Main

IgM serves as the first line of defence against infections in humoral immunity. The predominant form of IgM in serum is a pentameric assembly (pIgM), with five IgM monomers joined together by the joining chain (J-chain). In the absence of the J-chain, hexameric IgM can be formed. The Fc tailpiece has a vital role in IgM polymer formation and J-chain incorporation. In addition to functioning in polymeric antibody forms, the monomeric membrane-bound IgM (mIgM) participates in the BCR complex, together with Igα (also known as CD79A) and Igβ (also known as CD79B), and has a key role in antigen recognition and B cell activation. A transmembrane segment is present in mIgM for membrane localization in place of the tailpiece.

Different immunoglobulin classes feature different Fc regions, which are recognized by specific Fc receptors to elicit distinct effector mechanisms. For example, polymeric immunoglobulin receptor (pIgR) selectively recognizes J-chain-containing pIgM6,7, as well as J-chain-containing dimeric, tetrameric or pentameric IgA8,9, and escorts them to the mucosal surface. The ectodomain of pIgR, known as the secretory component (SC), remains bound to pIgM, resulting in the formation of the secretory IgM (sIgM) complex10. Two other IgM Fc (Fcμ) receptors are present in humans in addition to pIgR: FcαμR and FcμR11. FcαμR is involved in the internalization of IgM-coated microorganisms12, and similar to pIgR, FcαμR binds both IgM and IgA. By contrast, FcμR binds IgM exclusively and is the only IgM-specific receptor in mammals.

Previously known as Toso or Faim3, FcμR was first identified as an inhibitor of Fas-mediated apoptosis in T cells13 before it was demonstrated to be the long-sought Fc receptor for IgM2,14. Fcmr-knockout mice produce a myriad of phenotypes, suggesting that FcμR has critical functions in B cell development and immune tolerance4,15,16,17,18,19,20,21. FcμR is also involved in T cell activation22 and T helper 17 cell pathogenicity23. In addition to these lymphocyte-related functions, FcμR has also been implicated in the activation of myeloid cells and their responses to cancer24,25,26, although this functional aspect is under debate27,28. In addition, FcμR participates in mucosal immunity by mediating the retrotransport of sIgM–antigen complexes back from the mucosa5. Of note, FcμR is highly expressed on chronic lymphocytic leukaemia B cells2,29,30,31,32, and targeting FcμR using either an IgM–toxin conjugate33 or engineered chimeric antigen receptor T cells34 could provide new treatment strategies for chronic lymphocytic leukaemia.

FcμR interacts with all forms of IgM, including pentameric and hexameric IgM, as well as mIgM within the BCR2,3,4,35. It is involved in the internalization of soluble IgM–antigen complexes, but not the BCR, suggesting that it binds differently to polymeric and monomeric IgM3,4,32. Furthermore, as FcμR carries out the retrotransport of SC-containing sIgM5, it can bind pIgM together with SC. Here we investigate the molecular mechanism of the FcμR–IgM interaction.

Crystal structure of FcμR-D1–Fcμ-Cμ4

The extracellular domain (ECD) of FcμR contains an immunoglobulin-like domain (D1) followed by an intrinsically disordered, highly O-linked glycosylated stalk region (Fig. 1a). We first present the crystal structure of FcμR-D1 in complex with the Fcμ-Cμ4 domain at 3.0 Å resolution (Fig. 1b and Extended Data Table 1). Surface plasmon resonance (SPR) experiments demonstrated that FcμR-D1 binds Fcμ-Cμ4 with high affinity, exhibiting a dissociation constant (Kd) value of approximately 3.2 nM (Fig. 1c and Extended Data Fig. 1). The structure reveals a 2:1 assembly, in which two FcμR-D1 molecules bind to an Fcμ-Cμ4 dimer from both sides. This is consistent with the previous finding that each IgM heavy chain contains an FcμR-binding site36. The interaction between FcμR-D1 and Fcμ-Cμ4 buries a surface area of 850 Å2 and mainly focuses on the Fcμ-Cμ4 residues Asn465–Glu468 and Glu526 (Fig. 1d). Arg45FcμR from CDR-like loop 1 of FcμR forms an ion pair with Glu468. In CDR-like loop 2, Phe67FcμR packs against the aliphatic portion of Glu468, whereas Lys69FcμR coordinates Glu526. Thr110FcμR in CDR-like loop 3 forms hydrogen bonds with the main chain groups of Asn465 and Leu466, whereas Asp111FcμR interacts with Arg467. Thr57FcμR and Thr60FcμR from the C′ strand of FcμR also mediate hydrogen bond contacts with Leu466–Arg467. Asn465–Glu468 and Glu526 are uniquely present in Fcμ among the five classes of human immunoglobulins, which explains the specificity of FcμR for Fcμ (Extended Data Fig. 2a). A previous study has shown that the Q510R mutant of Fcμ abrogated the binding of FcμR-D1 (ref. 36). In the structure, Gln510 is in close proximity to Lys73FcμR; thus, the Q510R mutant may disrupt the binding via charge repulsion.

Fig. 1: Crystal structure of the FcμR-D1–Fcμ-Cμ4 complex.
figure 1

a, Domain organizations of FcμR and Fcμ. TM, transmembrane; tp, tailpiece. b, Overall structure of the 2:1 FcμR-D1–Fcμ-Cμ4 complex. FcμR-D1 and Fcμ-Cμ4 are shown as ribbons, and the N and C termini of each molecule are indicated. c, SPR analyses of the FcμR-D1–Fcμ-Cμ4 interaction performed by passing twofold serial dilutions of purified Cμ4 (from 80 nM to 5 nM) to immobilized FcμR-D1. All SPR experiments in this paper were repeated at least two times. Fcμ-Cμ4 binding to FcμR-D1 Kd = 3.17 ± 0.63 nM. d, The FcμR-D1–Fcμ-Cμ4 interface. The dashed lines indicate polar interactions. e, Structural comparisons of the Fcγ–FcγRI (PDB ID: 4W4O), Fcε–FcεRI (PDB ID: 1F6A), Fcα–FcαRI (PDB ID: 1OW0) and Fcμ–FcμR complexes. The composite Fcμ–FcμR model was generated based on the FcμR-D1–Fcμ-Cμ4 crystal structure and the Fcμ structure (PDB ID: 6KXS). f, A composite model of two FcμR molecules bound to IgM BCR shown in two views. The model is generated by superimposing the structure of the FcμR-D1–Fcμ-Cμ4 complex onto that of IgM BCR (PDB ID: 7XQ8). The stalk and transmembrane regions of FcμR are sketched hypothetically.

Full size image

The binding mode between FcμR-D1 and Fcμ-Cμ4 is unique among the Fc–FcR complexes (Fig. 1e). The Fc receptors for IgG, including FcγRI (also known as CD64), FcγRII (also known as CD32) and FcγRIII (also known as CD16), form a 1:1 complex with Fcγ and bind to the lower hinge region on top of Cγ2 (ref. 37). FcεRI, the receptor for IgE, binds Fcε in a similar manner38. By contrast, the IgA-specific receptor FcαRI (also known as CD89) forms a 2:1 complex with Fcα and targets the Cα2–Cα3 junction39. Here, FcμR also binds Fcμ with 2:1 stoichiometry; nonetheless, it interacts with the lower half of Cμ4 and approaches the Cμ4 dimer from a completely different angle.

The C termini of the two FcμR and two Fcμ molecules are located on the same side within the complex; thus, this structural arrangement would enable the interaction between FcμR and mIgM on the membrane. The long-awaited structure of IgM BCR has been recently unveiled40,41,42. Superimposing the structure of the FcμR-D1–Fcμ-Cμ4 complex onto that of IgM BCR revealed that the FcμR-binding site next to Igβ was completely exposed (Fig. 1f). The binding site next to Igα is also largely accessible, although an Asn97-glycan on Igα may slightly interfere with the positioning of FcμR at this location. Indeed, FcμR co-precipitates with mIgM in the presence or absence of Igα and Igβ (Extended Data Fig. 1b), consistent with previous studies3,4.

Cryo-EM structure of FcμR–Fcμ–J

Circulating IgM exists as both pentameric and hexameric assemblies, and a cell-based assay has suggested that FcμR can bind to both forms with nanomolar affinity or avidity2,35. Indeed, immobilized FcμR-D1 absorbs Fcμ–J, that is, the pentameric IgM core containing an Fcμ pentamer and the J-chain, with an apparent Kd of approximately 0.3 nM (Fig. 2a). The presence of antigen-binding fragments has no influence on this tight interaction, as FcμR-D1 binds to a recombinant full-length IgM (anti-CD20)43 with similar affinity. To gain further insights into the structural basis of the FcμR–IgM interaction, we assembled a complex consisting of FcμR-ECD and Fcμ–J and performed cryo-electron microscopy (cryo-EM) analyses (Extended Data Table 2 and Extended Data Figs. 3 and 4). Both 1:1 and 4:1 FcμR–Fcμ–J complexes were observed in the cryo-EM data, and their structures were determined at resolutions of 3.4 Å and 3.7 Å, respectively.

Fig. 2: Cryo-EM structure of the FcμR-ECD–Fcμ–J complex.
figure 2

a, SPR analyses of the interactions between FcμR-D1 and Fcμ–J (top) or anti-CD20 IgM (bottom), performed by passing twofold serial dilutions of purified Fcμ–J or anti-CD20 IgM (both from 2 nM to 0.125 nM) to immobilized FcμR-D1. Fcμ–J binding to FcμR-D1 Kd = 0.33 ± 0.11 nM, and anti-CD20 IgM binding to FcμR-D1 Kd = 0.26 ± 0.13 nM. b, Overall structure of the 1:1 FcμR-ECD–Fcμ–J complex (left). The sIgM core structure is shown on the right for comparison. The Fcμ molecules are shown in two shades of blue, whereas the J-chain is shown in magenta. FcμR-D1 is shown in yellow, so is the D1 domain of SC. The rest of SC is shown in white. c, Structure of the 4:1 FcμR-ECD–Fcμ–J complex shown in two views. The four FcμR molecules are shown in different shades of yellow and green and labelled R1–R4. The β2–β3 loop of the J-chain, which prevents the binding of FcμR to Fcμ5, is highlighted with a dashed circle. d, Unmodelled densities (grey) are present in the 4:1 FcμR-ECD–Fcμ–J complex, which most likely correspond to the stalk regions of the FcμR molecules.

Full size image

In the 1:1 structure, a single FcμR occupies almost the same site on Fcμ–J as the D1 domain of SC in sIgM6,7 (Fig. 2b). Similar to SC-D1, this FcμR-D1 interacts with multiple regions in Fcμ–J. First, it binds to the Cμ4 domain of Fcμ1B (the ten Fcμ chains in the Fcμ pentamer are labelled as previously described, with the Fcμ chain attached to the C-terminal hairpin of the J-chain as Fcμ1A6) in a similar manner to that seen in the crystal structure. There is still no contact between FcμR-D1 and the Cμ3 domains or the Cμ3–Cμ4 junctions. Second, Met42FcμR–His43FcμR appears to pack on Tyr134J in the C-terminal region of the J-chain (Extended Data Fig. 4j). Last, Arg112FcμR contacts the tailpiece of Fcμ5B (Extended Data Fig. 4k). Together, this composite interface ensures the highly selective binding of FcμR-D1 at this site (R1 site).

In the 4:1 structure, four FcμR molecules are arranged into a semicircle (Fig. 2c). One of them binds at the R1 site, whereas the other three bind sequentially to Fcμ2–4 (R2–R4 sites). The binding of FcμR to Fcμ5 is prevented by the β2–β3 loop of the J-chain. At the R2–R4 sites, FcμR docks onto the Cμ4 domains of Fcμ2B, Fcμ3B and Fcμ4B, as seen in the crystal structure. The C–C′ loops of Fcμ1B-Cμ4, Fcμ2B-Cμ4 and Fcμ3B-Cμ4 from the adjoining Fcμ units also contribute to binding at these sites (Extended Data Fig. 4l). The four FcμR molecules all bind on the same side of Fcμ–J. This is in sharp contrast to the crystal structure, in which two FcμR molecules bind to both sides of the Cμ4 dimer. There is no notable contact among the four FcμR-D1 domains. Nevertheless, prominent densities are present within the FcμR-D1 semicircle and below the FcμR-D1 domains (Fig. 2d). These densities could not be unambiguously modelled but most likely belong to the stalk regions of FcμR, as the entire FcμR-ECD was used in the cryo-EM study. It is likely that the stalk regions bridge intermolecular interactions between the four FcμR molecules and facilitate their binding on the same side of Fcμ–J. This hypothesis is further supported by our observation that no 4:1 structure was visualized when FcμR-D1 was used to prepare the complex in place of FcμR-ECD, even when FcμR-D1 was present in 200-fold excess (Extended Data Fig. 5).

Cryo-EM structure of FcμR–Fcμ–J–SC

The fact that the R1 site overlaps with the binding site of SC-D1 raises the question of how FcμR enables the retrotransport of SC-containing sIgM5. First, we confirmed that FcμR can indeed bind the core sIgM (Fcμ–J–SC) (Fig. 3a). Then, we determined the cryo-EM structure of FcμR in complex with Fcμ–J–SC at 3.2 Å (Extended Data Table 2 and Extended Data Fig. 6). We initially thought that FcμR might bind to the R2–R4 sites in this complex. To our surprise, four FcμR molecules instead bound to the opposite side of Fcμ–J (R1′–R4′ sites; Fig. 3b). The binding environments at the R2′–R4′ sites are highly similar to those of R2–R4. The R1′ site is slightly different. In addition to mainly targeting Fcμ5A-Cμ4, R1′ FcμR also contacts Ser65J–Asp66J in the J-chain as well as Glu570 in the Fcμ1A tailpiece via Arg112FcμR (Extended Data Fig. 6g).

Fig. 3: Cryo-EM structure of the FcμR-ECD–Fcμ–J–SC complex.
figure 3

a, SPR analyses of the interactions between FcμR-D1 and Fcμ–J–SC performed by passing twofold serial dilutions of Fcμ–J–SC (from 40 nM to 2.5 nM) to immobilized FcμR-D1. Fcμ–J–SC binding to FcμR-D1 Kd = 2.85 ± 0.16 nM. b, Overall structure of the 4:1 FcμR-ECD–Fcμ–J–SC complex shown in two views. The four FcμR molecules are shown in yellow and green and labelled R1′–R4′. SC is shown in pink. c, An FcμR molecule bound to the R5′ site (black) would lead to unfavourable packing with the R4′ FcμR (green), as indicated by the dashed circle.

Full size image

Similar to the situation in the FcμR–Fcμ–J complex, a maximum of four FcμR molecules was observed to attach to Fcμ–J–SC. No FcμR is bound to Fcμ1A, even though the binding site appears exposed at first glance. Superimposing an FcμR-D1–Fcμ-Cμ4 pair from the crystal structure to Fcμ1A suggests that the binding of this fifth FcμR would lead to unfavourable packing with the R4′ FcμR (Fig. 3c). Fcμ1A is bound to the C-terminal hairpin of the J-chain, and this interaction apparently leads to a slight tilting of Fcμ1 within the Fcμ–J plane. Therefore, FcμR would have to approach the Fcμ–J platform with a slightly different angle at this site when compared with R2′–R4′, leading to a clash with the R4′ FcμR.

FcμR mutants with reduced binding to IgM

To assess the functional relevance of the molecular interactions described above, we designed FcμR mutants and performed functional experiments. The R45A mutant of FcμR-D1 failed to pull down monomeric Fcμ (Fig. 4a). Likewise, R45A, F67A;K69A, N109A;R112A, R45A;F67A, R45A;R112A and R45A;F67A;R112A all displayed reduced or abolished interactions with Fcμ–J (Fig. 4b) and Fcμ–J–SC (Fig. 4c). To examine the FcμR–IgM interaction in a cellular context, we fused wild-type (WT) FcμR or the R45A;F67A mutant to GFP and expressed them in HeLa cells. As shown in Fig. 4d, cells bearing WT FcμR readily absorbed Fcμ–J on the surface and quickly internalized it at 37 °C, consistent with previous observations14,32. By contrast, IgM binding and internalization were not observed in the R45A;F67A-expressing cells. We also generated HEK293T and Jurkat cell lines stably expressing WT or the R45A;F67A FcμR. Although their cell-surface levels were similar, R45A;F67A-expressing cells captured substantially less Fcμ–J, as evaluated by flow cytometry (Fig. 4e and Extended Data Fig. 7). Together, these data corroborate our structural analyses and demonstrate the critical roles of these FcμR residues in binding to IgM.

Fig. 4: FcμR mutants display reduced binding to IgM.
figure 4

a, The FcμR-D1 R45A mutant did not interact with monomeric Fcμ in a pull-down assay. For gel source data in this paper, see Supplementary Fig. 1. All pull-down experiments in this paper were repeated two times. b, FcμR-D1 mutants displayed reduced or abolished interaction with Fcμ–J. c, FcμR-D1 mutants were also defective in binding to Fcμ–J–SC. d, HeLa cells expressing FcμR(R45A;F67A) did not bind and internalize twin-strep-tagged Fcμ–J, as examined by confocal fluorescence microscopy. A twin-strep tag is present on the N terminus of Fcμ, and binding of Fcμ–J was examined using PE-labelled streptavidin (red). Expression of FcμR was monitored with C-terminally fused GFP (green). These images are representatives of 131 of 171 cells (131 Fcμ–J+ cells out of 171 GFP+ cells) examined for FcμR–GFP at 0 °C; 22 of 27 cells for FcμR–GFP at 37 °C; 0 of 48 cells for FcμR(R45A;F67A)–GFP at 0 °C; 0 of 16 cells for FcμR(R45A;F67A)–GFP at 37 °C; 0 of 33 cells for GFP only at 0 °C; and 0 of 19 cells for GFP only at 37 °C; respectively. This experiment was repeated two times with similar results. e, Flow cytometry analyses suggested that HEK293T (left) and Jurkat (right) cells stably expressing FcμR(R45A;F67A) exhibit reduced binding to Fcμ–J. Cell-surface FcμR levels were examined using the HM14 anti-FcμR antibody. The experiment was repeated three times. The black dots indicate individual data points for n = 3 biological replicates. Data were plotted as the mean ± 1 s.e.m. Statistical analysis was performed by two-tailed, unpaired Student’s t-test, and the P values are indicated in the graphs. For data underlying this graph, see the source data.

Source data

Full size image

Discussion

Our results show how FcμR binds mIgM, pIgM and sIgM. We first showed that FcμR can form a 2:1 complex with mIgM. FcμR physically associates with IgM BCR3 and limits its cell-surface delivery by causing retention in the Golgi apparatus4. A universal phenotype manifested by the Fcmr-knockout mice is autoantibody production, demonstrating the critical function of FcμR in balancing BCR signalling. In light of the recently reported IgM BCR structure, we showed that the FcμR-binding sites are accessible in IgM BCR. Future work is needed to determine how FcμR, particularly the stalk region and transmembrane segment of FcμR, interacts with Igα and Igβ to regulate BCR activity.

We further showed that FcμR can form a 4:1 complex with both Fcμ–J and Fcμ–J–SC, albeit binding on opposite faces of the Fcμ–J platform. Among the eight different binding sites, the R1 site appears to have the highest affinity, as this site is always occupied in the FcμR–Fcμ–J samples. Furthermore, FcμR-D1 binds with higher affinity to Fcμ–J than to Cμ4, which also reflects the stronger binding at the R1 site, as interactions at the R2–R4 sites largely resemble those seen in the FcμR-D1–Cμ4 structure, whereas FcμR interacts with three Fcμ–J subunits at the R1 site. The R1′ site appears to be the second preferred site. When the R1 site is occupied by SC, FcμR binds to the other side of Fcμ–J instead of the R2–R4 positions. Compared with the R2–R4 or R2′–R4′ sites, the R1′ site also features additional interactions between FcμR and the J-chain, as well as the Fcμ1A tailpiece.

The formation of receptor clusters is a common means to initiate signalling events. Our results strongly suggest that four FcμR molecules tend to cluster on one side of pIgM. Once an FcμR binds the R1 or R1′ site, it may help to recruit the other three molecules to the same side via the stalk region. We envision that the formation of such an FcμR tetramer would bring their transmembrane segments and cytosolic regions together to trigger effector functions. The binding of a fifth FcμR is not allowed on either side of pIgM: on the R1–R4 side, it is hindered by the β2–β3 loop of the J-chain, whereas on the R1′–R4′ side, the modest tilting of Fcμ1 caused by the alliance with the J-chain prevents the binding. Therefore, in addition to preventing IgM hexamer formation, the J-chain appears to break another layer of symmetry, permitting pIgM to bind to only four FcμR molecules on either side. It remains to be determined whether an IgM hexamer that lacks the J-chain can recruit six FcμR molecules and whether this would lead to a different signalling outcome.

Whether pIgM would engage eight FcμR molecules simultaneously on both sides and thereby bridge the contact between two FcμR-bearing cells remains unknown. We did not observe particles with FcμR-ECD or FcμR-D1 bound to both sides of Fcμ–J, even though FcμR-ECD and FcμR-D1 were both present in high excess during sample preparation. Size-exclusion chromatographies were performed after FcμR-ECD or FcμR-D1 was incubated with Fcμ–J, which probably removed FcμR-ECD and FcμR-D1 that bound weakly. Structural analyses suggest that the presence of R1–R4 FcμR renders a conformational change of the Fcμ–J platform and the R1′–R4′ sites are slightly displaced (Extended Data Fig. 8). Therefore, binding of FcμR at the R1–R4 sites probably reduces its binding to the other side.

Compared with human FcμR, mouse FcμR binds IgM less tightly2,44. Differences of two residues may contribute to this difference in binding behaviour (Extended Data Fig. 2b). First, mouse FcμR lacks Asn66, which may impact the positioning of Phe67, which is involved in all of the binding sites described above. Second, Arg112 is replaced by a Lys in mice, which probably affects the binding at the R1 or R1′ sites.

In summary, we delineated the complex mechanism of IgM recognition by its specific receptor FcμR. These results are important steps towards further understanding the elusive effector functions of IgM.

Methods

Cell culture

Sf21, High Five, HEK293F, HEK293T and HeLa cells were originally purchased from the American Type Culture Collection. Human Jurkat T cells, originally purchased from the American Type Culture Collection, were kindly provided by J. Hu (Peking University). These cell lines were not otherwise authenticated. The cells were regularly tested for free of mycoplasma contamination by standard PCR method. Sf21 and High Five cells were maintained using a nonhumidified shaker at 27 °C in SIM-SF and SIM-HF media (Sino Biological), respectively. HEK293F cells were cultured in a humidified shaker with 5% CO2 and 55% humidity at 37 °C in SMM 293T-I (Sino Biological). HeLa and HEK293T cells were cultured in DMEM (Gibco) supplemented with 1% penicillin–streptomycin (Gibco) and 10% FBS (Gibco) in a humidified incubator at 37 °C with 5% CO2. Human Jurkat T cells were cultured in RPMI-1640 medium supplemented with 2 mM GlutaMAX (Gibco), 15% FBS, 10 mM HEPES (Gibco), 1 mM sodium pyruvate (Gibco), 1% MEM nonessential amino acid solution (Gibco) and 1% penicillin–streptomycin at 37 °C and 5% CO2.

Protein expression and purification

DNA fragments encoding the FcμR-D1 domain (residues 18–124, UniProtKB: O60667) or FcμR-ECD (residues 18–251) were cloned into the pFastBac vector with an N-terminal melittin signal peptide followed by a His10 tag. Bacmids were generated using the Bac-to-Bac system (Invitrogen). Recombinant baculoviruses were generated and amplified using Sf21 insect cells. For protein production, High Five cells were infected at a density of 1.5–2.0 million cells per millilitre. After 48 h, the conditioned medium was collected by centrifugation and dialysed to exchange into the binding buffer (25 mM Tris-HCl, pH 8.0, and 150 mM NaCl). The recombinant proteins were then isolated using Ni-NTA affinity purification and eluted with the binding buffer supplemented with 500 mM imidazole. The FcμR-D1 domain and FcμR-ECD were further purified using a Superdex 75 column (GE Healthcare) and a Superdex 200 Increase column (GE Healthcare), respectively. The final buffer used for the gel filtration was the same as the binding buffer. FcμR-D1 mutants were expressed and purified similarly to WT FcμR-D1. The DNA fragment encoding Fcμ residues 446–558 was cloned into a modified pcDNA vector with an N-terminal IL-2 signal peptide and a C-terminal His8 tag. The plasmid was transiently transfected into HEK293F cells using polyethylenimine (PEI; Polysciences). Four days after transfection, the Fcμ-Cμ4 domain was purified from the conditioned medium by a similar method to FcμR-D1. The Fcμ–J complex was expressed and purified as previously described6. For the pull-down assay, Fcμ constructs with an N-terminal twin-strep tag were transiently transfected into HEK293F cells and cultured for 4 days. The protein was retrieved from the conditioned medium using the Strep-Tactin resin (Smart Lifesciences) and then further purified using a Superdex 200 Increase column in buffer containing 25 mM Tris-HCl, pH 7.4, and 150 mM NaCl.

SPR

SPR experiments were performed using a Biacore T200 (GE Healthcare). Purified FcμR-D1 was diluted to 0.02 mg ml−1 with 10 mM sodium acetate (pH 5.0), and immobilized on a CM5 chip (Cytiva) to 800–1,100 resonance units (RU) using standard amine coupling chemistry according to the manufacturer instructions. Binding studies were performed by passing twofold serial dilutions of purified Cμ4 (from 80 nM to 5 nM), Fcμ–J (from 2 nM to 0.125 nM), anti-CD20 IgM (from 2 nM to 0.125 nM) or Fcμ–J–SC (from 40 nM to 2.5 nM) over the immobilized FcμR-D1. The running buffer contains 10 mM HEPES, pH 7.4, 150 mM NaCl, 3 mM EDTA and 0.05% (v/v) surfactant P20. The SPR results were analysed with Biacore Evaluation Software and fitted using a 1:1 binding model. Each SPR experiment was repeated at least two times.

Crystallization and structure determination

Purified Fcμ-Cμ4 was incubated with excess FcμR-D1 overnight on ice, and the FcμR-D1–Fcμ-Cμ4 complex was then isolated using a Superdex 75 column and eluted in the final buffer. The purified complex was concentrated to 6–7 mg ml−1 for crystallization. Diffraction-quality crystals were grown at 18 °C by the sitting-drop vapour diffusion method using a 1:1 ratio of protein:reservoir solution. The reservoir solution contains 0.1 M ammonium citrate tribasic (pH 7.0) and 12% (w/v) PEG 3,350. For data collection, crystals were transferred into a solution containing 0.1 M ammonium citrate tribasic (pH 7.0), 14% (w/v) PEG 3,350 and 24% ethylene glycol before being cryo-preserved in liquid nitrogen. The diffraction data were collected at the National Facility for Protein Science Shanghai (beamline BL19U). Data were processed using HKL2000 (HKL Research). The crystal structure was determined by molecular replacement using the program Phaser45. The Fcμ-Cμ4 domain in the published structure of Fcμ–J–SC (PDB ID: 6KXS) and the model of the FcμR-D1 domain generated by Swiss-Model46 were used as search models. The structure model was then manually adjusted in Coot47 and refined using Phenix48. The final structure was validated with the wwPDB server49.

Cryo-EM sample preparation and data collection

To obtain the FcμR-ECD–Fcμ–J complex for cryo-EM analysis, purified Fcμ–J and FcμR-ECD were mixed in a 1:10 molar ratio and incubated on ice for 2 h. After that, the complex was further purified using a Superose 6 Increase column and eluted with the final buffer. The FcμR-ECD–Fcμ–J–SC complex was obtained similarly. For the FcμR-D1–Fcμ–J complexes, purified Fcμ–J and FcμR-D1 were mixed in 1:10 or 1:200 molar ratios and incubated on ice for 2 h. Each sample was also passed through the Superose 6 increase column and eluted using the final buffer. The above samples were then concentrated to 0.6 mg ml−1 and crosslinked with 0.05% glutaraldehyde (Sigma-Aldrich) for 10 min at 20 °C. Cryo-grids were prepared using the Vitrobot Mark IV (FEI). A 4 μl protein sample was applied onto holey-carbon gold grids (R1.2/1.3, Quantifoil) or graphene oxide-coated holey-carbon gold grids (R1.2/1.3, EMR), blotted with filter papers (Whatman No. 1) at 4 °C and 100% humidity, and then plunged into liquid ethane. The grids were first screened using a 200 kV Talos Arctica microscope equipped with a Ceta camera (FEI). Data collection was carried out using a Titan Krios electron microscope (FEI) operated at 300 kV for the FcμR-ECD–Fcμ–J and FcμR-ECD–Fcμ–J–SC complexes, or the 200 kV Talos Arctica microscope for the FcμR-D1–Fcμ–J complexes. Movies were recorded on a K2 or a K3 summit direct electron detector. Serial EM50 and EPU (E Pluribus Unum, Thermo Scientific) were used to collect the cryo-EM data.

Cryo-EM data processing and model building

Movie frames were motion-corrected and dose-weighted using the MotionCor2 program51. Contrast transfer function correction was performed using Gctf52. The rest of the image processing was performed in RELION53 or cryoSPARC54. High-quality micrographs were selected manually, and particles were autopicked by template picking. Particles were initially subjected to several rounds of 2D classification to exclude inaccurate particles and then further subjected to 3D classifications to choose the correct conformation. The favourite classes were selected for 3D refinement to generate the final 3D reconstruction. The local resolution map was analysed using ResMap55 and displayed using UCSF ChimeraX56. The cryo-EM structure of Fcμ–J or Fcμ–J–SC (PDB ID: 6KXS) and the FcμR-D1 crystal structure were docked into the cryo-EM density map using UCSF Chimera. The structure models were then adjusted using Coot and refined using the real-space refinement in Phenix57.

Immunoprecipitation

Codon-optimized DNAs encoding the heavy chain (variable domain of VRC0158 and constant domain of μ-chain) and light chain (variable domain of VRC01 and constant domain of κ-chain) of mIgM were cloned into the pcDNA vector, separated by a P2A-peptide motif. A Flag tag was present at the N terminus of the light chain. The Igα/Igβ expression vector was constructed similarly, with each of them carrying a C-terminal twin-strep tag. A truncated form of FcμR (residues 1–358; to remove the internalization motif) with a haemagglutinin (HA) tag was also cloned into the pcDNA vector. These plasmids were co-transfected into HEK293T cells using PEI. Cells were harvested 18–24 h later, washed with PBS and resuspended in the lysis buffer containing 25 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM phenylmethylsulfonyl fluoride (PMSF) and 1× protease inhibitor cocktail (B14002, Biotool). The resuspended cell pellets were then solubilized in the lysis buffer supplied with 1% n-dodecyl-β-d-maltoside (DDM; Anatrace) and 0.1% cholesteryl hemisuccinate (CHS; Anatrace) at 4 °C for 2 h with mild rotation. After centrifugation, the supernatants were incubated with the anti-Flag M2 affinity gel (A2220, Sigma-Aldrich) for 2 h with mild rotation. The beads were then washed three times using the binding buffer containing 25 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% DDM and 0.01% CHS. The immunoprecipitated proteins were eluted using the binding buffer supplemented with 200 μg ml−1 3× Flag peptide (NJP50002, NJPeptide). The results were analysed by immunoblotting using antibodies to the strep tag (1:3,000; HX1816, HuaxingBio), Flag tag (1:1,000; F3165, Sigma-Aldrich), HA tag (1:1,000; H3663, Sigma-Aldrich) and goat anti-mouse IgG horseradish peroxidase (1:5,000; HS201-01, TransGen Biotech).

Pull-down assay

A twin-strep tag is present on the N terminus of Fcμ. To examine the interaction between the Fcμ monomer or Fcμ–J and FcμR-D1, 40 μg of monomeric Fcμ or Fcμ–J was mixed with purified WT FcμR-D1 or mutants and incubated on ice for 1 h. The mixtures were then incubated with Strep-Tactin resins (Smart Lifesciences) in binding buffer supplemented with 0.1% Triton X-100 at 4 °C for another hour. After incubating at 4 °C for another hour with mild rotation, the Strep-Tactin beads were spun down and washed three times using the binding buffer. Proteins retained on the beads were then eluted using the binding buffer supplemented with 10 mM desthiobiotin (IBA Lifesciences). To examine the interaction between Fcμ–J–SC and FcμR-D1, the Flag tag on the N terminus of SC was used to perform a pull-down experiment with Flag agarose (Sigma-Aldrich). The experimental process was similar to that in the strep pull-down assay described above, except that the binding proteins were eluted using 200 μg ml−1 Flag peptide. The results were analysed by immunoblotting using antibodies to the strep tag (1:3,000; HX1816, HuaxingBio) and His tag (1:3,000; HT501, TransGen Biotech).

Confocal fluorescence microscopy

HeLa cells grown on coverslips were transfected with the indicated constructs using PEI and cultured for 24 h. The twin-strep-tagged Fcμ–J complex was first incubated with P-phycoerythrin (PE)-labelled streptavidin (1:333; 12-4317-87, Invitrogen) on ice for 30 min and then added to the cell culture to a final concentration of 15 μg ml−1. After incubation on ice or at 37 °C for an additional 30 min, the cells were washed twice using ice-cold or 37 °C PBS as indicated and fixed with 4% paraformaldehyde. The final coverslips were washed twice and mounted on slides with the fluorescence mounting medium (Dako). Images were acquired using a Nikon Live SR CSU W1 confocal microscope equipped with the ×100/1.4 objective. Images were processed using Fiji (ImageJ) with brightness and contrast adjustment for the whole image.

Preparation of FcμR stable cell lines and flow cytometry

GFP only, FcμR–GFP and the R45A;F67A–GFP mutant were cloned into the pQXCIP vector. These constructs were transfected into the HEK293T cell line together with the helper plasmid pCL10A1 using X-tremeGENE 9 (Roche). Two days later, the culture supernatants containing viruses were collected, filtered and used to infect Jurkat cells. Fresh medium with 4 μg ml−1 puromycin was added to the remaining HEK293T culture to screen for stable HEK293T cell lines overexpressing FcμR and the mutant. For the Jurkat T cell line, polybrene (Sigma-Aldrich) was added to a final concentration of 10 μg ml−1 before infecting cells at a ratio of 1 × 106 cells per millilitre of the supernatant. Transfection was performed by centrifugation at 500g for 90 min. Jurkat cells expressing comparable levels of GFP were maintained in the medium containing puromycin at 1 μg ml−1 and enriched three times from each transductant using a BD FACSAria III cell sorter (BD Biosciences). Flow cytometric analyses of cell-surface levels of FcμR were performed using the PE-labelled anti-FcμR monoclonal antibody (1:20; 563018, BD Biosciences). To examine IgM binding, cells were incubated with the twin-strep-tagged Fcμ–J complex at a concentration of 15 μg ml−1 for 30 min on ice, washed and then incubated with PE-labelled streptavidin. Stained cells were measured by the CytoFlEX S system (Beckman Coulter), and the flow cytometric data were analysed with FlowJo software (Tree Star) and GraphPad Prism.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.