Incorporation of long CDR3s into V domains: implications for the structural evolution of the antibody-combining site

Available data suggest that 'primitive' antibody-combining sites often include longer than average HCDR3s. Long HCDR3 sequences have been reported in diverse vertebrates, including humans, cattle, camels and sharks. These long HCDR3 segments contain unusual sequence features such as stretches of Gly or Pro residues and multiple Cys residues. We examined how longer than average HCDR3s were accommodated in the V domains of human, murine and camel antibodies with known three-dimensional structures. The main conclusions were that (1) HCDR3s longer than 12 residues should protrude outward from the V domains; (2) descending HCDR3 polypeptides may utilize VL (including LCDR3) constituents as a platform, supporting the protruding segments; (3) intra- and inter-HCDR disulfides are frequently formed to rigidify the structure of HCDR3 or the combining site, and (4) V and C domains were possibly more similar in primordial antibodies than they are in their present day counterparts.

Three-dimensional structure of an immunoglobulin light-chain dimer with amyloidogenic properties

The X-ray structure of an immunoglobulin light-chain dimer isolated from the urine as a "Bence-Jones protein" from a patient with multiple myeloma and amyloidosis (Sea) was determined at 1.94 A resolution and refined to R and R(free) factors of 0.22 and 0.25, respectively. This "amyloidogenic" protein crystallized in the orthorhombic P2(1)2(1)2(1) space group with unit-cell parameters a = 48.28, b = 83.32, c = 112.59 A as determined at 100 K. In the vital organs (heart and kidneys), the equivalent of the urinary protein produced fibrillar amyloid deposits which were fatal to the patient. Compared with the amyloidogenic Mcg light-chain dimer, the Sea protein was highly soluble in aqueous solutions and only crystallized at concentrations approaching 100 mg ml(-1). Both the Sea and Mcg proteins packed into crystals in highly ordered arrangements typical of strongly diffracting crystals of immunoglobulin fragments. Overall similarities and significant differences in the three-dimensional structures and crystalline properties are discussed for the Sea and Mcg Bence-Jones proteins, which together provide a generalized model of abnormalities present in lambda chains, facilitating a better understanding of amyloidosis of light-chain origin (AL).

Recognition of IgG-derived peptides by a human IgM with an unusual combining site

A monoclonal immunoglobulin (Ig)M cryoglobulin (Mez) with interesting binding behaviour was isolated from a Waldenström's macroglobulinemia (WM) patient. It demonstrated very strong binding to peptides derived from the sequences of human IgG. However, when tested for binding to intact IgG, this antibody (Ab) did not show any rheumatoid factor (RF) activity. We propose several nonexclusive structural interpretations of the Mez-binding propensities, based on the orientations and solvent accessibilities of ligand residues and the nature of the Ab-binding site. To further characterize the structural features of Mez-peptide binding, IgG-derived octapeptides were docked into the Mez fragment variable (Fv)-binding site, revealing additional reasons for Mez-binding selectivity based on the interactions of the docked peptides with the Mez Fv. The problem was also approached from an immunological perspective. Comparisons of Mez variable region of the light chain (VL)/variable region of the heavy chain (VH) sequences with those of human germlines and known IgM RFs allowed us to provide a possible outline tracing the structural and functional origins of the Mez IgM. Coupled with examinations of interactions in docked complexes, this analysis led us to propose that the potential for RF activity, demonstrated through Mez binding to IgG-derived peptides, was owing to the inherent sequence and structure of the Mez IgM, rather than to somatic mutations. Thus, Mez IgM may occupy an intermediate niche between IgMs with and without RF activity.

Anti-sm autoantibodies in systemic lupus target highly basic surface structures of complexed spliceosomal autoantigens

Autoantibodies directed against spliceosomal proteins are a common and specific feature of systemic lupus erythematosus. These autoantibodies target a collection of proteins, including Sm B, B', D1, D2, and D3. We define the common antigenic targets of Sm D2 and D3 and examine their role in spliceosomal autoimmunity. Our results define nine major common epitopes, five on Sm D2 and four on Sm D3. These epitopes have significantly higher (more basic) isoelectric points than do nonantigenic regions. In fact, this association is of sufficient power to make isoelectric point an excellent predictor of spliceosomal antigenicity. The crystallographic structure of Sm D2 and D3 is now partially described. The anti-Sm D2 and D3 antigenic targets are located on the surface of the respective three-dimensional complexed proteins, thereby suggesting that these epitopes are accessible in the native configuration. All but one of these nine epitopes conspicuously avoid the specific regions involved in intermolecular interactions within the spliceosomal complex. One of the D3 epitopes (RGRGRGMGR) has significant sequence homology with a major antigenic region of Sm D1 (containing a carboxyl-terminal glycine-arginine repeat), and anti-D3 Abs cross-react with this epitope of Sm D1. These results demonstrate that spliceosomal targets of autoimmunity are accessible on native structure surfaces and that cross-reactive epitopes, as well as structural associations of various spliceosomal Ags, may be involved in the induction of autoimmunity in systemic lupus.

Crystallization and preliminary X-ray diffraction analysis of monkey dimeric dihydrodiol dehydrogenase

Dihydrodiol dehydrogenase catalyzes the NADP(+)-linked oxidation of trans-dihydrodiols of aromatic hydrocarbons to corresponding catechols and exists in multiple forms in mammalian tissues. The dimeric form of mammalian dihydrodiol dehydrogenase has a primary structure distinct from the previously known mammalian enzymes and may constitute a novel protein family with the prokaryotic proteins. Monkey kidney dimeric dihydrodiol dehydrogenase was crystallized from buffered ammonium phosphate solution using the hanging-drop vapour-diffusion method. The crystals diffract to 2.65 A resolution in the laboratory and belong to the hexagonal P6(1)22 or P6(5)22 space group, with unit-cell parameters a = b = 122.8, c = 121.3 A, alpha = beta = 90, gamma = 120 degrees.

Immunoglobulin cross-reactivity examined by library screening, crystallography and docking studies

Antibodies are extremely diverse with respect to their specificities and affinities for target molecules. Despite rigorous selection, some antibodies are cross-reactive whereby they recognize their natural antigens along with other molecules. In this review, we discuss our efforts toward understanding the cross-reactivity of selected immunoglobulins. Investigations that are discussed employed screens of combinatorial peptide libraries, crystallography of ligand-protein complexes, and computer-based peptide docking simulations. In the first example, two different antibodies (NC6.8 and NC10.14) bound the same trisubstituted guanidine (NC174) with similar affinities, but utilized predominantly dissimilar binding strategies. However, there was one common binding strategy, in which the cyanophenyl portion of NC174 was inserted end-on into the binding crevices of the NC6.8 and NC10.14 antibodies. In the second example, scanning of peptide libraries and X-ray crystallography were used to design and test synthetic peptides for binding to the Mcg L chain dimer. Again, end-on insertion was favored for all peptides larger than dipeptides in the voluminous Mcg binding cavity. Finally, automated docking was used for rapid predictions of complexes for the Fv molecule from a broadly cross-reactive human IgM (Mez) and nearly two thousand peptides. Certain amino acids, including the aromatic residues Trp and Phe, functioned as anchoring groups in automated docking. Anchoring groups acted in most of the peptides that were otherwise accommodated by a variety of binding strategies in the docked complexes. We suggest that anchoring of at least a portion of a ligand in a binding site is a common mechanism for antibody recognition.

Binding of synthetic peptides by a human monoclonal IgM with an unusual combining site structure

Using X-ray crystallography, a human monoclonal IgM cryoglobulin (Mez) was found to have an unusual combining site topography. Analysis of the unliganded Fv at 2.6 A resolution revealed that the HCDR3 had partitioned the active site into two compartments [Ramsland PA et al. 2000. Mol. Immunol. 37: 295-310]. The two cavities had dimensions and chemical properties that were compatible with the binding of peptides. In this study, libraries of peptides were prepared using solid-phase synthesis. Binding of the intact Mez IgM to these peptides was tested by enzyme-linked immunoassays. Screening of 400 dipeptides revealed that binding was markedly skewed toward amino acids with aromatic side-chains (Phe and Trp), especially when located in the second position. Preferential recognition of aromatic side-chains by Mez IgM was confirmed with larger peptides of three to five residues, but C-terminal positioning was not favored in these peptides. Mez IgM also showed binding propensities for acidic residues (Asp and Glu) as well as several other side-chains with different chemical properties, including His, Pro, Asn and Gln. Mez IgM recognized sets of overlapping octapeptides representing the sequences of the constant domains of human IgG1 heavy chains. These peptides represented similar stretches of polypeptide on the three-dimensional structures of all three constant domains (CH1, CH2 and CH3). Thus, Mez IgM may recognize structurally homologous regions of immunoglobulin domains, which were conserved during the evolution of the immune system.

Docking of combinatorial peptide libraries into a broadly cross-reactive human IgM

A monoclonal IgM cryoglobulin with diverse binding behavior was isolated from a patient (Mez) with Waldenström's macroglobulinemia. It gave very high titers in the binding of combinatorially synthesized libraries of peptides ranging in size from two to eight residues. The crystal structure of Mez Fv revealed that the binding site was divided into two cavities of unequal volumes with dimensions and chemical properties that were compatible with the binding of peptides. Access to this unique combination of structural information and peptide binding data led us to carry out Mez-peptide docking simulations to gain insight into the Mez binding propensities. In the present article, the results for docking of five peptide libraries are combined with discussions of the methods and approximations involved in the docking process. We analyze the origins of peptide binding affinity for Mez IgM in terms of its cross-reactivity and its structural preferences.

The three-dimensional structure of a complex of a murine Fab (NC10. 14) with a potent sweetener (NC174): an illustration of structural diversity in antigen recognition by immunoglobulins

The three-dimensional structure of a complex of an Fab from a murine IgG2b(lambda) antibody (NC10.14) with a high potency sweet tasting hap- ten, N-(p-cyanophenyl)-N'-(diphenylmethyl)-N"-(carboxymethyl)guan idine (NC174), has been determined to 2.6 A resolution by X-ray crystallography. This complex crystallized in the triclinic space group P1, with two molecules in the asymmetric unit. In contrast to a companion monoclonal antibody (NC6.8) with a kappa-type light chain and similar high affinity for the NC174 ligand, the NC10.14 antibody possessed a large and deep antigen combining site bounded primarily by the third complementarity-determining regions (CDR3s) of the light and heavy chains. CDR3 of the heavy chain dominated the site and its crown protruded into the external solvent as a type 1' beta-turn. NC174 was nested against HCDR3 and was held in place by two tryptophan side-chains (L91 and L96) from LCDR3. The diphenyl rings were accommodated on an upper tier of the binding pocket that is largely hydrophobic. At the floor of the site, a positively charged arginine side-chain (H95) stabilized the orientation of the electronegative cyano group of the hapten. The negative charge on the acetate group was partially neutralized by a hydrogen bond with the phenolic hydroxyl group of tyrosine H58. Comparisons of the modes of binding of NC174 to the NC6.8 and NC10.14 antibodies illustrate the enormous structural and mechanistic diversity manifest by immune responses.

Binding of nascent collagen by amyloidogenic light chains and amyloid fibrillogenesis in monolayers of human fibrocytes

Light (L) chain dimers expressed by multiple myeloma cells and collected as Bence-Jones proteins from the urine of human subjects were tested for their ability to form deposits in fibroblast monolayer cell cultures. Bence-Jones proteins from subjects with primary amyloidosis associated with L chains were shown to form fibrillar deposits by the in vitro assay introduced in this report. Filaments interspersed with nascent collagen could be detected after only 48 h. Deposition of L chains continued over a period of 72 h culminating in the appearance of dense fibrils with widths of 80-100 A and a variety of lengths. Formation of amyloid-like fibrils was accompanied by interference with the maturation of the collagen produced by the fibroblast cells. Fibrils composed of the Mcg lambda-type L chain were deposited between collagen fibers, thus expanding them laterally and leading to their partial disintegration. Mature collagen was completely missing from fibroblast monolayers exposed to the Sea lambda chain and the Jen kappa chain. Collagen with the characteristic striped pattern matured normally in control samples, such as those not dosed with amyloid precursors or those treated with a non-amyloidogenic Bence-Jones protein (e.g., the Hud lambda chain dimer). By immunochemical techniques using fluorescein- and gold-labeled anti-L chain antibodies, amyloidogenic L chains were shown to decorate the strands of nascent collagen. This observation suggests that amyloidogenic L chains are concentrated in the extracellular matrix by monovalent antigen-antibody type reactions. The capacity of the Mcg L chain dimer to bind collagen-derived sequences was tested by soaking crystals with a collagenase substrate, PZ-Pro-Leu-Gly-Pro-D-Arg. Difference Fourier analysis at 2.7 A resolution indicated that the PZ-peptide is a site-filling ligand. It could not be removed from the active site by perfusion of the crystal with ammonium sulfate crystallizing media. Similar experiments with the collagen-derived peptide (Pro-Pro-Gly)(5) showed substantial hysteresis effects extending from one end of the Mcg dimer to the other. After the ligand was withdrawn, the active site of the Mcg dimer could no longer bind the PZ-peptide. However, if the active site was first blocked by the PZ-peptide and subsequently exposed to the (Pro-Pro-Gly)(5) peptide, the difference Fourier map was indistinguishable from that obtained with the PZ-peptide alone. We concluded that amyloidogenic L chains such as the Mcg dimer could be concentrated in the perivascular space by binding to normal tissue constituents. These components include nascent collagen, which can be deterred from maturing as a result of this binding. Participation in such pathological activity is also self-destructive to the amyloidogenic L chains, which lose their binding capabilities for collagen-derived peptides and also become susceptible to irreversible conversion to amyloid fibrils. All of these events may be prevented by prior treatment of the amyloidogenic L chains with site-filling ligands. (c) 2000 John Wiley & Sons, Ltd.

An unusual human IgM antibody with a protruding HCDR3 and high avidity for its peptide ligands

The crystal structure of the Fv molecule from a human monoclonal IgM cryoglobulin (Mez) was determined at 2.6 A resolution. Amino acid sequences of framework regions (FR) of the Mez light (L) and heavy (H) chain variable domains (VL and VH) are highly similar to their counterparts in another human Fv (Pot) previously subjected to X-ray analysis in our laboratory. As expected, the three-dimensional (3-D) structures of FR are quite similar in the two proteins, as are four of the six complementarity-determining regions (CDRs): CDRs 1 and 2 for both L and H chains. Absence of Pro 95L from the LCDR3 loop in Mez VL (relative to Pot LCDR3) results in compression of this loop and creates more space in the VL-VH interface. In the two IgMs, HCDR3 conformations differ significantly from all previously defined conformations for these loops. Pot has a 12-residue HCDR3 that collapses to fill all available space in the VL-VH domain interface, resulting in the formation of a relatively flat platform for antigen binding. In Mez, the HCDR3 is two residues longer and is comprehensively different. A semi-rigid ascending segment dominated by a Pro-Pro-Tyr sequence protrudes out into solvent. The descending portion has the sequence Gly-Trp-Gly-Gly-Gly, which promotes high local flexibility. This segment folds across the VL-VH domain interface to interact with residues in LCDR3. These features partition the Mez active site into two compartments, a large cavity between VL and VH and a smaller cavity lined entirely by constituents of the three heavy chain CDRs. Such an unusual topographical feature indicates why the Mez IgM does not bind to the Fc portion of intact human IgG antibodies in immunoassays yet interacts with high avidity with many Fc-derived octapeptides. The cavities are expected to be the repositories for the Fc-derived peptides, while the semi-rigid protrusion of the Mez HCDR3 prevents the close approach of another macromolecule (e.g. intact IgG) to the active site.

Interference of rheumatoid factor activity by aspartame, a dipeptide methyl ester

Circulating autoimmune complexes of IgM rheumatoid factors (RF) bound to the Fc portions of normal, polyclonal IgG antibodies are frequently present in humans with rheumatoid arthritis (RA). The sweet tasting methyl ester of L-Asp-L-Phe (aspartame or APM) was found to relieve pain and improve joint mobility in subjects with osteo- and mixed osteo/rheumatoid arthritis [Edmundson, A. B. and Manion, C. V. (1998). Clin. Pharmac. Ther. 63, 580-593]. These clinical observations prompted the testing of the inhibition by APM of the binding interactions of human IgM RFs with IgG Fc regions. The propensity of APM to inhibit IgM RF binding was assessed by competitive enzyme immunoassays with solid-phase human IgG. Ten RA serum samples and three purified monoclonal cryoglobulins, all of which had RF activity, were tested in this system. We found that the presence of APM significantly reduced the binding of IgM RFs. The inhibitory propensity of APM with monoclonal RF cryoglobulins was increased by the addition of CaCl(2) to the binding buffer. Similar inhibition of the binding of RA derived RFs to IgG was observed for Asp-Phe and its amidated derivative, indicating that the methyl ester is not required for APM's interaction with IgM antibodies. A human (Mez) IgM known to bind octameric peptides derived from the Fc portion of a human IgG(1) antibody was tested for binding of dipeptides by the Pepscan method of combinatorial chemistry. The relative binding constants of Asp-Phe and Phe-Asp were ranked among the highest values for 400 possible combinations of the 20 most common amino acids. Possible blocking interactions of APM were explored by computer-assisted docking studies with the model of a complex of an RF Fab with the Fc of a human IgG(4) antibody. Modeling of ternary immune complexes revealed a few key residues, which could act as molecular recognition sites for APM. A structural hypothesis is presented to explain the observed interference with RF reactivity by APM. Extrapolations of the current results suggest that APM may inhibit the binding of IgG in a substantial proportion of IgM RFs. Interference of RF reactivity, especially in RA patients, may alleviate the pain and immobility resulting from chronic inflammation of the joints.

The acid-labile subunit of the serum insulin-like growth factor-binding protein complexes. Structural determination by molecular modeling and electron microscopy

The acid-labile subunit (ALS) is a glycosylated 85-kDa member of the leucine-rich repeat (LRR) protein superfamily and circulates in ternary complexes with the insulin-like growth factors (IGFs) and their binding proteins (IGFBPs). These complexes are thought to regulate the serum IGFs by restricting IGF movement out of the circulation. However, little is known about how ALS binds to IGFBP-3 or -5, which link the IGFs to ALS. To investigate potential sites of interaction, the ALS structure has been modeled with the crystal structure of the LRR protein porcine ribonuclease inhibitor as a template. ALS is predicted to be a donut-shaped molecule with an internal diameter of 1.7 nm, an external diameter of 7.2 nm, and a thickness of 3.6 nm. These dimensions are supported by rotary shadowing electron microscopy of ALS. The internal face is lined with a substantial region of electronegative surface potential that could interact with the positively charged region on IGFBP-3 known to be involved in ALS binding. The model also predicts that three potential N-linked oligosaccharide sites within the LRR domain are clustered together, which may be important in light of recent studies showing ALS glycan involvement in complex formation with IGFBP-3.

Interference of rheumatoid factor activity by aspartame, a dipeptide methyl ester

Circulating autoimmune complexes of IgM rheumatoid factors (RF) bound to the Fc portions of normal, polyclonal IgG antibodies are frequently present in humans with rheumatoid arthritis (RA). The sweet tasting methyl ester of L-Asp-L-Phe (aspartame or APM) was found to relieve pain and improve joint mobility in subjects with osteo- and mixed osteo/rheumatoid arthritis [Edmundson, A. B. and Manion, C. V. (1998). Clin. Pharmac. Ther. 63, 580-593]. These clinical observations prompted the testing of the inhibition by APM of the binding interactions of human IgM RFs with IgG Fc regions. The propensity of APM to inhibit IgM RF binding was assessed by competitive enzyme immunoassays with solid-phase human IgG. Ten RA serum samples and three purified monoclonal cryoglobulins, all of which had RF activity, were tested in this system. We found that the presence of APM significantly reduced the binding of IgM RFs. The inhibitory propensity of APM with monoclonal RF cryoglobulins was increased by the addition of CaCl(2) to the binding buffer. Similar inhibition of the binding of RA derived RFs to IgG was observed for Asp-Phe and its amidated derivative, indicating that the methyl ester is not required for APM's interaction with IgM antibodies. A human (Mez) IgM known to bind octameric peptides derived from the Fc portion of a human IgG(1) antibody was tested for binding of dipeptides by the Pepscan method of combinatorial chemistry. The relative binding constants of Asp-Phe and Phe-Asp were ranked among the highest values for 400 possible combinations of the 20 most common amino acids. Possible blocking interactions of APM were explored by computer-assisted docking studies with the model of a complex of an RF Fab with the Fc of a human IgG(4) antibody. Modeling of ternary immune complexes revealed a few key residues, which could act as molecular recognition sites for APM. A structural hypothesis is presented to explain the observed interference with RF reactivity by APM. Extrapolations of the current results suggest that APM may inhibit the binding of IgG in a substantial proportion of IgM RFs. Interference of RF reactivity, especially in RA patients, may alleviate the pain and immobility resulting from chronic inflammation of the joints.

Modeling of acanthoxin A1, a PLA2 enzyme from the venom of the common death adder (Acanthophis antarcticus)

The phospholipase A2 enzyme, acanthoxin, found in the venom of the common death adder (Acanthophis antarcticus) as with other snake PLA2 enzymes displays neurotoxic activity. It is unclear whether this neurotoxic activity particular to some snake PLA2 enzymes is a result of structural differences solely within the catalytic sites or at a distant location upon the molecules. We have predicted the three-dimensional structure of one of the two predominant isoforms of acanthoxin (A1) using comparative protein modeling techniques. Given the high degree of homology and the availability of a high quality crystallographic structure, notexin was used as a molecular template to construct an all atom model of acanthoxin. The model was made using the program MODELLER3 and then refined with X-PLOR. Comparison between the predicted structure of acanthoxin and several X-ray structures of toxic and nontoxic PLA2 enzymes has led to a testable two-step proposal of neurotoxic PLA2 activity; involving the favorable binding to acceptor molecules followed by enzymatic intrusion upon the target membrane. The electrostatic potentials across the molecular surfaces of toxic and nontoxic PLA2 enzymes were calculated (GRASP) and it was found that the toxic PLA2 enzymes possessed a charge distribution on the noncatalytic surface not identified in the nontoxic PLA2 enzymes. Thus we have identified residues potentially involved in the interaction of the PLA2 enzymes with their acceptor molecules. Furthermore, the proposed acceptor molecule recognition site is distant from the catalytic site which upon binding of the PLA2 to the acceptor molecule may enhance the enzymatic ability of the toxic PLA2 enzymes on particular cell types.

Structural aspects of human IgM antibodies expressed in chronic B lymphocytic leukemia

BACKGROUND

Malignant B cells from patients with chronic B lymphocytic leukemia (B CLL) generally express both surface IgM and the pan T cell antigen CD5, a characteristic of the B1 population of B lymphocytes. The IgM on the surface of these B CLL cells is frequently polyreactive with respect to its capacity to recognize multiple structurally dissimilar antigens (Ag).

OBJECTIVES

To understand the structural characteristics of the polyreactive binding sites of human IgM molecules expressed on B CLL cells by: (1) analyzing the nucleotide and protein sequences of the variable (V) domains of five IgM molecules expressed in cases of B CLL and; (2) utilizing these sequences to generate three-dimensional (3D) models of Fv (VL - VH) molecules.

STUDY DESIGN

Peripheral blood leukocytes obtained from five cases of B CLL were tested for polyreactive binding properties by assessing their capacity to bind mouse IgG by indirect immunofluorescence. The V region genes of light and heavy chains were amplified using the polymerase chain reaction, subsequently cloned and their nucleotide sequences obtained. Translated amino acid sequences of the V domains were used to generate homology models of the Fv molecules.

RESULTS

Low affinity binding of mouse IgG was demonstrated for all B CLL samples examined, confirming the polyreactive nature of the IgM expressed on these cells. There was an absence or minimal mutation within V region genes when compared to germline Ig genes. Junctional diversity was not observed for VL regions, although truncations and insertions were frequent in D minigenes of VH regions. The binding sites were predicted to form either relatively flat surfaces with occasional protrusions or cavities at the VL - VH domain interface. Aromatic side chains covered a large proportion of the potential binding surfaces in the models of B CLL Fv components.

DISCUSSION

Primary DNA sequences can be categorized as germline, suggesting that the B cells involved in B CLL are germline or naive in origin. The medium to large HCDR3s provide the majority of probable contact residues for antigens. While prominent aromatic residues are likely to engage in binding patterns which are conserved (e.g. mouse Ig reactivity), the diverse binding sites predicted for B CLL-derived IgMs also have properties which are conducive to polyreactive antigen binding.

Identification of a homologue of CD59 in a cyclostome: implications for the evolutionary development of the complement system

We have employed a COS cell expression cloning procedure to isolate a full length cDNA clone encoding a hagfish leukocyte-associated membrane protein (HLMP1). The protein, which is identified by a monoclonal antibody (JB3) generated in our laboratory, is present on the majority of hagfish leukocytes and is also expressed on erythrocytes. The cDNA clone contained an open reading frame encoding a 120 residue polypeptide which exhibits 33% amino acid sequence identity with the precursor protein of human CD59, a leukocyte-associated membrane protein which regulates the action of the complement membrane attack complex on homologous cells. CD59 belongs to a family of structurally related glycoproteins which includes the Ly-6 proteins expressed on mouse lymphocytes. In addition to significant overall sequence homology HLMP1 shows conservation of 8 key cysteine residues with members of the CD59/Ly-6 family. Comparison of the hagfish sequence with that of the mature human CD59 protein suggested a processed protein consisting of 74 amino acids associated with the cell membrane via a GPI anchor. The latter was confirmed by immuno-flow cytometry following treatment of transfected COS cells with phospholipase. Phylogenetic analysis and tissue distribution of this protein in the hagfish are consistent with HLMP1 being a homologue of CD59. A three-dimensional model of HLMP1, constructed using the NMR-determined structure for human CD59 as a template, indicated conservation of a core structure of five strands of beta-sheet and a short helix stabilised by four disulfide bonds. These findings, when taken together with our previous identification of C5a-like chemotactic activity in LPS-activated serum, provide indirect evidence for the existence of the terminal lytic complement pathway (C5 to C9) in these primitive vertebrates.

Comparison of the three-dimensional structures of a humanized and a chimeric Fab of an anti-gamma-interferon antibody

The objective of this work is to compare the three-dimensional structures of "humanized" and mouse-human chimeric forms of a murine monoclonal antibody elicited against human gamma-interferon. It is also to provide structural explanations for the small differences in the affinities and biological interactions of the two molecules for this antigen. Antigen-binding fragments (Fabs) were produced by papain hydrolysis of the antibodies and crystallized with polyethylene glycol (PEG) 8,000 by nearly identical microseeding procedures. Their structures were solved by X-ray analyses at 2.9 A resolution, using molecular replacement methods and crystallographic refinement. Comparison of these structures revealed marked similarities in the light (L) chains and near identities of the constant (C) domains of the heavy (H) chains. However, the variable (V) domains of the heavy chains exhibited substantial differences in the conformations of all three complementarity-determining regions (CDRs), and in their first framework segments (FR1). In FR1 of the humanized VH, the substitution of serine for proline in position 7 allowed the N-terminal segment (designated strand 4-1) to be closely juxtaposed to an adjacent strand (4-2) and form hydrogen bonds typical of an antiparallel beta-pleated sheet. The tightening of the humanized structure was relayed in such a way as to decrease the space available for the last portion of HFR1 and the first part of HCDR1. This compression led to the formation of an alpha-helix involving residues 25-32. With fewer steric constraints, the corresponding segment in the chimeric Fab lengthened by at least 1 A to a random coil which terminated in a single turn of 310 helix. In the humanized Fab, HCDR1, which is sandwiched between HCDR2 and HCDR3, significantly influenced the structures of both regions. HCDR2 was forced into a bent and twisted orientation different from that in the chimeric Fab, both at the crown of the loop (around proline H52a) and at its base. As in HCDR1, the last few residues of HCDR2 in the humanized Fab were compressed into a space-saving alpha-helix, contrasting with a more extended 310 helix in the chimeric form. HCDR3 in the humanized Fab was also adjusted in shape and topography. The observed similarities in the functional binding activities of the two molecules can be rationalized by limited induced fit adjustments in their structures on antigen binding. While not perfect replicas, the two structures are testimonials to the progress in making high affinity monoclonal antibodies safe for human use.

Diverse binding site structures revealed in homology models of polyreactive immunoglobulins

We describe here computer-assisted homology models of the combining site structure of three polyreactive immunoglobulins. Template-based models of Fv (VL-VH) fragments were derived for the surface IgM expressed by the malignant CD5 positive B cells from three patients with chronic lymphocytic leukaemia (CLL). The conserved framework regions were constructed using crystal coordinates taken from highly homologous human variable domain structures (Pot and Hil). Complementarity determining regions (CDRs) were predicted by grafting loops, taken from known immunoglobulin structures, onto the Fv framework models. The CDR templates were chosen, where possible, to be of the same length and of high residue identity or similarity. LCDR1, 2 and 3 as well as HCDR1 and 2 for the Fv were constructed using this strategy. For HCDR3 prediction, a database containing the Cartesian coordinates of 30 of these loops was complied from unliganded antibody X-ray crystallographic structures and an HCDR3 of the same length as that of the B CLL Fv was selected as a template. In one case (Yar), the resulting HCDR3 model gave unfavourable interactions when incorporated into the Fv model. This HCDR3 was therefore modelled using an alternative strategy of construction of the loop stems, using a previously described HCDR3 conformation (Pot), followed by chain closure with a beta-turn. The template models were subjected to positional refinement using energy minimisation and molecular dynamics simulations (X-PLOR). An electrostatic surface description (GRASP) did not reveal a common structural feature within the binding sites of the three polyreactive Fv. Thus, polyreactive immunoglobulins may recognise similar and multiple antigens through a diverse array of binding site structures.