Abstract
Expression of a germline VH3609/D/JH2 IgH in mice results in the generation of B1 B cells with anti-thymocyte/Thy-1 glycoprotein autoreactivity by coexpression of Vk21-5/Jk2 L chain leading to production of serum IgM natural autoantibody. In these same mice, the marginal zone (MZ) B cell subset in spleen shows biased usage of a set of Ig L chains different from B1 B cells, with 30% having an identical Vk19-17/Jk1 L chain rearrangement. This VH3609/Vk19-17 IgM is reactive with intestinal goblet cell granules, binding to the intact large polymatrix form of mucin 2 glycoprotein secreted by goblet cells. Analysis of a μκ B cell AgR (BCR) transgenic (Tg) mouse with this anti–goblet cell/mucin2 autoreactive (AGcA) specificity demonstrates that immature B cells expressing the Tg BCR become MZ B cells in spleen by T cell–independent BCR signaling. These Tg B cells produce AGcA as the predominant serum IgM, but without enteropathy. Without the transgene, AGcA autoreactivity is low but detectable in the serum of BALB/c and C.B17 mice, and this autoantibody is specifically produced by the MZ B cell subset. Thus, our findings reveal that AGcA is a natural autoantibody associated with MZ B cells.
This article is featured in In This Issue, p.489
Introduction
Antibodies present in serum of normal animals in the absence of specific Ag immunization are called natural Abs. Among these, Abs binding to self-antigens, predominantly IgM Igs encoded by germline genes, are termed “natural autoantibodies” (1–3). Natural autoantibodies that bind to intracellular constituents, such as DNA, nuclear proteins, and cytoskeletal components, and to plasma proteins are common in vertebrates at all ages, from newborn to adult (2, 4). The presence of autoantibodies to apoptotic or senescent cells, which expose such intracellular constituents, and to oxidized low-density lipoprotein in serum, suggests that a fundamental role for natural autoantibody may be rapid elimination of damaged cells and clearance of degraded self-molecules (1, 5, 6). Furthermore, cross-reactivity of natural autoantibodies to determinants present on bacteria or viruses enables a rapid protective response to infection (7). Thus, the presence of natural autoantibodies contributes both a housekeeping function and also defensive immunity. Although it is known that genetic background, such as MHC-linked genes, affects the natural autoantibody repertoire (8), the details of how such natural autoantibodies are generated and controlled remain a subject of continued debate. In mice, one clear source is B1 B cells. These B cells are generated by self-ligand–mediated signaling, thereafter serving as a source of natural autoantibodies (9). We show in this study that marginal zone (MZ) B cells also make a natural autoantibody, producing IgM with autoreactivity to mucin 2 (Muc2), a major component of intestinal goblet cell granules and secreted intestinal mucus.
The MZ is a region in spleen between the lymphoid-rich white pulp and the red pulp that consists of an open circulatory network that filters the blood (10). B cells residing in this MZ site encounter and trap pathogens circulating in blood, with or without the aid of Ag-presenting dendritic cells (DCs), and rapidly respond, serving as a defensive barrier (11). Large polymerized Muc2 that bears abundant and variable glycans (12) is the secreted mucin in gut, a major component of intestinal mucus that functions to block microbacterial invasion (13). Such Muc2 in the gut lumen is constantly sampled by DCs in the intestine (14). Our data demonstrate that developing B cells with autoreactivity to this heavily glycosylated intestinal mucin become MZ B cells and accumulate at this site. This process is dependent on Btk, a kinase involved in B cell AgR (BCR) signaling. Btk is essential for IgM and IgG3 natural Ab production in serum (15, 16). Thus, our data demonstrate BCR-ligand–mediated selection leads to autoreactive MZ B cell generation and natural autoantibody production. This bears a striking similarity to Btk-dependent positive selection of B1 B cells by Ag, which contribute a different natural autoantibody in the same animal. In humans, the presence of anti-goblet cell Ab in serum has been recognized for decades, originally discovered in colitis patients (17, 18). Our data in mice suggest that such anti-goblet cell Abs in humans may also include a natural autoantibody, as described in the Discussion.
Materials and Methods
Mice
The VH3609/D/JH2 targeted insertion (knock-in) mouse line (VH3609t) was made by homologous recombination in ES cells as previously reported (19), with a slight modification to the targeting construct, by extending the short (right side) homology arm from 0.8 kB to 2.5 KB, transfecting the R1 ES line (129/Sv × 129/Sv-CP F1 origin). The floxed NeoR gene was eliminated by crossing the knock-in founder to CMV-CreTg.BALB mice. VH3609μ transgenic (Tg) mice have been described previously (9). To establish VH3609/Vk19-17 μκTg mouse lines, we cloned a Vk19-17/Jk1 rearranged κ L chain gene from a VH3609μ+ MZ B cell–derived hybridoma, P16-8F5, by long-distance PCR. Vk19-17 is identical to GenBank MUSIGLAFE. Generation of μκTg mouse lines was carried out as previously described (20), by injection into (C3H × C57BL/6)F1 eggs. One μκTg founder line, EP67, was used in this study as representative of six anti–goblet cell/mucin2 autoreactive (AGcA) μκ lines (all showed increased MZ B cell generation and Tg IgM production in serum). The VH3609/Vk19-17 μκ Tg (EP67 μκTg) mouse line and the VH3609t mouse line were backcrossed to C.B17 mice more than six generations before use in experiments. Btk mutant Xid and Rag-1 knockout mice were both on a BALB/C background. Muc2 and Muc1 knockout mice were on a C57BL/6 background. C.B17, BALB/cAnN, C57BL/6 ICR, and C.B17 SCID were bred and maintained in our laboratory animal facility. Germ-free (Swiss Webster) mice and Sprague Dawley rats were purchased from Taconic. All animal experiments were conducted under a protocol approved by the Fox Chase Cancer Center Institutional Animal Care and Use Committee.
Flow cytometry analysis and reagents
Multicolor flow cytometry analysis, sorting, and mAb reagents have been described previously (21). Rat anti-mouse Ig idiotype Abs used in this study were made as previously described, by using SM6C10 IgM (VH3609/Vk21-5) as immunizing Ag (20). The anti-VH3609SM6C10 idiotype Ab “VH3609id” is P9-10C7, a rat IgG2a. The P9-13H8 Ab 13H8k, a rat IgG1, reacts with a restricted set of mouse κ L chains, including Vk21-5SM6C10 and Vk19-17P16-8F5. The rat P9-19A4 anti-thymocyte/Thy-1 glycoprotein autoreactivity (ATA) idiotype Ab (20) does not react with VH3609/Vk19-17 IgM, distinguishing AGcA from ATA among 13H8κhi+ B cells.
Single-cell Igκ-chain sequencing
For VH3609μ Tg mouse data in Supplemental Table I, cDNA was prepared and amplified from individual sorted B cells, MZ B (CD21hiCD23−AA4− VH3609id+) and follicular (FO) B cells (CD21medCD23+AA4−VH3609id+) from spleen and immature B (AA4+CD21−CD23−) cells from bone marrow (BM) and spleen, for κ gene sequencing by nested PCR as described previously (9). The Vk-Ck PCR product was cloned using TA vector (Invitrogen, Carlsbad, CA); then individual plasmid clones were sequenced. Sequencing efficiency from individual cells under this procedure was 80% for all B cell preparations. To assess mouse-to-mouse variation of VH3609t mouse Igκ usage by MZ B and FO B cell subsets, we sorted individual Igκ+cells directly onto AmpliGrid AG480F slides (Beckman Coulter) using a FACS-VantageSE flow cytometer or a FACSAria II (Becton Dickinson), followed by a reverse transcription (RT) reaction and two rounds of PCR amplification. RT and first-round PCR were performed using the Qiagen OneStep RT-PCR kit (Qiagen) in a 1-μl reaction containing Vκ forward primer, 5′-GACATTGTGATGACCCAGTCTC-3′ and Cκ reverse primer, 5′-CCATTTTGTCGTTCACTGCCATC-3′, which was also used as the RT primer for cDNA synthesis. Second-round PCR amplification used the same Vκ primer and a nested Cκ primer 5′-GAAGCACACGACTGAGGCACC-3′. The second-round PCR fragments were purified and then sequenced using a nested Cκ primer. The efficiency of obtaining sequence was between 70 and 90%. Kappa L chain genes were classified according to the ImMunoGeneTics Information system database (http://www.imgt.org) and a report from Thiebe et al. (22).
BM cell transfer
Cell sorter purified B/T lineage–negative, stem/progenitor cell–enriched fractions (CD19− IgM− CD5− and low side scatter) from BM of EP67μκ Tg C.B17 mice and non-Tg C.B17 mice were cotransferred to C.B17 SCID mice (lightly irradiated, 3 Gy, 1 d before) by i.v. injection (1:10 transfer ratio; 5 × 104 cells from EP67 and 5 × 105 cells from C.B17, per recipient). Transfer of EP67μκ Tg mouse BM stem/progenitors alone (5 × 105 cells/recipient) was used as a control. Analysis was done 5–8 wk after transfer.
Quantitative RT-PCR assay
Gene expression was quantitated by real-time PCR, in duplicate, using TaqMan assays from Applied Biosystems, an ABI 7500 real-time thermal cycler, and ABI software (Life Technologies). Relative gene expression levels were normalized using β-actin values as a standard.
Measurement of serum transgene Ab
To measure transgene-derived Ab in EP67 mouse serum by ELISA, we coated the assay plate with P9-10C7(VH3609id), then detected it by either P9-13H8, in conjunction with biotin mouse anti-rat IgG1 (MCA194B; Serotec, Raleigh, NC) or biotin mouse anti-IgMa, followed by alkaline phosphatase–conjugated avidin. Quantitation of total IgM/κ, IgMa, and IgMb levels by ELISA was described previously (20).
Intestinal tissue cryosection immunofluorescence analysis
Dissected intestine was gently flushed of stool material by ice-cold PBS in a 1-ml syringe, then frozen. Staining of ethanol-fixed frozen sections (8-μm thickness for intestine and 5–6 μm for other tissues) and microscope imaging was done as described previously (23). First, for comparing IgM produced by SP2/0 cotransfected with VH3609μ IgH and Vk19-17κ IgL (MK19) versus IgM produced by SP2/0 cotransfected with VH3609μ IgH and Vk21-5κ IgL (MK21) staining (both IgMa+), C.B17 (IgMb) mouse colon cryosections were incubated with IgMa-containing samples (at 2–5 μg/ml) for 1.5 h at 37°C, followed by incubation with fluorochrome-conjugated anti-IgMa Ab, together with different fluorochrome-conjugated lectins (30 min, at room temperature). Second, natural autoantibodies were tested at 1/100 dilution of sera. RagKO mouse intestine was used to comparably test different IgM allotype serum samples from normal mouse strains, in combination with fluorochrome-conjugated anti-IgM Ab. Third, IgM in 3- to 4-d culture supernatant of LPS-activated B cells was adjusted to 2–10 μg/ml for staining. Fluorochrome (fluorescein or Alexa Fluor 350)-conjugated lectins for cryosection staining were isolectin B4 IB4 (BSI-B4) (Sigma-Aldrich), peanut agglutinin (PNA; Vector Laboratories), Dolichos biflorus agglutinin (DBA; Vector Laboratories), and wheat germ agglutinin (WGA; Invitrogen). DAPI (Thermo Scientific) was used to visualize nuclei. Stained slides were viewed with a Nikon Optiphot epifluorescence microscope. Images were recorded using the QImaging Retiga-1300 cooled CCD camera and processed with Openlab software (Improvision). Images were obtained with either a ×10 or ×20 objective lens (original camera magnification, ×264).
Isolation of colonic crypts
After opening the colon longitudinally and washing in Dulbecco’s PBS without Ca2+ and Mg2+, specimens were incubated with 0.04% sodium hypochlorite for 15 min at room temperature, then incubated with 3 mM EDTA for 90 min at room temperature, with moderate shaking throughout the incubation. Then samples were vigorously shaken for 2 min, removing stromal materials, and colonic crypts/mucous granules were pelleted by centrifugation at 480 × g for 3 min at 4°C.
Electrophoresis and Western blotting
Crypts were transferred into a new tube, washed, incubated at room temperature for 5 min, then centrifuged at 8000 rpm for 5 min to release mucus into the supernatant. Crypt supernatant was treated with 1% NP-40 lysis buffer and applied to 3.3% SDS-PAGE with a 2.5% stacking gel, following a standard procedure, in the presence of 2-ME. Also, 10% and 15% SDS-PAGE were used to test for the presence of MK19-specific bands. After blocking membranes with 5% nonfat dry milk/TBS/0.05% Tween 20 for 1 h at room temperature, Muc2 and MK19 immunoblotting was carried out with goat anti-Muc2 (N terminus peptide) in combination with HRP-donkey anti-goat IgG Ab (both from Santa Cruz Biotechnology), and by MK19 IgM in combination with HRP-goat anti-mouse IgM (Santa Cruz Biotechnology), respectively. For lectin blotting, biotin conjugated-DBA lectin (Vector Laboratories) was used in combination with HRP-streptavidin (Southern Biotech).
Cross-linking of Ab and immunoprecipitation
RagKO mouse colonic crypts were incubated with MK19 or MK21 IgM for 1.5 h at 37°C, washed with PBS, then incubated with PBS containing 1 mM cross-linker, dithiobis succinimidyl propionate (DSP; Thermo Scientific), for 30 min at 37°C, and finally quenched by adding 20 mM Tris-HCl. For immunoprecipitation, cells were lysed in NP-40 lysis buffer, and soluble lysate was incubated with rat anti-mouse IgM Ab (331.12) bound protein G-Dynabeads (Invitrogen) for 18 h. Beads were washed four times in 0.5% NP-40 lysis buffer. Bound material was eluted with 1% NP-40 lysis buffer containing Laemmli buffer with 2-ME, boiled for 5 min; then supernatant was applied to SDS-PAGE, for IgM, Muc2, and DBA lectin blotting.
Experimental colitis
Mice were treated by oral administration of 4% dextran sulfate sodium (DSS; 36–50 kDa molecular mass; MP Biomedicals, Solon, OH) in drinking water for 5 d, which was thereafter replaced by normal distilled water. Examination of the intestine was carried out 3 (the declined body weight), 8 (recovered body weight), and 14 d after cessation of DSS water.
Results
BCR associated with the MZ B cell subset
Expression in mice of a germline VH3609/D/JH2 μ H chain as a transgene (VH3609μTg) promotes generation of B1 B cells with anti–Thy-1 autoreactivity (9). ATA recognizes highly glycosylated Thy-1(CD90) that is predominantly expressed on the membrane of immature thymocytes (24), in both native and also dying cells. ATA IgM is secreted into serum as a natural autoantibody by B1 B cells (9, 20). This ATA BCR uses a Vk21-5/Jk2 κ L chain, generated by rearrangement of the Ig L chain locus (9). In targeted VH3609/D/JH2 insertion mice (knock-in, VH3609t), as shown in Fig. 1A, among CD5+ B cells (B1a) in the peritoneal cavity, VH3609+ B cells expressing the engineered IgH are present (VH3609id+) with a large proportion being 13H8khi+. This anti-κ Ab recognizing limited set of κ L chains including Vk21-5 and Vk19-17 (“13H8k”), which we developed by immunization with ATA IgM, reacts with a restricted set of mouse κ L chains at various levels, including Vk21-5/Jk2 κ. Vk21-5/Jk2 κ expression in 13H8khi+ B1a B cells, as ATA B cells, was confirmed by single-cell sequence analysis (31/31 = 100%). In contrast, the majority of CD5− B220hi B cells (B2) in the peritoneal cavity show less staining by 13H8k than ATA B cells (Fig. 1A) and do not express Vk21-5/Jk2 κ.
MZ B cell–associated BCR, VH3609/Vk19-17. (A) Analysis of B cell subsets (gated rectangles color coded on plots) in the peritoneal cavity washout cells (PerC) and spleen from a VH3609t mouse. The VH3609id+ cell gate is marked by a vertical line. The majority of 13H8khi/med Igκ B cells are also VH3609id+ in the MZ B subset (arrow). (B) Igκ single-cell sequence data from MZ B and FO B subsets of four individual VH3609t mice. Percentages are from a total of 25–36 samples per mouse B cell subset gated as VH3609id(10C7)+Igλ−. Three Igκs enriched in the MZ B cell subset, in contrast with FO B cells (empty bars), are shown. Vk19-17/Jk1 Ig L chains used by individual MZ B cells were all identical. (C) Selected Igκ-chain usage data comparing the immature, FO B, and MZ B cells in VH3609μTg mice (all data are provided in Supplemental Table I). Confirmation of highest usage of identical Vk19-17/Jk1 Ig L chains by MZ B cells (in CDR3, the Vk-Jk junction encodes a proline [P] residue).
In the spleen of the same mouse, an alteration of the 13H8k binding pattern diversity is seen in VH3609id+ B cells from the transitional immature stages (CD21lo CD24hiAA4+) that can progress to become either mature FO B cells or MZ B cells. In contrast with VH3609id + FO B (13H8kmed/lo and 13H8k−), MZ B cells, including CD21hiCD23+ pre-MZ B cells, show predominant 13H8khi/med cells (arrow in Fig. 1A, Spl B/MZ B). Single-cell sequence data from four individual VH3609t mice showed frequent usage by MZ B cells of an identical κ-chain, Vk19-17/Jk1, followed by Vkba9/Jk5 and Vk19-25/Jk1 (Fig. 1B). Further single-cell Igκ sequence analysis using VH3609μ Tg mice, comparing VH3609id+ MZ B cells (CD21hiCD23−) with immature and FO B cells, is shown in Supplemental Table I; the selected Igκ set associated with MZ B is shown in Fig. 1C. These results revealed that ∼30% of MZ B cells have an identical Vk19-17/Jk1 rearrangement (CDR3 region is shown in Fig. 1C), in contrast with much lower and the lowest frequency in FO B and immature B cells, respectively. This Vk19-17/Jk1 Igκ is another L chain strongly recognized by13H8k.
To assess whether the accumulation of B cells with this BCR in the MZ B subset occurs directly from immature transitional stage cells or a secondary outcome of mature FO B cell activation, we made the μκ Tg mouse line, EP67, expressing this VH3609/Vk19-17 BCR. In EP67 mice, the majority of newly generated immature B cells in BM express this Tg BCR. At 3 wk of age, early generated μκ Tg+ CD21−CD24hi immature B cells progressed to become CD21hi (and CD1d+ CD9+) in spleen without intermediate generation of FO B cells (Fig. 2A, asterisk), comprising a significant fraction of the CD21hiCD23− MZ B cell pool (Fig. 2A). In adult (2 mo) animals, μκ Tg+ B cells became both MZ B cells and FO B cells when competitor Tg− B cells were absent as assessed by surface phenotype by flow cytometry analysis, colocalization by immunohistology staining, and expression of MZ B cell–specific genes (Fig. 2B). However, the addition of non-Tg B cells in chimeric BM transferred mice restricted μκTg+ B cells to predominantly assume a CD21hi MZ B cell fate (Fig. 2C, asterisk). Furthermore, introducing Btk deficiency in the context of μκTg EP67 mice (EP67.Xid) resulted in nearly complete loss of 13H8khi μκ Tg+ MZ B cells (Fig. 2D), whereas neither CD40 deficiency nor complete elimination of T cells by Rag-1 deficiency had this effect (Supplemental Fig. 1A). Taken together, these data indicate that a common self-antigen or environmental Ag present from the neonatal through adult age promotes VH3609/Vk19-17 MZ B cell development directly from transitional stage cells in a BCR-dependent manner, without a requirement for T cell help.
Btk-dependent MZ B cell development by VH3609/Vk19-17 BCR B cells. (A) Spleen B cell FACS analysis of the VH3609/Vk19-17 μκTg mouse line, EP67, at 3 wk of age. The CD21hi B cell area is marked by red dotted line. Predominance of CD21hi pre-MZ B and MZ B cells in Tg+ mice (starred) relative to Tg− littermate. (B) FACS analysis of 2 mo adult EP67 spleen B cells. Tgμκ+ B cells, the predominant population, comprise 40–50% in spleen, and half become either FO B or MZ B cells. MZ B cell phenotype cells are colocalized in the MZ with SIGNR1+ macrophages (ER-TR9+), and FO B cell phenotype cells are present in B220+ follicles with CD21hi FO DCs (20× objective lens). Quantitative PCR analysis of selected genes was compared in sorted VH3609id+ MZ B and FO B cells from EP67 mice (n = 3 each, mean and SE) with wild type C.B17 mouse FO B cells. (C) Mixed cell transfer of adult mouse BM hematopoietic stem cell–enriched fractions using a 1:10 Tgμκ EP67/non-Tg ratio. Six weeks after transfer, spleen B cell FACS analysis was performed, gating for VH3609id+ (Tg+, IgMa+) cells (9% of total B cells) and VH3609id− (Tg−, IgMb+) B cells. Representative data of four recipients (n = 4) from two separate transfer experiments are shown. All mice showed predominant MZ B cell generation by the Tg+ B cells. Transfer of EP67 BM alone generated both MZ B and FO B cells, as found in intact adult EP67 mice (data not shown). (D) Analysis of BM AA4+ immature B and spleen B cells in 2-mo-old EP67 μκTg mice without or with Xid (Btk mutant) background. In contrast with BM, in spleen there is a reduction of total B cell numbers (2-fold lower in Xid), together with reduction of μκTg B cell frequency, and an absence of VH3609/Vk19-17 MZ B cells (red ellipse region) in EP67.Xid mice. Representative data are shown from analyses of four mice.
Anti-goblet cell autoantibody to Muc2
Previously we found that Thy-1 expressed at aberrantly low levels compared with physiologic results in ATA MZ B cell generation rather than B1 B cells (21). Thus, the possibility of weak anti–Thy-1 reactivity was the first explanation considered for MZ B cell generation in 13H8khi VH3609/Vk19-17 BCR Tg mice. However, the ATA idiotype Ab (20) did not bind this IgM, and Thy-1 deficiency did not alter MZ B cell generation (Supplemental Fig. 1A), in contrast with production of VH3609/Vk21-5 ATA B cells that are Thy-1 dependent (20, 21). Therefore, to investigate its reactivity, we cotransfected the VH3609μ and Vk19-17κ genes into the SP2/0 hybridoma line to produce secreted IgM, referred to as “MK19.” This Ab was compared with the IgM produced by SP2/0 cotransfected with ATA VH3609μ and Vk21-5κ, termed “MK21.” Both transfected IgMs are IgMa. The binding specificity of MK19, testing for possible autoreactivity associated with MZ B generation, was first assessed by standard procedures, including cell/tissue/cell line staining and ELISA. Although MK19 binding was below detection in most assays used, including assessment of reactivity to the cytoplasm and nucleus of various tissues and cell lines (Supplemental Table IIA), it intensely stained intestinal mucosal cell suspensions. This is in sharp contrast with thymocyte-restricted staining by MK21, an Ab that shares the same IgH (Fig. 3A). When we used ethanol-fixed cryosection staining, MK19 bound to intestinal goblet cells, together with contiguous luminal mucus, with staining distinctively highest in colon, followed by small intestine, but absent in stomach tissue (Fig. 3B, 3C).
AGcA by MZ B cell–associated VH3609/Vk19-17 IgM (MK19). (A) Colon mucosal cell suspension was stained by MK19, in contrast with thymus cell suspension that was stained by MK21. (B) Cryosection staining of colon. Goblet cells and contiguous luminal mucus were stained with MK19, but not MK21. IB4 (BSI-B4) using a 10× objective lens. (C) MK19 staining of transverse sections of intestines and stomach using a 20× objective lens. (D) Colon goblet cells (horizontal section) were shown to be PAS+ (paraffin section staining). MK19-stained granules in the goblet cells, which were distinct from PNA and IB4 bound supranuclear areas and nuclei stained by DAPI, as assessed using a 20× objective lens. (E) Apical goblet cell staining by MK19 was similar to DBA lectin binding in colon. MK19 staining (bottom middle) was merged with PNA and WGA lectins (top middle). (F) MK19 staining of Muc1- and Muc2-deficient mouse colon demonstrates predominantly Muc2-dependent binding of MK19. (G) Absence of MK19 staining of rat colon goblet cells. (E)–(G) using a 10× objective lens.
The major constituent of intestinal goblet cell granules is mucin, and Muc2 is the predominant secretory mucin found in colon, small intestine, but not stomach (25). Therefore, we hypothesized that VH3609/Vk19-17 Ab reacts with Muc2. Muc2 dimerizes in the endoplasmic reticulum and becomes heavily O-glycosylated in the Golgi, followed by polymerization, generating a long-chain glycoprotein structure (26). The Muc2 macromolecule rapidly expands during secretion, forming a highly polyionic network, constituting the intestinal mucus. MK19 binds goblet granules with a high content of O-linked glycans, recognized by labeling with periodic acid–Schiff stain (PAS), and separate from PNA-labeled areas that identify an early glycosylated form of mucin in the supranuclear Golgi region (Fig. 3D). In colon, there is preferential MK19 staining at the apical region of colonic crypts, rather than the areas labeled by WGA (Fig. 3E, middle panel); this apical region staining is similar to DBA binding areas that are enriched for α-N-acetylgalactosamine glycan (Fig. 3E, right panel). Most MK19 binding was Muc2 dependent, as shown by a large reduction of staining in Muc2 gene knockout mice, whereas transmembrane Muc1 deficiency did not alter binding (Fig. 3F). MK19 goblet cell reactivity was found with intestinal tissue from various mouse strains (data not shown), whereas no reactivity was observed with rat (Fig. 3G), similar to ATA IgM (MK21) sharing IgH, which has reactivity to thymocytes of various mice, but not rat (27). Intestinal goblet cells from germ-free mice also contain granules stained by MK19 (Supplemental Fig. 2). MK19-reactive granules in germ-free colon are on the apical region toward secretion into lumen as found in normal mice.
The importance of a fully glycosylated polymeric antigenic form for MK19 recognition, rather than either glycan or protein alone, was initially suggested by the absence of MK19 binding to paraffin-embedded intestine section samples, regardless of deglycosylation treatment (data not shown). By Western blotting, a ∼600-kDa band was detected corresponding to the major Muc2 glycoprotein precursor, comprising the majority of Alcian blue/PAS binding colon mucin and the majority of DBA+ Muc2 (Fig. 4A). In contrast, MK19 IgM failed to reveal a Muc2-dependent band using 3.3–15% SDS-PAGE, by this direct immunoblot procedure. Because most MK19 binding is Muc2 dependent, and MK19 binds to the surface of granules in a goblet cell suspension, we hypothesized that its binding requires a native Muc2 polymatrix glycoform normally expressed in mucus that is lost in the denatured form present in SDS-PAGE, which differs from reactivity of the Muc2 anti-peptide Ab. Therefore, we used a DSP cross-linking approach with MK19 preincubated goblet cell granules/mucus to examine MK19 linked material by anti-IgM immunoprecipitation. Formation of Muc2 polymatrix renders a large fraction of mucin insoluble to extraction (28) (Fig. 4B, right panel), and most MK19 IgM-bound material was also insoluble (Fig. 4B, left and middle panels) and, therefore, intractable to immunoprecipitation analysis. However, a small amount of MK19-containing material was soluble, as was a small amount of Muc2 (Fig. 4B). Immunoprecipitation with anti-IgM beads of this small amount of soluble MK19 bound lysate confirmed that the ∼600-kDa DBA+ Muc2 glycoprotein was directly bound by MK19, but not by MK21 control (Fig. 4C). This MK19-Muc2 binding is not to glycan alone, however, because it does not stain rat colon goblet cells (Fig. 3G) that contain DBA+ Muc2 that is similar to mouse (29). Therefore, we termed IgM with this “anti-goblet cell autoreactivity,” predominantly to intact Muc2, as “AGcA.”
Predominant MK19 AGcA binding to intact Muc2 glycoprotein. (A) Western blot of colon NP-40 lysate. Mucus released from crypts by centrifugation was suspended in 1% NP-40 and applied to 3.3% SDS-PAGE with a 2.5% stacking gel for detection of glycosylated colon mucin(s). Comparison of wild-type B6 versus Muc2KO.B6. Anti-Muc2 (N terminus peptide). (B) Colonic crypts from Rag1 null mice were incubated with MK19 (or MK21), cross-linked (CL) with DSP, then suspended in NP-40 buffer for SDS-PAGE. Left set is Western blot of IgM H chain band using soluble and insoluble lysate, showing the presence of MK19 IgM CL colon granules detected by anti-IgM. Right set of the same lysates preincubated with MK19 shows copresence of Muc2. MK19 IgM and Muc2 were both predominantly in the insoluble fraction. (C) Soluble material from MK19 IgM prebound (MK21 as a control) colon granules, followed by cross-linking, was incubated with anti-IgM–bound protein G Dynabeads for 18 h, then washed, eluted, and applied to SDS-PAGE analysis. Presence of MK19 bound to ∼600 kDa DBA+Muc2 is shown. IB, immunoblot; LB, lectin blot.
Natural AGcA production from MZ B
In the AGcA μκTg line EP67, Tg+ B cells (recognized by staining for the Tg IgMa allotype) become plasma cells (PCs) in red pulp (Fig. 5A, PC). Secreted transgene IgM comprises most of the IgM in serum, and this secretion is T cell independent (Fig. 5B). Anti-goblet cell autoreactivity by IgMa in EP67 mouse serum was confirmed by colon staining, in comparison with ATA μκTg 3369 mouse serum, where thymocyte-reactive IgMa is predominant, produced by B1 B cells (20) (Fig. 5C). Thus, AGcA B cells differentiate into cells secreting autoantibody into serum, as with ATA B cells.
Natural AGcA IgM production from MZ B cells. (A) The presence of AGcA PCs (high IgMa+ and CD19−) in EP67 mouse spleen. CD19+ B cells are also stained with IgMa+ at a lower level than PCs. (B) Predominant AGcA IgM in serum of EP67 mice was independent of T cells (EP67.RagKO). μκTg IgM is IgMa on a C.B17 (IgMb) mouse background. n = 9 each. (C) Colon and thymus cryosection staining comparison between EP67 AGcA μκTg mouse and 3369 ATA μκTg mouse sera, both with Tg IgMa predominance in serum. (D) AGcA reactivity by IgM produced from normal mouse MZ B cells. Colon from Rag1 null mice stained by LPS-stimulated B cell culture supernatant IgM (adjusted to 2–10 μg/ml) from BALB/c mice. B1 (B1a) B cells in the peritoneal cavity and MZ B cells, and FO B cells in spleen were purified from the same mouse. Representative of three experiments (two with BALB/c, one with C.B17). (A), (C), and (D) using a 20× objective lens. RP, red pulp; WP, white pulp.
In both mice and humans, the major fetal colonic mucin is Muc2, as in adults (30). Thus, AGcA-reactive granules, predominantly Muc2, are normally present in gut from the fetal/neonatal stage throughout life as an abundant self-antigen. Because of this natural abundance, to assess whether AGcA B cell generation is also occurring in non-Tg normal mice, LPS mitogen-stimulated culture supernatant from different B cell subsets was tested for AGcA activity. Colon staining analysis showed that IgM(s) with goblet cell reactivity, similar to VH3609/Vk19-17 AGcA, is produced by MZ B cells of BALB/c (and C.B17) mice, but not by B1 or FO B cells (Fig. 5D). Furthermore, AGcA autoantibody was detectable in serum, particularly in BALB/c, followed by C.B17 (Table I). In contrast with BALB/c or C.B17 strains, the AGcA reactivity was below detection limits in C57BL/6 mice, both in LPS-stimulated B cell subset supernatant (data not shown) and in serum (Table I). However, generation of AGcA MZ B cells and autoantibody production is not blocked in C57BL/6 mice. This was shown in an experiment backcrossing EP67 μκ Tg mice, originally on a C.B17 background, for four generations to C57BL/6, whereupon similar AGcA MZ B cell generation and autoantibody production occurred (Supplemental Fig. 1B). Thus, generation of AGcA-autoreactive B cells can occur if the germline BCR repertoire is appropriate. In conclusion, AGcA belongs to the class of strain-biased natural autoantibodies that is produced by MZ B cells.
Function of natural AGcA autoantibody
The presence of natural IgM autoantibody has been shown to play a protective role (5–7). However, it also carries a risk to initiate injury in conjunction with complement (31). Because autoreactivity to goblet cell granules was originally found in children with colitis (17), the question naturally arises whether the presence of AGcA, normally produced as a natural autoantibody, carries a risk for initiating disease. Although AGcA B cells preferentially differentiate to become MZ B cells in the spleen environment, some B cells with AGcA BCRs circulate and are found in intestinal tissue, where the Muc2 autoantigen is present. In adult EP67 μκTg mice, without competing non-Tg B cells, a significant fraction of AGcA B cells become FO B cells in addition to MZ B cells (Fig. 2B). These B cells circulate to form follicles in both peripheral and mesenteric lymph nodes, and AGcA B cells are also found in intestinal tissue, including the colon, forming isolated lymphoid follicles, located nearby epithelial goblet cells that contain the target Ag (Fig. 6A, upper panel). AGcA (IgMa) PCs also appear in the intestinal lamina propria with abundance of AGcA IgM in serum. However, the secreted multimeric IgM does not permeate intact epithelial walls to reach the goblet cells. This is clear from a lack of staining of goblet cell sections by anti-IgMa reagent, unless goblet cells exposed by sectioning are first incubated with AGcA-containing autoserum from an EP67 mouse, as shown in Fig. 6A (lower panel). Thus, even abnormal excessive production of natural AGcA IgM autoantibody in the EP67 Tg mice does not result in goblet cell loss, and mucosal pathology is not enhanced over a 15-mo period (data not shown).
Natural AGcA autoantibody in intestine. (A) The presence of AGcA B cells (IgMa+) in isolated lymphoid follicles in EP67 mouse colon, and absence of goblet cell prebound AGcA IgM (upper panel), unless the tissue section is first incubated with serum from EP67 μκTg mice containing AGcA autoantibody (lower panel). (B) Analysis of DSS-treated (8 d after ceasing DSS treatment) EP67 and 3369 mouse intestine. AGcA (but not ATA) plasma cells are increased in intestinal lamina propria (selected area, enlarged at right). (C) DSS-treated EP67 colon section staining (8 d after ceasing DSS treatment). AGcA B cell follicle (iLF, right side of the IgMa-stained figure) and AGcA PC infiltration into the damaged region, in comparison with the PNA+ undamaged goblet cell area. (A) and (C) using a 20× objective lens. (B) using a 10× and 20× objective lens.
To further assess how AGcA plays in mucosal immunity/disease, we treated EP67 Tg mice, in comparison with 3369 Tg mice, with an oral administration of DSS. DSS is toxic to intestinal epithelial cells, causing an epithelial barrier integrity defect. Body-weight decrease occurred 3 d after a 5-d DSS treatment was discontinued, followed by recovery after 3 more days, in both mouse lines. Three to 8 d after ceasing DSS treatment, whereas 3369 mice normally produce natural ATA as the predominant IgM in serum (20), there was a decrease in IgM normally found in the intestine of DSS-treated 3369 mice, including ATA IgMa PCs (Fig. 6B, lower panel). In contrast, DSS-treated EP67.CB17 mice showed microvascular dilation with increased red blood infiltration throughout the intestine as occurred during ongoing DSS treatment, together with an increase in AGcA PCs in the lamina propria (Fig. 6B, upper panel). In the colon, there was no sign of accelerated colitis; rather, we observed that most goblet and epithelial cells regenerated 8 d after ceasing DSS treatment. AGcA PCs surround the degenerated epithelial area, in contrast with the intact area in EP67 mouse colon (Fig. 6C, left side versus upper right side), suggesting clearance of damaged goblet cells. Surveying for an additional 2 wk showed that chronic ulcerative colitis (UC) was not induced in DSS-treated EP67.CB17 mice (data not shown). In this DSS colitis model, strain differences are well-known, such that chronic colitis occurs more often in C57BL/6 than BALB/c backgrounds (32). The EP67 Tg BCR on BALB/c IgM allotype-congenic C.B17 mice had abundant serum AGcA and also showed similar resistance to chronic colitis development as BALB/c mice.
Discussion
We show in this study that B cells with AGcA become MZ B cells in spleen and produce natural autoantibody in serum. This AGcA binds predominantly to intact large Muc2 glycoprotein. In the same mouse, expression of a BCR with anti-thymocyte autoreactivity results in B1 B cell generation. Both AGcA and ATA react with highly glycosylated proteins, specific to murine determinants, showing no binding to corresponding rat glycoproteins (27). Accumulation of AGcA MZ B cells occurs from the transitional immature stage and is dependent on Btk. Therefore, a BCR-ligand signal is required for AGcA MZ B cell development, as with ATA B1 B cells where again functional Btk is critical (9). How immature AGcA B cells are exposed to Ag, thereby generating MZ B cells and not B1 B cells, remains to be determined. However, in AGcA MZ B cell generation, DCs and blood vessels in spleen seem likely to play roles in presenting self-antigen, followed by the Notch 2 signal that is key in MZ B cell development (33). Goblet cells continually secrete Muc2 into the intestinal mucosa, providing a mucosal barrier against bacterial invasion. As with bacteria in the mucosal lumen, DC dendrites also continually internalize Muc2 from mucus (14, 34). If these DCs move into the lymphatic vessels and enter blood circulation, then transport of Muc2 to the spleen can occur. This may explain the early initiation of AGcA MZ B cell development in spleen, filtering the blood, promoted by the presence of Muc2 in neonates as well as adults. Thus, differences in BCR signaling because of variation in affinity (21) and/or the site of Ag availability may generate MZ rather than B1 B cells.
Our data also demonstrate genetic influences on AGcA B cell generation. Variation of MZ B–associated natural AGcA autoantibody levels between mouse strains resembles B1 B cell–associated natural autoantibody, T15 idiotype positive anti-phosphorylcholine natural autoantibody (T15id+ anti-PC) that is higher in BALB/c compared with C57BL/6 (35, 36). However, generation of T15id+ anti-PC B cells does occur in C57BL/6 mice, although at a much lower frequency (37). This IgM is one of the most studied natural autoantibodies in mice. Binding to apoptotic cells leads to the initial generation of T15id+ anti-PC B cells, with anti-PC autoantibody production, and cross-reactivity of this Ab to oxidized low-density lipoprotein helps to prevent atherosclerosis (38). Furthermore, reactivity to phosphorylcholine expressed by various microorganisms plays a protective role against pathogens (35). We show in this study that MZ B–associated AGcA is also higher in BALB/c than C57BL/6, again demonstrating the importance of genetic background in generation of the natural autoantibody repertoire. One possibility is that the VH or VL segment required for the AGcA BCR is different enough in C57BL/6 mice to reduce its affinity below that required for selection into the MZ B cell pool. Alternatively, the frequency of B cells with an appropriate AGcA BCR is lower in C57BL/6 mice, as found with T15id+ anti-PC B cells.
Understanding the presence of natural autoantibodies in vertebrates, their roles, and genetic influences have been a subject of continuing study. As with B1 B cell–derived natural autoantibodies, the production of AGcA IgM may normally function as a scavenger natural autoantibody, as suggested by experiments with DSS-treated mice. Because cells of the intestinal tract are constantly sloughing off and renewed by intestinal stem cells, the presence of AGcA IgM in animals with appropriate genetic background may help to promote clearance of damaged cells. When B cells producing natural AGcA switch to IgA in intestine, the secreted Ig will reach the mucosal lumen and bind to Muc2. This may enhance the physical barrier provided by Muc2 and aid in preventing microbial invasion, an interesting possibility to pursue. Considering the higher AGcA titer in BALB/c mice, the presence of AGcA autoreactive B cells may moderate or prevent development of chronic colitis, potentially acting as a regulatory B cell. In addition, AGcA may have cross-reactivity to infectious organisms, thereby playing a protective role as part of the MZ B cell subset. However, in turn, when aberrant Ag exposure or autoreactive T cell generation induces Ig class switching of high-affinity AGcA B cells, their presence has the potential to result in goblet cell loss by required ability to permeate epithelium, leading to pathogenicity.
In human pathology, the relevance of anti-goblet cell autoantibody to UC has been an issue for decades, as a potential disease-inducing autoantibody (17). Sera from children with UC contain Ab that reacts with fetal colon cells (17, 18), and anti-Muc2 reactivity in UC patient serum has been demonstrated (39). The presence of anti-goblet cell autoantibody among first-degree relatives of UC patients without obvious sign of disease has been interpreted as indicating a genetic risk for development of UC (40). However, a role for AGcA in enteropathy has not been clearly established; rather, AGcA is often detected in the serum of healthy individuals (41). As demonstrated in this study with mice, anti-goblet cell autoantibodies in relatives of UC patients may in part be composed of natural Igs at genetically regulated levels, as found for BALB/c versus C57BL/6 strains. The mouse model we describe enables such further study of the significance of natural autoantibodies, testing whether it functions to suppress inflammatory disease, but carrying a risk for promoting disease. In addition, in considering reactivity to the highly glycosylated Muc2 molecule, potential cross-reactivity to pathogenic bacteria or viruses to enable defensive immunity by MZ B cells is another area that merits further investigation.
Disclosures
The authors have no financial conflicts of interest.
Acknowledgments
We thank Gunner C. Hansson for advice on analysis of Muc2, Sandra Gendler for Muc1-deficient mice, Li-Jun Wen for VH3609t mouse generation, Anthony Yeung for advice on colon crypt preparation, Catherine Renner for help with histologic analysis, Yue-Sheng Li for supervising single-cell sequencing, and Kerry Campbell for discussion and comments on the manuscript.
Footnotes
This work was supported by National Institutes of Health/National Cancer Institute Grants AI49335 (to K.H.), AI026782 (to R.R.H.), and CA129330 (to K.H.), and the Fox Chase Cancer Center Blood Cell Development and Cancer Keystone Program.
The online version of this article contains supplemental material.
Abbreviations used in this article:
- AGcA
- anti–goblet cell/mucin2 autoreactive
- ATA
- anti-thymocyte/Thy-1 glycoprotein autoreactive
- BCR
- B cell AgR
- BM
- bone marrow
- DBA
- Dolichos biflorus agglutinin
- DC
- dendritic cell
- DSP
- dithiobis succinimidyl propionate
- DSS
- dextran sulfate sodium
- EP67 μκTg
- VH3609/Vk19-17 μκ Tg
- FO
- follicular
- 13H8k
- anti-κ Ab recognizing limited set of κ L chains including Vk21-5 and Vk19-17
- IB4 (BSI-B4)
- isolectin B4
- MK19
- IgM produced by SP2/0 cotransfected with VH3609μ IgH and Vk19-17κ IgL
- MK21
- IgM produced by SP2/0 cotransfected with VH3609μ IgH and Vk21-5κ IgL
- Muc2
- mucin 2
- MZ
- marginal zone
- PAS
- periodic acid–Schiff
- PC
- plasma cell
- PNA
- peanut agglutinin
- RT
- reverse transcription
- Tg
- transgenic
- UC
- ulcerative colitis
- VH3609id
- anti-VH3609SM6C10 idiotype Ab
- VH3609t
- VH3609/D/JH2 targeted insertion (knock-in) mouse line
- WGA
- wheat germ agglutinin.
- Received September 18, 2014.
- Accepted November 5, 2014.
- Copyright © 2015 by The American Association of Immunologists, Inc.
This article is distributed under The American Association of Immunologists, Inc., Reuse Terms and Conditions for Author Choice articles.