docking to T cell-specific adapter
protein. J. Biol. Chem. 283, 21909–21919.
Hasbold, J., Corcoran, L.M., Tarlinton, D.M., Tangye, S.G., and Hodgkin, P.D.
(2004). Evidence from the generation of immunoglobulin G-secreting cells that
stochastic mechanisms regulate lymphocyte differentiation. Nat. Immunol. 5,
55–63.
Heise, N., De Silva, N.S., Silva, K., Carette, A., Simonetti, G., Pasparakis, M.,
and Klein, U. (2014). Germinal center B cell maintenance and differentiation
are controlled by distinct NF-kB transcription factor subunits. J. Exp. Med.
211, 2103–2118.
Inamine, A., Takahashi, Y., Baba, N., Miyake, K., Tokuhisa, T., Takemori, T.,
and Abe, R. (2005). Two waves of memory B-cell generation in the primary immune response. Int. Immunol. 17, 581–589.
Jacob, J., Kelsoe, G., Rajewsky, K., and Weiss, U. (1991). Intraclonal generation of antibody mutants in germinal centres. Nature 354, 389–392.
Kaji, T., Ishige, A., Hikida, M., Taka, J., Hijikata, A., Kubo, M., Nagashima, T.,
Takahashi, Y., Kurosaki, T., Okada, M., et al. (2012). Distinct cellular pathways
select germline-encoded and somatically mutated antibodies into immunological memory. J. Exp. Med. 209, 2079–2097.
Kasturi, S.P., Skountzou, I., Albrecht, R.A., Koutsonanos, D., Hua, T., Nakaya,
H.I., Ravindran, R., Stewart, S., Alam, M., Kwissa, M., et al. (2011).
Programming the magnitude and persistence of antibody responses with
innate immunity. Nature 470, 543–547.
Kerfoot, S.M., Yaari, G., Patel, J.R., Johnson, K.L., Gonzalez, D.G., Kleinstein,
S.H., and Haberman, A.M. (2011). Germinal center B cell and T follicular helper
cell development initiates in the interfollicular zone. Immunity 34, 947–960.
Khalil, A.M., Cambier, J.C., and Shlomchik, M.J. (2012). B cell receptor signal
transduction in the GC is short-circuited by high phosphatase activity. Science
336, 1178–1181.
Kitano, M., Moriyama, S., Ando, Y., Hikida, M., Mori, Y., Kurosaki, T., and
Okada, T. (2011). Bcl6 protein expression shapes pre-germinal center B cell
dynamics and follicular helper T cell heterogeneity. Immunity 34, 961–972.
Le, T.-V.L., Kim, T.H., and Chaplin, D.D. (2008). Intraclonal competition inhibits
the formation of high-affinity antibody-secreting cells. J. Immunol. 181, 6027–
6037.
Linterman, M.A., Beaton, L., Yu, D., Ramiscal, R.R., Srivastava, M., Hogan,
J.J., Verma, N.K., Smyth, M.J., Rigby, R.J., and 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.
Liu, D., Xu, H., Shih, C., Wan, Z., Ma, X., Ma, W., Luo, D., and Qi, H. (2015).
T-B-cell entanglement and ICOSL-driven feed-forward regulation of germinal
centre reaction. Nature 517, 214–218.
Manz, R.A., Thiel, A., and Radbruch, A. (1997). Lifetime of plasma cells in the
bone marrow. Nature 388, 133–134.
Nutt, S.L., Hodgkin, P.D., Tarlinton, D.M., and Corcoran, L.M. (2015). The generation of antibody-secreting plasma cells. Nat. Rev. Immunol. 15, 160–171.
Obukhanych, T.V., and Nussenzweig, M.C. (2006). T-independent type II immune responses generate memory B cells. J. Exp. Med. 203, 305–310.
Oracki, S.A., Walker, J.A., Hibbs, M.L., Corcoran, L.M., and Tarlinton, D.M.
(2010). Plasma cell development and survival. Immunol. Rev. 237, 140–159.
Pape, K.A., Taylor, J.J., Maul, R.W., Gearhart, P.J., and Jenkins, M.K. (2011).
Different B cell populations mediate early and late memory during an endogenous immune response. Science 331, 1203–1207.
Paus, D., Phan, T.G., Chan, T.D., Gardam, S., Basten, A., and Brink, R. (2006).
Antigen recognition strength regulates the choice between extrafollicular
plasma cell and germinal center B cell differentiation. J. Exp. Med. 203,
1081–1091.
Phan, T.G., Gardam, S., Basten, A., and Brink, R. (2005). Altered migration,
recruitment, and somatic hypermutation in the early response of marginal
zone B cells to T cell-dependent antigen. J. Immunol. 174, 4567–4578.
Ramiscal, R.R., and Vinuesa, C.G. (2013). T-cell subsets in the germinal center.
Immunol. Rev. 252, 146–155.
Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc. 129
Rao, S.P., Vora, K.A., and Manser, T. (2002). Differential expression of the
inhibitory IgG Fc receptor FcgammaRIIB on germinal center cells: implications
for selection of high-affinity B cells. J. Immunol. 169, 1859–1868.
Tomayko, M.M., Steinel, N.C., Anderson, S.M., and Shlomchik, M.J. (2010).
Cutting edge: Hierarchy of maturity of murine memory B cell subsets.
J. Immunol. 185, 7146–7150.
Schittek, B., and Rajewsky, K. (1990). Maintenance of B-cell memory by longlived cells generated from proliferating precursors. Nature 346, 749–751.
Toyama, H., Okada, S., Hatano, M., Takahashi, Y., Takeda, N., Ichii, H.,
Takemori, T., Kuroda, Y., and Tokuhisa, T. (2002). Memory B cells without somatic hypermutation are generated from Bcl6-deficient B cells. Immunity 17,
329–339.
Schwickert, T.A., Alabyev, B., Manser, T., and Nussenzweig, M.C. (2009).
Germinal center reutilization by newly activated B cells. J. Exp. Med. 206,
2907–2914.
Schwickert, T.A., Victora, G.D., Fooksman, D.R., Kamphorst, A.O., Mugnier,
M.R., Gitlin, A.D., Dustin, M.L., and Nussenzweig, M.C. (2011). A dynamic
T cell-limited checkpoint regulates affinity-dependent B cell entry into the
germinal center. J. Exp. Med. 208, 1243–1252.
Victora, G.D., and Nussenzweig, M.C. (2012). Germinal centers. Annu. Rev.
Immunol. 30, 429–457.
Wang, Y., and Bhattacharya, D. (2014). Adjuvant-specific regulation of longterm antibody responses by ZBTB20. J. Exp. Med. 211, 841–856.
Shih, T.-A.Y., Meffre, E., Roederer, M., and Nussenzweig, M.C. (2002). Role of
BCR affinity in T cell dependent antibody responses in vivo. Nat. Immunol. 3,
570–575.
Wollenberg, I., Agua-Doce, A., Hernández, A., Almeida, C., Oliveira, V.G., Faro,
J., and Graca, L. (2011). Regulation of the germinal center reaction by Foxp3+
follicular regulatory T cells. J. Immunol. 187, 4553–4560.
Shlomchik, M.J., and Weisel, F. (2012). Germinal center selection and the
development of memory B and plasma cells. Immunol. Rev. 247, 52–63.
Wrammert, J., Smith, K., Miller, J., Langley, W.A., Kokko, K., Larsen, C.,
Zheng, N.-Y., Mays, I., Garman, L., Helms, C., et al. (2008). Rapid cloning of
high-affinity human monoclonal antibodies against influenza virus. Nature
453, 667–671.
Smith, K.G., Light, A., Nossal, G.J., and Tarlinton, D.M. (1997). The extent of
affinity maturation differs between the memory and antibody-forming cell
compartments in the primary immune response. EMBO J. 16, 2996–3006.
Takahashi, Y., Dutta, P.R., Cerasoli, D.M., and Kelsoe, G. (1998). In situ studies
of the primary immune response to (4-hydroxy-3-nitrophenyl)acetyl. V. Affinity
maturation develops in two stages of clonal selection. J. Exp. Med. 187,
885–895.
Takahashi, Y., Ohta, H., and Takemori, T. (2001). Fas is required for clonal selection in germinal centers and the subsequent establishment of the memory B
cell repertoire. Immunity 14, 181–192.
Takemori, T., Kaji, T., Takahashi, Y., Shimoda, M., and Rajewsky, K. (2014).
Generation of memory B cells inside and outside germinal centers. Eur. J.
Immunol. 44, 1258–1264.
Taylor, J.J., Pape, K.A., and Jenkins, M.K. (2012). A germinal center-independent pathway generates unswitched memory B cells early in the primary
response. J. Exp. Med. 209, 597–606.
130 Immunity 44, 116–130, January 19, 2016 ª2016 Elsevier Inc.
Yoon, J., Feng, X., Kim, Y.-S., Shin, D.-M., Hatzi, K., Wang, H., and Morse,
H.C., 3rd (2014). Interferon regulatory factor 8 (IRF8) interacts with the B cell
lymphoma 6 (BCL6) corepressor BCOR. J. Biol. Chem. 289, 34250–34257.
Zotos, D., and Tarlinton, D.M. (2012). Determining germinal centre B cell fate.
Trends Immunol. 33, 281–288.
Zotos, D., Coquet, J.M., Zhang, Y., Light, A., D’Costa, K., Kallies, A., Corcoran,
L.M., Godfrey, D.I., Toellner, K.-M., Smyth, M.J., et al. (2010). IL-21 regulates
germinal center B cell differentiation and proliferation through a B cell-intrinsic
mechanism. J. Exp. Med. 207, 365–378.
Zuccarino-Catania, G.V., Sadanand, S., Weisel, F.J., Tomayko, M.M., Meng,
H., Kleinstein, S.H., Good-Jacobson, K.L., and Shlomchik, M.J. (2014).
CD80 and PD-L2 define functionally distinct memory B cell subsets that are independent of antibody isotype. Nat. Immunol. 15, 631–637.
Immunity
Supplemental Information
A Temporal Switch in the Germinal Center
Determines Differential Output
of Memory B and Plasma Cells
Florian J. Weisel, Griselda V. Zuccarino-Catania, Maria Chikina, and Mark J. Shlomchik
#2
10 4
10 4
10 3
10 3
10 3
10 3
10 3
0
0
0
0
0
0.00
0 10
10
5
10
4
10
3
10
3
10
4
10
5
0.00
10 3
10 4
5
10
4
10
4
10
3
10
3
10 5
0.03
0 10
10
0
0 10 2
10 5
2
10 3
10 4
10 5
0.04
10 5
10 4
10 4
10 3
10 3
10 3
0
0
0
10 3
10 4
10 5
3
10
4
10
5
0.00
0 10 2
10 4
0 10 2
10
0
0 10 2
10 5
2
0 10 2
10 3
10 4
10 5
10 3
10 4
10 5
0.04
0 10
2
10
3
10
B
20
15
4
10
5
0 10
2
10
3
10
4
10
5
10
5
0
0 10 2
10 3
10 4
10 5
t
5
w
10
cJ
4
tx
10
no
3
ce
ll
tx
5
10
B
10
2
0.09
10 5
10 4
# of NP+ B cells
per spleen (x103)
B cell tx
No tx
0.08
10 5
10 4
0
Balb/cJ wt
0.09
10 5
#5
10 4
0 10
NP
0.06
10 5
#4
b/
0.09
10 5
#3
al
#1
B
A
lambda light chain
Figure S1, related to Fig. 1 and 2. Adoptive transfer system to adjust NP reactive precursor frequency to WT
levels. AM14 transgenic x Vk8R genetic ally targeted BALB/cJ mice harbor an irrelevant monoclonal population of B
cells and therefore are unable to mount endogenous NP-­specific immune responses but otherwis e dis play normal
lymphoid architecture (Prak and Weigert, 1995;; Shlomchik et al., 1993) with reduced numbers of B cells. These mice
were adoptiv ely transferred with an equivalent of 2x105 NP reactive B cells from B1-­8i+/-­ genetically targeted BALB/cJ
mice, carrying a germline encoded, unmutated Vh186.2 site-­directed transgene that encodes a BCR with moderate
affinity for the hapten NP when paired with λ1 light chains (Sonoda et al., 1997), to allow donor B cell derived NP-­
specific immune responses. (A) 24h after adoptive transfer recipients were sacrificed and splenocytes were stained to
determine the frequency of Igλ + NP-­binding B cells. FACS plots are gated on live CD19+ cells and each plot represents
an individual AM14 transgenic x Vk8R genetically targeted BALB/cJ mouse either transferred with NP-­reactive B cells
(top row;; B cell tx) or injected with transfer buffer only (middle row;; No tx). Naive BALB/cJ WT mic e served as control
(bottom row). Numbers adjacent to outlined areas indic ate percent gated population. (B) Quantification of frequencies
of NP-­reactiv e B cell displayed in (A) indic ate a WT-­like precursor frequency of on average of about 12.5x103 NP-­
specific B cells per recipient spleen at the time point of immunization.
150K
150K
100K
100K
100K
50K
50K
50K
41.74
40.02
250K
0
50K
250K
0
200K
200K
150K
150K
150K
100K
SSC'W
0
100K
150K
200K
250K
50K
100K
150K
200K
250K
0
10 5
10 5
4
4
4
10
10
10 3
95.76
50K
100K
150K
10 5
200K
250K
10 4
150K
200K
250K
50K
100K
150K
200K
0
50K
100K
150K
10 5
10 4
150K
150K
150K
150K
100K
100K
100K
100K
100K
100K
50K
50K
50K
79.55
0
0
50K
100K
200K
250K
10 3
10 2
10 2
10 2
0
10 2
CD38
10 5
10 3
10 4
10 5
200K
10 3
10 4
10 5
10 2
0
10 3
10 4
150K
94.01
NA
10 2
0
10 5
IgG1
FMO$IgG1
10 5
10 4
10 4
41.80
100K
50K
50K
50K
10 4
0
0
50K
100K
150K
200K
250K
5
50K
10 2
10 3
10 4
B220
10 5
10 5
10 3
10 3
10 2
10 2
0
0
B220
10 3
10 4
10 4
10 5
100K
150K
200K
10 4
10 3
10 3
2
2
10
0
10 2
0
10 3
95.44
0
0
50K
100K
150K
200K
250K
50K
100K
150K
200K
10 2
10 3
10 4
10 5
0
10 2
10 3
10 4
10 5
250K
97.26
0
50K
100K
150K
200K
10 3
10 3
0
0
0
3.70
10 2
10
10 3
10 4
10 2
0
10
10 3
10 4
5
10
10 3
10 3
10 3
10 2
0
0
10 5
10 2
0
92.88
10 3
10 4
150K
100K
100K
100K
150K
200K
250K
10 4
NA
10 3
10 5
IgG1+
150K
200K
250K
0
10 5
10 4
10 4
10 4
3
3
10 3
10 2
10 3
10 4
10 5
100K
150K
200K
250K
10 2
31.10
0
10 2
50K
36.69
0
0
10 2
10 3
10 4
10 5
0
10 2
10 3
10 5
10 4
10 5
10 5
0.00
0.00
10 4
10 4
10 4
10 3
10 3
10 3
10 2
10 2
0
10
5
10
4
10 4
10 5
10
PD'L2
NA
10 3
10
2
10
3
10
4
1.39
10
5
0
0
10
2
10
3
10
4
10
5
0
10
2
10
3
10
4
10
5
19.46
NA
10 3
•
10 2
0
0
NA
2
0
19.00
0
10
5
60.24
10
2
10
3
10
4
0.67
10
5
3.19
10 2
0
10 2
10 3
10 4
<Pacific Blue-A>: IgM (intra)
0
10
10 3
10 4
0
10 2
10 3
10 4
<Pacific Blue-A>: IgM (intra)
10 2
0
10 5
10 2
0
100K
10 5
36.41
250K
93.93
50K
10 5
0
200K
0
0
10
150K
50K
0
50K
100K
100K
91.51
50K
0
50K
10 5
BrdU
10 5
•
10 5
0
0
200K
0.00
10 2
0
10 3
10 4
0.00
10 2
10 4
10 3
5
10 4
10 4
10 2
0
10 4
10 3
250K
2.13
10 5
10 4
10 2
200K
150K
0.51
2.84
10 5
5
150K
10 4
10 3
0
100K
200K
10 2
CD19
50K
150K
10
250K
43.68
0
0
200K
10 5
10 5
10 2
10 3
10 4
<Pacific Blue-A>: IgM (intra)
0
10 5
d15'17$
PBS$i.p.
5
10
10 4
BrdU
0
0
200K
250K
SSC'W
250K
10 5
10 4
10 5
0.39
150K
250K
0
10 3
10 2
10 2
10 4
10 5
100K
250K
5
d15'17$
BrdU i.p.
10 5
20.26
35.44
0
50K
50K
0
250K
10 4
IgM
50K
40.84
0
0
0.48
10 4
10
10 3
10 5
10 4
10 2
10 2
d6'8$PBS$i.p.
19.26
0
BrdU
0
10 4
10 5
IgG1+
250K
10 2
<FITC-A>: IgG1 (intra)
0
d6'8$BrdU i.p.
BrdU
200K
10 5
10 5
0
10 5
IgM+
FSC'A
250K
10 4
0
<FITC-A>: IgG1 (intra)
10 4
150K
10
10 3
IgG1
10 3
100K
5
97.51
0
0.02
10 2
0
10 2
50K
10 4
10 3
10 2
0
200K
0
NP$int
46.80
64.01
0
150K
0
0
10
10 3
IgG1
10 3
10 2
100K
94.37
0
0.89
FMO$IgM
10 5
0.04
48.75
IgM
50K
90.03
100K
0
10 3
150K
93.53
100K
10 4
NP$sur
NA
10 4
0
200K
CD138
10 3
250K
10 5
18.40
10 2
200K
200K
FSC'A
0.00
0
10 2
0
10 3
0
77.65
0
150K
200K
10 2
78.71
10 5
100K
250K
250K
10 4
CD95
50K
250K
150K
B220
10 3
0
0
10 4
10 3
0
80.52
0
150K
10 5
B220
50K
250K
10
250K
0.05
150K
SSC'W
0
0
10 5
2.61
100K
95.26
0
0
50K
10 3
97.58
0
200K
150K
FSC'A
250K
0
0
10 3
200K
91.74
10 5
10
150K
50K
0
50K
100K
250K
200K
0
92.51
50K
0
50K
100K
93.42
EMA
200K
200K
50K
NP
150K
250K
100K
FSC'A
100K
250K
250K
200K
NP
200K
250K
200K
CD80
150K
250K
200K
10 3
10 2
10 2
0
PD'L2
0.00
0
0
10
10
4
10
5
FMO$
CD80
10 3
10 2
0
27.01
0
10 5
10 4
8.35
10 4
5
10 3
3
CD80
100K
+ NP-­CGG
txTx+$NP'CGG
+ PBS
+$PBS
250K
CD80
50K
250K
Tx+$Alum$
+ Alum
tx
+ BrdU
+$BrdU
200K
<FITC-A>: IgG1 (intra)
0
0
Tx+$NP'CGG$
+ NP-­CGG
tx
+ BrdU
+$BrdU
250K
50K
37.02
0
C
SSC'A
150K
No tx + NP-­CGG
NCC$+$NP'CGG
SSC'H
200K
Tx
tx + Alum
+$Alum
EMA
250K
200K
tx + NP-­CGG
+$NP'CGG
Tx
SSC'A
250K
200K
0
B
NCC$+$NP'CGG
No tx + NP-­CGG
250K
FSC'A
SSC'H
tx
Tx+$Alum
+ Alum
SSC'H
SSC'A
tx +$NP'CGG
Tx
+ NP-­CGG
EMA
A
0
10
3
10
4
10
5
10
FMO$
PD'L2
PD'L2
69.13
10 2
10 3
10 4
32.67
10 5
0.73
4
10 3
10 2
0
66.06
0
0.54
10 2
10 3
10 4
10 5
+ MBC, +Figure
related
to detection
Fig.
1. Gatingof strategy
the detection
BrdU+ LLPC
8 weeks
ed to Fig.
1. Gating
forto
theFig. detection
of BrdUstrategy
MBC 8S3,
weeks
after
immunization
of
Figure
S4, after
related
to Fig.
1. of
Gating
strategy for of the tdetection
BrdU+ MBC subsets 8 w
Figure
S2, strategy
related
1. Gating
for
the
BrdUfor
MBC ofsubsets
and
LLPC, 8 wimmunization
eeks
after immunization
ransfer of
recipients.
transfer
(left column) orofAlum
only recipients.
(middle
ts. Transfer recipients were immuniz ed with NP4CGG in Alum
(left recipients.
column) orTransfer
Alum onlrecipients
y (middle were immuniz ed with NP4CGG in Alumimmunization
transfer
Transfer recipients were immuniz ed with NP4CGG in Alum (lef
Transfer recipients
with
NP-­‐CGG in Alum ( left
r Alum nly or(middlecolumn) 2only
4h (middle
aasfter NP-­‐
B cell transfer
andand
i.p. column)
24h after
NP4reactive
B cell
transfer
and i.p. o
injected
withoBrdU
PBS atcolumn)
indicated
outlined
in reactive
NP4reactive
B cell transfer
and i.p. were
injected immunized
with BrdU or PBS
at indicated
time
points
as outlined
incolumn) or time
Alumpoints
column)
24h after NP4reactive
B cell transfer
i.p.i njected
injected with BrdU (left
Fig. 1A.
Mice
immunized
NP4CGG
in Alum without NP4specific B cell transfercolumn)
(right columnI
no
cell control)
munizedwith
with NP4CGG
in
Alum
without
NP4specific Btime cell transfer
(right
columnI
NCC:with
no cell
control)
orlum PBSNCC:
(right
column)
at indicated timeB points
as
outlined in (right
Fig. 1A. column
Single4cell suspensions
of red
BrdU
or
PBS a
t i
ndicated
p
oints
as
outlined
in F
ig. 1
A. M
ice
immunized
with
NP-­‐
C
GG i
n A
w
ithout
NP-­‐specific
c
ell
transfer
in A served as were
control
s . Single4cell
suspensions
cells were
analyzedwere
8 weeks
after 8 weeks after immunization in flow cytometric analy sis to determine the
ls. Single4cell suspensions of red blood cell4depleted splenocytes
analyzed
8 weeks
after of red blood cell4depleted bone marrow
depleted
splenocytes
analyzed
+
+
4 NP
+to determine
immunization
in flow
cytometri
cEMA
analysis
the frequency splenocytes
of BrdU cells.
LLPC
stringently
defined
+ cells.
4, NP+ CD19+8 + cells after and B; no tx) erved as
ingle-­‐cell
suspensions
red
blood
cell-­‐depleted
(A, C) were
oThe
r bone-­‐marrow
(B) wofere
w eeks
ow cytometric
analysis
to sdetermine
the controls. frequency of S
BrdU
cells. MBC
were
defined
asof of BM
BrdU
dis tribution
of CD80cells
and PD4L2
EMAanalyzed
BrdU
was assessed. Arro
4 B2204 CD138+ NPsurface4 NPintracellular+ IgG1+ cells. Arrows indic ate subsequent gating of populations and
as EMA
954 and IgG1+ or IgM+ cells. Arrows indic ate subsequent gating
of populations
and numbers next to
subsequent gating of populations
and+numbers +next to outlined
areas indicate
percent
gated population. F
+ cells. ( A) MBC w ere defined
-­‐ NP+ B220
-­‐ and IgG1
+ or IgM
+ cells. immunization
by
flow
cotometry
to
determine
the
frequency
of B
rdU
as
EMA
CD38
CD95
( B) numbers
nexts towere
outlined
areas
(NA)(FMO)
is depicted
of
dicate percent gated population. Fluorescent minus one (FMO)
control
included
to indicate
prove percent gated population. Not applicable
minus one
controlsinstead
were included
to prove specificity of the Ab intentionally excluded from the stainin
-­‐
-­‐
+
surface-­‐
intracellular+
+
-­‐
+
+
+
subsequent
gating
if
parental
gate
contains
virtually
no
cells.
Not applicable
(NA) is of depicted
instead
subsequent
gate contains
no cells.
Ab intentionally
excluded
the stainingdefined
solution. Not
(NA) isCD138
depic ted NP
instead of NP
BM LLPC w erefrom
stringently
asapplicable
EMA B220
IgG1 cells. The distribution
CD80 andofPD-­‐L2 of Egating
MAif, parental
NP CD19
BrdUvirtually
cells
g if parental gate contains virtually no cells.
was assessed in (C). Arrows indicate subsequent gating of populations and numbers next to outlined areas indicate percent gated population. F luorescent
minus one (FMO) controls were included to prove specificity of the Ab intentionally excluded from the staining solution. Not applicable (NA) is depicted
instead of subsequent gating if parental gate contains virtually no cells.
Area measurement
B
Consecutive sections
day 6
day 4
day 4
A
B220 CD4 EdU
B220 CD4 EdU Igλ
C
d3 No EdU i.p.
d4 No NP-­CGG i.p.
PNA B220 EdU
200
µm
d4 No B cell transfer
200
µm
B220 CD4 EdU Igλ
Figure S3, related to Fig. 3. Validation and quantification of histological analysis. Transfer recipients were
injected with 0.85mg EdU at 6.5, 3.5 and 0.5h prior to harvesting spleens at indic ated days post NP-­CGG
immunization. (A) Representative example of region definition and measurement. Regions of interest (ROI) were
drawn around white pulps and the area of B cell zones was determined and measured by staining intensity using the
count and measure analy sis module of cellSens Dimensions software (Olympus). Areas of T cell zones where
measured in analogous fashion (not depicted). Proliferating Ag specific (Igλ + EdU+) B cells were counted within each
area. (B) Clustered proliferation of Ag specific B cells within B cell follicles (color-­coded, dotted regions) was defined as
GC reaction if counterstaining for peanut agglutinin (PNA) was positive on consecutive sections. The markers B220,
EdU and Igλ (not depicted in right panel) were used to identify these regions on the consecutive sections (color-­coded,
dotted regions). Examples for d4 (upper panel) and d6 (lower panel) are shown. (C) Validation of specificity of
measured response. Transfer recipients not given EdU injection (left panel) or without NP-­CGG immuniz ation (middle
panel) or without B1-­8 B cell transfer (right panel) did not show significant expansion of Igλ + B cells. These controls
indicate that the observed expansion of Igλ + B cells was due to Ag-­mediated activ ation of adoptively transferred B1-­8 B
cells. Scale bars are 200µm.
A
NP)CGG
250K
200K
200K
150K
150K
100K
100K
50K
50K
52.39
100K
150K
200K
250K
0
250K
250K
200K
200K
150K
150K
100K
100K
100K
150K
200K
250K
50K
10 4
10 3
10 3
0
150K
200K
250K
100K
10 5
150K
200K
250K
96.79
0
10
10 3
10 2
10 2
0
0
10 2
10 3
10 4
150K
100K
100K
50K
58.96
0
0
50K
100K
100K
150K
200K
200K
250K
150K
95.57
100K
10 5
0
10 2
10 3
10 4
50K
100K
150K
200K
250K
10 3
IgG1
NP.sur
10 4
10 5
10 5
FMO*IgG1
10 5
10 4
10 4
10
0.00
3
51.55
10
50K
100K
150K
200K
250K
5
59.77
10
10 2
10 2
0
0
10 3
10 4
10 5
NP.int
3
0
10 2
0
10 2
10 3
10 4
10 5
10.03
10 5
0
10 2
10 3
10 4
10 5
10
3
10
3
10
2
10
2
10 4
3
10 3
B220
BrdU
150K
200K
250K
10 2
1.24
0 10
2
10
3
10
4
10
0
0.93
5
0 10 2
10 5
10 5
10 4
10 4
0.36
2
10 3
10 4
10 5
10 3
0.00
10 2
0
0
10 3
0
10 4
10 5
10 3
0
10 4
10 5
95.97
10 4
10 3
NA
2.01
10 2
IgM
0
0
10 2
10 3
10 4
10 5
d36#38.
BrdU i.p.
0
0
IgG1
50K
10 4
0
+
0.00
10 5
10 4
BrdU
0
10
10 5
10 2
10 3
10 4
10 5
10 5
0
10 2
10 3
10 4
10 5
10 5
10.55
0.00
10 4
10 4
10 3
10 3
10 2
10 2
0
0
10 5
IgG1
B220
IgM
100K
93.12
0.00
d9)11*BrdU*i.p. d9)11*PBS*i.p.
+
250K
5
10 4
10
36.33
3
10 2
0
200K
0
0
10 3
FMO*IgM
10 5
10 4
34.31
IgM
NA
0
10 3
150K
50K
10 2
0
10
IgG1
CD38
NA
2.81
10 2
0
10 4
93.77
0
0
100K
96.13
10 5
CD138
10 3
CD95
250K
10 5
95.71
10 2
200K
0
0
10 2
10
150K
50K
0
10
10 5
4
100K
100K
10 4
FSC#A
50K
200K
150K
10 2
0.01
0
250K
200K
250K
63.24
0
150K
250K
10 3
Figure S4, related to Fig. 4. Gating
strategy for the detection of BrdU+ MBC
and LLPC, 8 weeks after immunization
of WT mice. BALB/cJ WT mice were
immunized with NP-­CGG in Alum (left
column) or Alum only (right column) and
i.p. injected with BrdU or PBS at indicated
time points as depic ted in Fig. 1A. Single-­
cell suspensions of red blood cell-­depleted
splenocytes (A) or bone marrow cells (B)
were analy zed 8 weeks after immunization
by flow cytometry to determine the
frequency of BrdU+ cells. (A) MBC were
defined as EMA-­ NP+ B220+ CD38+ CD95-­
and IgG1+ or IgM+ cells. (B) BM LLPC
were stringently defined as EMA-­ B220-­
CD138+ NPsurface-­ NPintracellular+ IgG1+ cells.
Arrows indicate subsequent gating of
populations and numbers next to outlined
areas indicate percent gated population.
Fluorescent minus one (FMO) controls
were included to prove specificity of the Ab
intentionally excluded from the staining
solution. Not applicable (NA) is depicted
instead of subsequent gating if parental
gate contains virtually no cells.
4
10 3
0
50K
10 5
0.03
4
150K
SSC#W
0
50K
200K
10 5
10 2
97.17
100K
5
10 4
250K
200K
50K
0
10
0
NP
250K
EMA
EMA
50K
5
10 2
B220
200K
0
0
10
150K
50K
0
FSC)A
100K
90.61
50K
10
50K
250K
FSC#A
0
50K
SSC#H
0
Alum
Alum
50K
48.22
91.53
SSC)W
NP#CGG
NP-­CGG
SSC#A
250K
0
SSC)H
Alum
B220
SSC)A
FSC)A
B
Alum
NP-­CGG
IgG1+
BrdU
10 2
10 3
10 4
10 5
0
10 2
10 3
10 4
10 5
16.78
10 4
10 4
10 3
10 3
2
10 2
10
0
0.00
0
0 10
0
d36#38.
PBS.i.p.
2
10
3
10
4
10
5
0 10
2
10
3
10
4
10
5
10 5
d to Fig. 4. Gating strategy for the detection of BrdU+ MBC
8 weeks
after immunization
of WT
Figure
S7, related
to Fig. 4. Gating
strategy for the detection of BrdU+ BM LLPCs 8 weeks after immunization of
WTormice.
WTcolumn)
mic e were
with NP6CGG in Alum (left column) or Alum only (right column) and i.p.
T mice were immunized with NP6CGG in Alum (left column)
AlumBALB/cJ
only (right
andimmunized
i.p.
injected with
BrdU or PBS
indicated
or PBS at indicated time points as depicted in Fig. 1A. Single6cell
suspensions
of redatblood
cell6 time points as depicted in Fig. 1A. Single6cell suspensions of red blood cell6
depleted
cells
analyzed 8 weeks after immunization in flow cytometric analysis to determine the
es were analyzed 8 weeks after immunization in flow cytometric
analybone
sis to marrow
determine
thewere
frequency
Antigens, antibodies and detection reagents
Chicken γ globulin (CGG; Sigma-Aldrich) or Bovine serum albumin (BSA) was haptenated with
nitrophenyl (NP)-hydroxysuccinimide ester (Cambridge Research Biochemicals).
Allophycocyanin, Phycoerythrin and BSA were haptenated with nitro-iodo-phenyl (NIP) –
hydroxysuccinimide ester. The haptenation ratios of NP or NIP to proteins were determined by
spectrometry. NP-33-CGG was used for immunizations. The following reagents were prepared
and/ or conjugated in our laboratory: NIP5-BSA-Alexa Fluor 680; anti-IgM (B7-6) Alexa Fluor
680 and Pacific Blue Ab; anti-CD19 (1D3.2) Pacific Blue or Alexa Fluor 647 Ab; anti-CD80
(16-10A1) Alexa Fluor 488 Ab; Anti-B220 (RA3-6B2) Alexa Fluor 488 or Alexa Fluor 647;
anti-kappa (187.1) Pacific Blue; Alexa Fluor 647 or unconjugated anti-CD16/CD32 (2.4G2) Ab.
Unconjugated polyclonal Goat anti-lambda Ab was purchased from Southern Biotech and
conjugated to Alexa Fluor 680 and Pacific Blue in our laboratory. Unconjugated peanut
agglutinin (PNA) was purchased from Vector Laboratories and conjugated to Alexa Fluor 488 in
our laboratory. Anti-PDL2 (TY-25) biotin Ab, anti-CD38 (90) PE or biotin Ab were purchased
from Biolegend. Anti-IgG1 (A85-1) FITC or V450 Ab and anti-CD138 (281-2) biotin or PE Ab
were ordered from BD Biosciences. Anti-B220 (RA3-6B2) APC-Cy7 Ab, anti-CD19 (ID3)
APC-Cy7 Ab, anti-CD95 (Jo2) Pe-Cy7 Ab and anti-CD73 (TY/23) PE Ab were from BD
Pharmingen. Anti-BrdU (MoBU1) Alexa Fluor 647 Ab was from Invitrogen. Polyclonal antiIgG1-AP Ab was purchased from Southern biotech. Ab-producing cell lines 174.2 (rat IgM antimouse CD4) and 31-68.1 (rat IgM anti- mouse CD8) were grown in RPMI 1640 media to
generate antibody containing supernatants (Ceredig et al., 1985) for complement mediated cell
depletion.
3
BrdU detection by multicolor flow cytometry
1x107 red blood cell depleted splenocytes were incubated with anti-CD16/CD32 Abs and EMA
in staining buffer (SB; 1xPBS, 3%FCS, 2mM EDTA, 0.02% NaN3) for 15min on ice followed
by 10min exposure to fluorescent light. Cells were washed and stained for surface antigens for
30min on ice in SB and then washed in SB. Cell pellets were resuspended in 500µl cold 0.15M
NaCl and 1.2ml of pre-cooled 100% ethanol were dropwise added while vortexing gently and
incubated for 40min on ice. Cells were pelleted, washed once in SB and resuspended in 1ml 1%
PFA supplemented with 0.05% Tween20 and incubated for 30min at RT followed by overnight
incubation at 4°C. Cells were pelleted, washed once in SB and then resuspended in 1ml of 0.15M
NaCl, 4.2mM MgCl2 supplemented with 100Kunitz DNaseI (Sigma) for digestion of DNA. After
30min incubation at 37°C cells were pelleted and incubated for 20min on ice in 40µl SB with
10% rat- and 10% mouse serum (Equitech-bio, Inc) to block unspecific binding. 50µl SB with
anti-BrdU Alexa Flour 647 Ab and any Abs for intracellular staining were added and incubated
overnight at 4°C. Cells were washed twice in SB and analyzed at the LSRII cytometer.
EdU/ BrdU double detection by multicolor flow cytometry
1x107 red blood cell depleted splenocytes were incubated with anti-CD16/CD32 Abs and EMA
in SB for 15min on ice followed by 10min exposure to fluorescent light. Cells were washed and
surface-stained with anti-CD38 Alexa Fluor 680 Ab and anti-IgG1 V450 Ab for 30min at 4°C.
Cells were washed once in SB followed by an additional washing step in 1xPBS supplemented
with 1%BSA. Pellets were resuspended in 120µl fixative (component D of Click-it® EdU Flow
cytometry assay kit, Invitrogen) and incubated for 20min at RT and 15min at 4°C. 3ml PBS/BSA
were added and cells were pelleted. After one additional washing step in PBS/BSA cells were
4
resuspended in 2ml Perm/wash buffer (PW) of the click-it® EdU Flow cytometry assay kit and
incubated for 5min at RT. Cells were pelleted and resuspended at a final volume of 80µl in PW.
After 10min incubation at RT 400µl of the click-it reaction (prepared according to the
manufacturers instructions) was added to the cells. After 40min incubation at RT cells were
washed twice in PW followed by 2 additional washing steps in PBS/BSA and 2x SB. Cells were
stained with NIP-PE and anti-CD95 PE-Cy7 Ab in SB for 30min at 4°C and then washed with
3ml SB. Cell pellets were resuspended in 500µl cold 0.15M NaCl and 1.2ml of pre-cooled 100%
ethanol were dropwise added while vortexing gently and incubated for 40min on ice. Cells were
pelleted, washed once in SB and resuspended in 1ml of 0.15M NaCl, 4.2mM MgCl2
supplemented with 100Kunitz DNaseI. After 30min incubation at 37°C cells were pelleted and
incubated for 20min on ice in 45µl SB with 10% rat and 10% mouse serum. 60µl SB with antiB220 APC-Cy7 Ab and anti-BrdU Alexa Fluor 647 Ab were added and incubated overnight at
4°C. Cells were washed twice in SB and analyzed at the LSRII cytometer.
Gene expression profiling
mRNA was isolated from 3x105 FACS Aria sorted cell with an RNeasy Micro kit (Qiagen).
Biotinylated cRNA was generated with the Illumina TotalPrep RNA Amplification Kit (Life
Technologies) and was hybridized to Illumina MouseWG-6 v2.0 Expression BeadChip arrays at
the Yale Keck Microarray Facility. Data were analyzed with packages in software of the R
project for statistical computing. Raw expression data were normalized by the quantile method
provided by the lumi package in R/Bioconductor. Genes with different expression in the early
versus late GC subset were defined by two criteria: an absolute difference in expression of ≥1.7,
and a statistically significant change in expression as determined by false-discovery rate q < 0.05
5
(Alan Dabney and John D. Storey). qvalue: Q-value estimation for false discovery rate control. R
package version 1.38.0.)
6
SUPPLEMENTAL REFERENCES
Ceredig, R., Lowenthal, J., Nabholz, M., and MacDonald, H. (1985). Expression of interleukin-2
receptors as a differentiation marker on intrathymic stem cells. Nature 314, 98–100.
Prak, E.L., and Weigert, M. (1995). Light chain replacement: a new model for antibody gene
rearrangement. J Exp Med 182, 541–548.
Shlomchik, M.J., Zharhary, D., Saunders, T., Camper, S.A., and Weigert, M.G. (1993). A
rheumatoid factor transgenic mouse model of autoantibody regulation. Int Immunol 5, 1329–
1341.
Sonoda, E., Pewzner-Jung, Y., Schwers, S., Taki, S., Jung, S., Eilat, D., and Rajewsky, K. (1997).
B cell development under the condition of allelic inclusion. Immunity 6, 225–233.
7