- Email: [email protected]
Formation and stability of β-hairpin structures in polypeptides
107
Formation and stability of p-hairpin structures in polypeptides Francisco Blanco*, Marina Ramirez-Alvarado t and Luis Serrano+ Experimental work on peptide models with ~-hairpin structures has provided new insights into the formation and stability of this secondary str~cture element. Both the turn region and the antiparallel strand residues not only affect the overall stability of the hairpin, but also determine the o/pe of hairpin formed. These results agree reasonably well with those from experimental and statistical analyses of ~-sheet structures in proteins.
<1>,'11 analysis of the protein database [7], although there is an important context effect caused due to the protein environment [5,7,8].
Figure 1
lie
(a) Addresses Biostructure and Biocomputing, European Molecular Biology Laboratory, Meyerhofstrasse 1, Heidelberg 069117, Germany ·e-mail: [email protected] te-mail: [email protected] *e-mail: [email protected] Correspondence: luis Serrano
NH3+ -Arg-Gly
H
0
Val
N
-{y T~\ ~
coo-- A,g·G~
H
-
•
I I I
Current Opinion in Structural Biology 1998, 8:107-111
Iffir
H
\ ,~
I I I
Tyr
H
'j""
Asn
N
0I
I I I
I I
H
I'XY.J.X),Y1G~
Jv~ I~
IH
N
vH
0
H
Lys
H
0
http://biomednet.com/elecref/0959440X00800107
© Current Biology ltd ISSN 0959-440X
(b)
Abbreviations BPTI bovine pancreatic trypsin inhibitor NOE nuclear Overhauser enhancement
Introduction The ~ hairpin is the simJ1lest form of-an antiparallel ~ sheet and is defined as a turn region flanked by two strands with a defined backbone hydrogen-bonding pattern (Figure 1). There are several types of hairpins depending of the type of turn. In proteins, 2:2 hairpins are the most common (Figure 1a), followed by 3:5 hairpins (Figure 1b), 4:4 hairpins (Figure Ie) and other less abundant conformations. The two number nomenclature was introduced by Sibanda et 01. [1] and it refers to the number of the residues in the turn (or not in strand conformation) \inder two different criteria. The first is the number of strand residues whose amide or carboxyl moiety is involved in the typical pattern of ~ sheets. The second is the number of strand residues with both the amide and carboxyl participating in the ~-sheet hydrogen-bonding pattern. Unlike formation of a helices, formation of ~ sheets is not yet well understood, because of its inherent complexity (distinct substructures must be considered: the turn; the strands and their relative interplay) and the lack of simple model systems akin to those used for studying a-helix formation. Proteins have been used to derive experimental scales of the propensity of ~-sheet formation for the 20 natural amino acids [2-5] as well as to study of the effect of different residue pairs in an antiparallel ~ sheet [6]. The intrinsic propensities correlate well among themselves and with a scale derived from
(c)
Current Opinion in Structural Biology
Schematic representation of ~-hairpin structures observed in isolated linear peptides in solution. (a) A 2:2 ~ hairpin with a type I' ~ turn. The sequence of the peptide was analyzed by Ramirez-Alvarado et a/. [15 aal. (b) A 3:5 ~ hairpin with a type I ~ turn and a ~ bulge. The sequence comes from a peptide analyzed by de Alba et a/. [20 aa]. (c) A 4:4 ~ hairpin with with a type I ~ turn of the same sequence as (b).
Model peptides should enable analysis of the sequence determinants of secondary structure formation in a context-
108
Folding and binding
free environment, as has been achieved with a helices. Most natural peptides encompassing hairpins, however, are mainly devoid of structure in water or form aggregates [9,10], with one exception-the 41-56 residue fragment of the protein G Bl domain [11). This is not surprising since secondary structure propensities are normally low in natural sequences [12) and the same forces that could lead to intramolecular ~-sheet formation can also drive intermolecular aggregation. Recently, the development of simple peptide models of ~-hairpin structures [13,14,15° has enabled useful information to be obtained on this otherwise elusive secondary structure element. This review focuses on the experimental work on these peptide models and discusses the prospects and challenges of hairpin analysis. 0
)
Peptide models of hairpin structures The first reported case of a ~-hairpin-forming linear peptide is a modified sequence of a 2:2 ~ hairpin from the a-amylase inhibitor tendamistat [13). The native sequence forms transiently populated turns at the central region but no ~-hairpin structure. Replacement of the turn residues by the sequence with the highest tendency to form type I ~ turns (Asn-Pro-Asp-Gly), results in the formation of a 3:5 non-native hairpin with a three-residue turn (type I turn plus a ~ bulge) (Figure Ib). This causes a shift of one residue in the antiparallel alignment of the strands, altering the pairing between the interstrand residues. Similar results were found with a 16-residue peptide derived from the first 2:2 ~ hairpin of ubiquitin, in which the same sequence was introduced in the turn region [14). The presence of these non-native hairpins could be due to the incompatibility of the geometry of a type I turn with the right-handed twist typical of ~ sheets in proteins (16). Type I' or II' turns are normally associated with 2:2 hairpins in proteins, with a geometry that matches the right-handed twist of the ~ sheet [1). This hypothesis is strongly supported by the de novo design of a linear peptide (BH8) based on an analysis of 2:2 hairpins in the protein database (Figure la) [15°°). In this peptide, the sequence Asn-Gly was selected for the central turn residues as it is the most abundant in 2:2 hairpins with type I' turns. The strand residues were selected based on apparently favored combinations of residues in the strands of protein hairpins, also matching criteria to avoid aggregation. This designed peptide folds into the expected 2:2 ~-hairpin structure with a population around 35% and is an ideal system by which to analyze the sequence determinants of 2:2 ~-hairpin formation. Other direct evidence of the turn sequence that favors type I' or II' ~ turns in 2:2 hairpin formation was found by replacing the L-Pro residue at the turn of the ubiquitin-derived peptide with D-Pro, with opposite chirality. In this case, a 2:2 hairpin was formed with type I' or II' turns, depending on the residue accompanying the Pro at position i+ 1 (I' for D-Pro-D-Ala, and II' for
D-Pro-Gly and D-Pro-Ala) [17°°]. Similar results were found in an analogous study of a different peptide sequence containing D-Pro-Gly at the turn region [18].
The turn region The importance of turn stability in hairpin formation is clearly shown by the analysis of a series of peptides based on the BH8 sequence. The peptides all contain the sequence Xxx-Gly in the turn region, with Xxx being either asparagine, aspartate, serine, alanine or glycine. All the peptides form 2:2 ~ hairpins but with different populations with respect to the random-coil conformations [19·°]. The order of hairpin stabilization is Asn:2: Asp> Gly > Ser:2: Ala. The relative positions of asparagine, serine and alanine within this hierarchy can be explained by their intrinsic residue preferences in populating the backbone dihedral angles corresponding to the first position of a type I' turn. Aspartate and glycine were found to be more and less stabilizing, respectively, than expected from their CI>,'P propensities. Exactly the same relative abundancies, however, were found at that position in 2:2 ~ hairpins in proteins (Figure 2). In the case of aspartate, there is a specific sidechain rotamer preference that creates a favorable electrostatic interaction with its own amide group when it is preceded by a residue in an extended conformation. The analysis of mutants of a designed sequence that adopts different populations of 3:5 and 4:4 ~ hairpins (Figure 1b,c), enabled de Alba et 01. [20··,21,22] to identify several factors that could selectively increase the type of ~ hairpin formed. The observed effects were again consistent with the statistical analysis of the protein database. An interaction between asparagine and threonine at the beginning of the turn region (Figure Ib) stabilized the 3:5 hairpin. A residue with a higher intrinsic preference in adopting backbone dihedral angles in the UR region of a Ramachandran plot than asparagine stabilized the 4:4 hairpin (Figure lc). The introduction of the native turn sequence of the ~ hairpin from the protein bovine pancreatic trypsin inhibitor (BPTI) leads to the formation of a single 4:4 hairpin, as in BPTI. A sequence that almost exclusively adopts 2:2 hairpins was obtained when the turn residues were replaced by the sequence Tyr-Asn-Gly-Lys.
The strands Mutagenesis analysis of the BH8 peptide has shown that interstrand sidechain-sidechain interactions contribute significantly to hairpin stability, although it is difficult to discriminate these interactions from the change in intrinsic secondary structure propensities [15°°]. Recently, the BH8 sequence was elongated by placing oppositely charged residues at the first and last positions of the hairpin structure, resulting in a significant increase in stability (M RamIrez-Alvarado, F] Blanco, L Serrano, unpublished data). This supports the hypothesis that interstrand sidechain-sidechain interactions are important in the formation of hairpin structure.
Formation and stability of
~·hairpin
Figure 2
rnc:
--
25
Y = -1~.499 + 0.87314x R= 0.98261
BHa
~
:::3
20
•
Q)
Co
~
-
.s::.
.~
BH9
15
(/)
'in .s::.
10
BH13
.~ (/)
Q)
109
compnsmg the hairpin structure are conformationally constrained and display correlated
c:
'2-
structures in polypeptides Blanco, Ramirez-Alvarado and Serrano
5
(/)
CO ()
BH12
• • BH11
0 15
20
25
30
35
40
% Hairpin population Current Opinion in Structural Biology
Correlation between the hairpin population of the corresponding peptides and relative residue frequencies at the first position of the ~ turn in protein ~ hairpins. The peptides are based on the sequence of the peptide described in Figure 1a. Mutations have been made at position L1 of the type I' turn of the ~ turn in BH8 [19 001; BH9 has aspartate at position L1; BH11 has alanine; BH12 has glycine; and BH13 has serine. The y-axis indicates the number of ~ hairpins with a type I' turn and a glycine at position L2, as found in the database of protein structures. The x-axis reflects the average ~-hairpin population estimated from NMR measurements.
Rico and co-workers [23 have explored the effect of the strand residues on the equilibrium between the 3:5 and 4:413-hairpin structures and the random-coil conformations in their peptide systems. Several modifications of the sequence of the N-terminal 13 strand showed that the effect on the stability of one hairpin structure is independent of the effect on the other and that 3:5 hairpins are less susceptible to mutations, suggesting they are more stable than 4:4 hairpi.ns. The higher overall preference for 3:5 hairpins was r~lated to increased efficiency of sidechain packing and burying of hydrophobic surface area when compared to 4:4 hairpins. 0
]
Kinetic analysis of the folding reaction of protein G residues 41-56 has been analyzed by the temperature jump method [26 The folding of this hairpin is slower than in the case of polyalanine based a helices (microseconds as opposed to nanoseconds) and theoretical analysis suggests that the distance between the favorable sidechain-sidechain interactions and the 13 turn could strongly influence the folding speed. The analysis of more peptides with different sequences and turns is needed before generalizing all hairpin and sheet structures. 00
The description of the ~·hairpin formation: population quantification While quantification of the structured population in helical peptides is reasonably straightforward, it is not the case for 13 hairpins. Far-UV circular dichroism spectra analysis is not the most appropriate method for estimating the hairpin population, in contrast to a-helical structures, because of the much smaller signal and strong influence of the l3-turn conformation [15 On the other hand, different NMR parameters can be used for estimating population at a residue level. The structural NOE intensities, the chemical shifts of the Cu protons and Ca proton-amide proton coupling constants can be used to estimate the hairpin population, assuming a simple two-state model, with hairpin and random-coil conformations in fast exchange [14,15 .,20 Each parameter has its own drawbacks, but combined they can yield an estimate of the hairpin population as accurate as the two-state assumption allows it to be [19. 00
0
0
Dynamic and kinetic properties of peptides
].
].
00
].
].
~·hairpin
13C-NMR relaxation data as well as data from extensive molecular dynamic simulations have been determined [24,25] for the modified Tendamistat peptide [13]. The combined data suggest that the peptide has significantly populated hydrogen bonds, and transient loose interactions occur in related but distinct conformations between hydrophobic and terminal charged groups. Recently, the BH8 peptide system was selectively 13C-enriched at different positions and 13C-NMR relaxation experiments were performed. It was found that those residues
Implications for protein stability In the case of a helices, it has been shown that by locally increasing helical propensities it is possible to significantly increase the resistance of proteins to thermal and chemical denaturation [27,28]. de Grado and co-workers [29 have shown that it is possible to increase the thermal stability of a protein by locally modifying the l3-turn region of a 13 hairpin. Recently, it was discovered that changing the Asn-Asp sequence of a 2:2 13 hairpin of an SH3 domain to the more favorable type I' l3-turn sequence, Asn-Gly, significantly increases the stability of the protein 00
]
110
Folding and binding
to chemical and thermal denaturation, as well as the folding rate (J MartInez, L Serrano, unpublished data). Once we acquire the level of knowledge that we have already for a helices, it should be possible to rationally modify protein stability in ~-sheet-containing proteins.
7.
Munoz v, Serrano L: Intrinsic secondary structure propensities of the amino acids, using statistical phi-psi matrices. Comparison with experimental scales. Proteins 1994, 20:301311.
8.
Otzen DE, Fersht AR: Side-chain determinants of stability. Biochemistry 1995, 34:5718-5724.
9.
Ramirez-Alvarado M, Serrano L, Blanco FJ: Conformational analysis of peptides corresponding to all the secondary structure elements of protein L 81 domain: Secondary structure propensities are not conserved in proteins with the same fold. Protein Sci 1997, 6:162-174.
10.
Viguera AR, Jimenez MA, Rico M, Serrano L: Conformational analysis of peptides corresponding to ~-hairpins and a ~ sheet, that represent the entire sequence of a-spectrin SH3domain. J Mol Bioi 1995, 255:507-521.
11.
Blanco FJ, Rivas G, Serrano L: A short linear peptide that folds into a native stable ~-hairpin in aqueous solution. Nat Struct Bioi 1994, 1:399-409.
12.
Munoz v, Cronet P, L6pez-Hernimdez E, Serrano L: Analysis of the effect of lOcal interactions in protein stability. Fold Des 1996, 1:167-1 78.
13.
Blanco FJ, Jimenez MA, Rico M, Santoro J, Herranz J, Nieto JL: Evidence of a short linear peptide that folds into native stable ~-hairpin in aqueous solution. JAm Chem Soc 1993, 115:58875888.
14.
Searle MS, Williams DH, Packman LC: A short linear peptide derived from the N-terminal sequence of ubiquitin folds into a water-stable non-native ~-hairpin. Nat Struct Bioi 1995, 2:9991006.
Conclusions There are already adequate peptide models of ~-hair pin structures that provide information about specific interactions at both the turn region and the strands. The overall stability of the hairpin structure is the result of many different contributions, including mainchain hydrogen bonding, secondary structure propensities and sidechain-sidechain interactions. Depending on the sequence, some of these factors are more important than others. In general, there is a good correlation between the experimental data and the statistical analysis of the protein database. As more experimental information becomes available, a description of hairpin formation in an algorithmic form will be more feasible and could be used to rationally stabilize ~-sheet-containing proteins. The description of hairpin formation will be more difficult than for a helices, because the number of different interactions that need to be evaluated is larger. There are different hairpin types that must be considered as alternative structures for a single sequence, and the population quantification methods need to be cross-checked. The development of theoretical approaches that allow exploration of the conformational space of different sequences in fixed frameworks within an adequate forcefield, in combination with the experimental analysis of model peptides, will enable us, in the future, to reach the same level of knowledge that we presently have for a helices.
Acknowledgements We thank Maria Angeles Jimenez and colleagues for sending us preprints of their work. Francisco Blanco is a fellow of the Spanish Ministerio de Educaci6n y Ciencia.
~-sheet
Ramirez-Alvarado M, Blanco FJ, Serrano L: De novo design and structural analysis of a model ~-hairpin peptide system. Nat Struct Bioi 1996, 3:604-612. This paper describes a peptide designed completely de novo that folds as a monomeric ~ sheet. Substitution of the strand residues by alanine results in abolishment of the hairpin structure, suggesting that sidechain-sidechain interactions contribute significantly to hairpin stability. 15.
16.
Chothia C: Conformation of twisted ~-pleated sheets in proteins. J Mol Bioi 1973, 75:295-302.
Haque TS, Gellman SH: Insights on ~-hairpin stability in aqueous solution from peptides with enforced type I' and type II' ~-turns. JAm Chem Soc 1997, 1197:2303-2304. The authors have done an interesting study incorporating D-stereoisomers residues at one or both turn positions of a 16-residue peptide that corresponds to the N-terminal segment of ubiquitin, in order to create "mirror image turns". Their results indicate that type I' and type II' ~ turns are required for the formation of a two-residue turn ~ hairpin. 17.
18.
Karle IL, Awasthi SK, Balaram P: A designed ~-hairpin peptide in crystals. Proc Natl Acad Sci USA 1996, 93:8189-8193.
Ramirez-Alvarado M, Blanco FJ, Niemann H, Serrano L: Role of ~-turn residues in ~-hairpin formation and stability in designed peptides. J Mol Bioi 1997, 273:898-912. Mutagenesis of the first central residue in the type I' ~ turn from a model peptide system that folds as a ~ hairpin shows the importance of the turn in hairpin stability and illustrates the idea that the presence of the turn does necessarily alter tHe pairing of the strands. Interestingly, there is a very good correlation between the effect of the different residues on hairpin stability and their relative abundance in the protein database. 19.
References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: • ••
of special interest of outstanding interest
1.
Sibanda BL, Blundell TL, Thornton JM: Conformation of ~ hairpins in protein structures. A systematic classification with applications to modelling by homology, electron density fitting and protein engineering. J Mol Bioi 1989, 229:759-777.
2.
Kim CA, Berg JM: Thermodynamic ~-sheet propensities measured using a zinc-finger host peptide. Nature 1993, 362:267-270.
3.
Minor DL, Kim PS: Measurement of the ~-sheet-forming propensities of amino acids. Nature 1994, 367:660-663.
4.
Smith CK, Withka JM, Regan L: A thermodynamic scale for the ~-sheet forming tendencies of the amino acids. Biochemistry 1995, 33 :5510-551 7.
20.
De Alba E, Jimenez MA, Rico M: Turn residue sequence determines ~-hairpin conformation in designed peptides. J Am Chem Soc 1997, 119:175-183. Different peptides were designed to fold into different ~-hairpin conformations in aqueous solution and were analyzed in order to further the understanding of the role of the turn sequence in defining ~-hairpin structure. 21.
De Alba E, Blanco FJ, Jimenez MA, Rico M, Nieto JL: Interactions responsible for the ~-hairpin conformation population formed by a designed linear peptide. Eur J Biochem 1995, 233:283292.
22.
De Alba E, Jimenez MA, Rico M, Nieto JL: Conformational investigation of designed short linear peptides able to fold into ~-hairpin structures in aqueous solution. Fold Des 1996, 1:122-144.
23.
De Alba E, Rico M, Jimenez MA: Cross-strand side chain interactions versus turn conformation in ~-hairpins. Protein Sci 1997, 6:2548-2560.
~-sheet
5.
Minor DL, Kim PS: Context is a major determinant of propensity. Nature 1994, 371 :264-267.
6.
Smith CK, Regan L: Guidelines for protein design. The energetics of ~-sheet side chain interactions. Science 1995, 270:980-982.
Formation and stability of [3-hairpin structures in polypeptides Blanco, Ramirez-Alvarado and Serrano
111
The relative stability of 3:5 and 4:4 hairpins are investigated by mutating certain residues at the strand regions.
a theoretical model that explains the kinetic results and could be used in the future to quantitatively analyze hairpin formation.
24.
27.
Munoz V, Cronet P, Lopez-Hernandez E, Serrano L: Analysis of the effect of local interactions in protein stability. Fold Des
28.
Virtudes V, Viguera AR, Aviles FX, Serrano L: Stabilisation of proteins by rational design of a-helix stability using helix/coil transition theory. Fold Des 1996, 1:29-34.
29.
Zhou HX, Hoess RH, deGrado WF: In vitro evolution of thermodynamically stable turns. Nat Struct 810/1996, 3:446-
Constantine KL, Mueller L, Andersen NH, Tong H, Wandler CF, Friederichs MS, Bruccoleri RE: Structural and dynamic properties of a [3-hairpin-forming linear peptide. 1. Modeling using ensemble-averaged constraints. JAm Chem Soc 1995,
117:10841-10854. 25.
Friederichs MS, Stouch TR, Bruccoleri RE, Mueller L, Constantine KL: Structural and dynamic properties of a [3-hairpin-forming linear peptide. :l. 13C NMR relaxation analysis. JAm Chem Soc
1995,117:10864-10885. 26.
Munoz V, Thompson PA, Hofrichter J, Eaton WA: Folding dynamics and mechanism of ~-hairpin formati6n. Nature 1997,
390:196-199. The authors have determined, for the first time, the kinetics of folding of a ~-hairpin peptide by a temperature-jump method and compared the results with those obtained for a model helical system. Interestingly, they described
1997, 1:167-178.
••
451. Random sequences differing in amino acid type as well as number of residues were substituted for the native turn sequences in a series of mutants from the immunoglobulin G binding domain of protein G. Thermodynamic measurements showed that turn sequences selected under stringent conditions result in the most stable proteins. This suggests that the sequences of ~ turns in a variety of proteins are far from random, and are under evolutionary pressure to favor thermodynamically stable structures.
Formation and stability of p-hairpin structures in polypeptides Francisco Blanco*, Marina Ramirez-Alvarado t and Luis Serrano+ Experimental work on peptide models with ~-hairpin structures has provided new insights into the formation and stability of this secondary str~cture element. Both the turn region and the antiparallel strand residues not only affect the overall stability of the hairpin, but also determine the o/pe of hairpin formed. These results agree reasonably well with those from experimental and statistical analyses of ~-sheet structures in proteins.
<1>,'11 analysis of the protein database [7], although there is an important context effect caused due to the protein environment [5,7,8].
Figure 1
lie
(a) Addresses Biostructure and Biocomputing, European Molecular Biology Laboratory, Meyerhofstrasse 1, Heidelberg 069117, Germany ·e-mail: [email protected] te-mail: [email protected] *e-mail: [email protected] Correspondence: luis Serrano
NH3+ -Arg-Gly
H
0
Val
N
-{y T~\ ~
coo-- A,g·G~
H
-
•
I I I
Current Opinion in Structural Biology 1998, 8:107-111
Iffir
H
\ ,~
I I I
Tyr
H
'j""
Asn
N
0I
I I I
I I
H
I'XY.J.X),Y1G~
Jv~ I~
IH
N
vH
0
H
Lys
H
0
http://biomednet.com/elecref/0959440X00800107
© Current Biology ltd ISSN 0959-440X
(b)
Abbreviations BPTI bovine pancreatic trypsin inhibitor NOE nuclear Overhauser enhancement
Introduction The ~ hairpin is the simJ1lest form of-an antiparallel ~ sheet and is defined as a turn region flanked by two strands with a defined backbone hydrogen-bonding pattern (Figure 1). There are several types of hairpins depending of the type of turn. In proteins, 2:2 hairpins are the most common (Figure 1a), followed by 3:5 hairpins (Figure 1b), 4:4 hairpins (Figure Ie) and other less abundant conformations. The two number nomenclature was introduced by Sibanda et 01. [1] and it refers to the number of the residues in the turn (or not in strand conformation) \inder two different criteria. The first is the number of strand residues whose amide or carboxyl moiety is involved in the typical pattern of ~ sheets. The second is the number of strand residues with both the amide and carboxyl participating in the ~-sheet hydrogen-bonding pattern. Unlike formation of a helices, formation of ~ sheets is not yet well understood, because of its inherent complexity (distinct substructures must be considered: the turn; the strands and their relative interplay) and the lack of simple model systems akin to those used for studying a-helix formation. Proteins have been used to derive experimental scales of the propensity of ~-sheet formation for the 20 natural amino acids [2-5] as well as to study of the effect of different residue pairs in an antiparallel ~ sheet [6]. The intrinsic propensities correlate well among themselves and with a scale derived from
(c)
Current Opinion in Structural Biology
Schematic representation of ~-hairpin structures observed in isolated linear peptides in solution. (a) A 2:2 ~ hairpin with a type I' ~ turn. The sequence of the peptide was analyzed by Ramirez-Alvarado et a/. [15 aal. (b) A 3:5 ~ hairpin with a type I ~ turn and a ~ bulge. The sequence comes from a peptide analyzed by de Alba et a/. [20 aa]. (c) A 4:4 ~ hairpin with with a type I ~ turn of the same sequence as (b).
Model peptides should enable analysis of the sequence determinants of secondary structure formation in a context-
108
Folding and binding
free environment, as has been achieved with a helices. Most natural peptides encompassing hairpins, however, are mainly devoid of structure in water or form aggregates [9,10], with one exception-the 41-56 residue fragment of the protein G Bl domain [11). This is not surprising since secondary structure propensities are normally low in natural sequences [12) and the same forces that could lead to intramolecular ~-sheet formation can also drive intermolecular aggregation. Recently, the development of simple peptide models of ~-hairpin structures [13,14,15° has enabled useful information to be obtained on this otherwise elusive secondary structure element. This review focuses on the experimental work on these peptide models and discusses the prospects and challenges of hairpin analysis. 0
)
Peptide models of hairpin structures The first reported case of a ~-hairpin-forming linear peptide is a modified sequence of a 2:2 ~ hairpin from the a-amylase inhibitor tendamistat [13). The native sequence forms transiently populated turns at the central region but no ~-hairpin structure. Replacement of the turn residues by the sequence with the highest tendency to form type I ~ turns (Asn-Pro-Asp-Gly), results in the formation of a 3:5 non-native hairpin with a three-residue turn (type I turn plus a ~ bulge) (Figure Ib). This causes a shift of one residue in the antiparallel alignment of the strands, altering the pairing between the interstrand residues. Similar results were found with a 16-residue peptide derived from the first 2:2 ~ hairpin of ubiquitin, in which the same sequence was introduced in the turn region [14). The presence of these non-native hairpins could be due to the incompatibility of the geometry of a type I turn with the right-handed twist typical of ~ sheets in proteins (16). Type I' or II' turns are normally associated with 2:2 hairpins in proteins, with a geometry that matches the right-handed twist of the ~ sheet [1). This hypothesis is strongly supported by the de novo design of a linear peptide (BH8) based on an analysis of 2:2 hairpins in the protein database (Figure la) [15°°). In this peptide, the sequence Asn-Gly was selected for the central turn residues as it is the most abundant in 2:2 hairpins with type I' turns. The strand residues were selected based on apparently favored combinations of residues in the strands of protein hairpins, also matching criteria to avoid aggregation. This designed peptide folds into the expected 2:2 ~-hairpin structure with a population around 35% and is an ideal system by which to analyze the sequence determinants of 2:2 ~-hairpin formation. Other direct evidence of the turn sequence that favors type I' or II' ~ turns in 2:2 hairpin formation was found by replacing the L-Pro residue at the turn of the ubiquitin-derived peptide with D-Pro, with opposite chirality. In this case, a 2:2 hairpin was formed with type I' or II' turns, depending on the residue accompanying the Pro at position i+ 1 (I' for D-Pro-D-Ala, and II' for
D-Pro-Gly and D-Pro-Ala) [17°°]. Similar results were found in an analogous study of a different peptide sequence containing D-Pro-Gly at the turn region [18].
The turn region The importance of turn stability in hairpin formation is clearly shown by the analysis of a series of peptides based on the BH8 sequence. The peptides all contain the sequence Xxx-Gly in the turn region, with Xxx being either asparagine, aspartate, serine, alanine or glycine. All the peptides form 2:2 ~ hairpins but with different populations with respect to the random-coil conformations [19·°]. The order of hairpin stabilization is Asn:2: Asp> Gly > Ser:2: Ala. The relative positions of asparagine, serine and alanine within this hierarchy can be explained by their intrinsic residue preferences in populating the backbone dihedral angles corresponding to the first position of a type I' turn. Aspartate and glycine were found to be more and less stabilizing, respectively, than expected from their CI>,'P propensities. Exactly the same relative abundancies, however, were found at that position in 2:2 ~ hairpins in proteins (Figure 2). In the case of aspartate, there is a specific sidechain rotamer preference that creates a favorable electrostatic interaction with its own amide group when it is preceded by a residue in an extended conformation. The analysis of mutants of a designed sequence that adopts different populations of 3:5 and 4:4 ~ hairpins (Figure 1b,c), enabled de Alba et 01. [20··,21,22] to identify several factors that could selectively increase the type of ~ hairpin formed. The observed effects were again consistent with the statistical analysis of the protein database. An interaction between asparagine and threonine at the beginning of the turn region (Figure Ib) stabilized the 3:5 hairpin. A residue with a higher intrinsic preference in adopting backbone dihedral angles in the UR region of a Ramachandran plot than asparagine stabilized the 4:4 hairpin (Figure lc). The introduction of the native turn sequence of the ~ hairpin from the protein bovine pancreatic trypsin inhibitor (BPTI) leads to the formation of a single 4:4 hairpin, as in BPTI. A sequence that almost exclusively adopts 2:2 hairpins was obtained when the turn residues were replaced by the sequence Tyr-Asn-Gly-Lys.
The strands Mutagenesis analysis of the BH8 peptide has shown that interstrand sidechain-sidechain interactions contribute significantly to hairpin stability, although it is difficult to discriminate these interactions from the change in intrinsic secondary structure propensities [15°°]. Recently, the BH8 sequence was elongated by placing oppositely charged residues at the first and last positions of the hairpin structure, resulting in a significant increase in stability (M RamIrez-Alvarado, F] Blanco, L Serrano, unpublished data). This supports the hypothesis that interstrand sidechain-sidechain interactions are important in the formation of hairpin structure.
Formation and stability of
~·hairpin
Figure 2
rnc:
--
25
Y = -1~.499 + 0.87314x R= 0.98261
BHa
~
:::3
20
•
Q)
Co
~
-
.s::.
.~
BH9
15
(/)
'in .s::.
10
BH13
.~ (/)
Q)
109
compnsmg the hairpin structure are conformationally constrained and display correlated
c:
'2-
structures in polypeptides Blanco, Ramirez-Alvarado and Serrano
5
(/)
CO ()
BH12
• • BH11
0 15
20
25
30
35
40
% Hairpin population Current Opinion in Structural Biology
Correlation between the hairpin population of the corresponding peptides and relative residue frequencies at the first position of the ~ turn in protein ~ hairpins. The peptides are based on the sequence of the peptide described in Figure 1a. Mutations have been made at position L1 of the type I' turn of the ~ turn in BH8 [19 001; BH9 has aspartate at position L1; BH11 has alanine; BH12 has glycine; and BH13 has serine. The y-axis indicates the number of ~ hairpins with a type I' turn and a glycine at position L2, as found in the database of protein structures. The x-axis reflects the average ~-hairpin population estimated from NMR measurements.
Rico and co-workers [23 have explored the effect of the strand residues on the equilibrium between the 3:5 and 4:413-hairpin structures and the random-coil conformations in their peptide systems. Several modifications of the sequence of the N-terminal 13 strand showed that the effect on the stability of one hairpin structure is independent of the effect on the other and that 3:5 hairpins are less susceptible to mutations, suggesting they are more stable than 4:4 hairpi.ns. The higher overall preference for 3:5 hairpins was r~lated to increased efficiency of sidechain packing and burying of hydrophobic surface area when compared to 4:4 hairpins. 0
]
Kinetic analysis of the folding reaction of protein G residues 41-56 has been analyzed by the temperature jump method [26 The folding of this hairpin is slower than in the case of polyalanine based a helices (microseconds as opposed to nanoseconds) and theoretical analysis suggests that the distance between the favorable sidechain-sidechain interactions and the 13 turn could strongly influence the folding speed. The analysis of more peptides with different sequences and turns is needed before generalizing all hairpin and sheet structures. 00
The description of the ~·hairpin formation: population quantification While quantification of the structured population in helical peptides is reasonably straightforward, it is not the case for 13 hairpins. Far-UV circular dichroism spectra analysis is not the most appropriate method for estimating the hairpin population, in contrast to a-helical structures, because of the much smaller signal and strong influence of the l3-turn conformation [15 On the other hand, different NMR parameters can be used for estimating population at a residue level. The structural NOE intensities, the chemical shifts of the Cu protons and Ca proton-amide proton coupling constants can be used to estimate the hairpin population, assuming a simple two-state model, with hairpin and random-coil conformations in fast exchange [14,15 .,20 Each parameter has its own drawbacks, but combined they can yield an estimate of the hairpin population as accurate as the two-state assumption allows it to be [19. 00
0
0
Dynamic and kinetic properties of peptides
].
].
00
].
].
~·hairpin
13C-NMR relaxation data as well as data from extensive molecular dynamic simulations have been determined [24,25] for the modified Tendamistat peptide [13]. The combined data suggest that the peptide has significantly populated hydrogen bonds, and transient loose interactions occur in related but distinct conformations between hydrophobic and terminal charged groups. Recently, the BH8 peptide system was selectively 13C-enriched at different positions and 13C-NMR relaxation experiments were performed. It was found that those residues
Implications for protein stability In the case of a helices, it has been shown that by locally increasing helical propensities it is possible to significantly increase the resistance of proteins to thermal and chemical denaturation [27,28]. de Grado and co-workers [29 have shown that it is possible to increase the thermal stability of a protein by locally modifying the l3-turn region of a 13 hairpin. Recently, it was discovered that changing the Asn-Asp sequence of a 2:2 13 hairpin of an SH3 domain to the more favorable type I' l3-turn sequence, Asn-Gly, significantly increases the stability of the protein 00
]
110
Folding and binding
to chemical and thermal denaturation, as well as the folding rate (J MartInez, L Serrano, unpublished data). Once we acquire the level of knowledge that we have already for a helices, it should be possible to rationally modify protein stability in ~-sheet-containing proteins.
7.
Munoz v, Serrano L: Intrinsic secondary structure propensities of the amino acids, using statistical phi-psi matrices. Comparison with experimental scales. Proteins 1994, 20:301311.
8.
Otzen DE, Fersht AR: Side-chain determinants of stability. Biochemistry 1995, 34:5718-5724.
9.
Ramirez-Alvarado M, Serrano L, Blanco FJ: Conformational analysis of peptides corresponding to all the secondary structure elements of protein L 81 domain: Secondary structure propensities are not conserved in proteins with the same fold. Protein Sci 1997, 6:162-174.
10.
Viguera AR, Jimenez MA, Rico M, Serrano L: Conformational analysis of peptides corresponding to ~-hairpins and a ~ sheet, that represent the entire sequence of a-spectrin SH3domain. J Mol Bioi 1995, 255:507-521.
11.
Blanco FJ, Rivas G, Serrano L: A short linear peptide that folds into a native stable ~-hairpin in aqueous solution. Nat Struct Bioi 1994, 1:399-409.
12.
Munoz v, Cronet P, L6pez-Hernimdez E, Serrano L: Analysis of the effect of lOcal interactions in protein stability. Fold Des 1996, 1:167-1 78.
13.
Blanco FJ, Jimenez MA, Rico M, Santoro J, Herranz J, Nieto JL: Evidence of a short linear peptide that folds into native stable ~-hairpin in aqueous solution. JAm Chem Soc 1993, 115:58875888.
14.
Searle MS, Williams DH, Packman LC: A short linear peptide derived from the N-terminal sequence of ubiquitin folds into a water-stable non-native ~-hairpin. Nat Struct Bioi 1995, 2:9991006.
Conclusions There are already adequate peptide models of ~-hair pin structures that provide information about specific interactions at both the turn region and the strands. The overall stability of the hairpin structure is the result of many different contributions, including mainchain hydrogen bonding, secondary structure propensities and sidechain-sidechain interactions. Depending on the sequence, some of these factors are more important than others. In general, there is a good correlation between the experimental data and the statistical analysis of the protein database. As more experimental information becomes available, a description of hairpin formation in an algorithmic form will be more feasible and could be used to rationally stabilize ~-sheet-containing proteins. The description of hairpin formation will be more difficult than for a helices, because the number of different interactions that need to be evaluated is larger. There are different hairpin types that must be considered as alternative structures for a single sequence, and the population quantification methods need to be cross-checked. The development of theoretical approaches that allow exploration of the conformational space of different sequences in fixed frameworks within an adequate forcefield, in combination with the experimental analysis of model peptides, will enable us, in the future, to reach the same level of knowledge that we presently have for a helices.
Acknowledgements We thank Maria Angeles Jimenez and colleagues for sending us preprints of their work. Francisco Blanco is a fellow of the Spanish Ministerio de Educaci6n y Ciencia.
~-sheet
Ramirez-Alvarado M, Blanco FJ, Serrano L: De novo design and structural analysis of a model ~-hairpin peptide system. Nat Struct Bioi 1996, 3:604-612. This paper describes a peptide designed completely de novo that folds as a monomeric ~ sheet. Substitution of the strand residues by alanine results in abolishment of the hairpin structure, suggesting that sidechain-sidechain interactions contribute significantly to hairpin stability. 15.
16.
Chothia C: Conformation of twisted ~-pleated sheets in proteins. J Mol Bioi 1973, 75:295-302.
Haque TS, Gellman SH: Insights on ~-hairpin stability in aqueous solution from peptides with enforced type I' and type II' ~-turns. JAm Chem Soc 1997, 1197:2303-2304. The authors have done an interesting study incorporating D-stereoisomers residues at one or both turn positions of a 16-residue peptide that corresponds to the N-terminal segment of ubiquitin, in order to create "mirror image turns". Their results indicate that type I' and type II' ~ turns are required for the formation of a two-residue turn ~ hairpin. 17.
18.
Karle IL, Awasthi SK, Balaram P: A designed ~-hairpin peptide in crystals. Proc Natl Acad Sci USA 1996, 93:8189-8193.
Ramirez-Alvarado M, Blanco FJ, Niemann H, Serrano L: Role of ~-turn residues in ~-hairpin formation and stability in designed peptides. J Mol Bioi 1997, 273:898-912. Mutagenesis of the first central residue in the type I' ~ turn from a model peptide system that folds as a ~ hairpin shows the importance of the turn in hairpin stability and illustrates the idea that the presence of the turn does necessarily alter tHe pairing of the strands. Interestingly, there is a very good correlation between the effect of the different residues on hairpin stability and their relative abundance in the protein database. 19.
References and recommended reading Papers of particular interest, published within the annual period of review, have been highlighted as: • ••
of special interest of outstanding interest
1.
Sibanda BL, Blundell TL, Thornton JM: Conformation of ~ hairpins in protein structures. A systematic classification with applications to modelling by homology, electron density fitting and protein engineering. J Mol Bioi 1989, 229:759-777.
2.
Kim CA, Berg JM: Thermodynamic ~-sheet propensities measured using a zinc-finger host peptide. Nature 1993, 362:267-270.
3.
Minor DL, Kim PS: Measurement of the ~-sheet-forming propensities of amino acids. Nature 1994, 367:660-663.
4.
Smith CK, Withka JM, Regan L: A thermodynamic scale for the ~-sheet forming tendencies of the amino acids. Biochemistry 1995, 33 :5510-551 7.
20.
De Alba E, Jimenez MA, Rico M: Turn residue sequence determines ~-hairpin conformation in designed peptides. J Am Chem Soc 1997, 119:175-183. Different peptides were designed to fold into different ~-hairpin conformations in aqueous solution and were analyzed in order to further the understanding of the role of the turn sequence in defining ~-hairpin structure. 21.
De Alba E, Blanco FJ, Jimenez MA, Rico M, Nieto JL: Interactions responsible for the ~-hairpin conformation population formed by a designed linear peptide. Eur J Biochem 1995, 233:283292.
22.
De Alba E, Jimenez MA, Rico M, Nieto JL: Conformational investigation of designed short linear peptides able to fold into ~-hairpin structures in aqueous solution. Fold Des 1996, 1:122-144.
23.
De Alba E, Rico M, Jimenez MA: Cross-strand side chain interactions versus turn conformation in ~-hairpins. Protein Sci 1997, 6:2548-2560.
~-sheet
5.
Minor DL, Kim PS: Context is a major determinant of propensity. Nature 1994, 371 :264-267.
6.
Smith CK, Regan L: Guidelines for protein design. The energetics of ~-sheet side chain interactions. Science 1995, 270:980-982.
Formation and stability of [3-hairpin structures in polypeptides Blanco, Ramirez-Alvarado and Serrano
111
The relative stability of 3:5 and 4:4 hairpins are investigated by mutating certain residues at the strand regions.
a theoretical model that explains the kinetic results and could be used in the future to quantitatively analyze hairpin formation.
24.
27.
Munoz V, Cronet P, Lopez-Hernandez E, Serrano L: Analysis of the effect of local interactions in protein stability. Fold Des
28.
Virtudes V, Viguera AR, Aviles FX, Serrano L: Stabilisation of proteins by rational design of a-helix stability using helix/coil transition theory. Fold Des 1996, 1:29-34.
29.
Zhou HX, Hoess RH, deGrado WF: In vitro evolution of thermodynamically stable turns. Nat Struct 810/1996, 3:446-
Constantine KL, Mueller L, Andersen NH, Tong H, Wandler CF, Friederichs MS, Bruccoleri RE: Structural and dynamic properties of a [3-hairpin-forming linear peptide. 1. Modeling using ensemble-averaged constraints. JAm Chem Soc 1995,
117:10841-10854. 25.
Friederichs MS, Stouch TR, Bruccoleri RE, Mueller L, Constantine KL: Structural and dynamic properties of a [3-hairpin-forming linear peptide. :l. 13C NMR relaxation analysis. JAm Chem Soc
1995,117:10864-10885. 26.
Munoz V, Thompson PA, Hofrichter J, Eaton WA: Folding dynamics and mechanism of ~-hairpin formati6n. Nature 1997,
390:196-199. The authors have determined, for the first time, the kinetics of folding of a ~-hairpin peptide by a temperature-jump method and compared the results with those obtained for a model helical system. Interestingly, they described
1997, 1:167-178.
••
451. Random sequences differing in amino acid type as well as number of residues were substituted for the native turn sequences in a series of mutants from the immunoglobulin G binding domain of protein G. Thermodynamic measurements showed that turn sequences selected under stringent conditions result in the most stable proteins. This suggests that the sequences of ~ turns in a variety of proteins are far from random, and are under evolutionary pressure to favor thermodynamically stable structures.