Evidence for similar function of transmembrane segments in receptor and membrane-anchored proteins

Guidelines for membrane protein engineering derived from de novo designed model peptides

Notwithstanding great advances in the engineering and structural analysis of globular proteins, relatively limited success has been achieved with membrane proteins--due largely to their intrinsic high insolubility and the concomitant difficulty in obtaining crystals. Progress with de novo synthesis of model membrane-interactive peptides presents an opportunity to construct simpler peptides with definable structures, and permits one to approach an understanding of the properties of the membrane proteins themselves. In the present article, we review how our laboratory and others have used peptide approaches to assess the detailed interactions of peptides with membranes, and primary folding at membrane surfaces and in membranes. Structural studies of model peptides identified the existence of a "threshold hydrophobicity," which controls spontaneous peptide insertion into membranes. Related studies of the relative helicity of peptides in organic media such as n-butanol indicate that the helical propensity of individual residues--not simply their hydrophobicity--may dictate the conformations of peptides in membranes. The overall experimental results provide fundamental guidelines for membrane protein engineering.

Uncoupling hydrophobicity and helicity in transmembrane segments. Alpha-helical propensities of the amino acids in non-polar environments

Although the chains of amino acids in proteins that span the membrane are demonstrably helical and hydrophobic, little attention has been paid toward addressing the range of helical propensities of individual amino acids in the non-polar environment of membranes. Because it is inappropriate to apply soluble protein-based structure prediction algorithms to membrane proteins, we have used de novo designed peptides (KKAAAXAAAAAXAAWAAXAAAKKKK-amide, where X indicates one of the 20 commonly occurring amino acids) that mimic a protein membrane-spanning domain to determine the alpha-helical proclivity of each residue in the isotropic non-polar environment of n-butanol. Peptide helicities measured by circular dichroism spectroscopy were found to range from theta222 = -17,000 degrees (Pro) to -38,800 degrees (Ile) in n-butanol. The relative helicity of each amino acid is shown to be well correlated with its occurrence frequency in natural transmembrane segments, indicating that the helical propensity of individual residues in concert with their hydrophobicity may be a key determinant of the conformations of protein segments in membranes.

Sequence-related behaviour of transmembrane domains from class I receptor tyrosine kinases

2H NMR spectroscopy and freeze-fracture electron microscopy were used to compare the transmembrane domains of two Class I protein receptor tyrosine kinases (the EGF receptor and Neu/erbB-2) regarding overall behaviour in fluid lipid bilayer membranes. The 34-residue peptide, EGFRtm, was synthesised to contain the 23 amino acid hydrophobic stretch (Ile622 to Met644) thought to span the membrane of the human EGF receptor, plus the first 10 amino acids (Arg645 to Thr654) of the cytoplasmic domain. Deuterium probes replaced selected 1H nuclei at sites corresponding to Ala623, Met644, and Val650. The 38-residue peptide, Neutm, was synthesised having the 21 residue hydrophobic stretch (Ile660 to Ile680) calculated to span the membrane in rat Neu/erbB-2, plus residues Lys681 to Thr691 of the contiguous cytoplasmic domain. Deuterium probes replaced selected 1H nuclei at Ala661, Leu667, and Val676. A third peptide, Neutm*, was also prepared, corresponding to the transmembrane domain of a constitutively-activating Neu/erbB-2 transformant in which Val664 is replaced by Glu: it was deuterated in a manner identical to Neutm. Peptides were studied by 2H NMR spectroscopy at 1 mol% and 6 mol% in unsonicated fluid bilayers of 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) and in POPC containing 33 mol% cholesterol, over the range 12 degrees to 65 degreesC. Overall motion was found to be different for each of the three peptides under a given set of conditions. EGFRtm spectra were characteristic of axially symmetric motion in membranes of POPC alone, and in POPC/cholesterol at 35 degreesC and above. In contrast, spectra of the transmembrane peptides, Neutm and Neutm*, were characteristic of significantly axially asymmetric motion under all conditions studied (and regardless of sample preparation method). Addition of 33% cholesterol to membranes was accompanied by spectral changes consistent with increased formation of peptide dimers/oligomers in all cases. The transformant peptide, Neutm*, showed greater spectral evidence of immobilisation than did the wild type - probably reflecting a greater tendency to form large oligomers. Sequence-related details within the transmembrane domains of Class I receptor tyrosine kinases appear to exert important control over their associations within membranes. Freeze-fracture electron microscopy of the NMR samples demonstrated their liposomal nature. Peptide-related intramembranous particles (IMPs) were present which likely represent oligomers of the transmembrane peptide. IMP size and distribution were similar under a given set of conditions for all three peptides, suggesting that the differences seen by NMR spectroscopy reflect structures smaller than the 2 nm resolution limit of freeze-fracture EM and peptide relationships within its 20 nm accuracy of identifying lateral position.

Solution and solid state conformation of the human EGF receptor transmembrane region

The epidermal growth factor receptor (EGFR) is a member of the tyrosine kinase family of signalling cell surface molecules. Signalling by this protein is mediated through binding of epidermal growth factor to its extracellular region ultimately leading to phosphorylation of several residues on the intracellular portion of the receptor. The only means of communication between the intracellular and extracellular domains is via the transmembrane region of the protein. In this work we describe the first structural studies of a 34-residue synthetic peptide (hEGFRp), representative of the human EGFR transmembrane region, using two-dimensional and 2H wideline NMR and CD spectroscopies. In water the peptide demonstrated a lack of regular secondary structure and existed as oligomers. Addition of the lipomimetic solvent, trifluoroethanol (TFE), led to the production of monomeric structured species. Analysis of NMR spectra of the hEGFRp indicated that an alpha-helix was present between residues M626 and R647. This observation was reinforced by solid state 2H NMR studies in lipid bilayers which showed typical 'Pake' spectra indicating axially symmetric motion. The helical region in hEGFRp commences four residues later than predicted via hydrophobicity profiles, and extends to include several charged arginine residues which would lie on the cytosolic side of the membrane. These observations provide the first evidence that the transmembrane alpha-helical region in EGFR may not only traverse the membrane but may continue to the cytosolic region near T654, an important phosphorylation site.

A family of extracytoplasmic proteins that allow transport of large molecules across the outer membranes of gram-negative bacteria

Seventeen fully sequenced and two partially sequenced extracytoplasmic proteins of purple, gram-negative bacteria constitute a homologous family termed the putative membrane fusion protein (MFP) family. Each such protein apparently functions in conjunction with a cytoplasmic membrane transporter of the ATP-binding cassette family, major facilitator superfamily, or heavy metal resistance/nodulation/cell division family to facilitate transport of proteins, peptides, drugs, or carbohydrates across the two membranes of the gram-negative bacterial cell envelope. Evidence suggests that at least some of these transport systems also function in conjunction with a distinct outer membrane protein. We report here that the phylogenies of these proteins correlate with the types of transport systems with which they function as well as with the natures of the substrates transported. Characterization of the MFPs with respect to secondary structure, average hydropathy, and average similarity provides circumstantial evidence as to how they may allow localized fusion of the two gram-negative bacterial cell membranes. The membrane fusion protein of simian virus 5 is shown to exhibit significant sequence similarity to representative bacterial MFPs.

Computer-aided analyses of transport protein sequences: gleaning evidence concerning function, structure, biogenesis, and evolution

Three-dimensional structures have been elucidated for very few integral membrane proteins. Computer methods can be used as guides for estimation of solute transport protein structure, function, biogenesis, and evolution. In this paper the application of currently available computer programs to over a dozen distinct families of transport proteins is reviewed. The reliability of sequence-based topological and localization analyses and the importance of sequence and residue conservation to structure and function are evaluated. Evidence concerning the nature and frequency of occurrence of domain shuffling, splicing, fusion, deletion, and duplication during evolution of specific transport protein families is also evaluated. Channel proteins are proposed to be functionally related to carriers. It is argued that energy coupling to transport was a late occurrence, superimposed on preexisting mechanisms of solute facilitation. It is shown that several transport protein families have evolved independently of each other, employing different routes, at different times in evolutionary history, to give topologically similar transmembrane protein complexes. The possible significance of this apparent topological convergence is discussed.

Conformations of proline residues in membrane environments

Although noted as hydrophilic residues with helix-breaking potential, proline residues are observed in putatively alpha-helical transmembrane (TM) segments of many channel-forming integral membrane proteins. In addition to the recognized property of X-Pro peptide bonds (where X = any amino acid) to occur in cis as well as trans isomeric states, the tertiary amide character of the X-Pro bond confers increased propensity for involvement of its carbonyl group in specific H-bonded structures (e.g., beta- and gamma-turns) and/or liganding interactions with positively charged species. To examine this latter situation in further detail, we identified Leu-Pro-Phe as a consensus sequence triad based on actual occurrences of intramembranous Pro residues in transport protein TM segments. Accordingly, we have undertaken the synthesis of hydrophobic peptides with potential membrane affinity, of which t-butyloxycarbonyl-L-Ala-L-Ala-L-Ala-L-Leu-L-Pro-L-Phe-OH (t-Boc-AAALPF-OH) is an initial compound. Partitioning of this peptide into model membrane environments composed of lipid micelles induces specific conformation(s) for the membrane-bound hexapeptide, as monitored by 75-MHz 13C-nmr spectral behavior of 13C-enriched Leu and Pro carbonyl carbons, and by 300-MHz 1H-nmr spectra of peptide alpha, beta, and aromatic protons. Data are interpreted in terms of an intramolecularly H-bonded inverse gamma-turn conformation in the membrane environment involving the Leu-Pro-Phe triad. The inherent structural instability of a Pro-containing segment in a TM helix due to the multiplicity of possible local conformations is discussed as a functional aspect of membrane-buried prolines in transport proteins.

Cloning and sequence analysis of the gene encoding invertase from the yeast Schwanniomyces occidentalis

The gene encoding invertase (INV) has been cloned from Schwanniomyces occidentalis. The enzyme consists of 533 amino acids, 8 potential glycosylation sites and has a 45% identity with the invertase from Saccharomyces cerevisiae. The proenzyme has a 22 amino acid signal sequence that has a high alpha-helical transmembrane potential which differs significantly from that predicted for the Saccharomyces cerevisiae enzyme.