Based on observations of solubility and folding properties of peptide 33-mers derived from the &sheet domains of platelet factor-4 (PF4), interleukin-8 (IL-8), and growth related protein (Gro-a), as well as other 6-sheet-forming peptides, general guidelines have been developed to aid in the design of water soluble, self-association-induced 0-sheet-forming peptides. CD, 'H-NMR, and pulsed field gradient NMR self-diffusion measurements have been used to assess the degree of folding and state of aggregation. PF4 peptide forms native-like @-sheet tetramers and is sparingly soluble above pH 6. IL-8 peptide is insoluble between pH 4.5 and pH 7.5, yet forms stable, nativelike &sheet dimers at higher pH. Gro-a peptide is soluble at all pH values, yet displays no discernable @-sheet structure even when diffusion data indicate dimer-tetramer aggregation. A recipe used in the de novo design of water-soluble 0-sheet-forming peptides calls for the peptide to contain 40-50~0 hydrophobic residues, usually aliphatic ones (I, L, V, A, M) (appropriately paired and mostly but not always alternating with polar residues in the sheet sequence), a positively charged (K, R) to negatively charged (E, D) residue ratio between 4/2 and 6/2, and a noncharged polar residue (N, Q, T, S) composition of about 20% or less. Results on four de novo designed, 33-residue peptides are presented supporting this approach; Under near physiologic conditions, all four peptides are soluble, form &sheet structures to varying degrees, and self-associate. One peptide folds as a stable, compact 0-sheet tetramer, whereas the others are transient 6-sheet-containing aggregates.

Native platelet factor-4 (PF4) is an asymmetrically associated, homo-tetrameric protein (70 residues/subunit) known for binding polysulphated glycosaminoglycans like heparin. PF4 N-terminal chimeric mutant M2 (PF4-M2), on the other hand, forms symmetric tetramers [Mayo, Roongta, Ilyina, Milius, Barker, Quinlan, La Rosa and Daly (1995) Biochemistry 34, 11399-11409] making NMR studies with this 32 kDa protein tractable. PF4-M2, moreover, binds heparin with a similar affinity to that of native PF4. NMR data presented here indicate that heparin (9000 Da cut-off) binding to PF4-M2, while not perturbing the overall structure of the protein, does perturb specific side-chain proton resonances which map to spatially related residues within a ring of positively charged side chains on the surface of tetrameric PF4-M2. Contrary to PF4-heparin binding models which centre around C-terminal alpha-helix lysines, this study indicates that a loop containing Arg-20, Arg-22, His-23 and Thr-25, as well as Lys-46 and Arg-49, are even more affected by heparin binding. Site-directed mutagenesis and heparin binding data support these NMR findings by indicating that arginines more than C-terminal lysines, are crucial to the heparin binding process.

Native human platelet factor 4 (PF4) is a homotetrameric protein (70 residues/subunit) known for its anticoagulant heparin binding activity. 2D 15N--1H HSQC NMR experiments of native PF4 in solution show the presence of conformational heterogeneity consistent with the formation of asymmetric homo-tetramers as observed in the X-ray crystal structure of both human and bovine PF4. A chimeric mutant of PF4 (called PF4-M2) which substitutes the first 11 N-terminal residues for the first eight residues from homologous interleukin-8 forms symmetric homo-tetramers with essentially the same heparin binding activity as native PF4. The solution structure of PF4-M2 has been investigated by using two- and three-dimensional 1H- and 15N-NMR spectroscopy and NOE-restrained simulated annealing molecular dynamics. As with other members of the CXC chemokine family whose structures are known, the PF4-M2 subunit monomer consists of a mostly hydrophobic, triple-stranded antiparallel beta-sheet onto which is folded an amphipathic C-terminal helix and a less periodic N-terminal domain. Although N-terminal substitution with the less acidic interleukin-8 sequence most affects the quarternary structure relative to native PF4 at the AC and AD dimer interfaces, AB dimer stability is weakened as reflected in reduced equilibrium association binding constants.

Previous studies have produced conflicting interpretations regarding the aggregation state of BPTI in solution. Here, pulsed-field gradient NMR self-association measurements have been performed with BPTI under a variety of temperature, pH, salt, urea conditions, and protein concentrations. Relative to the standard proteins, lysozyme, ribonuclease, and ubiquitin, diffusion constants indicate that BPTI dimerizes at concentrations above about 3 mg/mL and below 280 K. At higher temperatures, a marked self-association is observed above 10 mg/mL. The apparent lack of significant effects from variations in pH and NaCl concentration suggests minimal contribution to the aggregation process from charge-charge interactions. In contrast, in nondenaturing concentrations of urea (2 M), BPTI behaves as a monomer, suggesting that hydrophobic and polar residues modulate BPTI association. The BPTI surface shows that while one side is highly charged, the opposite side, composed mostly of hydrophobic and some hydrophilic residues, is feasible as an interface for BPTI self-association.

A de novo designed 33-residue polypeptide folds as a compact beta-sheet sandwich tetramer in aqueous solution. NMR structural analysis shows that although monomer subunits have the same three-stranded antiparallel beta-sheet fold, two equally populated conformational states are identified. Conformational heterogeneity arises from formation of two distinct dimer folds. Each dimer is formed by continuing the monomer beta-sheet into a six-stranded sheet similar to that found in alpha-chemokines. Dimer heterogeneity arises primarily from a two-residue shift in the alignment of interfacial strands. NOE-based conformational modeling has yielded well-defined structures for both dimer types. While the tetramer beta-sheet sandwich most probably results from association of hydrophobic surfaces from two amphipathic dimers, dimers could combine to form either two types of homotetramers and/or one heterotetramer composed of both dimer types. Even though interdimer NOEs could not be unambiguously identified to resolve this point, thermodynamic arguments based on observation of equal populations of both dimer types favor formation of heterotetramers.