A Structurally Distinct Human Mycoplasma Protein that Generically Blocks Antigen-Antibody Union

See allHide authors and affiliations

Science  07 Feb 2014:
Vol. 343, Issue 6171, pp. 656-661
DOI: 10.1126/science.1246135

Easy M

Our immune systems can produce a vastly diverse repertoire of antibody molecules that each recognize and bind to a specific foreign antigen via a hypervariable region. However, there are a few bacterial antigens—such as Protein A, Protein G, and Protein L—that instead bind to the antibody's conserved regions and can bind to a large number of different antibodies. These high-affinity broad-spectrum antibody-binding properties have been widely exploited both in the laboratory and in industry for purifying, immobilizing, and detecting antibodies. Grover et al. (p. 656) have now identified Protein M found on the surface of human mycoplasma, which displays even broader antibody-binding specificity. The crystal structure of Protein M revealed how Protein-M binding blocks the antibody's antigen binding site. This mechanism may be exploited by mycoplasma to escape the humoral immune response.


We report the discovery of a broadly reactive antibody-binding protein (Protein M) from human mycoplasma. The crystal structure of the ectodomain of transmembrane Protein M differs from other known protein structures, as does its mechanism of antibody binding. Protein M binds with high affinity to all types of human and nonhuman immunoglobulin G, predominantly through attachment to the conserved portions of the variable region of the κ and λ light chains. Protein M blocks antibody-antigen union, likely because of its large C-terminal domain extending over the antibody-combining site, blocking entry to large antigens. Similar to the other immunoglobulin-binding proteins such as Protein A, Protein M as well as its orthologs in other Mycoplasma species could become invaluable reagents in the antibody field.

Clonal B cell proliferation, as well as lymphomas and myelomas, can result from chronic infections with organisms such as Escherichia coli, Helicobacter pylori, and Hepatitis C virus. (16). The main feature seems to be the approximation of two systems, each capable of sustained replication in which the replicating microbe induces proliferation and selection of members of the replicating B cell repertoire. In this regard, we investigated mycoplasma infection because it has the features of chronicity (7) and, as an obligatory parasite, is largely confined to the surface of cells (8, 9). In the course of our experiments, we discovered that some human mycoplasmas produce a protein that binds to immunoglobulins (Igs) with high affinity. This protein, which we refer to as Protein M, has a structure that differs from all others in the Protein Data Bank (PDB).

Because we were interested in clonal B cell proliferation in the context of chronic infections, we investigated whether monoclonal antibodies produced as a result of multiple myeloma react with mycoplasma antigens. We tested the ability of plasma from 20 multiple myeloma patients (10) to bind to total cellular extracts from multiple mycoplasma species, including human pathogens or commensal organisms and others that infect nonhuman vertebrates. Remarkably, these experiments showed that the antibodies in the plasma all reacted strongly with molecules present in human, but not nonhuman, pathogenic mycoplasmas (Fig. 1A and fig. S1A). The main reactivity was with a protein with an apparent molecular weight of ~50 kD in Mycoplasma genitalium and with several proteins with apparent molecular weights of 40 to 65 kD in Mycoplasma penetrans (Fig. 1A). We focused our attention on the protein from M. genitalium because it appeared to be more homogeneous as determined with gel electrophoresis (Fig. 1A). The Ig reactivity with the M. genitalium protein was similar for all patients’ plasma tested (fig. S1A). To substantiate that the clonal multiple myeloma Ig is the component responsible for binding to the M. genitalium protein, as opposed to a highly reactive protein that copurifies with it, the Fab′ (fragment antigen-binding) of the primary monoclonal antibody in the plasma of multiple myeloma patient 13PL (13PL Fab′) was highly purified by chromatography followed by crystallization, and its reactivity was studied by using dissolved crystals as a source of the antibody (Fig. 1, B to D). The 13PL Fab′ from the dissolved crystals bound to the same antigen in mycoplasmas as antibodies isolated from whole sera (Fig. 1C). To confirm that the crystals contained an antibody, the x-ray structure was determined at 1.2 Å resolution, and only a Fab′ was present (Fig. 1D). To test whether a similar reactivity could be found in blood from nonmyeloma normal donors (fig. S1B), samples from random donors were studied. The sera from these normal donors also surprisingly reacted with the same mycoplasma proteins as did the clonal myeloma Igs, indicating that the ability of human Igs to react with this mycoplasma protein was not confined to those produced in multiple myeloma. We therefore termed the M. genitalium protein that reacts with Ig as Protein M.

Fig. 1 Igs selectively bind to proteins in human mycoplasma.

(A) (Left) Western blot analysis of the reactivity of plasma from multiple myeloma patient 13PL with cell extracts from Mycoplasma alligatoris, Mycoplasma crocodyli, Mycoplasma fermentans, M. genitalium, Acholeplasma laidlawii, Mycoplasma mycoides, M. penetrans, Mycoplasma pneumoniae, and Mycoplasma pulmonis. All mycoplasma cells were grown in appropriate media. Cells were lysed according to manufacturer’s protocol by using lysis buffer from Sigma Aldrich (St. Louis, MO). Nucleic acids were degraded by means of treatment with deoxyribonuclease and ribonuclease. A protease inhibitor cocktail (Roche, Basel, Switzerland) was added to prevent proteolytic degradation. The extracts from the same number of cells were separated on SDS-PAGE gels and transferred to nitrocellulose membranes for Western blot analysis. (Right) Ponceau red-stained protein bands of the cell extracts. (B) Crystals of 13PL Fab′ from a multiple myeloma patient’s monoclonal immunoglobulin. (C) Western blot analysis of the reactivity of 13PL Fab′ from the plasma of a multiple myeloma patient with the same cell extracts in (A). The 13PL Fab′ was purified by means of crystallization. The extracts were separated on SDS-PAGE gels as described in (A). (D) Crystal structure of 13PL Fab′ shown in ribbon diagram, with the light and heavy chains colored in light and dark gray, respectively. The loops corresponding to CDRs L1, L2, and L3 are colored red, whereas CDRs H1, H2, and H3 are colored blue. The relatively rare disulfide in human CDR3s is colored green.

At this point, the possibilities were that Protein M was an antigen to which most people make an antibody or was a protein that binds to Ig domains or other features that are present in most antibodies. To study these possibilities, we first isolated Protein M using an affinity column constructed from antibody 13PL. The affinity-purified Protein M was separated on SDS–polyacrylamide gel electrophoresis (SDS-PAGE) gels followed by Western blot analysis by using a different myeloma antibody to confirm the presence of the binding protein. The band on the SDS-PAGE gel corresponding to Protein M was excised, and proteomics analysis with mass spectrometry was carried out (fig. S2A). These studies showed that Protein M was M. genitalium protein MG281, which is an uncharacterized membrane protein (UniProtKB accession no. P47523) (fig. S2B) of 556 amino acids with a predicted transmembrane domain (residues 16 to 36) (fig. S2C). Furthermore, homologs of Protein M are present in other mycoplasma strains, such as Mycoplasma pneumonia, Mycoplasma iowae, and Mycoplasma gallisepticum (UniProtKB database). Antibodies did not bind to mycoplasma extracts from a Protein M–null M. genitalium mutant, again suggesting that Protein M might be the molecule to which antibodies bind (fig. S3A). To establish that Protein M alone is sufficient for antibody binding, a His-tagged Protein M lacking the membrane-spanning region (recombinant Protein M, residues 37 to 556) was cloned, expressed in E. coli, and purified by means of affinity chromatography and size-exclusion chromatography. Western blot analysis of purified Protein M showed that it reacted strongly with the monoclonal Ig from a multiple myeloma patient (fig. S3B). Protein M also bound to all isotypes of human IgGs (fig. S4A) as well as mouse, rat, rabbit, goat, and bovine IgGs (fig. S4B). To further elucidate the minimum sequence responsible for antibody binding, the Protein M and 13PL IgG complex (mixed in a 1:1.1 molar ratio) was incubated with trypsin for 5 hours. SDS-PAGE gel analysis showed that a truncated protein remained intact after 5 hours as compared with uncomplexed Protein M, which was totally digested into smaller fragments. The trypsin-digested Protein M (Protein M TD) was found by means of mass spectroscopy to contain residues 74 to 482 (fig. S5). A His-tagged Protein M TD consisting of residues 74 to 468 (recombinant Protein M TD) was then cloned, expressed in E. coli, and purified by means of affinity chromatography and size-exclusion chromatography. Protein M and Protein M TD showed similar binding affinities to a panel of Igs or Fabs with binding affinity (Kd) values in the nM range (fig. S6), as determined by using Biolayer Interferometry (11).

Ig binding to Protein M was confined to the Fab domain of the antibody molecule, as shown with Biolayer Interferometry (fig. S6). Because a variety of antibodies with different complementarity-determining regions (CDRs) all bind to Protein M, specific interaction with the combining site of the antibody molecule appeared to be excluded. To understand the molecular basis for this interaction, crystal structures of recombinant Protein M TD in complex with two antibody Fabs PGT135 against HIV-1 gp120 with a κ light chain (Kd = 3.7 nM) (12) and Fab CR9114 against influenza hemagglutinin with a λ light chain (Kd = 1.9 nM) (13) were determined to 1.65 and 2.50 Å resolution, respectively (Fig. 2 and table S1). Although the 13PL Fab′–Protein M TD complex could not be crystallized, we were able to obtain the structure through electron microscopy, which showed a similar mode of binding (fig. S7). The Protein M structure is very different from any other known Ig-binding proteins—such as Protein G, Protein A, and Protein L—or indeed any other structures in PDB ( (14). Protein M TD comprises a large domain (residues 78 to 440) that includes a leucine-rich repeat (LRR)–like subdomain and a smaller domain (residues 441 to 468) (Fig. 2A). Protein M TD binds predominantly to the variable light (VL) domains of both PGT135 Fab (Fig. 2B) and CR9114 Fab (Fig. 2C) but makes some very limited interactions with the other three Fab domains. The Fab–Protein M TD interactions bury total solvent-accessible surface areas of 3590 Å2 and 2870 Å2 for PGT135 Fab (Fig. 2D and figs. S8A and S9A) and CR9114 Fab (Fig. 2E and figs. S8B and S9B), respectively, mainly from the VL domains of the Fabs (15). The common interacting positions, which are about two-thirds occupied by hydrophilic residues in both antibodies, are located on one edge of VL (Fig. 2F and tables S2 and S3). Ten conserved hydrogen bonds and one salt bridge are made from Protein M TD to each Fab VL, including six hydrogen bonds to the main chain of VL residues 15, 16, 18, 54, and 77, with the rest to the side chains of VL Arg61, Gln79, and Glu81 (tables S2 and S3), almost all of which are highly conserved among human antibodies with both κ and λ light chains (table S4). Other residues at the common paratope positions, which make van der Waals contacts and nonconserved H-bonds or salt bridges, are less conserved except for VL Gln37 and Pro59 (table S5) as well as the completely conserved CL Ser168 and VH1 Ser168. However, some of the nonpolar interactions may be conserved even with different amino acids. The N- and C-terminal fragments (residues 37 to 74 and residues 469 to 556), which were truncated in Protein M TD as compared with Protein M, are likely disordered because the three-dimensional reconstructions of a Fab in complex with Protein M and Protein M TD by using negative-stain electron microscopy are nearly identical (figs. S10 and S11).

Fig. 2 Crystal structures of recombinant protein M TD in complex with PGT135 Fab and with CR9114 Fab.

(A) Overall structure of Protein M TD from its complex with PGT135 Fab in ribbon representation. Protein M TD appears to have two main domains: a larger domain I in green (residues 78 to 440), which includes a LLR-like structure in purple (residues 240 to 330) and a smaller domain II in light blue (residues 441 to 468). (B) Overall structure of Protein M TD in complex with PGT135 Fab. Protein M TD is colored in green, and PGT135 Fab is colored in light gray for the light chain and dark gray for the heavy chain. Protein M TD predominantly binds VL of PGT135 Fab but also interacts to a lesser extent with VH, CL, and CH1. (C) Overall structure of Protein M TD in complex with CR9114 Fab. The coloring scheme is similar to that of (B). Residues 455 to 468 in domain II of Protein M TD were flexible and not modeled. Protein M TD predominantly binds VL of CR9114 Fab. (D) Molecular surface representation of PGT135 Fab with the paratope of Protein M TD in red, Fab light chain in light gray, and Fab heavy chain in dark gray. (E) Molecular surface representation of CR9144 Fab with the paratope of Protein M TD and colored as in (D). (F) The common paratope residue locations (in red) of Protein M TD for PGT135 Fab and CR9114 Fab are shown on PGT135 Fab, including VL residues 14 to 18 of FR1; 37, 45, 47, and 49 of FR2; 53 to 56 of CDR2; 57 to 61, 76, 77, 79, and 81 of FR3; residues 107 and 109 connecting VL and CL; as well as residues 168 and 170 of CL and residue 168 of CH1.

To determine the scope of Protein M binding to antibodies with different light-chain configurations and allotypes, the Kd of Protein M TD to 24 different light chains in a variety of formats were determined. Because the heavy chains potentially could alter the light-chain conformations, we studied the same germline light chain paired with three different heavy chains. The particular heavy chains made little difference, and the Kd varied between 1.5 and 4.8 nM for four VH/VL combinations (fig. S12, A to C). Similarly, when the same heavy chain was paired with four different light chains, Protein M binding ranged from 1.8 to 2.4 nM (fig. S12, D to G). We evaluated the effect of allotypic variation using five κ chain allotypes (κ1, κ2, κ3, κ4, and κ6) and three λ chain allotypes (λ2, λ5, and λ11). Allotypic variation had little effect, and the Kd for the allotypic variants ranged between 1.0 and 4.8 nM (fig. S12 and table S6). The preservation of binding affinity in the presence of allotypic variation is to be expected because the critical Protein M contacts are largely conserved among allotypes (tables S4 and S5). To determine whether antigen specificity affected Protein M, we determined Kd for Protein M TD binding to a panel of eight affinity-matured monoclonal antibodies against the same HIV-1 gp120 antigen but with different epitopes and found that Kd only varied between 0.7 and 3.8 nM (fig. S12, Q to X, and table S6).

Last, we assessed the percentage of polyclonal human Igs from the plasma of normal blood donors that was capable of binding to Protein M. After two passages through a column containing Protein M TD immobilized on Ni–nitrilotriacetic acid matrix at a flow rate of 1 ml/min, >90% of all the Igs were removed, which is in agreement with our data that showed Protein M binds to all human monoclonal antibodies that we have tested to date.

These structural studies suggested that Protein M should preclude the ability of the antibody to bind to its antigen because it displaces or distorts the CDRs and/or may use its C-terminal domain to sterically block entrance to the antibody-combining site (fig. S13). We tested the ability of recombinant Protein M and Protein M TD to block antigen-antibody union for six different antigen-antibody pairs, including two polyclonal auto-antibodies. The monoclonal antibodies used were generated against human influenza virus (13), HIV-1 (16), human Ebola (17), and mouse Ebola (18); polyclonal antibodies were purified from Goodpasture’s disease patient serum (19) and lupus mouse serum (20). Blocking of the binding of serum polyclonal antibodies to antigens by Protein M is important because such sera represents a collection of antibodies rather than a single monoclonal species. Prior incubation of the antibodies with Protein M or Protein M TD (in a 1:8 molar ratio) strongly inhibited antibody binding to its cognate antigen (Fig. 3), but the order of addition is critical. Once antigen-antibody union has occurred for high-affinity antigens, Protein M does not disrupt the antibody-antigen complex (fig. S14).

Fig. 3 Protein M blocks antigen-antibody union.

(A) Binding of CR9114 IgG, a human broadly neutralizing antibody against influenza virus to one of its antigens, H5 hemagglutinin [influenza virus strain A/Viet Nam/1203/2004 (H5N1)] was evaluated by using enzyme-linked immunosorbent assay (ELISA) after precomplexing with recombinant Protein M (red) and Protein M TD (blue) (at a 1:8 molar ratio). Binding of the IgG to the HA in the absence of protein M was used as control (purple). The extent of binding was analyzed with a colorimetric assay. The curves were obtained with a nonlinear regression analysis in which the data were fit to a four-parameter logistic equation based on a simple binding model. Error bars represent SD of duplicate measurements. (B) Binding of PGT135 IgG, a human broadly neutralizing antibody against its antigen HIV-1 gp120 (JR-FL gp120 core construct) was evaluated (purple) after precomplexing PGT135 IgG with Protein M (red) and Protein M TD (blue) (at a 1:8 molar ratio). ELISA assay was performed as in (A). (C) Binding of KZ52 IgG, a human broadly neutralizing antibody against its Ebola antigen glycoprotein (purple), was evaluated as in (B). ELISA assay was performed as in (A). (D) Binding of 13C6 IgG, a mouse broadly neutralizing antibody against its Ebola glycoprotein antigen (purple) was evaluated as in (B). (E) Binding of antibody to COL4A3 human polyclonal serum from a patient with Goodpasture’s disease to its antigen COL4A3 (purple) was evaluated as in (B). (F) Binding of anti-DNA polyclonal serum from a mouse with lupus to its antigen chromatin (purple) was evaluated as in (B).

This discovery, of a heretofore unknown high-affinity Ig-binding protein in human M. genitalium, should be considered in the context of other known Ig-binding proteins, such as Protein G, Protein A, and Protein L (2123), which have been invaluable reagents and tools in the antibody field. The Protein M structure is very different from these other Ig-binding proteins and is also very different from any other known protein structures. Unlike Protein G, Protein A, and Protein L—which all contain multiple, small, Ig-binding domains—Protein M has a large domain of 360 residues, which binds principally to antibody VL domains, as well as a LRR-like motif (24) that faces away from the antibody molecule and may have an as-yet uncharacterized function. Protein M also contains a 115-residue C-terminal domain that likely protrudes over the antibody combining site. To our knowledge, compared with other known Ig-binding proteins, the Protein M TD–antibody Fab buried surface area is the largest (25).

Protein M binds to antibodies with either κ or λ light chains using conserved hydrogen bonds and salt bridges, from backbone atoms and conserved side chains, and some conserved van der Waals interactions as well as other nonconserved interactions. These conserved interactions provide a structural basis for the broad reactivity with Fvs, Fabs, or Igs. In contrast, the primary binding site for Protein G and Protein A is the antibody Fc domain, although secondary lower-affinity binding sites include the CH1 domain of IgG for Protein G (26) or VH of the human VH3 gene family for Protein A (22). Protein L binds only to the VL of most human κ light chains, except for the VκII subgroup (Fig. 4) (23). Thus, this new broad-scope, high-affinity antibody-binding protein, which binds both κ and λ chains, is likely to find a myriad of applications in immunochemistry. In addition to its general use, Protein M may be particularly important for large-scale purification of therapeutic antibodies. Last, although Protein M may be important for the host-parasite relationship, further studies are necessary to elucidate the consequences of its expression on the parasite surface.

Fig. 4 Structural comparison of the antibody-binding sites of Protein M and other Ig-binding proteins.

All antibody Fabs or Fcs are shown in ribbon representation with light chain in light gray and one heavy chain in dark. (A) Protein M TD is depicted in green ribbon representation in its complex with PGT135 Fab. Protein M mainly binds to the VL domain. (B) One of the Ig-binding domains of Protein L (orange) bound to VL of an antibody Fab (PDB code 1HEZ). Noticeably, although the Protein L domain can bind to two Fabs with similar paratope sites (23), only one binding site is shown here. (C) One of the Ig-binding domains of Protein A (blue) bound to VH (weaker secondary site) of an antibody Fab (PDB code 1DEE). (D) One of the Ig-binding domains of Protein G (pink) bound to CH1 (weaker secondary site) of an antibody Fab (PDB code 1IGC). (E) One of the Ig-binding domains of Protein A (blue) bound to the antibody Fc region (primary site) near the CH2 and CH3 domain interface (PDB code 1FC2). (F) One of the Ig-binding domains of Protein G (pink) also bound to antibody Fc region (primary site) near CH2 and CH3 domain interface (PDB code 1FCC). Although Protein M and Protein L predominantly bind the VL domain, their binding sites are very different, with only one common residue (position 18 of VL). These Ig-binding proteins appear to have different ranges of affinities for antibodies. Generally, for human antibodies, Protein M binds strongly to all the types; Protein G binds strongly to the antibody Fc region and weakly to the CH1 domain; Protein A binds strongly to antibody Fc region (except to the IgG3 subtype) and weakly to VH of VH3 gene family; and Protein L binds strongly to VL of κ light chains, except the VκII subgroup.

Supplementary Materials

Materials and Methods

Figs. S1 to S14

Tables S1 to S6

References (2748)

References and Notes

  1. ForteBiol; a division of Pall Life Sciences. All rights reserved, 2013.
  2. A structural similarity search performed by using the Dali server ( indicated that Protein M TD is not substantially similar to other structures, except for the LLR-like motif.
  3. The Fab-Protein M TD interactions bury total solvent-accessible surface areas of 2490 Å2 for PGT135 VL domain (1280 Å2 from Protein M TD and 1210 Å2 from PGT135) and 2380 Å2 for CR9114 VL domain (1230 Å2 from Protein M TD and 1150 Å2 from CR9114), respectively.
  4. Acknowledgments: We thank B. Cravatt and G. Siuzdak for mass spectroscopy studies, M. Lee and C. Locke for assistance in Western blot and ELISA studies, W. Yu for protein purification, and H. Tien of the Robotics Core at the Joint Center for Structural Genomics ( for automated crystal screening. X-ray diffraction data sets were collected at the Stanford Synchrotron Radiation Lightsource beamlines 11-1 and 12-2. The b12 IgG and PGT135 IgG were kindly provided by B. Moldt in the laboratory of D. Burton, and the other anti-HIV antibody Fabs were provided by J.-P. Julien and Y. Hua. The Lupus mouse serum and mouse chromatin were kindly provided by R. Baccala. The plasmapheresis fluid from a patient with Goodpasture’s disease and human recombinant COL4A3 antigen were provided by V. Pedchenko. The KZ52 IgG, 13C6 IgG, and Ebola GP were kindly provided by E. Saphire. The electron microscopy studies were supported by startup funds from The Scripps Research Institute (to A.B.W.) and conducted at the National Resource for Automated Molecular Microscopy at The Scripps Research Institute, which is supported by NIH through the National Center for Research Resources’ P41 program (RR017573). We also thank NIH RO1 AI042266 (I.A.W.), NIH 5 R21 AI098057-02 (J.I.G.), NIH R01 AG020686 (M.C.), NIH U19 AI06360 (D.R.S.), and NIH K08 AR063729-01 (A.M.) for support. Coordinates and structure factors are deposited in PDB under accession codes 4NZR for Protein M TD/PGT135 Fab complex, 4NZT for Protein M TD/CR9114 Fab complex, and 4NZU for 13PL Fab′. The electron microscopy reconstructions are deposited in the Electron Microscopy Data Bank under accession codes EMD-5834 for Fab of human mAb b12 in complex with Protein M, EMD-5835 for Fab of multiple Myeloma patient human mAb 13PL in complex with Protein M TD, and EMD-5836 Fab of human mAb b12 in complex with Protein M TD. A patent covering this work (official application PCT 13/050656) entitled “Immunoglobulin-binding of human mycoplasma antigens and methods of use thereof” has been filed. This is publication 25063 from The Scripps Research Institute.

Stay Connected to Science

Navigate This Article