Blood-borne human plasma cells in steady state are derived from mucosal immune responses

Protective humoral immunity is provided by plasma cells through the production of antibodies. Because the half-life of the secreted immunoglobulin (Ig) is limited to a maximum of 2 to 3 weeks,¹  persisting humoral immunity must be regulated through mechanisms controlling generation, survival, and homeostasis of plasma cells, that is, terminally differentiated B cells that are generated from activated B cells.^2–5  In a secondary systemic immune response to a protein antigen such as tetanus toxoid, antigen-specific IgG-secreting plasmablasts with somatically mutated V_H gene rearrangements are generated from memory B cells.^6,7  After leaving the secondary lymphoid tissues, they are detectable in human blood between days 6 and 8 after vaccination. At this time, they are migratory and attracted by CXCL12, that is, could migrate to tissues, such as bone marrow.⁶  In the bone marrow, plasmablasts, expressing high human leukocyte antigen–DR (HLA-DR), proliferating and being migratory, can differentiate into long-lived plasma cells, which are no longer migratory, express low HLA-DR, and do not proliferate and continue to secrete their antibodies, thus maintaining humoral memory,^8,9  independent of circulating CD20⁺ B cells, as shown in patients treated with anti-CD20 (rituximab).^10,11  Humoral immunity can also be driven by persisting antigen and the continuous differentiation of B cells into short-lived plasma cells.^3,12 

Survival of plasma cells in vivo depends on specific signals from their environment, the plasma cells' niche.¹³  The number of niches most probably limits the number of plasma cells in the body.¹⁴  Entry into and egress from survival niches thus probably determine the homeostasis of plasma cells specific for a certain antigen. Recently, in accordance with this hypothesis, the mobilization of plasma cells of diverse specificities by tetanus toxoid-specific plasmablasts after tetanus vaccination has been reported.⁶  Trafficking and localization of plasmablasts and plasma cells in the body/tissue are mediated by homing receptors and chemokine receptors.¹⁵  The adhesion molecules CD62L (L-selectin) and α₄β₇ integrin initiate transmigration from blood into tissue by transient interactions with carbohydrates (eg, peripheral lymph node addressin [PNAd]) expressed by endothelial cells in the peripheral lymph nodes, as for CD62L,^16,17  and with the mucosal addressin cell-adhesion molecule-1 (MAdCAM-1) expressed on intestinal endothelial cells, as for α₄β₇ integrin.^18–21  Plasma cells isolated from human lamina propria express high levels of α₄β₇ integrin.²²  Further, rotavirus-specific IgD⁻/CD38^high/CD27^high plasma cells induced in the gut-associated lymphoid tissue express α₄β₇ integrin and C-C motif receptor 10 (CCR10),²³  a receptor for chemotaxis toward CCL28 of mucosal tissues and the skin.²⁴  Interaction between α₄β₇ integrin and MAdCAM-1 initiates transmigration^25,26  and is indispensable for maintenance of secretory IgA levels.^20,27  It has also been demonstrated that CCR9 and its ligand CCL25 mediate homing of plasmablasts expressing this chemokine receptor into intestinal tissue.^28,29  Plasmablasts generated in systemic immune responses do not express CCR9 or CCR10 but do express CXCR4.^30,31  CXCL12, the ligand of CXCR4, is expressed abundantly in human tissues, including spleen, bone marrow, lymph nodes,³²  and mucosal tissues,³³  and mediates recruitment of CXCR4⁺ plasmablasts into bone marrow.^30,31  In short, subsets of plasmablasts have been described expressing distinct chemokine receptors or combinations thereof, enabling them to migrate to mucosal tissue, bone marrow, or inflamed tissue, settle there, and differentiate into plasma cells with different function and fate.

Based on phenotype, human antibody-secreting cells of different anatomic localization, that is, spleen, blood, and bone marrow, have been viewed as distinct successive developmental stages of plasma cell differentiation.^34,35  Here, we provide evidence for an additional layer of complexity, namely, the mucosal versus systemic origin of plasmablasts and plasma cells in blood, indicating the independent chronic generation of mucosal plasmablasts versus the induced generation of parenteral plasmablasts. In steady state, the few antibody-secreting cells detectable express IgA, β₇ integrin and CCR10, and most of them, but not all, are HLA-DR^high, qualifying them as plasmablasts of mucosal immune reactions. Between days 6 and 8 after parenteral (systemic) vaccination with tetanus/diphtheria toxoid, additional prominent populations of IgG⁺, HLA-DR^high, CCR10⁻, CXCR4⁺, vaccine-specific plasmablasts and HLA-DR^low plasma cells with different specificity appear in the blood. Plasmablasts migrate toward gradients of the chemokines they have receptors for, in contrast to plasma cells, confirming our earlier notion that these plasma cells might be no longer migratory and have been dislocated from the plasma cells niche by systemic immune reaction.⁶ 

Citrate or heparinized whole blood from healthy donors (18-57 years of age; average, 35 years) was collected, and peripheral blood mononuclear cells (PBMCs) were isolated by density gradient centrifugation as described before,⁶  using lymphocyte separation medium (PAA Laboratories, Coelbe, Germany).

Some donors were immunized with tetanus/diphtheria vaccine (Sanofi-Aventis, Bridgewater, NJ) after informed consent had been obtained from each donor.

Leukocyte filters were obtained from the blood bank of the Charité University Hospital, Institute of Transfusion Medicine, immediately after preparation. They were flushed reversely with 20 mL cold phosphate-buffered saline (PBS), and PBMCs were isolated by density gradient centrifugation as described before.⁶  All donors were healthy and fulfilled criteria for blood donation. Initial studies of a cohort of blood donors (shown in Figure 1, and Figures S1C,S3, available on the Blood website; see the Supplemental Materials link at the top of the online article) were further expanded by detailed phenotypic and functional analyses of additional healthy volunteers before and after vaccination.

Bone marrow cells were obtained from patients (55-80 years of age; average, 68 years) undergoing hip joint endoprosthetic surgery at the Department for Orthopedics at the Charité Berlin. As previously reported, different age of blood and bone marrow donors does not cause a significant change of total IgG- or IgA-secreting cells in the body.³⁶  The material obtained was flushed with cold PBS supplemented with 0.5% bovine serum albumin (BSA) and 5 mM ethylenediaminetetra-acetic acid (PBS/BSA/ethylenediaminetetra-acetic acid; Merck, Darmstadt, Germany) and was filtrated using cell strainers (70 μm; BD Biosciences Discovery Labware, Bedford, MA). Bone marrow mononuclear cells were isolated by subsequent density gradient centrifugation as described before for PBMCs.⁶  No enzymatic digestion was used.

Blood serum was collected from healthy donors using the Vacutainer system according to the manufacturer's instructions (BD Biosciences, San Jose, CA).

The ethics committee of the medical faculty of the Humboldt University (Charité) approved the study, and patients' informed consent was obtained before enrollment in accordance with the Declaration of Helsinki.

Surface antigens were stained by coincubation of mononuclear cells with monoclonal antibodies (mAbs) for 10 minutes at 4°C in PBS/BSA. Cells were washed once. Dead cells were electronically excluded by adding 4,6 diamidino-2-phenylindole (DAPI; Invitrogen, Carlsbad, CA) directly before acquisition and subsequent electronic gating on DAPI⁻ cells.

Intracellular (ic) antigens were stained after surface staining. Cells were washed twice in PBS, resuspended in 2% formaldehyde solution (Merck), and fixed for 20 minutes at room temperature. Cells were then washed twice in PBS. Saponin (Sigma Chemie, Deisenhofen, Germany) was used as permeabilizing agent at 0.5% solution (saponin buffer) in PBS/BSA containing 0.02% sodium azide (PBS/BSA/azide) for the intracellular staining and at 0.1% solution in PBS/BSA/azide for the washing steps. Cells were labeled with biotinylated or digoxigenated mAb in 0.5% saponin buffer for 15 minutes at room temperature. Cells were then washed with 0.1% saponin buffer. For secondary detection, the procedure was repeated using streptavidin–peridinin chlorophyll protein (PerCP; BD PharMingen, San Diego, CA), streptavidin–allophycocyanin (APC)–Cy7 (Invitrogen), and antidigoxigenin (Roche Diagnostics, Mannheim, Germany) coupled to Alexa 350 (Invitrogen). Finally, cells were analyzed cytometrically on an LSRII cytometer (BD Biosciences) equipped with an additional UV laser and a DivaSoft operation system (BD Biosciences). Data were analyzed using FlowJo software (TreeStar, Ashland, OR). Cell aggregates were excluded according to peak vs area of the forward scatter signal. Cytometric fluorescence data are displayed as 2-color plots in log10 scale, light scatter in a linear scale.

Antibodies used included the following: CD19–phycoerythrin (PE; clone HD37; DakoCytomation, Glostrup, Denmark), CD19-PerCP or -PE-Cy7 (SJ25C1; BD Biosciences), CD27-Cy5 or –fluorescein isothiocyanate (FITC; 2E4; kind gift from René van Lier, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands), biotinylated and FITC-labeled and unlabeled κ (G20-193; BD Biosciences) and λ Ig light chain (JDC-12; BD Biosciences), CD38-FITC, -PE or -APC (HIT2; BD Biosciences), CD20-FITC (2H7; BD Biosciences), CD20-PerCP (L27; BD Biosciences), HLA-DR, coupled to FITC or Cy5 (L243; Deutsches Rheumaforschungszentrum [DRFZ], Berlin, Germany), β₇ integrin–PE (FIB504; BD Biosciences), CD62L-FITC (145; Miltenyi Biotec, Auburn, CA), CD62L-PE-Cy5 (Dreg-56; BD Biosciences), Ki-67–FITC (MIB-1; Dako North America, Carpinteria, CA), CD138-PE (B-B4; Chemicon International, Temecula, CA), IgG-biotin or -FITC (G18-145; BD Biosciences), IgA-biotin (G20-359; BD Biosciences), IgM-biotin (G20-127; BD Biosciences), CD45-PerCP (2D1; BD Biosciences), CD3-FITC (UCHT-1; DRFZ), and CD14-Cy5 (TM1; DRFZ), CD3-PacificBlue (UCHT1; BD Biosciences), CD14-Pacific Blue (M5E2; BD Biosciences), IgM-PE (G20-127; BD Biosciences), CCR10-PE or -APC (314305; R&D Systems Europe, Abingdon, United Kingdom), CCR9-PE (112509; R&D Systems Europe), IgA-FITC (M24A; Millipore, Billerica, MA), and α_E integrin–FITC (Ber-ACT8; BD Biosciences).

On human blood B lymphocytes, β₇-integrin expression is identical with expression of α₄β₇-integrin dimer.³⁷  α₄ integrin (CD49d) is expressed abundantly on blood CD19⁺/CD38^high plasmablasts/plasma cells.³⁴  α_E integrin, a potential alternative dimerization partner for β₇ integrin, was not expressed by peripheral CD19⁺/CD27^high B cells (data not shown). FIB504 staining therefore provided a measure for β₇-integrin and α₄β₇-integrin expression.

For assessment of chemotactic attraction, a chemotaxis assay was used as described previously.^6,30,38  Briefly, 24-well plates with transwell inserts (6.5-mm diameter, 5-μm pore size; Corning Life Sciences, Acton, MA) and RPMI 1640 medium (Invitrogen) supplemented with 0.5% BSA (low endotoxin; Sigma-Aldrich, St Louis, MO) were used. The inserts were coated with 100 μL human fibronectin solution (Invitrogen) at a concentration of 10 μg/mL in distilled water and incubated for 1 hour at 37°C and 5% CO₂. The solution was removed and the inserts were dried for 2 hours at 37°C.

PBMCs or bone marrow mononuclear cells were isolated as described in “Preparation of blood and bone marrow samples,” using prewarmed RPMI 1640 instead of buffers. For some experiments, B cells were enriched using RosetteSep technology (StemCell Technologies, Vancouver, BC). Cells were counted in a Neubauer chamber. The lower transwell chamber was filled with 600 μL assay medium with or without the human recombinant chemokines CXCL12 (stromal cell–derived factor 1α, 10 nM; R&D Systems, Minneapolis, MN), CCL25 (thymus-expressed chemokine or Ckβ-15, 300 nM; PeproTech, Rocky Hill, NJ), CCL28 (mucosae-associated epithelial chemokine, 300 nM, PeproTech) at optimal concentrations^6,39 ; 0.5 or 2 × 10⁶ PBMCs or bone marrow mononuclear cells were added to the upper chamber. Cells were then allowed to migrate for 90 minutes at 37°C in a humid atmosphere (5% CO₂). Finally, cells were collected from upper and lower wells and plasmablasts/plasma cells were enumerated cytometrically. Frequencies of migrated cells (migrated cells counted in the lower chamber divided by cells counted in the upper and lower chambers) are indicated.

Data were analyzed using GraphPad Prism (GraphPad Software, San Diego, CA). Frequencies of various cell populations were calculated with FlowJo software. P values were calculated by Mann-Whitney test for unpaired observations and Wilcoxon test for paired data (both 95% confidence interval, 2-tailed).

As shown in Figure S1, blood plasmablasts/plasma cells were iden-tified cytometrically as intracellular Ig^high (icIg^high) cells, CD19⁺/CD27^high(/CD20^low)⁴⁰  or CD19⁺/CD38^high/CD20^low cells.³⁴  Cells identified as icIg^high cells were stained brightly and exclusively by κ- or λ-light chain antibodies, and counterstaining for appropriate surface antigens confirmed the accuracy of this staining approach (Figure S1A). Less than 0.001% κ light chain⁺/λ light chain⁺ double-positive cells were detected. Antigen specificity of plasmablasts/plasma cells was analyzed as described⁶  using recombinant tetanus toxin C fragment (rTT.C) coupled to digoxigenin. The specificity of staining for icIg and rTT.C was confirmed by staining in the absence of saponin and by inhibition of the staining with 10-fold excess of unlabeled primary antibody and antigen, respectively, resulting in a 10-fold reduction of staining intensity (data not shown). Cytometric quantification of CD45⁺/SSC^low/CD19⁺/CD20^+/− B cells, including plasmablasts/plasma cells, was performed using the TruCount system (BD Biosciences) according to the manufacturer's instructions and antibodies detecting CD45, CD19, and CD20. Numbers of subfractions of plasmablasts and plasma cells were calculated based on previous findings^6,40  that these cells are CD19⁺/CD27^high/CD20^low/icIg^high cells.

In steady state, plasmablasts and plasma cells were detected readily in the blood of healthy persons in the absence of apparent or intentional activation of the immune system. The median frequency of plasmablasts/plasma cells (either CD19⁺/CD27^high lymphocytes or icIg^high cells, Figure S1A,B) was 0.14% (± 39%; SD) of PBMCs (range, 0.03%-2.39%, 49 donors tested; Figure 1). Twenty-three persons were tested 6 to 7 days after secondary systemic vaccination with tetanus/diphtheria toxoid. In these donors, a significantly increased frequency of plasmablasts and plasma cells was observed (P < .001, Mann-Whitney test, 95% confidence interval) with a median of 0.46% (± 0.5%) of PBMCs (range, 0.08%-2.06%). The absolute count of blood-borne plasmablasts/plasma cells in steady state was 2307/mL (± 657/mL; 19 donors). In all donors, when absolute cell numbers were assessed directly before and 7 days after vaccination with tetanus/diphtheria toxoid, increased absolute counts of plasmablasts/plasma cells were observed after immunization (Figure S1C).

Plasmablasts/plasma cells were detected according to expression of icIg light chains (icIg^high) and analyzed for expression of intracellular IgG (icIgG), IgA (icIgA), and IgM (icIgM) (Figure 2A). In steady state, an average of 84% of plasmablasts/plasma cells expressed icIgA (range, 60%-92%, 8 donors). In the same representative persons, icIgG⁺ and icIgM⁺ cells were detected at median frequencies of 12% (range, 1%-38%) and 5% (range, 1%-10%) of total plasmablasts/plasma cells, respectively. Systemic tetanus/diphtheria vaccination resulted in a substantial relative increase of icIgG⁺ cells on day 7 after vaccination (median, 81%; range, 73%-90%, 4 donors), whereas the frequencies of icIgA⁺ and icIgM⁺ cells were 15% (range, 8%-24%) and 4% (range, 1%-8%), respectively (Figure 2B). Before vaccination, the median frequency of icIgG⁺ cells among PBMCs was 0.04% (0.00%-0.11%), of icIgA⁺ cells 0.16% (0.07%-0.38%) and of icIgM⁺ cells 0.01% (0.00%-0.02%), respectively. Seven days after vaccination, the frequencies of icIgA⁺ and icIgM⁺ cells among PBMCs remained constant (icIgA⁺ 0.10%; range, 0.04%-0.31%, P = .283; icIgM⁺ 0.03%; range, 0.01%-0.04%, P = .049), whereas frequencies of icIgG⁺ cells were 10-fold increased to 0.67% (range, 0.37%-1.06%, P = .004) (Figure 2C).

Peripheral steady-state plasmablasts/plasma cells expressed β₇ integrin (31.7% ± 23.1%; range, 7%-95%, 29 donors), indicating the potential of these plasmablasts/plasma cells to bind to MAdCAM-1 on high endothelial venules of gut-associated lymphoid tissues. After parenteral tetanus/diphtheria vaccination, the numbers of β₇-integrin⁺ plasmablasts/plasma cells remained stable (median, 1.259/mL before and 1.702/mL after vaccination), whereas the numbers of β₇-integrin⁻ cells increased significantly on day 7 after vaccination (median, 1.393/mL before and 8.814/mL after vaccination). Of note, rTT.C-binding plasmablasts, representative of vaccine-specific plasmablasts,⁶  did not express β₇ integrin (Figure 3A). In the same assay, CD62L was stained to discriminate antibody-secreting cells of systemic, not mucosal, origin. As expected, most of the vaccination-induced icIg^high plasmablasts/plasma cells expressed the lymph node homing marker CD62L (69%; range, 54%-88%), in particular nearly all plasmablasts binding to rTT.C, cells that were completely absent before vaccination (Figure 3).

In steady state, most peripheral CD19⁺/CD27^high cells expressed the CCR10 (56% ± 19%; range, 16%-79%, 7 donors; Figure 3D), a chemokine receptor functionally expressed by lymphocytes in mucosal tissues of airways and gut,^39,41  mammary gland,⁴²  and skin.^24,43  Counterstaining for IgA showed a quantitative correlation between expression of IgA and CCR10, with a difference in median fluorescence intensity of the CCR10 signal (IgA⁺: 12.407 ± 1.905; range, 10.127-14.327; IgA⁻: 4.438 ± 1.932; range, 2.834-6.996, P = .029, 4 donors; Figure 3E). Seven days after tetanus/diphtheria vaccination, blood-borne, vaccine-specific plasmablasts did not express CCR10. At this time point, the overall frequency of CCR10-expressing cells among total plasmablasts/plasma cells was significantly reduced (18% ± 11%; range, 9%-33%, 4 donors; Figure 3F), resulting from the transient appearance of rTT.C-specific CCR10⁻ plasmablasts. In steady state, all circulating IgA⁺ plasmablasts/plasma cells expressed CCR10 and thus qualified as antibody-secreting cells from mucosal immune reactions: 55% (± 15%; range, 31%-69%) of those coexpressed β₇ integrin (Figure 3G), but not CCR9.⁶ 

Consistent with their expression of CCR10, steady-state plasmablasts spontaneously migrated toward a gradient of CCL28 (300 nM; Figure 4A), in an in vitro transwell migration assay. Individual frequencies of migratory CD19⁺/CD27^high plasmablasts were 16%, 21%, 37%, and 34% (4 donors). A total of 18%, 25%, 12%, and 30% of plasmablasts migrated toward a gradient of CXCL12 (10 nM), and less than 2% migrated toward the intestinal chemokine CCL25, the ligand of CCR9. Steady-state plasmablasts migrating toward CCL28 all expressed CCR10 (2 donors tested). Of the plasmablasts migrating toward CXCL12, most IgA⁺ cells also expressed CCR10, but only few of the IgA⁻ cells (Figure 4D).

After vaccination with tetanus/diphtheria toxoid, the frequencies of plasmablasts migrating toward CXCL12 were 15% (± 13%; 10 donors, median ± SD), and barely any cells could be detected migrating toward CCL25 (< 1%), compared with assay medium (< 1%), in 2 donors tested (Figure 4B). At this time, antigen-specific cells and most of all plasmablasts/plasma cells did not express CCR10, implicating a significantly lower migration as response toward CCL28 compared with steady state.

As expected, less than 0.6% of plasma cells from bone marrow migrated to any of the chemokines tested, that is, more than 99% of bone marrow plasma cells were nonmigratory (Figure 4C), despite the finding that bone marrow plasma cells do express CXCR4 and CCR10, but not CCR9 (Figure 6B).

The mature phenotype of bone marrow plasma cells, as shown in Figure 5, was consistent with previous results obtained in mice.³⁰  Bone marrow plasma cells expressed high levels of CD38, CD138, and icIg (Figure S2). Few, if any, expressed Ki-67 (5% ± 3%), high levels of HLA-DR (12% ± 6%), or CD62L (2% ± 5%) (Figure 5B). Expression of β₇ integrin was detectable on 49% (± 9%) of total bone marrow plasma cells. In peripheral blood, the CD62L⁻/β₇ integrin⁻, HLA-DR^low, Ki-67⁻ plasma cells had a phenotype similar to bone marrow plasma cells (Figure 5B). In contrast, blood plasmablasts expressing CD62L or β₇ integrin were Ki-67⁺ and expressed high levels of HLA-DR (Figure 5A). Only HLA-DR^high plasmablasts were capable of spontaneous migration toward 10 nM of CXCL12 in transwell-migration assays (Figure 5C).

The overall contribution of mucosal plasmablasts to the population of bone marrow plasma cells can be estimated according to expression of IgA and CCR10 by bone marrow plasma cells (Figures 6, S4). A total of 39.5% (± 9.8%; range, 27.6%-56.1%) of bone marrow plasma cells expressed icIgA, 55.1% (± 9.1%; range, 36.4%-63.2%) icIgG, and 6.7% (± 4.7%; range, 3.6%-16.3%) icIgM (Figure S2D). A total of 37% (± 10%; range, 28%-52%, 5 donors) of all CD38^high bone marrow plasma cells, including IgA⁺ plasma cells, expressed CCR10 (Figure 6B). Of the IgA-secreting cells in the bone marrow, 43% (range, 25%-59%) expressed CCR10 and 33% (range, 25%-45%) β₇ integrin (Figure S4). Thus, approximately 40% of IgA⁺ bone marrow plasma cells have a phenotype consistent with their mucosal origin. Migratory, mucosal steady-state plasmablasts are thus apparently not able to extinguish systemic humoral memory. Figure S3 demonstrates that most, if not all, occurrences of increased peripheral numbers of plasmablasts/plasma cells in apparently healthy donors (Figure 1) are the result of mucosal immune reactions.

The homeostasis of antibody-secreting cells providing humoral immunity is still poorly understood. Plasma cells are residing mainly in bone marrow, but also in secondary lymphoid tissue and mucosal tissue. Their survival apparently depends on signals provided by their environment, the plasma cell niche.¹³  It remains controversial whether these plasma cells are continuously replaced^3,12  or only as a consequence of subsequent immune reactions.^6,44  Evidence for the mobilization of memory plasma cells into the blood in the course of an immune reaction has been provided by a previous study.⁶  It remained unclear whether such a mobilization does also occur continuously in steady state, implying a corresponding constant (chronic) generation of new plasmablasts to maintain the observed stability of humoral memory.^3,45 

Antibody-secreting cells of blood in steady state have been analyzed phenotypically before. Arce et al identified IgG-secreting cells of blood by the cytometric secretion assay and showed that these cells are heterogeneous with respect to expression of HLA-DR and CD38,³⁵  and speculated that the CD38^low expressing IgG-secreting cells in blood might be plasmablasts.^35,46  Moreover, Johansen et al have detected circulating CD19^dim/IgA^dim cells in steady state and speculated on their mucosal origin.⁴⁷ 

Here we have analyzed the phenotype and migratory potential of distinct antibody-secreting cells of peripheral blood to determine their possible origin and destination by applying different degrees of immune activation. In accordance with previous data, we found that, in steady state, 11.5 × 10⁶ antibody-secreting cells are circulating in 5 L of blood, compared with 5.5 × 10⁸ residing in bone marrow and 6.5 × 10⁹ in the gut-associated lymphoid tissue.^48,49  More than 80% of the circulating antibody-secreting cells in steady state express IgA, α₄β₇ integrin, or CCR10. In contrast, 7 days after systemic immunization with tetanus toxoid, most of the antibody-secreting cells in blood express IgG and neither α₄β₇ integrin nor CCR10. This indicates that in steady state most, if not all, antibody-secreting cells are derived from mucosal immune reactions. It is doubtful that these cells contribute to humoral memory provided by long-lived plasma cells of the bone marrow, for 3 reasons: first, IgA-secreting memory plasma cells of the bone marrow secrete monomeric IgA and not dimeric IgA.⁵⁰  Second, we do not find a mobilization of IgG⁺ bone marrow plasma cells in steady state, as we did in the context of systemic immune reactions, when newly generated plasmablasts compete with resident memory plasma cells of the bone marrow for survival niches. The 5.75 × 10⁵ IgG⁺/HLA-DR^low plasma cells we find in the present study in the blood during steady state (Document S1) are probably not long-lived plasma cells from the bone marrow because, if they were, the half-life of humoral memory would be less than 1 year, much shorter than actually observed. The half-lives of humoral memory for a variety of pathogens has been determined as ranging between 11 and more than 10 000 years.⁴⁵  Third, the continuous influx of steady-state “mucosal” plasmablasts into bone marrow would result in 80% of the plasma cells in the bone marrow secreting IgA. This is not observed. The frequency of IgA-secreting bone marrow plasma cells is approximately 40% (Figure 6). Therefore, the chronic production of IgA⁺ plasmablasts does not substantially impact on IgG⁺ plasma cell memory. α₄β₇-integrin expression of steady-state antibody-secreting cells allows adhesion and initiation of transendothelial cell migration in the high endothelial venules of gut tissue^18,21,51  and represents a homing marker for the gut mucosa.¹⁵  Both, intentional and unintentional antigenic challenges in the mucosa induce circulating IgA⁺ plasmablasts expressing β₇ integrin.^23,52  Expression of CCR10 is induced in mucosal immune reactions of the airways and the gut where it serves for navigation inside various mucosal tissues, where it mediates chemotaxis along gradients of its ligand CCL28.^24,39,41,53  Coherently, steady-state PBMCs produce mainly secretory IgA in vitro.⁵⁰  The absence of CCR9 expression^28,29  excludes the small intestine as a major contributor of steady-state antibody-secreting cells in blood. Whereas antibody-secreting cells expressing α₄β₇ integrin in steady state could stem from gut-associated lymphoid tissue (except intestine), α₄β₇-integrin⁻ cells probably stem from immune reactions in alveolar mucosal tissue.^54,55  Among steady-state antibody-secreting cells are both dividing, Ki-67⁺, HLA-DR^high plasmablasts as well as nondividing, Ki-67⁻, HLA-DR^low plasma cells. A total of 25% of steady-state antibody-secreting cells are migrating toward gradients of CCL28 in transwell-migration assays. In vivo, this would allow them to home to the gut. Whether the presence of mature plasma cells in steady-state blood reflects the mobilization of resident mucosal plasma cells, in analogy to the mechanism postulated for long-lived IgG-secreting plasma cells of the bone marrow,⁶  could not be analyzed here because the specificity of the plasmablasts and plasma cells could not be compared.

Steady-state IgA⁺/CCR10⁺ antibody-secreting cells also express CXCR4 and are attracted by CXCL12 gradients. Interaction of CXCR4 with its chemokine ligand CXCL12 is involved in localization of plasma cells to the bone marrow³¹  but probably is also involved in recruitment and maintenance of mucosal IgA-secreting cells.⁵⁶  For murine antibody-secreting cells, it has been shown that plasmablasts are attracted by CXCL12 and migrate in response to it. Plasma cells do not but instead use CXCL12 as a survival signal.^13,30  We show here, for the first time, that human HLA-DR^high plasmablasts migrate in response to CXCL12 gradients, whereas HLA-DR^low plasma cells of blood and bone marrow plasma cells do not. This confirms our original notion⁶  that plasma cells from blood are destined to death by neglect, whereas blood-borne plasmablasts have the potential to home to a niche providing survival signals.

For plasmablasts generated in systemic immune responses and expressing CXCR4 but not CCR10, it has been shown that their preferred homing organ is the bone marrow.⁵⁷  Plasmablasts of steady state express both CXCR4 and CCR10 and would have a choice to home to either bone marrow or mucosa. Although 40% of bone marrow plasma cells express IgA or CCR10, it is doubtful that steady-state plasmablasts contribute to this population. Why plasmablasts of steady state express functional CXCR4 but do not home to bone marrow (as discussed above), remains unclear. It has been shown that bone marrow resident IgA-secreting cells secrete monomeric IgA, whereas steady-state antibody-secreting cells secrete dimeric, secretory IgA.⁵⁰  IgA⁺, CCR10⁺ bone marrow plasma cells are probably derived from distinct mucosal immune reactions. In 2 of 50 healthy blood donors analyzed, we detected significantly enhanced numbers of blood-borne antibody-secreting cells. These cells were probably not generated in an unintentional systemic immune response, but rather in an unintentional mucosal immune response, because they expressed α₄β₇ integrin (Figure S3). The recruitment of IgA-secreting cells from distinct mucosal immune responses has been described for mice infected with rotavirus.⁵⁸  The difference between steady-state plasmablasts and plasmablasts generated during infection with regard to their competence to join the pool of bone marrow memory plasma cells appears to be crucial for the development of mucosal vaccines and our understanding of immunity to mucosal virus challenge.

The authors thank Karin Reiter and Kristin Kemnitz for skillful technical assistance, Katharina Raba and Toralf Kaiser for FACS expertise, and all volunteers.

This study was supported by Deutsche Forschungsgemeinschaft (grant SFB 650/TP16 and grant Do491/7-1), Charité University Hospital funding, and the Berlin Senate.

Contribution: H.E.M. and W.S. performed research; H.E.M. analyzed results and made the figures; H.E.M., A.R., and T.D. designed research and wrote the manuscript; and T.Y., F.H., K.T., and R.A.M. discussed results and provided vital material for the study.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Henrik E. Mei, B Cell Memory Group, Deutsches Rheumaforschungszentrum Berlin, Charitéplatz 1, 10117 Berlin, Germany; e-mail: mei@drfz.de.