Identification of residues in the Cmu4 domain of polymeric IgM essential for interaction with Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1)

Abstract

The binding of nonspecific human IgM to the surface of infected erythrocytes is important in rosetting, a major virulence factor in the pathogenesis of severe malaria due to Plasmodium falciparum, and IgM binding has also been implicated in placental malaria. Herein we have identified the IgM-binding parasite ligand from a virulent P. falciparum strain as PfEMP1 (TM284var1 variant), and localized the region within this PfEMP1 variant that binds IgM (DBL4beta domain). We have used this parasite IgM-binding protein to investigate the interaction with human IgM. Interaction studies with domain-swapped Abs, IgM mutants, and anti-IgM mAbs showed that PfEMP1 binds to the Fc portion of the human IgM H chain and requires the IgM Cmu4 domain. Polymerization of IgM was shown to be crucial for the interaction because PfEMP1 binding did not occur with mutant monomeric IgM molecules. These results with PfEMP1 protein have physiological relevance because infected erythrocytes from strain TM284 and four other IgM-binding P. falciparum strains showed analogous results to those seen with the DBL4beta domain. Detailed investigation of the PfEMP1 binding site on IgM showed that some of the critical amino acids in the IgM Cmu4 domain are equivalent to those regions of IgG and IgA recognized by Fc-binding proteins from bacteria, suggesting that this region of Ig molecules may be of major functional significance in host-microbe interactions. We have therefore shown that PfEMP1 is an Fc-binding protein of malaria parasites specific for polymeric human IgM, and that it shows functional similarities with Fc-binding proteins from pathogenic bacteria.

FIGURE 1

Characterization of TM284var1 and expression of recombinant DBL domains. (A) IFA on live TM284 infected erythrocytes using mouse anti-human IgM (1/1000) and Alexa Fluor™ 488 conjugated goat anti-mouse secondary antibody (1/500). An IgM-positive rosetting infected erythrocyte shows punctuate surface fluorescence, while a non-rosetting infected erythrocyte is negative. Parasite nuclei stained with DAPI. (B) Protein extracts from TM284 or IT/R29 parasites, in the presence or absence of trypsin, were incubated with either protein-L (to pull out IgM) or protein-G (to pull out IgG) beads. Immunoprecipitated products were run on 3-8% acrylamide tris-acetate gels and transferred to PVDF membranes for blotting with the anti-PfEMP mAb 6H1. The product at >250 kDa represents PfEMP1. (C) Domain structure of the PfEMP1 variant encoded by the TM284var1 gene. (D) Northern-blot of ring-stage rosetting (R+) and non-rosetting (R−) TM284 parasites probed with a DBL4β probe specific for TM284var1 (left panel) or an exon II probe recognising all var genes (right panel). Ethidium bromide stained gel is shown as an indication of loading (lower panel). Bars indicate the position of the 9 Kb and 6 Kb bands of the RNA Millenium™ markers (Ambion). (E) IFA on COS-7 cells transfected with TM284 var1CIDRβ (upper panel) and TM284 var1DBL4β (lower panel). COS-7 cells were incubated with mAbs 1D3 or DL6 to determine transfection efficiency (left hand column), with secondary antibody alone as negative control (middle column) or with mouse mAb to human IgM (right hand column). (F) Western blot analysis of purified DBL domains of TM284var1 expressed in E. coli. PVDF membranes were incubated with peroxidase conjugated anti-HIS as described under ‘Experimental Procedures’. An equivalent SimplyBlue-stained SDS-PAGE 4-15% gradient gel is shown on the left. 2 μg of protein was loaded per lane.

FIGURE 2

Overview of antibody heavy chain constructs. Constant heavy chain domains/sequences are shown as blue (μ), yellow (α) or red (γ) ovals. TP = secretory tailpiece. L309C marks a single amino acid replacement of the leucine from IgG to the cysteine from the homologous position in IgM, to create a domain-swap antibody with greater ability to polymerize (38).

FIGURE 3

Binding of IgA/IgM and IgG/IgM domain-swaps to DBL domains. ELISA analysis of recombinant DBL4β from the TM284var1 variant binding to IgA/IgM (A) or IgG/IgM (B) domain-swap antibodies immobilized in microtitre wells (left panels). No binding was seen with control DBL3ε (R29var1 variant) (30) or DBL5ε (Tm284var1 variant) in the same ELISA. Binding of domain-swaps was also investigated by IFA with COS-7 cells expressing DBL4β or DBL5/3ε (right panels). Positive transfectants were detected with mAb DL6 reactive with the HSV-glycoprotein D tag expressed C-terminally of the DBL domain. DL6 was then detected with a FITC labeled anti-mouse IgG (green). Binding of domain-swap antibodies was detected with a phycoerythrin labeled anti-mouse λ (red) that binds to the common light chain shared by all the domain-swap antibodies. Only transfected cells incubated with IgM, α/Cμ2,3,4, α/Cμ4, γ/L309C-Cμ4 and γ/Cμ3,4 bound DBL4β by IFA and this recapitulated data seen by ELISA. Binding of domain-swap antibodies co-localized with DBL4β seen on the surface of unfixed COS-7 cells (C). No binding of antibodies was seen to control COS-7 cells expressing either DBL3ε or DBL5ε.

FIGURE 4

The polymerization status of IgM is crucial for binding to DBL domains. Anti-NIP recombinant human IgM and various point mutants deficient in polymerization (IgM C414S and IgM C575S) were examined for binding to DBL4β by ELISA (left panel) and by IFAs (right panel) as described for Fig. 3. A reduced binding was seen for IgM C414S and no binding was seen with IgM C575S. IgM Fab’s or polymeric IgG (IgG L309C-μTP) showed no binding when used at equivalent molar concentrations to IgM in both assays.

FIGURE 5

Localization of the DBL4β binding site in the Cμ4 domain of IgM. (A) Amino acid alignment of human, mouse and bovine IgM with human IgA1 and IgG1 at two exposed interdomain loop regions within the Cμ4 domain (see Fig. 8B). The IgA1 or IgM sequences are numbered according to the commonly adopted schemes used for IgA1 Bur (59) or IgM (60), whereas Eu numbering is used for IgG (61). Bold residues are conserved between species. Human IgG1 and IgA1 residues involved in contact with protein A and FcαR (CD89) respectively are underlined (41,49). (B) Binding of anti-NIP human IgM and the Cμ4 mutants, NR445-446HL and PLSP394-397LLPQ, were examined for binding to DBL4β by IFA as described for Fig. 3. Neither mutant was seen to bind DBL4β. (C) Binding and detection of IgM point mutants to NIP-BSA immobilized onto microtiter plates. Antibodies were detected with peroxidase conjugated anti-λ of anti-μ as described in ‘Experimental Procedures’. (D) Epitope mapping of mAb binding to the IgM point mutants NR445-446HL and PLSP394-397LLPQ by immunoblotting. (E) Size exclusion chromatograms of recombinant human IgM, γ/L309C-Cμ4 and IgM NR445-446HL antibodies run on a Superdex GL200 column.

FIGURE 6

Epitope mapping of anti-human IgM mAbs and their effect on IgM binding to DBL4β. (A) Domain-swapped and point mutated antibodies were serially diluted from left to right onto nitrocellulose membranes and blotted against a panel of anti-human IgM mAbs from a previous study (42,43). mAbs 4-3, 5D7 and 196.6b bound to the Cμ3 domain, whereas mAbs 1F11 and 1G6 bound to the Cμ4 domain containing antibodies. Binding of antibodies was as for Western blots described in ‘Materials and methods’. (B) ELISA showing blocking of DBL4β binding to IgM by monoclonal antibodies 4-3, 1G6, 1F11, 2F11 and 196.6. IgM was coated onto microtiter plates at 5 μg/ml prior to incubation with varying concentrations of mAb. After washing, a fixed concentration of DBL4β (25 μg/ml) was added and binding detected as described in ‘Experimental Procedures’. (C) Blocking of IgM binding to DBL4β COS-7 cell transfectants by mAb 2F11, in contrast with mAb 1X11 which was unable to block binding at equivalent concentrations (200 ng/ml), as seen in ELISAs. IgM binding was detected by an anti-λ (top panel) or anti-human Fcμ (bottom panel) phycoerythrin labeled antibodies.

FIGURE 7

Development and disruption of ‘pseudo-rosettes’ by Cμ4 specific monoclonal antibodies. (A) Binding of IgM, IgA1 or α/Cμ4 antibodies opsonized onto NIP-coated erythrocytes to DBL4β transfected COS-7 cells assessed by “pseudo”-rosette formation. The results are normalized by expressing specific rosette formation as a percentage of that seen with erythrocytes coated with anti-NIP human IgM (Serotec) at 100 μg/ml. (B) The binding of Ab-coated erythrocytes to DBL4β transfectants (stained with acridine orange) to form rosettes (arrowed) were visualized by white light / fluorescence microscopy and a rosette defined as a DBL4β transfected COS-7 cell surrounded by five or more erythrocytes. (C) IgM coated erythrocytes incubated in the presence of serum or plasma to allow opsonization and activation of components of the classical complement pathway still formed rosettes with DBL4β transfectants. Incubation of erythrocytes opsonized with IgM and complement components to mAb 4-3 totally blocked rosette formation.

FIGURE 8

Molecular models of human IgM highlighting residues interacting with Fc-binding proteins. (A) Model of human IgM (kindly provided by Professor Stephen Perkins, University College, University of London) annotating sites in the Cμ4 domain involved in DBL4β (magenta) binding and sites in the Cμ3 domain important for interaction with C1q (turquoise) (2). (B) Molecular model (PyMol) showing the two sides of IgM-Fc based on the known crystal structure of the analogous IgE-Fc (10OV.pdb)(​62). Exposed highlighted residues Pro³⁹⁴-Pro³⁹⁷ (orange) and Pro⁴⁴⁴-Val⁴⁴⁷ (red) in the Cμ4 domain have been shown to be involved in the interaction with DBL4β. The C-terminal tailpieces are omitted for clarity.