The Immunoglobulin D-binding Part of the Outer Membrane Protein MID from Moraxella catarrhalis Comprises 238 Amino Acids and a Tetrameric Structure*

Moraxella catarrhalis IgD-binding protein (MID), a 200-kDa outer membrane protein comprising 2,139 amino acids, has recently been isolated and shown to display a unique and specific affinity for human IgD. To identify the IgD-binding region, MID was digested with proteases. In addition, a series of truncated fragments of MID were manufactured and expressed in Escherichia coli followed by analysis for IgD binding in Western and dot blots. The smallest fragment with essentially preserved IgD binding was comprised of 238 amino acid residues (MID962–1200). Shorter recombinant proteins gradually lost IgD-binding capacity, and the shortest IgD-binding fragment comprising 157 amino acids (MID985–1142) displayed a 1,000-fold reduced IgD binding compared with the full-length molecule. The truncated MID962–1200 was efficiently attracted to a standard IgD serum and to purified myeloma IgD(κ) and IgD(λ) sera but not to IgG, IgM, or IgA myeloma sera. Furthermore, the fragment specifically bound to peripheral blood B lymphocytes, and the binding was inhibited by preincubation with anti-IgD-Fab polyclonal antibodies. Results obtained by introducing five amino acids randomly into MID962–1200 using transposons suggested that α-helix structures were important for IgD binding. Ultracentrifugation experiments and gel electrophoresis revealed that native MID962–1200 was a tetramer. Interestingly, tetrameric MID962–1200 attracted IgD more than 20-fold more efficiently than the monomeric form. Thus, a tetrameric structure of MID962–1200 is crucial for optimal IgD-binding capacity.

tarrhalis and H. influenzae activate human B cells through interactions with surface IgD (5). We have recently purified and characterized a 200-kDa outer membrane protein from M. catarrhalis designated Moraxella IgD-binding protein (MID) 1 (6). The deduced amino acid sequence of MID consists of 2,139 residues. MID has specific IgD-binding properties and interacts with two purified human IgD myeloma proteins, four IgD myeloma sera, and one IgD standard serum. In contrast, no binding is detected to myelomas of other immunoglobulin classes including IgG, IgM, IgA, and IgE. MID also binds to the surface-expressed B cell receptor (BCR) IgD but not to other membrane molecules on human peripheral blood lymphocytes. Furthermore, MID is mitogenic for purified B cells and induces Ig production in the presence of Th2 cytokines (7).
IgD is a unique immunoglobulin, which was first discovered in 1965 by Rowe and Fahley (8). Because soluble IgD is easily degraded by proteolytic enzymes and therefore difficult to purify, the specific role of IgDs has not yet been completely defined (9). Soluble IgD represents approximately 0.25% of the total serum immunoglobulins. In contrast, cell surfacebound IgD serves as an antigen BCR and is co-expressed with IgM on mature B lymphocytes. Interestingly, an increased number of IgD molecules has been observed in human bronchus-associated lymphoid tissue and tonsils (10), suggesting a possible interaction between the immune system and the IgD-binding respiratory pathogens M. catarrhalis and H. influenzae (4,6).
In the present study, we show that a recombinant protein corresponding to a 238-amino acid sequence between positions 962 and 1200 of MID essentially has a preserved IgD-binding capacity on a molar basis compared with the fully intact MID 1-2139 molecule. The truncated protein formed a tetramer as revealed by gel electrophoresis and ultracentrifugation analyses. The tetrameric molecule bound IgD more than 20-fold efficiently compared with the monomeric counterpart.

MATERIALS AND METHODS
Reagents-M. catarrhalis Bc5 was a clinical isolate from a nasopharyngeal swab culture obtained from our department (6). E. coli DH5␣ or BL21(DE3) (Novagen; Darmstadt, Germany) were used for most transformations. For vectors containing fragments A and B, the host BL21(DE3)-pLysS (Novagen) was used. Transformed bacteria were grown in Luria Bertani broth supplemented with kanamycin. In the case of BL21(DE3)-pLysS, chloramphenicol was also added. The expression vector pET-26b(ϩ) was from Novagen, and the pMAL expression vector was purchased from New England Biolabs (Beverly, MA). The human IgD myeloma whole sera IgD() and IgD() were obtained from The Binding Site (Birmingham, England), and the IgD standard serum OTRD 02/03 was from Behringwerke (Marburg, Germany). Myeloma whole sera IgG, IgM, and IgA were included as described previously (6). Horseradish peroxidase (HRP)-conjugated goat anti-human IgD polyclonal antibodies were purchased from BioSource International (Camillo, CA), and HRP-conjugated anti-humanor -light chain polyclonal antibodies were from Dakopatts (Glostrup, Denmark). To produce a specific anti-MID 902-1200 antiserum, rabbits were immunized intramuscularly with 200 g of purified recombinant protein emulsified in complete Freund's adjuvans (Difco, Becton Dickinson; Heidelberg, Germany) and boosted on days 18 and 36 with the same dose of protein in incomplete Freund's adjuvans. Blood was drawn 2 to 3 weeks later. HRP-conjugated swine anti-rabbit pAb (Dakopatts) were used as detection antibodies. Antibodies used for flow cytometry were R-phycoerythrin (RPE)-labeled anti-human CD19 (Dakopatts) mAb and the Ig-fraction of rabbit anti-human IgD-Fab (4), which was further purified by passing through a polyclonal Ig-Sepharose column.
Peptide Cleavage of MID and Amino Acid Sequence Analysis-MID was isolated and purified from M. catarrhalis Bc5 as described previously (6). Native MID in 0.05 M Tris-HCl (pH 8.8) containing 0.1% Empigen ® was digested with trypsin or chymotrypsin at an enzyme/ protein ratio of 1:10 at 37°C overnight or with Endoproteinase LysC at an enzyme/protein ratio of 1:50 (Sigma) at 30°C overnight. The digestion mixtures were run on SDS-PAGE, and protein bands were then transferred electrophoretically to Immobilon-P membranes (Millipore, Bedford, MA) and analyzed for IgD binding. The fragments were sequenced with an Applied Biosystems (Foster City, CA) 470A gas-liquidsolid-phase sequenator (11). The resulting sequences were for trypsin TAQANTESSIAVG, GNTATNFSVNSGDDNALIN, and QGINED-NAFVKGLEK, and for chymotrypsin PSTVKADN. When MID was digested with LysC, the sequence MID 533-1172 was obtained by mass spectrometry (12).
DNA Cloning and Protein Expression-All truncated MID constructs were manufactured using PCR-amplified fragments. Taq DNA polymerase was from Roche Molecular Biochemicals, and PCR conditions were as recommended by the manufacturer. The open reading frame of the mid gene from M. catarrhalis (pET26-MID) was used as template (6). All MID constructs, expect for MID 367-590 (fragment C), were amplified by PCR using specific primers introducing the restriction enzyme sites BamHI and HindIII. Because fragment C has an internal HindIII site, an XhoI was used instead of HindIII at the 3Ј-end. The PCR products were cloned into pET26b(ϩ), except for fragment I, which was cloned into pMAL-c2. To avoid presumptive toxicity, the resulting plasmids were first transformed into the host E. coli DH5␣. Thereafter, the plasmids encoding for the MID fragments were transformed into the expressing hosts BL21(DE3) or BL21(DE3)-pLysS. All constructs were sequenced using the BigDye Terminator Cycle Sequencing v. 2.0 Ready reaction sequencing kit (PerkinElmer Life Sciences).
To produce recombinant proteins, bacteria were grown to mid-log phase (OD 600 0.5-1.0) followed by 3.5 h of induction with 1 mM isopropyl-1-thio-␤-D-galactoside, resulting in overexpression of the proteins. Bacteria were sonicated, and the recombinant proteins were purified on columns containing a nickel resin (Novagen) according to the manufacturer's instructions for native conditions. Fragment I, which was cloned into the pMAL-c2 vector, was purified on amylose resin columns (New England Biolabs). The concentrations of eluted proteins were determined using the BCA Protein Assay kit (Pierce). The resulting proteins were then examined by SDS-PAGE, Western, and dot blots.
Gel Electrophoresis and Detection of Proteins on Membranes (Western Blot)-SDS-PAGE (12%) was run as described previously (6). Samples of purified proteins were also mixed with Tris-HCl sample buffer (pH 8.8) at room temperature and run at native conditions in Tris-HCl PAGE (7.5%) in Tris-glycine native running buffer (pH 8.3) at 150 constant voltage using running and blotting instruments from Bio-Rad. Gels were stained with Coomassie Brilliant Blue R-250 (Bio-Rad). In parallel, the proteins were electrophoretically transferred (30 V for 2-3 h) from a gel to an Immobilon-P membrane (Millipore). Gels with full-length MID were blotted at 30 V overnight. After transfer, the membrane was blocked with 5% milk powder in PBS with 0.1% Tween 20 (PBS-Tween) for 2 h. After four washings in PBS-Tween, the membrane was incubated with IgD myeloma serum (0.5 g/ml human IgDmyeloma protein, The Binding Site) for 1 h. HRP-conjugated goat antihuman IgD (BIOSOURCE International), diluted 1:1,000, was added after four washes and incubated for 45 min. The Ig proteins were diluted in PBS-Tween containing 2% milk powder, and the incubations were kept at room temperature. Finally, the membrane was washed with PBS-Tween four times. Development was performed with ECL Western blotting detection reagents (Amersham Biosciences) and visualized with a Personal Molecular Imager FX (Bio-Rad). Analysis of tetrameric and monomeric forms of MID 962-1200 (F2) on Coomassiestained gels and Western blots was performed with the Quantity One densitometry software from Bio-Rad.
Dot Blot Assays-Purified MID-derived proteins diluted in 4-fold steps in 100 l of 0.1 M Tris-HCl (pH 9.0) were manually applied to nitrocellulose membranes (Schleicher & Schü ll, Dassel, Germany) using a dot blot device. To compare the truncated proteins with each other and with full-length MID, the same amount on a molar basis was added to the membranes. After a 1-h incubation at room temperature, membranes were treated as the Western blot membranes. In separate experiments, the strongly IgD-binding recombinant protein MID 962-1200 was diluted in 4-fold steps and applied to membranes. After blocking, the lanes were separated and incubated with different human myeloma sera (2.8 g with reference to Ig) for 1 h, followed by four washes and incubation with anti-humanand -light chain polyclonal antibodies. The membranes were then washed four times and developed as described above.
Enzyme-linked Immunosorbent Assay (ELISA)-The interaction between IgD and the truncated MID-derived proteins was also tested in ELISA. Microtiter plates (F96 Maxisorp, Nunc-Immuno Module; Roskilde, Denmark) were coated with an IgD myeloma serum (8 g of IgD()) diluted in 4-fold steps in 0.1 M Tris-HCl (pH 9.0) at 4°C overnight. The plates were washed four times with PBS-Tween and blocked for 2 h at room temperature with PBS-Tween supplied with 1.5% ovalbumin (blocking buffer). After four washings, the plates were incubated for 1 h at room temperature with 2 g of the MID-derived proteins diluted in blocking buffer. Thereafter, the plates were washed and incubated with rabbit anti-MID 902-1200 antiserum diluted 1:500 in blocking buffer. After 1 h at room temperature, HRP-conjugated swine anti-rabbit pAb diluted 1:1,000 was added and incubated at room temperature for 1 h. After four additional washings, the plates were developed and measured at 450 nm.
Flow Cytometry-Human peripheral blood lymphocytes (PBLs) were isolated from heparinized blood by Lymphoprep (Nycomed; Oslo, Norway) density-gradient centrifugation. Flow cytometric analyses were performed as described previously (6). Briefly, recombinant MID 962-1200 (fragment F2, 5 mg/ml) in PBS adjusted with 0.1 M carbonate buffer to pH 9.2, was incubated with 0.15 g/ml FITC solubilized in DMSO. After 45 min at room temperature and constant stirring, the sample was diluted and subjected to a PD10 column (Amersham Biosciences) preequilibrated with PBS, pH 7.4. The resulting MID 962-1200 -FITC was used for binding studies. 10 g of the labeled fragment was incubated with PBLs in 100 l of PBS containing 1% fetal calf serum on ice for 30 min followed by incubation with specific RPE-conjugated anti-human CD19 mAbs. In blocking experiments, the PBLs were preincubated with 30 g of anti-human IgD-Fab polyclonal antibodies for 30 min on ice and were thereafter incubated with recombinant MID 962-1200 protein and RPE-conjugated anti-human CD19 mAbs. After final washes, the samples were analyzed in an EPICS XL-MCL flow cytometer (Beckman Coulter, Hialeah, Florida).
Ultracentrifugation-Sedimentation equilibrium analysis was performed in a Beckman Optima XL-A analytical ultracentrifuge equipped with absorbance optics, using an An50Ti rotor. MID 962-1200 was equilibrated in PBS buffer, whereas 0.1 M NH 4 HCO 3 containing 0.5% Empigen ® was used for native purified MID. Short column experiments were performed at two consecutive speeds (9,000 and 11,000 rpm). After sedimentation equilibrium was reached, absorbance scans were taken with 0.001-cm step size, 8 averages, at appropriate wavelengths (231, 236, 280, or 286 nm). Six-hole centerpieces of charcoal-filled Epon (12-mm optical path) were used, and the equilibrium temperature was 20°C. High-speed sedimentation at 48,000 rpm was conducted afterward for baseline correction. Estimates of the buoyant molar mass of the protein were determined by fitting a sedimentation equilibrium model for a single sedimenting solute to individual data sets with the program EQASSOC (13). These values were converted to the corresponding average molar masses using the partial specific volume calculated from the amino acid composition of the protein (14). Several models of self-association were globally fitted to multiple sedimentation equilibrium using the MULTEQ3B software (kindly supplied by Dr. Allen Minton, National Institutes of Health, Bethesda, MD).
Mutation Generation Using Transposons-Five extra amino acids were randomly inserted into the IgD-binding region of MID (fragment F2, MID 962-1200 ) using the Mutation Generation System (Finnzymes, Espoo, Finland) according to the manufacturer's instructions. Because the pET26b(ϩ) vector has an internal NotI site, MID 962-1200 was first subcloned into pUC18 (CLONTECH, Palo Alto, CA) using the restric-tion enzyme sites BamHI and HindIII. The artificial transposon Entranceposon was inserted with an in vitro transpositon reaction catalyzed by a single purified MuA transposase. The transposition reaction proceeded for 1 h at 30°C followed by transformation of HB101 (Bio-Rad) by electroporation. Since the Entranceposon codes for chloramphenicol resistance the clones could easily be selected. After plasmid preparation (Qiagen, Hilden, Germany), the insert was digested with BamHI and HindIII and separated by agarose gel electrophoresis. The inserts containing the Entranceposon were further cloned into pUC18 followed by transformation into DH5␣. The resulting clones were pooled, and the plasmid DNA was extracted. The Entranceposon was removed by digestion with NotI leaving 15 base pairs. DNA was ligated and the plasmids transformed into BL21(DE3). The resulting clones containing five extra amino acids at different positions were tested for specific IgD-binding capacity using colony-blots. Nitrocellulose blotting membranes (Sartorius; Göttingen, Germany) were incubated on the plates with E. coli clones. After 20 min, bacteria on the membranes were lysed with chloroform treatment for 30 min. The membranes were washed, blocked, and incubated with IgD as described above for Western blots. To determine the precise position of the five extra amino acids, the F2 fragment was amplified by PCR using M13 forward and reverse primers. The resulting PCR products were sequenced using capillary electrophoresis on a Beckman CEQ 2000 and a dye-terminator cycle sequencing kit (CEQ DTCS kit, BeckmanCoulter, Stockholm, Sweden). Editing and alignment of the resulting DNA sequences were performed using PHRED (CodonCode, Deadham) and SEQUENCHER (MedProbe, Oslo, Norway).
Protein Structure Analysis-The secondary structure of the MID 962-1200 (F2) fragment was predicted using a software at the home page for Computational Genomics Group (http://genomic.sanger. ac.uk/) (15).

Protease Digestion of MID Indicates the Localization of the
IgD-binding Region-In a previous paper, we demonstrated that MID specifically binds human IgD (6). To investigate which part of MID interacts with IgD, native MID isolated from M. catarrhalis strain Bc5 was digested with trypsin, chymotrypsin, or LysC. The protein mixtures were applied to SDS-PAGE and transferred to membranes followed by incubation with IgD and HRP-conjugated anti-IgD polyclonal antibodies. Several bands of different molecular weights were obtained. After amino acid sequencing, individual positions in MID could be discerned (Fig. 1A). Trypsin digestion resulted in a 40-kDa IgD-binding band and 35-and 80-kDa bands that did not interact with IgD. Furthermore, chymotrypsin treatment resulted in, inter alia, a 25-kDa non-IgD-binding band, whereas digestion with LysC gave, among others, a 66-kDa IgD-binding band. Sequencing of the 66-kDa MID fragment confirmed the position of the IgD-binding sequence by analogy with the trypsin-derived 40-kDa band. These initial experiments thus indi- Fragments attracting IgD (ϩ) or not binding IgD (Ϫ) are indicated to the right. In panel A, native MID was treated with proteases overnight, followed by analyses by SDS-PAGE and Western blot. Specific bands were excised from the gels, and the proteins were sequenced by Edman degradation or mass spectrometry. In panel B, 11 fragments were cloned and expressed in E. coli. The recombinant proteins were purified using nickel or amylose resins and analyzed for IgD binding in Western and dot blots. After blotting, the resulting filters were incubated with human IgD myeloma serum followed by incubation with a secondary HRP-conjugated polyclonal antibody and subsequent development using ECL.
cated that an IgD-binding region was located approximately between amino acid residues MID 864 and MID 1172 .
IgD Binding Is Preserved in 238 Amino Acids of MID-To determine in more detail the MID IgD-binding region, 11 sequences derived from the full-length MID were cloned and expressed in E. coli. The recombinant proteins covered the entire MID sequence, and the individual lengths and positions were as shown in Fig. 1B. The recombinant proteins comprising amino acid residues 69 -1111 (void of the signal peptide) or 1011-2139 of MID did not bind IgD as revealed in Western and dot blots. In contrast, the protein MID 902-1200 (protein fragment F1) attracted IgD, strongly suggesting that the single IgD-binding region of MID was within that particular sequence.
To pinpoint the sequence responsible for the IgD binding, the truncated MID 902-1200 was systematically shortened at the N-and C-terminal ends (Fig. 2). Equimolar concentrations of the various recombinant proteins were compared with native full-length MID 1-2139 isolated from M. catarrhalis. The different recombinant proteins were diluted in 4-fold steps, added to membranes, and incubated with human IgD. On a molar basis, an essentially preserved IgD-binding capacity was detected for the truncated MID protein stretching from amino acid residue 962 to 1200. The shortest truncated protein still interacting with IgD was localized between MID 985 and MID 1142 (fragment F6). The IgD binding property was lost when the N terminus was reduced to the MID 1000 residue (fragment F4) or when the C-terminal was shortened to MID 1130 (fragment F7). Finally, fragment F8 (MID 902-1130 ), with a longer N-terminal and a shorter Cterminal (compared with MID 985-1200 , fragment F3), was also manufactured and analyzed. However, this truncated MID did not interact with IgD, suggesting that the binding capacity was depending on a longer C-terminal.
To further characterize the specific MID-dependent IgD binding, an ELISA was constructed using human IgD as bait.

FIG. 3. The IgD-binding fragment MID 962-1200 (F2) is a tetramer as revealed by ultracentrifugation and PAGE at native conditions.
A, MID 962-1200 was subjected to ultracentrifugation, and the obtained results were compared with theoretical values for a tetramer and dimer (broken lines). B, MID 962-1200 ran as a tetramer at non-reducing conditions. The ultracentrifugation was performed on a Beckman Optima XL-A analytical ultracentrifuge with absorbance optics as described in detail under "Material and Methods." In B, recombinant MID 962-1200 was analyzed by gel electrophoresis under native conditions and a corresponding Western blot using IgD() as probe and secondary HRP-conjugated anti-IgD antibodies. Transferrin and albumin were included as molecular mass markers running at 80 and 66 kDa, respectively. DNA encoding for the various truncated MID proteins were cloned into the expression vector pET26b(ϩ) and produced in E. coli. The recombinant proteins containing histidine tags were purified and dot blotted onto a nitrocellulose membrane. The membrane was probed with human IgD() followed by secondary HRP-conjugated polyclonal antibodies that were used for detection.
All of the recombinant truncated MID fragments were subjected to ELISA followed by incubation with a specific rabbit anti-serum directed against MID 902-1200 . The ELISA was developed using HRP-conjugated goat anti-rabbit polyclonal antibodies. The same pattern as with the dot blot (Fig. 2) was observed, i.e. fragments F4, F7, and F7 were not attracted to the solid-phase IgD, whereas the other fragments bound to a variable degree compared with full-length MID (not shown).

Native MID and the Truncated MID 962-1200 Are Tetramers as Shown by Sedimentation Equilibrium Analysis and Gel
Electrophoresis-In a previous publication, we estimated with chromatography and SDS-PAGE that native MID is aggregating and exists as an oligomer (6). To confirm these results, a sedimentation equilibrium analysis on MID was performed. Purified MID, which was solubilized in the detergent Empigen ® (0.5%), was subjected to gradient centrifugation at two different speeds in an ultracentrifuge supplied with absorbance optics. An apparent molar mass of ϳ1,000 kDa was obtained, and it is concluded that native MID forms an oligomer.
In addition to full-length MID, we subjected the truncated MID 962-1200 (F2) to a sedimentation equilibrium analysis. The data obtained were compared with the calculated values for a tetramer or dimer. The experimental gradient for MID 962-1200 protein (27 M) as well as the theoretical ones for dimer and tetramer species are shown in Fig. 3A. The predominant MID 962-1200 species sedimented at equilibrium with an average molecular mass of 110 Ϯ 3 kDa. This was compatible with a protein tetramer because MID 962-1200 including a histidine tag has a molecular mass of 27 kDa as calculated from the amino acid composition. To analyze the putative quaternary structure of MID 962-1200 (F2) for IgD binding, the recombinant protein was subjected to analysis in gel electrophoresis under native conditions. In addition, a corresponding IgD Western blot analysis was performed. MID 962-1200 migrated as a protein with an apparent molecular mass of ϳ100 kDa in the native gel (Fig. 3B), confirming the ultracentrifugation experiments. Thus, in addition to the sedimentation equilibrium analysis, the experiment with the native gel suggested that MID 962-1200 in its native form was a tetramer consisting of four molecules. Furthermore, using native conditions, the tetrameric structure of MID 962-1200 strongly bound IgD (Fig. 3B).
Optimal MID 962-1200 -IgD Interaction Is Dependant on Tetramer Structure-To further shed light upon the need for a tetramer structure to obtain an optimal IgD binding, MID 962-1200 (F2) was incubated at 60 or 100°C followed by analysis on SDS-PAGE and Western blots. MID 962-1200 formed both a monomer and a tetramer after pretreatment at 60°C (Fig. 4A). The tetrameric structure was, however, dis-  4. A tetrameric structure of MID 962-1200 (F2) is a prerequisite for optimal IgD binding. A, SDS-PAGE of MID 962-1200 after treatment at 60°C separates monomers and tetramers. After heat treatment at 100°C monomers only can be detected. B, Corresponding Western blot with IgD as probe reveals weak IgD binding to monomers. C, mean IgD binding to tetramers and monomers, respectively, in six different experiments. IgD binding is shown as arbitrary units/g protein. Error bars indicate S.D. MID 962-1200 was treated in SDS sample buffer at 60 or 100°C for 10 min and subjected to SDS-PAGE and Western blot analysis. No difference was found in blotting efficiency between the monomeric and tetrameric forms. The resulting Coomassie-stained gel and Western blot were analyzed by densitometry. The percentage of protein migrating as tetramer or monomer was calculated and compared with the IgD-binding capacity. rupted at 100°C and resulted in a monomeric form, which displayed a considerably weaker binding to IgD when examined in Western blots (Fig. 4, A and B). To investigate the capability of the tetramer to bind IgD in comparison with the monomeric form, the MID 962-1200 fragment was subjected to analysis at 60°C in six different experiments. The heattreated protein was subjected to SDS-PAGE, and the IgD binding activity was analyzed by Western blots. Resulting gels and filters were analyzed by densitometry, and the protein concentration (density) of the monomer was divided with the corresponding tetramer concentration. The obtained value (%) was related to the concentration (g) of total protein loaded on the gels. Interestingly, when IgD binding to the tetrameric respectively monomeric forms were compared, a 23-fold more efficient binding to IgD was found with the tetrameric MID 962-1200 (Fig. 4C).
The Specific IgD Binding of MID 962-1200 (F2) Includes ␣-Helix Structures-The MID 962-1200 sequence was subjected to computer analysis (Fig. 5A). A possible secondary structure of the truncated MID including four ␣-helices and several ␤-strands was observed. Data based upon sequence analysis suggested that a reduction in the number of amino acids at the N-and C-terminals of MID 962-1200 would thus destroy the secondary structure, causing a weaker IgD binding that would eventually disappear. Therefore, five amino acid residues were randomly introduced into MID 962-1200 using transposons and a MuA transposase in vitro. A fraction of the resulting E. coli clones (n ϭ 90) were examined for IgD binding, and their MID 962-1200 -containing plasmids were sequenced. Two different groups were consequently found, i.e. binding and nonbinding clones (Fig. 5B). When the sequence MID 982-1059 was interrupted with the five extra amino acids, IgD binding was completely abolished. Another sequence (MID 1100 -1145 ) that was upstream of the molecule was also found to be important for IgD binding. Interestingly, those sequences included ␣-helix structures. In contrast, MID 962-1200 still attracted IgD when five extra amino acid residues were introduced near the N-terminal region or between MID 1059 and MID 1100 , i.e. two sequences not containing ␣-helices. Insertion of extra amino acids within the end of the C-terminal-flanking region did not considerably influence IgD binding, a finding that was partly consistent with the truncated MID fragments shorter than MID 962-1200 . Computer analysis revealed that when five amino acid residues were introduced into sequences responsible for ␣-helix formation, that particular structure was disrupted. Thus, the secondary structure analysis (Fig. 5A) compared with the data obtained with transposons confirmed the results seen with the truncated MID protein series (Fig. 2).
MID 962-1200 (F2) Specifically Binds to Human IgD and Is Attracted to the IgD BCR-Full-length MID 1-2139 attracts human IgD but not immunoglobulins from the other subclasses (6). To verify the specificity of the interaction between IgD and recombinant MID 962-1200 , a dot blot experiment was performed. The membrane was coated with the truncated MID protein followed by incubation with IgD, IgG, IgM, or IgA myeloma sera. As shown in Fig. 6, MID 962-1200 interacted with the three IgD myeloma sera, whereas MID 962-1200 -dependent binding was not detected to myeloma sera representing the IgG, IgM, or IgA subclasses. The next step was to investigate whether truncated MID also was attracted to membrane-bound IgD, i.e. the IgD BCR. MID 962-1200 was conjugated to FITC followed by incubation with purified PBLs. In parallel, cells were labeled with RPEconjugated anti-CD19 monoclonal antibodies. After washes, PBLs were subjected to flow cytometry analysis. Interestingly, FIG. 6. MID 962-1200 (F2) specifically binds to human IgD but not to IgG, IgM, or IgA. The truncated MID 962-1200 protein was recombinantly expressed in E. coli and purified. Thereafter, MID 962-1200 was applied in 4-fold dilutions to a nitrocellulose membrane. Resulting filters were incubated with different myeloma sera representing the different Ig classes followed by incubation with HRP-conjugated antihumanand -light chain polyclonal antibodies. IgD B.W. is a polyclonal standard serum from Behringwerke.  (CD19-RPE). B, CD19 ϩ lymphocytes specifically attracted FITC-conjugated MID 962-1200 . C, decreased MID 962-1200 -FITC binding was detected when the IgD B cell receptors were blocked with polyclonal antibodies (anti-IgD-Fab-pAb). PBLs were purified from heparinized human blood using Ficoll gradients. Incubation with Abs and MID 962-1200 -FITC was followed by washings and analysis by flow cytometry. A typical experiment out of three performed is shown.
FITC-conjugated MID 962-1200 significantly bound to the CD19 ϩ B lymphocyte population (Fig. 7B). To ensure that MID 962-1200 -FITC specifically interacted with the IgD BCR, PBLs were also preincubated with anti-human IgD-Fab polyclonal antibodies. As seen in Fig. 7C, the MID 962-1200 -FITC-dependent IgD binding was completely abolished when the IgD BCR was blocked by anti-IgD-Fab pAb. These results were in agreement with the previously published data on full-length MID (6). DISCUSSION The novel outer membrane protein MID from M. catarrhalis recently has been identified and shown to display a strong and specific affinity for human IgD (6). In the present paper, we use proteolytic fragments of MID and recombinantly expressed proteins to demonstrate that the IgD binding was essentially preserved in a truncated recombinant protein located between amino acid residues 962 and 1200 of MID (Fig. 2). The molecular mass of the smallest protein still binding IgD comprised 157 amino acids. The IgD-binding capacity of the recombinant proteins was related to individual lengths, i.e. the binding decreased dramatically with the length. This may be because the binding is highly dependant on a specific molecular structure of the IgD-binding region. This hypothesis was further strengthened by our ultracentrifugation and gel electrophoresis experiments. We have shown previously that native MID isolated from M. catarrhalis, in addition to the 200-kDa band, forms a band that seemingly is an oligomer with a molecular mass of ϳ1,000 kDa in SDS-PAGE (6). When purified MID solubilized in the detergent Empigen ® (0.5%) was analyzed by ultracentrifugation, an apparent molecular mass of ϳ1,000 kDa was observed. This molecular mass was consistent with a tetramer of MID in the presence of Empigen ® . The smallest truncated MID protein (MID 962-1200 , fragment F2) with an essentially preserved IgD binding activity (compared with fulllength MID) and which could be analyzed by ultracentrifugation in the absence of detergent was shown to be a tetramer under native conditions. Interestingly, when tetrameric and monomeric forms of MID 962-1200 were separated by SDS-PAGE and analyzed for IgD binding, the tetramer was shown to attract IgD Ͼ20-fold more efficiently compared with the monomeric form. Thus, the tetrameric structure was considerably more active compared with the monomer. We cannot fully exclude, however, that the monomeric protein was completely unreactive because it may refold to a tetrameric structure during the blotting process.
The MID region required for IgD binding is large compared with other bacterial surface proteins that bind immunoglobulins. Nonimmune Ig binding activity was first observed in Staphylococcus aureus with the discovery of protein A that attracts IgG (16). Protein A interacts with the Fc part of IgG, but only with subclasses 1, 2, and 4 (17). In addition, protein A binds a fraction of Ig molecules of all classes because of so-called alternative binding to the variable heavy chains (18). Protein A is a 45-kDa protein composed of five extracellular domains as tandem repeats and an additional transmembrane domain. Each of the extracellular domains consists of 58 -61 amino acids. When one of the domains (D) were cloned and expressed in E. coli, it was discovered that this small protein exhibited both the VH3 Fab-and Fc-binding specificities (19). In addition, competition ELISA showed that VH3 binding of this single domain did not interfere with Fc binding or vice versa. In contrast to protein A, protein G found in group C and G streptococci binds with high specificity to all human subclasses of IgG Fc (20). The smallest synthetic peptide of protein G still binding IgG Fc derived from the C-terminal region is only 11 residues long and corresponds to amino acid residues 34 -44 in the C1 domain (21). However, partly in parallel with our results with MID, the C1 domain in itself exhibits weaker binding compared with protein G as a whole (22).
Protein L derived from Peptostreptococcus magnus binds to -light chains of IgG, IgM, and IgA (23). The Ig binding is mediated through several highly homologous domains (24), each consisting of 72-76 amino acid residues. Another Igbinding protein, protein Sir derived from group A streptococci, binds to human IgA and IgG of all subclasses (25). The IgA-binding domain of protein Sir recently has been characterized in detail and shown to consist of 29 amino acids (26). However, protein Sir was found to contain the same 29 residues as another streptococcal protein (Arp4), which did not interact with IgA in itself. When 10 extra amino acid residues were added at each end of the 29-amino acid residue sequence and in addition a C-terminal cysteine residue, a coiled-coil dimer could be formed, and the resulting 50-amino acid long peptide strongly interacted with IgA.
Certain structural features may contribute to optimal MID 902-1200 -dependent IgD binding and can be dissected into three possible factors: i) the intrinsic binding capacity of monomers to bind IgD; (ii) oligomerization, independent of binding activity; and (iii) new determinants formed as a result of oligomer formation that contribute added binding avidity. The fact that IgD binding to truncated MID proteins requires 157 amino acid residues or more does not necessarily mean that all of those amino acids are involved in the IgD binding. However, when we randomly introduced five amino acid residues using transposons, the extra amino acids in two different sequence stretches completely abolished IgD binding. No similarities between the two amino acid stretches were found. The transposon results support together with the ultracentrifugation experiments that the overall tetrameric structure of the molecule is of high importance for the IgD interaction. Taken together, our data suggest that the tetramer formation of MID 962-1200 results in the formation of new determinants maximizing the binding avidity.
In conclusion, we have shown that the IgD-binding region of the MID protein comprises a 238-amino acid long sequence that is not repeated in MID. Furthermore, tetramer formation is of fundamental importance for optimal IgD binding.