Protein Science (1993), 2, 1574-1590. Cambridge University Press. Printed in the USA. Copyright 0 1993 The Protein Society Identification, classification, and analysis of beta-bulges in proteins A.W. EDITH CHAN,',2 E. GAIL HUTCHINSON,' DANIEL HARRIS,' AND JANET M. THORNTON' ' Biomolecular Structure and Modelling Unit, Department of Biochemistry and Molecular Biology, University College, Cower Street, London WClE 6BT, United Kingdom Italfarmaco-Research Center, via dei Lavoratori 54, Milan 22092, Italy (RECEIVEDFebruary 24, 1993; ACCEPTEDJuly 21, 1993) Abstract A @-bulgeis a region of irregularity in a &sheet involving two @-strands.It usually involves two or more residues in the bulged strand opposite to a single residue on the adjacent strand. These irregularities in 0-sheets were iden- tified and classified automatically, extending the definition of &bulges given by Richardson et al. (Richardson, J.S., Getzoff, E.D., &Richardson, D.C., 1978, Proc. Nud. Acud. Sci. USA 75,2574-2578). A set of 182 protein chains (170 proteins) was used, and a total of 362 bulges were extracted. Five types of @-bulgeswere found: clas- sic, G1, wide, bent, and special. Their characteristic amino acid preferences were found for most classes of bulges. Basically, bulges occur frequently in proteins; on average there are more than two bulges per protein. In general, @-bulgesproduce two main changes in the structureof a @-sheet:(1) disrupt the normal alternation of side-chain direction; (2) accentuate the twist of the sheet, altering the direction of the surrounding strands. Keywords: @-bulges;classification; proteins; protein structure Beta-pleated sheets, being one of the two major structural &Bulges, like @-turns, affect the directionality of 0- elements found in globular proteins, are formed from strands, but inmuch less drastic a manner. Thebulge is two or more @-strandsaligned side-by-side and hydrogen caused by the extraresidues on thebulged strand, which bonded. In each strand, the polypeptide chain is in an increase the backbone length, thus causing the strandto almost fully extended conformation, with typical 4, $J tor- bulge out of the plane ofthe sheet. At thesame time, the sion angles of - 122" and 143", respectively (Richardson twist of the sheet is slightly accentuated. et al., 1978). The hydrogen bonds between adjacent strands In this paper we present a systematic studyand classi- are formedbetween the main-chain CO and NH groups. fication of @-bulges and aim to explain how and why The regular arrangements of the strandsin @-sheetscan the bulges are formed. All bulges are identified automat- be distorted by "@-bulges,"which have been classified as ically using a computer program. Many moreX-ray crys- follows: classic,G1, wide, parallel, and some pseudobulges tal structures have become available since the original (Richardson et al., 1978; Richardson, 1981). This classi- classification was made. As a result, some new classes are fication was based largely on an examinationof protein introduced to add to the original bulge definitions. backbone drawings in the Atlas of Molecular Structureon Microfiche (Feldmann, 1976) and the hydrogen-bonding diagrams in published reports of solved X-ray structures. Analysis A @-bulgeis defined as theregion between two consecu- tive @-type hydrogen bonds,which include two residues Before defining @-bulges,let us describe P-sheets in more on one strand opposite a single residue on the other detail. Adjacent &strands can eitherin therun same direc- strand. tion (parallel @-sheet)or in opposite directions (antipar- allel @-sheet). Theirside chains extend above and below __~.~ .~ the sheet, with the C,-C, bond of each residue being ap- Reprint requests to: Janet M. Thornton, Biornolecular Structure and Modelling Unit, Biochemistry and Molecular Biology Department, Uni- proximately perpendicular to the planeof the sheet. The versity College, Cower Street, London WClE 6BT, United Kingdom. direction ofthese side chains alternates as onegoes along 1574 Beta-bulges: Classification and analysis 1575 a strand but is in registeron adjacent strands. The strands has been extended to any irregularity in the hydrogen- in a @-sheetare connected by hydrogen bonds. bonding pattern of a @-sheet,where the regular pattern Kabsch and Sander (1983) defined residues i and j, is disrupted by at most one extra residue on one strand which form the appropriatehydrogen bonds between ad- and at most four extra residues on the other strand.The jacent strands (or are covalently bonded to residues that irregularities were classifiedinto: classic, wide,bent, and form the required hydrogenbonds), and have termedthis special bulge types in parallel and antiparallel @-sheets. a bridge. There are two types of bridges: paralleland anti- However, even this extended definition does not include parallel. Their hydrogen-bondingpatterns are depicted in cases wherea bulge occurs outsidethe definitionof &sheet. Figure 1. Nevertheless, we have chosen to restrict the definition to In antiparallel @-sheet,the hydrogen-bonding pattern cases within the P-sheets only. between a given pair of strands has alternately wide and The bulges werefirst classified by the number of extra narrow spacing. This is becausethe hydrogen bonds from residues in each strand. Then we determined whether they a given strand alternate between the two strands either belonged to antiparallel or parallel @-sheets.Finally, the side of it as one moves from one residue position to the hydrogen-bonding pattern in the bulge residues was com- next. Thus, for example, all the even-numbered residues pared with the patterns illustrated in Figures 2-4. In ad- will havehydrogen bonds from their CO and NH groups dition, for classic bulges;the +, $ angles of residue I must to one adjacent strand, while the odd-numbered residues fall into an aR conformation. will bond in the opposite direction to the other adjacent The following examples illustratethe classification pro- strand. In parallel &sheet, on the other hand, the hydro- cess for bulges inantiparallel P-sheets. If there is one extra gen bonds between the bridges are evenly spaced, going residue on one strandbetween the two (perfect) bridges, across the sheet at an angle to the strands. but none on the adjacent one, then the bulge can be de- For each protein, the main-chain hydrogen-bonding fined as classic or wide by the specific hydrogen-bonding strengths and torsion angles were calculated using a pro- patterns illustrated in Figure 2. If there is one extra resi- gram called SSTRUC (D.K. Smith, unpubl.). The pat- due on each strand (at the same location), then it isa bent terns of hydrogen bonds were used to assign secondary bulge. Special bulgesare defined to have between twoand structures using a modified implementationof the Kabsch four extra residues on one strand. and Sander (1983) algorithm, whereby a residue that lies Because residue 1 in a G1 bulge doesnot form part of a at either end of a secondary structure is included if it has bridge, a different algorithm has been developedto search one “correct” hydrogen bond. In a &bulge, the two resi- for the G1 bulges. First, we have excluded allthe classic dues on the bulged side are labeled “1” and “2”, whereas bulges becauseboth G1 and classic bulgeshave identical the residue on the opposite strand is labeled“X” (Richard- hydrogen-bond patterns. Then we searched for all exam- son et al., 1978). Here, the original definition of &bulge ples that match the hydrogen patterns illustrated for G1G t Antiparallel Parallel Fig. 1. Description of antiparallel and parallel &sheet, with illustration of Kabsch and Sander’s (1983) definition for bridge, 1576 A.W.E. Chan et al. +"TI+ $- c+(150) I ' C- (15) I -+ ibi- ++++ - -ST+ + 1 A-bent (6) I W (30) I I ' G1G (50) t G1T (33) Fig. 2. Hydrogen-bonding diagrams for @-bulgesin antiparallel @-sheets.The name of each subclass is listed underneath each figure. The number in parentheses denotes the total number of bulges found in that class. Arrows indicate an NH to CO hy- drogen bond. Rectangles represent residues in &strands, and ellipsoids correspond to any residue not in a @-strand.The con- formation of each residue is indicated inside each box. Residues involved in the bulge are labeled as X, 1, and 2. The "+" and 'I-" signs indicate the orientations of the side chains into and out of the paper. G is glycine. Beta-bulges: Classification and analysis 1577 PC (18) P-bent (4) Fig. 3. Hydrogen-bond diagrams for @-bulgesin parallel @-sheets. and G1T in Figure 2, with the additional condition that an average of greater than two bulges per protein in the residues X and 2 must form partof a bridge, but allowing data set. Most examples involveantiparallel strands, with residue 1 to have any conformation. fewer than 10% of 0-bulges being betweenparallel strands The hydrogen-bonding patterns were visualized sche- even though the ratio of antiparallel to parallel @-sheets matically using diagrams produced by the HERA pro- is 4.5 to 1 in our dataset. Hence, &bulges are more com- gram (Hutchinson & Thornton, 1990), and the proteins mon in antiparallel sheets. All classes exceptthe G1 were themselves wereexamined visually usingthe QUANTA"" found in both parallel and antiparallel &sheets, whereas computer graphics program. the G1 class was found only at the edges of antiparallel &sheets. Results and discussion Almost all bulges occur at the edge of a sheet, with the bulged strand being the outermost one. The distortion Classification of @-bulges produced by the bulge is most pronounced in the special, Table 1 lists the numbers of &bulges in each of the five bent, and G1 bulges, and least pronounced in the classic main classes and their subclasses. Our data set included and wide bulges. These distortions are demonstrated in 362 bulges from 170 proteins (Table 2). This represents Figure 5 and kinemages. The common structural effect of 1578 A. W.E. Chan et al. I I + & A+ + - + + + + - + + s3 s4 4 2+ ' SPW I SP3 Fig.