Evidence for Plasticity and Structural Mimicry at the Immunoglobulin Light Chain-Protein L Interface

Abstract

The multidomain bacterial surface protein L (PpL) is a virulence factor expressed by only 10% ofPeptostreptococcus magnus strains, and its expression is correlated with bacterial vaginosis. The molecular basis for its ability to recognize 60% of mammalian immunoglobulin light chain variable regions (VL) has been described recently by x-ray crystallography, which suggested the presence of two VL binding sites on each protein L domain (Graille, M., Stura, E. A., Housden, N. G., Beckingham, J. A., Bottomley, S. P., Beale, D., Taussig, M. J., Sutton, B. J., Gore, M. G., and Charbonnier, J. (2001) Structure9, 679–687). Here, we report the crystal structure at 2.1 Å resolution of a protein L mutant complexed to an Fab′ fragment with only 50% of the VL residues interacting with PpL site 1 conserved. Comparison of the site 1 interface from both structures shows how protein L is able to accommodate these sequence differences and therefore bind to a large repertoire of Ig. The x-ray structure and NMR results confirm the existence of two VL binding sites on a single protein L domain. These sites exhibit a remarkable structural mimicry of growth factors binding to their receptors. This could explain the protein L superantigenic activity on human B lymphocytes. The multidomain bacterial surface protein L (PpL) is a virulence factor expressed by only 10% ofPeptostreptococcus magnus strains, and its expression is correlated with bacterial vaginosis. The molecular basis for its ability to recognize 60% of mammalian immunoglobulin light chain variable regions (VL) has been described recently by x-ray crystallography, which suggested the presence of two VL binding sites on each protein L domain (Graille, M., Stura, E. A., Housden, N. G., Beckingham, J. A., Bottomley, S. P., Beale, D., Taussig, M. J., Sutton, B. J., Gore, M. G., and Charbonnier, J. (2001) Structure9, 679–687). Here, we report the crystal structure at 2.1 Å resolution of a protein L mutant complexed to an Fab′ fragment with only 50% of the VL residues interacting with PpL site 1 conserved. Comparison of the site 1 interface from both structures shows how protein L is able to accommodate these sequence differences and therefore bind to a large repertoire of Ig. The x-ray structure and NMR results confirm the existence of two VL binding sites on a single protein L domain. These sites exhibit a remarkable structural mimicry of growth factors binding to their receptors. This could explain the protein L superantigenic activity on human B lymphocytes. Gram-positive bacterial proteins that interact with immunoglobulins have been widely studied and utilized in immunology and biotechnology. Among the best known are Staphylococcus aureus protein A (SpA) 1The abbreviations used are: SpA, Staphylococcus aureus protein A; SpG, streptococcal protein G; PpL, Peptostreptococcus magnus protein L; VL, variable light chain region; VH, variable heavy chain region; CL, constant light chain region; CH, constant heavy chain region; Fv, variable fragment; Fc, constant fragment; PpLwt, wild type protein L domain; PpLD55A, protein L domain containing the Asp to Ala substitution at position 55; MPEG 5000, methoxypolyethylene glycol 5000; rmsd, root mean square deviation; TROSY, transverse relaxation optimized spectroscopy; HSQC, heteronuclear single quantum correlation 1The abbreviations used are: SpA, Staphylococcus aureus protein A; SpG, streptococcal protein G; PpL, Peptostreptococcus magnus protein L; VL, variable light chain region; VH, variable heavy chain region; CL, constant light chain region; CH, constant heavy chain region; Fv, variable fragment; Fc, constant fragment; PpLwt, wild type protein L domain; PpLD55A, protein L domain containing the Asp to Ala substitution at position 55; MPEG 5000, methoxypolyethylene glycol 5000; rmsd, root mean square deviation; TROSY, transverse relaxation optimized spectroscopy; HSQC, heteronuclear single quantum correlation (1Langone J.J. Adv. Immunol. 1982; 32: 157-252Google Scholar), streptococcal protein G (SpG) (2Bjorck L. Kronvall G. J. Immunol. 1984; 133: 969-974Google Scholar), and Peptostreptococcus magnus protein L (PpL) (3Bjorck L. J. Immunol. 1988; 140: 1194-1197Google Scholar). They contain from two to five successive and highly homologous Ig-binding domains and recognize distinct regions on a wide portion of Ig. Protein L, a surface protein from Peptostreptococcus magnus, is 76–106 kDa in size and contains four to five (depending on the strain) highly homologous consecutive extracellular domains (4Murphy J.P. Duggleby C.J. Atkinson M.A. Trowern A.R. Atkinson T. Goward C.R. Mol. Microbiol. 1994; 12: 911-920Google Scholar, 5Kastern W. Sjobring U. Bjorck L. J. Biol. Chem. 1992; 267: 12820-12825Google Scholar). It binds with high affinity and high specificity to mammalian κ light chain variable (VLκ) regions (6Akerstrom B. Bjorck L. J. Biol. Chem. 1989; 264: 19740-19746Google Scholar, 7Nilson B.H. Solomon A. Bjorck L. Akerstrom B. J. Biol. Chem. 1992; 267: 2234-2239Google Scholar). Although expressed by a minority ofP. magnus strains, PpL appears to be a virulence factor because its presence is correlated with enhanced vaginal colonization (8Ricci S. Medaglini D. Marcotte H. Olsen A. Pozzi G. Bjorck L. Microb. Pathog. 2001; 30: 229-235Google Scholar) and the concomitant occurrence of bacterial vaginosis (9Kastern W. Holst E. Nielsen E. Sjobring U. Bjorck L. Infect. Immun. 1990; 58: 1217-1222Google Scholar). PpL, like SpA and SpG, is an important tool for purification, recognition, and removal of immunoglobulins, but it has a different and unique specificity. SpA and SpG domains recognize a common site at the interface between the CH2 and CH3 domains on the Fc of IgG (Fcγ) (10Deisenhofer J. Biochemistry. 1981; 20: 2361-2370Google Scholar, 11Sauer-Eriksson A.E. Kleywegt G.J. Uhlen M. Jones T.A. Structure. 1995; 3: 265-278Google Scholar), but they also interact with functionally distinct regions on the Fab portion of particular Ig families or subclasses (SpA does not bind to most IgG3, whereas SpG does) (12Derrick J.P. Wigley D.B. Nature. 1992; 359: 752-754Google Scholar, 13Graille M. Stura E.A. Corper A.L. Sutton B.J. Taussig M.J. Charbonnier J.B. Silverman G.J. Proc. Natl. Acad. Sci. U. S. A. 2000; 97: 5399-5404Google Scholar). SpG recognizes the Fab portion through binding to the CH1 domain of IgG, whereas SpA interacts with the VH domain on 12% of mouse and 50% of human Fv fragments. Although these two latter interactions have relatively low affinity (K d of 200 nm), they have been used in specific cases for Fab purification and phage library validation. PpL does not share Fc binding ability with SpA and SpG, but its high affinity (K d of 1 nm) for the variable light chain regions of more than two thirds of mouse and ∼50% of human Ig, irrespective of their class, makes it particularly interesting for the purification of Fab fragments and single chain antibodies (scFv). To understand better how PpL can bind a large population of light chains, we recently reported the crystal structure of the complex between a PpL single domain and a human Fab (14Graille M. Stura E.A. Housden N.G. Beckingham J.A. Bottomley S.P. Beale D. Taussig M.J. Sutton B.J. Gore M.G. Charbonnier J. Structure. 2001; 9: 679-687Google Scholar). Surprisingly, in this structure the PpL domain is sandwiched between two antibody light chains in a manner reminiscent of cytokines binding to their receptors at the cell surface (15Cunningham B.C. Ultsch M. De Vos A.M. Mulkerrin M.G. Clauser K.R. Wells J.A. Science. 1991; 254: 821-825Google Scholar, 16de Vos A.M. Ultsch M. Kossiakoff A.A. Science. 1992; 255: 306-312Google Scholar, 17Syed R.S. Reid S.W. Li C. Cheetham J.C. Aoki K.H. Liu B. Zhan H. Osslund T.D. Chirino A.J. Zhang J. Finer-Moore J. Elliott S. Sitney K. Katz B.A. Matthews D.J. Wendoloski J.J. Egrie J. Stroud R.M. Nature. 1998; 395: 511-516Google Scholar). From the two sites identified, only site 1 was already documented (18Wikstrom M. Sjobring U. Drakenberg T. Forsen S. Bjorck L. J. Mol. Biol. 1995; 250: 128-133Google Scholar, 19Beckingham J.A. Bottomley S.P. Hinton R. Sutton B.J. Gore M.G. Biochem. J. 1999; 340: 193-199Google Scholar, 20Beckingham J.A. Housden N.G. Muir N.M. Bottomley S.P. Gore M.G. Biochem. J. 2001; 353: 395-401Google Scholar). To investigate the functional relevance of site 2, we have designed a PpL mutant (PpL D55A; hereafter PpLD55A) where substitution of Asp-55 by Ala should disrupt a salt bridge and a hydrogen bond that was believed to be crucial for the interaction with PpL site 2. Thus, the PpLD55A mutant should retain only site 1 binding for VL. This mutant was built on the Y64W template, where Trp-64 is an efficient fluorescent probe to measure K d by stopped flow methods (20Beckingham J.A. Housden N.G. Muir N.M. Bottomley S.P. Gore M.G. Biochem. J. 2001; 353: 395-401Google Scholar). Here, we present the complex between the PpLD55A mutant and a mouse Fab′ fragment bearing a VLκ9 region to 2.1Å resolution. The structure shows how PpL is able to accommodate differences in light chain sequences. The PpLD55A mutant indeed shows a single binding mode in the crystal. To confirm the relevance of the second site in solution, we have studied the interaction between VL chains and different PpL domains by heteronuclear NMR spectroscopy. Finally, we compare the residues involved in the first and second interfaces and reveal a strong homology between the two PpL interaction patches. Cloning of the 19D9D6 antibody (IgG1,κ) light chain was performed by standard methodology using classical PCR amplification as described previously (21Bettsworth F. Monnet C. Watelet B. Battail-Poirot N. Gilquin B. Jolivet M. Menez A. Arnaud M. Ducancel F. J. Mol. Recognit. 2001; 14: 99-109Google Scholar). The cDNA encoding the Fd domain of the 19D9D6 antibody was obtained using the rapid amplification of cDNA ends (RACE) method, as described previously (22Ruberti F. Cattaneo A. Bradbury A. J. Immunol. Methods. 1994; 173: 33-39Google Scholar). The PCR products were obtained using the Taq DNA polymerase (Amersham Biosciences) and directly ligated into the pCR®2.1 cloning vector using a TA cloning kit (Invitrogen) according to the manufacturer's instructions. The sequences of cloned DNA were determined by sequencing on an ABI 310 automatic sequencer using a Dye Terminator Cycle Sequencing Ready Reaction Kit (Applied Biosystems). The antibody 19D9D6 (IgG1,κ) was produced in vitro and purified as follows (23Jolivet-Reynaud C. Dalbon P. Viola F. Yvon S. Paranhos-Baccala G. Piga N. Bridon L. Trabaud M.A. Battail N. Sibai G. Jolivet M. J. Med. Virol. 1998; 56: 300-309Google Scholar). Briefly, the supernatant was diluted 1:2 (v/v) in a high salt on a protein A (Amersham Biosciences) and at the manufacturer's instructions. The fragment was obtained by of the IgG by The purified IgG was with in an and for at The was by The fragment was in the supernatant and purified by size on a 200 (Amersham The Fab′ fragment was obtained by with and of the residues with The Fab′ was purified by size on The PpLD55A and were by on the Y64W or wild type DNA (20Beckingham J.A. Housden N.G. Muir N.M. Bottomley S.P. Gore M.G. Biochem. J. 2001; 353: 395-401Google Scholar) using the and The sites of the are in were by DNA sequencing using the DNA sequencing kit were purified as as described previously S.P. Beckingham J.A. J.P. Atkinson M. Atkinson T. Hinton Gore M.G. 1995; Scholar). with the were by in were and in with and Protein expression was by of a 1 at with were a PpL was purified as described in S.P. Beckingham J.A. J.P. Atkinson M. Atkinson T. Hinton Gore M.G. 1995; NMR were on a NMR for and were and 1 in and were was to the to of and and correlation were using the sequence by and were for and for the The Fab′ was in the by of Fab′ 19D9D6 with of PpLD55A were by at in by of a containing 10% MPEG 5000, with of the protein for were by using by E.A. J. 1991; Scholar). were into a containing of of 50% MPEG 5000, and of and 50% MPEG were at K from to and to 2.1Å resolution on and were using the W. Scholar). to the are in and cell for the is in and in the are and and in the are and of for not in the for the is in and in the are and of for not in the in a The Å was used to the structure by the molecular with the J. A. 1994; Scholar). the from the variable and constant regions of the Fab′ 19D9D6 structure E.A. M. Housden N.G. Gore M.G. Biol. 58: Scholar) as the two Fab′ present in the were to position the two PpL were an of this were with the A. C. Scholar). These to a single PpL domain in the of the VL region from both Fab′ The PpL domains were into these by on both Fab′ the previously structure of the complex between a human Fab and the domain (14Graille M. Stura E.A. Housden N.G. Beckingham J.A. Bottomley S.P. Beale D. Taussig M.J. Sutton B.J. Gore M.G. Charbonnier J. Structure. 2001; 9: 679-687Google Scholar) with the E.A. J. Mol. Biol. Scholar). The two present in the were first the Å resolution with of to the with the of the domains involved in the interfaces and from to 2.1 Å resolution The structure was with the P. J. M. T. Biol. 1998; Scholar), and the was in for the structure are in The contains and glycol from the residues from to for the first PpLD55A domain and from to for the second PpLD55A domain. for PpL domain structures to (14Graille M. Stura E.A. Housden N.G. Beckingham J.A. Bottomley S.P. Beale D. Taussig M.J. Sutton B.J. Gore M.G. Charbonnier J. Structure. 2001; 9: 679-687Google Scholar, M. Drakenberg T. Forsen S. Sjobring U. Bjorck L. Biochemistry. 1994; Scholar, D. Zhang Biol. 2001; Scholar), the residues are and from the the of from the CH1 region and to a the region the is in the but from each complex chain that regions of the as by the J. Scholar). The only is Ala from the which has been to in crystal structures of Fab fragments. The VL is between the two Fab′ with a root mean square of as by the E.A. J. Mol. Biol. Scholar). both PpLD55A domains on each with an of with the for the and which is in The PpLD55A domain the structure as the (14Graille M. Stura E.A. Housden N.G. Beckingham J.A. Bottomley S.P. Beale D. Taussig M.J. Sutton B.J. Gore M.G. Charbonnier J. Structure. 2001; 9: 679-687Google Scholar), and the domain to resolution according to D. Zhang Biol. 2001; with of for each were to be at the interface they two with from the S. Biol. 1994; is with a from the the is or it at in the The surface were with the from S. Biol. 1994; Scholar). bond were using the S. Biol. 1994; Scholar) with a of Å and according to the of and J. Mol. Biol. 1994; Scholar). The Fab′ 19D9D6 a VL chain to the which is to the human The contains two Fab′ by The two on each with a of Å with the for the PpL with on the with the domain which two Fab in the complex with human Fab (14Graille M. Stura E.A. Housden N.G. Beckingham J.A. Bottomley S.P. Beale D. Taussig M.J. Sutton B.J. Gore M.G. Charbonnier J. Structure. 2001; 9: 679-687Google Scholar), the PpLD55A mutant binds to only Fab′ This interaction PpL site 1 on the of the domain from the The interactions by PpLD55A with the VL region of Fab′ 19D9D6 are different from for Fab complex as a of differences in their κ chain sequences. of the VL residues involved in the Fab complex are different in the VLκ9 their is is in Ala is is is is and is 1 These not the structure of the which of Å The PpL and the VL A are and are through hydrogen between chain from both a This a from the VL region into the PpL as was in the Fab structure (14Graille M. Stura E.A. Housden N.G. Beckingham J.A. Bottomley S.P. Beale D. Taussig M.J. Sutton B.J. Gore M.G. Charbonnier J. Structure. 2001; 9: 679-687Google Scholar). Although the binding mode is interactions are residues from Fab′ 19D9D6 and from PpLD55A) are involved in interactions in this structure with in the complex from the Fab and from the of which is an in Fab of these were already A and hydrogen of are that the of the and PpL to the This that these hydrogen between PpL and the VL chain are important for the of a large population of regions in a interactions in the VL PpLD55A complex of the complex of the between the human and VLκ9 are in in a between the human and VLκ9 are in Surprisingly, the and regions are in Fab′ the hydrogen from to and from to in the Fab complex are in this structure because of the of both hydrogen for this These are by the of the of PpL and of the region the VL to the domain to These have been to be the most PpL and Fab′ NMR have that PpL has a than and M. Drakenberg T. Forsen S. Sjobring U. Bjorck L. Biochemistry. 1994; Scholar, M. Forsen S. Drakenberg T. J. Biochem. Scholar). the two crystal this PpL two and M. Stura E.A. Housden N.G. Beckingham J.A. Bottomley S.P. Beale D. Taussig M.J. Sutton B.J. Gore M.G. Charbonnier J. Structure. 2001; 9: 679-687Google Scholar). the the portion of PpL is to the VL A and makes two hydrogen The first the Ala to the PpL the interaction by a hydrogen bond and The second PpL to the a the of PpL is more than in the Fab structure the of the chain of from the region with the of its chain a hydrogen bond between and VL and previously to PpL in the 19D9D6 at position the from in Fab to in the Fab′ 19D9D6 a with by the a in the in Fab′ 19D9D6 in Fab the interaction between position a from the and from in crystal as a of a in crystal the of the Fab′ complexed to PpL mutant is and position its in site G. Gore, in This that this interaction is of for complex position a an from site 1 in the Fab complex is involved in interaction with PpL and makes a salt bridge to from light chain to site the Fab′ the chain of is as by the of strong and does not This is not to the of site 2, because in the complex with PpL mutant that binds to two the from the Fab′ 19D9D6 to site 1 does not the Fab′ 19D9D6 to PpL site This results in a in the from for to for VLκ9 at site the two VL regions in complex with PpL share only 50% sequence the the interface is conserved. and chain the but a of residues from both the ability of PpL to its This of as as 50% of the human immunoglobulin because the and only VL the interaction is sequence and only light chains with an PpL and the structure the PpL region for site and of the is and is only involved in a crystal that the has an important for site Thus, the of the Asp-55 this mutant is to interact through the large surface in the Fab complex but a crystal could be to explain the of site it could be a crystal and it could be to sequence differences between the VL regions of Fab and Fab′ are by the of the crystal of PpL in complex with the Fab′ E.A. M. Housden N.G. Gore M.G. Biol. 58: Scholar). The two PpL mutant domains with site residues to at position or to at position are both in site are sandwiched between two Fab′ 19D9D6 as for the complex between human Fab and M. Stura E.A. Housden N.G. Beckingham J.A. Bottomley S.P. Beale D. Taussig M.J. Sutton B.J. Gore M.G. Charbonnier J. Structure. 2001; 9: 679-687Google and The where residues Asp-55 and have been to Ala and in a different crystal and binds to the Fab′ only through site 1 This strong to the that the at position a in site 2, and its by an is to disrupt this To x-ray we have used heteronuclear NMR to the VL binding site of PpL with human in with the K. R. G. K. Proc. Natl. Acad. Sci. U. S. A. Scholar). obtained by this should a of in the complex with a in with NMR (18Wikstrom M. Sjobring U. Drakenberg T. Forsen S. Bjorck L. J. Mol. Biol. 1995; 250: 128-133Google Scholar). have the on two different PpL and not PpLD55A because of VL binding at two different and a both and a of complexed and as from the of at the and the of a of and A of the to a the with the of from the residues This is with the of a complex of molecular that does not the residues as by (14Graille M. Stura E.A. Housden N.G. Beckingham J.A. Bottomley S.P. Beale D. Taussig M.J. Sutton B.J. Gore M.G. Charbonnier J. Structure. 2001; 9: 679-687Google Scholar). the at the This is a that the binding of to chains has been to an of the VL binding sites on PpL, the and at to be the most have on residues from PpL site Asp or and these residues in a of with the these are in the a for both and is that at site 2. The NMR results different binding for and They also confirm the existence of site 2, which is by the Asp-55 to substitution most by the Asp to This the of VL binding sites on the PpL from five to from four to on the P. magnus its PpL was to only VL binding site on each and the were on this These results are important and a more of the activity of these two PpL is able to interact with antibodies the site in a manner reminiscent of SpG and in the binding by SpA and SpG site binding PpL to which to the bacterial protein and the from this in the of these to be like SpA, PpL can as a through its Fab binding A. J.P. G. B. Bjorck L. G. Infect. Immun. 2000; Scholar). the Fab PpL the VL to 1 B from both in the complex and a for site 1 and site of the VL region of the two in complex with both PpL sites into a strong between the two Asp-55 on on and the residues have a This that both of which have been to binding at site J.A. Bottomley S.P. Hinton R. Sutton B.J. Gore M.G. Biochem. J. 1999; 340: 193-199Google Scholar, E.A. M. Housden N.G. Gore M.G. Biol. 58: Scholar). The structural mimicry between the two interfaces is also at the structure PpL binding sites a and of an A from the VL region is to PpL but to PpL These interactions are of the type with four hydrogen between site 1 of PpLD55A and Fab′ 19D9D6 with only two hydrogen between site of and Fab The can a of to which residues in a PpL domain are in each of the two light chain binding sites and or not the two sites share It is to that the two sites not to be the of but the of functionally structural Although a strong structural mimicry has not been have two binding sites to the The extracellular of 1 like human growth and bind to their receptors with a (15Cunningham B.C. Ultsch M. De Vos A.M. Mulkerrin M.G. Clauser K.R. Wells J.A. Science. 1991; 254: 821-825Google Scholar, 16de Vos A.M. Ultsch M. Kossiakoff A.A. Science. 1992; 255: 306-312Google Scholar, 17Syed R.S. Reid S.W. Li C. Cheetham J.C. Aoki K.H. Liu B. Zhan H. Osslund T.D. Chirino A.J. Zhang J. Finer-Moore J. Elliott S. Sitney K. Katz B.A. Matthews D.J. Wendoloski J.J. Egrie J. Stroud R.M. Nature. 1998; 395: 511-516Google Scholar). the two different sites on the regions of the receptors. in the of and the strong mimicry described is not The interactions better the interaction of with its because of the of and the between the two binding sites of for Aoki K.H. T. J. Biochemistry. are the of a PpL domain binding to two the presence of a second binding site a to a protein that already has to five Ig binding the of it has been that not to Stura E.A. Liu M.A. W. S. Biol. 1998; Scholar). It is that a between the manner in which PpL interacts with B cell receptors and the that cytokines interact with their receptors. From a structural of a of domains not a in the that receptors The is and these results of the by PpL to interact with of the human B cell repertoire sequence at the VL of x-ray and NMR we confirm the existence of two VL binding sites on a single PpL domain. the from and at the are to F. A. and S. for at B. Gilquin and A. for and and for in of this