Rapid B Cell Receptor-induced Unfolded Protein Response in Nonsecretory B Cells Correlates with Pro- Versus Antiapoptotic Cell Fate*

The adaptive unfolded protein response (UPR) is essential for the development of antibody-secreting plasma cells. B cells induced by lipopolysaccharide (LPS) to differentiate into plasma cells exhibit a nonclassical UPR reported to anticipate endoplasmic reticulum stress prior to immunoglobulin production. Here we demonstrate that activation of a physiologic UPR is not limited to cells undergoing secretory cell differentiation. We identify B cell receptor (BCR) signaling as an unexpected physiologic UPR trigger and demonstrate that in mature B cells, BCR stimulation induces a short lived UPR similar to the LPS-triggered nonclassical UPR. However, unlike LPS, BCR stimulation does not induce plasma cell differentiation. Furthermore, the BCR-induced UPR is not limited to cells in which BCR induces activation, since a UPR is also induced in transitional immature B cells that respond to BCR stimulation with a rapid apoptotic fate. This response involves sustained up-regulation of Chop mRNA indicative of a terminal UPR. Whereas sustained Chop expression correlates with the ultimate fate of the BCR-triggered B cell and not its developmental stage, Chop–/– B cells undergo apoptosis, indicating that CHOP is not required for this process. These studies establish a system whereby a terminal or adaptive UPR can be alternatively triggered by physiologic stimuli.


The adaptive unfolded protein response (UPR) is essential for the development of antibody-secreting plasma cells. B cells induced by lipopolysaccharide (LPS) to differentiate into plasma cells exhibit a nonclassical UPR reported to anticipate endoplasmic reticulum stress prior to immunoglobulin production. Here we demonstrate that activation of a physiologic UPR is not limited to cells undergoing secretory cell differentiation. We identify B cell receptor (BCR) signaling as an unexpected physiologic UPR trigger and demonstrate that in mature B cells, BCR stimulation induces a short lived UPR similar to the LPS-triggered nonclassical UPR. However, unlike LPS, BCR stimulation does not induce plasma cell differentiation. Furthermore, the BCR-induced UPR is not limited to cells in which BCR induces activation, since a UPR is also induced in transitional immature B cells that respond to BCR stimulation with a rapid apoptotic fate. This response involves sustained up-regulation of Chop mRNA indicative of a terminal UPR. Whereas sustained Chop expression correlates with the ultimate fate of the BCR-triggered B cell and not its developmental stage, Chop؊/؊ B cells undergo apoptosis, indicating that CHOP is not required for this process. These studies establish a system whereby a terminal or adaptive UPR can be alternatively triggered by physiologic stimuli.
The unfolded protein response (UPR) 4 is an evolutionarily conserved endoplasmic reticulum (ER)-to-nucleus signaling pathway typically initiated in situations in which ER protein folding demand overwhelms folding capacity (1)(2)(3)(4). The goal of the UPR is to reduce ER stress by restoring a balance to client protein load and folding capacity. A multifaceted response controlled by the ER transmembrane molecules ATF6 (activating transcription factor 6), IRE1 (inositol-requiring enzyme 1), and protein kinase-like ER kinase accomplishes this goal primarily through transcriptional activation of a large number of UPR target genes encoding proteins involved in nearly all aspects of protein folding, modification, trafficking, and degradation.
The UPR has primarily been examined in situations of pharmacologically induced or pathophysiologic stress. However, in mammals, this pathway is also activated during the normal physiologic development and differentiation of professional secretory cells, such as ␤ islet cells of the pancreas and plasma cells (5)(6)(7)(8)(9)(10)(11)(12)(13). Plasma cells are terminally differentiated B lymphocytes (B cells) whose function is to secrete Ig, a critical element of the humoral immune response. During plasma cell differentiation, the enormous increase in Ig production was initially believed to flood the ER with protein and induce a UPR via this classical mechanism (7). However, recent studies indicate that unlike the situation in pharmacologically induced UPRs, initial activation of UPR pathways during B cell differentiation is selective and may occur through a mechanism independent of ER stress, since it precedes increased Ig production (5,11,12). Consequently, a model has been proposed in which ER expansion during the differentiation of professional secretory cells is a twophase process. According to this model, the initial phase of the UPR during differentiation is protein load-independent and preparative, whereas the second phase is driven by the classical mechanism of protein accumulation within the ER.
Although the relationship between the UPR and LPS-induced plasma cell differentiation has been the subject of intensive study, the association of the UPR with antigen-specific adaptive immune responses has not been explored. Antigen-specific immune responses require multiple activation and differentiation signals, beginning with the specific interaction of a naive B cell with its cognate antigen through the clonotypic B cell receptor for antigen (BCR) (14 -16). Since the LPS-driven plasma cell differentiation process is associated with activation of adaptive UPR pathways prior to significant Ig accumulation and thus has been speculated to protect cells via anticipatory UPR induction, we became interested in whether a similar process occurs in BCR-dependent activation. Moreover, although the BCR signal is a critical first step for the generation of antigen-specific plasma cells, BCR signals alone do not lead to plasma cell differentiation, except in isolated, highly specialized cell populations (17). Thus, our studies allow for evaluation of the association of the UPR with cellular responses independent of secretory cell differentiation. Subsequent to BCR stimulation, activated B cells may receive cytokine differentiation signals, which have been associated with UPR induction (5,11). We hypothesized that anticipation of upcoming ER stress might begin from the earliest BCR-mediated stimulatory signal, despite the dissociation between BCR signals and plasma cell differentiation.
Here we identify BCR signaling as a novel physiologic UPR trigger. We demonstrate that primary mature B cells stimulated through the BCR initiate an adaptive UPR similar to the UPR induced by LPS during the initial 24 h of the differentiation process, as ascertained by examination of the transcriptional responses associated with activation of the UPR. This response is characterized by up-regulation of prototypical UPR targets encoding ER chaperones, transient Chop (C(EBP) homol-ogous protein) up-regulation, and minimal XBP-1 (X-box-binding protein 1) activation.
Although the outcome of UPR induction is often adaptation to ER protein load, in cases of overwhelming ER stress, UPR induction leads to apoptosis. This process, the terminal UPR, is not fully understood but is associated with activation of dedicated apoptotic molecules, including CHOP (18 -25). Engagement of the BCR on transitional immature B cells, precursors of mature B cells, leads to a rapid apoptotic fate (26 -30). Our studies demonstrate activation of UPR signaling pathways in BCR-stimulated primary transitional immature B cells. In these cells, prolonged Chop up-regulation and relatively poor activation of chaperone genes characterize the BCR-induced UPR. This pattern is consistent with activation of a terminal, or proapoptotic, UPR. Co-stimulation of transitional immature B cells with physiologic survival signals rescued these cells from BCR-induced apoptosis and transformed the observed UPR, reducing Chop expression and increasing expression of chaperone-encoding genes.

EXPERIMENTAL PROCEDURES
B Lymphocyte Isolation-Wild-type BALB/c mice were obtained from the Jackson Laboratory (Bar Harbor, ME) or NCI, National Institutes of Health (Bethesda, MD). Splenic B cells were isolated from 8 -12-week-old female mice as previously described (31). These populations are typically 95-98% B220 pos , IgM pos B cells consisting of 85-90% mature CD23 pos AA4.1 neg B cells. The remaining small subset represent maturing IgM pos transitional immature B cells (typically 10 -15%). Transitional immature cells were isolated from the spleens of autoreconstituting mice that received 500 rads of whole body irradiation 14 days prior to animal sacrifice, as previously described (30,32,33). Transitional immature B cells isolated in this manner were IgM high , IgD int , and AA4.1 pos . For some experiments directly comparing the responses of transitional immature B cells with mature B cells, mature B cells were isolated from irradiated mice allowed to autoreconstitute for 27-36 days. For electron microscopy of transitional versus mature cells, T1 (B220 high , CD23 neg , AA4.1 pos ) and mature (B220 high , CD23 pos , AA4.1 neg ) cells were sorted by flow cytometry and were determined to be greater than 98% pure.
B Lymphocyte Functional Assays-Purified B cells were cultured in RPMI containing 10% fetal calf serum (Hyclone Laboratories, Logan, UT), ␤-mercaptoethanol (Sigma), nonessential amino acids, OPI (oxaloacetic acid/pyruvate/insulin) and L-glutamine at 37°C and 5% CO 2 . All reagents were obtained from Invitrogen except as indicated. Cells were stimulated with 25 g/ml goat F(abЈ) 2 fragments (HC-and LC-directed) purchased from Jackson ImmunoResearch (West Grove, PA) to cross-link the BCR or with Salmonella typhosa LPS at 10 g/ml (Sigma). T cell help signals were provided, using 10% X-4 supernatant containing IL-4, as previously described (34), or 10 g/ml anti-CD40 (HM40-3; Pharmingen, San Diego, CA). Differentiation and activation markers were examined by flow cytometry of cell surface markers using Pharmingen (San Diego, CA) antibodies. Antibody secretion was assessed by an enzyme-linked immunospot assay (KPL, Gaithersburg, MD) as detailed by the manufacturer. Total Ig was examined using rabbit anti-mouse F(abЈ) 2 fragments conjugated to horseradish peroxidase (Jackson ImmunoResearch, West Grove, PA). Plates were read and photographed using Image-Pro Plus (Hitech Instruments, Edgemont, PA), and counts were confirmed manually. Apoptosis was assessed by propidium iodide staining for DNA content as directed by the manufacturer (Calbiochem). Subdiploid percentages were determined by flow cytometry in triplicate.
Electron Microscopy-Electron microscopy was performed at the University of Pennsylvania Biomedical Imaging Core Facility. Cells were fixed in 2.5% glutaraldehyde plus 2% paraformaldehyde in 0.1 M sodium cacodylate overnight at 4°C. The next day, pellets were rinsed in 0.1 M sodium cacodylate buffer and then postfixed with 2% osmium tetroxide, dehydrated in graded ethanol, and embedded in Epon. 70-nm thin sections were examined with a JEOL JEM 1010 electron microscope after uranyl acetate and bismuth subnitrite staining. Using a Hamamatsu CCD camera and AMT 12-HR software, the images were captured at various magnifications. All supplies were purchased from Electron Microscopy Sciences (Fort Washington, PA).
Semiquantitative RT-PCR Analysis-Total RNA was extracted and reverse transcribed as described above. Spliced and unspliced Xbp-1 were amplified as previously described (11).
Western Blotting-B cells were lysed in radioimmune precipitation assay buffer and run on 11% SDS-PAGE gels. Protein concentration was determined by bicinchronic acid assay (Sigma), and equal amounts of protein were loaded (50 g). Anti-mouse XBP-1 (M-186) was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Goat antimouse secondary antibody conjugated to horseradish peroxidase was obtained from Pierce. Protein detection was performed using enhanced chemiluminescence (Amersham Biosciences).
Chop-deficient Mice-Chop Ϫ/Ϫ mice bearing the CHOP.KO2 allele have been previously described (35). For these studies, the CHOP.KO3 allele was used. The identical region of the Chop gene is deleted in the two alleles, namely the entire coding region (except the C-terminal last DECEMBER 2, 2005 • VOLUME 280 • NUMBER 48 34 amino acids) located between the genomic PmlI and NheI sites. It is replaced by a LacZ reporter gene in the CHOP.KO3 allele. The PGK.Neo cassette used to select the targeted allele was removed by loxP recombination. The mutant allele directs no expression of CHOP protein. All experiments were conducted with sex-matched littermate controls in a sixth generation backcross of the mutant allele into a C57BL/6 genetic background.

BCR Stimulation Leads to Morphologic Changes in the Absence of
Plasma Cell Differentiation-Unlike LPS stimulation, which drives polyclonal proliferation and differentiation of B cells into antibody-secreting plasma cells, engagement of the BCR by antigen initiates an adaptive immune response, directed against the specific antigen (14,16). Although BCR signals alone do not lead to plasma cell differentiation (36 -39), the BCR-antigen interaction is a critical initiator of the humoral immune response. It is only after BCR engagement that B cells interact with, activate, and receive differentiation signals from T cells. We were interested in whether BCR signals could play a role in preparing cells for eventual plasma cell differentiation, possibly by initiating preparative ER remodeling via UPR activation. To investigate this hypothesis, we directly examined changes in the amount and organization of ER-Golgi networks induced by BCR signals in primary splenic B cells using electron microscopy (EM). We found that although stimulation through the BCR with anti-BCR does not lead to plasma cell differentiation, changes in cell morphology do accompany the BCR activation signal.
For these experiments, primary splenic B cells were treated with either anti-BCR to cross-link the BCR, thereby acting as a surrogate antigen, or with LPS. Stimulated populations were then examined for evidence of plasma cell differentiation. Plasma cells can be identified both by expression of specific cell surface markers, including Syndecan-1, and by particular morphologic characteristics (10, 40, 41). Differ-entiation of resting naive B cells into plasma cells in response to anti-BCR and LPS was examined using flow cytometric analysis and EM. LPS stimulation led to the emergence of B220 pos Syndecan-1 pos plasma cells after 72 h of culture; however, B cells cultured with anti-BCR for 96 h remained Syndecan-1 neg (data not shown).
Similarly, a direct examination of the morphology of BCR-and LPSstimulated splenic B cells revealed that, as expected, the morphologic changes indicative of plasma cell differentiation, a highly polarized appearance, and well ordered ER-Golgi network, were observed in LPSstimulated B cells but were not apparent in BCR-stimulated B cells (Fig.  1A). However, subtle morphologic changes do occur in response to BCR signals. Resting B cells have a small, poorly developed cytoplasm ( Fig.  1B, left). BCR-stimulated B cells increased in both overall size and cytoplasm/nucleus ratio by 72 h poststimulation (Fig. 1B, middle). In addition, the amount of cytoplasmic membranes visibly increased, suggesting an increase in the size of the ER-Golgi network (Fig. 1B, middle, arrowheads). As compared with LPS-stimulated cells (Fig. 1B, right), the increase in cytoplasmic membranes was modest, and membranes appeared poorly organized. For these experiments, a representative cell is shown, and 82-99% of live cells demonstrated the depicted morphology.
Plasma cells participate in the immune response by secreting large amounts of Ig. To verify that BCR signals did not lead to secretion of antibody in the absence of typical features of plasma cell differentiation, antibody secretion was tested directly. Enzyme-linked immunospot assays for total Ig, IgM, and IgG1 failed to detect Ig-secreting cells in BCR-stimulated primary splenic B cell cultures, whereas numerous secretory cell spots, indicating the presence of Ig-secreting cells, were observed in wells containing cells cultured in the presence of LPS for 72 h (Fig. 1C) (data not shown). The absence of BCR-induced plasma cell differentiation events was important to assess and document in our system for the studies that follow.

Prototypical UPR Chaperone-encoding Genes are Up-regulated in Response to BCR Signals in Populations Consisting of Predominantly
Mature B Cells-Physiologic expansion of the cytoplasm and ER is a characteristic outcome of the adaptive UPR. Recently, several groups have reported evidence that a unique, anticipatory UPR occurs during early phases of LPS-triggered plasma cell differentiation, preceding the increased expression of Ig (5,11,12). To investigate whether a similar response could be initiated by the BCR-induced activation signal, we treated primary splenic B cells with anti-BCR and examined whether UPR pathways were activated.
The most highly conserved indicator of UPR activation is the transcriptional up-regulation of genes encoding ER chaperones. This outcome is observed in pharmacologically induced UPRs as well in all reported examples of physiologic UPRs during B cell differentiation. As indicators of activation of the UPR, we measured the transcriptional up-regulation of the prototypical UPR target genes BiP, Grp94, and calreticulin in response to BCR signals in primary B cells. Splenic B cells were cultured with either anti-BCR or LPS for up to 72 h, and mRNA levels of each gene were measured by quantitative RT-PCR.
BiP/Grp78 is a prototypical UPR target gene encoding a critical ER chaperone (42). Like the known UPR inducer LPS, BCR stimulation with anti-BCR led to transcriptional up-regulation of BiP mRNA in primary B cells (Fig. 2, left). BCR-triggered BiP up-regulation occurred early in the response, with 2-3-fold increases in BiP transcripts at 24 h poststimulation. BiP up-regulation by LPS was somewhat less than that induced by BCR signals at 24 h, but B cells cultured with LPS increased expression of BiP mRNA to a maximum of ϳ3-fold by 72 h poststimulation. In contrast, BCR-induced transcriptional up-regulation of BiP was transient, returning rapidly to unstimulated levels by 48 h post-stimulation.
Up-regulation of Grp94 and calreticulin transcription, other well characterized UPR target genes encoding ER luminal chaperones, was also evident in BCR-stimulated cells (Fig. 2, middle and right). Similar to BiP mRNA, Grp94 and calreticulin mRNA levels increased to ϳ3-fold higher at 24 h than initially observed at 0 h. LPS stimulation also up-regulated these targets, with maximal levels achieved late in the response. Again, the response in BCR-stimulated cells was relatively short lived, and levels of Grp94 and calreticulin mRNA were reduced by 72 h, whereas LPS-stimulated cells continued to increase expression of these chaperone genes over time.
To exclude the possibility that this response reflected simply a general activation process associated with induction of cell cycle, we examined cell line models for BCR-triggered UPR target gene activation. As in primary B cells, BCR stimulation with anti-Ig induced rapid up-regulation of BiP, Grp94, and calreticulin in Bal-17 cells (data not shown). These data indicate that the induction of UPR target genes in BCRstimulated cells is not due solely to the activation of quiescent cells.

BCR Signals Up-regulate Xbp-1 Transcription and Lead to Detectable mRNA Splicing and Protein Expression but Do Not Induce Significant
XBP-1 Activity-The transcription factor XBP-1 is an important amplification loop for the UPR and is critical for plasma cell differentiation (9,(43)(44)(45). Production of the active form of XBP-1 requires an IRE1-dependent mRNA splicing event, yielding Xbp-1(s), which encodes spliced XBP-1 protein (sXBP-1). Chaperone induction in the absence of significant XBP-1 activation has previously been observed in the early, nonclassical phase of the LPS-triggered UPR (11,12). Since the extensive ER remodeling associated with plasma cell differentiation requires sXBP-1 (46), we hypothesized that XBP-1 was not highly active in the BCR-triggered UPR.
To test this hypothesis, we measured BCR-induced transcriptional activation of Xbp-1, assessed Xbp-1 splicing, and examined sXBP-1 protein levels and activity. Xbp-1 transcription was up-regulated ϳ3-fold at 24 h post-stimulation in cells treated with anti-BCR (Fig. 3A). BCRinduced Xbp-1 transcription was more rapid than that of LPS-stimulated cells, but significantly elevated levels of Xbp-1 transcripts were not maintained at 48 h. This response contrasts with LPS-stimulated cells, which continued to increase Xbp-1 expression through 72 h, eventually achieving very high levels of Xbp-1 mRNA. Thus, BCR signals induce Xbp-1 transcription more modestly than does LPS. The BCR-induced Xbp-1 response is maximal at 24 h post-stimulation, coinciding with maximal expression of the chaperone target genes BiP, Grp94, and calreticulin.
We were interested in whether Xbp-1 splicing occurred in BCR-stimulated cells and therefore examined levels of spliced Xbp-1 mRNA (Xbp-1(s)). The presence of Xbp-1(s) mRNA was measured both by quantitative RT-PCR using primers designed to span the excised portion of Xbp-1 mRNA as well as using a previously described semiquantitative RT-PCR technique (11). The more sensitive quantitative RT-PCR technique revealed that levels of Xbp-1(s) are increased in both LPS-and anti-BCR-treated cells at 24 h, an increase that was not detectable by semiquantitative RT-PCR (Fig. 3B). Although Xbp-1(s) mRNA levels increased in BCR-stimulated B cells, the increase was significantly less than that observed in cells cultured with LPS for 48 -72 h.
To determine whether sXBP-1 protein was expressed, we measured protein levels by immunoblot. Levels of sXBP-1 protein increased slightly in both BCR-and LPS-stimulated cells at early time points, but expression in the BCR-stimulated cells was not maintained. In contrast, LPS-stimulated cells continued to increase expression levels of sXBP-1 protein, eventually achieving levels significantly higher than those observed in BCR-or LPS-stimulated cells at 24 h (Fig. 3B, bottom). Thus, transient increases in both Xbp-1(s) mRNA and sXBP-1 expression levels were observed in BCR-stimulated B cells, but BCR-induced expression of Xbp-1(s) and sXBP-1 was significantly lower than that induced late in the LPS response (Fig. 3B). The functional significance of sXBP-1 protein expression was assessed by examining the transcriptional activation of the well characterized XBP-1 target gene Edem-1 by quantitative RT-PCR. Edem-1 mRNA levels were not significantly up-regulated in BCR-stimulated cells, whereas LPS-stimulated cells demonstrated significant Edem-1 up-regulation by 72 h post-stimulation (Fig. 3C). These results indicate that whereas BCR signals up-regulate Xbp-1 transcription and lead to measurable Xbp-1 splicing and protein expression, they are not associated with sXBP-1 transcriptional activity. In each assay examining XBP-1 pathway activation, BCR-stimulated and LPS-stimulated B cells responded equivalently at 24 h poststimulation. Thus, the BCR-triggered UPR is very similar to the adaptive UPR that precedes increased protein load in LPS-stimulated B cells and appears to consist of up-regulation of UPR target genes encoding Xbp-1 and chaperones and minimal XBP-1 activation.  (32,33,(47)(48)(49). Self-reactive immature stage B cells are eliminated in the bone marrow and spleen by negative selection, a process initiated by the antigen-BCR interaction (26 -30). Thus, BCR signals in transitional immature B cells trigger an additional nonsecretory fate, apoptosis. We were interested in determining whether UPR activation was a general feature of BCR stimulation or was more closely associated with cells undergoing an activation response. An exploration of a possible BCRinduced UPR in transitional immature B cells also was interesting, because it represented an opportunity to study a physiologic UPR in an apoptotic context. Programmed cell death is a known consequence of prolonged unresolved UPR induction, but it has been studied only in response to pharmacologic or pathophysiologic UPR induction. We hypothesized that if BCR signals in transitional immature B cells initiated a UPR that played a role in their apoptotic fate, this UPR would be distinct from that observed in mature B cells and would instead share similarities with previously described proapoptotic UPRs.
To directly observe the effect of BCR signaling on B cells at both the transitional immature and mature developmental stages, we compared the cell morphology of BCR-stimulated primary B cells by EM. Splenic populations of B cells were sorted by flow cytometry. Cells were then stimulated with anti-BCR and subjected to EM analysis to observe morphologic changes associated with BCR signaling in cells of each maturation stage at time points corresponding to early (4 h) and late (10 h) stages of the BCR-induced transitional B cell apoptotic response. Resting transitional immature and mature B cells appeared very similar in size and morphology (Fig. 4, left). Early in the BCR response, both developmental stages demonstrated small increases in cell size (Fig. 4, far right) and in cytoplasmic content by visual inspection (Fig. 4, middle). However, in contrast to mature B cells, by 10 h poststimulation, transitional immature B cells had taken on a classic apoptotic appearance with condensed nuclei and fragmented cell membranes (Fig. 4, upper right). Cells examined in this manner were 75-99% morphologically homogeneous, and a representative cell is shown.
We next sought to determine whether BCR signals in transitional immature B cells led to induction of a UPR. As transitional immature B cells rapidly undergo apoptosis in response to BCR signals, we examined B cells up to 8 h post-stimulation, a time point that represents the final phase of fate determination for BCR-stimulated immature B cells (50). Mature B cells were also examined during this time frame to determine whether the BCR-induced UPR previously observed at 24 h is also evident within 8 h of stimulation and to serve as a point of comparison for the transitional immature B cells. This approach allowed us to compare cells in which BCR cross-linking induces an adaptive activating signal (mature B cells) to cells in which BCR stimulation induces an apoptotic response (transitional immature B cells).
An examination of transcriptional activation of UPR chaperone genes revealed that BCR stimulation with anti-BCR led to up-regulation of BiP and calreticulin mRNA within 2-4 h in mature B cells (Fig. 5A). Transitional immature B cells also up-regulated BiP, but these cells were slower to achieve significant levels of BiP up-regulation, reaching 3-fold induction only at 8 h post-stimulation (Fig. 5A, left). Calreticulin mRNA was minimally and transiently up-regulated by transitional immature B cells during the BCR response, in contrast to the mature B cells, which showed continued up-regulation through 4 -8 h (latest time point shown) (Fig. 5A, right). These observed differences in chaperone gene induction were not related to the irradiation process used to enrich for transitional immature B cells, since mature B cells obtained from irradiated mice as described under "Experimental Procedures" exhibited the same pattern of gene induction as mature B cells from unmanipulated mice (data not shown).
The role of XBP-1 in the transitional immature B cell BCR-triggered UPR was examined by measuring levels of total Xbp-1 transcripts, assessing Xbp-1 splicing, and investigating activation of the XBP-1 tar-get gene Edem-1, as described above. Xbp-1 transcription was not highly induced in the early hours of the BCR-triggered UPR in either mature or transitional immature B cells (Fig. 5B, left). Significant splicing of Xbp-1 was not detected by either quantitative RT-PCR or semiquantitative RT-PCR; nor was XBP-(s) protein up-regulated as measured by immunoblot (Fig. 5B, right) (data not shown). In addition, no significant XBP-1 activity was detected as measured by Edem-1 transcription levels (Fig. 5C).   DECEMBER 2, 2005 • VOLUME 280 • NUMBER 48

Chop Is Transcriptionally Induced by BCR Signals in Mature B Cells but Is Maintained at High Levels in Only Transitional Immature B Cells-
Cells unable to productively undergo an adaptive UPR may subsequently initiate a default proapoptotic program involving activation of dedicated apoptotic pathways (18 -25). This process is designated as a terminal UPR. Very little is known regarding the commitment of a cell to an adaptive versus apoptotic fate during the UPR, and the distinction between these two responses, beyond the ultimate cell fate, is poorly understood (51). We hypothesized that the BCR-induced UPR in transitional immature cells represented a terminal UPR. Since the UPR target gene Chop encodes a C(EBP) family transcription factor that has been associated with apoptotic fate in cells undergoing a terminal UPR (35,(52)(53)(54), we chose to investigate whether Chop was involved in the BCR-induced UPR and, if so, whether it was differentially induced in mature versus transitional immature B cells.
We utilized quantitative RT-PCR to determine whether BCR signals induced Chop transcription and to compare the responses of mature and transitional immature B cells. Chop mRNA was up-regulated in both mature and transitional immature cells stimulated through the BCR with anti-BCR (Fig. 5D). However, increased Chop mRNA levels were maintained only in BCR-stimulated transitional immature B cells, not in mature B cells. Sustained Chop up-regulation in transitional immature B cells, destined for apoptosis, suggested a possible inductive role for CHOP in the BCR-induced apoptosis of these cells.
To examine whether Chop maintenance in transitional immature B cells reflected BCR-induced events or was simply a consequence of downstream events induced by apoptosis, we cultured cells in DEVDfluoromethyl ketone, an agent known to block caspase-3 activation (55). Activation of caspase-3 has been shown to be a necessary and irreversible step in the BCR-induced apoptosis of transitional immature B cells (30). Consequently, apoptosis was attenuated in cells receiving both anti-BCR and DEVD-fluoromethyl ketone (data not shown). Nevertheless, Chop mRNA levels were maintained and even elevated through 8 h after anti-BCR stimulation in cells receiving DEVD-fluoromethyl ketone as well as anti-BCR (data not shown). Our results indicate that Chop expression is not induced as a consequence of apoptosis but rather lies upstream of the terminal caspase-dependent events in BCRmediated apoptosis.

Survival Signals Generated through CD40, IL-4 Receptor, and TLR4 (Toll-like receptor 4) Result in Alteration of the BCR-induced Transi-
tional Immature B Cell UPR-Since the UPR is known to control two divergent fates, apoptosis and adaptation to stress, we hypothesized that the differences observed in the BCR-induced UPR in transitional immature versus mature B cells might play a determining role in their developmentally regulated fates. To investigate this possibility, we manipulated the fate of BCR-stimulated transitional immature B cells using physiologic survival signals and examined the associated UPR.
The developmental stage-associated differences in Chop and chaperone target gene expression levels observed at 8 h poststimulation coincide with an important window of opportunity for BCR-stimulated transitional immature B cells. Prior to 8 -10 h post-BCR-stimulation, transitional immature B cells can be diverted from an apoptotic fate by a variety of physiologic survival signals that are likely to be present during an ongoing immune response (50). Beyond this stage, however, the apoptotic fate decision is irreversible. Our results indicate that the UPRs in transitional immature and mature B cells are qualitatively different during this fate decision window, with the transitional immature B cell response characterized by high levels of Chop mRNA coincident with relatively poor chaperone gene activation.
The addition of survival signals transmitted through CD40 (anti-CD40), IL-4 receptor (IL-4), and TLR4 (LPS) at 4 h after anti-BCR stimulation prevented BCR-induced apoptosis of transitional immature B cells (50) (Fig. 6, A and B). We reasoned that if the UPR determines cell fate, then changes known to redirect cell fate in transitional immature B cells would be the result of changes in the BCR-induced UPR that could be observed by examining activation of UPR pathways. We hypothesized that the protected B cells would exhibit a UPR that now appeared more like that of BCR-stimulated mature B cells. In particular, we predicted that changing the fate of a B cell would be associated with decreased Chop expression and increased chaperone gene expression during the fate decision. To test these predictions, we compared transcriptional responses in transitional immature B cells receiving anti-BCR stimulation alone with those receiving both anti-BCR and a survival signal. In BCR-stimulated transitional immature B cells, Chop mRNA is up-regulated ϳ3-fold at 4 h poststimulation as compared with resting cells (Fig. 5D). Anti-CD40, IL-4, or LPS was added to BCR-stimulated transitional B cell cultures at 4 h post-BCR-stimulation, and Chop mRNA expression levels were measured after 4 h of co-incubation (8 h post-BCR-stimulation). We wished to determine whether redirecting cell fate with these survival signals was associated with decreased Chop expression at the time of cell fate determination. Strikingly, the addition of anti-CD40, IL-4, or LPS resulted in decreased Chop mRNA levels in BCR-stimulated transitional immature B cells at 8 h poststimulation as compared with cells treated with anti-BCR alone (Fig. 6C, left). At this time point, transitional immature B cells co-stimulated with anti-BCR and a survival signal expressed Chop mRNA at base-line levels.
Redirection of cell fate in transitional immature B cells was also associated with changes in the activation of chaperone genes. Transitional immature B cells stimulated through the BCR and then co-activated with anti-CD40, IL-4, or LPS at 4 h post-stimulation were found to express higher levels of the chaperones BiP, Grp94, and calreticulin at 8 h poststimulation than those receiving the BCR signal alone (Fig. 6C,  middle and right). These data indicate that transitional immature B cells in which apoptosis is prevented exhibit a UPR induction pattern similar to that of BCR-stimulated mature B cells, favoring expression of chaperone genes over expression of Chop. Thus, under positive stimulatory conditions (anti-BCR or anti-BCR plus co-stimulation, for mature and transitional immature B cells, respectively), Chop expression is transient although detectable, and chaperone gene expression is high. These data indicate that the pattern of UPR gene expression induced by antigen receptor signaling correlates with the ultimate fate of the BCRstimulated B cell rather than simply the maturation stage.
CHOP Is Not Required for BCR-induced Apoptosis-To determine whether there is an absolute requirement for CHOP expression in the apoptotic fate of BCR-stimulated transitional immature B cells, we examined in vivo B cell development in mice genetically engineered to lack Chop. We hypothesized that if CHOP is essential for antigen-triggered immature B cell apoptosis, its loss would manifest in significant aberrations in the numbers of B cells, particularly bone marrow immature and transitional immature B cells. However, when age-matched, sex-matched littermate control animals and Chop-deficient mice were compared, both exhibited equivalent percentages, and absolute numbers of all B cell stages in the bone marrow and spleen and B cells were indistinguishable based upon morphologic markers and cell size (Figs. 7, A and B) (data not shown). Chop Ϫ/Ϫ mice exhibit normal numbers and percentages of two developmental stages of transitional immature B cells (T1 and T2 in Fig. 7), mature follicular B cells (FO), and marginal zone B cells (MZ). These data demonstrate the lack of an effect of CHOP expression on splenic B cell compartments. Likewise, no differences were observed in thymocyte or T cell percentages and distribution (data not shown).
To determine whether CHOP is an essential component of the BCRtriggered apoptosis of transitional immature B cells in vitro, transitional immature cells from Chop-deficient mice were isolated and stimulated with anti-BCR. Both Chop ϩ/Ϫ and Chop Ϫ/Ϫ transitional immature B cells underwent apoptosis in response to anti-BCR (Fig. 7C). Responses compared across several independent experiments failed to establish a consistent or significant difference between Chop-sufficient and -deficient B cells. These data indicate that CHOP is not required for BCRtriggered apoptosis of transitional immature B cells.

DISCUSSION
Using an ex vivo primary B cell model, we have identified an unexpected association between BCR signals and the UPR, documenting for the first time that physiologic activation of UPR pathways occurs in nonsecretory cells. Although BCR signals induce activation and proliferation of B cells, without additional non-BCR signals these activated B  DECEMBER 2, 2005 • VOLUME 280 • NUMBER 48 cells fail to differentiate into antibody-secreting plasma cells. Using anti-BCR to stimulate mature B cells through the BCR, we have shown transcriptional activation of prototypical UPR target genes encoding Xbp-1 as well as the chaperones BiP, Grp94, and calreticulin during the first 24 -48 h poststimulation. This pattern of observed UPR pathway induction in response to BCR signals closely resembles that previously reported in B cells stimulated with LPS, a differentiation signal.

BCR-induced UPR in Nonsecretory B Cells
In recent years, several laboratories have examined the involvement and importance of the UPR in the terminal differentiation of B cells into plasma cells, primarily utilizing LPS-stimulated B cells or B cell lines as model systems (5,11,12,51,56). The association of this differentiation program with initiation of a UPR is predictable, as plasma cell function is dependent upon the ability to secrete large amounts of Ig. However, there is a growing body of evidence that selective induction of UPR pathways in the process of plasma cell differentiation occurs prior to the influx of Ig into the ER. Consequently, the early LPS-or cytokine-triggered plasma cell UPR has been proposed to anticipate rather than respond to ER stress. In support of this model, several groups have demonstrated that up-regulation of UPR chaperones (5,11,12), and, in some cases, XBP-1 activation (5) occurs prior to significant up-regulation of Ig production. A corollary of this model is that the scope of the physiologic UPR may extend to numerous processes in which preparative ER remodeling may desensitize cells to anticipated stress.
The importance of our work with regard to this model is in its ability to dissociate physiologic UPR initiation from obligatory initiation of an antibody secretion response. We confirm that BCR stimulation does not lead to plasma cell differentiation, yet we demonstrate that BCR signals initiate UPR signaling pathways. Thus, the physiologic UPR is not limited to situations in which professional secretory cells will surely be generated. The BCR signal is an activation signal that with subsequent differentiation signals may lead to plasma cell differentiation, but the likelihood of this outcome for a particular antigen-stimulated B cell is low.
The transient but rapid UPR response, observed from 2 to 24 h post-BCR stimulation, occurs during the window in which BCR-stimulated B cells migrate to specialized lymphoid microenvironments to await T cell second signals. Therefore, this response may be a mechanism for activated B cells to prepare for possible plasma cell differentiation signals. An increase in cytoplasmic membranes, suggesting expansion of the ER, a modification that enhances folding capacity and could serve a protective function in the event of increased ER protein load, is observed in cells receiving BCR signals alone. Entry into an XBP-1 predominant UPR, as observed in the second phase of the LPS response (48 -72 h poststimulation), may occur following interaction with T cells from subsequent activation and differentiation signals. Indeed, CD40 and IL-4 signaling have been demonstrated to induce chaperone up-regulation and XBP-1 activation in B cells (11).
Our demonstration that the BCR signal initiates a UPR similar to that induced in the first day of the LPS differentiation process adds support to the concept that this early phase of the LPS response may be anticipatory. However, the idea of an anticipatory UPR remains speculative, and the mechanism by which B cells "anticipate" upcoming ER stress in response to LPS or anti-BCR remains unclear. Indeed, it is possible that this adaptive UPR may be triggered by a classical mechanism, although it precedes Ig accumulation. BCR-triggered calcium release from the ER may allow the B cell activation signal to immediately begin ER preparation via a classical UPR induction mechanism. Future studies will be required to elucidate the mechanisms by which early, selective activation of UPR pathways are initiated and regulated during BCR-triggered activation and LPS-or cytokine-triggered plasma cell differentiation.
In addition to demonstrating that a physiologic activation signal initiated UPR signaling in the absence of plasma cell differentiation, we have shown that UPR pathways are activated during physiologic apoptosis. Whereas prolonged unresolved ER stress resulting from pharmacologic or pathophysiologic induction of a UPR is known to induce apoptosis, no evidence for a physiologic terminal UPR has previously been reported. Since many developmental processes involve regulated cell death, it makes teleological sense that the UPR, a response every cell is capable of mounting, has been co-opted for this developmental purpose as well. Transitional immature B cells stimulated with anti-BCR undergo apoptosis in the context of UPR pathway activation with the clear association with a proposed UPR-specific death pathway, consistent with activation of a terminal UPR.
In contrast to mature B cells, transitional immature B cells induce a UPR that is characterized by elevated Chop mRNA rather than chaperones or Xbp-1, particularly during the phase of final fate determination. The observation that distinctive UPR pathways with known implications for cell fate dominated the BCR-induced UPR observed in mature as compared with transitional immature B cells suggested that differences in the UPR could control cell fate in response to BCR signals. Experiments in which apoptosis of BCR-stimulated transitional immature B cells is prevented with physiologic survival signals suggest that the selective activation and maintenance of particular UPR pathways are closely associated with cell fate. In these experiments, transitional immature B cells were co-stimulated with anti-CD40, IL-4, or LPS. This co-stimulation prevented apoptosis of the BCR-stimulated cells and was associated with predictable changes in the observed UPR. Under these conditions, transitional immature B cells demonstrated reduced Chop expression and increased chaperone gene expression during a critical phase of the response, when cell fate is being determined. It is possible that the mechanism by which these distinct signals promote transitional immature B cell survival involves converting a terminal UPR into an adaptive UPR.
Our data support a model in which cell fate during a UPR is determined not by one aspect of the adaptive or terminal UPR but by the relative balance of pathways induced and maintained through critical phases of the response. Our studies in Chop Ϫ/Ϫ mice indicate that CHOP does not function as a master regulator of apoptosis during the BCR-induced UPR in transitional immature B cells. Indeed, Chop Ϫ/Ϫ transitional immature B cells retain the ability to die in response to BCR stimulation. Because of the tight correlation between cell fate and Chop mRNA expression in our co-stimulation assays, this result was somewhat surprising; however, it is consistent with previous reports that indicate that redundancies exist in the terminal UPR. Embryonic stem cells lacking protein kinase-like ER kinase exhibit significantly blunted Chop induction in response to pharmacologic initiation of the UPR (18) yet undergo higher levels of apoptosis than wild-type cells (57). In addition, Chop Ϫ/Ϫ mouse embryonic fibroblasts die in response to pharmacologic UPR induction, albeit with slower kinetics than control cells (35). Thus, whereas CHOP does not appear to be required for BCRinduced death, it remains possible that this molecule may still contribute to the terminal fate of BCR-stimulated transitional cells.
The implications of these studies extend beyond the field of B lymphocyte biology. Our system provides the first opportunity to directly observe and modulate both the terminal and the adaptive UPR in response to an identical trigger. Using this system, we have identified characteristic features of physiologic UPRs triggered under adaptive versus proapoptotic conditions. Although many of the pathways involved in the mammalian UPR are now known, it is clear that the mammalian UPR, particularly the physiologic UPR, is a highly complex, multifactorial response. There are probably additional molecules involved, unknown triggers, and previously unexpected roles for UPR signaling during physiologic processes. Since the kinetics of the BCR response are well characterized and BCR-triggered death is reversible, BCR signaling may prove to be a unique means for studying the importance and reversibility of a variety of adaptive and terminal UPR pathways in a physiologic context.