Site-specific labeling uncovers differences in levels and distribution of B-cell receptors of different isotypes on primary B cells

Abstract

Expression of the BCR is essential for survival, development, and effector functions of B cells. Naive B cells express surface IgM and IgD, while surface IgG1 is expressed by class-switched (memory) B cells. Despite similar overall structures, the different BCR isotypes show differences in distribution and expression levels. The dynamics of BCR behavior have been difficult to explore owing to a lack of appropriate tools that can track the BCR without causing concomitant activation. Using CRISPR-Cas9, we inserted a sortase recognition motif (LPETG [LeuProGluThrGly]) at the C-terminus of the OB1 transnuclear ovalbumin-specific Cκ chain (Igκ-LPETG mice). The surface BCR from Igκ-LPETG mice is fully functional and can be labeled site-specifically with biotin or fluorophores. Igκ-LPETG mice show near-normal B-cell development, with an increase in Igλ-producing cells, presumably due to massive contraction of the κ locus V-region cluster upon V-J recombination to generate the OB1 light chain. Using the Igκ-LPETG mice, we compared organization and density of BCRs on the surface of IgM/IgD⁺ B cells bearing a wild-type (WT) heavy chain locus and IgG1 B cells in the OB1 model. The density of IgG1 BCRs is much reduced compared to IgM/IgD BCRs on primary B cells. Upon activation, IgM/IgD BCRs are found in detergent-insoluble domains, whereas IgG1 BCRs are not. The isotype of the Ig heavy chain thus contributes to surface expression and nanoscale organization of the BCR.

Introduction

The immune system can recognize and respond to a near infinite number of Ags. The BCR recognizes Ags and contributes to a specific and durable defense against a wide range of pathogens.¹ The BCR is a complex of proteins that consists of membrane immunoglobulin (mIg), composed of 2 identical heavy (HC) and light (LC) chains, each with a variable (V) and a constant (C) region, together with the Igα (CD79a) and Igβ (CD79b) accessory proteins.² The latter contain ITAMs required for signal transduction.^3–5 The variable portion of mIg is responsible for Ag binding, while the C regions determine the BCR isotype and effector functions of their secreted counterparts.³ Naive mature B cells express 2 Ig isotypes on their surface: IgM and IgD. In response to Ag and costimulatory signals, B cells engage in class-switch recombination to access other isotypes (IgG, IgA, or IgE) with retention of the HC’s V region sequence. Ig LCs are not altered in the process. It has been suggested that the isotype of the mIg component of the BCR in addition to its V region influence Ag specificity and affinity.⁶ The C regions of the mIg isotypes may differentially regulate B-cell Ag-specific signaling through distinct interactions with the extracellular domain of Igα/Igβ. For example, IgG-Cγ3 and IgM-Cμ4 show a head-to-tail and side-by-side interaction with Igα/Igβ, respectively.⁷

Upon binding of Ag to the variable portion of the BCR, B cells launch a cascade of intracellular signals⁸ essential for their proliferation. BCR engagement also leads to internalization, processing, and presentation of the captured Ag by Class II MHC molecules. The display of peptide-Class II MHC molecules is essential for activation of T cells that provide help in class switching and in the differentiation of B cells into antibody-secreting cells.⁹ BCR-triggered events control a series of cell-fate decisions at several checkpoints during B-cell development.¹ Therefore, BCR engagement with Ag must be tuned to respond appropriately to execute each of these critical functions: (i) survival upon receipt of “tonic” signals, (ii) production of antibodies/presentation of Ag to T cells upon activation, and (iii) terminal differentiation into a specific cell type (plasmablasts and plasma cells) for secretion of Ig. The molecular events that govern fate decisions in the periphery are not well understood. Likewise, the detailed molecular features that allow signal transduction upon BCR engagement remain to be clarified. A growing body of evidence points to changes in nanoscale organization of the BCR as critical for activation, survival, and fate decisions,^10,11 to which differences in isotype might well contribute.

Visualization of surface receptors can provide new insights into their function. Direct labeling of the BCR without concomitant activation remains a challenge for live cell imaging experiments. Current approaches to visualize the BCR rely on the use of either fluorescently labeled antibody reagents or affibodies directed to the BCR.^12,13 Alternatively, they involve the use of fluorescently labeled Ags.¹⁴ Neither approach covalently tags the BCR. The use of fluorescently labeled antibodies and Ag can activate B cells and thus preclude an accurate assessment of BCR organization prior to and upon Ag binding. Other approaches rely on genetic fusions of BCR-accessory proteins with bulky fluorescent proteins (∼25 kDa for Green Fluorescent Protein), which may affect BCR organization¹⁵ and behavior. Therefore, there is a need to develop a strategy to label BCRs on live primary B cells without activating them.

Here we introduce a generalizable method to achieve covalent modification of the BCR in a minimally perturbing manner. We exploited CRISPR-Cas9 to engineer a sortase A (SrtA) recognition motif (LPETG [LeuProGluThrGly]) into the rearranged κ locus of OB1 mice, which bear an IgG1 BCR specific for ovalbumin.^16,17 These mice were generated by somatic cell nuclear transfer and contain rearranged IgH and κ loci at their endogenous locations. The OB1 mice thus retain the capacity for secondary κ and λ LC rearrangements, as well as class-switch recombination for the IgH locus and somatic hypermutation of the rearranged IgH and LC loci.

The thus-modified OB1 κ locus encodes a LC that can be modified enzymatically and site-specifically with any small molecule substituent of choice, including biotin and fluorophores. BCRs on B cells derived from Igκ-LPETG mice are fully functional after labeling with SrtA, as evident from downstream signaling events upon Ag binding. Using this strategy, we compared the organization of BCRs on the surface of B cells of different isotypes. We observed reduced levels of surface IgG1 BCRs compared to those of IgM/IgD BCRs, notwithstanding the fact that transcriptional control elements do not obviously differ for these isotypes. In addition, we showed that IgG1 BCRs were present in detergent-soluble domains at rest and upon activation. In contrast, IgM/IgD BCRs migrate to the detergent-insoluble fraction upon activation. The Igκ-LPETG LC is an example of how site-specific labeling using sortase can be applied to study the dynamics and organization of surface receptors.

Materials and methods

Mice

Transnuclear OB1 RAG-proficient (C57BL/6 background)¹⁷ κ-LPETG mice (mixed OB1 LC C57BL/6 and 129 background) were generated as described,¹⁶ and OB1 HC κ-LPETG were obtained by crossing κ-LPETG mice with OB1 RAG-proficient mice. All animals were maintained under specific pathogen–free conditions at the Whitehead Institute and the Boston Children’s hospital, and experiments were performed with approval of the Committee for Animal Care and the Association for the Assessment and Accreditation of Laboratory Animal Care. C57BL/6 were purchased from Jackson Laboratory and maintained at our facilities (Whitehead Institute and Boston Children’s hospital). All experiments were performed on 6-to-12-week-old mice.

Reagents, antibodies, and flow cytometry

Reagents and antibodies were obtained from Thermo Fisher (β-actin, clone BA3R; streptavidin-HRP, ref. 21130), Southern Biotech (αIg, ref. 1010-01; αIgλ, ref. 1060-05; αIgM, ref. 1021-01; αIgκ, ref. 1050-05 and ref. 1170-09L; αIgλ, ref. 1060-05 and ref. 1175-02), BioLegend (CD23, B3B4; CD21, 7E9; CD43, S11;), BD (CD19, 1D3; B220, RA3-6B2; CD16/CD32, 2.4G2; αIgλ, ref. 744523). Cell viability dye 7-AAD (Via-Probe) was purchased from BD Biosciences and used following manufacturer’s instructions. Flow cytometry data were acquired on an LSRFortessa or LSR II (BD) instrument and analyzed with the FlowJo software package (Tree Star). The imaging flow cytometry data were acquired on an Amnis ImageStreamX Mk II with 60×/0.9NA objective and analyzed with IDEAS software (v6.2). Surface markers were stained at 4 °C for 25 min in the presence of Fc block (BD) in PBS + 1 mM EDTA + 0.5% bovine serum albumin. Indo-1 AM was purchased from Invitrogen. Anti-IgM and anti-IgG1 (unlabeled and conjugated to HRP) were obtained from Southern Biotech.

Sortase labeling

Sortagging reactions were performed using the hepta-mutant version of Sortase A (SrtA) from Staphylococcus aureus, which is calcium independent and exhibits an enhanced activity.^18,19 The enzyme was produced as described elsewhere.²⁰ SrtA probes (GGG-biotin, GGG-Alexa-488, GGG-Alexa-647, and GGG-Cy5) were produced in house as described.²¹ Three to 5 × 10⁶ B cells were labeled with 60 μM of SrtA and 600 μM of probe in complete RPMI on ice for 1.5 h or the indicated time. Labeled cells were washed 3 times with cold PBS. For immunoglobulin labeling, serums from the indicated mice were collected and used in a SrtA reaction as described.¹⁷ Immunoglobulins were then isolated using protein-G beads and the beads containing the labeled immunoglobulins boiled in SDS sample buffer to release the bound material. The sample was then submitted to SDS-PAGE, blotted using streptavidin-HRP.

B-cell purification

Untouched B cells were purified by negative selection using CD43 Dynabeads (Life Technologies) according to the manufacturer’s recommendations (purity was always >90%).

RNA-sequencing library preparation and sequencing

One thousand CD19⁺Igκ⁺Igλ^- cells were sorted using a FACS Aria II cell sorter and lysed in a guanidine thiocyanate buffer (Qiagen) supplemented with 1% β-mercaptoethanol. RNA was isolated using solid-phase reversible immobilization bead cleanup, reverse-transcribed into complementary DNA and pre-amplified as previously described.^22–24 Sequencing was performed on an IlluminaHiSeq2000 at the genome technology core of the Whitehead Institute. Vκ gene segments were assigned using IMGT.²⁵

Ca²⁺ flux

Naive primary B cells were resuspended in serum-free RPMI and labeled with Indo-1 AM for 30 min at 37 °C, following manufacturer’s instructions. The cells were then washed and left at room temperature (RT) for 20 min to enable desterification of the dye. Cells were equilibrated at 37 °C for 1 to 2 min before flow cytometry analysis. Signals at 485 nm (HI/Ca²⁺ bound) and 410 nm (LO/Ca²⁺ free) were collected for ∼1 min to record baseline Ca²⁺ levels (ratio of Indo-1 LO/HI). The cells were then stimulated with the indicated reagents and Ca²⁺ levels recorded for 7-min total time. Data were collected in a FACS LSR II instrument (BD) and analyzed using FlowJo software (Tree Star).

Serum ELISA

Serum was collected from age-matched mice. Ninety-six–well high-binding plates (Costar) were coated with 1 μg/ml of anti-Ig (Southern Biotech) in PBS overnight at 4 °C, rinsed 3× with PBST (0.05%Tween-20 in PBS) and blocked in 10% FBS in PBS for 1 h at RT. Serum were incubated neat and in 10-fold dilutions for 2 h in 10% FBS in PBS. The plates were rinsed 5× with PBS and incubated with HRP-coupled secondary antibodies recognizing Igκ and Igλ (1:1,000, Southern Biotech) for 1 h. After rinsing, the plates were developed using OptEIA TMB substrate reagent kit (BD), the reaction was stopped with 1 M hydrochloric acid, and the plates were read at 450-nm absorbance.

Metabolic labeling and immunoprecipitation

B cells were cultured in complete RPMI with 10% FBS, and LPS (20 μg/mL) was added to the culture medium. Plasmablasts were starved for 1 h in methionine- and cysteine-free medium and pulsed for 10 min with [³⁵S] methionine/cysteine (PerkinElmer). Supernatants were harvested, and the cells lysed in Nonidet P-40 buffer (1% NP-40 in HEPES buffer pH 6.8 containing protease inhibitors [Roche]). B-cell receptors and immunoglobulins were immunoprecipitated as described.²⁶

Immunoblotting

For p-ERK, B cells were rested overnight at 37 °C in complete RPMI. The next day, the cells were equilibrated at 37 °C before adding the stimuli. Cells were lysed using RIPA buffer (20 mM Tris pH 7.5, 1 mM EDTA, 100 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, and 0.1% SDS), and lysates analyzed by SDS-PAGE and immunoblotting using α-P-ERK, α-total ERK (Cell Signalling Technologies), α-β-actin-HRP, and streptavidin-HRP (Thermo Fisher). Blots were imaged on a ChemiDoc MP (Bio-Rad). For lipid raft analysis, after sortase labeling, cells were lysed on ice for 30 min in buffer (20 mM MES, 100 mM NaCl, 30 mM Tris-HCl pH 7.5) + 1% Triton X-100 or 100 mM β n-octyl glucoside (Sigma-Aldrich) and HALT protease inhibitors (Thermo Fisher). Insoluble fractions were isolated by centrifugation at 20,000 × g for 10 min at 4 °C. Supernatant was transferred to a fresh tube (soluble fraction) while the pellet (insoluble fraction) was resuspended in MNT (20 mM MES, 100 mM NaCl, 30 mM Tris-HCl pH 7.5) + 2% SDS and heated at 95 °C for 10 min. Samples were passaged through a U-100 insulin needle (BD Biosciences) to shear any DNA and then processed for immunoblotting as above.

Statistical analysis

Mean and standard deviation (SD) values were calculated with GraphPad Prism (GraphPad Software). Unpaired Student’s t test was used to compare 2 variables, while one-way ANOVA was used for multiple comparisons. P values of <0.05 were considered significant and are indicated in the figure or figure legends.

Results

Site-specific labeling of the B-cell receptor using Sortase A

Using somatic cell nuclear transfer, we generated a line of transnuclear mice that have B cells of defined specificity.¹⁶ The OB1 line produces an IgG1 that recognizes ovalbumin, the precise epitope of which has been identified.²⁷ To create a model that allows labeling and engagement of the BCR in a manner that would avoid its activation, we used CRISPR-Cas9 gene editing to generate germline-modified OB1 mice bearing a modified κ locus that contains an LPETG motif, preceded by a 6-residue linker, at the C-terminus of the Cκ domain (Fig. 1A). The OB1 κ chain rearrangement involves Vκ1-135, which is located at the 5′ end of the Vκ cluster, with only a single functional Vκ gene, Vκ 2-137, remaining 5′ of Vκ1-135. This rearrangement produced a massively contracted Igκ locus. Regardless of the Vκ segment used, Vκ1-135 or Vκ 2-137, the resulting κ chain and the BCRs in which it participates should be suitable sortase (SrtA) substrates, as they would both use the LPETG-tagged Cκ exon.

Sortase A (SrtA) is a transpeptidase that allows the attachment to the LPXTG SrtA recognition motif of any substituent bearing an NH₂-GlyGlyGly (GGG) at its NH₂-terminus (Fig. 1B). A molecular model of a full-sized IgG1 molecule shows that the C-terminus of the κ–LC is exposed²⁸ as confirmed by the recent cryo-EM structure of the BCR.⁷ By extending the κ C-terminus with an LPETG motif, we hypothesized that the SrtA recognition site would be accessible to exogenously added SrtA. We incubated serum antibodies isolated from Igκ-LPETG mice with recombinant SrtA in the presence or absence of a GGG-biotin nucleophile. Successful modification results in the appendage of the GGG-biotin moiety to the C-terminus of the modified κ LC. No purification of Ig from serum is required to achieve site-specific labeling. When using serum isolated from wild-type (WT) mice, we observed no labeling of κ chains under any condition. Immunoglobulins from animals that carry the modified OB1 κ LCs showed labeling only when both SrtA and the GGG-biotin nucleophile were included. Detection of the biotinylated κ LC with streptavidin-HRP conjugate on immunoblots yielded a single labeled polypeptide of the appropriate molecular weight (Fig. 2A).

Because mammalian cells do not possess naturally occurring surface proteins with an exposed LPXTG motif,²⁰ we expected that B cells from WT mice would be refractory to SrtA labeling unless they carry the κ-LPETG motif. Indeed, labeling at 0 °C of the Igκ-LPETG BCR on naive B cells occurs within minutes and is specific (Fig. 2B and C). A plateau is reached after ∼30 min of incubation (Fig. 2C). In addition to GGG-biotin, Igκ-LPETG BCRs were readily labeled with other substituents, including GGG-Alexa-647 and GGG-Cy5 (Fig. 2D). Primary B cells from Igκ-LPETG can thus be efficiently modified with a diverse set of substituents.

Preferential association of the OB1 κ-chain with the OB1 IgG1 heavy chain

The VJ rearrangement in the OB1 mice involves Vκ1-135, a Vκ gene near the very 5′ end of the locus, with only a single functional Vκ segment, Vk2-137, remaining at the 5′ end of the Vκ cluster. All the intervening Vκ gene segments have been deleted in the course of rearrangement (Fig. 3A and B). The Jκ segment used is Jκ1, with 3 additional 3′ J segments remaining, an organization that allows secondary rearrangements involving the single remaining Vκ2-137.²⁹ Breeding experiments in which we placed the modified κ locus on a WT IgH background showed that the 5′-most Vκ segment is indeed functional and can be accessed by a secondary Vκ rearrangement, as confirmed by DNA sequence analysis (see below). Keeping in mind that allelic exclusion at the κ locus is imperfect,²⁹ such secondary rearrangement places the Vκ2-137 segment in a functionally correct arrangement with the modified Cκ exon and thus yields a LC that is also a SrtA substrate. With the modified κ locus on a WT IgH background, B cells produce a mixture of the OB1 κ LC and a distinct κ chain that is derived from a secondary rearrangement, as shown by blotting for SrtA-labeled biotinylated κ LCs (Fig. 3C). The migration on SDS-PAGE of the Vκ2-137-bearing LC is clearly distinct from that of the OB1 LC. The LC from Igκ-LPETG mice runs as 2 sharp distinct bands. These represent the Vκ1-135 and Vκ2-137-bearing LCs compared to the polyclonal LC band present in WT mice which is more diffuse (Fig. 3F). In contrast, if the modified κ locus is placed on the OB1 HC background, the secondarily rearranged κ is no longer detected (Fig. 3C), indicative of preferential association of the OB1 IgG1 HC with the OB1 κ chain. A single copy of the OB1 HC locus suffices, as allelic exclusion at the HC locus ensures its preferential expression. Deep sequencing of the Vκ repertoire indicates that 5% to 10% of all Vκ sequences in mice with and without the OB1 HC use the Vκ2-137 segment (Fig. 3D and E). This suggests that the difference in surface disposition of the distinct V-bearing BCRs is controlled posttranscriptionally.

Metabolic labeling experiments in which we compared IgM synthesis in plasmablasts from WT C57BL/6J and Igκ-LPETG mice show comparable levels of IgM secretion, but Igκ-LPETG mice show somewhat decreased levels of surface IgM (Fig. 3F). This is likely due to differences in the ability of the polyclonal HC repertoire to efficiently pair with the 2 available κ LCs.

B-cell development in OB1 κ-LPETG mice shows an aberrant κ to λ ratio

We performed flow cytometry to characterize B-cell development in Igκ-LPETG mice. Compared to WT mice, Igκ-LPETG mice displayed an increase in the frequency of λ-producing cells in the spleen as well as a decreased κ to λ ratio for serum immunoglobulins (Fig. 4A–E). In WT mice, fewer than 5% of mature B cells are Igκ⁺. However, Igκ LPETG^+/+ mice have 20% to 40% of Igλ⁺ B cells in the spleen. Interestingly, we identified a population of splenic Igκ⁺Igλ⁺ B cells in Igκ-LPETG mice (Fig. 4E) and confirmed their presence with spectrally divergent dyes by imaging flow cytometry (Fig. 4F). In the bone marrow, the B-cell compartment derived from Igκ-LPETG mice shows a slight increase in pre-B cells compared to WT cells, although not significantly different from OB1 LC^+/+ (Fig. 4G). There is a slight decrease in splenic follicular B cells when compared to WT cells but not in comparison with OB1 LC^+/+. We conclude that apart from the decreased κ to λ ratio, the B-cell compartment shows no major aberrations. Presumably not all newly generated VDJ heavy chain combinations pair equally well with Vκ1-135 or Vκ2-137, thus favoring outgrowth of B cells with λ light chains produced by continued light chain rearrangements.

Labeling of the κ-LPETG chain does not activate B cells

We examined whether the enzymatic conjugation of a biotin label to the C-terminus of the κ LC would trigger signaling via the BCR. We took advantage of the fact that the SrtA variant used for labeling retains enzymatic activity at 0 °C.³⁰ We could thus perform labeling experiments at 0 °C, remove the enzyme together with the added excess nucleophile, and return cells to 37 °C, followed by an interrogation of ERK phosphorylation as an indicator of B-cell activation. For reference purposes, we incubated cells with a polyclonal α-IgM/G to cross-link the BCR and trigger subsequent phosphorylation of ERK. We observed biotinylation only for B cells from mice that carry the modified Igκ-LPETG chain and only when both enzyme and GGG nucleophile were present (Fig. 5A). The levels of phospho-ERK were comparable for B cells left untreated and B cells labeled using SrtA and the GGG-biotin nucleophile. The increase in phospho-ERK required the inclusion of a cross-linking α-IgM and was similar for WT and Igκ-LPETG^+/+ B cells, regardless of whether the κ chains had been enzymatically biotinylated or not. An examination of Ca²⁺ flux yielded similar results (Fig. 5B). Cross-linking of the BCR with α-IgM/G produced a significant Ca²⁺ flux, regardless of labeling status. The presence of the LPETG motif and the attached biotin itself produced no Ca²⁺ flux. B cells of Igκ-LPETG mice are therefore functional with respect to BCR signaling. We conclude that the SrtA motif present on the κ LC does not obviously affect activation of B cells, nor does SrtA labeling promote or inhibit B-cell activation.

IgM/IgD and IgG1 BCRs are organized differently on the surface of B cells

BCR density and nanoscale organization influence B-cell fate.^31–33 While expression of IgG1 BCR supports development of B cells in most models, some subsets of B cells can be adversely affected, suggesting that the presence of an IgG1 instead of an IgM/IgD BCR could impact B-cell fate.^16,34 Density and organization of the BCR differ between IgG1-carrying B cells and those with IgM/IgD.^27,35,36 Owing to differences in the cytoplasmic tail of the IgM/D and IgG1 BCR, we hypothesized that the differences observed in distribution could be due to association of BCRs with detergent-resistant membranes (DRMs) and their associated proteins, also referred to as lipid rafts or microdomains. Other studies have identified IgD BCRs in DRMs under resting conditions, while IgM BCRs move to DRMs upon activation.³⁷ The role of DRMs in BCRs of the IgG1 class remains unknown.¹⁰ B cells from Igκ-LPETG^+/+ and OB1 Igκ-LPETG^+/+ mice were incubated with SrtA and GGG-biotin to biotinylate surface-exposed BCRs. Cells were then lysed with buffer containing either β-n-octylglucoside (OG) or Triton X-100 (TX100). OG should fully solubilize proteins sequestered in DRMs while TX100 lysis keeps DRMs intact.³⁸ Blots were then probed with streptavidin-HRP to examine distribution of surface-exposed BCRs (Fig. 6A and B). As expected, some IgM/D BCRs were present in the TX100 insoluble fraction yet solubilized by treatment with OG. In contrast, IgG1 BCRs were fully solubilized by TX100. To determine whether BCR localization changes upon activation, we stimulated B cells with α-IgM/G before detergent extraction to detect possible changes in organization of the BCR upon activation (Fig. 6C and D). After activation, a substantial portion of IgM/D BCR is present in the TX100 insoluble fraction while IgG1 BCRs remain in the TX100 soluble fraction. Overall, our results suggest that IgM/IgD BCRs find themselves in a distinct lipid environment upon activation, in contrast to IgG1 BCRs, which do not obviously change their distribution upon activation.

Isotype-dependent differences in BCR expression at the cell surface

Next, we tested whether the surface expression of BCRs was altered in B cells carrying different isotypes. We isolated splenic B cells from WT HC Igκ-LPETG^+/+ and OB1 HC Igκ-LPETG^+/+ and stained them for surface markers for flow cytometry. Confirming previously published results in WT mice,³⁹ IgG1-carrying splenic (primary) B cells derived from the monoclonal OB1 HC Igκ-LPETG^+/+ mice showed a reduction in Igκ median fluorescence intensity (MFI) compared to IgM/IgD⁺ naive WT splenic B cells (Fig. 7A and Fig. S1). Ig LCs are produced in molar excess over HCs.^40,41 This suggests that it is the level of the Ig HC that is the limiting factor. Alternatively, accessibility of the anti-Igκ to the BCR may be affected by differences in nanoscale organization between IgM/IgD and IgG1 BCRs. Quantitative PCR (qPCR) analysis of Vκ, μ_mem, δ_mem, and γ1_mem mRNA levels in WT HC Igκ-LPETG^+/+ and OB1 HC Igκ-LPETG^+/+ mice showed no significant difference in transcript levels for light chain mRNA or μ_mem and γ1_mem HC mRNA (Fig. 7B and C). δ_mem transcript levels make up such a small proportion of HC mRNA in Igκ-LPETG^+/+ cells that even when considered together with γ1mem transcripts, the total level would still be similar for Igκ-LPETG^+/+ and OB1 HC Igκ-LPETG^+/+ mice. As the identity of the control elements that drive transcription of the Ig heavy chain locus are preserved, but their relative distances are altered in the course of class-switch recombination, this is not entirely unexpected. Some form of posttranscriptional control must therefore be responsible for the differences in BCR levels when comparing IgM/D- and IgG1-expressing B cells (Fig. 7B and C). Using imaging flow cytometry, we similarly observed a small but consistent difference in Igκ levels between WT HC Igκ-LPETG^+/+ and OB1 HC Igκ-LPETG^+/+ mice by comparison of fluorescence intensity of 3500 cells (Fig. 7D and E). Interestingly, the WT HC Igκ-LPETG^+/+ BCR showed a bimodal distribution during SrtA/GGG-Cy5 labeling as compared to OB1 HC Igκ-LPETG^+/+ BCR (Fig. 7F and G and Fig. S2). Further studies will be required to assess whether such differences are attributable to variations in accessibility to sortase and/or density of Igκ-LPETG between the IgM and IgD subtypes, changes in the organizations of the respective BCRs, or to—as yet unidentified—B-cell subsets.

A Western blot performed on SrtA/GGG-biotin–labeled B cells showed that total (intracellular and surface) Igκ levels were higher in OB1 HC Igκ-LPETG^+/+ compared to WT HC Igκ-LPETG^+/+ cells (Fig. 7H and Fig. S3). Despite this difference, there was more sortase labeling of surface BCRs in WT HC Igκ-LPETG^+/+ cells, indicating that the quantity and/or accessibility of surface-disposed Igκ differs between the 2 genotypes. The SrtA labeling disparity may be associated with the structural difference between IgM and IgG1 BCRs, as the IgM-Cμ2-Cμ4 could enhance SrtA accessibility of κ-LPETG in comparison to IgG-Cγ2-Cγ3 (Fig. 1A). Given structural similarities between Cγ and Cδ this may explain the bimodal labelling observed (Fig. 7G).

Discussion

We developed a model that enables site-specific labeling of the BCR on naive B cells without obvious signs of activation, as judged by the absence of downstream signaling events (pERK, Ca²⁺ flux) upon covalent modification of κ LCs in the BCR. The presence of the SrtA motif installed on Cκ allowed labeling with biotin or a fluorophore and did not affect B-cell activation or differentiation. Using this model, we found that IgM/IgD BCRs were partially located in detergent-insoluble domains at rest and migrated to detergent-insoluble domains upon activation, unlike IgG1 BCRs, which remained detergent soluble under all conditions.

The rearranged Ig loci of OB1 mice are in their natural configuration: no genetic manipulation was involved in their creation. Our labeling experiments show the preferential association of the OB1 κ LC, which carries Vκ1-135, with the OB1 γ1 HC. When the modified OB1 Ck locus is placed on a WT IgH background, we detect the presence of 2 labeled LC species, one of which is the Vκ1-135 OB1 κ chain. Because there is only a single functional Vκ, Vκ2-137, remaining in the rearranged OB1 Cκ locus, any secondary κ rearrangements necessarily involve Vκ2-137.

The differences in the levels of IgM/IgD versus IgG1 BCRs on the surface of B cells is unlikely to be due to differences in transcription of the κ locus, as it is identical in both settings. Retention of the Eμ enhancer as well as enhancer elements downstream of the Cα constant region, important for regulation of rearrangement and transcription,⁴² cannot explain the difference we observe. The levels of BCRs on the surface of B cells likely affect BCR function, owing to differences in cytoplasmic tails and/or receptor organization. We measured the expression levels on the surface of B cells by flow cytometry and found fewer IgG1 than IgM/IgD BCRs. Also, the κ-LPETG of the IgM/IgD BCR isotype was more amenable to SrtA labeling than the IgG1, suggesting a difference in Igκ-LPETG accessibility and/or density. Lower surface expression of IgG1 BCRs has also been seen in mice carrying polyclonal IgG1 B cells.³⁹ These polyclonal IgG1 mice show a ∼10-fold reduction in the number of peripheral B cells. Conversely, another IgG1 polyclonal model that enhances surface IgG1 expression (by deletion of the intronic polyA site to enhance membrane IgG1 over secreted IgG1) shows no obvious anomalies in B-cell development.³⁴ Monoclonal IgG1-bearing OB1 B cells show near-normal B-cell development.¹⁶ We observed a bimodal distribution of labeling intensity of Igκ-LPETG in the HC WT/Igκ-LPETG mice, with one population demonstrating a similar intensity to OB1 IgG-bearing cells. This may be the result of differences in accessibility for IgM/IgD-containing BCRs, driven by structural differences between HC isotypes. The number of constant region domains (4 vs. 3 for IgM vs. IgD) is one such factor. If so, this would suggest that subpopulations of naive B cells exist with varied IgM/IgD levels. This has been observed previously in circulating and splenic B cells.^43,44 Together, these findings underscore how BCR density and specificity are critical for B-cell fate decisions. The Igκ-LPETG mice presented here offer a flexible strategy to investigate the mechanistic relationship between Ag specificity and BCR organization, assuming that the fixed light chain is compatible with recognition of the Ag of interest.

With a few notable exceptions,^45–47 most of our knowledge concerning Ag-specific BCR in primary B cells pertains to IgM BCRs, one of the 2 isotypes present in resting B cells. However, IgM lacks a substantive cytoplasmic tail and therefore relies entirely on Igα/Igβ for signal transduction.⁴⁶ In contrast, all other isotypes (IgG1, IgG2a, IgG2b, IgG3, IgE, and IgA) possess longer, highly conserved, cytoplasmic tails implicated in downstream signaling.^34,48–50 Experiments that explore surface BCRs of isotypes different than IgM and of known specificity have mostly relied on in vitro class-switch recombination. This approach does not usually provide homogenous B-cell populations.⁵¹ The extracellular portion of the constant region is also structurally diverse for the different Ig isotypes. While the different extracellular regions of the different BCR isotypes possess a similar overall structure, the length and flexibility of their hinge regions are very different. For example, the hinge region of IgD restricts the response of this isotype to only polyvalent Ags, while IgM BCRs can recognize both monovalent and polyvalent Ags.⁴⁷ The length of the IgG1 cytoplasmic tail, and its ability to directly engage in signaling without additional factors, provides a possible explanation for its underrepresentation in detergent-insoluble domains, implicated in the organization of signaling complexes.⁵² Further work is required to address differences in the nanoscale organization of BCRs at the cell surface and the temporal changes therein upon Ag engagement. The B cells interrogated in this study were naive, and it is therefore possible that changes in BCR organization may be found in memory B cells as evidenced by their different biochemical responses to stimuli.^53,54 The Igκ-LPETG model provides an excellent tool with which to undertake such studies as it allows stochiometric labeling under native conditions and does not impede activation.

Acknowledgments

We thank the WIBR Flow Cytometry and genome core facilities for sorting cells, and analyzer use and for the sequencing, I. Barrassa (Department of Bioinformatics and Research Computing, WIBR) for assistance with RNA-seq analysis, Tom DiCesare with illustrations, and J. Jackson and S. Kolifrath for assistance with mouse husbandry genotyping and colony management. We are thankful to M. Hagiwara for help with pulse chases. We are in debt to G. Victora and L. Mesin for invaluable discussions and suggestions on this project and A. Avalos for initial guidance. We thank the Ploegh lab for suggestions and fruitful discussions.

Author contributions

D.B., E.M.O., H.L.P., and N.M. designed the research and wrote the paper. D.B., E.M.O., and N.M. performed the research and analyzed the data. Z.L., T.F., and R.W.C. contributed new reagents and provided guidance.

Supplementary material

Supplementary material is available at The Journal of Immunology online.

Funding

This work was supported by a grant from the National Institutes of Health (5R01AI087879-11) and D.B received support from Boehringer Ingelheim fund.

Conflicts of interest

None declared.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

References

Author notes