Uncoupling Hydrophobicity and Helicity in Transmembrane Segments
1998; Elsevier BV; Volume: 273; Issue: 37 Linguagem: Inglês
10.1074/jbc.273.37.23645
ISSN1083-351X
Autores Tópico(s)Mitochondrial Function and Pathology
ResumoAlthough 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 usedde 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 α-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 θ222 = −17,000 ° (Pro) to −38,800 ° (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. 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 usedde 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 α-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 θ222 = −17,000 ° (Pro) to −38,800 ° (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. transmembrane Chou-Fasman 9-fluorenylmethoxycarbonyl high pressure liquid chromatography. Membrane-spanning segments of proteins are overwhelmingly α-helical (1Cheng A. van Hoek A.N. Yeager M. Verkman A.S. Mitra A.K. Nature. 1997; 387: 627-630Crossref PubMed Scopus (243) Google Scholar, 2Walz T. Hirai T. Murata K. Heymann J.B. Mitsuoka K. Fujiyoshi Y. Smith B.L. Agre P. Engel A. Nature. 1997; 387: 624-627Crossref PubMed Scopus (378) Google Scholar, 3Picot D. Loll P.J. Garavito R.M. Nature. 1994; 367: 243-249Crossref PubMed Scopus (1135) Google Scholar, 4Kuhlbrandt W. Wang D.N. Fujiyoshi Y. Nature. 1994; 367: 614-621Crossref PubMed Scopus (1525) Google Scholar, 5Pebay-Peyroula E. Rummel G. Rosenbusch J.P. Landau E.M. Science. 1997; 277: 1676-1681Crossref PubMed Scopus (820) Google Scholar). Previous studies of protein transmembrane (TM)1 segments have therefore tended to regard their helicity as a given parameter and focused principally on their hydropathy properties (6Engelman D.M. Steitz T.A. Goldman A. Annu. Rev. Biophys. Biophys. Chem. 1986; 15: 321-353Crossref PubMed Scopus (1188) Google Scholar, 7Kyte J. Doolittle R.F. J. Mol. Biol. 1982; 157: 105-132Crossref PubMed Scopus (16899) Google Scholar). Yet considering the array of residues occurring in TM domains, including those considered classically as helix breakers (i.e. Gly, Ile, Val, Thr, and Pro) in aqueous-based proteins, it appears unlikely that all residues will possess identical helical propensities in the membrane environment. Previous studies of peptide/protein secondary structures in aqueous solution indicate that individual amino acids have distinct conformational preferences (α-helical and β-sheet) that influence protein structure and folding (8Chou P.Y. Fasman G.D. Biochemistry. 1974; 13: 211-222Crossref PubMed Scopus (1820) Google Scholar, 9Chakrabartty A. Kortemme T. Baldwin R.L. Protein Sci. 1994; 3: 843-852Crossref PubMed Scopus (587) Google Scholar, 10Blaber M. Zhang X.-J. Matthews B.W. Science. 1993; 260: 1637-1640Crossref PubMed Scopus (434) Google Scholar, 11Minor Jr., D.L. Kim P.S. Nature. 1994; 367: 660-663Crossref PubMed Scopus (540) Google Scholar, 12Munoz V. Serrano L. Proteins. 1994; 20: 301-311Crossref PubMed Scopus (275) Google Scholar, 13O'Neil K.T. DeGrado W.F. Science. 1990; 250: 646-651Crossref PubMed Scopus (1111) Google Scholar). However, in membranes, fatty acyl chains of lipid molecules present a hydrophobic environment to embedded proteins, and consequently, protein folding in membranes is subject to different rules versus those governing proteins in the aqueous milieu (14Chang D.K. Chien W.J. Arunkumar A.I. Biophys. J. 1997; 72: 554-566Abstract Full Text PDF PubMed Scopus (8) Google Scholar, 15Kothekar V. Mahajan K. Raha K. Gupta D. J. Biomol. Struct. Dyn. 1996; 14: 303-316Crossref PubMed Scopus (7) Google Scholar, 16Lee S. Kiyota T. Kunitake T. Matsumoto E. Yamashita S. Anzai K. Sugihara G. Biochemistry. 1997; 36: 3782-3791Crossref PubMed Scopus (24) Google Scholar). The need for such information has become more crucial, because the genomes of various organisms recently analyzed indicate that large proportions (20–40%) of the genes correspond to membrane proteins. We have now examined this situation using a set of de novodesigned Ala-based peptides. Features of these peptides mimic a prototypical single-spanning membrane domain with respect to length, hydrophobicity, and residue occurrence. An excellent correlation is found between the experimentally measured helicity in the non-polar phase and the occurring frequency of a given residue computed from a data base of native protein TM helices. These findings identify a novel pathway through which the amino acid composition of a membrane protein can assist its accommodation in a hydrophobic environment. Peptides were synthesized by the continuous flow Fmoc solid phase method (17Liu L.-P. Deber C.M. Biochemistry. 1997; 36: 5476-5482Crossref PubMed Scopus (104) Google Scholar). C termini of peptides were aminated after cleavage from NovaSyn KR 125 resin. Purification of peptides was carried out on a reverse-phase Vydac-C4 semi-preparative HPLC (10 × 250 mm, 300 Å), using a linear gradient of acetonitrile in 0.1% trifluoroacetic acid. Purified peptides were characterized by analytical HPLC, amino acid analysis, and mass spectrometry. The aggregation states of peptides were monitored by CD measurements and by size exclusion HPLC, from which it was found that through the concentration range of 5–250 μm, peptides remained monomeric in all experimental media. Concentrations of peptides were determined in triplicate through quantitative amino acid analysis using Ala recovery as the standard and BCA protein assay (enhanced protocol). Peptides were stored as solid powders at −20 °C. To avoid the complexity of synthesizing a multiple Cys-containing peptide, only the central X residue was substituted by Cys, and the other two X residues were replaced by Leu. CD measurements were performed on a Jasco-720 spectropolarimeter using a 1-mm path length quartz cell at 25 °C. Each spectrum was the average of four scans with buffer background subtracted. Curves reported are based on triplicate measurements; standard deviation is ± 1%. Peptide concentration was typically 30 μm in aqueous buffer and in n-butanol. The aqueous buffer was prepared from 10 mm Tris-HCl, 10 mm NaCl, pH 7.0. The rationale for the design of the peptides used in the present study, of sequence KKAAAXAAAAAXAAWAAXAAAKKKK-amide (whereX indicates one of the 20 commonly occurring amino acids) (17Liu L.-P. Deber C.M. Biochemistry. 1997; 36: 5476-5482Crossref PubMed Scopus (104) Google Scholar, 18Liu L.-P. Li S.-C. Goto N.K. Deber C.M. Biopolymers. 1996; 39: 465-470Crossref PubMed Scopus (68) Google Scholar), is given as follows: (i) The hydrophobic segment of peptide is comprised of 19 amino acids, which when folded into an α-helical conformation is of sufficient length to span a phospholipid bilayer (19Reithmeier R.A.F. Curr. Opin. Struct. Biol. 1995; 5: 491-500Crossref PubMed Scopus (143) Google Scholar). (ii) Distributions of the three "guest" residues Xhave been designed to preserve both angular and longitudinal symmetry around the helix, thereby minimizing any bias from amphipathic character that may arise when X is a polar or charged residue. In addition, triple substitutions of guest residueX in the hydrophobic core serve to amplify the effect of guest replacements, ensuring that the spectroscopic measurements can detect their structural impact. (iii) Ala, the most appropriate background residue as demonstrated by previous studies (20Deber C.M. Li S.C. Biopolymers. 1995; 37: 295-318Crossref PubMed Scopus (113) Google Scholar, 21Li S.C. Deber C.M. Nat. Struct. Biol. 1994; 1: 368-373Crossref PubMed Scopus (139) Google Scholar), was chosen as the "host" residue. (iv) A Trp residue was incorporated into the hydrophobic segment as a fluorescence probe to monitor characteristics of the local microenvironment. (v) Lys residues were added at N and C termini to enhance the water solubility of peptides, whereas the C terminus was aminated to eliminate potential electrostatic attractions that might occur inter- or intramolecularly. The solubility of these peptides in a wide variety of media make it feasible to investigate and compare specifically the conformational preferences of each residue in aqueous buffer versusnon-polar media. Because the only difference among the present peptides is the guest X residues, it is reasonable to assume that variations in peptide helicity can be interpreted as a manifestation of the propensity for each X amino acid to form an α-helical structure in a specific environment. CD spectroscopy, which is sensitive to secondary structure of proteins and polypeptides in solution, was used to monitor the conformations of model peptides. In aqueous buffer, peptides formed partial or non-helical conformations as shown by θ222data in Table I and CD spectra shown for selected peptides (Fig. 1 A); a generally good correlation was obtained when comparing peptide helicities with Pα values predicted by Chou and Fasman (Fig. 1 B) (8Chou P.Y. Fasman G.D. Biochemistry. 1974; 13: 211-222Crossref PubMed Scopus (1820) Google Scholar), as well as with those obtained in several experimental systems (9Chakrabartty A. Kortemme T. Baldwin R.L. Protein Sci. 1994; 3: 843-852Crossref PubMed Scopus (587) Google Scholar, 10Blaber M. Zhang X.-J. Matthews B.W. Science. 1993; 260: 1637-1640Crossref PubMed Scopus (434) Google Scholar, 12Munoz V. Serrano L. Proteins. 1994; 20: 301-311Crossref PubMed Scopus (275) Google Scholar). Because this group of algorithms is based on data from water-based peptides and proteins, these results imply that the present peptide design is an appropriate model to derive the helix propensities of each amino acid. That the Asp peptide showed relatively high helical propensity in this series of peptides, as does the Glu peptide, may be due to the fact that salt bridges potentially formed between these residues and that the terminal Lys residues may extend and stabilize peptide helices. However, any such overestimation of helicity for these two peptides is probably minor, because fairly good correlation is obtained when plotting their helicity in aqueous solution versus the C.-F. Pα parameter (Fig. 1 B).Table IHelical-forming tendency of model peptides KKAAAXAAAAAXAAWAAXAAAKKKK-amide in nonpolar and aqueous media, and α-helical preferences of amino acids derived from protein transmembrane segmentsRank orderInn-butanolX residue [P α (TM)]In Aq.X residue (−θ222 × 10−4)X residue (−θ222 × 10−4)1I (3.88)I (2.69)E (2.25)2L (3.84)L (2.53)A (2.06)3V (3.82)V (2.32)L (1.80)4F (3.78)A (1.86)I (1.54)5A (3.72)F (1.63)D (1.47)6M (3.67)G (1.36)R (1.35)7G (3.44)M (1.26)M (1.33)8Y (3.33)C (0.76)Q (1.19)9C (3.28)T (0.70)Y (1.16)10T (3.27)S (0.63)W (1.13)11W (3.20)W (0.45)F (1.12)12S (2.99)Y (0.39)K (1.12)13H (2.90)P (0.30)V (1.03)14Q (2.87)H (0.27)C (0.91)15R (2.84)Q (0.11)T (0.90)16N (2.82)N (0.10)H (0.79)17D (2.66)R (0.08)S (0.79)18K (2.65)K (0.08)N (0.59)19E (2.54)D (0.06)G (0.34)20P (1.70)E (0.05)P (0)θ222nm in CD spectra is used as a measure of helicity. Values reported are based on triplicate measurements, with a standard deviation of ±1%. Peptide concentrations were 30 μm.P α was derived from the data of si reported in Ref. 38Jones D.T. Taylor W.R. Thornton J.M. Biochemistry. 1994; 33: 3038-3049Crossref PubMed Scopus (700) Google Scholar; see text for details.Aq., aqueous buffer (10 mm Tris-HCl, 10 mm NaCl, pH 7.0). Open table in a new tab θ222nm in CD spectra is used as a measure of helicity. Values reported are based on triplicate measurements, with a standard deviation of ±1%. Peptide concentrations were 30 μm. P α was derived from the data of si reported in Ref. 38Jones D.T. Taylor W.R. Thornton J.M. Biochemistry. 1994; 33: 3038-3049Crossref PubMed Scopus (700) Google Scholar; see text for details. Aq., aqueous buffer (10 mm Tris-HCl, 10 mm NaCl, pH 7.0). Organic solvents with dielectric constants between pure water and the hydrocarbon interior of biological membranes have been widely used to mimic the non-polar environments of membranes (22Chung L.A. Anal. Biochem. 1997; 248: 195-201Crossref PubMed Scopus (22) Google Scholar, 23Li S.-C. Goto N.K. Williams K.A. Deber C.M. Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 6676-6681Crossref PubMed Scopus (198) Google Scholar, 24Ben-Efraim I. Bach D. Shai Y. Biochemistry. 1993; 32: 2371-2377Crossref PubMed Scopus (43) Google Scholar, 25Vinogradova O. Badola P. Czerski L. Sonnichsen F.D. Sanders C.R. Biophys. J. 1997; 72: 2688-2701Abstract Full Text PDF PubMed Scopus (63) Google Scholar). To create a quasi-membranous yet homogenous (isotropic) environment and to eliminate the complexities involved in protein/peptide-lipid interfacial interaction (viz. partitioning, electrostatics), peptides were dissolved in n-butanol, a moderate non-polar solvent of dielectric constant 17.8 at 25 °C (26Lide D.R. CRC Handbook of Chemistry and Physics. 72nd Ed. CRC Press, Inc., Boca Raton, FL1991Google Scholar), and their CD spectra were recorded. n-Butanol was chosen as a membrane-mimetic environment on the basis that (i) it effectively dissolved all 20 peptides, and (ii) in comparison to iso-propanol,n-propanol, and iso-butanol, n-butanol produced only minimum background absorbance at the lower wavelength region (<200 nm) of the CD measurements. Experiments with selected peptides in the above four solvents demonstrated that the CD spectra for a given peptide were essentially superimposable, indicating that the measured ellipticity is a property largely determined by the amino acid sequence of a peptide in this set of related solvents. Similar results were obtained for alcohol-induced helix formation of melittin, where end point ellipticities were essentially independent of the alcohol used (27Hirota N. Mizuno K. Goto Y. J. Mol. Biol. 1998; 275: 365-378Crossref PubMed Scopus (225) Google Scholar). In contrast to results obtained in aqueous buffer, all peptides formed predominantly α-helical conformations in n-butanol (Table I; spectra shown for selected peptides in Fig. 2 A). CD spectra were found to be independent of peptide concentration over the range 7.5–120 μm (not shown), and thus the helicity differences observed among the peptides are not caused by their differential solubility but by their intrinsic propensities to form the α-helical conformation in a non-polar environment. Aromatic residues are known to contribute specifically to CD spectra in regions ascribed to secondary structure (28Chakrabartty A. Kortemme T. Padmanabhan S. Baldwin R.L. Biochemistry. 1993; 32: 5560-5565Crossref PubMed Scopus (330) Google Scholar). Although a Trp residue is incorporated at the midsegment of all peptides as an integral part of the host sequence, no specific bias or distortion of spectra was observed. However, in those peptides containing multiple aromatic residues (i.e. when X indicates Trp, Phe, and Tyr at guest positions), aromatic contributions may be amplified, as suggested by the "off-line" points observed for Trp, Phe, and Tyr peptides in Figs. 1 B and 2 B. The observation that there is a wide range of helical preferences for individual amino acids (∼35% drop from Ile to Glu, and ∼56% drop from Ile to Pro) in n-butanol appears to be a direct function of the ability of the solvent to "solvate" a given side chain. For example, amino acids with relatively non-polar side chains (Ile, Leu, Val, Ala, Gly, and Phe) showed higher helical-forming tendency than those with polar and charged side chains. Comparing the value of θ222 of each peptide in aqueous bufferversus n-butanol, we found that there are drastic differences regarding the rank order of helical propensity for these residues in the two media (Table I). Environment is thus a key regulatory factor in determining the conformation adopted by a given protein sequence. As a further example, although Gly and Pro are among the most destabilizing residues of α-helices in globular proteins and polypeptides (8Chou P.Y. Fasman G.D. Biochemistry. 1974; 13: 211-222Crossref PubMed Scopus (1820) Google Scholar, 9Chakrabartty A. Kortemme T. Baldwin R.L. Protein Sci. 1994; 3: 843-852Crossref PubMed Scopus (587) Google Scholar, 10Blaber M. Zhang X.-J. Matthews B.W. Science. 1993; 260: 1637-1640Crossref PubMed Scopus (434) Google Scholar, 29Altmann K.H. Wojcik J. Vasquez M. Scheraga H.A. Biopolymers. 1990; 30: 107-120Crossref PubMed Scopus (41) Google Scholar), they do display a considerable tendency to form α-helices in membrane environments (see Table I). Li et al. (23Li S.-C. Goto N.K. Williams K.A. Deber C.M. Proc. Natl. Acad. Sci. U. S. A. 1996; 93: 6676-6681Crossref PubMed Scopus (198) Google Scholar) also found that in a membrane environment, suitably placed Pro tends to protect rather than break α-helical structures. Based on these data, it should not be unexpected that one finds Gly and Pro quite commonly in native transmembrane helices, especially Pro in transporter proteins (30Deber C.M. Glibowicka M. Woolley G.A. Biopolymers. 1990; 29: 149-157Crossref PubMed Scopus (55) Google Scholar, 31Rao J.K.M. Argos P. Biochim. Biophys. Acta. 1986; 869: 197-214Crossref PubMed Scopus (406) Google Scholar, 32Brandl C.J. Deber R.B. Hsu L.C. Woolley G.A. Young X.K. Deber C.M. Biopolymers. 1988; 27: 1171-1182Crossref PubMed Scopus (14) Google Scholar). Notably, Ser and Thr display a higher tendency to occur in TM helices than in aqueous-based helices. These residues can satisfy their hydrogen-binding capacity by forming H bonds with the carbonyl oxygen in the preceding turn of the α-helix, permitting such residues to occur in helices buried within a hydrophobic milieu (33MacKenzie K.R. Prestegard J.H. Engelman D.M. Science. 1997; 276: 131-133Crossref PubMed Scopus (865) Google Scholar, 34Engelman D.M. Steitz T.A. Cell. 1981; 23: 411-422Abstract Full Text PDF PubMed Scopus (597) Google Scholar, 35Gary T.M. Matthews B.W. J. Mol. Biol. 1984; 175: 75-81Crossref PubMed Scopus (223) Google Scholar). Several workers have analyzed the membrane inclusion preference of individual amino acids based on known membrane proteins (36Landolt-Marticorena C. Williams K.A. Deber C.M. Reithmeier R.A. J. Mol. Biol. 1993; 229: 602-608Crossref PubMed Scopus (325) Google Scholar, 37Taylor W.R. Jones D.T. Green N.M. Proteins. 1994; 18: 281-294Crossref PubMed Scopus (95) Google Scholar, 38Jones D.T. Taylor W.R. Thornton J.M. Biochemistry. 1994; 33: 3038-3049Crossref PubMed Scopus (700) Google Scholar), analogous to the C.-F. algorithm for globular proteins (8Chou P.Y. Fasman G.D. Biochemistry. 1974; 13: 211-222Crossref PubMed Scopus (1820) Google Scholar); for example, Jones et al. (38Jones D.T. Taylor W.R. Thornton J.M. Biochemistry. 1994; 33: 3038-3049Crossref PubMed Scopus (700) Google Scholar) used a dynamic programming algorithm to provide a membrane topology model from single sequence information. To obtain a membrane-based predictor comparable with the conventional C.-F. Pα parameter, we converted the "log likelihood" parameter of Jones et al.(si) to: P α(TM) =qi/pi, where P α(TM), the preference of an individual amino acid for occurrence in a TM helix, is the ratio of qi, the relative frequency of occurrence of the amino acid i in a particular structural class, topi, the relative frequency of occurrence of the amino acidi in all the sequences in the data set. The resulting values of P α(TM) derived from the "helix middle" of single-spanning membrane proteins are listed in Table I. There is good correlation (r = 0.93 when Pro was excluded) between the θ222 (n-butanol) and theP α(TM) (Fig. 2 B). Thus, rather than simple identification of highly hydrophobic segments in the primary sequences of membrane proteins, the uncoupling of hydrophobicity from helicity in transmembrane domains allows a clearer delineation as to where helices are highly probable as a function of both protein sequence and environment. The correspondence observed between the experimentally determined helical propensity for individual amino acids and their non-random occurring frequency in protein TM helices suggests that the high frequency of occurrence in membranes of residues such as Leu, Val, Ile, and Phe derives not only from their hydrophobicity but also from their intrinsic propensity to form the α-helical conformation in the non-polar environments of membranes.
Referência(s)