Protein Engineering, Vol. 14, No. 8, 543-547,
August 2001
© 2001 Oxford University Press
Stabilization of local structures by 
–CH and aromatic–backbone amide interactions involving prolyl and aromatic residues
Department of Biomedical Sciences, School of Medicine, Creighton University, 2500 California Plaza, Omaha, NE 68178, USA
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Weakly polar interactions between the side-chain aromatic rings and hydrogens of backbone amides (Ar–HN) and CHn of aliphatic groups (
–CH) are known to form local structures and to stabilize secondary structure in peptides and proteins. To investigate the existence of these interactions and to explore their possible role in constraining the structures of Pro–Xaa and Xaa–Pro fragments in proteins, a database search was performed in a non-redundant set of proteins from the Brookheaven Protein Data Bank for –CH and Ar–HN interactions in Pro–Xaa and Xaa–Pro fragments (where Xaa is either Phe, Tyr or Trp). In Xaa–Pro fragments, the percentage of –CH interactions and Ar–HN interactions, respectively, was 20.6 and 3.2%, in Pro–Xaa fragments 26.8, 8.6 and 4.0% of the Pro–Xaa fragments contained both interactions, while no Xaa–Pro fragments had both. The protein fragments containing Ar–HN and/or –CH interactions were clustered on the basis of similarity of selected torsion angles. The clustering resulted in well defined clusters. Thus, –CH and Ar(i)–HN(i) interactions were able to constrain individual conformations of the Pro–Xaa and Xaa–Pro fragments. These local structures were found to be independent of the secondary structure of the polypeptide chains in which the fragments were found.
Keywords: aromatic residue/proline/secondary structure/weakly polar interactions
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Interactions between the aromatic rings of amino acids and either CHn of aliphatic groups of residues (
–CH interaction) (Nishio et al., 1995
) or the backbone amide (Ar–HN interaction) (Burley and Petsko, 1986; Levitt and Perutz, 1988) can have a role in the stabilization of protein structure and of protein–protein and protein–ligand interactions. The major source of the attraction energy of –CH interaction is due to the dispersion interaction, while the relatively small but highly orientation-dependent electrostatic interaction controls the orientation of the C–H bond to the aromatic ring (Tsuzuki et al., 2000a). The Ar–HN interaction is similar in nature to the –CH interaction, although the electrostatic interaction is more significant in the Ar–HN interaction (Tsuzuki et al., 2000b). Examples of these interactions in peptide and protein structures include the restriction of the conformation of serine protease inhibitors (Shimohigashi et al., 1999) by side chain–side chain –CH interactions, the stabilization of the N-terminal side of the central parallel ß-sheet of cutinase (Prompers et al., 1997) by Ar–HN interactions and the stabilization of intermediates in the folding of BPTI (Kemmink and Creighton, 1993).
A previous protein database search for Ar–HN interactions found that the propensity of the aromatic rings to interact with the backbone amide in the same residue [Ar(i)–HN(i) interactions] in Pro–Xaa protein fragments is three times higher than expected (Tóth et al., 2001). It was therefore proposed that –CH interactions between aromatic rings and prolyl side chains are responsible for the high incidence of the Ar(i)–HN(i) interaction. Others have also reported that prolyl–aromatic interactions lead to local structures in peptide fragments (Nardi et al., 1997, 2000; Pal and Chakrabarti, 1999). This investigation was undertaken to establish the basis for this phenomenon.
The incidence of the cis peptide bond in Yaa–Pro fragments, where Yaa can be any amino acid, in proteins is the highest if Yaa is an aromatic amino acid (Stewart et al., 1990; Pal and Chakrabarti, 1999). Investigations of model peptides by 1H NMR spectroscopy found that in Xaa–(cis)Pro fragments the aromatic ring current causes a decrease in the value of the chemical shift of the Pro side chain hydrogens due to the stacking of the aromatic ring on the side chain of Pro (Yao et al., 1994; Wu et al., 1998). Pal and Chakrabarti proposed that the attractive force between the aromatic ring and the Pro side-chain hydrogens, the –CH interaction, could be responsible for such stacking and for stabilizing the cis peptide bond (Pal and Chakrabarti, 1999).
About 10% of the Pro–Yaa fragments in proteins are (cis)Pro peptide bonds and more than half of the (cis)Pro–Xaa fragments have conformations in which the aromatic ring is close to the prolyl side chain (Nardi et al., 2000). Such contacts were identified (Nardi et al., 1997) in the model peptide Ala–Pro–Tyr using 1H NMR and molecular dynamics (Nardi et al., 2000). Among the folded conformations of this peptide, the (cis)Pro form had interactions between the aromatic ring and the prolyl side chain.
To investigate the effect of the –CH and Ar(i)–HN(i) interactions on the formation of local structures in Pro–Xaa and Xaa–Pro protein fragments, a protein database was searched. The conformation of Pro–Xaa and Xaa–Pro protein fragments with –CH and/or Ar(i)–HN(i) interactions were investigated by cluster analyses and by determining the secondary structure of the polypeptide chains in which the fragments were located.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
When the aromatic ring is close to a proton then the delocalized electrons of the aromatic ring change the local magnetic field of the proton. In NMR measurements this phenomenon causes an anomalous shift referred to as the ring shift (
ring). The ring value is influenced by the interaction geometry of the Ar–HN and –CH interaction. To identify both the Ar–HN and the -CH interactions, the ring of the amide hydrogens and the hydrogens of C–Hn groups, respectively, was determined. Fragments with amide hydrogen ring of –0.5 p.p.m. or lower (Worth and Wade, 1995) were selected to contain Ar(i)–HN(i) interaction, while fragments with proline side chain hydrogen ring of –0.25 p.p.m. or lower (Shimohigashi et al., 1999; Nardi et al., 2000) were selected to contain CH– interactions.
A database of 560 coordinate files of proteins from the Protein Data Bank (Bernstein et al., 1977) with <25% sequence similarity (Hobohm et al., 1992) and a resolution of 3 Å or better (lists of the redundant proteins were downloaded from the EMBL file server: ftp.embl-heidelberg.de) was generated using SYBYL 6.2 (Tripos, St. Louis, MO). The database was searched for Xaa–Pro and Pro–Xaa fragments, where Xaa was either Phe, Tyr or Trp. SYBYL script #1 was used to execute the program TOTAL (Williamson and Asakura, 1993) to calculate ring.
SYBYL script #2 was used to measure selected torsion angles of the fragments. The DSSP program (Kabsch and Sander, 1983) was used to determine the secondary structure of the fragments. Multiple copies of particular Ar–HN interactions, due to structural analogs in the same protein, were ruled out on the basis of similarities in amino acid sequence and secondary structure.
Selected Pro–Xaa and Xaa–Pro fragments were grouped on the basis of the type of interactions found therein. Then, they were clustered using the Partitioning Around Medoids (PAM) method (Kaufman and Rousseeuw, 1990). This method finds groups of related conformations based on their pairwise dissimilarities. Dissimilarities between conformations were constructed by calculating the torsion angle root mean square (r.m.s.) distance (Becker, 1997) using the equation
 | (1) |
where N is the number of torsion angles and 
k(i) and k(j) are the values of the torsion angle k in structures (i) and (j), respectively. The values of 
Xaa1 and 
Xaa torsion angles were used to calculate the dij for the group with Ar(i)–HN(i) interactions in Xaa–Pro fragments, and Xaa1 and 
Xaa for Yaa-Pro–Xaa fragments, where Yaa can be any amino acid. The values of 
Yaa–Pro, Pro, Xaa and Xaa1 torsion angles were used to calculate the dij for the group with –CH interactions in Pro–Xaa fragments, and Xaa1, Xaa and Xaa–Pro for Xaa–Pro fragments. The dissimilarity matrix was constructed from the torsion angles by a Perl script and was used as input file for the clustering program PAM. Clustering was performed using a minimum of two and maximum of 10 clusters.
The number of clusters into which the conformations can be divided was selected on the basis of the silhouette width function, s(i). The value of the silhouette width of an individual conformation can be between –1 and 1. It is defined by the average dissimilarity, a(i), between conformation i and all other conformations within cluster A and the average dissimilarity, b(i), between conformation i and all other conformations within the neighboring cluster B.
 | (2) |
 | (3) |
 | (4) |
The number of clusters representing the optimal clustering of the system is based on the highest average silhouette width of all clusters. The highest average silhouette width of all clusters is defined as the silhouette coefficient of the system. Silhouette width values between 0.71 and 1.00 represent a strong cluster, between 0.51 and 0.70 a reasonable cluster, between 0.26 and 0.50 a weak cluster that could be artificial and < 0.26 no cluster (Watts et al., 2001).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Some 6.3% of the peptide bonds between Yaa–Pro in Yaa–Pro–Xaa fragments were in a cis orientation, whereas 7.4% of the Xaa–Pro peptide bonds in Xaa–Pro fragments were in a cis orientation. The occurrence of
–CH and Ar–HN interactions in the Xaa–Pro and Pro–Xaa fragments is summarized in Table I. Some 50% of the (cis)Pro–Ar fragments and 25.3% of the (trans)Pro–Ar were involved in CH– interactions. 68.3% of the Xaa–(cis)Pro fragments had CH– interactions, while 16.8% of the Ar–(trans)Pro fragments had CH– interactions. Ar(i)–HN(i) interactions were found in both protein fragments, but they existed simultaneously with CH– interactions only in Pro–Ar fragments.
|
Table IIA
shows selected averaged torsion angles of clusters of Pro–Xaa protein fragments. The clustering of Pro–Xaa fragments with –CH interactions resulted in weaker clusters compared with Pro–Xaa fragments with Ar(i)–HN(i) and/or –CH interactions on the basis of their silhouette coefficients. Pro–Xaa fragments with –CH interactions resulted in one strong, five reasonable and two weak clusters, Pro–Xaa fragments with Ar–HN interactions resulted in one strong and three reasonable clusters and Pro–Xaa fragments with –CH and Ar(i)–HN(i) interactions resulted in one strong, three reasonable and one weak clusters (Table IIA).
|
Table IIB
shows selected averaged torsion angles of the clusters from Xaa–Pro fragments. In Xaa–Pro fragments, no structures were found with both –CH and Ar(i)–HN(i) interactions. The silhouette coefficients from the clustering of Xaa–Pro fragments with either –CH or Ar(i)–HN(i) interactions were high, 0.80 and 0.76, respectively. Strong and reasonable clusters were identified here (Table IIB).
The secondary structure of the polypeptide chains in which the protein fragments were found are summarized in Table III. Among (cis)Pro–Xaa fragments, 75% of the structures in the PAa2 cluster were in turns and bends and no fragment was in random coil, whereas 50% of the structures in the PAa7 cluster found in turns and bends and 38% were in random coil. The averaged torsion angles in PAa7 matched the backbone dihedrals of the right-handed polyproline I helices (,) = (–75,160°) (MacArthur and Thornton, 1991). Some 61% of the structures in cluster PAa5 were in helical conformations, while the rest were in turns and bends. More than half of the structures in clusters PAa4 and PAa8 were in random coil. The averaged torsion angles from the latter two clusters were similar to the torsion angle space for left-handed polyproline II helices (,) = (–75,145°) (MacArthur and Thornton, 1991). Ar(i)–HN(i) interactions in Pro–Xaa fragments were most often found in turns, bends and random coil and they were rarely found in ß-structures. Pro–Xaa fragments containing both Ar(i)–HN(i) and –CH interactions were found mostly in turns, bends and helices. The only exception was cluster PAab2, in which 80% of the structures were in random coil. Averaged torsion angles from PAab2 were similar to the backbone dihedral angles of polyproline I helix.
|
Among Xaa–Pro fragments containing
–CH interactions, 38% of those in cluster APa1 were in random coil and over 70% of those in clusters APa3 and APa4 were in a helical conformation; 95% of the fragments in cluster APa2 were in either turn or bend structures. Among Xaa–Pro fragments with Ar(i)–HN(i) interactions, fragments in cluster APb1 were found only in bends. |
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
The increase in the frequency of Ar(i)–HN(i) interactions in Pro–Xaa fragments (Table I
) can be attributed to the synergistic effect of both the Ar(i)–HN(i) and –CH interactions in constraining the conformation in 4% of these fragments. Conversely, in Ar–Pro fragments, the aromatic side chain has to be in different conformations to form either an Ar(i)–HN(i) or a –CH interaction. These two interactions could also be regarded as competing forces.
The high number and the low average silhouette coefficient (0.46) of PAa1–8 clusters suggest that the conformational space sampled by Pro–Xaa fragments with –CH interactions is large. Conversely, Xaa–Pro protein fragments with –CH interactions had fewer clusters than had Pro–Xaa fragments and all resulting clusters were strong. This suggests that the conformation of Xaa–Pro fragments with –CH interactions was more limited than the conformation of Pro–Xaa fragments with –CH interactions. –CH and/or Ar(i)–HN(i) interactions together with the pyrrolidine ring in Pro (MacArthur and Thornton, 1991) could have caused constraints in the conformational space of Pro–Xaa and Xaa–Pro protein fragments. In most clusters, fragments usually favored a particular secondary structure, although fragments in one cluster were also found in either in helices, turns or random structures. This suggests that the averaged dihedral angles and standard deviations describing the conformation of the clusters are independent of secondary structure (Figure 1). The only exception to this was cluster APa2, which was found to be in bends and turns 95% of the time. Furthermore, protein fragments in cluster APa2 contained only cis Xaa–Pro peptide bonds.
|
-helix (from Escherichia coli RNA polymerase; PDB access code 1sig). (C) Fragment Pro51–Trp52 in an extended type of structure (from NADH oxidase; PDB access code 1nox). (D) Fragments represented in (B) and (C), superimposed on the fragment represented in (A). Ribbons in (A), (B) and (C) represent the turn, helix and random meander conformations, respectively.Several studies (Montelione et al., 1984
; Oka et al., 1984; Yao et al., 1994; Reimer et al. 1998; Kang et al., 1999; Nardi et al., 2000) of the conformation of the ...–Xaa–Pro–... and ...–Yaa–Pro–Xaa–... peptides have led to reports mainly focusing on the conformation of the (cis)Pro peptides. The data therein, however, also provide evidence for Ar(i)–HN(i) interactions in (trans)Pro–Xaa fragments. For example, Yao et al. studied the effect of amino acid replacement at positions 2, 4 and 5 in the peptide Ser–Tyr–Pro–Tyr–Asp–Val on the stability of type VI turn using 1H NMR (Yao et al., 1994). When the fourth residue was Phe, Tyr and Trp, the chemical shift of the amide hydrogen of that residue was found to be less by 0.48, 0.66 and 1.03 ppm, respectively, in Ser–Tyr(trans)Pro–Xaa–Asp–Val than in Ser–Tyr(cis)Pro–Xaa–Asp–Val, indicating an Ar(i)–HN(i) interaction. In another study, the conformation of Ala(cis)Pro–Tyr was explored in detail (Nardi et al., 2000), but the conformation of the Ala(trans)Pro–Tyr was not explored. The data presented also included evidence for the existence of –CH and Ar(i)–HN(i) interactions in Ala(trans)Pro–Tyr as follows: (1) the chemical shifts of the Tyr amide hydrogen and the Pro side chain Hß3 and H
2 are 0.30, 0.32 and 0.64 ppm lower, respectively, in Ala(trans)Pro–Tyr than in the Ala(cis)Pro–Tyr; (2) ROESY cross peaks exist between the and 
protons of the Tyr side chain and the ß protons of Pro, suggesting CH– interactions, and between the protons of the Tyr side chain and the amide hydrogen of the same residue, suggesting Ar(i)–HN(i) interactions. These results further support the conclusion of this study, that –CH interactions can play a significant role in stabilizing local structures of polypeptides and that peptides with Xaa–Pro or Pro–Xaa fragments can have conformations similar to the conformations identified by the cluster analyses. |
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Well defined conformations were identified by clustering Pro–Xaa and Ar–Xaa fragments containing
–CH and Ar(i)–HN(i) interactions. It is concluded that –CH and Ar(i)–HN(i) interactions constrain the structure of these fragments in individual conformations. These conformations were identified as local structures of the Pro–Xaa and Xaa–Pro fragments, which were independent of the secondary structures of the polypeptide chains in which the fragments were found. An exception was the Xaa–(cis)Pro fragments, which were found only in turns and bends.
|
Notes |
|---|
1 To whom correspondence should be addressed. E-mail: vasz{at}bif1.creighton.edu
|
Acknowledgments |
|---|
This work was supported a grant from the National Science Foundation (EPS-9720643) and the Carpenter Chair in Biochemistry, Creighton University.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Becker,O.M. (1997) Proteins, 27, 213–226.[Medline]
Bernstein,F.C., Koetzle,T.F., Williams,G.J., Meyer,E.E.,Jr, Brice,M.D., Rodgers,J.R., Kennard,O., Shimanouchi,T. and Tasumi,M. (1977) J. Mol. Biol., 112, 535–542.[Web of Science][Medline]
Burley,S.K. and Petsko,G.A. (1986) FEBS Lett., 203, 139–143.[Web of Science][Medline]
Hobohm,U., Scharf,U., Schneider,R. and Sander,C. (1992) Protein Sci., 1, 409–417.[Web of Science][Medline]
Kabsch,W. and Sander,C. (1983) Biopolymers, 22, 2577–2637.[Web of Science][Medline]
Kang,Y.K., Jhon,J.S. and Han,S.J. (1999) J. Pept. Res., 53, 30–40.[Medline]
Kaufman,L. and Rousseeuw,P.J. (1990) Finding Groups in Data: an Introduction to Cluster Analysis. Wiley, New York.
Kemmink,J. and Creighton,T.E. (1993) J. Mol. Biol., 234, 861–878.[Web of Science][Medline]
Levitt,M. and Perutz,M.F. (1988) J. Mol. Biol., 201, 751–754.[Web of Science][Medline]
MacArthur,M.W. and Thornton,J.M. (1991) J. Mol. Biol., 218, 397–412.[Web of Science][Medline]
Montelione,G.T., Arnold,E., Meinwald,Y.C., Stimson,E.R., Denton,J.B., Huang,S.G., Clardy,J. and Scheraga,H.A. (1984) J. Am. Chem. Soc., 106, 7946–7958.
Nardi,F., Worth,G.A. and Wade,R.C. (1997) Fold. Des., 2, 62–68.
Nardi,F., Kemmink,J., Sattler,M. and Wade,R.C. (2000) J. Biomol. NMR, 17, 63–77.[Medline]
Nishio,M., Umezawa,Y., Hirota,M. and Takeuchi,Y. (1995) Tetrahedron, 51, 8665–8701.[Web of Science]
Oka,M., Montelione,G.T. and Scheraga,H.A. (1984) J. Am. Chem. Soc., 106, 7959–7969.
Pal,D. and Chakrabarati,P. (1999) J. Mol. Biol., 294, 271–288.[Web of Science][Medline]
Prompers,J.J., Groenewegen,A., van Schaik,R.C., Pepermans,H.A.M. and Hilbers,C.V. (1997) Protein Sci., 6, 2375–2384.[Web of Science][Medline]
Reimer,U., Scherer,G., Drewello,M., Kruber,S., Schutkowski,M. and Fischer,G. (1998) J. Mol. Biol., 279, 449–460.[Web of Science][Medline]
Shimohigashi,Y., Nose,T., Yamauchi,Y. and Maeda,I. (1999) Biopolymers, 51, 9–17.[Medline]
Stewart,D.E., Sarkar,A. and Wampler,J.E. (1990) J. Mol. Biol., 214, 253–260.[Web of Science][Medline]
Tóth,G., Watts,C.R., Murphy,R.F. and Lovas,S. (2001) Proteins: Struct. Funct. Genet., 43, 373–381.[Web of Science][Medline]
Tsuzuki,S., Honda,K., Uchimaru,T., Mikami,M. and Tanabe,K. (2000a) J. Am. Chem. Soc., 122, 3746–3753.
Tsuzuki,S., Honda,K., Uchimaru,T., Mikami,M. and Tanabe,K. (2000b) J. Am. Chem. Soc., 122, 11450–11458.
Watts,C.R., Tóth,G., Murphy,R.F. and Lovas,S. (2001) J. Mol. Struct.535, 171–182.
Williamson,P.W. and Asakura,T. (1993) J. Magn. Reson., Ser. B, 101, 67–71.
Worth,G.A. and Wade,R.C. (1995) J. Phys. Chem., 99, 17473–17482.
Yao,J., Feher,V.A., Espejo,B.F., Reymond,M.T., Wright,P.E. and Dyson,H.J. (1994)J. Mol. Biol., 243, 736–753.[Medline]
Wu,W.-J. and Raleigh,D.P. (1998) Biopolymers, 45, 381–394.[Medline]
Received November 15, 2000; revised March 7, 2001; accepted May 11, 2001.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  K. Huang, S. J. Chan, Q.-x. Hua, Y.-C. Chu, R.-y. Wang, B. Klaproth, W. Jia, J. Whittaker, P. De Meyts, S. H. Nakagawa, et al. The A-chain of Insulin Contacts the Insert Domain of the Insulin Receptor: PHOTO-CROSS-LINKING AND MUTAGENESIS OF A DIABETES-RELATED CREVICE J. Biol. Chem., November 30, 2007; 282(48): 35337 - 35349. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. C. Pedersen, J. Dong, F. Taniguchi, H. Kitagawa, J. M. Krahn, L. G. Pedersen, K. Sugahara, and M. Negishi Crystal Structure of an alpha 1,4-N-Acetylhexosaminyltransferase (EXTL2), a Member of the Exostosin Gene Family Involved in Heparan Sulfate Biosynthesis J. Biol. Chem., April 11, 2003; 278(16): 14420 - 14428. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||