Here, There, and Anywhere? Arguments for and against the Physical Plasma Cell Survival Niche

The induction of immunity due to infection or vaccination often results in long-term Ab-mediated protection. Indeed, vaccine-induced Ab titers induced for many common viral Ags are exceptionally durable, with estimated half-lives surpassing 1000 years (1). Until 1997, it was thought that all Ab-secreting plasma cells were short-lived cells, with half-lives that failed to exceed more than a week. Consequently, because Ab molecules are intrinsically short lived (2, 3), it was proposed that long-term maintenance of Ab titers required the periodic stimulation of Ag-specific memory B cells (4), which in turn generated waves of short-lived plasma cells. However, over the past two decades several studies have shown that many plasma cells can persist indefinitely (5–7). In fact it now appears that a large fraction of serum Abs derive from long-lived plasma cells residing in the bone marrow (BM) (1). Yet many newly formed plasma cells appear to die within days of their generation (8–11). These observations raise the intriguing question of why so many newly formed plasma cells die whereas others persist.

It is widely accepted that plasma cells must continuously capture and process a variety of unique signals from their external environment to avoid death. Perhaps because many plasma cells home swiftly to the BM soon after their generation (12, 13), it is generally thought that plasma cells must integrate into specialized BM microenvironments, or niches, where they must remain indefinitely to survive (6, 14). However, plasma cells exhibiting hallmarks of long-lived cells have also been captured in the spleen of mice and people (6, 15–17), and the lungs of influenza-infected mice (18), and recent data indicate that IgA-secreting plasma cells in the gut are also long lived (19). What then, if anything, is unique about these tissues? In this study we will review current knowledge on the cell-extrinsic and cell-autonomous processes associated with plasma cell survival, and consider the strengths and limitations of the physical niche survival model.

Before considering the survival issue, let us consider past work placing long-lived plasma cells as key components in adaptive humoral immunity. During the latter half of the 19th century, an intense debate emerged concerning the basic mechanisms underlying natural and adaptive immunity. The fundamental issue at hand was whether host defense is mediated by the activity of cells versus soluble mediators in the blood (or humor). Despite clear evidence showing that macrophages and other phagocytic cells digest and destroy microbes (20), by the early 1900s it was also clear that Abs against exotoxin mediated immunity to tetanus and diphtheria (21). Soon thereafter, it was shown that passive transfer of serum from immunized donors alone could protect naive recipients from subsequent infections (22). As it became undeniable that Abs are central protective elements for many pathogens, questions about the events governing Ab production and how Ab structure influences Ab function began to take hold. Major insights into both issues came many years later from the study of histological analyses of peripheral lymphoid tissues. A variety of pathologists had noted that the red pulp of the spleen was routinely enriched for plasma cells identified readily due to their characteristic expanded endoplasmic reticulum (ER). Notably, although implied by many earlier studies, the connection between plasma cells and Ab production came in 1947 when Fagraeus showed that placement of splenic red pulp fragments from twice immunized rabbits into culture led to the easy detection of Ag-specific Ab in the supernatant (23). Subsequently it was confirmed using more direct histological techniques that in situ plasma cells contained and released large quantities of Ab (24–26), and thus serve as key components of humoral immunity. In parallel, it was noted that mouse plasmacytoma cell lines secrete enormous quantities of Ig and also possessed a widely expanded ER (27–29), leading to additional questions about how the unique morphology of these cells facilitates robust Ab synthesis and secretion. Together these observations set the stage for studies attempting to elucidate the basic biology of Ab-secreting cells, including their relationship with resting and activated B cells and their role in establishing and maintaining humoral immunity.

When during B cell differentiation do early plasma cells become receptive to cell-extrinsic and cell-autonomous survival cues? Early during primary Ab responses Ag-engaged B cells undergo clonal expansion, and then subsequently yield an initial wave of memory B cells and plasma cells (30–32). In parallel, other activated B cells initiate and localize within germinal centers (GCs), unique and largely T cell dependent anatomic structures enriched for cells undergoing robust clonal expansion along with class-switch recombination, somatic hypermutation, and affinity-based selection (33, 34). It is been proposed that most long-lived plasma cells arise from GCs (31, 35). This idea certainly has merit; due to robust clonal expansion within GCs, it is easy to envision that GC-derived cells dominate Ag-specific plasma cell pools. But does this mean that immature plasma cells only become receptive to life-sustaining signals when or after experiencing GC microenvironments? We suggest otherwise. Over the past few years several studies have shown that a variety of T cell–independent Ags, which fail to evoke meaningful GC responses (36, 37), readily induce durable Ab responses and long-lived plasma cells (30, 38, 39). Moreover, prevention of GC reactions early in responses to T cell–dependent Ag leads to fewer Ag-specific BM plasma cells, but the resulting cells are clearly long lived (30). Thus, although many, and perhaps most, long-lived plasma cells induced by protein-rich T cell–dependent Ags arise from GC-experienced B cells, it is unlikely that GCs provide the unique environments needed for plasma cells to become receptive to life-sustaining signals. When, then, during differentiation do early plasma cells become receptive to requisite survival signals? And do all early plasma cells become receptive, or do many new plasma cells die simply because they fail to respond to needed cell-extrinsic and cell-intrinsic pathways? To answer these questions, we will need to consider the unique signals and events employed by plasma cells to avoid apoptosis.

Early experiments focused on peripheral lymphoid organs revealed that plasma cell populations in these tissues experience a high degree of turnover, thus lending to the idea that plasma cells are short lived, with half-lives ranging from a few days to 2–3 wk at most (25, 40–42). Despite the dominance of this idea for many years, two classic papers subsequently established that reasonable numbers of newly generated plasma cells survive to become long-lived cells without input from naive or memory B cells (5, 6). Consequently, it is now generally believed that plasma cells that manage to avoid apoptosis during the early phases of Ag-induced differentiation go on to survive substantially longer than naive lymphocyte populations. How, then, is this achieved? Survival mechanisms for plasma cells are likely to be quite distinct from those employed by other long-lived immune cells, such as memory B cells. For starters, plasma cells secrete as many as 10,000 Abs per second (43, 44), suggesting the need for plasma cells to enact appropriate biochemical pathways to coordinate the huge energy demands needed to synthesize large quantities of proteins, presumably without pause. In this context, a key question is to what extent biochemical events needed for plasma cell survival are enacted from within versus from extracellular cues derived from cell-cell interactions within dedicated tissues such as the BM. As described further below, the answer is probably both: plasma cell survival appears to involve unique biochemical processes induced by intracellular activation of the unfolded protein response (UPR), but it also requires implementation of additional antiapoptotic pathways stimulated by cytokines and other extracellular factors.

One pathway employed by plasma cells to avoid death due to the stress associated with constant robust protein secretion is the UPR. Indeed, early plasma cell differentiation requires activation of the UPR-associated transcription factor XBP1 as well as the ER sensor, IRE1 (45–47). IRE1 is activated in response to the increased ER protein load inherent to plasma cell function (48). Remarkably, upon activation IRE1 migrates into the nucleus where it serves as a largely XBP-1–specific RNA splicing enzyme, causing a needed frame shift in the XBP-1 transcript required to code for a functional transcription factor (49). The cytokine IL-4 has been implicated in activation of the UPR (45), however, there is no reason to consider IL-4 an essential plasma cell survival factor. In fact, because the UPR initiates in response to increased intracellular concentrations of misfolded proteins, the resulting survival pathways might be considered cell autonomous. In fact, to our knowledge none of the proposed extracellular plasma cell survival factors described below have been shown to regulate the UPR.

Plasma cells possess additional unique features that imply they can readily adapt to events and scenarios that would cause other B lineage cells to die. For instance, unlike naive and memory B cells, the vast majority of immature and mature plasma cells are resistant to DNA damage-invoked cell death pathways (6, 30). Perhaps similarly, whereas inhibition of mTOR activity causes activated B cells to die, plasma cells adapt to loss of mTOR signaling by restraining Ab synthesis and increasing autophagy (50). Therefore, it appears that plasma cells can use a variety of cell-autonomous mechanisms to adjust to challenges that cause many other B lineage cells to perish.

Unless provided with stromal cells and a variety of additional factors, plasma cells die almost immediately when placed in culture (51), emphasizing their dependence on additional extracellular inputs. Indeed, in our hands plasma cells die much faster than naive B cells when cultured in simple tissue culture medium. Therefore, to avoid apoptosis, plasma cells are likely to use additional non-cell–autonomous survival mechanisms. Contemporary models predict that plasma cells only gain access to needed survival cues when localized within specialized regulatory microenvironments, or niches (52). Indeed, several studies suggest that disruption of putative survival niches causes many plasma cells to die (53–58). However, although it is clear that plasma cell survival requires inputs from cytokines and other factors, and recent work suggests that plasma cells in the BM are not highly migratory (59), to our minds that plasma cells must remain indefinitely in discreet unique microenvironments to gain access to these signals remains an open question. In this study we will review current information on extrinsic plasma cell survival cues, and consider opposing models for where and how they access such signals.

A wide variety of cytokines, integrins, and chemokines have been implicated in plasma cell survival. The cytokine IL-6 is associated with Ab titer maintenance due to its capacity to enhance plasma cell survival in vitro (60). The chemokine CXCL12 also promotes plasma cell survival in vitro (61), lending to models wherein CXCL12-expressing stromal cells guide plasma cells toward unique environments in the BM important for their survival (see below) (14). Consistent with this possibility, newly generated plasma cells fail to colonize the BM of mice lacking the CXCL12 receptor CXCR4 (62, 63). Similarly, disruption of adhesive interactions mediated by the integrins LFA-1 and VLA-4 causes a temporary loss of BM plasma cells (64). Thus CXCL12 may attract new plasma cells to unique microenvironments rich in anti-apoptotic survival factors where they are retained by integrins such as LFA-1 and VLA-4. However, plasma cell survival is not affected dramatically in IL-6–null mice (60), the impact of LFA-1/VLA-4 blockade appears to be transient (64), and whether in vivo CXCR4 or CXCL12 deletion causes mature plasma cells to die has not been tested directly.

Perhaps the most convincing evidence for a nonredundant cell-extrinsic plasma cell survival signal stems from work with mice lacking the cytokine receptor BCMA. The BLyS cytokine family includes three receptors, BR3, TACI, and BCMA, and the two cytokines APRIL and BLyS (or BAFF) (65). Unlike IL-6 deficiency, BCMA deletion causes profound loss of BM plasma cells (66). Furthermore, BCMA appears to facilitate plasma cell survival by promoting the activity of the Bcl-2 family member Mcl-1 (67). Notably, BCMA binds effectively to BLyS and APRIL, suggesting that either cytokine can activate BCMA and facilitate plasma cell survival. Thus a key question that emerges is whether BCMA facilitates plasma cell survival by binding to APRIL or BLyS. Or will either cytokine suffice? Indeed, whereas transfer of cells into APRIL^−/− mice compromises plasma cell survival (68, 69), Benson et al. (70) showed that in intact APRIL^−/− animals plasma cell numbers only decline after neutralizing systemic BLyS levels. Hence, these results indicate that to survive plasma cells need access to either APRIL or BLyS. However Belnoue et al. (69) demonstrated that plasma cell frequencies can be compromised in APRIL-deficient mice, either 48 h after transfer of early splenic plasmablast-like cells into APRIL-deficient hosts, or 7 d after secondary immunization of intact APRIL^−/− mice. Why the discrepancy? Although more work is needed in this area, it is worth noting that whereas Benson et al. examined plasma cell frequencies in APRIL^−/− mice 3 mo postimmunization, Belnoue et al. employed much shorter time points ranging from 2 to 7 d. It is tempting to speculate therefore that these disparate results reflect (at least in part) an enhanced need for BCMA/APRIL interactions during earlier phases of plasma cell differentiation, perhaps before newly formed plasma cells begin to migrate to the BM. This leads us to the question of where plasma cells access these and other relevant factors.

Current models predict that plasma cells only gain access to requisite survival signals such as APRIL within dedicated size-limited regulatory microenvironments, or niches (52). Perhaps we should discuss the term niche in this context. Typically in physiology the term niche evokes visions of unique physical microenvironments within tissues and organs. However, the size and composition of a particular cell population can also be constrained by factors irrespective of physical microenvironments. For example, take mature recirculating B cells, which rely heavily on access to sufficient concentrations of BLyS. The cytokine BLyS is readily detected in serum in mice and people (71, 72). Nonetheless, the size of the naive recirculating B cell pool is clearly controlled by BLyS availability: increasing BLyS levels leads to an increased pool size, and decreasing serum BLyS concentrations leads to a decreased pool size (73). Therefore, concentrations of a systemic cytokine can control life and death decisions for B cells throughout the body, even though the majority of these cells recirculate throughout the blood and lymphatic systems every few days (74).

In contrast, it is generally thought that new plasma cells must migrate to and remain in specialized physical niches, in the BM or elsewhere, to access critical survival cues and maintain Ab titers for the long term (see Fig. 1A). This idea is consistent with recent imaging data suggesting that most BM plasma cells form durable physical interactions with stromal cells (59), and other data illustrate that plasma cells form grape-like clusters throughout the BM (75). However, these results do not exclude the possibility that long-lived plasma cells enter the circulation from time to time, especially during systemic inflammatory responses known to disrupt many aspects of BM homeostasis. Indeed, such events might cause mature BM plasma cells to relocate to the spleen or other sites. As mentioned above, mature plasma cells can be readily captured in the spleen at steady state (6, 15–17), with decay rates that mirror their counterparts in the BM (6, 15–17), and drug-induced blockade of bone remodeling has been shown to divert plasma cell (and megakaryocyte) homing from the BM to the spleen without impairing serum Ab responses (76). Given that CXCL12-expressing stromal cells are also found in the splenic red pulp (62), it is tempting to predict that early in terminal B cell differentiation, cohorts of newly formed plasma cells lodge in CXCL12-rich regions of the spleen where they may remain for extended periods as long-lived cells. It should also be noted that numbers of mature splenic plasma cells appear to increase in NZB/W mice, which experience a lupus-like syndrome, and in patients with immune thrombocytopenia (9, 77). Although the migratory properties of these cells remain undefined, it would appear that not only can the spleen harbor long-lived plasma cells, but it may also serve as a reservoir for mature BM plasma cells during systemic inflammatory responses.

Plasma cells are also known to invade inflamed sites, such as the kidneys of lupus-prone NZB/W mice (9, 78). Notably, in one study many plasma cells captured in inflamed kidneys were shown to be specific for exogenous Ags used to immunize these mice much earlier, before the onset of overt inflammation (78). This result suggests that pre-existing plasma cells, most likely in the BM, can migrate into additional alternative tissues during inflammatory states. Indeed, cells within inflamed sites are also known to express CXCL12 and APRIL. Therefore it is likely that disruption of normal marrow homeostasis due to systemic inflammation causes an unknown fraction of BM plasma cells to relocate into alternative CXCL12-rich tissues.

Another layer of complexity stems from recent work showing that in mice and people as many as 40–50% of BM plasma cells are immature and largely short-lived cells (11, 79). These results suggest that life and death decisions for immature plasma cells can occur after entry into the BM. Furthermore, these findings are consistent with the notion that plasma cell death in the BM reflects the failure of these cells to gain access to relevant physical niches orchestrated by CXCL12-expressing cells. In this regard, CXCL12 is expressed by a variety of cells throughout the BM including osteoclasts, hematopoietic stem cells (HSCs), and perivascular reticular and endothelial cells (80). Given that osteoclasts reportedly produce BLyS and APRIL (81), one possibility is that plasma cell niches in the BM are orchestrated by APRIL and CXCL12-producing osteoclasts. From this point of view, mature plasma cells would be predicted to localize specifically in the bone-proximal osteoclast-rich endosteal regions of the marrow. This might make sense, given that malignant plasma cells in multiple myeloma are known to perturb normal bone homeostasis (82). Nonetheless, plasma cell biologists thus far have focused mainly on hematopoietic cells as sources of APRIL. In fact, basophils, megakaryocytes, and eosinophils have each been implicated in plasma cell survival due to their expression of relevant cytokines such as APRIL and IL-6 (53–55, 76).

At first glance, that eosinophils and other hematopoietic cells regulate plasma cell survival might be surprising given the very different population dynamics of the cells in question. Unlike plasma cells (59), eosinophils are short-lived cells known to migrate swiftly from the BM into the blood and peripheral lymphoid tissues to promote various aspects of innate immunity (83). Likewise, megakaryocytes and basophils are also short-lived motile cells. Therefore, maintenance of Ab-titers would require a multitude of constant cell-cell interactions between static long-lived plasma cells and these rather dynamic short-lived cell types. A second issue concerning these ideas is the lack of information on potential underlying mechanisms. One possibility is that hematopoietic cells implicated in plasma cell survival each express abundant levels of APRIL and/or BLyS. Although eosinophils and megakaryocytes appear to express APRIL (53, 54), the impact on Ab responses of abrogating APRIL expression specifically by these cell types has not been reported. Moreover, APRIL and BLyS appear to be produced by many cell types, and as mentioned above, BLyS can be detected in the serum (71). Therefore if plasma cell survival does indeed require that they integrate into unique regions of the BM, additional niche-specific survival cues may be at play. Finally, it should also be noted that depletion of eosinophils, basophils, and megakaryocytes each caused only a 2-fold reduction in plasma cell numbers, suggesting a large degree of redundancy in the cell-cell interactions supporting plasma cell niches. Alternatively, given that only 50–60% of BM plasma cells are long lived, depletion of each of these plasma cell–supporting cell types might result in a selective impact on long-lived plasma cell numbers, leading to a more impactful effect on humoral immunity. Or instead, these manipulations might lower plasma cell input into the BM, lessening the impact on truly long-lived cells.

Other work suggests that plasma cells routinely interact with cells bearing ligands for the T cell costimulatory receptor CD28. In naive mature B cells, the transcription factor Pax5 silences a battery of genes whose function is inconsistent or at least dispensable for B cell function (84). However, early during plasma cell differentiation activated cells downregulate Pax5, leading to de-repression of a variety of genes including CD28 (85) and Flt3, a cytokine receptor needed for very early B cell development (86). Data from Lee and colleagues (87) suggest that plasma cells exploit CD28 signaling to avoid apoptosis, thus raising the possibility that dendritic cells and/or perhaps other cells bearing CD28 ligands are important components of plasma cell niches. However, data from Jacobs and colleagues paint a very different picture, as in their hands CD28-deficient plasma cells appeared to produce higher concentrations of Ab, a result that was recapitulated when B lineage cells lacked canonical CD28 ligands (88). Evidently, additional work is needed in this area. Interestingly, recent data indicate that depletion of regulatory T cells (Tregs) from the BM also causes loss of plasma cells (89). In this regard, past work suggests that Tregs protect HSCs from inflammatory signals with the potential of disrupting hematopoiesis (90). Thus this function may extend to other nearby cells. This may also reveal information on the location of BM plasma cells. HSCs were once thought to reside preferentially in bone-proximal endosteal regions of the BM (91). However, it is now clear that HSCs reside mainly in perivascular regions of the marrow where they may experience increased concentrations of CXCL12 (80). Perhaps BM plasma cells, Tregs, and HSCs preferentially home to CXCL12-rich perivascular spaces where Tregs, stimulated by BM dendritic cells, regulate poorly defined aspects of their survival and function.

IgA-secreting plasma cells along the gastrointestinal tract have long been considered a key immune mechanism for advancing symbiosis with commensal microbes (92). Recent work suggests that IgA Abs preferentially target potentially dangerous bacterial taxa (93). Indeed, prevention of class switching by AID deletion causes intestinal dysbiosis, allowing anaerobic bacteria to breach the intestinal epithelium (94).

The composition of the intestinal microbiota can fluctuate due to numerous factors including aging and disease (95). Therefore, it is likely that mucosal B cells must rapidly produce new IgA-secreting cells when colitogenic taxa initially integrate into complex bacterial communities in the gut. One possibility is that all IgA-secreting plasma cells are short lived, thus allowing for constant remodeling of the IgA repertoire in response to an ever-evolving microflora. However, recent data suggest that the bulk of IgA-secreting cells in the gut are long lived (19). This result raises a variety of questions. To survive, must IgA-secreting plasma cells also remain in size-limited niches in the small intestine lamina propria? What extrinsic factors drive plasma cell survival in the gut, and to what extent do these signals overlap with those employed by BM plasma cells? Remarkably, aside from one report proposing that eosinophils are also components of a small intestine lamina propria plasma cell niche, virtually no information on this issue is available. Thus an important focus going forward will be to discern the factors controlling plasma cell lifespan in the gut, and to evaluate the impact of perturbing these processes on intestinal health.

We suggest that it is reasonable to approach the notion of physical plasma cell survival niches with some skepticism. It is clear that most BM plasma cells rely heavily on access to APRIL or BLyS (66, 70), and it appears that mature plasma cells are relatively stationary (59). However, to us, that plasma cells must remain indefinitely in physical survival niches to survive is less obvious. To begin, if plasma cells can indeed use APRIL or BLyS interchangeably (70), because BLyS is by and large a systemic cytokine (96), it is difficult to imagine scenarios where plasma cells require life-long residence in physical APRIL-rich survival niches, in the BM or elsewhere. Second, it should be emphasized that, whereas systemic inflammatory responses can severely deplete BM plasma cell pools (89, 97), likely through temporary loss of CXCL12 expression (98), there is also evidence that such cells can relocate into alternative CXCL12-rich tissues (78). This general assertion also applies to experiments wherein integrin blockade led to transient loss of BM plasma cells (64). Therefore, CXCL12 may attract many newborn plasma cells into the BM, but such cells may also migrate elsewhere and survive to secrete more Ab. Third, whereas depletion of eosinophils or megakaryocytes reportedly drives down the number of Ag-specific plasma cells (53, 54), whether depletion of either cell type leads to selective loss of long-lived cells remains to be shown directly. Indeed, in each case, the described loss in plasma cell numbers was far from dramatic, and it remains difficult to track the fate of pre-existing long-lived cells in these experiments. Additionally, perhaps eosinophil or megakaryocyte depletion coincides with increases in long-lived plasma cells at alternative sites such as the spleen. Indeed, splenic plasma cells appear to increase modestly in mice lacking eosinophils (53), though the lifespan of these cells remains an open question. In contrast, mice lacking megakaryocytes did not appear to possess increased numbers of splenic plasma cells, but it is also notable that plasma cell numbers were compromised early after immunizing megakaryocyte-deficient mice, although at later time points BM plasma cell numbers were relatively intact (54). At the very least, these issues and inconsistencies raise questions about which BM cells produce APRIL and IL-6 and whether all mature long-lived plasma cells must remain in close proximity to these cells to avoid death. Therefore, although plasma cells appear to colocalize preferentially with stromal cells that express CXCL12, such interactions could in principle occur in a variety of disparate sites including those lacking abundant numbers of APRIL-producing cells (see Fig. 1B).

How then should we move forward to resolve these issues? Let us place the issues described above within two different frameworks. For the first we will focus on the heterogeneity of BM plasma cell pools. To our understanding, a strict or extreme interpretation of the physical niche model predicts that entry into specialized microenvironments is a necessary and sufficient step for newly formed plasma cells to avoid apoptosis. Thus, it would be predicted that the vast majority of plasma cells located in these microenvironments would be long lived. Or conversely, the majority of short-lived plasma cells would be found elsewhere, perhaps scattered throughout the BM and other tissues (see Fig. 1A). The same reasoning can be applied to long-lived splenic plasma cell niches. Alternatively, entry into microenvironments harboring long-lived plasma cells is neither necessary nor sufficient to ensure longevity. In this scenario, newly formed and mature long-lived plasma cells might enjoy equal access to such sites. Given that expression of the CXCL12 receptor CXCR4 initiates early during B cell activation in peripheral lymphoid tissues (62), one might predict that the majority of immature BM plasma cells, including those destined to die within days of their generation, would be found intermixed with their long-lived counterparts (Fig. 1B). Although methods to selectively identify dying versus long-lived plasma cells in such studies would require development, clues to this end might lie in the observation that plasma cell survival is intimately tied to Mcl-1 activity (67).

A second framework for testing the physical niche model may stem from a better understanding of plasma cell homing and retention mechanisms. It is generally thought that plasma cells remain in the BM for lengthy periods without routinely recirculating into other tissues. Perhaps methods could be developed to force mature long-lived plasma cells to routinely recirculate throughout the blood and lymphatic systems, ideally in APRIL-null mice. To be sure, this idea is simple in concept but highly complex in execution. Nonetheless, clear results from such studies would either elevate attention toward fully elucidating the components of plasma cell niches, or inspire the development of alternative models for understanding why some plasma cells survive when so many others die.

We thank Dr. Mark Shlomchik and Dr. Michael Cancro for critically helpful discussions, and Dr. Shlomchik for providing uniquely useful mouse models.

The authors have no financial conflicts of interest.