Content uploaded by Mark Jay Shlomchik
Author content
All content in this area was uploaded by Mark Jay Shlomchik
Content may be subject to copyright.
Maintenance of the plasma cell pool is independent
of memory B cells
Anupama Ahuja*, Shannon M. Anderson
†‡
, Ashraf Khalil*, and Mark J. Shlomchik*
†§
Departments of *Laboratory Medicine and †Immunobiology, Yale University, New Haven, CT 06510
Communicated by Martin G. Weigert, University of Chicago, Chicago, IL, January 23, 2008 (received for review December 10, 2007)
Humoral memory to an antigen (Ag) is maintained for several
decades in the form of memory B cells and serum Ab. In fact, plasma
cells (PCs) that secrete Ab are known to be long-lived and could be
solely responsible for maintaining the long-lived Ab titers. Alter-
natively, it has been proposed that the PC compartment is main-
tained for long periods by the differentiation of memory cells into
long-lived PCs as a result of nonspecific stimulation. This model
predicts accelerated decay of PC numbers in the absence of mem-
ory cells for the same Ag. To address this prediction, we have
developed a mouse model system that combined the ability to
deplete B cells with the ability to detect Ag-specific memory and
PCs. After establishing an immune response, we depleted Ag-
specific memory B cells with an anti-hCD20 mAb and determined
the effect on the PC compartment over 16 weeks. Using a combi-
nation of surface markers, we demonstrated that memory B cells
remained depleted over the course of the experiment. However,
despite this absence of memory cells for an extended duration, PC
numbers in spleen and bone marrow did not decline, which
indicates that the PC compartment does not require a significant
contribution from memory B cells for its maintenance and instead
that PCs are sufficiently long-lived to maintain Ab titers over a long
period without renewal. This observation settles an important
controversy in B cell biology and has implications for the design of
vaccines and for B cell depletion therapy in patients.
CD20 兩antibody-forming cell 兩serum antibody 兩B cell depletion
Immune response to a T-dependent antigen (Ag) leads to
activation and differentiation of B cells into plasma cells (PCs)
and memory cells. It is thought that the combination of serum
Ab and memory B cells together provide long-lasting immunity
to a variety of pathogens. Memory B cells are known to be
long-lived, although they undergo homeostatic self-renewal (1–
3). However, how serum Ab is maintained for long periods is less
clear. In some cases, chronic exposure to Ag could be respon-
sible, although it is unlikely that it explains long-lived Ab to
transient Ags such as tetanus toxoid and lymphocytic chorio-
meningitis virus (LCMV) (4, 5). A second alternative is that
long-lived PCs are sufficient to maintain long-lived Ab titers for
a lifetime (6–8). Indeed, there is substantial evidence that PCs
have long half-lives, estimated to be 138 days in mice (9). A third
possibility is that memory cells renew the long-lived PC com-
partment on a regular basis (10). The stimulation of these
memory cells to differentiate could be stochastic, from cross-
reactive specific Ags or from nonspecific innate immune [e.g.,
Toll-like receptor (TLR)] signals. In support of this latter idea
are several pieces of evidence. First, the measured half-lives of
PCs suggest that they need to be renewed to maintain the very
long-standing Ab titers that can ensue after a single immuniza-
tion (11). Further, memory cells are more sensitive to TLR and
possibly B cell receptor signals and are prone to differentiate into
Ab-forming cells (AFCs) in response to a variety of stimuli (12,
13). Critically for this argument, Bernasconi et al. (10) found a
tight correlation between the frequency of circulating memory
cells specific for a given Ag and the serum Ab titers for that same
Ag in humans. They interpreted this finding in causal terms in
that the result was consistent with the notion that such memory
cells supported the PC compartment (which in turn secreted the
serum Ab). Otherwise, they argued, one would not expect to find
such a close relationship between memory cell frequency and
serum Ab level. However, it is quite difficult to test such a
proposition in humans.
If memory cells were to contribute to PCs on a regular basis,
then depletion of memory cells (but not PCs) should have a
significant effect on PC numbers and Ab titers. Anticipating such
a possibility, Slifka and Ahmed (7) had used sublethal irradiation
to eliminate memory B cells, which they demonstrated by a
clonal expansion assay. They were able to detect PCs and serum
Ab for ⬎1 year, although the levels declined continuously in the
interim; interestingly, this decline was more pronounced in the
irradiated compared with the nonirradiated mice. This decline
could be because irradiation is not a specific way to eliminate B
cells and had other undesirable effects, perhaps on bone marrow
(BM) stromal cells that provide PC niches and on PCs them-
selves. Thus, this experiment, although demonstrating long PC
half-lives, could be interpreted in favor of either model for
long-lived serum Ab maintenance.
The question of how the PC compartment turns over with
respect to memory B cells is not only of basic interest in B cell
biology but also is important in the context of B cell depletion
therapy that is being used for treatment of lymphoma and
autoimmune diseases (14, 15). In patients, B cell depletion has
been mainly accomplished by using rituximab, a chimeric mAb
to human CD20 (hCD20), which depletes mature B cells includ-
ing (CD27
⫹
) memory B cells (16, 17). The extent of depletion in
secondary lymphoid tissues is poorly understood, although sec-
ondary Ab responses are impaired after treatment (18 –20).
Preestablished Ab titers to certain vaccine Ags have not been
affected in the short-term in patients, although titers of several
autoantibodies decline after treatment (21, 22). Because the
source of such autoantibodies is unclear, as is the extent of
depletion of memory cells and other potential precursors (17,
23), it is not possible to make conclusions about memory cell and
PC relationships in patients at this time. Indeed, dissecting this
relationship further in patients will be difficult if not impossible
until the extent of memory cell depletion in lymphoid tissues can
be reliably assessed. Thus, a murine model is needed to explore
the question further. Anti-CD20-mediated B cell depletion
murine models developed by our group (22) and others (24, 25)
are ideal for such studies.
To address the role of memory B cells in PC maintenance, we
crossed our hCD20 transgenic (Tg) mice (22) with the B1-8
knockin mice (26). The B1-8 allele, in combination with endog-
enous V
1, confers specificity to the (4-hydroxy-3-nitrophenyl)
acetyl (NP) hapten, resulting in a population of ⬇2% B cells with
Author contributions: A.A., S.M.A., and M.J.S. designed research; A.A. and A.K. performed
research; A.A., A.K., and M.J.S. analyzed data; and A.A. and M.J.S. wrote the paper.
The authors declare no conflict of interest.
‡Present address: Department of Microbiology and Immunology, University of California,
San Francisco, CA 94143.
§To whom correspondence should be addressed. E-mail: mark.shlomchik@yale.edu.
© 2008 by The National Academy of Sciences of the USA
4802–4807
兩
PNAS
兩
March 25, 2008
兩
vol. 105
兩
no. 12 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0800555105
NP specificity. The hCD20 Tg mice express hCD20 exclusively
on B cells. Using the hCD20 Tg system, we can deplete all
populations of B cells, in particular memory B cells, using an
anti-hCD20 mAb. This treatment should not affect the PC pool,
which does not express hCD20 (27). The presence of the B1-8
allele allowed us to generate a robust population of NP-specific
memory B cells that was amenable to reproducible quantitation
upon immunization with NP-chicken gamma globulin (NP-
CGG). In contrast, immunization of WT mice leads to a variable
and small number of memory B cells (⬇40,000 per spleen), which
would be difficult to assess, particularly after depletion (28).
Furthermore, unlike previous studies that used functional assays
that depend on various external factors, we directly enumerated
the expanded Ag-specific cells, using IgG1 isotype expression,
BrdU labeling, and newly defined surface markers to identify
memory B cells (3). We found that depletion of memory B cells
for prolonged periods of time did not have any detectable effect
on the PC population, indicating that a continuous input from
memory B cells is not required for maintenance of PCs and
humoral memory for the duration and conditions studied.
Results
Experimental Design. The experimental design is depicted in Fig. 1.
To test the hypothesis that maintenance of PCs is independent of
memory B cells, we crossed hCD20 Tg mice (22) with B1-8 knockin
mice (26) to create hCD20 ⫻B1-8 Tg mice. The hCD20 ⫻B1-8 Tg
mice were immunized with NP-CGG. After allowing 15–19 weeks
for the memory B cell pool to become stably established (refs. 2 and
3 and as confirmed in the ‘‘predepletion cohort’’), we treated the
mice with anti-hCD20 mAb for 2 weeks to deplete B cells. Using this
strategy, we expected to deplete memor y B cells but not PCs
because memory B cells are CD20
⫹
, whereas PCs are not. We
confirmed this outcome in a subset of mice (‘‘depletion cohort’’).
Moreover, a cohort of the immunized mice was injected with BrdU
during the peak of germinal center response (day 9–day 15). With
this strategy, all of the cells that are dividing during the labeling
period take up BrdU (2, 3). However, once the labeling is stopped,
BrdU is lost from the cells that continue to divide and is retained
only by cells that stopped dividing. Therefore, remaining BrdU
⫹
cells ⬎15 weeks after labeling represent long-lived cells that were
generated during the initial response (Fig. 1). We then waited 16
weeks after the termination of depletion therapy to allow time for
the lack of putative memory cell input into the PC compartment to
manifest, potentially as reduced numbers of PCs in the depleted
cohort. Prior data (9) had suggested that this time frame should be
more than sufficient to observe detectable effects. We then ana-
lyzed the depleted mice for recovery of naı¨ve and memory B cells
16 weeks after depleting the memory B cells (‘‘recovery cohort’’)
and most importantly evaluated the effect of long-term absence of
memory B cells on AFCs.
Generation of Memory Cells and PCs in hCD20 ⴛB1-8 Tg Mice.
Immunization of hCD20 ⫻B1-8 double Tg mice led to a
significant increase in the frequency of NIP
⫹
/Kappa
⫺
(NIP
⫹
)B
cells (Fig. 2 a,Middle, Ctrl-Rx mice, and b) when analyzed 17–21
weeks later. Among NIP
⫹
B cells, there was an increase in the
frequency and numbers of NIP
⫹
/Kappa
⫺
/IgG1
⫹
(IgG1
⫹
) B cells
and NIP
⫹
/Kappa
⫺
/IgG1
⫺
/CD80
hi
(IgG1
⫺
) B cells (Fig. 2 a,
Right, Ctrl-Rx mice, and b) compared with the alum-treated (Fig.
2a, Alum mice) or unimmunized mice (data not shown). Because
the GC response has ended by this time (refs. 3 and 28; discussed
below), the IgG1
⫹
B cells are deemed ‘‘true’’ memory B cells.
Moreover, in agreement with a recent report on the identity of
memory B cells, a significant fraction (⬇45%) of IgG1
⫹
cells
express high levels of CD73 and CD80 (ref. 3 and data not
shown). Although the IgG1
⫺
subset includes some contaminat-
ing naı¨ve B cells, it also expanded as a result of immunization,
indicating that it is substantially comprised of memory B cells of
IgM and/or other non-IgG1 isotypes (3). This notion was further
confirmed by coexpression of CD73 and CD80 on the majority
of IgG1
⫺
cells (data not shown). Additionally, the presence of
mutations in V
1, a hallmark of memory, was confirmed in both
Ag-specific populations in the immunized mice (see Methods and
data not shown). These results were in agreement with published
studies (3, 29, 30). Also in accord with published data (2, 3), in
the mice that were labeled with BrdU during the peak of GC
response, ⬇4% of NIP
⫹
cells were BrdU
⫹
, and ⬇50% of them
were IgG1
⫹
(Fig. 2 c, Ctrl-Rx mice, and d) and CD80
hi
(data not
shown).
A stable pool of Ag-specific PCs, as revealed by the ELISPot
AFC assay, was also observed in immunized mice (Fig. 3a,
Ctrl-Rx mice). A phenotypic Ag-specific PC population, defined
as CD138
⫹
/intracellular NIP
hi
/surface NIP
int
/CD45
⫺
, was also
noted, comprising ⬇0.022% of total splenocytes 17–21 weeks
after immunization (Fig. 3b, Ctrl-Rx mice). It should be noted
that at this time point, only 0.2% of NIP
⫹
cells are PNA
hi
/CD95
⫹
(data not shown), indicating that, as reported (3, 28), the GC
response had essentially ceased, and few if any additional
memory cells or PCs were being generated.
Depletion of Ag-Specific Memory B Cells with Anti-hCD20 Ab Treat-
ment. After establishing the presence of Ag-specific memory
cells and PCs, a cohort of immunized mice was treated with 2 mg
per week of the murine anti-hCD20 mAb, 2H7, for 2 weeks, i.p.
(Fig. 1). Treatment with 2H7 resulted in ⬇94% depletion of
splenic B cells (Fig. 2 a, Rx mice, and b;P⫽0.0079) compared
with control mice treated with mouse gamma globulin (Fig. 2 a,
Ctrl-Rx mice, and b). This result is in accord with previous data
using similar systems (22, 24, 25). NIP
⫹
B cells were also
depleted to a comparable extent (90%, P⫽0.0079; Fig. 2a,Rx
mice, and b). Interestingly, anti-hCD20 treatment resulted in
almost complete depletion of both IgG1
⫹
(98%, P⫽0.0079) and
IgG1
⫺
(92%, P⫽0.0079) subsets of memory B cells (Fig. 2b).
Extensive depletion of Ag-specific BrdU
⫹
cells (97%, P⫽
0.0079) and IgG1
⫹
/BrdU
⫹
cells (98%, P⫽0.0079; Fig. 2 cand
d) confirmed these results. Thus, in our model system, the extent
of depletion of memory B cells was comparable with that of naı¨v e
Fig. 1. Experimental design for determining the effect of memory B cell
depletion on PCs. The hCD20 ⫻B1-8 mice were immunized with NP-CGG. After
15–19 weeks, the immunized mice were treated with anti-hCD20 mAb 2H7 for
2 weeks. During the recovery period of 16 weeks, the AFCs were allowed to
decay in the absence of any input from memory B cells. Instead of anti-hCD20
mAb, the control-treated mice (Ctrl-Rx) were given mouse gamma globulin.
Alum-treated and unimmunized mice received no further treatment. BrdU
was injected in a cohort of unimmunized, alum-treated, and NP-CGG-treated
mice every day from day 9 through day 15 (d9-d15) after immunization.
Cohorts of mice were analyzed 15–19 weeks after immunization (predeple-
tion cohort), immediately after anti-hCD20 therapy (depletion cohort), and 16
weeks after termination of therapy (recovery cohort) for numbers of total B
cells, memory B cells, and PCs.
Ahuja et al. PNAS
兩
March 25, 2008
兩
vol. 105
兩
no. 12
兩
4803
IMMUNOLOGY
B cells and possibly exceeded it, perhaps because of resistance of
certain naı¨ve B cell subsets, chiefly MZ B cells (25).
Short-Term Effect of B Cell Depletion Therapy on AFCs. Because most
if not all long-lived PCs are believed to down-regulate CD20
expression (27), we expected little if any immediate effect on these
cells. Indeed, B cell depletion caused no reduction in NP-specific
splenic AFCs in the ELISpot assay (Fig. 3a). Similarly, the average
number of splenic PCs, identified by FACS (Fig. 3b), was similar in
the treated mice (P⫽0.5476) immediately after depletion to that
in the control treated mice (Fig. 3 band c). Taken together, these
results demonstrate that initial depletion of B cells had minimal
effect on splenic PCs.
Long-Term Effect of B Cell Depletion on Memory Cells. Sixteen weeks
after termination of treatment, the numbers of total splenic B
cells (P⫽0.8279) and naı¨ve NIP
⫹
cells (P⫽0.0695) were
restored to levels similar to those in control-treated, immunized
mice (Fig. 4a). In contrast, depletion remained substantial, with
a 90% reduction of Ag-specific IgG1
⫹
cells compared with
controls (P⬍0.0001) and 65% depletion of Ag-specific IgG1
⫺
cells (P⫽0.0005; Fig. 4b), although the number of splenic
memory B cells in the recovery cohort was slightly higher than
the depletion cohort. Most likely, the apparently better recon-
stitution of the IgG1
⫺
B cell subset is attributable to inclusion in
the FACS gate of some naı¨ve cells, which recover almost
completely during this time period (see above). Furthermore,
NIP
⫹
/BrdU
⫹
cells (92% depletion; P⫽0.0043) and NIP
⫹
/
BrdU
⫹
/IgG1
⫹
cells (89% depletion; P⫽0.0043) remained
depleted (Fig. 4c). That memory cells were irreversibly depleted
was as expected but nonetheless rules out that the memory cell
Fig. 2. Depletion of memory B cells immediately after treatment with
anti-hCD20 mAb 2H7 (depletion cohort). (a) Representative FACS plots show-
ing the gating strategy for identification of memory B cells in alum-treated
(Alum, n⫽5; Top); immunized, control mouse gamma globulin-treated
(Ctrl-Rx, n⫽5; Center), and immunized, 2H7-treated (Rx, n⫽5; Bottom) mice.
The parent gates are shown on Top. (b) Total numbers of B cells, NIP⫹/Kappa⫺
cells, NIP⫹/IgG1⫹cells, and NIP⫹/IgG1⫺cells in spleens of Ctrl-treated (open
symbols) and 2H7-treated (filled symbols) mice. (c) Representative FACS plots
of NIP⫹cells from alum-treated, Ctrl-treated, and 2H7-treated mice stained
with IgG1 and BrdU. The percentages of total BrdU⫹(gate with thin line) and
BrdU⫹/IgG1⫹(gate with thick line) are shown. (d) Comparison of total num-
bers of NIP⫹/BrdU⫹and NIP⫹/BrdU⫹/IgG1⫹memory subsets in spleens of
Ctrl-treated (open symbols) and 2H7-treated (filled symbols) mice. (band d)
Horizontal lines represent median. **,P⬍0.01, Mann–Whitney Utest.
Fig. 3. Short-term effect of depletion of B cells on PCs (depletion cohort). (a)
Comparison of the number of NP-specific ELISPots per million splenocytes
from Ctrl-treated mice (open symbols, n⫽5) and 2H7-treated mice (filled
symbols, n⫽5). (b) Representative FACS plots showing phenotyping of PCs as
CD138⫹/intracellular NIPhi/surface NIPint/CD45⫺in the Ctrl-Rx (Upper) and 2H7
Rx (Lower) immunized mice. (c) Total number of PCs, identified by FACS, in
spleens of Ctrl-treated (open symbols) and 2H7-treated (filled symbols) mice.
(aand c) Horizontal lines represent median.
4804
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0800555105 Ahuja et al.
compartment can renew itself to any significant extent after
depletion.
Long-Term Effect of B Cell Depletion Therapy on AFCs. If memory
cells are required for continuous generation of PCs, a long-term
absence of memory cells should lead to reduced numbers of PCs.
To test this possibility, we evaluated the number of AFCs 16
weeks after completion of B cell depletion in immunized mice.
Strikingly, even after prolonged depletion of memor y B cells, the
number of splenic AFCs, determined by NP-specific ELISPot
assays, was indistinguishable from control-treated mice, in which
memory B cells remained (Fig. 4d;P⫽0.9100). This finding was
confirmed by the presence of similar numbers of intracellular
NIP
hi
/CD138
⫹
/surface NIP
int
/CD44
hi
/CD45
⫺
cells in the treated
and control-treated groups of mice (Fig. 4e;P⫽0.3530). Most
importantly, the numbers of long-lived AFCs in the BM of
treated mice were comparable with those of control mice, as
assessed by NP-specific ELISpot assay (Fig. 4f;P⫽0.2821) and
FACS (Fig. 4g;P⫽0.6623). These data indicate that the PC
compartments of both spleen and BM remain intact in the
absence of memory cells for extended periods of time, probably
because of their long-lived nature.
Discussion
Although the t1
兾
2of serum IgG is on the order of a few weeks,
serum titers of Ab are maintained long after exposure to an Ag.
This paradox is explained in part by constant secretion of Ab
from the long-lived PC pool. Some have suggested that this alone
is sufficient to explain the longevity of serum Ab titers after a
single exposure to Ag (6–8). Alternatively, because PCs appear
to have a finite life span, it has been suggested that their numbers
are maintained by continued differentiation from the memory B
cell pool, perhaps by nonspecific stimulation (10). Here, we
attempted to distinguish the two possibilities by depleting Ag-
specific memory B cells for extended periods and analyzing the
effect on PCs. Our data clearly show that a 2-week treatment of
hCD20 ⫻B1-8 Tg mice with anti-CD20 mAb resulted in
effective depletion of both mature and memory B cells, whereas
PCs were not significantly affected. Strikingly, numbers of PCs
did not decrease over 4 months, even in the absence of memory
B cells, indicating that the PCs do not need to be continuously
generated from memory B cells. Our data differ from a previous
analysis of LCMV-infected mice, which were irradiated to
deplete memory B cells; PC numbers in such mice declined
4-fold in 4 months (7). Given our results, this decline could be
attributed to the pleiotropic effects of radiation on other com-
ponents of immune system. In our experiments, we specifically
targeted B cells (including memory B cells) with anti-hCD20
mAb, minimizing any undesirable side effects. Another advan-
tage of our approach is that we directly enumerated memory B
cells using surface markers in FACS instead of functional assays
as in previous studies. Doing so ensures that our results are not
confounded by factors external to the memory cells per se. In this
regard, we were able to show depletion of not only IgG1
⫹
memory cells but also IgG1
⫺
memory cells that presumptively
Fig. 4. Long-term effect of depletion of B cells on memory B cells and PCs
(recovery cohort). (a–c) Total numbers of B cells and NIP⫹/Kappa⫺cells (a),
NIP⫹/IgG1⫹cells and NIP⫹/IgG1⫺cells (b), and NIP⫹/BrdU⫹cells and NIP⫹/
BrdU⫹/IgG1⫹cells (c) in spleens of alum-treated (open triangles, nⱖ3),
Ctrl-treated (open symbols, nⱖ5), and 2H7-treated (filled symbols, nⱖ5)
mice. (d) Comparison of the number of NP-specific ELISPots per million spleno-
cytes from alum-treated (open triangles, n⫽7), Ctrl-treated (open symbols,
n⫽16), and 2H7-treated (filled symbols, n⫽15) mice. (e) Total number of PCs,
identified by FACS, in spleens of alum-treated (open triangles, n⫽7), Ctrl-
treated (open symbols, n⫽16), and 2H7-treated (filled symbols, n⫽15) mice.
(f) Comparison of the number of NP-specific ELISPots per million BM cells from
alum-treated (open triangles, n⫽7), Ctrl-treated (open symbols, n⫽16), and
2H7-treated (filled symbols, n⫽15) mice. (g) Percentage of PCs, identified by
FACS, in the BM of alum-treated (open triangles, n⫽3), Ctrl-treated (open
symbols, n⫽5), and 2H7-treated (filled symbols, n⫽6) mice. (a–g) Horizontal
lines represent median. **,P⬍0.01; ***,Pⱕ0.0005, Mann–Whitney Utest.
Ahuja et al. PNAS
兩
March 25, 2008
兩
vol. 105
兩
no. 12
兩
4805
IMMUNOLOGY
included IgM
⫹
cells, now appreciated to be a substantial com-
ponent of murine (and human) memory (3, 16, 31). To do this,
we used CD80 expression, supplemented with CD73 expression
(data not shown), both of which are elevated on IgG and IgM
memory cells. Furthermore, the almost complete elimination of
NP-specific cells that were labeled with BrdU during the immu-
nization phase and that contained both IgM and IgG cells (ref.
3, Fig. 2d, and data not shown) argues that our results cannot be
explained by an unexpected failure to detect a memory cell
subset.
A potential explanation for sustained levels of PCs even upon
depletion of memory B cells could be an increase in levels of B
cell-activating factor of the TNF family (BAFF), which was seen
in humans as a result of B cell depletion (32, 33). BAFF and
APRIL (a proliferation inducing ligand) are known to promote
the survival of B cells and PCs, respectively (34). BAFF eleva-
tions alone, as would transiently occur after B cell depletion, are
unlikely to promote abnormal PC survival. Most likely, APRIL
is sufficient to promote PC survival because inhibition of BAFF
and APRIL led to reduction in BM PC numbers in a B cell
maturation Ag-dependent fashion (35), whereas inhibition of
BAFF alone did not affect long-lived PC levels and the Ab they
secreted (M. Tomayko, M. Cancro, and M.J.S., unpublished
observations). Whereas BAFF levels increase 2- to 3-fold during
rituximab-mediated depletion of B cells in lupus and rheumatoid
arthritis patients, APRIL levels actually decrease in lupus pa-
tients and remain unchanged in rheumatoid arthritis (32, 33).
Moreover, BAFF levels return to pretreatment levels with
recovery of B cells. In mice, B cell recovery after depletion
occurs in several weeks (25). These considerations together
suggest that it is unlikely that BAFF levels would play a
significant role in maintenance of PC pool in our work.
Over a 4-month period because we saw no reduction in PC
numbers even after averaging results from ⱖ15 individual mice
from each group, it is clear that the actual PC half-life is
substantially longer. This point holds regardless of whether
memory cells are present. For example, were the half-life to be
32 weeks, we should have seen at least a 30% decrement in PC
numbers in 16 weeks. Thus, it seems that in mice PCs likely are
able to live as long as the life of the animal, and the PC
compartment does not require any significant contribution from
memory cells to maintain serum Ab titers. Nonetheless, we
cannot rule out the possibility that a much longer follow-up
period would have revealed a very slow decay of PCs in depleted
mice.
Our data show that there is no need to invoke memory cell
recruitment for PC renewal. However, they do not rule out that
such recruitment can occur under certain circumstances, for
example, after PC depletion (although a physiologic situation for
this has yet to be defined). It has been suggested that a small
fraction of memory B cells divide every day (1, 36). The reported
slow decay of BrdU
⫹
memory cells in the face of stable overall
numbers of these cells (3) and as also seen in our control-treated
mice, is most consistent with homeostatic memory cell renewal
resulting in dilution of the label over time. This renewal has been
suggested to be consistent with some differentiation of memor y
cells into AFCs in vivo (10). Moreover, in vitro data suggest that
dividing memory cells can differentiate into PCs upon nonspe-
cific stimulation or bystander T help. Therefore, it is possible that
stimulation of TLR4 and TLR9 may result in proliferation and
differentiation of memory cells even under physiological condi-
tions in vivo. However, such nonspecific stimulation of memory
cells in WT mice has recently been shown to generate short-lived
plasmablasts without their recruitment to the long-lived PC
compartment in BM (37). In fact, the Ag-specific long-lived PC
pool was shown, if anything, to contract as a consequence of
nonspecific stimulation, which suggests that spontaneous turn-
over of memory cells is unlikely to be a major factor in
preservation of serum Ab titers.
It is well established that long-lived PCs play an important role
in sustaining humoral memory. However, the total number of
‘‘survival niches’’ in BM is thought to be constant and is
estimated to support a maximum of 10
6
PCs per mouse (38).
Moreover, to maintain protective serum Ab titers, the repertoire
of humoral memory is estimated to include ⬇1,000 different
specificities. Therefore, to explain how additional long-lived PCs
can be generated with each new infection, a model has been
proposed in which old PCs are displaced from BM at a rate slow
enough not to lose the old specificities (38, 39). The newly
generated plasmablasts are then recruited into the long-lived PC
pool caused by the opening of the niches. In our experiments, we
did not attempt to address this issue, which although related, is
independent from the relationship between memory cells
and PCs.
Unexpectedly, during the course of the experiment, we noticed
an increase in total AFC numbers in the NP-specific ELISpot of the
immunized mice as time progressed. Total splenic AFC numbers in
both control treated (open symbols) and anti-hCD20 treated (filled
symbols) immunized mice were ⬇4-fold higher at the recover y time
point (Fig. 4d) compared with the depletion time point (Fig. 3a).
Considering that, as just demonstrated, differentiation from mem-
ory cells does not increase PC numbers, a possible explanation is
that not all of the PCs were fully mature at the depletion time point
(even though we waited 15–19 weeks after immunization), and as
time progressed, more PCs matured. Alternatively, it is possible that
developed PCs divide at a very slow rate at least for some time after
initial formation, which could explain the increased numbers.
The finding that differentiation of memory B cells into PCs is
not a major pathway for maintenance of the PC pool implies that
the immune system has evolved two independent strategies to
maintain long-term memory: PCs and memory B cells. Each has
its own signals for formation and its own signals and niches for
survival (39). This strategy represents a more robust defense
against immunosuppressive interventions, whether pathogen-
mediated or iatrogenic, than if the PC compartment were to rely
on the memory compartment for its renewal. However, in the
case of Ag-specific restimulation, the memory compartment
could potentially contribute further to the long-term PC com-
partment, thus shaping it further, although the main product of
restimulation of memory cells is a short-lived AFC (37).
With regard to clinical application, our data suggest that
depletion of memory B cells with rituximab should have no effect
on long-lived PC numbers, which is in agreement with the fact
that total serum IgG and Abs to microbes and vaccine Ags either
decline slowly or not at all even after prolonged B cell depletion
(21, 22). To the extent that some decay of Ig levels occurs with
B cell depletion, it could be attributed to a component derived
from short-lived plasmablasts (which are reliant on CD20
⫹
precursors for renewal), a limit on the life span (albeit a long
one) of established PCs, and/or an effect of niche displacement
caused by subsequent immunization (39). The maintenance of
the serum Ab component of humoral memory despite B cell
depletion could account for the low infectious complication rate
in chronically depleted individuals. Indeed, the lack of infections
in such B cell-depleted individuals highlights the two-pronged
and independent strategy of the immune system to achieve
long-term humoral immunity.
Methods
Mice and Immunization. The human CD20 (hCD20) Tg (22) and B1-8 knockin
mouse strains (26) on BALB/c backgrounds were crossed to generate hCD20 ⫻
B1-8 Tg mice. The F1 mice from these crosses were immunized i.p. at 8–15
weeks of age with either 50
g of alum-precipitated NP-CGG or alum precip-
4806
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0800555105 Ahuja et al.
itate alone as control. All animal experiments were approved by the Yale
Institutional Animal Care and Use Committee.
FACS Analysis. For FACS analysis, NIP-haptenated PE (NIP-PE), and APC (NIP-
APC) reagents and mAbs against murine CD80 (1610A1), Kappa (187.1), and
CD44 (Pgp-1) were prepared as described in ref. 40. Abs against IgG1 (A85-1),
CD73 (Ty/23), CD19 (1D3), CD138 (281-2), CD45 (RA3-6B2), CD95 (Jo-2) were
purchased from BD Biosciences. Abs against BrdU (PRB-1; Molecular Probes),
Lambda (Southern Biotech), streptavidin (SA)-PE/Cy7 (eBiosciences), and PNA-
FITC (Vector Laboratories) were also purchased. Samples were stained and
analyzed on LSRII or FACSAria (BD Biosciences).
Immunodepletion. The mAb against hCD20, 2H7, was used for B cell depletion.
2H7 is a mouse IgG2b that binds an epitope on hCD20 similar to that bound by
rituximab and therefore, mimics rituximab treatment of humans. 2H7 was
purified from culture supernatants (the hybridoma was a generous gift from
E. Clark, University of Washington, Seattle) as described in ref. 22. Mice were
injected i.p. with 2 mg per week of either 2H7 or control mouse IgG (Rockland
Immunochemicals) in sterile PBS for 2 weeks, administered twice a week.
BrdU Administration and Detection. A cohort of immunized and alum-treated
mice were labeled with BrdU by injecting 3 mg/ml BrdU (Sigma–Aldrich) i.p.,
every 24 h from day 9 through day 15 after immunization. BrdU was detected,
at the time of sacrifice, as described in ref. 41.
ELISPot. NP-specific AFCs were quantitated by ELISpot assay as described in ref.
40, with plates coated with NP26-BSA and visualized with polyclonal anti-
-
alkaline phosphatase Ab (Southern Biotech) and bromo-4-chloro-3-indolyl
phosphate substrate (AMRESCO).
Sorting and Sequence Analysis. To analyze for presence of mutations in
chain,
the memory subsets were sorted and sequenced as described in ref. 3.
Statistical Analysis. The Mann–Whitney two-tailed Utest was used to deter-
mine statistical significance.
ACKNOWLEDGMENTS. We thank Dr. Klaus Rajewsky for B1-8 knockin mice;
Jinping Wang, Jonathan Shupe, Cuiling Zhang, and Haowei Wang for expert
technical assistance; Anja Hauser and Kim Good for helpful discussions; Geoff
Lyon of the Yale Cell Sorting Facility for help with sorting; and Michelle
Horniak, Terrence Hunt, and the rest of the staff of the Yale Animal Resource
Center for excellent animal care. This work was supported by National Insti-
tutes of Health Grants AI43603 and AR44077 (to M.J.S.). A.A. was supported
by an Arthritis Foundation Postdoctoral Fellowship.
1. Schittek B, Rajewsky K (1990) Maintenance of B-cell memory by long-lived cells
generated from proliferating precursors. Nature 346:749–751.
2. Anderson SM, Hannum LG, Shlomchik MJ (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.
3. Anderson SM, Tomayko MM, Ahuja A, Haberman AM, Shlomchik MJ (2007) New
markers for murine memory B cells that define mutated and unmutated subsets. J Exp
Med 204:2103–2114.
4. Gatto D, et al. (2007) Regulation of memory antibody levels: The role of persisting
antigen versus plasma cell life span. J Immunol 178:67–76.
5. Zinkernagel RM, et al. (1996) On immunological memory. Annu Rev Immunol 14:333–
367.
6. Manz RA, Radbruch A (2002) Plasma cells for a lifetime? Eur J Immunol 32:923–927.
7. Slifka MK, Ahmed R (1998) Long-lived plasma cells: A mechanism for maintaining
persistent antibody production. Curr Opin Immunol 10:252–258.
8. Slifka MK, Matloubian M, Ahmed R (1995) Bone marrow is a major site of long-term
antibody production after acute viral infection. J Virol 69:1895–1902.
9. Slifka MK, Antia R, Whitmire JK, Ahmed R (1998) Humoral immunity due to long-lived
plasma cells. Immunity 8:363–372.
10. Bernasconi NL, Traggiai E, Lanzavecchia A (2002) Maintenance of serological memory
by polyclonal activation of human memory B cells. Science 298:2199–2202.
11. Traggiai E, Puzone R, Lanzavecchia A (2003) Antigen-dependent and -independent
mechanisms that sustain serum antibody levels. Vaccine 21 (Suppl 2):S35–S37.
12. Bernasconi NL, Onai N, Lanzavecchia A (2003) A role for Toll-like receptors in acquired
immunity: Up-regulation of TLR9 by BCR triggering in naive B cells and constitutive
expression in memory B cells. Blood 101:4500– 4504.
13. Raman VS, Lind EF, Benson MJ, Noelle RJ (2007) Strategies for selective priming of
memory B cells. Immunol Lett 109:93–100.
14. Sanz I, Anolik JH, Looney RJ (2007) B cell depletion therapy in autoimmune diseases.
Front Biosci 12:2546–2567.
15. Grillo-Lopez AJ (2002) Monoclonal antibody therapy for B cell lymphoma. Int J Hema-
tol 76:385–393.
16. KleinU, Rajewsky K, Kuppers R (1998) Human immunoglobulin (Ig)M⫹IgD⫹peripheral
blood B cells expressing the CD27 cell surface antigen carry somatically mutated
variable region genes: CD27 as a general marker for somatically mutated (memory) B
cells. J Exp Med 188:1679–1689.
17. Leandro MJ, Cambridge G, Ehrenstein MR, Edwards JC (2006) Reconstitution of pe-
ripheral blood B cells after depletion with rituximab in patients with rheumatoid
arthritis. Arthritis Rheum 54:613–620.
18. van der Kolk LE, Baars JW, Prins MH, van Oers MH (2002) Rituximab treatment results
in impaired secondary humoral immune responsiveness. Blood 100:2257–2259.
19. Bearden CM, et al. (2005) Rituximab inhibits the in vivo primary and secondary
antibody response to a neoantigen, bacteriophage
X174. Am J Transplant 5:50–57.
20. Gonzalez-Stawinski GV, Yu PB, Love SD, Parker W, Davis RD, Jr (2001) Hapten-induced
primary and memory humoral responses are inhibited by the infusion of anti-CD20
monoclonal antibody (IDEC-C2B8, rituximab). Clin Immunol 98:175–179.
21. SfikakisPP, Boletis JN, Tsokos GC (2005) Rituximab anti-B-cell therapy in systemic lupus
erythematosus: Pointing to the future. Curr Opin Rheumatol 17:550–557.
22. Ahuja A, et al. (2007) Depletion of B cells in murine lupus: Efficacy and resistance.
J Immunol 179:3351–3361.
23. Anolik JH, et al. (2004) Rituximab improves peripheral B cell abnormalities in human
systemic lupus erythematosus. Arthritis Rheum 50:3580–3590.
24. Uchida J, et al. (2004) The innate mononuclear phagocyte network depletes B lym-
phocytes through Fc receptor-dependent mechanisms during anti-CD20 antibody
immunotherapy. J Exp Med 199:1659–1669.
25. GongQ, et al. (2005) Importance of cellular microenvironment and circulatory dynam-
ics in B cell immunotherapy. J Immunol 174:817–826.
26. TakiS, Meiering M, Rajewsky K (1993) Targeted insertion of a variable region gene into
the immunoglobulin heavy chain locus. Science 262:1268–1271.
27. Uchida J, et al. (2004) Mouse CD20 expression and function. Int Immunol 16:119–129.
28. Takahashi Y, Ohta H, 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.
29. McHeyzer-Williams MG, Nossal GJV, Lalor PA (1991) Molecular characterization of
single memory B cells. Nature 350:502–505.
30. SmithKG, Nossal GJ, Tarlinton DM (1995) FAS is highly expressed in the germinal center
but is not required for regulation of the B cell response to antigen. Proc Natl Acad Sci
USA 92:11628–11632.
31. White H, Gray D (2000) Analysis of immunoglobulin (Ig) isotype diversity and IgM/D
memory in the response to phenyl-oxazolone. J Exp Med 191:2209–2220.
32. CambridgeG, et al. (2006) Circulating levels of B lymphocyte stimulator in patients with
rheumatoid arthritis after rituximab treatment: Relationships with B cell depletion,
circulating antibodies, and clinical relapse. Arthritis Rheum 54:723–732.
33. Vallerskog T, et al. (2006) Differential effects on BAFF and APRIL levels in rituximab-
treated patients with systemic lupus erythematosus and rheumatoid arthritis. Arthritis
Res Ther 8:R167.
34. Mackay F, Schneider P, Rennert P, Browning J (2003) BAFF and APRIL: A tutorial on B
cell survival. Annu Rev Immunol 21:231–264.
35. O’Connor BP, Cascalho M, Noelle RJ (2002) Short-lived and long-lived bone
marrow plasma cells are derived from a novel precursor population. J Exp Med
195:737–745.
36. Macallan DC, et al. (2005) B cell kinetics in humans: Rapid turnover of peripheral blood
memory cells. Blood 105:3633–3640.
37. Xiang Z, et al. (2007) Fc
␥
RIIb controls bone marrow plasma cell persistence and
apoptosis. Nat Immunol 8:419– 429.
38. Hofer T, et al. (2006) Adaptation of humoral memory. Immunol Rev 211:295–302.
39. Radbruch A, et al. (2006) Competence and competition: The challenge of becoming a
long-lived plasma cell. Nat Rev Immunol 6:741–750.
40. Hannum LG, Ni D, Haberman AM, Weigert MG, Shlomchik MJ (1996) A disease-related
rheumatoid factor autoantibody is not tolerized in a normal mouse: Implications for
the origins of autoantibodies in autoimmune disease. J Exp Med 184:1269–1278.
41. Allman DM, Ferguson SE, Lentz VM, Cancro MP (1993) Peripheral B cell maturation. II.
Heat-stable antigen(hi) splenic B cells are an immature developmental intermediate in
the production of long-lived marrow-derived B cells. J Immunol 151:4431–4444.
Ahuja et al. PNAS
兩
March 25, 2008
兩
vol. 105
兩
no. 12
兩
4807
IMMUNOLOGY
