CD138 and APRIL regulate plasma cell survival, competition, and retention in the bone marrow niche

SUMMARY

Durable serological protection is maintained through the persistence of antigen-specific plasma cells (PCs), but key factors regulating the survival of nascent PCs remain unclear. Previously, we reported that bone marrow (BM) PCs partially organize into clusters that are enriched for long-lived PCs, suggesting that clusters are survival niches. Here, we report that acute blockade of a proliferation-inducing ligand (APRIL) and B cell activating factor (BAFF) using transmembrane activator and CAML interactor (TACI)-Fc rapidly disrupts clusters and mobilizes BM PCs. CD138, a surface co-receptor that is abundant on PCs and binds APRIL but not BAFF, regulates PC retention in the BM and adhesion and motility on fibronectin. Cell-intrinsic CD138 levels control competition for survival between nascent CD138^low PCs and mature CD138^high PCs, and enhanced survival of CD138^high PCs correlates with retention in clusters. Collectively, these results indicate that PC clusters are survival niches and that dynamic competition between new and pre-existing PCs regulates the survival of new PCs and the durability of antibody responses.

Graphical Abstract

In brief

Survival of new PCs after vaccination is highly variable, inefficient, and poorly understood. Park et al. demonstrate that higher cell-intrinsic CD138 levels control PC survival in competitive environments, correlating with enhanced PC localization in clusters. These clusters are disrupted by APRIL inhibition, suggesting they are survival niches.

INTRODUCTION

Circulating serum antibodies are present in all vertebrates and provide immediate, potentially life-saving protection against pathogens.¹ Plasma cells (PCs) are terminally differentiated, non-proliferative antibody-secreting effector cells of the B lineage that are the main source of circulating antibodies.² Long-term survival of PCs is therefore critical for humoral immune memory. During an infection or immunization, B cell subsets in secondary lymphoid organs are activated by cognate antigen encounter and can differentiate into plasmablasts (PBs). PBs (CD138^highB220⁺) are short-lived precursors of PCs (CD138^highB220⁻), and together, PBs and PCs constitute antibody-secreting cells (ASCs).³ Nascent ASCs migrate to specialized niches in distant organs, most commonly the bone marrow (BM), where only a small fraction survive to become mature long-lived PCs (LLPCs), through processes that are not yet fully resolved.^2–6

The soluble tumor necrosis factor (TNF) superfamily cytokines APRIL (a proliferation-inducing ligand) and BAFF (B cell activating factor) play central roles in BM PC survival.⁷ APRIL and BAFF bind TACI (transmembrane activator and calcium-modulator and cyclophilin ligand [CAML] interactor) and BCMA (B cell maturation antigen) receptors, inducing nuclear factor κB (NF-κB) signaling and increasing levels of anti-apoptotic Bcl-2 family proteins.^8–12 APRIL and BAFF co-blockade depletes BM PCs within days to weeks after blockade initiation.^13–15 While BAFF inhibition alone is insufficient to reduce BM PC numbers,^14,16 loss of BCMA, the more abundant receptor on PCs, which binds APRIL with greater affinity than BAFF, is sufficient.^12,17 APRIL may therefore be the more critical factor regulating PC survival in the BM niche.

While numerous hematopoietic-derived BM cell types can produce APRIL, the organization and dynamics of APRIL-producing niches remain unknown.⁴ In principle, APRIL may be passively diffusing throughout the parenchyma or restricted in limited quantities to discrete physical niches¹⁸ for which lodging might be competitive.¹⁹ In support of a physical survival niche model,^19,20 we and others have found that 15%–40% of PCs organize into grape-like clusters throughout the BM.^21–23 Based on intravital imaging studies using two-photon laser scanning microscopy (2PLSM), we also found that BM PCs are motile but decelerate upon entering clusters, which dynamically form and break apart over a timescale of hours.²² Interestingly, PC clustering is reduced in the BM of APRIL-deficient mice, and LLPCs are enriched in PC clusters.^22,23 Taking these results together, we hypothesize that PC clusters nucleate around APRIL-producing cells that support PC survival and possibly maturation as well. Furthermore, cluster formation may be limited by APRIL availability, and access to clusters may therefore be competitive among PCs.³

In addition to BCMA and TACI, efficient APRIL-mediated pro-survival signaling depends on the transmembrane proteoglycan CD138 (syndecan-1), which is highly expressed on the surface of ASCs.²⁴ Through its heparan sulfate groups, CD138 binds and oligomerizes APRIL but not BAFF, enhancing BCMA and/or TACI signaling.^10,14,25,26 CD138 also binds interleukin (IL)-6 to augment IL-6 receptor signaling in a similar fashion, and by increasing sensitivity to both pro-survival cytokines, CD138 enhances the survival of vaccine-generated nascent ASCs in lymph nodes.²⁷ Proteins that bind CD138 heparan sulfate groups belong to various families and include extracellular matrix proteins, cell adhesion molecules, integrins, and additional cytokines, enabling CD138 to exert a multifaceted influence over cellular behavior.²⁸ Among cell types, CD138 surface levels are highest on mature PCs^24,27 and are also dynamic, rapidly dropping in the absence of serum factors through endocytic trafficking.^29,30 However, a role for CD138 in regulating BM PC homeostasis and survival has not been determined. Given the potentially limited nature of APRIL-rich niches, whether CD138 regulates ASC competition and the potential for ASCs to become LLPCs remains unknown.

Here, we found that high CD138 levels on BM PCs correlated with greater functionality. Furthermore, CD138 afforded survival advantages to BM PCs by regulating adhesion, motility, and positioning in PC clusters. Importantly, PC clusters were rapidly dispersed, and the survival of the remaining BM PCs was compromised upon interfering with APRIL and BAFF in vivo. These findings suggest that PC clusters represent survival niches and provide support for a competitive physical survival niche model in the regulation of BM PC homeostasis and serological memory duration.

RESULTS

CD138 expression supports BM ASC fitness and survival

Despite being a canonical ASC marker, CD138 surface staining spans over 1.5 fluorescent logs on freshly isolated bona fide mature BM PCs analyzed by flow cytometry (Figure 1A, left). To determine whether heterogeneous CD138 surface levels could be a rheostat of in situ PC physiology, BM PCs were divided based on CD138 levels into four classes (Figure 1A, right), whose function and phenotype were assessed. Compared to PCs with the lowest CD138 levels, those with the highest levels were enriched for the CD93⁺CD98⁺ subset by 1.45-fold (Figures 1B and 1C) and displayed 2-fold higher in vivo uptake of the fluorescent glucose analog 2-NBDG (2-(N-(7-Nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-deoxyglucose) (Figures 1D and 1E). These features are associated with metabolic activity important for antibody production.³¹ PCs with the highest CD138 levels also exhibited 2-fold higher surface levels of CXCR4 (Figures 1F and 1G), a critical chemoreceptor for BM PC retention and antibody titer maintenance.²³ These results suggested that CD138 surface levels could play active roles in BM PC function, retention, and survival.

Figure 1. CD138 confers functional and cell-intrinsic advantages to BM PCs.

(A) Gating of CD138^highB220⁻ PCs in BM samples, followed by breakdown of PCs into CD138-level percentiles.

(B) Representative flow plots of CD93 and CD98 expression by BM PCs with the lowest 10% and highest 10% of CD138 levels.

(C, E, and G) Frequency of PCs that are CD93⁺CD98⁺ (C), normalized 2-NBDG uptake (E), and CXCR4 levels (G) among BM PCs in indicated CD138-level percentiles.

(D and F) Representative histograms of 2-NBDG uptake (D) and CXCR4 levels (F) among BM PCs according to indicated CD138 level percentiles.

(H) Schematic for generating CD138 mixed BM chimeras.

(I) Frequencies of indicated cell types out of total live BM cells in Tom⁺ or Tom⁻ compartments within individual chimeras. Data from Tom⁺WT:WT and Tom⁺WT: KO chimeras are on the left and right of the dotted midline, respectively.

(J) Total live BM cells from chimeras were separated into Tom⁺ and Tom⁻ compartments, and Tom⁺:Tom⁻ count ratios were calculated for total live cells (‘‘total’’) and indicated cell types.

Each datapoint represents one mouse, and data from the same mouse are connected by lines (C, E, G, I, and J). Lines on floating bars represent the minimum, mean, and maximum (I). Data are pooled from two or more independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and ns, not significant by paired ANOVA (C, E, and G) or paired t test (I and J). (C) n = 23; (E) n = 14; (G) n = 6; (I and J) n = 8–10; n = 3 for LLPC data, which are from one experiment. See also Figure S1.

To assess the function of CD138 in BM ASCs, we bred the ASC reporter allele B lymphocyte-induced maturation protein 1 (Blimp1)-YFP into CD138^+/+ (wild type [WT]) and CD138^−/− (knockout [KO]) mice. The Blimp1-YFP reporter labels ASCs with YFP and has allowed us to identify ASCs equally well in WT and KO mice without relying on CD138 expression^27,32 (Figures S1A and S1B). In unperturbed naive KO mice, there were no marked impairments in ASC number, frequency, or phenotype compared to WT controls (Figures S1C–S1G), suggesting that ASCs can survive or mature in the complete absence of CD138. However, we suspected that CD138 may play a cell-intrinsic role in PC survival in competitive environments, based on its known functions and our previous work.²⁷

To study the cell-intrinsic function of CD138 in competitive settings, we generated Tom⁺WT:KO mixed BM chimeric mice by injecting a 1:1 ratio of hematopoietic stem cell (HSC)-containing BM from tdTomato-expressing CD138^+/+ Blimp1-YFP (Tom⁺WT) and KO (CD138^−/− Blimp1-YFP⁺) donor mice into lethally irradiated recipients (Figure 1H). Donor-derived CD138^+/+ ASCs were Tom⁺YFP⁺, whereas CD138^−/− ASCs were YFP⁺ (Figure S1H). As controls, Tom⁺WT:WT chimeras were generated, in which both Tom⁺YFP⁺ and YFP⁺ ASCs were CD138^+/+. After allowing 6 months for BM recovery, BM B cells (YFP⁻B220^high), PBs (YFP⁺B220⁺), PCs (YFP⁺B220⁻), and LLPCs (identified using previously established markers³³; Figure S1I) from Tom⁺ and non-tdTomato-expressing (Tom⁻) compartments were analyzed by flow cytometry. In Tom⁺WT:KO chimeras, WT PB and PC frequencies within the Tom⁺ compartment were significantly higher than KO PB and PC frequencies within the Tom⁻ compartment, but no such differences were observed for B cells (Figure 1I). Overall, WT ASCs outnumbered KO ASCs by approximately 5-fold. Tom⁺:Tom⁻ ratios were highest in PBs and remained elevated in PCs and LLPCs, although to lesser extents (Figure 1J). No differences in B cell, PB, PC, or LLPC frequencies or numbers were observed between Tom⁺ and Tom⁻ compartments in control chimeras.

Although variability in immunoglobulin (Ig) isotype skewing between WT and KO ASCs in Tom⁺WT:KO chimeras could contribute to differences in ASC composition and abundance,³ we did not observe significant differences in isotype makeup between WT and KO ASCs (Figure S1J). Within spleens (SPs), WT ASC frequencies and numbers only slightly surpassed those of KO ASCs in Tom⁺WT:KO chimeras, while Tom⁺ and Tom⁻ compartments did not substantially differ in control chimeras (Figures S1K and S1L). Taken together, CD138 contributes to PC enrichment and survival in the BM in a competitive, cell-intrinsic, and isotype-agnostic fashion.

Since CD138 plays important roles throughout ASC maturation, we also assessed whether CD138 regulates PC maintenance in the BM. We generated mice expressing tamoxifen (TAM)-inducible PC-specific Blimp1-controlled cre (BEC)²³ with cre-inducible YFP fluorescent reporters in both CD138-sufficient (BEC Rosa26^LSLYFP) and deficient (CD138^−/− BEC Rosa26^LSLYFP) backgrounds. In these mice, treatment with TAM induces irreversible labeling, i.e., timestamping, of ASCs with approximately 70% efficiency (Figures 2A and 2B). The survival and maturation of timestamped Tom⁺YFP⁺ ASCs into LLPCs can be tracked in relation to bulk, unlabeled Tom⁺ PCs that represent a heterogeneous population comprising immature and mature PCs arising after the TAM treatment period.³ Tracking of timestamped PCs in intact BEC Rosa26^LSLYFP and CD138^−/− BEC Rosa26^LSLYFP mice revealed that timestamped CD138^−/− PCs persisted and acquired an LLPC maturation phenotype, based on the upregulation of CD93 and CXCR4,²³ similarly to timestamped CD138^+/+ PCs (Figures 2C–2E). These results indicate that CD138-deficient ASCs can survive and mature normally, specifically under non-competitive conditions in which CD138 is completely absent in the host.

Figure 2. CD138 promotes cell-intrinsic maturation and survival of BM PCs in competitive settings.

(A) Experimental setup for measuring steady-state PC turnover rates in CD138^+/+ and CD138^−/− mice.

(B) Representative flow plots of timestamped and non-labeled (bulk) PC populations at days 5 and 90.

(C) Percentage of timestamped PCs out of total PCs on days 5 and 90 in CD138^+/+ and CD138^−/− mice. Data are presented as mean ± SEM.

(D and E) CD93⁺ percentage (D) and surface CXCR4 levels (E) of timestamped PCs. CD93⁺ percentage increases and fold increases of CXCR4 levels in CD138^+/+ PCs relative to CD138^−/− PCs are indicated.

(F) Experimental setup for generating timestamping CD138 mixed BM chimeras and assessing turnover of timestamped PCs.

(G) Representative flow plots of timestamped BM PCs from chimeric compartments on days 5 and 30.

(H) Chimerism–defined as CD138^+/+:CD138^−/− compartment cell count ratios–for total live BM cells and timestamped PCs.

(I and J) CD93⁺ percentage (I) and surface CXCR4 levels (J) of timestamped PCs.

(K) Absolute numbers of timestamped BM PCs, per leg.

(L) Percentage of timestamped BM PCs remaining on day 30 relative to mean day 5 values.

(M and N) Intracellular Mcl-1 (M) and Bcl-2 (N) levels of timestamped PCs.

Each datapoint represents one mouse, and data from the same mice are connected by lines (C–E and H–N). Data are pooled from two independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and ns, not significant by paired t test (D, E, and H–N) or unpaired t test (C). (C) n = 3–10; (D and E) n = 8; (H–N) n = 7.

Next, to assess the role of CD138 in PC maintenance under competitive conditions, we generated 1:1 mixed BM chimeras using CD138^+/+ and CD138^−/− timestamping donor mice. In these experiments, Blimp1-YFP BEC Rosa26^LSLTomato CD45.1/2 and CD138^−/− BEC Rosa26^LSLYFP CD45.2 mice were used to reliably track CD138^+/+ and CD138^−/− timestamped PC subsets without interference (Figures 2F and 2G). Timestamped CD138^+/+ and CD138^−/− PCs were analyzed at peak numbers 5 days after TAM treatment, and subsequent decay was quantified on day 30. Although total donor cell chimerism was approximately 1:1, PC timestamping efficiency was 10-fold higher within the CD138^+/+ subset on day 5, and differences between total and timestamped PC chimerism were exacerbated on day 30 (Figure 2H). This was in part due to the higher recombination efficiency of the Rosa26^LSLTomato locus compared to the Rosa26^LSLYFP locus (90% versus 70%, respectively)²³ but mostly due to the difference in CD138 expression between PC subsets. On day 5, timestamped CD138^−/− PCs exhibited compromised maturation based on CD93 and CXCR4 levels (Figures 2I and 2J). By day 30, timestamped CD138^−/− PCs had decayed more rapidly than CD138^+/+ counterparts (Figures 2K and 2L) and continued to express lower levels of CD93 and CXCR4 (Figures 2I and 2J). Levels of the anti-apoptotic protein Mcl-1 were also reduced in CD138^−/− PCs, while Bcl-2 levels were not impaired (Figures 2M and 2N). Therefore, CD138 is cell-intrinsically required for competitive PC survival in the BM.

Intrinsic CD138 levels control competition between new and pre-existing PCs

The formation of durable humoral memory after vaccination requires nascent ASCs, which express low but heterogeneous levels of CD138, to successfully access limited pro-survival factors in the BM. This may involve competition among waves of nascent CD138^low ASCs and/or competition with pre-existing mature CD138^high PCs encountered in the BM niche. We developed a vaccination model system to resolve these two possibilities, in which CD138 levels were manipulated on host and donor-derived ASCs (Figure 3A) based on the fact that CD138 surface levels of CD138^+/+ PCs were approximately 2-fold higher than those of CD138^+/− PCs (Figure 3B). Mice expressing the B1–8^hi nitrophenyl (NP)-specific Ig³⁴ and Blimp1-YFP alleles (BHY) were bred in a CD138^+/+ background with the CD45.2 congenic marker (WT BHY) or in a CD138^+/− background with the CD45.1/2 congenic markers (HET BHY). Naive B cells from these two mice were co-transferred into C57BL/6 recipients (B6 hosts). Hosts were immunized with alum-adjuvanted NP-KLH, and the fates of vaccine-generated antigen-specific WT BHY and HET BHY PC subsets were tracked in the BM (Figure 3C).

Figure 3. Higher CD138 levels confer a survival advantage to nascent vaccine-generated PCs.

(A) Schematic for adoptive transfer immunizations.

(B) Representative histogram of CD138 expression by endogenous PCs in B6 and B6 HET hosts.

(C) Gating for vaccine-generated BHY PCs and WT donor and HET donor subsets.

(D) CD138 levels of indicated PCs 2 weeks post-immunization. Mean levels of donor-derived PCs are summarized as percentages of mean host PC levels.

(E) Numbers of WT BHY PCs and HET BHY PCs in BM harvested from two legs of B6 and B6 HET hosts.

(F) Bcl-2 levels in WT BHY PCs and HET BHY PCs in B6 and B6 HET hosts.

Each datapoint represents one mouse, and data from the same mouse are connected by lines (D–F). Lines on floating bars represent the minimum, mean, and maximum (D and F). Data are pooled from two or more independent experiments. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, and ns, not significant by paired t test (D–F). (D) n = 7–10; (E, left) n = 13–23; (E, right) n = 15–19; (F) n = 9–11.

As expected, CD138 levels on nascent WT BHY PCs were 2-fold higher than on nascent HET BHY PCs, and both nascent PC subsets had lower CD138 levels than endogenous, mature PCs of B6 hosts (Figure 3D, left side). WT BHY PCs also outnumbered HET counterparts in B6 hosts at 2 and 8 weeks post-immunization (Figure 3E). No differences were seen between donor subsets at the germinal center stage (data not shown), consistent with what we previously reported in a similar model.²⁷ WT BHY PCs had higher levels of Bcl-2 than HET BHY PCs (Figure 3F). Therefore, WT BHY PCs possessed a survival advantage over HET BHY counterparts, indicating that higher CD138 levels enhance the survival of nascent PCs in a cell-intrinsic fashion.

Next, to determine whether CD138 levels mediated ASC competition among nascent donor-derived PCs or between nascent and endogenous pre-existing PCs, we repeated these experiments using CD138^+/− mice in C57BL/6 backgrounds, referred to as B6 HET hosts. In B6 HET hosts, CD138 levels on nascent WT BHY and HET BHY PCs were similar to those observed in B6 hosts (Figure 3D). However, differences in CD138 levels between endogenous mature PCs of B6 HET hosts and nascent WT BHY PCs or HET BHY PCs were abolished or reduced, respectively. At 2 and 8 weeks post-immunization, similar numbers of WT BHY and HET BHY PCs were observed in the BM of B6 HET hosts, suggesting that CD138-mediated competition among donor-derived PCs was lost (Figure 3E). Of note, however, intrinsic Bcl-2 levels remained higher in WT BHY PCs than in HET BHY PCs, as was observed in WT hosts (Figure 3F). Therefore, although donor-derived WT BHY PCs could express supraphysiological CD138 levels compared to endogenous PCs in B6 HET hosts, this somehow did not further enhance survival.

We conclude that CD138 expression regulates competition between nascent PCs and pre-existing mature PCs, although potential effects by other host-endogenous cells on nascent PC survival have not been ruled out. CD138 levels on host cells may thereby establish the minimal threshold that nascent, maturing CD138^low PCs need to compete and survive in the BM niche.

CD138 influences BM PC positioning and dynamics

BM PC maturation is associated with increased clustering and reduced motility.^23,35 Since CD138 promoted cell-intrinsic PC survival in the BM, we next examined whether CD138 expression influenced PC spatial organization and/or dynamics in the BM. Using 2PLSM, we visualized fluorescent ASC subsets within the BM of live, anesthetized Tom⁺WT:WT and Tom⁺WT:KO chimeras (described in Figure 1). Based on flow cytometry analyses in Figure 1I, BM PC:BM PB ratios were approximately 10:1 for Tom⁺ and Tom⁻ subsets in both sets of chimeras, indicating that the majority of fluorescent BM ASCs being imaged were PCs. To identify PC clusters, total YFP⁺ PCs, spanning both Tom⁺ and Tom⁻ subsets, were input into the DBSCAN (density-based spatial clustering of applications with noise) algorithm³⁶ (Figures 4A and 4B). DBSCAN-identified clusters were further annotated according to Tom⁺ and Tom⁻ PC subsets (Figure 4C). For each subset, relative abundance in clusters was assessed independently, e.g., the percentage of clustered PCs within Tom⁺ PCs only. Overall, the percentage of total YFP⁺ PCs in clusters was not significantly different between the two chimera groups (Figure 4D). While no differences in clustering were observed between Tom⁺WT and WT PCs of control chimeras, Tom⁺WT PCs were significantly enriched (1.5-fold) in clusters as compared to KO PCs in Tom⁺WT: KO chimeras (Figures 4C and 4E). These findings suggest that CD138 expression promotes access to or retention in PC clusters in a cell-intrinsic fashion.

Figure 4. CD138 controls entry of PCs into survival niches.

(A) Representative BM tile from a Tom⁺WT:KO chimera. Blimp1-YFP⁺ PCs and tdTomato⁺ BM cells are in green and red channels, respectively. Scale bar, 100 μm.

(B) DBSCAN-processed version of (A). Magenta spots represent non-clustered PCs, and non-magenta spots represent clustered PCs. Different colors indicate distinct clusters.

(C) Zoomed-in views of boxed regions in (B). Yellow spots represent Tom⁺WT PCs, and green spots represent KO PCs. Scale bar, 20 μm.

(D) Percentage of total YFP⁺ PCs in clusters in chimeras. Data are presented as mean ± SD.

(E) Percentage of each chimeric PC subset in clusters. Data from Tom⁺WT:WT and Tom⁺WT:KO chimeras are on the left and right of the dotted midline, respectively.

(F) Time-lapse images highlighting migration trajectories of Tom⁺WT and KO PCs in a Tom⁺WT:KO chimera. Scale bar, 50 μm. Time is shown as h:min:s.

(G) Tracks of all Tom⁺WT and KO PCs plotted at a common origin from the Tom⁺WT:KO chimera in (F).

(H) Track displacement velocities of individual PCs. Bars indicate means.

(I) Average displacement velocity of chimeric PC subsets on a per-mouse basis.

(J) Mean-squared displacement (MSD) rates of chimeric PC subsets, pooled from 4 chimeras each.

Lines represent mean values. Each datapoint represents one mouse (D, E, and I) or cell (H), and data from the same mouse are connected by lines (E and I). Data are pooled from two or more independent experiments. *p < 0.05, **p < 0.01, ****p < 0.0001, and ns, not significant by unpaired t test (D and H), paired t test (E and I), or ANCOVA comparing slopes of simple linear regression analyses (J). (D and E) n = 6–8; (H) n = 56–214; (I and J) n = 4.

We next assessed whether CD138 influences BM PC dynamics by performing time-lapse 2PLSM. In Tom⁺WT:KO chimeras, KO PCs were significantly more motile than Tom⁺WT PCs, as evidenced by elongated trajectories (Figures 4F and 4G) in addition to 3-fold higher displacement velocities and a 28-fold higher mean-squared displacement (MSD) rate (Figures 4H–4J). By contrast, both Tom⁺WT PCs and WT PCs in control Tom⁺WT: WT chimeras displayed similar low motility with comparable displacement velocities and MSD rates (Figures 4H–4J). Overall, CD138 regulates PC behavior by improving cluster localization and reducing motility.

CD138 controls PC retention in the BM and adhesion to fibronectin

Using 2PLSM, we previously reported that PC migration within the BM is mediated in part by the chemokine receptor CXCR4 and integrins LFA-1 and VLA-4.²² Specifically, VLA-4 mediates cell arrest and BM retention, while antibody-mediated blockade or chemical inhibition of VLA-4 components increases cell motility and mobilizes BM PCs into the blood. Based on these results and previous studies reporting that CD138 can indirectly activate VLA-4 through cis interactions,³⁷ we sought to determine whether CD138 participates in BM PC retention. WT (Blimp1-YFP) mice were treated with anti-CD138 antibody or isotype control, and peripheral blood was analyzed before and after treatments (Figure 5A). The blood of unperturbed mice is typically enriched for immature PBs with detectable B220 and major histocompatibility complex (MHC) class II levels (Figure S2A). 24 h after anti-CD138 treatment, the number and frequency of mature PCs in peripheral blood increased by 5- to 25-fold (Figures 5B and 5C), consistent with mobilization from the BM. There was no increase in circulating immature PBs, which are CD138^low. Interactions between CD138 and VLA-4 can be abrogated using a peptide inhibitor, synstatin, that consists of the CD138 VLA-4-binding site (SSTN_VLA-4).³⁸ Mice treated with SSTN_VLA-4 displayed an increase in mature PCs in the blood 24 h after treatment (Figures S2B and S2C).

Figure 5. CD138 promotes retention of BM PCs and adhesion and motility on fibronectin.

(A) Experimental design.

(B and C) PC and PB numbers (B) and frequencies (C) out of total live cells in 50 μL blood before and after treatments.

(D) Representative fluorescence and corresponding IRM images of Blimp1-YFP⁺ PBs. Adherent PBs and corresponding IRM signals are indicated by magentafilled arrowheads. The far-right PB is non-adherent, based on the lack of a corresponding IRM signal in the space indicated by the white-filled arrowhead. Scale bar, 20 μm.

(E) Individual IRM cell tracks plotted from a common origin of WT, KO, and αCD138-treated WT PBs on fibronectin- or VCAM-1-coated glass from one representative time-lapse movie each.

(F and G) Percentage of adhered cells (F) and displacement velocities (G) for WT, KO, and αCD138-treated WT PBs imaged on fibronectin- or VCAM-1-coated glass.

Each datapoint represents one mouse (B and C), movie (F), or cell (G), and data from the same mouse are connected by lines (B and C). Data are presented as mean ± SEM (F and G). Data are pooled from two or more independent experiments. *p < 0.05, **p < 0.01, ****p < 0.0001, and ns, not significant by paired t test (B and C) or unpaired t test (F and G). (B and C) n = 5; (F) n = 3–5; (G) n = 30–132. See also Figure S2.

To better understand how VLA-4 regulates BM PC motility and retention, we performed interference reflection microscopy (IRM)-based live-cell in vitro imaging to visualize ASC adhesion to VLA-4-interacting BM substrates such as VCAM-1 and fibronectin.²² We had found that ASCs adhere strongly to VCAM-1-coated substrates, upon which they remain arrested, whereas on fibronectin, ASCs are highly motile. Importantly, adhesion to these two substrates was abrogated upon chemical inhibition of VLA-4, although adhesion to ICAM-1, which binds LFA-1, but not VLA-4, was unaffected.

We subsequently examined whether CD138 regulates VLA-4-dependent ASC adhesion and migration by performing IRM of Blimp1-YFP⁺ ASCs (Figure 5D) on fibronectin-, VCAM-1-, or ICAM-1-coated substrates. Whereas WT ASCs adhered to and migrated on fibronectin and VCAM-1 in a manner consistent with our previous studies, KO ASCs largely failed to adhere to and migrate on fibronectin while behaving similarly on VCAM-1 to WT counterparts (Figures 5E–5G). Similarly, pre-treatment with anti-CD138 antibody prevented WT ASCs from adhering to and migrating on fibronectin but did not affect their behavior on VCAM-1. Furthermore, CD138 did not play a role in LFA-1-mediated interactions with ICAM-1 (Figures S2D–S2F). Collectively, the data indicate that CD138 on ASCs regulates BM retention and VLA-4-mediated adhesion and motility on fibronectin-coated substrates.

BM PC survival niches are dynamic and co-dependent on APRIL and BAFF

The results so far indicate that CD138 expression on BM PCs is associated with superior BM maintenance and localization in clusters that may represent physical APRIL-producing survival niches. CD138 may thereby regulate APRIL-dependent survival of BM PCs by enhancing retention within and sensitivity to APRIL-producing niches. To determine whether APRIL signaling plays a requisite role in BM PC maintenance and clustering, we sought to directly interfere with APRIL through TACI-Fc treatment. Soluble TACI-Fc can compete with cells for binding of free APRIL and BAFF, lowering the availability of these cytokines in vivo. BM PCs in live, anesthetized WT (Blimp1-YFP) mice were hence imaged by 2PLSM before and after treatment with TACI-Fc (Figure S3A). As controls, these experiments were also performed with matching isotype (human IgG1) or BAFF receptor (BAFFR)-Fc, which only scavenges BAFF.

Approximately 4.5 h after TACI-Fc treatment, total PC numbers in the BM were reduced by 50%, and PC clusters were reduced by 80% (Figures 6A–6C). Control IgG1 treatment did not alter PC numbers, and BAFFR-Fc treatment led to a smaller 20% reduction in total PC counts but no change in PC clustering. Immediately after TACI-Fc/IgG1/BAFFR-Fc treatment, PC dynamics and clustering were also tracked in real time by time-lapse 2PLSM (Figure 6D; Videos S1 and S2). Following TACI-Fc treatment, a progressive, linear decrease in total PC numbers was observed, resulting in 50% depletion by 4 h post-treatment (Figure 6E). This was accompanied by a 78% loss of PC clustering, with an average cluster half-life of 122 min (Figure 6F). Similar effects were not observed following control treatments or prior to TACI-Fc/IgG1/BAFFR-Fc treatment (Figures S3B and S3C), indicating that reductions in BM PC numbers and clustering were specific to APRIL and BAFF co-blockade. Comparing decay rates of PCs that were either initially clustered or non-clustered following TACI-Fc treatment revealed that non-clustered PCs were lost more rapidly than clustered PCs (Figure S3D). During time-lapse imaging, loss of PCs was not observed in conjunction with migration beyond imaging borders or apoptosis, suggesting that PCs were exiting the BM by egressing into nearby sinusoids.

Figure 6. APRIL and BAFF maintain PCs in their survival niche.

(A) Representative BM tiles from before and after TACI-Fc treatment. Left: Blimp1-YFP⁺ PCs and autofluorescence are in the green and red channels, respectively. Right: DBSCAN-processed tiles. Scale bars, 100 μm.

(B and C) PC counts per mm³ of BM (B) and percentage of PCs in clusters (C) in tiles before and after treatments.

(D) Time-lapse images of PC clusters, encircled in yellow, after treatments. Scale bars, 25 μm. Time is shown as h:min:s.

(E and F) Left, number of total PCs (E) or PCs in clusters (F) over time, as a percentage of initial, in time-lapse images after treatments. Data are pooled from 3 mice per condition and presented as mean ± SEM. Right, percentage of initial total PCs (E) or clustered PCs (F) left 4 h after treatments.

(G) PC and PB counts in BM (2 legs) and SP harvested from treated or untreated mice.

(H) Blimp1-YFP⁺ ASC (PCs and PBs) frequency out of total live cells in 50 μL blood before and 24 h after treatments.

(I) IκBα levels in BM PCs 6 h after treatments.

(J) Bcl-2 and Mcl-1 levels in BM PCs 24 h after treatments.

Each datapoint represents one mouse (B, C, E, and F, right sides, and G–J), and data from the same mouse are connected by lines (B, C, and H). Data are presented as mean ± SD (E and F, right sides, G, I, and J). Data are pooled from two or more independent experiments (B–C and E–G) or show one representative experiment (H–J). *p < 0.05, **p < 0.01, ***p < 0.001, and ns, not significant by paired t test (B and C), ratio paired t test (H), or unpaired t test (E and F, right sides, G, I, and J). (B and C) n = 3–4; (E and F) n = 3; (G) n = 4–6; (H) n = 5–6; (I and J) n = 3–5. See also Figure S3.

To determine whether BM PCs were indeed mobilized, we enumerated cells in the BM and SP after TACI-Fc, IgG1, and BAFFR-Fc treatments by flow cytometry. By 6 h after TACI-Fc treatment, BM PC numbers and frequencies fell by about 45% compared to control (untreated) mice and slightly further at 24 h (Figures 6G and S3E), when increased ASC frequencies in peripheral blood were observed (Figures 6H and S3F). Similar effects were seen for BM PBs, which were mobilized more effectively (Figures 6G, S3E, and S3F). Overall, the effects of TACI-Fc did not appear to be isotype specific (Figure S3G), although IgA⁺ LLPCs trended toward enrichment 24 h after TACI-Fc treatment (Figure S3H). SP PC counts and frequencies were unaltered by TACI-Fc treatment, but PB counts and frequencies increased by 2-fold at 6 h before returning to baseline levels at 24 h (Figures 6G and S3E). IgG1 and BAFFR-Fc treatment did not affect PC or PB counts and frequencies in BM and SP. Total BM and SP cellularities were unaltered by TACI-Fc treatment as well as IgG1 and BAFFR-Fc treatments (Figure S3I), indicating that TACI-Fc treatment specifically affects ASCs.

As critical survival factors for ASCs, APRIL and BAFF are inducers of NF-κB signaling and anti-apoptotic factors like Bcl-2 and Mcl-1. Incubating BM PCs with APRIL in vitro led to a reduction in IκBα levels, signifying induction of canonical NF-κB signaling (Figures S3J and S3K). Inversely, IκBα levels were 1.3-fold higher in BM PCs after TACI-Fc treatment compared to IgG1 and BAFFR-Fc treatments (Figure 6I). Moreover, TACI-Fc treatment reduced Bcl-2 and Mcl-1 levels in BM PCs by 30% compared to IgG1-treated controls (Figure 6J), although apoptosis was not yet observed (Figures S3L–S3M). Taken together, the results show that acute APRIL and BAFF co-blockade directly impedes PC retention within BM niches, leading to mobilization and impaired survival.

DISCUSSION

The formation of durable serological memory requires nascent ASCs to survive and fully differentiate into LLPCs in the BM. These processes are dependent on the acquisition of survival factors that are either freely diffusing or physically limited in niches.¹⁸ Our previous work established that PC clusters are APRIL dependent and enriched for LLPCs.^22,23 Here, our finding that acute co-blockade of APRIL and BAFF with TACI-Fc dispersed PC clusters further supports the notion that APRIL may be produced within PC clusters that represent putative BM PC survival niches rather than freely diffusing. While we cannot rule out potential involvement by BAFF, APRIL is more likely to be the primary factor supporting BM PC survival since BAFFR-Fc treatment failed to recapitulate the effects of TACI-Fc treatment. Our findings are therefore consistent with previous studies showing that BM PCs are directly associated with APRIL-producing cells such as eosinophils and megakaryocytes.^21,39,40 Within clusters, LLPCs may have enhanced ‘‘VIP’’ access to APRIL-producing cells through unknown mechanisms, which may allow them to survive indefinitely.³ However, LLPC privileges may come at the expense of newly minted PCs seeking the same survival factors.

Mobilization of BM ASCs after TACI-Fc treatment also indicates that APRIL may control the retention of these cells in the BM. In the absence of survival signals, ASCs may recirculate to other niches, such as the SP, in search of available signals. Immature PCs and PBs are more commonly found in the blood than mature PCs. This may reflect the impaired ability of immature ASCs to home to limited survival niches, ultimately leading to apoptosis of these cells from neglect. Expanding BM PC survival niches may improve universal vaccine durability, as a recent study demonstrated that megakaryocyte activation through thrombopoietin (TPO) administration enhances the longevity of specific antibody titers elicited by a variety of vaccines.⁴¹ Future work may better define BM PC survival niche composition, function, and capacity for accommodating nascent ASCs, as well as potential mechanisms by which the niche promotes the maturation and survival of LLPCs.

Given our results here, we hypothesize that CD138 modulates ASC competition in the BM in a top-down manner,³ such that pre-existing CD138^high PCs limit the survival of incoming waves of nascent ASCs that are initially CD138^low prior to full maturation. CD138^high PCs display superior localization in clusters, enhanced survival, and greater metabolic activity, which is suggestive of clustered CD138^high PCs producing more antibodies. However, the influence of CD138 on the BM ASC population as a whole was complex and dependent on competitive settings. In particular, CD138-mediated ASC competition was most evident at the PB stage, suggesting that CD138 may function at an early checkpoint regulating PB-to-PC maturation. One potential underlying mechanism might be through CD138-controlled access to APRIL-rich niches, which could limit the number of nascent ASCs that can join the CD138^high LLPC pool. Taking into consideration the initially surprising result that supraphysiological CD138 levels on nascent donor-derived ASCs relative to endogenous host PCs did not further increase antigen-specific PC output, CD138 levels on PCs may establish the threshold specifically for the entry of nascent ASCs into clusters. This is in line with previous work where we did not detect higher CD138 levels on timestamped LLPCs compared to bulk PCs.²³ Furthermore, while CD138 was required for fitness during competition, some CD138^−/− PCs could persist, suggesting a role for CD138-independent compensatory survival mechanisms, such as BAFF-, CXCL12-, and CD28-mediated signaling.

Within clusters, CD138 may work in concert with VLA-4 (α4/β1),⁴² other integrins such as β3, β5, and αv,⁴³ and other adhesive or chemotactic surface proteins that are upregulated on LLPCs and may function in isotype-specific niches.^3,23,33 Still, interfering with CD138 was sufficient to mobilize ASCs into the blood. This may have resulted from compromised APRIL signaling, as with TACI-Fc treatment, or may be due to disruption of VLA-4-mediated adhesion to fibronectin. While in vitro, fibronectin mediated ASC migration, the impact of fibronectin and other ligands such as VCAM-1 and ICAM-1 on BM ASC dynamics and positioning in situ is likely to be more complex due to variations in ligand density and organization. Within rodent BM, fibronectin organizes into septa that compartmentalize the marrow into discrete areas enclosing hematopoietic cell clusters,⁴⁴ which could regulate directed cell migration, mediate PC arrest within APRIL-rich niches, and/or provide additional pro-survival signals.⁴⁵ Our model explains why newly minted PCs fail to survive in the BM but also predicts that bolstering intrinsic CD138 levels on vaccine-generated ASCs should alleviate CD138-dependent maturation and survival bottlenecks, leading to an enhanced antibody response durability following vaccination.

Limitations of the study

Various experiments in this study relied on using cells from CD138 KO mice to infer the role of CD138 in PC survival, but the findings could be confounded by the influence of CD138 on early ASC maturation. The generation of conditional CD138 alleles that could be deleted at the PC stage would be necessary to fully clarify the role of CD138 in PC survival. Additionally, while we have strong reason to believe that the effects of TACI-Fc treatment on BM PCs are mediated directly and primarily through reductions in available APRIL, TACI-Fc can also bind BAFF, which leaves open the possibility that BAFF may also support BM PC organization and maintenance. Moreover, while TACI-Fc treatment rapidly disrupted BM PC clusters, direct visualization of APRIL and BAFF in situ is needed to test whether survival factors, particularly APRIL, are enriched within clusters to directly maintain these survival niches.

RESOURCE AVAILABILITY

Lead contact

Inquiries for further information or requests for resources and reagents should be directed to and will be fulfilled by the lead contact, David R. Fooksman (david.fooksman@einsteinmed.edu).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• This study did not generate data that fall under Cell Press’s category of standardized data types, and therefore, the data were not deposited into a public repository. However, data are available from the corresponding author upon request.

• The code supporting the current study has not been deposited in a public repository because the scripts require Imaris and MATLAB software to run, but it is available from the corresponding author upon request.

• Inquiries or requests for all other items should be directed to and will be fulfilled by the corresponding author.

STAR★METHODS

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

Blimp1-YFP³² and B1–8^hi mice,³⁴ which are purchasable through Jackson laboratories, were generated previously in C57BL/6 backgrounds and bred in-house. CD138 KO (Sdc-1^−/−) mice were generated previously in a C57BL/6 background⁴⁶ and bred in-house. Mice with constitutive tdTomato expression in all cells were generated by crossing B6.Cg-Gt(Rosa)^{26Sortm14(CAG-LSL-tdTomato)/Hze}/J (Ai14 i.e., Rosa26^LSLTomato, #007914, Jackson laboratories) to B6.C-Tg(CMV-cre)1Cgn/J (CMV-Cre, #006054, Jackson laboratories). C57BL/6J mice and B6.129X1-Gt(ROSA)26Sor^tm1(EYFP)Cos/J (R26R-EYFP i.e., Rosa26^LSLYFP, #006148) mice were purchased from Jackson laboratories. B6-Ly5.1/Cr (CD45.1, #564) mice generated in C57BL/6 backgrounds were purchased from Charles River. BEC mice were generated previously in C57BL/6 backgrounds using CRISPR-Cas9 technology to knock-in a Cre^ERT2-IRES-tdTomato cassette downstream of exon 6 in the Prdm1 locus²³ and bred in-house. All mice were housed in groups of two to five animals per cage under specific pathogen free conditions at the Albert Einstein College of Medicine animal facility. Experiments were conducted as sex-unbiased, with sex- and age-matched adult mice that were at least 8 weeks old used for all experiments and same-sex littermates randomly assigned to experimental groups. The animal protocol in this study was approved by the Albert Einstein College of Medicine Institutional Animal Care and Use Committee (IACUC), protocol #0000–1021.

Cell culture

For PB cultures, naïve B cells from the spleens of Blimp1-YFP reporter mice (both females and males were used) were purified using CD43 microbeads according to the manufacturer’s instructions and cultured in L-glutamine-free RPMI 1640 media supplemented with 1% penicillin-streptomycin-glutamine, 10% FBS, 50 μM 2-mercaptoethanol, and 25 μg/mL LPS for 3 days at 37°C and 5% CO₂.

For primary BM cultures, BM was harvested from femurs and tibias of Blimp1-YFP reporter mice (both females and males were used) and red blood cells were lysed as described above. BM cells were incubated in L-glutamine-free RPMI 1640 media supplemented with 1% penicillin-streptomycin-glutamine, 10% FBS, and 50 μM 2-mercaptoethanol at 37°C and 5% CO₂. Multimeric APRIL was added at a concentration of 100 μg/mL.

METHOD DETAILS

Generation of mixed BM chimeras

5–7-week-old C57BL/6J or CD45.1 hosts were lethally irradiated with 950 rad. BM was harvested from the femur and tibia of donors, resuspended in PBS, and mixed 50:50. ≥ 2 × 10⁶ total cells of these mixtures were injected intravenously (i.v.) into irradiated hosts. Chimeras were maintained on water supplemented with sulfamethoxazole 200 mg-trimethoprim 40 mg at a 1:50 vol:vol ratio for the first 2 weeks and at least 8 weeks were allowed for BM recovery prior to experimentation.

Adoptive cell co-transfer and immunization

Spleens were harvested from donor mice and mashed through a 70 μm filter placed in PBS with 0.5% BSA and 2mM EDTA to obtain a single-cell suspension. Following red blood cell lysis, näıve B cells were isolated by magnetic-activated cell sorting (MACS) using CD43 MicroBeads according to the manufacturer’s instructions. Naı¨ve donor B cells were mixed 50:50 in PBS and 2 × 10⁶ total cells were injected i.v. into recipient mice. One day after transfer, recipient mice were immunized intraperitoneally (i.p.) with 50 μg of nitrophenyl-conjugated keyhole limpet hemocyanin (NP₃₂-KLH) emulsified in alum at a 2:1 vol:vol ratio in 150 μL volume.

Mouse treatments and injections

For 2-NBDG experiments, 50 μg of 2-NBDG dissolved in PBS was injected i.v. 15 min before sacrifice. For PC timestamping experiments, 4 mg of tamoxifen (TAM) dissolved in corn oil at a concentration of 20 mg/mL were administered by oral gavage for 3 consecutive days. For anti-CD138 treatments, 100 μg of anti-CD138 antibody (clone 281–2, BioLegend) or matching isotype control (rat IgG2a, clone RTK2758, BioLegend) were injected i.v. For SSTN_VLA-4 treatments, 620 ng of SSTN_VLA-4 reconstituted in 200 μL of PBS was injected i.v. and an equal volume of PBS was injected as the vehicle control. For human IgG1, TACI-Fc, and BAFFR-Fc treatments, 100 μg were injected i.v. in 100 μL volume. For intravital imaging, 5 μL of Qtracker 705 Vascular Labels were co-injected to confirm successful i.v. entry.

Flow cytometry

For BM, femurs and tibias were crushed in FACS buffer (PBS with 0.5% BSA and 1 mM EDTA) and the resulting mixture was passed through a 70 μm filter to obtain a single-cell suspension. Spleens were directly mashed through a 70 μm filter placed in FACS buffer to obtain a single-cell suspension. AccuCount fluorescent counting beads were added to cell suspensions according to the manufacturer’s instructions to obtain absolute cell counts. Cell suspensions were pelleted by centrifugation at 380 × g for 5 min. Red blood cells were removed by resuspending the pellet in ACK lysing buffer and incubating on ice for 3 min, followed by a wash with FACS buffer. UltraComp ebeads were used to create single-color controls for compensation and unmixing according to the manufacturer’s instructions. For single-color controls of yellow fluorescent protein and tdTomato, Blimp1-YFP mice and mice with constitutive tdTomato expression in all cells (generation described in the ‘‘mice‘‘ section) were bled.

For staining of cell surface proteins, cells were resuspended in FACS buffer containing anti-CD16/32 antibodies for Fc receptor blockade and fluorochrome-conjugated antibodies, followed by incubation for 30 min at 4°C. BD Horizon Brilliant Stain Buffer was added according to the manufacturer’s instructions when two or more antibodies conjugated to BD Horizon Brilliant fluorescent polymer dyes were used together. Cells were then washed with FACS buffer and analyzed directly or further processed for intracellular staining. For intracellular staining, cells were fixed and permeabilized with the BD Cytofix/Cytoperm Fixation/Permeabilization kit according to the manufacturer’s instructions. Cells were then resuspended in BD Perm/Wash buffer containing fluorochrome-conjugated antibodies and stained for 1 h at room temperature, followed by washes with BD Perm/Wash buffer and resuspension in FACS buffer prior to flow cytometry analysis. For intracellular IκBα staining, cells were immediately fixed with 4% formaldehyde, followed by surface staining, permeabilization with 0.3% Triton, and intracellular staining for 1 h at room temperature.

Samples were acquired on a BD FACSAria II (BD Biosciences) or a Cytek Aurora (Cytek Biosciences) and analyzed using FlowJo software v10 (FlowJo, LLC).

Multiphoton intravital imaging

Mice were surgically prepared for intra-tibial BM imaging. Briefly, mice were anesthetized with isoflurane gas and supinely secured to an imaging plate heated to 37°C. An incision was made in the leg and soft tissue superficial to the tibia was removed. A custom-built apparatus was used to immobilize the leg and expose the tibia, which was thinned with a microdrill to a thickness of under 200 μm to allow penetration and visualization of the underlying BM by 2-photon laser scanning microscopy (2PLSM). Prior to imaging, vacuum grease was applied to the apparatus on the area surrounding the exposed bone to create a water-tight immersion well that was filled with warm Lactated Ringer’s solution for the microscope water immersion objective.

Mice continued to be anesthetized with isoflurane during imaging. Imaging was conducted using an Olympus FVE-1200 upright microscope, 25 × 1.04 NA objective, and DeepSee MaiTai Ti:Sapphire pulsed laser (Spectra-Physics) tuned to 920 nm. To maintain mouse body temperature and minimize room light infiltration, the microscope was fitted with a custom-built incubator chamber that was maintained at a temperature of 35°C–37°C throughout imaging. Large-field BM images, i.e., tiles, were obtained by sequentially acquiring z stack images (5 μm step size) spanning the field of interest followed by automated stitching of individual images into one continuous image using Olympus software. For time-lapse movies, 90–105 μm deep Z-stacks (3 μm step size) with 1× zoom and 512 × 512 μm XY dimensions were acquired every 1.5 or 3 min for 4–5 h.

Intravital imaging analysis

All image analysis was performed using Imaris software 9.3.1 (Bitplane). Blimp1-YFP⁺ plasma cells were automatedly detected based on size (10 μm radius) and intensity and three-dimensionally tracked over time in time-lapse movies. Ratio and/or background-subtracted channels were generated to facilitate detection. For CD138 chimeras, background signal from the infared channel (Ch4) was subtracted from the YFP (Ch2) and tdTomato (Ch3) channels. A ratioed channel (background subtracted Ch3 divided by background subtracted Ch2) was used to distinguish Tom⁺ Blimp1-YFP⁺ plasma cells from Tom⁻ Blimp1-YFP⁺ plasma cells. A threshold of 1.1 in track intensity mean of the ratioed channel was applied such that Tom⁺ Blimp1-YFP⁺ plasma cells would exhibit values above the threshold while Tom⁻ Blimp1-YFP⁺ plasma cells would exhibit values below the threshold. To correct for drift, sessile autofluorescent macrophages detected in Ch4 were three-dimensionally tracked over time and XYZ registrations were automatedly corrected. For analyses of tiled and time-lapse images from Blimp1-YFP mice, Ch3 (empty) was used for background subtraction and drift correction.

Plasma cell clusters were identified in tiles and time-lapse movies with a custom Imaris XTension that used the DBSCAN (density-based spatial clustering of applications with noise) algorithm. DBSCAN defines clusters as groups of points that contain at least MinPts within Epsilon distance of a core point.³⁶ For detection of plasma cell clusters, MinPts of 5 and Epsilon of 30–40 μm were used. The volume of BM in images was calculated using the Imaris XTension SpotsVolume, which calculated the total volume occupied by detected cells in a three-dimensional image. In time-lapse movies, the displacement velocity of a given plasma cell track was calculated by dividing the distance between the first and last track positions by the total duration of the track in minutes. For each timepoint of a time-lapse movie, the mean square displacement (MSD) was computed as the mean of the squared distance between the initial and current positions for all present cell tracks. Analyses comparing MSD’s between groups were conducted on the linear portions of the MSD functions for accuracy.

Interference reflection microscopy and analysis

Interference reflection microscopy (IRM) imaging experiments used Blimp1-YFP⁺ PBs generated in vitro (described in the ‘‘cell culture‘‘ section) that were seeded onto Nunc Lab-Tek chambered coverglasses. Coverglasses were coated with 50 μg/mL ICAM-1 Fc, 50 μg/mL VCAM-1 Fc, or 300 μg/mL fibronectin by addition of a 5 μL drop to the center of the slide followed by incubation for 2 h at 37°C. The glass surface was washed with PBS and blocked with 5% BSA for 1 h at room temperature and then washed sequentially with PBS and imaging buffer (1× HBSS with 0.5 mM MgCl₂, 1 mM CaCl₂, 2% BSA, and 10 mM HEPES; pH 7.4). Prior to imaging, PBs were re-suspended in imaging buffer and added to chambers. For anti-CD138 treatments, anti-CD138 antibody (clone 281–2, BioLegend) was added at a concentration of 100 μg/mL just prior to imaging.

Time-lapse imaging was conducted on an inverted Leica SP5 confocal microscope (Einstein Analytical Imaging Facility) with the heated chamber set to 37°C. A 40× oil immersion objective was used and images were acquired every 15 s for 20 min. The dichroic mode was used to detect reflected light. IRM time-lapse movies were analyzed on Imaris. YFP⁺ PBs were automatedly detected as described in the ‘‘intravital imaging analysis‘‘ section and YFP⁺ PBs with IRM channel intensity values below background levels resulting from destructive interference of light reflected from adhered cell membranes and the glass surface were identified as adhered cells and selected for tracking.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analyses were conducted on GraphPad Prism (version 10). Two-tailed unpaired T-tests were used to compare two groups while paired T-tests or paired-ANOVA were used to compare two or more cell populations in the same mouse, respectively. All statistical parameters as well as tests and corresponding p-values are specified in figure legends. p-values less than or equal to 0.05 were considered statistically significant. Data are presented as minimum, mean, and maximum; mean ± SD; or mean ± SEM, as stated in figure legends.

Supplementary Material

Supplemental information can be found online at https://doi.org/10.1016/j.celrep.2025.116123.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Rabbit monoclonal anti-active caspase-3 BUV395 (clone C92–605.rMab)	BD Biosciences	Cat# 570180; RRID:AB_3685569
Rat monoclonal anti-B220 APC-Cy7 (clone RA3–6B2)	BioLegend	Cat# 103224; RRID:AB_313007
Mouse monoclonal anti-Bcl-2 AF647 (clone BCL/10C4)	BioLegend	Cat# 633510; RRID:AB_2274702
Mouse monoclonal anti-Bcl-2 PE-Cy7 (clone BCL/10C4)	BioLegend	Cat# 633512; RRID:AB_2565247
Rat monoclonal anti-CD3 PE/Dazzle™ (clone 17A2)	BioLegend	Cat# 100246; RRID:AB_2565883
Rat monoclonal anti-CD4 BV510 (clone GK1.5)	BioLegend	Cat# 100449; RRID:AB_2564587
Rat monoclonal anti-CD8alpha BV510 (clone 53–6.7)	BioLegend	Cat# 100752; RRID:AB_2563057
Rat monoclonal anti-CD16/CD32 (clone 2.4G2)	Bio X Cell	Cat# BE0307; RRID:AB_2736987
Rat monoclonal anti-CD38 BV510 (clone 90)	BD Biosciences	Cat# 740129; RRID:AB_2739886
Mouse monoclonal anti-CD45.1 AF700 (clone A20)	BD Biosciences	Cat# 561235; RRID:AB_10611577
Mouse monoclonal anti-CD45.1 PE (clone A20)	Thermo Fisher Scientific	Cat# 12–0453-83; RRID:AB_465676
Mouse monoclonal anti-CD45.1 PerCP-Cy5.5 (clone A20)	Thermo Fisher Scientific	Cat# 45–0453-82; RRID:AB_1107003
Mouse monoclonal anti-CD45.2 BV650 (clone 104)	BioLegend	Cat# 109836; RRID:AB_2563065
Mouse monoclonal anti-CD45.2 PB (clone 104)	BioLegend	Cat# 109820; RRID:AB_492872
Mouse monoclonal anti-CD45.2 PerCP-Cy5.5 (clone 104)	BioLegend	Cat# 109828; RRID:AB_893350
Rat monoclonal anti-CD93 BUV615 (clone AA4.1)	BD Biosciences	Cat# 751080; RRID:AB_2875116
Rat monoclonal anti-CD93 PE-Cy7 (clone AA4.1)	BioLegend	Cat# 136506; RRID:AB_2044012
Hamster monoclonal anti-CD95 PE-Cy7 (clone JO2)	BD Biosciences	Cat# 557653; RRID:AB_396768
Rat monoclonal anti-CD138 (clone 281–2)	BD Biosciences	Cat# 553712; RRID:AB_394998
Rat monoclonal anti-CD138 APC (clone 281–2)	BioLegend	Cat# 142506; RRID:AB_10962911
Rat monoclonal anti-CD138 BV785 (clone 281–2)	BioLegend	Cat# 142534; RRID:AB_2814047
Rat monoclonal anti-CD138 PE (clone 281–2)	BioLegend	Cat# 142504; RRID:AB_10916119
Rat monoclonal anti-CD138 PerCP-Cy5.5 (clone 281–2)	BioLegend	Cat# 142510; RRID:AB_2561601
Armenian hamster monoclonal anti-CD183 PE-Cy7 (clone CXCR3–173)	Thermo Fisher Scientific	Cat# 25–1831-82; RRID:AB_2573416
Rat monoclonal anti-CD184 BUV805 (clone 2B11/CXCR4)	BD Biosciences	Cat# 741979; RRID:AB_2871282
Rat monoclonal anti-CD184 BV421 (clone 2B11/CXCR4)	BD Biosciences	Cat# 562738; RRID:AB_2737757
Rat monoclonal anti-CD267 BV421 (clone 8F10)	BD Biosciences	Cat# 742840; RRID:AB_2741091
Rat monoclonal anti-CD326 BV711 (clone G8.8)	BioLegend	Cat# 118233; RRID:AB_2632775
Rat monoclonal anti-GL7 AF647 (clone GL7)	BioLegend	Cat# 144606; RRID:AB_2562185
Rat monoclonal anti-I-A/I-E AF700 (clone M5/114.15.2)	BioLegend	Cat# 107622; RRID:AB_493727
Rat monoclonal anti-I-A/I-E PE (clone M5/114.15.2)	BioLegend	Cat# 107608; RRID:AB_313323
Rat monoclonal anti-IgA APC (clone mA-6E1)	Thermo Fisher Scientific	Cat# 17–4204-80; RRID:AB_2848293
Rat monoclonal anti-IgD FITC (clone 11–26c.2a)	BD Biosciences	Cat# 553439; RRID:AB_394859
Rat monoclonal IgG2a kappa isotype control (clone RTK2758)	BioLegend	Cat# 400544; RRID:AB_11147167
Goat polyclonal anti-IgG AF647	Jackson ImmunoResearch Labs	Cat# 115–605-164; RRID:AB_2338913
Human IgG1 isotype control	Bio X Cell	Cat# BE0297; RRID:AB_2687817
Donkey polyclonal anti-IgM DyLight 405	Jackson ImmunoResearch Labs	Cat# 715–475-140; RRID:AB_2340838
Mouse monoclonal anti-IκBα APC (clone L35A5)	Cell Signaling Technology	Cat# 46001; RRID:AB_3697353
Rat monoclonal anti-Ly-6A/E AF700 (clone D7)	BioLegend	Cat# 108142; RRID:AB_2565959)
Rabbit monoclonal anti-Mcl-1 AF647 (clone D2W9E)	Cell Signaling Technology	Cat# 78471; RRID:AB_2799914
Rabbit monoclonal anti-Mcl-1 PB (clone D2W9E)	Cell Signaling Technology	Cat# 85415; RRID: AB_3697355
Mouse monoclonal anti-TIGIT BV421 (clone 1G9)	BioLegend	Cat# 142111; RRID:AB_2687311
Chemicals, peptides, and recombinant proteins
APRIL (mouse) (multimeric) (rec.)	Adipogen Life Sciences	Cat# AG-40B-0089
BAFF-R (human):Fc (human) (rec.)	Adipogen Life Sciences	Cat# AG-40B-0027
Bovine Fibronectin	Sigma-Aldrich	Cat# F1141
Bovine Serum Albumin, Fraction V (Modified Cohn)	Sigma-Aldrich	Cat# 12660
Corn oil	Sigma-Aldrich	Cat# C8267
EDTA (0.5 M), pH 8.0, RNase-free	Thermo Fisher Scientific	Cat# AM9261
16% Formaldehyde, Methanol-Free	Cell Signaling Technology	Cat# 126060S
Gibco™ ACK Lysing Buffer	Thermo Fisher Scientific	Cat# A1049201
Gibco™ Fetal Bovine Serum	Thermo Fisher Scientific	Cat# A3160601
Gibco™ HBSS (10×), calcium, magnesium, no phenol red	Thermo Fisher Scientific	Cat# 14065056
Gibco™ Penicillin-Streptomycin-Glutamine (100×)	Thermo Fischer Scientific	Cat# 10378016
HEPES	Sigma-Aldrich	Cat #H3375
Imject Alum Adjuvant	Thermo Fisher Scientific	Cat# 77161
Isoflurane Inhalation Solution	Henry Schein	Cat# 1182097
Lactated Ringer’s Solution	Henry Schein	Cat# 14792
Lipopolysaccharides from Escherichia coli O111:B4	Sigma-Aldrich	Cat# L-2630
2-Mercaptoethanol (50 mM)	Thermo Fisher Scientific	Cat# 31350010
2-NBDG (2-(N-(7-Nitrobenz-2-oxa-1, 3-diazol-4-yl)Amino)-2-Deoxyglucose)	Thermo Fisher Scientific	Cat# N13195
NP(32)-KLH	Biosearch Technologies	Cat# N-5060–5
PBS, 10X Powder, pH 7.4	Thermo Fisher Scientific	Cat# BP665–1
Qtracker™ 705 Vascular Labels	Thermo Fisher Scientific	Cat# Q21061MP
Recombinant Mouse ICAM-1/CD54 Fc Chimera Protein	R&D Systems™	Cat# 796IC050
Recombinant Mouse VCAM-1/CD106 Fc Chimera Protein	R&D Systems™	Cat # 643VM050
RPMI 1640 Media, No Glutamine	Cytiva	Cat# SH30096.01
Sulfatrim Pediatric Suspension (sulfamethoxazole and trimethoprim oral suspension USP)	PAI Pharmaceutical Associates, Inc.	NDC# 00121–0854-16
SynstatinVLA-4 custom peptide (CD138 core protein residues 210–236)	GenScript	N/A
TACI (human):Fc (human) (rec.)	Adipogen Life Sciences	Cat# AG-40B-0079
Tamoxifen	Sigma-Aldrich	Cat# T5648
Triton™ X-100 Surfact-Amps™ Detergent Solution	Thermo Fisher Scientific	Cat# 28314
Critical commercial assays
AccuCount Fluorescent Particles	Spherotech	Cat# ACFP705
BD Cytofix/Cytoperm ™ Fixation/Permeabilization Solution Kit	BD Biosciences	Cat# 554714
BD Horizon ™ Brilliant Stain Buffer	BD Biosciences	Cat# 566349
CD43 (Ly-48) MicroBeads, mouse	Miltenyi Biotec	Cat# 130–049-801
Nunc ™ Lab-Tek ™ Chambered Coverglass	Thermo Fisher Scientific	Cat # 155411PK
UltraComp eBeads ™ Compensation Beads	Thermo Fisher Scientific	Cat# 01222242
Experimental models: Organisms/strains
Mouse: Blimp1-ERT2-Cre-TdTomato (BEC)	Jing et al., 2024.	N/A
Mouse: Blimp1-YFP: B6.Cg-Tg(Prdm1-EYFP)1Mnz/J	The Jackson Laboratory	JAX Stock# 008828; RRID:IMSR_JAX:008828
Mouse: B1–8hi: CBy.129P2(B6)-Ightm1Mnz/J	The Jackson Laboratory	JAX Stock# 007775; RRID:IMSR_JAX:007775
Mouse: B6: C57BL/6J	The Jackson Laboratory	JAX Stock# 000664; RRID:IMSR_JAX:000664
Mouse: B6.C-Tg(CMV-cre)1Cgn/J	The Jackson Laboratory	JAX Stock# 006054; RRID:IMSR_JAX:006054
Mouse: CD45.1: B6-Ly5.1/Cr	Charles River Laboratories	NCI Stock# 564
Mouse: CD138 KO: Sdc-1−/−	Alexander et al.46	N/A
Mouse: Rosa26LSLTomato: B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J	The Jackson Laboratory	JAX Stock# 007914; RRID:IMSR_JAX:007914
Mouse: Rosa26LSLYFP: B6.129X1-Gt(ROSA)26Sortm1(EYFP)Cos/J	The Jackson Laboratory	JAX Stock# 006148; RRID:IMSR_JAX:006148
Software and algorithms
Adobe After Effects	Adobe	N/A
Excel	Microsoft	RRID:SCR_016137
FlowJo	FlowJo, LLC	RRID:SCR_008520
Imaris	Bitplane	RRID:SCR_007370
MATLAB	MathWorks	RRID:SCR_001622
Olympus Fluoview FV10-ASW	Olympus	RRID:SCR_014215
Prism	Graphpad	RRID:SCR_002798

Highlights.

• High CD138 levels on plasma cells (PCs) control survival in competitive environments

• CD138 promotes PC clustering, binding to fibronectin, and retention in the bone marrow

• Acute inhibition of APRIL using TACI-Fc disrupts bone marrow PC clusters and mobilizes PCs