Structure-function studies of human granulocyte-macrophage colony-stimulating factor. Identification of residues required for activity

Human granulocyte-macrophage colony stimulating factor (hGM CSF), a protein containing 127 amino acids, was chemically synthesized by using automated stepwise solid-phase methods. The unpurified synthetic hGM-CSF had the same range of actions on hemopoietic cells as the purified recombinant protein. The structural requirements for the activities of synthetic hGM-CSF were examined by the design and synthesis of fragments and analogs. The synthetic fragment, hGM-CSF (54-127), containing all four of the cysteine residues found in the intact protein, lacked detectable activity. Assays of fragments shortened at the N terminus showed that the residues 1-13 were not required for activity, but that the integrity of residues 14-25, particularly residues 16, 17, and 18, was critical for biologic activity. The 14-25 region is predicted to form the first alpha-helix in hGM-CSF. Synthetic peptides within the N-terminal 53 residue region lacked detectable activity. The synthetic analog hGM-CSF (1-121), which lacks the C-terminal 6 residues, had similar activity to hGM-CSF (1-127) indicating that residues 122-127 are not required for activity. An analog, [Ala88] hGM-CSF (14-96), which lacks the hydrophobic C-terminal region and 2 cysteine residues, had low but readily detectable activity suggesting that residues 14-96 are sufficient for detectable synthetic hGM-CSF activity, although the presence of residues 97-121 are required for full activity. No dissociation of the multiple biological activities of hGM-CSF was detected.     

Structure and stability of apolipoprotein a-I in solution and in discoidal high-density lipoprotein probed by double charge ablation and deletion mutation

To identify residues and segments in the central region of apolipoprotein A-I (apoA-I) that are important for the protein structure and stability, we studied the effects of four double charge ablations, D102A/D103A, E110A/E111A, R116V/K118A, and R160V/H162A, and two deletion mutations, Delta(61-78) and Delta(121-142), on the conformation and stability of apoA-I in the lipid-free state and in reconstituted discoidal phospholipid-cholesterol-apoA-I particles (rHDL). The findings suggest that D102/D103 and E110/E111 located in helix 4 and segment(s) between residues 61 and 78 are involved in maintenance of the conformation and stability of apoA-I in both the lipid-free state and in rHDL. R116/K118 located in helix 4 are essential for the conformation and stabilization of apoA-I in rHDL but not vital for the lipid-free state of the protein. The R160V/H162A substitutions in helix 6 lead to a less compact tertiary structure of lipid-free apoA-I without notable effects on the lipid-free or lipid-bound secondary conformation, suggesting involvement of R160/H162 in important interhelical interactions. The results on the Delta(121-142) mutant, together with our earlier findings, suggest disordered structure of a major segment between residues 121 and 143, likely including residues 131-143, in lipid-free apoA-I. Our findings provide the first experimental evidence for stabilization of rHDL by specific electrostatic interhelical interactions, in agreement with the double belt model. The effects of alterations in the conformation and stability of the apoA-I mutants on in vitro and in vivo functions of apoA-I and lipid homeostasis are discussed.     

Role of interchain alpha-helical hydrophobic interactions in Ca2+ affinity, formation, and stability of a two-site domain in troponin C.

We have previously shown that a 34-residue synthetic peptide representing the calcium-binding site III of troponin C formed a symmetric two-site dimer consisting of two helix-loop-helix motifs arranged in a head-to-tail fashion (Shaw, G.S., Hodges, R.S., & Sykes, B.D., 1990, Science 249, 280-283). In this study the hydrophobicities of the alpha-helices were altered by replacing L-98 and F-102 in the N-terminal region and/or I-121 and L-122 in the C-terminal region with alanine residues. Our results showed that substitution of hydrophobic residues either in the N- or C-terminal region have little effect on alpha-helix formation but resulted in a 100- and 300-fold decrease in Ca2+ affinity, respectively. Simultaneous substitution of both hydrophobes in the N- and C-terminal region resulted in a 1,000-fold decrease in Ca2+ affinity. Data from guanidine hydrochloride denaturation studies suggested that intermolecular interactions occur and that the less hydrophobic analogs had a lower overall conformational stability. These data support the contention that the hydrophobic residues are important in the formation of the two-site domain in troponin C, and this hydrophobic association stabilizes Ca2+ affinity.     

Characterization of discontinuous epitope of prion protein recognized by the monoclonal antibody T2

The anti-prion protein (PrP) monoclonal antibody T2 has previously been prepared using PrP-knockout mice immunized with mouse recombinant PrP residues 121-231, however its interaction mechanism to PrP antigen has not been cleared. Here we identified and characterized the epitope of T2 antibody. The competitive ELISA with 20-mer synthetic peptides derived from PrP121-231 showed that T2 antibody had no affinity for these peptides. The analysis with deletion mutants of PrP revealed that 10 amino acids in the N terminus and 66 amino acids in the C terminus of PrP121-231 were necessary for reactivity with T2. Two far regions are necessary for complete affinity of the T2 antibody for PrP; either region alone is not sufficient to retain the affinity. The epitope recognized by T2 antibody is discontinuous and conformational. We examined the effect of disulfide bond and salt bridges. Alkylation of cysteine residues in C terminus of PrP121-231, which breaks a disulfide bond and disrupts the structure, had diminished the reactivity. Mutations induced in the PrP121-231 to break the disulfide bond or salt bridges, markedly had reduced the reactivity with T2 antibody. It suggests that T2 antibody recognized the structure maintained by the disulfide bond and salt bridges.     

NMR structure of a variant human prion protein with two disulfide bridges

The nuclear magnetic resonance structure of the globular domain with residues 121-230 of a variant human prion protein with two disulfide bonds, hPrP(M166C/E221C), shows the same global fold as wild-type hPrP(121-230). It contains three alpha-helices of residues 144-154, 173-194 and 200-228, an anti-parallel beta-sheet of residues 128-131 and 161-164, and the disulfides Cys166-Cys221 and Cys179-Cys214. The engineered extra disulfide bond in the presumed "protein X"-binding site is accommodated with slight, strictly localized conformational changes. High compatibility of hPrP with insertion of a second disulfide bridge in the protein X epitope was further substantiated by model calculations with additional variant structures. The ease with which the hPrP structure can accommodate a variety of locations for a second disulfide bond within the presumed protein X-binding epitope suggests a functional role for the extensive perturbation by a natural second disulfide bond of the corresponding region in the human doppel protein.     

Interactions between subdomains in the partially folded state of staphylococcal nuclease

Staphylococcal nuclease can be roughly divided into a beta-subdomain in N-terminal and an alpha-subdomain in C-terminal. They fold sequentially under certain conditions, causing a partially folded intermediate state in which the native-like beta-barrel persists while alpha-helix regions largely disorder. To investigate the possible long-range interactions between the two subdomains in the intermediate, N-terminal fragments have been used as intermediate analogues, with polypeptide ending at positions 102, 110, 121 and 135 and with a tryptophan substitution at position 66 or 88 to facilitate the observation of the beta-barrel. Segment-resolved interactions between beta-barrel and residues 103-135 were identified by comparing their spectroscopic properties of fluorescence, circular dichroism and NMR and by their stability. Except for unstable V66W102, the guanidine and thermal denaturation of fragments are cooperative and well approximated by the two-state transition. Minimal stable structure units of both tryptophan-containing fragments comprise residues 1-110. With the main interaction in segment 103-135, residues 103-110 contribute approximate 2 kcal/mol to the stability. Elongation of C-terminal from 110 residue neither increases the stability nor alters the structure core of the G88W fragments. However, residues 111-121 influence the tertiary structure of the V66W fragments suggesting its minor interactions with beta-barrel.     

Identification of functionally important TonB-ExbD periplasmic domain interactions in vivo

In gram-negative bacteria, the cytoplasmic membrane proton-motive force energizes the active transport of TonB-dependent ligands through outer membrane TonB-gated transporters. In Escherichia coli, cytoplasmic membrane proteins ExbB and ExbD couple the proton-motive force to conformational changes in TonB, which are hypothesized to form the basis of energy transduction through direct contact with the transporters. While the role of ExbB is not well understood, contact between periplasmic domains of TonB and ExbD is required, with the conformational response of TonB to presence or absence of proton motive force being modulated through ExbD. A region (residues 92 to 121) within the ExbD periplasmic domain was previously identified as being important for TonB interaction. Here, the specific sites of periplasmic domain interactions between that region and the TonB carboxy terminus were identified by examining 270 combinations of 45 TonB and 6 ExbD individual cysteine substitutions for disulfide-linked heterodimer formation. ExbD residues A92C, K97C, and T109C interacted with multiple TonB substitutions in four regions of the TonB carboxy terminus. Two regions were on each side of the TonB residues known to interact with the TonB box of TonB-gated transporters, suggesting that ExbD positions TonB for correct interaction at that site. A third region contained a functionally important glycine residue, and the fourth region involved a highly conserved predicted amphipathic helix. Three ExbD substitutions, F103C, L115C, and T121C, were nonreactive with any TonB cysteine substitutions. ExbD D25, a candidate to be on a proton translocation pathway, was important to support efficient TonB-ExbD heterodimerization at these specific regions.     

Search for alpha-helical propensity in the receptor-bound conformation of glucagon-like peptide-1

To elucidate the receptor-bound conformation of glucagon-like peptide-1 (GLP-1), a series of conformationally constrained GLP-1 analogues were synthesized by introducing lactam bridges between Lys(i) and Glu(i)(+4) to form alpha-helices at various positions. The activity and affinity of these analogues to GLP-1 receptors suggested that the receptor-bound conformation comprises two alpha-helical segments between residues 11-21 and 23-34. It is notable that the N-terminal alpha-helix is extended to Thr(11), and that Gly(22) plays a pivotal role in arranging the two alpha-helices. Based on these findings, a highly potent bicyclic GLP-1 analogue was synthesized which is the most conformationally constrained GLP-1 analogue reported to date.     

Characterization of the {alpha}-helix region in domain 3 of the haemolytic lectin CEL-III: implications for self-oligomerization and haemolytic processes

CEL-III is a haemolytic lectin, which has two beta-trefoil domains (domains 1 and 2) and a beta-sheet-rich domain (domain 3). In domain 3 (residues 284-432), there is a hydrophobic region containing two alpha-helices (H8 and H9, residues 317-357) and a loop between them, in which alternate hydrophobic residues, especially Val residues, are present. To elucidate the role of the alpha-helix region in the haemolytic process, peptides corresponding to different parts of this region were synthesized and characterized. The peptides containing the sequence that corresponded to the loop and second alpha-helix (H9) showed the strongest antibacterial activity for Staphylococcus aureus and Bacillus subtilis through a marked permeabilization of the bacterial cell membrane. The recombinant glutathione S-transferase (GST)-fusion proteins containing domain 3 or the alpha-helix region peptide formed self-oligomers, whereas mutations in the alternate Val residues in the alpha-helix region lead to decreased oligomerization ability of the fusion proteins. These results suggest that the alpha-helix region, particularly its alternate Val residues are important for oligomerization of CEL-III in target cell membranes, which is also required for a subsequent haemolytic action.     

The solution structure of the C-terminal domain of TonB and interaction studies with TonB box peptides

The TonB protein transduces energy from the proton gradient across the cytoplasmic membrane of Gram-negative bacteria to TonB-dependent outer membrane receptors. It is a critically important protein in iron uptake, and deletion of this protein is known to decrease virulence of bacteria in animal models. This system has been used for Trojan horse antibiotic delivery. Here, we describe the high-resolution solution structure of Escherichia coli TonB residues 103-239 (TonB-CTD). TonB-CTD is monomeric with an unstructured N terminus (103-151) and a well structured C terminus (152-239). The structure contains a four-stranded antiparallel beta-sheet packed against two alpha-helices and an extended strand in a configuration homologous to the C-terminal domain of the TolA protein. Chemical shift perturbations to the TonB-CTD (1)H-(15)N HSCQ spectrum titrated with TonB box peptides modeled from the E.coli FhuA, FepA and BtuB proteins were all equivalent, indicating that all three peptides bind to the same region of TonB. Isothermal titration calorimetry measurements demonstrate that TonB-CTD interacts with the FhuA-derived peptide with a K(D)=36(+/-7) microM. On the basis of chemical shift data, the position of Gln160, and comparison to the TolA gp3 N1 complex crystal structure, we propose that the TonB box binds to TonB-CTD along the beta3-strand.     

Structural and thermodynamic consequences of burying a charged residue within the hydrophobic core of T4 lysozyme.

To determine the energetic and structural consequences of placing a charged group within the core of a protein, two "buried charge" mutants, Met 102----Lys (M102K) and Leu 133----Asp (L133D) were constructed in phage T4 lysozyme. Both proteins fold at neutral pH, although they are substantially less stable than wild type. The activity of M102K is about 35% that of wild type, while that of L133D is about 4%. M102K could be crystallized, and its structure was determined at high resolution. The crystal structure (at pH 6.8) of the mutant is very similar to that of wild type except for the alpha-helix that includes residues 108-113. In wild-type lysozyme, one side of this helix is exposed to solvent and the other contacts Met 102. In the M102K structure this alpha-helix becomes much more mobile, possibly allowing partial access of Lys 102 to solvent. The stability of M102K, determined by monitoring the unfolding of the protein with CD, is pH-dependent, consistent with the charged form of the substituted amino acid being more destabilizing than the uncharged form. The pKa of Lys 102 was estimated to be 6.5 both by differential titration and also by NMR analysis of isotopically labeled protein with 13C incorporated at the C epsilon position of all lysines. As the pH is lowered below pH 6.5, the overall three-dimensional structure of M102K at room temperature appears to be maintained to pH 3 or so, although there is evidence for some structural adjustment possibly allowing solvent accessibility to the protonated form of Lys 102.     

Predicted Folding of β-Structure in Myelin Basic Protein

Predictions of myelin basic protein secondary structure have not previously considered a major role for beta-structure in the organization of the native molecule because optical rotatory dispersion and circular dichroism studies have provided little, if any, evidence for beta-structure, and because a polycationic protein is generally considered to resist folding into a compact structure. However, the Chou-Fasman, Lim, and Robson algorithms identify a total of five beta-strands in the amino acid sequence. Four of these hydrophobic amino acid sequences (37-45, 87-95, 110-118, and 150-158) could form a hairpin intermediate that initiates folding of a Greek-key-type beta-structure. A second fold on the more hydrophobic side, with the addition of a strand from the N-terminus (residues 13-21), would complete the five-stranded antiparallel beta-sheet. A unique strand alignment can be predicted by phasing the hydrophobic residues. The unusual triproline sequence of myelin basic protein (100-102) is enclosed in the 14-residue hairpin loop. If these prolines are in the trans conformation, models show that a reverse turn could occur at residues 102-105 (Pro-Ser-Gln-Gly). Algorithms do not agree on the prediction of alpha-helices, but each of the two large loops could accommodate an alpha-helix. Myelin basic protein is known to be phosphorylated in vivo on as many as five Ser/Thr residues. Phosphorylation might alter the dynamics of folding if the nascent polypeptide were phosphorylated in the cytoplasm. In particular, phosphorylation of Thr-99 could neutralize cationic residues Lys-106 and Arg-108 within the hairpin loop. In addition, the methylation of Arg-108 might stabilize the hairpin loop structure through hydrophobic interaction with the side chain of Pro-97. The cationic side chains of arginine and lysine residues located on the faces of the beta-sheet (Arg-43, Arg-114, Lys-13, Lys-92, Lys-153, and Lys-156) could provide sites for interaction with phospholipids and other anionic structures on the surface of the myelin lipid bilayer.     

A contraceptive peptide vaccine targeting sulfated glycoprotein ZP2 of the mouse zona pellucida

In this study, we have mapped and characterized a B cell epitope of sulfated glycoprotein ZP2 (ZP2) as a step toward the development of a multi-epitope zona pellucida (ZP) vaccine. Recombinant polypeptides expressed by random deoxyribonuclease-digested fragments of ZP2 cDNA were screened for binding to IE-3, a monoclonal antibody to murine ZP2. Positive clones contained cDNA inserts encoding polypeptide corresponding to ZP2(103-134). When normal or ovariectomized female mice were immunized with three overlapping peptides that span this region of ZP2 (101-120, 111-130, 121-140), only ZP2(121-140) elicited IgG antibodies that reacted with mouse ovarian ZP, indicative of the presence of native B epitope and helper T cell epitope in ZP2(121-140). To more finely map the ZP2 B cell epitope, a random peptide display library was screened with the IE-3 antibody, and a consensus tetramer sequence VxYK that matched the ZP2(123-126) sequence VRYK was located. Competitive immunofluorescence analysis with single alanine-substituted VxYK peptides ranked the relative contribution of the three critical B cell epitope residues as Y > V > K. A chimeric peptide was constructed that contained the YRYK motif of ZP2 and a bovine RNase T cell epitope. Although (C57BL/6xA/J) F1 (B6AF1) female mice immunized with the chimeric peptide developed ZP antibody response, this peptide elicited antibody only in mice of the histocompatibility complex (MHC) H-2(k or b) haplotype. In contrast, ZP2(121-140) peptide elicited antibody in inbred mice with three additional mouse MHC haplotypes. Moreover, although ZP2(121-140) contained a T cell epitope, no oophoritis was observed after immunization of B6AF1 mice with ZP2(121-140) in complete Freund's adjuvant (CFA). In a preliminary trial, female B6AF1 mice immunized with ZP2(121-140) in CFA had reduced litter sizes as compared with mice injected with CFA alone.     

The mapping of epitopes of the 18-kDa protein of Mycobacterium leprae recognized by murine T cells in a proliferation assay

The 18-kDa protein of Mycobacterium leprae was purified from recombinant plasmids pUL108 and pML-3 grown in Saccharomyces cerevisiae and Escherichia coli, respectively. Significant lymphoproliferative responses were observed when T cells from immunized mice were challenged in culture with purified 18-kDa protein. Synthetic peptides have been prepared that span most of the 148 amino acid residues that constitute the sequence of the 18-kDa protein and used to map epitopes recognized by T cells. When mice were immunized with 18-kDa protein and lymph node cells subsequently prepared and challenged in microculture proliferative assays by using synthetic peptides, only one region of the intact protein appeared stimulatory. This T cell epitope was located between residues 116 and 121, adjacent to an epitope between residues 110 and 115 which we have previously shown to bind the L5 mAb. Immunization of mice with peptides, and subsequent challenge of lymph node cells in assays by using the 18-kDa protein as Ag revealed that residues 111-125 were the most effective in priming responses. Furthermore, the ability of 18-kDa primed lymph node cells to recognize determinants on both M. leprae and Mycobacterium tuberculosis indicates that in addition to possessing an M. leprae-specific B cell determinant, the 18-kDa protein contains a cross-reactive T cell epitope(s).     

NMR structure of an F-actin-binding "headpiece" motif from villin

A growing family of F-actin-bundling proteins harbors a modular F-actin-binding headpiece domain at the C terminus. Headpiece provides one of the two F-actin-binding sites essential for filament bundling. Here, we report the first structure of a functional headpiece domain. The NMR structure of chicken villin headpiece (HP67) reveals two subdomains that share a tightly packed hydrophobic core. The N-terminal subdomain contains bends, turns, and a four-residue alpha-helix as well as a buried histidine residue that imparts a pH-dependent folding. The C-terminal subdomain is composed of three alpha-helices and its folding is pH-independent. Two residues previously implicated in F-actin-binding form a buried salt-bridge between the N and C-terminal subdomains. The rest of the identified actin-binding residues are solvent-exposed and map onto a unique F-actin-binding surface.     

Alterations in the two globular domains or in the connecting alpha-helix of bacterial ribosomal protein L9 induces +1 frameshifts

The ribosomal 50S subunit protein L9, encoded by the gene rplI, is an elongated protein with an alpha-helix connecting the N- and C-terminal globular domains. We isolated rplI mutants that suppress the +1 frameshift mutation hisC3072 in Salmonella enterica serovar Typhimurium. These mutants have amino acid substitutions in the N-terminal domain (G24D) or in the C-terminal domain (I94S, A102D, G126V, and F132S) of L9. In addition, different one-base deletions in rplI altered either the final portion of the C terminus or removed the C-terminal domain with or without the connecting alpha-helix. An alanine-to-proline substitution at position 59 (A59P), which breaks the alpha-helix between the globular domains, induced +1 frameshifting, suggesting that the geometrical relationship between the N and C domains is important to maintain the reading frame. Except for the alterations G126V in the C terminus and A59P in the connecting alpha-helix, our results confirm earlier results obtained by using the phage T4 gene 60-based system to monitor bypassing. The way rplI mutations suppress various frameshift mutations suggests that bypassing of many codons from several takeoff and landing sites occurred instead of a specific frameshift forward at overlapping codons.     

Ultrastructural characterization of embryonic chick cartilage proteoglycan core protein and the mapping of a monoclonal antibody epitope.

The ultrastructure of embryonic chick cartilage proteoglycan core protein was investigated by electron microscopy of specimens prepared by low angle shadowing. The molecular images demonstrated a morphological substructural arrangement of three globular and two linear regions within each core protein. The internal globular region (G2) was separated from two terminally located globular regions (G1 and G3) by two elongated strands with lengths of 21 +/- 3 nm (E1) and 105 +/- 22 nm (E2). The two N-terminal globular regions, separated by the 21-nm segment, were consistently visualized in well spread molecules and showed little variation in the length of the linear segment connecting them. The E2 segment, however, was quite variable in length, and the C-terminal globular region (G3) was detected in only 53% of the molecules. The G1, G2, and G3 regions in chick core protein were 10.1 +/- 1.7 nm, 9.7 +/- 1.3 nm, and 8.3 +/- 1.3 nm in diameter, respectively. These results are similar to those described previously for proteoglycan core proteins isolated from rat chondrosarcoma, bovine nasal cartilage, and pig laryngeal cartilage (Paulsson, M., Morgelin, M., Wiedemann, H., Beardmore-Gray, M., Dunham, D., Hardingham, T., Heinegard, D., Timpl, R., and Engel, J. (1987) Biochem. J. 245, 763-772). However, a significant difference was detected between the length of the elongated strand (E2) of core proteins isolated from chick cartilage, E2 length = 105 +/- 22 nm, compared to bovine nasal cartilage, E2 length = 260 +/- 39 nm. The epitope of the proteoglycan core protein-specific monoclonal antibody, S103L, was visualized by electron microscopy, and the distance from the core protein N terminus to the S103L binding site was measured. The S103L binding site was localized to the E2 region, 111 +/- 20 nm from the G1 (N terminus) domain and 34 nm from the G3 (C terminus) domain. cDNA clones selected from an expression vector library of chicken cartilage mRNA also show this epitope to be located near the C-terminal region (R. C. Krueger, T. A. Fields, J. Mensch, and B. Schwartz (1990) J. Biol. Chem. 265, 12088-12097).     

Structural analysis of the heparin-binding site of the NC1 domain of collagen XIV by CD and NMR.

Type XIV collagen, a fibril-associated collagen with interrupted triple helices (FACIT), interacts with the surrounding extracellular matrix and/or with cells via its binding to glycosaminoglycans (GAGs). To further characterize such interactions in the NC1 domain of chicken collagen XIV, we identified amino acids essential for heparin binding by affinity chromatography analysis after proteolytic digestion of the synthetic peptide NC1(84-116). The 3D structure of this peptide was then obtained using circular dichroism and NMR. The NC1(84-116) peptide appeared poorly structured in water, but the stabilization of its conformation by the interaction with hydrophobic surfaces or by using cosolvents (TFE, SDS) revealed a high propensity to adopt an alpha-helical folding. A 3D structure model of NC1(84-116), calculated from NMR data recorded in a TFE/water mixture, showed that the NC1-heparin binding site forms a amphipathic alpha-helix exhibiting a twisted basic groove. It is structurally similar to the consensus spatial alpha-helix model of heparin-binding [Margalit et al. (1993) J. Biol. Chem. 268, 19228-19231], except that the GAG binding domain of NC1 may be extended over 18 residues, that is, the NC1(94-111) segment. In addition, the formation of a hydrophobic groove upon helix formation suggests the contribution of additional sequences to ensure the stability of the GAG-binding domain. Overall the NC1(84-116) model exhibits a nativelike conformation which presents suitably oriented residues for the interaction with a specific GAG.     

Comparison of structural requirements for interaction of the same peptide with I-Ek and I-Ed molecules in the activation of MHC class II-restricted T cells.

We have analyzed the interaction of the hen egg-white lysozyme (HEL) peptide 107-116 with the MHC class II molecule I-Ek, using truncated and single residue substitution analogues to measure activation of I-Ek-restricted, 107-116-specific T cell hybridomas and competition for Ag presentation by I-Ek molecules. These results have been compared with previous findings on the interaction of the same peptide with the I-Ed molecule. Stimulation of T cell hybridomas by truncated peptides defines the sequence 108-116 as the minimum epitope necessary for activation of both I-Ek- and I-Ed-restricted T cell hybridomas. Substitution analysis pinpoints three residues (V109, A110, and K116) in the sequence 108-116 as being critical for binding to I-Ek molecules and demonstrates the involvement of most other residues in recognition by T cells. Results previously obtained for binding of HEL 107-116 to I-Ed molecules indicated that peptide residues R112, R114, and K116 were critical for interaction with I-Ed. Comparison of these results indicates a difference in the likely MHC contact residues between the HEL sequence 108-116 and I-Ed or I-Ek molecules, suggesting that the same HEL peptide assumes a different conformation in the binding site of these two MHC molecules. This in turn affects residues interacting with the specific T cell receptor. According to the hypothetical tridimensional structure predicted for class II molecules, the difference in MHC contact residues observed within the sequence 108-116 can be related to polymorphic amino acids in the binding site of I-Ek and I-Ed molecules. A search through published binding data for a common pattern in this and other I-Ek-binding peptides has permitted us to derive a possible motif for predicting peptide binding to I-Ek molecules. This putative motif was tested by determining binding to I-Ek of an unbiased panel of about 150 synthetic peptides. Binding data indeed demonstrate the presence of this motif in the majority of good binders to I-Ek molecules.