| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Departments of Medicine, Molecular Biology and Genetics, and Oncology, Division of Rheumatology, Johns Hopkins University School of Medicine, Baltimore, MD 21205
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
or 
) gene rearrangement.
Pro-B cells that generate a productive (in-frame) H chain gene assemble
a signaling complex known as the pre-BCR (4). The pre-BCR
signals the pre-B cell to activate L chain gene rearrangement and
inactivate further H chain gene rearrangement, but it is unclear
whether the pre-BCR signal requires extracellular ligand (5, 6). If a pre-B cell generates a productive L chain gene
rearrangement, it uses the resultant Ig to assemble its BCR. The
subsequent stages of B cell maturation can be defined by the expression
of surface Ig (sIg). Immature B cells express sIgM, but not sIgD, while
mature B cells and most peripheral B cells express both sIgM and
sIgD. There are two subsets of peripheral B cells called B-2 and B-1 (reviewed in Ref. 7). These subsets express distinct Ig repertoires, presumably reflecting their different functions in the immune system (8). B-2 cells are the predominant B cell population in the blood, spleen, and lymph nodes. B-1 cells are largely restricted to the peritoneum and other body cavities and comprise the majority of B cells in these areas. B-1 cells have the capacity for self-renewal and are responsible for secreting most of the IgM present in the serum of unimmunized animals (9). In fact, a large fraction of the B-1 subset is composed of cells with measurable affinity for self Ags with repetitive structures, such as phosphatidylcholine (PC), Ig (rheumatoid factor), and DNA, as well as common bacterial carbohydrate Ags (9, 10, 11, 12, 13). Thus, it has been suggested that developing B cells bearing self-specific Ag receptors are actively selected into the B-1 cell repertoire (14), and that mature B-1 cells may be involved in T cell-independent responses to common environmental Ags. Recent evidence supports a model in which B-2 and B-1 cells derive from a single lineage, and that an uncommitted B cell (B-0) is induced to differentiate to a B-1 cell through interaction with T cell-independent Ags in the absence of T cell help (14, 15).
Although the concept of positive selection of T cells developing in the thymus is well established, only recently has the idea that a parallel event occurs during B cell development gained similar acceptance (1, 2). Thymic T cells receive a positive selection signal when their Ag receptor is "tickled" by a low intensity interaction with MHC plus self Ag, but undergo apoptosis if this interaction is too strong (reviewed in Refs. 1 and 16). Tickling of the TCR on mature T cells by MHC+ self Ag in the periphery is thought to enable the survival of memory T cells (17). Memory B cells might similarly survive if the BCR is tickled by the same self Ags responsible for their emigration from the bone marrow. There are three basic checkpoints at which developing B cells undergo selection (1). First, as noted above, pro-B cells fail to survive unless they generate a productive H chain gene rearrangement capable of pairing with surrogate L chains to form a pre-BCR (18, 19). There is evidence that the structure of the H chain variable domain can influence the outcome of pre-BCR signaling (19, 20, 21, 22). Second, at a later stage of development, immature B cells expressing a self-specific BCR are induced to edit that receptor by reactivation of the V(D)J recombinase, to undergo apoptosis, or to become anergic (23, 24, 25). Those cells with useful BCRs are positively selected to emigrate to the periphery. Finally, the earliest B cell emigrants from the bone marrow arrive in the spleen, where they enter follicles and acquire the ability to join the recirculating pool. BCR-mediated signaling is required for the maintenance (selection) of peripheral B cells (26).
One important subset of mature B cells resides in the marginal zone (MZ) of the spleen, a site bridging the red and the white pulp, where blood filters past macrophages, B cells, and dendritic cells before reaching the red pulp. MZ B cells are both phenotypically and functionally distinct from follicular B cells, expressing higher levels of complement receptors (CR1/2; CD21), but little or no CD23. MZ B cells are more sensitive to B cell mitogens than follicular B cells and differentiate extremely rapidly into Ab-secreting cells upon encounter with Ag (reviewed in Ref. 27). There is strong evidence that an appropriate quality BCR signal is required for B cells to become MZ cells, and that these B cells play an important role in T-independent Ab responses and in Ag retention in the MZ (28, 29). There is also evidence that recruitment to the MZ is a competitive process (30), and that, as for the B-1 compartment, it is dependent upon BCR H and L chain composition (28, 31). Thus, B cell clones are thought to join the long-lived MZ repertoire based on their BCR specificity and the nature of individual BCR-mediated encounters, although it is not yet clear whether this process requires a particular ligand.
Transgenic (Tg) model systems have clearly demonstrated that most self-reactive B cells failing to successfully edit their receptors are negatively selected at an immature stage of development (32, 33), analogous to thymocytes expressing a self-reactive TCR (34). However, the concept that an Ig-dependent positive selection event might also occur within the B cell compartment, either via a cell autonomous mechanism or mediated by ligand(s) in the BM microenvironment, is still a matter of debate (1, 2). At least two lines of evidence suggest that B-2 cells are positively selected. First, DNA sequence analyses have demonstrated that the Ig repertoires of pre-B, immature B, and mature peripheral B cells in normal mice are distinct. While the use of a broad array of VH genes characterizes B cells at early stages of development, mature splenic B cells are more restricted in their VH gene usage, suggesting that the expansion of Ig-expressing B cells is ligand dependent (35, 36, 37). Second, only a small proportion (10–20%) of newly formed B cells populate the peripheral lymphoid organs, indirectly suggesting that only actively selected cells migrate from the bone marrow or that only a subset of B cells receive signals enabling their survival in the periphery (38, 39, 40).
Evidence for Ag-driven positive selection of cells into the B-1 cell
lineage has been even more convincing. In normal unimmunized mice
PC-specific B cells appear to be exclusively B-1. They are driven to
expand from birth, and eventually account for 2–10% of the peritoneal
B-1 repertoire in normal adults. IgM specific for PC is encoded
predominantly by either of two combinations of VH
and VL: VH11 and
V9 (41) or
VH12 and V4
(42). In the case of
VH12+ PC-reactive B-1
cells, the VH12 H chain is further characterized
by a complementarity-determining region 3 (CDR3) that is invariably 10
aa in length (42). In mice carrying a PC-specific
VH12-DSP2.9-JH1 H chain
transgene, large numbers of
V4+
transgene+ PC-specific cells populate the spleen
and peritoneum, resembling double
(VH12/V4) Tg mice
provided with both H and L chain genes (43). This
similarity between H and H+L chain Tg mice lends credence to the model
that VH12+ B cells are
positively selected into the B-1 cell lineage. Positive selection of
self-reactive B cells was also observed in a transgenic mouse
expressing an IgH chain gene conferring reactivity to Thy-1
(44).
We recently reported some unexpected results arising from studies aimed at identifying B cells undergoing negative selection in the bone marrow by using the apoptosis marker, annexin V (AnV) (45). AnV is a member of a large family of Ca2+ and phospholipid binding proteins (46), and in the presence of physiological concentrations of Ca2+, AnV has high affinity for negatively charged phospholipids, especially phosphatidylserine (PS). The plasma membrane of a healthy cell typically exhibits an asymmetric distribution of its major phospholipids maintained via the activity of aminophospholipid translocase (47). Virtually all the PS and most of the phosphatidylethanolamine (PE) and phosphatidylinositol reside on the inner leaflet of the plasma membrane, with sphingomyelin largely confined to the outer leaflet, and PC distributed equally between both leaflets (48). During the early stages of apoptosis, cells lose their membrane phospholipid asymmetry and expose PS on the outer leaflet of the plasma membrane, triggering their phagocytosis by macrophages that bear PS receptors. Rapid phagocytosis prevents secondary necrosis and inflammation within the surrounding tissue (reviewed in Refs. 49 and 50).
Fluorescently labeled forms of recombinant AnV (rAnV) have been used previously in flow cytometry to identify apoptotic cells (51). We used FITC- or biotin-rAnV to stain BM cells from Tg mice bearing a potentially autoreactive BCR (32). Surprisingly, we found that rAnV stains nearly all the Tg BM B cells in mice on a background lacking the autoantigen, but at levels somewhat lower than those on apoptotic cells. rAnV-binding B cells are also present in nontransgenic (nonTg) mice among both B-1 and B-2 cell populations and do not display any of the other classic physiological changes associated with apoptosis (45). Instead, we show here that rAnV binding to the surface of B cells, both early and late in their developmental progression, correlates with BCR-dependent positive selection events.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Tg mice expressing the Ig H+L chain genes from the 3-83 BCR that confers reactivity to the mouse MHC class I Ags H-2Kb and -Kk (52) were provided by Dr. David Scott (American Red Cross, Rockville, MD) and maintained on an H-2d background by continuous backcrosses to BALB/c mice (National Cancer Institute, Frederick, MD). To obtain H-2bxd and H-2kxd Tg mice, H-2d Tg males were bred to either C57BL/6 (H-2b) or B10.A (H-2k) females obtained from The Jackson Laboratory (Bar Harbor, ME). IgH Tg mice (53) were obtained from Drs. Michel Nussenzweig (Rockefeller University, New York, NY) and Phil Leder (Harvard University, Boston, MA). VH12 Tg mice (43), which carry a rearranged Ig H chain gene from a PC-specific B-1a clone, were provided by Dr. Stephen Clarke (University of North Carolina, Chapel Hill, NC). VH12 Tg mice were maintained as heterozygotes by continuous backcrosses to C57BL/6 mice. Control nonTg mice were offspring from these same matings. Mice were maintained in specific pathogen-free conditions in our facility, and unless otherwise noted were used in experiments at 4–20 wk of age.
Reagents
Recombinant human AnV was cloned by PCR from a human placental cDNA library, expressed in Escherichia coli, then purified and conjugated to FITC or biotin as previously described (54). Our reagent gave results identical with those using commercially prepared versions of FITC-rAnV (Clontech, Palo Alto, CA) or biotin-rAnV (PharMingen, San Diego, CA). TUNEL assays were performed with a kit (Roche, Indianapolis, IN), according to the manufacturer’s instructions. 7AAD and Quantum Red (QR)-streptavidin (SA) were purchased from Sigma (St. Louis, MO). Anti-B220-PE (clone RA3-6B2), anti-CD5-PE (clone 53-7.3), anti-CD8-FITC (clone 53-6.7), anti-CD4-PE (clone GK1.5), and anti-CD19-biotin (clone 1D3) mAbs were obtained from PharMingen (San Diego, CA). Goat anti-mouse IgM-PE and goat anti-mouse IgD-biotin were purchased from Southern Biotechnology Associates (Birmingham, AL). Anti-CD43-biotin (clone S7) and anti-B220-biotin (clone RA3-6B2) were prepared in our laboratory. Both FITC- and biotin-conjugated versions of 54.1, the mAb that recognizes the 3-83 Id, were gifts from Terri Grdina (American Red Cross).
Cell lines
The H-2d stromal cell line S17 (55) was obtained from Dr. Ken Dorshkind (University of California, Los Angeles, CA), and op42 (56), an H-2Kb-bearing stromal cell line, was provided by Dr. Paul Kincade (Oklahoma Medical Research Foundation, Oklahoma City, OK). Both stromal cell lines were cultured at 37°C in a 5% CO2 incubator in RPMI 1640 supplemented with 10% FBS, 4 mM glutamine, 10 mM HEPES, and antibiotics (RP10).
Cell preparation and purification
BM cells were isolated from femurs and tibias by careful disruption in PBS using a mortar and pestle. Cells were resuspended, depleted of bone fragments by passive sedimentation, and pelleted at 1000 x g. Splenocytes were obtained by crushing spleens between glass slides, then resuspending and pelleting as described for BM. Peritoneal cavity washout cells (PerC) were obtained by repeatedly flushing the peritoneum with PBS and collecting the cells in a 5-cc syringe with an 18-gauge needle. To avoid the potential loss or gain of rAnV+ cells, BM and spleen samples were not depleted of RBC by Ficoll treatment or hypotonic lysis; instead, RBC and dead cells were gated out electronically after flow cytometric analysis. All cells were resuspended in FACS wash buffer (FACS WB; HBSS, 1% BSA, and 10 mM HEPES buffer, pH 7.4) at a concentration of 20–30 x 106 c/ml before staining. To enrich for B cells, BM or spleen preparations were stained with anti-CD19-biotin followed after a wash by SA-conjugated magnetic beads, then passed over a MiniMacs separation column, according to the manufacturer’s instructions (Miltenyi Biotec, Sunnyvale, CA).
Cell staining and flow cytometry
For staining with FITC-rAnV, 1–1.5 x 106 cells were transferred to 5-ml tubes and washed with 1 ml of FACS WB and 2 mM CaCl2 or with WB and 2 mM EGTA, then pelleted at 1000 x g. Cells were then incubated on ice for 20 min in the presence of saturating amounts of the appropriate PE- and/or biotin-conjugated mAbs and FITC- or biotin-rAnV in a total volume of 100 µl of FACS WB with Ca2+ or EGTA. Cells were washed with 1.5 ml of WB and Ca2+ or EGTA, pelleted, then resuspended in 100 µl WB and Ca2+ or EGTA containing a saturating amount of QR-SA. After another 20-min incubation on ice, the cells were washed and pelleted as before, then resuspended in 0.5 ml of WB and Ca2+ or EGTA and analyzed on a FACScan using CellQuest software (Becton Dickinson, Mountain View, CA). Detectors for forward (FSC) and side (SSC) light scatter were set on a linear scale, whereas logarithmic detectors were used for all three fluorescence channels (FL-1, FL-2, and FL-3), except for cell cycle analyses, where FL-3 was used on a linear scale. Compensation for spectral overlap between FL channels was performed for each experiment using single-color-stained cell populations. Wherever possible, instrument settings were saved to disk and used again with slight modifications if necessary in related experiments. All data were collected ungated to disk and were analyzed using CellQuest software. Unless otherwise noted, RBC and dead cells were excluded by electronically gating data on the basis of FSC vs SSC profiles; a minimum of 104 cells of interest were analyzed further. Cell sorting was performed at 4°C using a FACStarPlus (Becton Dickinson) with LYSYS II or CellQuest software or using a Coulter EPICS Elite flow cytometer (Miami, FL). To ensure retention of rAnV binding throughout the sort and reanalysis, CaCl2 was added to the sheath tank to a final concentration of 2 mM before sorting. To avoid the formation of calcium/phosphate precipitates during sorting, the standard PBS-based sheath fluid was replaced with HBSS (lacking phenol red) buffered with 10 mM HEPES. Approximately 103 sorted cells of each type were reanalyzed for purity. TUNEL staining of BM cell cultures was performed using a kit (Roche) after surface labeling the cells with anti-B220-PE and rAnV-biotin plus QR-SA, followed by fixation and permeabilization, according to the manufacturer’s instructions. For cell cycle analyses, BM cells were first stained with FITC-rAnV and anti-B220-PE, then fixed and permeabilized as for TUNEL staining, and incubated for 15 min on ice with 10 µg/ml 7-AAD before analysis.
Stromal cell cultures
To determine the relative viability of purified B cells upon culture with stromal cells, unseparated or column-enriched CD19+ nonTg or 3-83 Ig Tg BM B cells were cultured for 40–48 h on plates with the adherent H-2d stromal cell line S17 (55) or the H-2Kb stromal cell line op42 (56). Stromal cells were plated 2 days before use at a density of 3–6 x 104 cells/well in 24-well plates in RP10 and were essentially confluent when the B cells were added. After culture, the B cells were harvested by gentle resuspension, counted, and stained with FITC-rAnV, anti-B220-PE, and 7AAD or were stained with anti-B220-PE and subjected to the TUNEL assay. Stained cells were analyzed by flow cytometry; collected data were later reanalyzed by gating on B220+ cells.
In vitro stimulation of splenocytes
Spleen cells from C57BL/6 mice were cultured in 24-well flat-bottom plates for 2 days in 2 ml of RP10 alone or in RP10 supplemented with 25 µg/ml LPS (Sigma), 25 µg/ml LPS and 10 ng/ml rmIL-4 (R&D Systems, Minneapolis, MN), 5 µg/ml Con A (Vector, Burlingame, CA), or 0.5 ng/ml PMA and 500 ng/ml ionomycin (Calbiochem, La Jolla, CA). Two days later, cells were harvested, washed twice with HBSS and 1% BSA, and stained as described for flow cytometry.
PCR-FLP
DNA was isolated from sorted cells by lysing them in the presence of SDS and EDTA, incubating overnight with proteinase K, adding 20 µg of glycogen as carrier, and performing one phenol:CHCl3 and one CHCl3 extraction, followed by precipitation in ethanol/0.3 M ammonium acetate. For reading frame usage analysis of sorted pro-B cell populations, DNA from 1–4 x 104 cells was first amplified (20 cycles of 94°C for 1 min and 66°C for 2.5 min, followed by one cycle of 72°C for 10 min) with a primer downstream of JH3 (JHA, 5'-TGCCTCAGACTTCAAGCTTCAGTTCTGG-3') and a degenerate VHJ558 gene family-specific primer (VH558-FR1, 5'-ARGCCTGGGRCTTCAGTGAAG-3'). A portion of this reaction (1 µl of 25 µl total) was used in a second round of amplification (20–25 cycles under the same conditions as used in the first round) using VH558-FR3 (5'-CTGACWTCTGAGRACTCYGCRGTCYATT-3') and an end-labeled primer downstream of JH2 (JHB3; 5'-ACACACATTTCCCCCCCAACAAA-3'). Similar results were obtained using primers downstream of JH1 or JH3. For 32P end labeling of the oligonucleotide primers, T4 polynucleotide kinase (New England Biolabs, Beverly, MA) was used according to the manufacturer’s instructions. The labeled oligos were purified on a QiaQuick spin column (Qiagen, Valencia, CA) according to the manufacturer’s instructions. Three microliters of the second PCR were analyzed on a 6% acrylamide sequencing gel alongside a 1-bp ladder. The sequencing gels were subsequently analyzed using a PhosphorImager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA) or were visualized by autoradiography. PCR-FLP assays of IgH rearrangements in sorted immature B cell subsets were similar, except that the first PCR contained JHA and a mix of VH558-FR1 and VH12-FR1 (5'-TACCTGCTCTATTACTGGTTTCC-3'). The second PCR contained VH186.2-FR3 (5'-AGCAGCCTGACATCTGAGGACTC-3') or VH12-FR3a (5'-CCAGTTCTTTCTGCAATTGAACTC-3') and a primer just downstream of JH1 (JHB6-xba, 5'-GGCTCTA GAGTGTGGCAGATGGCCTGACA-3').
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We have shown previously that our preparations of FITC- or biotin-labeled rAnV clearly demarcate apoptotic B cells in vitro, using as a control a temperature-sensitive Abelson virus-transformed cell line, 103-bcl2, that can be induced to undergo apoptosis upon shifting the cells to the nonpermissive temperature (45, 54). Based on published reports describing the utility of rAnV to identify apoptotic cells (reviewed in Ref. 51), our original goal was to use this reagent to track apoptotic B cells undergoing negative selection in vivo. To do this, we obtained mice expressing transgenes encoding both the H and L chains of an Ab (clone 3-83) specific for the MHC class I molecule H-2K (52). The 3-83 Ab binds to H-2Kk with high affinity and to H-2Kb with moderate affinity, but does not bind with appreciable affinity to H-2Kd (57, 58). Thus, in H-2d mice bearing the 3-83 transgenes, nearly all the immature BM B cells and mature peripheral B cells express the Tg BCR and stain with the 3-83 Id-specific mAb 54.1 (59). Development of Ig Tg B cells is also accelerated, and there are very few pro- and pre-B cells present in H-2d 3-83 Tg BM (32) (data not shown). However, in H-2b or H-2k 3-83 Tg mice, B cell development is arrested at an immature stage, and those Tg B cells remaining are IgMlowIgD- (32). Nemazee and colleagues have shown that a subset of these arrested BM B cells can undergo receptor editing (60).
To quantify apoptosis in Tg mice on a negatively selecting background
(i.e., in H-2b and H-2k
3-83 Tg mice), we stained BM cells from these mice and from
H-2d 3-83 Tg mice with FITC-rAnV. Contrary to
expectation, instead of identifying a large population of Tg B cells
undergoing apoptosis in the
H-2bxd and
H-2kxd Tg mice, we found
that most of the "live-gated" B220+ cells in
the H-2d Tg BM bind rAnV (Fig. 1
). These Tg B cells bind rAnV at
intermediate levels (rAnVint), significantly
higher than background staining in the absence of
Ca2+, but lower than the levels found on dead or
dying cells (based on ungated cell profiles; see below). Furthermore, a
substantial proportion (
40–50%) of B cells in both nonTg BM and
H-2bxd and
H-2kxd Tg BM also bind
rAnV at intermediate levels (Fig. 1).
|
Given the presence of substantial numbers of Tg
Id+ B cells in the peripheral lymphoid organs of
H-2d 3-38 Tg mice (32) (data not
shown), it seemed unlikely that all the rAnVint
cells in the BM of these mice represent cells undergoing apoptosis.
However, to test whether other MHC molecules, for example class Ib
complexes (reviewed in Ref. 62), might trigger apoptosis
of 3-83 Tg B cells in BALB/c (H-2d) mice, we
crossed the 3-83 transgene onto the
2-microglobulin-deficient background
(63). B220+ BM cells from
H-2b
2-microglobulin-/-
3-83 Tg mice also bind rAnV at intermediate levels (45).
In addition, we used ligation-mediated PCR (64) to examine
sorted rAnV+7AAD- 3-83 Tg
BM B cells for dsDNA breaks in the Ig locus associated with receptor
editing. Although receptor editing was clearly evident in unsorted Tg
BM cells from the H-2kxd
and H-2bxd mice, we found
no evidence for ongoing rearrangement in the
H-2d Tg BM (data not shown), as predicted from
the literature (24). Furthermore, we confirmed our initial
suspicions that most rAnV-binding B cells in Ig Tg and nonTg BM are not
dying by testing rAnVint B cells in a wide
variety of assays for apoptosis (45). As described
previously (45), we found that nearly all
rAnVint B cells lack internucleosomal DNA
cleavage, and they do not exhibit a loss of mitochondrial transmembrane
potential (65) or caspase activity, even though the latter
is one of the earliest detectable events during apoptosis
(45). We have also shown that the affinity for rAnV is not
an idiosyncrasy of B cells in the 3-83 Tg mice, because B cells in
nonTg mice as well as in several other Ig Tg mouse strains, including
anti-hen egg lysosome (anti-HEL) Ig Tg mice (66),
also stain in a similar fashion with FITC-rAnV (Fig. 1
)
(45).
Recombinant AnV does not bind to positively selected T cells
To investigate whether rAnV binding might reflect a general
membrane alteration on all lymphocytes undergoing selection, we
obtained anti-H-Y TCR Tg mice, a well-characterized model of T cell
positive and negative selection (34). We compared
FITC-rAnV staining levels on viable double negative, double positive,
and mature SP thymocytes from female or male anti-H-Y TCR Tg mice
(Fig. 2). Although anti-H-Y TCR Tg T
cells are subject to negative selection in male mice, substantial
numbers of rAnV+ cells do not accumulate,
probably because apoptotic thymocytes are rapidly engulfed by resident
macrophages (67). Furthermore, we found that rAnV does not
bind to anti-H-Y TCR Tg thymocytes undergoing positive selection
(i.e., in H-2b SCID females; Fig. 2). Thus, PS
exposure on nonapoptotic cells does not occur for all developing
lymphocytes in vivo, but is instead a feature that is specific to the B
cell lineage.
|
Although it is clear that rAnV-binding B cells isolated directly
ex vivo are not apoptotic (45), we reasoned that they
still might be at the very earliest stage of apoptosis, undetectable by
our other assays and might not survive well after several days culture
in vitro. To test the survival capacity of purified
rAnVint B cells, we cultured unseparated or
immunoaffinity-purified CD19+ BM cells from
H-2d 3-83 Ig Tg mice or nonTg littermates on the
H-2d BM stromal cell line S17 (55)
or the H-2Kb-expressing BM stromal cell line op42
(56). We then assessed the viability of the
B220+ cells 2 days later using either a
combination of rAnV and 7AAD staining (Fig. 3A), or the TUNEL assay (Fig. 3B). In contrast to cell populations cultured at 37°C in
the absence of stroma, both nonTg and 3-83 Tg B cells cocultured with
S17 remained largely viable by both FACS assays (Fig. 3 and data not
shown). However, a substantial fraction of 3-83 Tg cells that were
cultured on the stromal line op42 did undergo apoptosis, as expected
(Fig. 3) (60). The percentages of input cells recovered
from each culture condition were comparable (no stroma, 52% for Tg and
30% for nonTg; S17, 37% for Tg and 29% for nonTg; op42, 38% for
Tg), arguing against the possibility that the stromal cells might
actively phagocytose a majority of the rAnVint B
cells. Thus, multiple experimental criteria confirm that the majority
of rAnV+ BM B cells we detect in both nonTg and
Ig Tg mice are not apoptotic (Fig. 3) (45).
|
2-fold upon
coculture with either S17 or op42 stromal cell lines (Fig. 3A, compare the mean fluorescence intensities of the live
cells in the lower left quadrants). The relative
decrease in rAnV binding on B cells in the stromal cell cocultures
probably does not reflect the loss of a signal provided in vivo,
because BM B cells cultured in the absence of stromal cells maintain
high levels of exposed PS (Fig. 3, upper panels). Given the
short time span of the assay and the recovered cell counts (see above),
it is also unlikely that this change in rAnV staining levels reflects a
rapid outgrowth of the small number of rAnV-
input cells and a coincident disappearance of the
rAnVint cells. Instead, the stroma present in the
cocultures seems to have induced the reversal of membrane alteration on
the rAnVint B cells that had been prematurely
removed from their normal in vivo microenvironment (see
Discussion). The proportion of cells binding rAnV consistently differs between B cell subsets
Having established that rAnV can recognize B cells that
alter their membrane in a process distinct from apoptosis, we next
investigated whether there might be a pattern of rAnV binding on B
cells at various stages of development. We stained BM and spleen from
nonTg mice for three-color flow cytometric analysis with FITC-rAnV and
pairs of several mAbs useful for distinguishing well-defined immature
and mature B cell subsets (Fig. 4, A and B) (68, 69, 70, 71). When we analyzed
these cells, we did, in fact, observe a trend whereby the proportion of
each BM B cell subpopulation that binds rAnV increases as the cells
mature, but consistently decreases once the cells exit the BM and
continue their maturation in the periphery (Fig. 4C). A
relatively small proportion (30 ± 8.5%) of the pro-B cell
(B220+CD43+) compartment in
the BM binds rAnV, while the proportion of cells that bind rAnV
steadily expands among increasingly mature sIg+
BM B cells (reaching a high of 68 ± 9%; Fig. 4C).
This trend was also apparent if we focused on the levels of B220 and
heat-stable Ag surface expression to define the stages of BM B cell
maturation (data not shown). Interestingly, the levels of rAnV binding
dropped fairly sharply on the mature recirculating
IgMlowDhigh BM B cell
subset (42 ± 10%; Fig. 4B). When we examined splenic
B cells for rAnV staining among the peripheral B cell subsets,
fractions III
(IgMhighIgDlow), II
(IgMhighIgDhigh), and I
(IgMlowIgDhigh), we found
that the proportion of rAnV-binding cells steadily decreased as the
cells transited to the next maturational stage (Fig. 4C).
Thus, a larger fraction of those cells thought to represent recent
emigrants from the BM
(IgMhighDlow or
B220lowheat-stable Aghigh)
bind rAnV than do the more mature peripheral B cell subsets, a portion
of which probably return to the BM bearing reduced levels of PS (Fig. 4B;
IgMlowIgDhigh).
|
B)
(70). We also stained CD43- spleen
cells (negatively selected on a magnetic bead column; >95%
B220+) with rAnV and Abs to CD21 and CD23 to
distinguish NF, follicular (FO), and MZ B cells (Fig. 4B)
(28, 29). As shown in Fig. 4C, the T2
population contains a higher fraction of rAnV-binding cells than do the
T1 or mature B subsets. More strikingly, however, an average of 87% of
the MZ B cells are rAnV+, significantly higher
than their NF and FO counterparts (Fig. 4C). Thus, PS is
apparently gradually reinternalized on B cells as they exit the BM or
shortly thereafter, but is again externalized on mature B cells
recruited to the MZ by a BCR-mediated interaction. Recombinant AnV binds to nearly all B-1 cells
Given the developmentally regulated patterns of rAnV binding
throughout B cell development in normal mice (Fig. 5) and our observation that a large
fraction of Ig Tg B cells bearing a selectable BCR bind rAnV (Fig. 1)
(45), we postulated that this reagent might be identifying
positively selected B cells. To investigate this possibility, we first
analyzed BM, spleen, and peritoneal washout cells (PerC) from nonTg
mice and from VH12 Tg mice in which the
rearranged IgH transgene efficiently mediates positive selection of
cells into the CD5+ B-1a cell subset
(43). These cells were stained for three-color flow
cytometric analysis with FITC-rAnV and mAbs specific for CD5 and B220.
As shown in Fig. 5, a large majority (80–95%) of the B-1a cells
(B220+CD5+) in the
peritoneum (upper panels), spleen (lower
panels), and BM (data not shown) of both nonTg and
VH12 Tg mice bound significant levels of rAnV,
analogous to what we had observed with 3-83 and anti-HEL Ig Tg BM
(Fig. 1) (45). Similar to the
rAnVint BM B cells from those mice,
rAnVint CD19+ cells
purified from VH12 Tg spleen are not apoptotic by
multiple criteria (data not shown) (45). Interestingly, in
both nonTg and VH12 Tg mice, the proportion of
cells binding rAnV in the B-1a lineage is greater than that in the B-2
population (B220+CD5-)
from the same individual (Fig. 5). Even more striking is the comparison
of rAnV levels on either B cell subset to those on T cells (Fig. 5;
B220-CD5+). Unlike the
staining pattern on B cells, rAnV binding to viable resting T cells is
rarely above the background staining seen in the absence of
Ca2+, even in the BM microenvironment (Fig. 5 and
data not shown).
|
Because many of the B cell subsets exhibiting extensive AnV
binding either contain dividing cells or are the successors of cells
that have divided recently (Figs. 4 and 5), we began to address the
issue of whether AnV binding is a consequence of cell division. We
stained bone marrow cells from C57BL/6 mice with anti-B220 and
rAnV-FITC, fixed and permeabilized the cells, then stained their DNA
with 7AAD before flow cytometric analysis (Fig. 6). Analysis of gated
B220+AnV- vs
B220+AnVint B cells did not
reveal any notable differences in the distribution of cells in the
G0/G1 or
G2/M/S phases of the cell cycle. This suggests
that if PS is exposed during cell division, it remains externalized
once the cells have exited the cell cycle.
|
Although resting T cells stained ex vivo are predominantly
AnV- (Figs. 2 and 7A and data not
shown), we questioned whether T cells activated in vitro might
externalize PS and bind rAnV. We also asked whether we might be able to
alter the rAnV binding properties of B cells if we deliberately forced
them into cycle. To address these questions, we isolated splenocytes
from C57BL/6 mice; cultured them for 2 days with LPS, LPS plus IL-4,
PMA plus ionomycin, or Con A; and compared the AnV staining patterns of
each cell type with that of unstimulated control cell cultures (Fig. 7A). We compared the fraction
of cells falling in the AnV-,
AnVint, and AnVhigh regions
among live-gated B220+ B or
CD3+ T cells (Fig. 7 and data not shown) for each
culture condition. We found that the extent of AnV binding to B cells
remained essentially unchanged under all stimulation conditions, except
for the LPS plus IL-4 treatment (86% AnVint/high
compared with 59% for the unstimulated cells). Interestingly, both Con
A and PMA/ionomycin stimulation induced a substantial increase in
AnV-binding cells within the CD3+ T cell
population (82 and 81% AnVint/high,
respectively, compared with 20% for the unstimulated cells; Fig. 7B). Whether these AnV-binding T cells expose PS on the
pathway leading to activation-induced cell death or are instead
mimicking the events occurring in B cells in vivo remains to be
determined (see Discussion).
|
To more rigorously test the idea that rAnV staining marks positive selection of B cells, we undertook a molecular analysis of the IgH repertoire of rAnV- vs rAnV+ B cells from nonTg mice. We looked for differences in cell populations using PCR-fragment length polymorphism (PCR-FLP) analysis of DNA from FACS-sorted BM cells (73). This assay consists of a PCR step using primers that amplify across the VDJ junction, followed by separation of the amplified products on a denaturing polyacrylamide gel, allowing an assessment of both CDR3 length heterogeneity and the distribution of in-frame and out-of-frame rearrangements within a population of cells.
We performed PCR-FLP analyses with DNA from
rAnV- or rAnVint pro-B
(B220+CD43+) cells sorted
from BALB/c BM using a degenerate PCR primer that recognizes the large
J558 family of VH genes and a primer that anneals
downstream of the JH1 or
JH2 gene segments. As noted above, pro-B cell
survival and maturation require the production of an in-frame Ig H
chain gene rearrangement. Interestingly, we reproducibly found that
while the rAnV- pro-B cell population bore many
out-of-frame IgH rearrangements, the rAnV+ pro-B
cells were enriched for in-frame rearrangements (Fig. 8A and data not shown).
|
B). Not
surprisingly, given that the VH186.2 family is
large and is characterized by no known CDR3 length preferences in
selected B cells, the PCR-FLP analyses of VH186.2
to DJH1 rearrangements did not reveal any obvious
differences between the rAnV- and
rAnV+ immature B cells (Fig. 8B,
upper panel). DNA sequence analysis of the IgH repertoire in
rAnV- vs rAnV+ populations
at various stages of B-2 cell development will ultimately be required
to identify potential differences among cells in this very diverse
compartment. However, when we analyzed rearrangements of the much
smaller VH12 family, we obtained a striking
result. As noted above, the predominant
VH12-containing H chain, found paired with a
V4+ L chain in the peritoneal anti-PC B-1
cell subset, consists of a
VH12-DH-JH1
rearrangement containing a 10-aa CDR3 (42). In three
independent experiments, we found that rAnV+
immature B cell populations consistently contain a dominant
VH12 to DJH1 rearrangement
that comigrates with the product from the VH12
transgene and with the corresponding positively selected
VH12 H chains in control nonTg PerC, day 6
spleen, and adult spleen (Fig. 8B, lower panel).
Conversely, the VH12 rearrangements present
within the rAnV- population are more diverse and
more closely resemble the pattern found in unsorted BM samples (Fig. 8B). Thus, rAnV marks a subset of immature BM B cells
selected based on the structure of its surface Ig. |
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We have previously reported our surprising finding that the phospholipid-binding protein AnV, generally a reliable marker of PS externalization during apoptosis, binds at appreciable levels to viable B cells (45). We showed that the bulk of B cells that bind rAnV at these so-called intermediate levels (higher than background staining in the absence of Ca2+, but lower than that on dead or dying cells) are not apoptotic by multiple criteria (45). Instead, we found that PS is mobilized to co-cap with IgM on anti-IgM-treated B cells and to colocalize with a marker of lipid rafts, dynamic membrane microdomains thought to function as platforms for the attachment of proteins whose localization and clustering may be critical for signal transduction (74, 75, 76, 77). Furthermore, because we could disrupt normal signaling through IgM by sequestering PS on rAnV+ B cells, we suggested that PS might play a role in BCR-mediated signaling events (45).
Here, we have described another unexpected facet of our studies using
rAnV by showing that the externalization of PS on viable B cells is
regulated during development, and that this membrane remodeling event
largely depends upon the expression by each B cell of a selectable BCR.
Recombinant AnV binding clearly increases on B cells as they traverse
the pro- to pre-B and immature to mature B transitions as assessed by
both cell surface (Figs. 4 and 5) and molecular genetic analyses (Fig. 8). Direct IgH repertoire analyses of sorted
rAnV+ and rAnV- B cell
subsets in normal mice strongly suggest that changes in the plasma
membrane leading to rAnV binding correlate with the generation of an
in-frame H chain gene rearrangement in pro-B cells (Fig. 8A)
and with a positive selection signal in immature B cells (Fig. 8B).
Our data reveal a general trend whereby the proportion of each BM B
cell subset that can bind rAnV increases as progenitor cells mature,
but steadily decreases once the cells exit the BM and continue their
maturation in the periphery (Fig. 4). Because a larger fraction of
those cells thought to represent cells about to exit from the BM (i.e.,
immature IgMhighIgD-) bind
rAnV than do the early splenic (NF/T1) B cell subsets (Fig. 4), this
membrane remodeling event might play a role in cell trafficking to the
peripheral lymphoid organs (Fig. 9).
However, it is not yet clear what induces the steady loss of rAnV
binding sites on increasingly mature (FO/fraction I) splenic B cells,
or what induces the re-externalization of PS as the cells are recruited
to the MZ (Fig. 4). These changes may be the result of changes in the
expression or activity of enzymes known to alter lipid asymmetry, such
as the scramblase and aminophospholipid translocase (see Ref.
49 for review).
|
In general, the immature and mature B cell subsets that exhibit
the highest proportion of positive staining by rAnV either contain
dividing cells or are the successors of cells that have divided
recently (i.e., immature
IgMhighIgD- BM B, B-1, and
MZ B cells; Figs. 1, 4, and 5). Because early selection processes in
the BM are accompanied by proliferation (78), it is
possible that this membrane alteration occurs when B cells divide.
Thus, we were initially led to consider whether PS is externalized as a
consequence of cell division. When we directly analyzed the cell cycle
status of normal or H-2d 3-83 Ig Tg BM B cells,
we found that the vast majority of cells binding rAnV at intermediate
levels are in G0/G1 (Fig. 6
and data not shown). Furthermore, the proportion of B cells binding
rAnV does not increase significantly if B cells are forced into cell
cycle in vitro (Fig. 7). It is also important to note that although
nearly 95% of the B-1 and MZ B populations bind rAnV (Figs. 4 and 5),
a much smaller fraction of these cells is actually cycling at any given
time (72, 79). Thus, at least for the B cell lineage, our
data suggest that if the externalization of PS requires cell division,
these PS residues remain exposed once the cell has exited the cell
cycle.
Interestingly, T cells (which are rAnV negative in vivo in normal and
TCR Tg mice; Fig. 2 and data not shown) stimulated in vitro with Con A
or PMA/ionomycin exhibit an increase in rAnV binding (Fig. 7). However,
whether this form of PS exposure represents a situation comparable to
that occurring on B cells in vivo or instead reflects the early stages
of T cell apoptosis (via activation-induced cell death) remains an open
question. Preliminary confocal microscopy experiments support the
former possibility, because PS colocalizes with Thy-1 molecules on Con
A blasts, or with TCR-CD3 complexes on splenocytes stimulated briefly
with soluble anti-CD3 mAb (data not shown). These data suggest
that, analogous to its role in B cells (45), PS may also
play a role as a component of lipid rafts in activated T cells.
Why do B cells alter their membrane lipid distribution during development?
B cells develop in intimate contact with a highly structured stromal cell environment that provides both specific adhesion contacts with the B cells and factors required for their growth and differentiation (80). Stromal cells are also important mediators of tolerance induction of sIg+ B cells. As B-lineage cells mature, they migrate in close proximity to stromal cells within the marrow. The earliest stem cells lie in the subendosteum, just adjacent to the bone surface, and as maturation proceeds the B cells move toward the central axis of the marrow cavity where they wait in sinuses for export to the periphery. Later stages of maturation become less dependent on stromal cells, and final transition from immature to mature B cell can actually occur either in the BM or in secondary lymphoid organs such as the spleen.
We propose that the change we observed in membrane structure on BM B
cells that accompanies differentiation (i.e., the expression of a
selectable receptor) may in part explain how a cell progresses to the
next maturational microenvironment (Fig. 9). PS+
pro-/pre-B cells with a functional pre-BCR might detach from the outer
zone of the BM and begin to migrate toward the central axis. Immature B
cells may be less dependent upon stromal cells because of their altered
membrane structure, and this change in membrane asymmetry may also play
a role in B cell export from the BM. Our data suggest that the overall
density of exposed PS may be important for export of immature B cells
from the BM, because FITC-rAnV staining increases as B cells mature in
nonTg BM (Fig. 4) and is brightest on Tg B cells expressing a
functional BCR in a permissive (nonnegatively selecting) environment
(Fig. 1) (45). Additional changes in adhesion molecule or
chemokine receptor expression, influenced in part by stromal cells and
their products, are likely to occur concomitantly with PS exposure
(1, 80, 81). It remains to be determined whether plasma
membrane remodeling is a crucial occurrence, or simply a byproduct of
these other maturational events.
Given that the levels of rAnV binding on viable B cells from 3-83 Tg or
nonTg BM consistently decreased upon coculture with either the S17 or
op42 stromal cell lines (Fig. 3A), it seems likely that
stroma present in these cultures induced the reversal of PS
distribution on the rAnVint immature B cells that
had been prematurely removed from their normal in vivo
microenvironment. It thus remains of further interest to determine
whether PS exposure accompanies an alteration in the adhesion
properties of B cells. Because thymocytes do not alter their membrane
in this manner, it may be that the externalization of PS during B cell
development is dependent upon the presence of unique BM stromal cell
cues, signals not provided by thymic stroma.
Our finding that nearly all B-1 and MZ B cells bind rAnV (Figs. 4 and 5) is compelling in light of the recent advances in understanding the
functional relatedness of these two subsets (reviewed in Ref.
72). It has been postulated that these two B cell subsets
have evolved to provide first-line responses for gut/peritoneum and
blood-borne Ags, and that restriction of the Ig repertoire of these
compartments may ensure rapid development of short term responses to a
limited number of conserved Ags. It has also been reported recently
that Pyk-2-deficient mice lack MZ B cells (29). The
authors suggest that the absence of Pyk-2 affects MZ B cell
localization to their specialized anatomical niches, possibly due to a
compromised response to one or several chemokines or to the dysfunction
in the adhesion to some matrix component or stromal cells. It will be
interesting to test whether mature B cells in these mice bind rAnV in a
manner similar to those in normal mice to begin to address whether
externalized PS promotes improved cell adhesion and/or motility of B
cells in addition to its apparent role in BCR signaling
(45).
The link between positive selection and rAnV binding in B-2 vs B-1 cell lineages
In the 3-83 Ig Tg model system, where the Tg BCR mediates B cell
selection into the conventional B-2 population in mice lacking
H-2Kb or k molecules (i.e.,
H-2d or MHC class I-/-),
we found that nearly all the Tg 3-83 Id+ B cells
bind rAnV (Fig. 1) (45). Those B cells arrested at the
immature IgMlowIgD- stage
in H-2b and H-2k Tg mice
actually bind lower levels of rAnV than their
H-2d counterparts (Fig. 1). Interestingly, the
few splenic B cells that have escaped negative selection in
H-2b or k Tg mice by replacing the Tg IgL chain
with an L chain encoded by an endogenous Ig rearrangement bind
increased levels of rAnV relative to the arrested Tg BM B cells (data
not shown) (45). This suggests that the edited cells
receive a positive signal through their modified BCR, allowing both a
membrane remodeling event and export to the periphery. Similarly, B-1a
cells in both nonTg and VH12 Tg mice also bind
rAnV at appreciable levels (Fig. 5). These data support our hypothesis
that rAnV binding and positive selection are linked, because B-1 cells
have been described as a subset that is more intensely positively
selected than conventional naive B-2 cells
(1) (see below). A third mouse model of B cell selection, the anti->HEL
Ig Tg strain (66), provided similar results in our rAnV
flow cytometry assays (45). Notably, the anergic B cells
in anti-HEL/soluble HEL double Tg mice also bind high levels of
rAnV (45), again consistent with the idea that this
membrane alteration occurs on B cells that have received a BCR signal
(in this case, a tolerogenic signal).
It is interesting to note that even mature peritoneal and splenic B-1
as well as mature MZ B cells retain the ability to bind rAnV (Fig. 5).
This distinguishes them from follicular B-2 cells, which tend to revert
to a relatively normal plasma membrane phospholipid arrangement as they
mature (Figs. 4 and 5) and may reflect in part the different anatomical
distributions of B-1 and MZ B cells. It is possible that the PS
receptors expressed on macrophages and other cells throughout the body
influence the trafficking of PS+ B cells. B-1
cells may receive frequent positive selection signals through recurrent
interactions with their cognate self-Ags or via their constitutively
active STAT3 protein (82, 83). Both B-1 and MZ B cells are
also known to differ dramatically from follicular B-2 cells in their
capacity to more rapidly respond to BCR signals (72, 79).
It has been argued that, in contrast to B-2 and B-1b
(CD5-) cells, the progenitors of B-1a
(CD5+) cells are abundant in the fetal omentum
and liver, but are absent from adult BM (84, 85, 86). However,
we have shown that immature
(IgM+IgD-) B cells that
bind rAnV in adult BALB/c BM include apparent B-1 cell precursors
carrying a rearranged canonical anti-PC
VH12+ Ig H chain (Fig. 8B). Although PS externalization is a prominent feature of
mature B-1 cells (Fig. 5), we do not believe that the canonical
VH12 rearrangements in
rAnV+ BM belong to mature recirculating B-1
cells, because our sort gates on immature B cells excluded those with
even low levels of IgD expression. Because we do not know whether the
selected VH12 H chains from sorted
rAnV+ immature
(IgM+IgD-) BM B cells have
paired with V4+ L
chains, it remains possible that these cells were positively selected
at the pre-B stage, but do not bind PC. In fact, Clarke and colleagues
have postulated a positive selection step at the pre-BI to pre-BII
transition of BM B cells that express this selected
VH12 H chain (21).
The nature of positive selection within the B cell lineage is not well understood. Current models postulate that B cell specificities are selected through interactions with endogenous or environmental Ag, because the IgH repertoire constricts at some point(s) between the pre-B cell stage and the mature IgD+ splenic B cell stage (reviewed in Refs. 1 and 2). Recent reports have demonstrated that continued BCR expression is required for the survival of mature B cells (26), and that signaling involving Syk is required for B cell maturation beyond the immature peripheral B cell stage (87). Thus, a BCR that can provide an adequate survival signal probably mediates the positive selection event that converts an immature to a mature B cell. After an immature B cell is selected to exit the BM, its BCR may continue to signal its survival and further maturation through repeated contact with Ag.
In contrast to the clear link between PS exposure and positive
selection of B-1 cells (Figs. 5 and 8B), a correlation
between PS externalization and positive selection of follicular B-2
cells has proven more difficult to substantiate at the molecular level
(Fig. 8B), largely because of the increased complexity of
the Ig repertoire of the B-2 population. It is not yet clear whether
positive selection based on BCR specificity acts on immature B-2 cells
in the BM, or whether the Ig repertoire narrows at a later stage of B-2
cell development, perhaps shortly after their arrival in the periphery
(36). However, the fact remains that rAnV binds to BM B-2
cells from two independent strains of Tg mice expressing rearranged
IgH+L genes encoding BCRs already proven capable of mediating positive
selection, because these transgenes were isolated from hybridomas of
mature B-2 clones (Fig. 1) (45).
Clearly, a substantially higher fraction of immature BM B cells binds
rAnV than actually progresses to join the mature peripheral B cell
pool. Estimates of survival of B cells attempting this developmental
transition range from 10 to 20% (38, 39, 40). Because in some
cases 70–90% of the immature BM B cell subset binds rAnV (Figs. 1 and 4) (45), it is apparent that although events associated
with positive selection lead to PS externalization, all rAnV-binding B
cells are not guaranteed successful emigration to the periphery.
We have previously shown that PS externalized on viable B cells serves as a component of lipid rafts and colocalizes with the BCR after IgM ligation (45). Thus, the increasing levels of PS exposure during B cell maturation may not be merely coincidental, but, rather, may be an important step in the critical process of rewiring the B cell to alter the signaling threshold of its Ag receptor or some quality of this signal. The potential link between PS exposure and B cell activation, proliferation, and adherence suggests that rAnV may represent a useful tool to identify B cell clones selected by Ag during a normal immune response or even to detect self-reactive clones mediating autoimmunity.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Current address: ZymoGenetics, 1201 Eastlake Avenue East, Seattle, WA 98102.
3 Current address: Department of Medicine, Pennsylvania Hospital, 800 Spruce Street, Philadelphia, PA 19147.
4 Address correspondence and reprint requests to Dr. Mark Schlissel at his current address, Department of Molecular and Cell Biology, 439 LSA, University of California, Berkeley, CA 94720-3200.
