Transcription factor MIST1 in terminal differentiation of mouse and human plasma cells

Abstract

Despite their divergent developmental ancestry, plasma cells and gastric zymogenic (chief) cells share a common function: high-capacity secretion of protein. Here we show that both cell lineages share increased expression of a cassette of 269 genes, most of which regulate endoplasmic reticulum (ER) and Golgi function, and they both induce expression of the transcription factors X-box binding protein 1 (Xbp1) and Mist1 during terminal differentiation. XBP1 is known to augment plasma cell function by establishing rough ER, and MIST1 regulates secretory vesicle trafficking in zymogenic cells. We examined morphology and function of plasma cells in wild-type and Mist1^−/− mice and found subtle differences in ER structure but no overall defect in plasma cell function, suggesting that Mist1 may function redundantly in plasma cells. We next reasoned that MIST1 might be useful as a novel and reliable marker of plasma cells. We found that MIST1 specifically labeled normal plasma cells in mouse and human tissues, and, moreover, its expression was also characteristic of plasma cell differentiation in a cohort of 12 human plasma cell neoplasms. Overall, our results show that MIST1 is enriched upon plasma cell differentiation as a part of a genetic program facilitating secretory cell function and also that MIST1 is a novel marker of normal and neoplastic plasma cells in mouse and human tissues.

the process of secretory cell differentiation involves a developmentally regulated cascade of events that requires multiple orchestrated changes in large cassettes of genes that mediate fundamental changes in specific organelles allowing the cell to assume its ultimate physiological function. Accumulating evidence suggests that these large-scale changes in gene expression patterns and their subsequent effects on cell architecture and function are controlled by a relatively small subset of transcription factors whose expression may be shared by developmentally diverse cells with similar functions (4, 5, 10, 42, 43, 51). To produce and secrete large quantities of protein, secretory cells expand their endoplasmic reticulum (ER), Golgi, and mitochondrial compartments, increase secretory vesicle biogenesis and membrane recycling, and establish an apical-basal pattern of polarization. Previous results have identified several transcription factors that regulate genes that in turn facilitate these cell structure changes in gastric zymogenic (chief) cells (ZCs). For example, X-box binding protein 1 (Xbp1), critically important in regulation of long-term plasma cell homeostasis (19, 52), has been shown to be necessary for ER expansion and homeostasis as precursor cells in the neck of gastric units migrate into the base and differentiate into digestive enzyme-secreting ZCs (20). However, in neither plasma cells nor ZCs is Xbp1 necessary for survival of the cells or for full development of cell identity, i.e., plasma cells still secrete immunoglobulin and undergo class switch, and ZCs still turn off expression of precursor genes and turn on digestive enzyme secretion. What is affected by the loss of these transcription factors is the efficiency of secretion over the life span of the cells (19, 52). Downstream of Xbp1 activation in ZCs, the basic helix-loop-helix (bHLH) transcription factor MIST1 (BHLHA15) regulates a highly conserved cohort of target genes that facilitate secretory vesicle maintenance and trafficking as well as apical cytoskeletal rearrangements (4, 39, 51). Although MIST1 can repress myoblast differentiation, it is not normally expressed in muscle tissue (27, 36). Rather, by all published accounts, from Drosophila through zebrafish and mammals, MIST1 expression is specific to secretory cells and is developmentally induced during terminal cell differentiation to augment vesicular trafficking in a given cell without affecting that cell's identity or fate (16–18, 34, 35, 51).

Thus previous literature suggests that even though cells may develop in diverse tissues, if they have similar function (e.g., secretory cells) they use similar transcriptional regulatory networks to regulate functionally specific gene cassettes. Elucidating these overlapping gene networks can help identify novel aspects of cell function or define behavior in cells whose function is not entirely clear.

Upon antigen binding to B cell receptors on naive B cells, signals relayed to the nucleus begin the process of preparing the cell for antibody production and secretion as mature, postmitotic plasma cells. During the early stages of plasma cell differentiation, transcription factors such as Blimp1 and Pax5 are part of a transcription factor circuitry that facilitates commitment to the plasma cell lineage (29, 32, 44, 46, 54, 55). This network of transcription factors culminates in induction of Xbp1, followed by increased antibody synthesis, transport, and secretion (1, 21, 25, 26, 40, 47). While the early developmental pathways that initiate plasma cell differentiation have been well characterized, less is known about the genetic events happening downstream of Xbp1 activation when plasma cells become terminally differentiated. These late-stage events in plasma cell formation are thought to be involved in the maintenance of cell architecture and the preservation of proper cellular function throughout the life span of the cell. Similar to ZCs, Mist1 expression at the mRNA level has been shown to be enriched in plasma cells in vitro in response to forced expression of XBP1, and Mist1 is expressed at the transcript level in mouse plasma cells (1, 3); however, the function of MIST1 in plasma cells remains unclear.

In this study we chose to address the role of MIST1 during the terminal differentiation of plasma cells. Here we show that Mist1 expression is enriched in plasma cells relative to all other B cell lineages wherein it regulates a cohort of genes that overlap with Mist1 targets in ZCs and that detection of plasma cells by MIST1 antibodies is an efficient method of phenotypically identifying plasma cells in tissue in both mice and humans. Furthermore, we use a cohort of 12 diverse plasma cell neoplasms to show that MIST1 can be used to identify plasma cell differentiation at least as reliably as the gold standard for plasma cell detection, CD138 (syndecan-1). Given the paucity of plasma cell-specific immunomarkers, we believe MIST1, with its nuclear labeling pattern and its expression only in rare other highly secretory cells, will be a valuable addition to the arsenal of tools to study these cells.

MATERIALS AND METHODS

Mice.

All experiments involving animals were performed according to protocols approved by the Washington University School of Medicine Animal Studies Committee. Germ line Mist1^−/− mice were generated as described previously (37) and maintained in a specific pathogen-free barrier facility. Wild-type mice were the littermates of Mist1^−/− mice resulting from heterozygote by heterozygote crosses. Mist1^CreERT mice were generated as previously described (20). mT/mG mice were obtained from Jackson Labs.

GeneChips.

GeneChip arrays used in these experiments were obtained from the Gene Expression Omnibus (GEO) as follows: gastric cell lineage (GEO accession: GSE5018), B cell lineage (GEO accession: GDS1695), and E_μ-xbp-1s mice (GEO accession: GSE6980).

Bioinformatic analysis.

Chip quality control and GeneChip-to-GeneChip comparisons to generate expression profiles were performed with dChip (28, 58). Cell lineage-specific profiles were generated by extracting those genes whose expression was increased in the given cell lineage relative to the other lineages (parameters: lower bound of 90% confidence of fold change ≥1.2, expression intensity difference ≥50).

Conversion of cell-specific gene lists into Gene Ontology distributions.

Lists of cell type-specific genes for each gastric epithelial cell lineage and B cell lineage were associated with Gene Ontology (GO) terms with GOurmet Vocabulary software (9). We then determined the distribution of GO terms associated with the gene lists for each population and performed unbiased hierarchical clustering of all the expression profiles, using GO term distributions allocated by the GOurmet Cartography software. Those GO terms whose fractional representations in the plasma cell and ZC gene lists were most increased relative to all the other expression profiles were determined by the software. Among those GO terms, those associated with specific subcellular domains or processes were identified and are depicted in Fig. 1.

Fig. 1.

Gastric zymogenic cells (ZCs) and plasma cells express a conserved cassette of genes and transcription factors that regulate protein processing in the endoplasmic reticulum (ER) and Golgi compartments. A: heat map representing the 269 gene probes that are enriched in gastric ZCs relative to other gastric epithelial cell lineages. B: heat map representing expression values of all 269 ZC transcripts on GeneChips from isolated B cell populations (3). Note the cluster of 75 ZC transcripts that are also enriched in plasma cells relative to other B cell lineages. C: fractional representation of genes categorized by specific Gene Ontology (GO) terms is shown for each gastric epithelial and B cell lineage. Individual GO terms that most significantly differentiate ZCs and plasma cells from their sibling cell populations are represented to show how the transcripts enriched in each expression profile match known cellular functions associated with those cell lineages. Genes described by the GO terms “endoplasmic reticulum,” “Golgi apparatus,” “endoplasmic reticulum membrane,” and “Golgi membrane” are highly represented in gene expression profiles of ZCs and plasma cells. D: gene expression of ZC-specific transcription factors is shown for all B cell lineages. Probes highlighted in red indicate ZC and plasma cell overlap. E: gene expression values for Mist1, Xbp1, and Creb3 are clustered by cell type as indicated. Each data point represents raw expression data from a single GeneChip. Note the significant (P < 0.001) lack of expression of the Mist1 gene in all immature B cell subsets. F: heat map representing the change in Mist1 gene expression in tissue from E_μ-xbp-1s mice, where Xbp1s tumor cells are cells from murine myelomas induced by overexpression of the spliced (active) Xbp1 transcript. Color chart below heat maps indicates fold change. *P < 0.05, **P < 0.01.

Spleen B cell cultures and cytospins.

Spleens were harvested from 8- to 10-wk-old mice and mechanically homogenized to generate a single-cell suspension. Red blood cells were depleted by hypotonic lysis, and whole splenic mononuclear cells were maintained at 37°C in 5% CO₂ in RPMI medium supplemented with 10% fetal bovine serum, 2 mM l-glutamine, 100 mg/ml glucose, 10 mM HEPES, 1 mM sodium pyruvate, and 100 ng/ml penicillin and streptomycin. Cells were plated at 2 × 10⁶ cells·ml medium⁻¹·well⁻¹ in 12-well tissue culture plates (TPP). Immediately upon plating, cells were treated with 20 μg/ml lipopolysaccharide (LPS; Sigma, St. Louis, MO). At 24-h intervals after LPS treatment, cells were removed from culture and prepared for flow cytometry or RNA was isolated (Qiagen RNeasy kit, Valencia, CA). Cytospins of cultured cells were prepared with the Shandon Cytospin 3 (Thermo Fisher Scientific, Waltham, MA) and stained with May Grunwald/Giemsa reagents (Sigma).

Quantitative PCR.

cDNA was synthesized from isolated RNA from splenic B cell cultures with SuperScript III (Invitrogen) and random primers. Quantitative real-time PCR (qRT-PCR) was performed with SYBR Green Mix (Abgene, Rochester, NY) and gene-specific primers on an Mx3000P detection system (Stratagene, La Jolla, CA). The following primers were used to quantify Xbp1 mRNA: forward AGC AGC AAG TGG TGG ATT TG, reverse TGA CAA CTG GGC CTG CAC CT; Mist1 mRNA: forward tgg tgg cta aag cta cgt gtc, reverse gac tgg ggt ctg tca ggt gt; and Mib1 mRNA: forward GTC ATC CCA GTC TCC AGG ATT CTG AA, reverse GGA CCA AAA GCC TAA CAA TCT GGG T.

For all qRT-PCR analysis, we used the following methods to analyze and plot data. In each PCR run, we used, as an internal control for every sample, 18S rRNA primers: forward CAT TCG AAC GTC TGC CCT ATC, reverse CCT GTG CCT TCC TTG GA. Differences in 18S RNA threshold cycle (C_t) between each sample were used to normalize each gene-specific C_t for overall starting RNA concentration. To establish a background C_t at which primer pairs amplified nonspecific message, we also ran a no-cDNA (“water”) control for each set of gene-specific primers. The background C_t was usually between 37 and 40 (which was the maximum number of cycles run). To plot “C_t above background,” the C_t from each gene-specific primer for each sample was normalized to all other samples by relative 18S C_t and then subtracted from the background C_t. In this way, plots are essentially log₂ scale (because each cycle means ∼2-fold increase in amplicon), and higher numbers represent higher amounts of starting RNA.

Flow cytometry.

For all flow cytometric experiments, cells were incubated with Fc block (MiltenyiBiotic, Auburn, CA) for 10 min at 4°C, followed by incubation with primary antibodies against CD138 (281-2), B220 (RA3-6B2), IgD (11-26c2a), IgM (G155-228), CD5 (53-7.3), or CD23 (B3B4) (BD Biosciences, San Diego, CA). Cells were analyzed on a FACScan flow cytometer (Becton Dickinson, Heidelberg, Germany).

Enzyme-linked immunosorbent assay.

For splenic B cell cultures, cell-free medium was isolated at 24-h intervals after LPS treatment and analyzed with a commercial enzyme-linked immunosorbent assay (ELISA) kit for murine IgM according to the manufacturer's instructions (Immunology Consultants Laboratory, Newberg, OR). For in vivo antibody secretion assays, serum was isolated from whole blood and fecal pellets were collected from age- and sex-matched mice and analyzed by ELISA kit for murine IgA (Immunology Consultants Laboratory).

Immunofluorescence of mouse tissues.

For murine tissue analysis, small intestine was harvested and fixed in 10% formalin overnight at 4°C, followed by multiple rinses in 70% EtOH, arrangement in 2% agar in a tissue cassette, and routine paraffin processing. Sections (5 μm) were deparaffinized and rehydrated, and then antigen retrieval was performed by boiling in Trilogy reagent (CellMarque, Rocklin, CA). Slides were blocked in 1% BSA, 0.3% Triton X-100 in PBS and then incubated in primary followed by secondary antibodies (see below). Slides were incubated 5 min in 1 μg/ml bisbenzimide (Invitrogen, Carlsbad, CA) before mounting in 1:1 glycerol-PBS.

The following primary antibodies were used for immunostaining in this study: rabbit anti-MIST1 [1:200, described previously (37), gift of Dr. Steve Konieczny, Purdue University], goat-anti IgA (1:100, STAR137F, AbD Serotec, Raleigh, NC), rat-anti mouse CD138 (1:500, 281–2, BD Biosciences), and human CD138. Secondary antibodies used were Alexa Fluor 488, 594-conjugated donkey anti-rabbit, anti-goat, or anti-mouse (1:500, Invitrogen). Cartoon traces of immunofluorescent images were performed with Adobe Illustrator (Adobe Systems, San Jose, CA).

Human plasma cell lesion study.

The Human Studies Committee Institutional Review Board at Washington University Medical Center approved testing of all human aspects of this study. A search of the database of the Lauren V. Ackerman Laboratory of Surgical Pathology at Barnes-Jewish Hospital (Washington University Medical Center, St. Louis, MO) was conducted for plasma cell neoplasms between 2007 and 2009. Ten cases of extramedullary disease and two cases with bone marrow involvement were randomly selected. All cases were previously evaluated by CD138 immunohistochemistry and in situ hybridization for kappa and lambda immunoglobulin light chains to establish clonality. For human MIST1 immunostaining, rabbit polyclonal antibody raised against recombinantly expressed human MIST1 lacking the DNA binding domain was used at a dilution of 1:400. Slides were stained in a Ventana Bench Mark XT automated slide stainer with the Ultra View Universal DAB Detection kit.

Microscopy.

All light photomicrographs were taken on an Olympus BX51 light microscope with DP12 12-megapixel digital camera. For multicolor immunofluorescence, a Zeiss Axiovert 200 inverted fluorescence microscope was used and photomicrographs were obtained via Axiocam MRM camera with Apotome optical sectioning filter (output as TIFF files; 1.4 megapixel: 1,388 × 1,040). Transmission electron microscopy was performed on a JEOL JEM 1200-EX microscope with the AMT Advantage HR (Advanced Microscopy Techniques, Danvers, MA) high-speed, wide-angle 1.3-megapixel transmission electron microscopy digital camera.

Graphing and statistics.

All graphs and statistics were performed in GraphPad Prism. Statistical analysis was, in the case of simple control versus experimental comparison, by Student's t-test. In some cases, post hoc analysis was performed with Bonferroni correction to ensure against multiple comparison bias.

RESULTS

A conserved cassette of functionally specific genes is enriched upon terminal differentiation of gastric zymogenic and plasma cells.

Genes enriched in gastric ZCs relative to other mature gastric epithelial cell lineages were determined as previously described (39). Briefly, ZCs, parietal cells, gastric pit cells (surface mucous cells), and juvenile (postnatal days 7–10) neck cells were isolated individually from frozen tissue sections of mouse (C57BL/6) stomach by laser capture microdissection (LCM). With dChip software, 269 genes with absolute expression values of >50 in ZCs that were also at least (with 90% confidence) 1.2-fold enriched relative to all non-ZC lineages were identified and are depicted as a heat map in Fig. 1A. To determine which ZC-specific genes were also enriched during terminal differentiation of immature B cells into plasma cells, expression values of each of the 269 ZC enriched genes were identified on GeneChips of isolated B cell populations as previously established by Bhattacharya et al. (3). Of the 269 ZC-specific genes with absolute expression values of >50 in plasma cells, 75 of these genes were at least 1.2-fold enriched (with 90% confidence) in plasma cells relative to all other B cell lineages (Fig. 1B). There was no pattern of enrichment of ZC transcripts in other B cell populations. These data suggest that ZCs and plasma cells may employ a conserved cassette of genes, despite their diverse origins.

We hypothesized that the shared gene cassette might comprise genes that facilitate the formation of the subcellular structures necessary for high-capacity secretion. Thus we looked to see whether the cassette of genes was enriched for any particular subcellular compartment or process related to secretion. We determined which GO terms were overrepresented in ZCs and plasma cells relative to their sibling cell populations with GOurmet, a program we designed for this purpose (9, 10). Interestingly, both ZC and plasma cell GO profiles displayed a significant increase in the fractional representation of transcripts categorized under the GO terms “endoplasmic reticulum” (example genes: Ero1lb, Os9; mean of ZC and plasma cell 14.22 ± 2.57% vs. 5.95 ± 0.88% mean for all other lineages, P ≤ 0.01), “Golgi apparatus” (examples: Tmed3, Hs2st1; mean of ZC and plasma cell 9.76 ± 1.03% vs. 4.87 ± 0.98% mean for all other lineages, P ≤ 0.05), “endoplasmic reticulum membrane” (examples: Pigz, Sec11c; mean of ZC and plasma cell 5.52 ± 0.67% vs. 2.17 ± 0.40% mean for all other lineages, P ≤ 0.01), and “Golgi apparatus membrane” (examples: Ica1, Ergic1; mean of ZC and plasma cell 4.14 ± 0.26% vs. 1.7 ± 0.48% mean for all other lineages, P ≤ 0.05) compared with all other gastric and B cell lineages (Fig. 1C). The results suggest that ZCs and plasma cells share a higher-order pattern of gene expression that may play a role in preparing these cells to handle large levels of protein synthesis and secretion from the cell.

Xbp1 and Mist1 gene expression is enriched upon terminal differentiation of gastric zymogenic cells and plasma cells.

Recent evidence has implicated a relatively limited number of transcription factors that are developmentally regulated and induced in terminally differentiated cells to activate expression of genes that organize specific subcellular compartments. For example, the bHLH transcription factor TFEB has recently been shown to induce transcription of a large cassette of diverse genes that enhance lysosome biogenesis in cells (42, 43). We reasoned that regulation of this shared gene cassette might be by specific transcription factors that were influencing expression of a broad cohort of subcellular effectors. Of the 75 transcripts that are unique to the ZC and plasma cell lineages, 5 of these transcripts encode proteins that regulate transcription. Expression profiles for those genes are shown in Fig. 1D. Previous reports have shown that one of those genes, Xbp1, plays an important role in plasma cell differentiation by maintaining ER homeostasis as the level of antibody production and secretion increases (1, 15, 19, 21, 26, 47). Furthermore, studies from our lab have shown that Xbp1 is also necessary for maturation of gastric zymogenic cells; in the absence of Xbp1, ZCs still develop, but they are much smaller, with considerable reduction in rough ER (rER) and secretory granules (20). Expression of the bHLH transcription factor Mist1 (Bhlha15) was also enriched upon plasma cell differentiation (Fig. 1, D and E).

The pattern of Mist1 expression in B cell differentiation was unique among shared ZC/plasma cell-specific transcription factors: its expression was undetectable (26.74 ± 3.7 mean expression value) in all immature B cell subsets, where other transcription factors, although enriched in plasma cells, still maintained moderate expression levels in the cell stages leading up to plasma cell differentiation (563.3 ± 49.33 and 1,089 ± 82.25 mean expression for Xbp1 and Creb3, respectively; P < 0.001) (Fig. 1E). Mist1 has been implicated as a downstream gene target of XBP1 in plasma cells in vitro and in ZCs in vitro and in vivo (1, 20). Consistent with those findings, we analyzed the pattern of Mist1 expression in a published report (6) of Xbp1 overexpression. Twenty-six percent of mice with transgenic expression of Xbp1 open reading frame under the control of the immunoglobulin V_H promoter and E_μ enhancer elements (E_μ-xbp-1s mice) developed multiple myeloma between 14 and 24 mo of age. Interestingly, our analysis shows that Mist1 gene expression is enhanced in those plasma cell myeloma samples derived from E_μ-xbp-1s mice (Fig. 1F). Analysis of Mist1 expression in published expression profiles of all other mouse immune cells showed no expression in any T cell or innate immune population (not shown). Collectively, the data suggest that Mist1 is a transcription factor unique in the immune system to terminally differentiated plasma cells that may play a role in determining secretory cell identity and function in the late stages of plasma cell differentiation.

Mist1 is expressed during plasma cell differentiation in vitro and regulates genes that function in maintaining cytoplasmic membranes.

LPS can stimulate plasma cell differentiation in vitro by signaling through Toll-like receptor complexes, thus activating B cells to proliferate, increase immunoglobulin expression, expand the secretory apparatus, and secrete antibody over the course of 3–4 days in cell culture (13, 24, 45, 47, 48). To prove that Mist1 was specifically expressed during plasma cell differentiation, whole spleens from wild-type C57BL/6 mice or Mist1^−/− mice were plated as single-cell suspensions and treated with a single dose of LPS (20 μg/ml medium). After 24-h exposure to LPS, wild-type and Mist1^−/− spleen cultures remained dominated by cells that by flow cytometry displayed intermediate levels of forward scatter and low levels of side scatter, indicative of the presence of small, agranular cells. Cytospins of these cultures confirmed the presence of small B lymphocytes (3–4 μm) with densely packed nuclei and a high nuclear-to-cytoplasm ratio (Supplemental Fig. S1B, representative of wild-type and Mist1^−/− cultures at day 1 after LPS).¹ These cultures contained few mature plasma cells as indicated by CD138 and B220 coexpression where plasma cells (CD138⁺, B220⁺) are contained within the indicated box (Supplemental Fig. S1A). However, after 3-day exposure to LPS, changes in light scattering properties were observed, with a threefold increase in the number of cells with high forward-scatter and intermediate side-scatter patterns, reflecting an increase in the number of cells with expanded cytoplasmic compartment and organelle content (Supplemental Fig. S1C). These data correlated with a significant increase in the percentage of CD138⁺, B220⁺ mature plasma cells relative to day 1 cultures [2.28 ± 0.38% vs. 18.97 ± 1.94% (P < 0.001) and 1.75 ± 0.26% vs. 18.31 ± 2.01% (P < 0.001) for wild-type and Mist1^−/− cultures, respectively; n = 3 separate experiments]. Furthermore, cytospins of cell cultures at day 3 after LPS indicated an increase in the number of morphologically mature plasma cells, identified as cells with expanded cytoplasm (black arrow), nuclei with patchy, open chromatin (white arrow), and a cytoplasmic Hoff (white arrowhead) (Supplemental Fig. S1D). Of note, no significant differences were observed in the formation or survival of mature plasma cells due to LPS treatment in Mist1^−/− spleen cultures, as no increase of a granular, small population representing apoptotic cells and cell fragments was observed. Differentiation of the culture into mature plasma cells correlated with an increase in the level of secreted IgM after 4 days in culture (Supplemental Fig. S1E). Interestingly, Mist1 gene expression was not necessary for the increased secretion of IgM protein.

Cells were harvested at 24-h intervals after LPS treatment, and mRNA was isolated to assess gene transcript levels of Xbp1 and Mist1 over time. Consistent with previous reports, Xbp1 gene expression increased gradually over the course of LPS treatment, with maximum levels achieved by day 3 (Fig. 2A). These findings were similar for wild-type and Mist1^−/− spleen cultures. As shown in Fig. 2B, Mist1 gene expression significantly increased after 2 days of exposure to LPS and reached peak levels by day 3. As expected, Mist1 expression was undetectable in Mist1^−/− spleen cultures. To determine the effect of loss of Mist1 on gene expression in plasma cells, we performed GeneChip array analysis of these ex vivo spleen cultures at 1 and 3 days with or without exposure to LPS. Bioinformatic analysis indicated a list of 700 genes whose expression was significantly changed (1.2-fold enriched) in wild-type plasma cell cultures at 3 days after LPS (peak time of Mist1 expression in wild-type cultures) compared with Mist1^−/− cultures and whose expression was specific for day 3 after LPS (peak time of plasma cell differentiation). Of these 700 genes, 218 genes were enriched in plasma cells compared with all other B cell lineages. GO profiles of these 218 genes displayed a significant increase in the fractional representation of transcripts categorized under the GO terms “integral to membrane,” “endoplasmic reticulum,” and “transport.” Histological assessment of intestinal plasma cells from Mist1^−/− mice revealed rER dilatation compared with wild-type plasma cells (Supplemental Fig. S2). In addition, this list of genes included several that have been previously described as MIST1 targets in ZCs (51), such as Serpini1, Arrdc3, and Ccpg1 (Fig. 2D). We also identified a novel MIST1 target in plasma cells, Mindbomb 1 (MIB1), an E3 ubiquitin ligase known to mediate the ubiquitination of the canonical Notch ligand Deltalike and the tumor suppressor Death associated protein kinase (DAPK). MIST1 dependence of Mib1 expression was confirmed by qRT-PCR (Fig. 2, D and E).

Fig. 2.

Mist1 is expressed in the late stages of plasma cell differentiation after treatment of whole spleen mononuclear cells with lipopolysaccharide (LPS) in vitro. A and B: quantitative RT-PCR (qRT-PCR) of spleen cultures from wild-type and Mist1^−/− mice for levels of Xbp1 and Mist1 gene expression. RNA was isolated from the cultures and analyzed for gene expression at 24-h intervals after LPS treatment. y-Axis represents log₂ change in expression relative to levels in well containing no cDNA after 18S normalization. C_t, threshold cycle. C: heat map representing the 700 gene probes that are significantly changed (1.2-fold enriched) in wild-type (WT) plasma cell cultures at 3 days after LPS (peak time of Mist1 expression in wild-type cultures) compared with Mist1^−/− cultures, and whose expression is specific for day 3 after LPS (peak time of plasma cell differentiation). Of these 700 genes, 218 show plasma cell specific expression compared with other B cell lineages. D: heat map of known MIST1 target genes and 1 unique target, Mindbomb 1, that are enriched in wild-type plasma cell cultures. E: qRT-PCR of spleen cultures from wild-type and Mist1^−/− mice for levels of Mib1 gene expression (2.33 ± 0.219-fold increase in wild type). *P < 0.05, **P < 0.01, ***P < 0.001.

Our data implicate MIST1 as a developmentally regulated transcription factor that is expressed during the late stages of plasma cell differentiation following or concomitant with induction of XBP1 that regulates genes that, similar to ZCs, have the function of facilitating vesicular trafficking and expansion of the cytoplasm.

Mist1^−/− mice display normal plasma cell development and antibody secretion in vivo.

There has been only one previously reported experiment on MIST1 in plasma cell differentiation, which noted that loss of MIST1 led to a slight, but statistically significant, decrease in NP-specific antibody secretion from plasma cells after nitrophenylated chicken gammaglobulin (NP-CGG) immunization in mice (3). To further characterize the function of MIST1 in plasma cells in vivo, levels of antibody secretion in wild-type and Mist1^−/− mice were assessed under homeostatic/standard housing conditions without exposure to antigenic stimulus. All experiments were performed in mice of the same age, kept in the same cages. Plasma cells resident in the small intestine are the primary source of IgA antibodies in the blood and intestinal lumen. As shown in Fig. 3A, blood serum levels of IgA were unchanged in Mist1^−/− mice compared with wild-type mice [102.3 ± 18.88 vs. 100.4 ± 10.03 μg/ml, respectively; P = not significant (n.s.), n = 8]. Similarly, levels of IgA in fecal pellets were unchanged in Mist1^−/− with respect to wild-type mice (7.75 ± 1.9 vs. 8.49 ± 1.77 μg/g, respectively; P = n.s., n = 8). To determine whether loss of MIST1 causes a delay in the maturation or differentiation of plasma cells in vivo, the number of preplasma B cells and mature plasma cells in the spleen was determined by flow cytometry. Splenic B lymphocyte subsets were determined as follows: follicular B cells (IgD^hi, IgM^lo), marginal zone (MZ) B cells (IgD^lo, IgM^hi), B-1a B cells (CD5^hi, IgM⁺), B-1b B cells (CD5^lo, B220^lo), B-2 B cells (CD23⁺, IgM⁺), and plasma cells (CD138⁺, B220⁺). As shown in Fig. 3B, each B cell subset was equally represented in wild-type and Mist1^−/− mice.

Fig. 3.

Mist1^−/− mice display normal plasma cell development and antibody secretion. A: blood serum and fecal pellets from wild-type and Mist1^−/− mice were assayed for IgA levels by ELISA. Each data point represents an individual mouse (n = 8 mice/group). Mice were matched by age and sex between wild-type and Mist1^−/− populations. B: fractional representation of B cell subsets in wild-type and Mist1^−/− spleens as determined by flow cytometry as follows: follicular B cells (IgD^hi, IgM^lo), marginal zone (MZ) B cells (IgD^lo, IgM^hi), B-1a B cells (CD5^hi, IgM⁺), B-1b B cells (CD5^lo, B220^lo), B-2 B cells (CD23⁺, IgM⁺), and plasma cells (CD138⁺, B220⁺). C: flow cytometric analysis of bone marrow harvested from wild-type (left) and Mist1^−/− (right) femurs. CD138⁺, B220⁺ plasma cells are shown in boxes (n = 4 mice/group).

The bone marrow acts as a reservoir for long-lived plasma cells responsible for maintaining steady-state levels of serum antibodies. Circulating plasma cells are recruited to the bone marrow stroma by specific cytokine interactions, such as SDF-1/CXCR₄, which facilitate long-term (up to 1 yr in mice) survival (11, 38, 49, 50, 53). To assess the effect of loss of MIST1 on the recruitment and/or survival of bone marrow plasma cells, femurs were collected from wild-type and Mist1^−/− mice and bone marrow mononuclear cells were isolated. Flow cytometry revealed no differences in the number of bone marrow plasma cells, represented by coexpression of CD138 and B220, between wild-type and Mist1^−/− mice (0.24 ± 0.026% vs. 0.22 ± 0.025% of total bone marrow mononuclear cells respectively; P = n.s., n = 4 mice/group) (Fig. 3C). Consistent with these findings, the number of IgA⁺ plasma cells in the small intestine was unchanged in Mist1^−/− mice (data not shown). Collectively, these data suggest that Mist1 expression is not necessary for normal plasma cell development and/or survival under homeostatic conditions.

Plasma cells can be sorted on the basis of Mist1 promoter expression.

We next reasoned that although Mist1 function may be redundant or dispensable in plasma cells, the highly specific pattern of Mist1 expression might still be of considerable import, because there are scant markers of plasma cell differentiation that are sensitive and specific. By crossing Mist1^CreERT mice (20) with mT/mG double-fluorescent Cre reporter mice (31) we were able to establish a line of reporter mice in which expression of a tamoxifen-inducible Cre gene was driven from the Mist1 promoter. In these mice, upon treatment with tamoxifen Cre recombinase translocates to the nucleus, where recombination converts a gene locus from constitutive expression of membrane-targeted tdTomato (mT) to constitutive expression of membrane-targeted EGFP (mG). Therefore, in theory, any cell expressing Mist1 would be mG positive while non-Mist1-expressing cells will be mT positive. We chose to use this system to test whether Mist1 expression could be used as a metric to sort plasma cells by flow cytometry. Mist1Cre^ERT-mT/mG mice were given an intraperitoneal injection of tamoxifen (1 mg/20 kg mouse) every other day for 5 days. Five days after the last injection of tamoxifen, bone marrow samples were harvested and mononuclear cells isolated. Using flow cytometry, we detected a distinct population of Mist1-expressing mG cells (G gate) in the bone marrow constituting 0.105 ± 0.007% (based on 2 independent experiments) of the total bone marrow mononuclear cell fraction that were not present in the vehicle-treated controls (Fig. 4A). As shown in Fig. 4B, expression of the plasma cell marker CD138 was >150-fold enriched in the Mist1-mG cell population compared with the cells expressing mT (as defined by G and R gates, respectively, in Fig. 4A) (50.53 ± 0.69% vs. 0.095 ± 0.007% CD138-expressing cells per group, respectively; P < 0.0001, n = 2 mice in independent experiments). The overlap between the Mist1-mG and plasma cell fractions was expected, though still striking, given the extremely small percentages of these related populations in the global marrow aspirate. Although there may be some error due to the small number of events recorded, two independent mice showed essentially identical results, suggesting that about half of the Mist1-mG cells identified in the bone marrow do not express CD138. Those cells may represent a precursor stage of plasma cell differentiation in which Mist1 expression precedes the induction of CD138 surface expression, or, more intriguingly, Mist1 expression may be maintained in a plasma cell population that has downregulated or lost CD138 expression; these cells would be missed during normal surveys of plasma cells (see other evidence of CD138⁻ plasma cells below). Additionally, these Mist1-expressing CD138⁻ cells may represent a unique cell population in the bone marrow outside the plasma cell lineage.

Fig. 4.

Mist1-expressing plasma cells can be sorted by flow cytometry. A: flow cytometric analysis of bone marrow harvested from Mist1^CreERT-mT/mG femurs after tamoxifen treatment. mT cells and mG cells are indicated by the R and G gates, respectively (n = 2 mice). B: fractional representation of plasma cells (as indicated by CD138 expression) for non-Mist1-expressing mT cells (mT+/mG⁻, R gate) and Mist1-expressing mG cells (mT−/mG⁺, G gate).

Mist1 identifies mature plasma cells in mouse and human tissue.

Currently, for both diagnostic pathologists and basic science investigation, the only commonly used protein marker is CD138 (syndecan-1), which also labels other cells. Thus we decided to take advantage of the highly coordinated expression pattern of Mist1 during B cell differentiation into plasma cells to investigate whether MIST1 could phenotypically identify plasma cells in situ. Figure 5A is a cross-sectional representation of mouse small intestine in which primary antibody staining for MIST1 is represented in red and IgA in green. Intestinal villi are outlined by white dashed lines, with white arrows indicating the nuclear staining pattern of MIST1⁺, IgA⁺ plasma cells. Cell counts taken from multiple mice indicate that 98% of all IgA⁺ plasma cells in the small intestine were also positive for MIST1 staining. As a negative control, small intestines from Mist1^−/− mice were stained as previously described; intravillous plasma cell nuclei in Mist1^−/− mice are indicated by white arrowheads (Fig. 5B). Because Paneth cells in the crypts of the small intestine have been shown to express MIST1 (36), we have excluded the crypts from our analysis. Figure 5C shows wild-type small intestine stained with primary antibodies against MIST1 (red) and CD138 (green). Interestingly, cell counts showed that nearly all CD138⁺ cells were also MIST1⁺. However, just as in the FACS of bone marrow, the converse was not true (76.95 ± 4.67% of Mist1⁺ cells were CD138⁺). MIST1⁺ cells that did not coexpress CD138 are indicated by dashed arrows (23.04 ± 4.67% Mist1⁺, CD138⁻) (Fig. 5C). These data suggest that MIST1 can positively identify nearly all IgA-secreting plasma cells in the mouse small intestine; however, not all MIST1-expressing cells had detectable levels of CD138 on their cell surface, suggesting that this population of mature plasma cells may represent plasma cells in an earlier or later state of differentiation.

Fig. 5.

MIST1 antibody staining positively identifies mature plasma cells in normal mouse and human tissue. A and B: fluorescent microscopy showing anti-MIST1 (red) and anti-IgA (green) antibody staining in the small intestine of wild-type (A) and Mist1^−/− (B) mice. Graph on right of A represents quantification of MIST1⁺, IgA⁺ and MIST1⁺, IgA⁻ cells. Positive nuclear staining for MIST1 is indicated by white arrows in wild-type sections, with absence of MIST1 antibody staining in Mist1^−/− sections shown with white arrowheads. C: sections of small intestine from wild-type mice stained for anti-MIST1 (red) and anti-CD138 (green) antibodies and quantification of MIST1⁺, CD138⁺, and MIST1⁺, CD138⁻ cells (shown by dashed arrows, right). Note the nonnuclear, nonspecific background staining with the MIST1 antibody in the small intestine villi. Of note, the small intestine is unusual in the degree of background immunofluorescence when our MIST1 antibody is used. However, the MIST1 label is nonnuclear and nonspecific in the epithelial membrane under these conditions. D: tissue sections of normal human stomach stained with anti-MIST1 (red), anti-CD138 (green), and DAPI (blue). Right: ×100 representation of the area indicated by white box, left, showing MIST1 expression in CD138⁺ plasma cells. White arrowheads, left, indicate MIST1⁺ gastric ZCs. Note that the antibody for CD138 stains, in addition to plasma cells, gastric parietal cells as well as surface epithelial cells. Scale bars, 50 μm.

Under normal conditions, the human stomach contains scant resident plasma cells. Figure 5D represents a cross section of a human stomach with no histopathological abnormality that has been stained with primary antibodies for MIST1 (red) and CD138 (green). The stomach lumen is to the left, and bases of gastric units are to the right. There are scattered CD138⁺ plasma cells in the subluminal mesenchyme that also show strong nuclear expression of MIST1, as indicated within the box (white arrows). Of note, CD138 antibody labels epithelial cells in many tissues and, in stomach, stained gastric parietal cells as well as surface epithelial cells, highlighting the necessity for novel markers to identify plasma cells more specifically in situ. Thus we show that costaining with anti-MIST1 and CD138 antibodies uniquely identifies plasma cells in the adult human stomach. As expected, gastric ZCs in the bases of the glands had a MIST1⁺, CD138⁻ phenotype (white arrowheads). These data demonstrate that MIST1 is expressed in adult human plasma cells and that antibodies against MIST1 can positively identify mature plasma cells in nonpathological human tissue.

During chronic infection with Helicobacter pylori, the continual cycles of chronic inflammation can lead to metaplastic transformation of the gastric epithelium. To determine whether MIST1 can identify plasma cells in pathological conditions in humans, we immunostained a tissue section from a patient with chronic atrophic gastritis and early metaplastic lesions of the two most common types: intestinal metaplasia and pseudopyloric metaplasia [aka spasmolytic polypeptide-expressing metaplasia (SPEM)]. Figure 6 shows rare foci of residual normal gastric glands with MIST1⁺ ZCs and MIST1⁻ parietal cells (another normal gastric epithelial cell) as indicated by white and black arrows, respectively. In addition, MIST1 antibody staining revealed a diffuse MIST1⁺ plasma cell infiltrate (see Fig. 6, inset for high-magnification photomicrograph of plasma cells positive for MIST1). Thus, during chronic atrophic gastritis, there is loss of epithelial MIST1 and increase in MIST1-expressing plasma cell infiltrate.

Fig. 6.

Mist1-expressing plasma cells are increased in metaplastic stomach tissue. Stomach sections from a patient with characteristic patterns of metaplastic tissue were stained with anti-MIST antibodies (brown). Inset: high-magnification image shows normal MIST1⁺ ZCs (white arrows) and MIST1⁻ gastric parietal cells (black arrows) and abundant expansion of infiltrating MIST1⁺ plasma cells (yellow box and picture at bottom) in mesenchyme.

MIST1 expression correlates with CD138 immunoreactivity in human plasma cell neoplasms.

MIST1 would be particularly useful diagnostically to pathologists if it were expressed in neoplastic cells exhibiting plasma cell differentiation. Thus we next examined a cohort of plasma cell lesions from patients with diverse plasma cell neoplasms in diverse tissues. Ten cases of extramedullary disease and two cases with bone marrow involvement were randomly selected from the tissue archived in the Division of Anatomic and Molecular Pathology at the Washington University School of Medicine after a search of tissue taken between 2007 and 2009. All cases had been previously evaluated by CD138 immunohistochemistry and in situ hybridization for kappa and lambda immunoglobulin light chains to establish clonality. Expression of MIST1 was scored as percentage of plasma cells with positive nuclear expression (0%, <10%, 10–25%, 25–50%, 50–75%, 75–90%, and >90%) (Table 1). As shown in Table 1, 10 of 12 patients displayed nearly complete overlap in plasma cell CD138 and MIST1 expression. Examples of these results are represented in Fig. 7, A and B, where biopsies from two of the above patients were stained with hematoxylin and eosin and CD138 and MIST1, respectively. Of note, strong nuclear expression of MIST1 in plasma cells was observed (black arrows), with little or no cytoplasmic staining. No immunoreactivity for MIST1 was found in epithelial cells, muscle, endothelial cells, and normal hematopoietic elements within the bone marrow [all cases were evaluated independently by three pathologists (J. M. Klco, J. L. Frater, and J. C. Mills)]. In two cases, CD138 and MIST1 immunoreactivity patterns were different: in one, more cells were positive for MIST1; in the other, the opposite pattern was observed. The data suggest that MIST1 immunoreactivity is as useful in diagnosing plasma cell lesions in humans as the gold standard, CD138. Furthermore, because MIST1 is nuclear and can occasionally identify a higher fraction of cells showing plasmacytoid differentiation, it might be an improvement over CD138 in certain diagnostically challenging cases.

Table 1.

MIST1 expression in human plasma cell neoplasms

Sex	Age, yr	Tissue	Final Diagnosis	CD138+ Plasma Cells, %	Mist+ Plasma Cells, %
F	44	Left humerus	Extramedullary plasma cell myeloma	>90	>90
F	61	Right neck	Plasmacytoma	>90	50–75
F	74	T3 vertebrae	Plasmacytoma	25–50	25–50
F	80	Sacrum	Extramedullary plasma cell myeloma	>90	>90
M	59	Right distal tibia	Extramedullary plasma cell myeloma	>90	>90
M	67	T1 vertebrae	Extramedullary plasma cell myeloma	50–75	>90
F	62	Right axillary lymph node	Plasmacytoma	25–50	25–50
F	63	Left pubic ramus	Plasmacytoma with anaplastic features	>90	>90
M	69	Liver	Extramedullary plasma cell myeloma	>90	>90
M	72	T6 vertebrae	Extramedullary plasma cell myeloma	>90	>90
F	48	Bone marrow (∼10% plasma cells)	Plasma cell myeloma	>90	>90
M	68	Bone marrow (∼10% plasma cells)	Plasma cell myeloma	>90	>90

Fraction of cells with plasma cell differentiation that were positive for CD138 and MIST1 expression was scored from biopsies of patients with various plasma cell lesions and diagnoses.

Fig. 7.

MIST1 is expressed in plasma cell neoplasms of multiple patients. A and B: biopsies from 2 patients are shown stained for hematoxylin and eosin (H & E; left), CD138 (center), and MIST1 (right) in serial tissue sections. Note that the cell population staining for the membrane marker CD138 also stains with nuclear MIST1. Inset photomicrographs represent MIST1 positive plasma cells in high magnification. Scale bars, 20 μm.

DISCUSSION

Here we have identified a cassette of genes expressed in common across two entirely different cell lineages. We show that the shared gene expression pattern includes coexpression of Mist1, which is a transcription factor known to regulate membrane trafficking in highly secretory cells from Drosophila to mammals. Gene array analysis and in vitro experiments confirmed that Mist1 expression goes from undetectable in preplasma B lymphocytes to highly enriched after Xbp1 activation and terminal plasma cell differentiation, at which point it upregulates a cohort of transcripts identified as direct targets in gastric ZCs. Although we could find no functional consequences of loss of MIST1 in plasma cells, we show that the specific, late-stage expression of MIST1 in plasma cells can be of tremendous advantage for both researchers and diagnosticians, as antibodies against MIST1 could uniquely identify mature plasma cells by immunohistochemistry and flow cytometry in tissue with a significant and striking overlap with CD138 staining, the gold standard for plasma cell identification.

MIST1 is a developmentally regulated transcription factor that is expressed in the terminal stages of differentiation of a handful of secretory cell lineages in diverse tissues (1, 3, 22, 23, 37, 39, 51, 56). Furthermore, the Mist1 gene is highly conserved from Drosophila and zebrafish to mammals. There is 78% identity across the DNA binding domain between mouse and fly Mist1, and MIST1's function as an enhancer of the cellular secretory apparatus with expression confined to scattered, diverse protein-secreting cells is also highly conserved. Even the MIST1 consensus cis-regulatory E-box sequence is conserved from fly to human (16–18, 34, 35, 51). Despite this high degree of cross-tissue, cross-species conservation, loss of Mist1 does not inhibit the cell fate determination of the cell lineages in which Mist1 expression is enriched. Instead, in all cases so far studied, loss of Mist1 function has been shown to cause profound changes in cell architecture and cell function. Gastric ZCs lacking MIST1 have small, unorganized secretory granules and disrupted apical/basal membrane polarity (39). Similarly, pancreatic acinar cells from Mist1^−/− mice and zebrafish have abnormalities in organelle localization and loss of normal calcium signaling (16, 30).

Hallmark features of secretory cell differentiation include expansion of the ER and Golgi compartments, increased secretory vesicle biogenesis, and cell membrane polarization. Plasma cells are unique among the MIST1-expressing secretory cells studied so far. Unlike ZCs, pancreatic acinar cells, and Drosophila peptidergic neurons, plasma cells do not contain abundant, large secretory vesicles, suggesting that protein secretion in plasma cells may be mechanistically different from other MIST1-expressing secretory cells. Furthermore, while most secretory cells mature within static tissue niches and derive their polarized architecture from interactions with the local tissue environment, plasma cell development takes place as the cells migrate through secondary lymphoid tissues and enter the circulation, therefore requiring plasma cells to remain fluid in their morphology. These differences may account for the fact that Mist1^−/− mice have seemingly normal plasma cell function. We have studied plasma cell morphology in situ in mouse intestine and in vitro in splenic cultures by immunofluorescence and electron microscopy, and, unlike all other normally MIST1-expressing cells, we see no consistent morphological differences in Mist1^−/− plasma cells during steady-state conditions (not shown). Only at the ultrastructural level did we detect MIST1-dependent difference in plasma cells, where Mist1^−/− plasma cells observed in mouse small intestine showed aberrant, dilated rER. Our findings suggest that either 1) the MIST1 transcriptional regulatory pathway in plasma cell is redundant with compensatory mechanisms for maintaining proper plasma cell function in the absence of MIST1 or 2) the importance of MIST1 is in a function of plasma cells that is not yet known or measurable. Thus further experiments will be necessary to determine whether Mist1 expression has alternative function in plasma cells, such as regulation of long-term survival in specific niches or, possibly, in response to certain types of stress. We favor the interpretation that MIST1 is important for plasma cell function, given that high levels of Mist1 expression are observed specifically upon terminal differentiation and that there are no known transcription factors with function redundant to that of MIST1.

Plasma cells can be recruited to a wide array of tissues throughout the body based upon cytokine signals emanating from the local tissue environment. Under normal conditions, plasma cells reside mainly within the bone marrow and the small intestine, where they mediate the primary humoral immune response to foreign antigen (33, 50). However, uncontrolled plasma cell infiltration and mislocalization can in many cases signify a disease state or progression of an extant condition (7, 8, 12, 14, 41, 57). Thus accurate identification of plasma cells within the tissue is important in the diagnosis and treatment of disease states. Cell surface markers such as CD138, CD56, CD28, and CD20 have been shown to identify plasma cells in various stages of disease [for review see Bataille et al. (2)]; however, cell surface expression of these and other membrane-bound plasma cell markers is transient and can be altered by enzymes in the local tissue environment or by experimental procedure. Furthermore, only CD138 is specific for plasma cells among all other immune cells, and as can be seen even in the present study, CD138 also labels nonimmune cells (e.g., parietal and surface epithelial cells in the gastric mucosa). Our finding that there are MIST1⁺, CD138⁻ cells with plasma cell morphology in normal intestine, in sorted bone marrow, and in plasmacytoid neoplasms suggests that there may be plasma cell populations that have been missed previously by investigators using only CD138 and a B cell marker (like B220) to identify plasma cells. Thus identifying plasma cells by flow cytometry and in situ with antibodies against a nuclear-specific protein such as MIST1 should provide a critically important alternative weapon in the relatively limited diagnostic arsenal for detecting plasma cell differentiation in neoplasms and in basic research.

GRANTS

Further support came from National Institutes of Health Grants RO1 DK-079798-01 (J. C. Mills) and T32 CA-009547-23 (B. J. Capoccia, A. J. Bredemeyer).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).