`
`JMB
`
`
`
`Cis Peptide Bonds in Proteins: Residues Involved,
`their Conformations, Interactions and Locations
`
`Debnath Pal and Pinak Chakrabarti*
`
`Department of Biochemistry,
`P-1/12 CIT Scheme VIIM
`Bose Institute, Calcutta
`700 054, India
`
`An analysis of a non-redundant set of protein structures from the Brook-
`haven Protein Data Bank has been carried out to find out the residue pre-
`ference,
`local conformation, hydrogen bonding and other stabilizing
`interactions involving cis peptide bonds. This has led to a’ reclassification
`of turns mediated by cis peptides, and their average geometrical par-
`ameters have been evaluated. The interdependence of the side and main-
`chain torsion angles of proline rings provided an explanation why such
`rings in cis peptides are found to have the DOWN puckering. A compari-
`son of cis peptides containing proline and non-proline residues show
`differences in conformation, location in the secondary structure and in
`relation to the centre of the molecule, and relative accessibilities of resi-
`dues. Relevance of the results in mutation studies and the cis-trans
`
`isomerization during protein folding is discussed.
`
`© 1999 Academic Press
`
`‘Corresponding author
`
`Keywords: cis peptide; pyrrolidine ring puckering; turn conformation;
`residue preference in cis peptides; weak interactions
`
`Introduction
`
`In proteins, the partial double bond character of
`the peptide bond results in two conformations
`depending on the value of the dihedral angle,
`on
`[CD,(1)-C(1)-N(1’)-C,,,(1’)]:
`cis and trans
`(with
`co = 0
`and 180°,
`respectively)
`(Pauling,
`1960;
`Ramachandran & Sasisekharan, 1968) (Figure 1(a)).
`The isomer with the two C“ atoms trans to each
`
`other is favoured overwhelmingly due to the lesser
`steric conflict involving the substituents at these
`positions, and only when a Pro residue is in posi-
`tion (1') is there a substantial steric clash involving
`the C“ atom at position (1) and C5 atom of Pro at
`position (1'), even in the trans conformation,
`to
`give the cis imide bond, X-Pro, a higher frequency
`of occurrence than what is observed for the amide
`
`bond, X-Xnp.
`approximately
`of
`energy
`in
`A difference
`2.5 kcal/mol between the trans and the cis isomers
`(corresponding to only 1.5% occurrence of the cis
`form), regardless of the solvent, and a rotational
`barrier of about 20 kcal/mol have been found for
`the peptide bond analog N-methylacetamide
`(LaPlanche 8: Rogers, 1964; Christensen et al., 1970;
`
`Abbreviations used: X, any amino acid residue; Xnp,
`any non-Pro amino acid residue; Ar, any aromatic
`residue.
`
`E-mail address of the corresponding author:
`pinak@boseinst.ernet.in
`
`0022-2836 /99 /460271-18 $30.00 /0
`
`1971; Perricaudet &
`Drakenberg & Forsén,
`Pullman, 1973; Radzicka et al., 1988; Iorgensen &
`Gao, 1988; Schnur et czl., 1989; Scherer et LZl., 1998).
`For an imide bond in Pro-containing peptides,
`however, the trans isomer is favoured over the cis
`by only 0.5 kcal/mol (Maigret et al., 1970), so that
`a higher abundance (10-30 %) of the cis form is
`observed (Brandts
`et
`al.,
`1975; Grathwohl &
`Wiithrich, 1976; Iuy et
`£11., 1983);
`the activation
`energy barrier for cis-trans isomerization is also
`less, 13 kcal/mol (Schulz & Schirmer, 1984). Using
`conformational energy calculations, Ramachandran
`& Mitra (1976) found expected frequencies for the
`cis isomer to be 0.1 % and 30% (corresponding to
`an enthalpy difference of 4.0 and 0.5 kcal/mol,
`respectively) for an Ala-Ala and Ala-Pro peptide
`bond, respectively. A survey of protein structures
`by Stewart et al. (1990) found only 0.05% of all X-
`Xnp, but 6.5 % of all X-Pro peptide bonds to occur
`in the cis conformation. The analysis of MacArthur
`& Thornton (1991) provided a value of 5.7% for
`the latter group, whereas a recent work (Weiss
`et al., 1998; ]abs et al., 1999) gave values of 0.03%
`and 5.2%, respectively, for the two types of pep-
`tide bonds.
`
`Due to the energy barrier, cz's—trans isomerization
`of peptide bond is a rather slow process at room
`temperature and has been shown to play an
`important role in protein folding (Brandts et al.,
`1975; Creighton, 1978; Schmid & Baldwin, 1978;
`Cook et al., 1979; Lin 8: Brandts, 1984; Brandts &
`
`WOCK- EXHIBIT 1026
`© 1999 Academic Press
`
`This material may be protected by Copyright law (Title 17 U.S. Code)
`
`
`
`272
`
`Cis Peptide Bonds
`
`cis peptide mediated turns and an evaluation of
`the torsion angles of the involved residues.
`We have recently investigated the interrelation-
`ship between the side-chain and the main-chain
`conformational angles in residues involved in the
`trans peptide units (Chakrabarti & Pal, 1998), and
`from this perspective it is worthwhile to study the
`relationship in cis peptide bonds. The pyrrolidine
`ring of Pro can be associated with two types of
`puckering, designated UP and DOWN, depending
`on the ring torsion angles (Ramachandran et al.,
`1970; Ashida & Kakudo, 1974). It has been noted
`by Milner-White et al. (1992) that the puckering of
`the ring when it is involved in the cis linkage is
`DOWN, and an explanation may be sought in
`terms of the interaction between the main-chain
`and side-chain atoms.
`The cis peptide bonds, especially the ones with
`non-Pro residues, are located near the active sites
`or are implicated to have roles in the function of
`the protein molecule (Herzberg & Moult, 1991;
`Stoddard & Pietrokovski, 1998; Jabs et al., 1999).
`Though important, some of the cis peptide bonds
`might have gone unreported in the structures
`determined at lower resolution (Weiss et al., 1998).
`To facilitate the identification of such overlooked
`cis peptide bonds it is important to characterize the
`location of known cis peptide units (both X-Pro
`and X-Xnp)
`in the three-dimensional structures
`and their solvent accessibility. A comprehensive
`analysis of these issues is made here, so as to
`understand the interactions that stabilize a cis pep-
`tide bond and possibly identify regions/sequences
`in protein structures that are likely to adopt a cis
`peptide linkage.
`
`Results and Discussion
`
`Residues forming the cis peptide linkage
`
`A total of 50 % (147 out of 294) of well-defined
`protein structures contain one or more cis peptide
`bonds; 0.3 % of all the bonds in the database exist
`in the cis form (231 in total). Most of them (87 %)
`are preceding Pro residues (5.7 % of X-Pro bonds
`have the cis conformation). The intrinsic prob-
`ability of a residue (X) to cause a cis conformation
`of the X-Pro linkage, given by the fraction of occur-
`rence of the bond in the cis form, is provided in
`Figure 2. Stewart
`et al.
`(1990)
`found Tyr-Pro
`sequence to be cis 25 % of time, while cis Trp-Pro
`was absent. A high occurrence (19 %) of Tyr in cis
`bonds was
`also reported by MacArthur &
`Thornton (1991). From a larger database we find
`that the percentage of Tyr occurring in cis bonds
`has been reduced considerably (9.7 %), and Trp has
`become equally conspicuous (10.4 %). A Pro-Pro
`bond has the highest frequency (11.2 %) to be in
`the cis form. The residue X in X-Pro that causes the
`bond to be cis at least 6 % of the time belongs to
`one of the following four groups: (i) aromatic resi-
`dues, (ii) small residues, Gly and Ala; (iii) polar
`residues Ser, Gln and Arg; and (iv) Pro provide
`
`Figure 1. Schematic representation of cis and trans
`conformations around X-Xnp and X-Pro peptide bonds
`(where X (cid:136) any residue, Xnp (cid:136) any non-Pro residue).
`(b) Convention for numbering residues flanking a cis
`peptide bond.
`
`Lin, 1986; Kim & Baldwin, 1990). An enzyme,
`prolyl isomerase is known to catalyze the cis-trans
`isomerization of X-Pro bonds (Schmid et al., 1993).
`Experimental data have been derived on the ther-
`modynamics and kinetics of cis-trans isomerization
`by substituting a Pro at (10) by a non-Pro residue
`(Schultz & Baldwin, 1992; Mayr
`et al., 1993;
`Tweedy et al., 1993; Odefey et al., 1995; Vanhove et
`al., 1996). However, to understand the structural
`effect of such mutations it is important to know
`the conformational
`features of residues in and
`around (Figure 1(b)) the X-Xnp cis peptide linkages
`vis-a`-vis the X-Pro bond. Moreover, although a cis
`peptide bond can cause reversal of chain direction
`(Lewis et al., 1973) leading to two types of turns
`with the two central residues having canonical f,c
`(degree) values of ((cid:255)60,120; (cid:255)90,0) and ((cid:255)120,120;
`(cid:255)60,0) (Richardson, 1981; Rose et al., 1985). These
`values, though widely quoted in literature (Wilmot
`& Thornton, 1988), need a reassessment as c of the
`second residue in the second set is usually signifi-
`cantly off the idealized value. Consequently, we
`thought it important to make a reclassification of
`
`
`
`Cis Peptide Bonds
`
`273
`
`residues are given in Table 2. Being most abun-
`dant, the X-Pro cases were analyzed after grouping
`them into two main turn types (VIa and VIb), as
`well as the four subgroups (VIa-1, VIa-2, VIb-1 and
`VIb-2) of the above types. In addition to consider-
`ing individual residues we also analyzed the occur-
`rence of groups of residues, like small (Gly and
`Ala), aromatic (Phe, Tyr, Trp and His), b-branched
`(Val, Ile and Thr) and short polar (Ser, Asp and
`Asn).
`There are interesting trends considering groups
`of residues (Table 2C). Taking X-Pro (VIa) as an
`example, aromatic residues have high occurrences
`at positions (1) and (20), which decrease sharply on
`moving outward. On the contrary, the b-branched
`residues are less at positions (1) and (20) (especially
`in the former position, which is also indicated in
`Figure 2), and increase along the outward locations
`(especially upstream). X-Pro (VIb) and Xnp-Xnp
`cases have very similar position-specific variations
`of these two groups of residues. In Pro-Pro cases,
`the aromatic residues are abundant at position (20)
`and the branched residues at position (2). Short,
`polar residues (Ser, Asp and Asn) are likely to be a
`constituent of the cis Xnp-Xnp bond, and also be a
`part of X-Pro (type VIb) bond.
`Small residues have relatively higher occurrences
`in all the positions of Xnp-Xnp, and also in pos-
`ition (20) of X-Pro (Table 2C). Although taken
`together as small residues, Gly and Ala are not
`always found in similar numbers. For example,
`Gly is more abundant in the location (1) of Xnp-
`Xnp, whereas Ala predominates in locations (10)
`and (20) (Table 2B). Likewise in X-Pro (turn type
`VIa), Gly is prominent at position (20) and is exclu-
`sively found in position (2), but does not occur at
`all in position (1). As to be discussed later, because
`of the conformations being distinct from other X-
`Pro cases, Gly-Pro cis peptides belong to different
`turn categories, VIb-3, VIc and VId (Table 3). How-
`ever, if taken together, they have equal preferences
`of aromatic and b-branched residues at positions
`(3) and (20).
`Among the different turn types involving the X-
`Pro cases (Table 2A), VIa-2 type, in comparison to
`VIa-1, has a high proportion of small residues,
`notably Gly in position (20). Relative to the above
`two types, VIb-1 has a greater presence of Pro
`
`Figure 2. Histogram showing the percentage of occur-
`rence of various residues in the cis conformation of the
`X-Pro peptide bond; the numbers of cis cases are given
`on top of each bar.
`
`61 % of the data points. Branched aliphatic residues
`Val, Ile, Thr and Leu are less frequent. Recently,
`Reimer et al. (1998) have also calculated the amino
`acid frequency of cis prolyl bonds for every single
`amino acid preceding Pro. Some of their values are
`smaller
`than ours, possibly because of
`their
`inclusion of lower resolution (3.5 A˚ ) data, where
`cis bonds are underestimated (Weiss et al., 1998).
`The number of observations of the X-Xnp bond
`in the cis form is rather small to make any definite
`statement. Out of 29 cases (Table 1) Gly, Ser, Trp,
`Ala and Asp have higher occurrences at position
`(10)
`(1) and Asn, Ala, Thr, Asp and Phe at
`(Table 2B).
`
`Residue preferences in the neighbourhood of
`cis peptide bonds
`When considering the neighbours ((cid:6)six residues)
`of prolyl residues and their physicochemical prop-
`erties, Fro¨ mmel & Preissner (1990) found six differ-
`ent patterns which contained 75 % of known X-Pro
`cases. To see if the local sequence has any influence
`on the occurrence of a X-Pro, Pro-Pro or Xnp-Xnp
`bond in the cis conformation the percentage com-
`position of residues at each position, from (3) to
`(30) (Figure 1(b)), was calculated, and the preferred
`
`Table 1. Percentage occurrence in the cis conformation of X-Xnp sequences
`
`Range (%)
`
`0.0-0.2
`0.2-0.4
`0.4-0.6
`0.6-0.8
`0.8-1.0
`1.0-1.2
`4.3
`
`Sequencea
`
`AA,GA
`DA,GT,AD,LT,AT,SV,VN,EI,GF,GG(2),DD
`QL,FS,RD,SY,DN,SF,SR,PN,NY
`PY
`WA,HT
`HF,CA
`WN(3)
`
`a Within a range, the sequences are in an ascending order of occurrence. The number of cases, if
`more than one, is given in parenthesis.
`
`
`
`274
`
`Cis Peptide Bonds
`
`Table 2. Preference of amino acid residues around various categories of cis peptide units
`(10)
`
`Categories
`
`(3)
`
`(2)
`
`(1)
`
`(20)
`
`(30)
`
`A. Xnp-Pro: different turn typesa
`VIa-1 (39)
`G(15), A,V(10)
`[F,E,P]
`T(23),V,I(15)
`I(10),V(9)
`I(25),A,D,S(17)
`
`VIa-2 (13)
`VIb-1 (100)
`VIb-2 (12)
`
`D(15),G(13)
`[A,N,R]
`V(31),G(15)
`T,P(11),G(9) [E]
`G(42)
`
`A(23),Y(13), E(10)
`[G,T,N]
`L,S,W(15)
`S(10)
`N(25),T,H(17)
`
`B. Different possible sequences forming cis peptideb
`X-Pro (VIa) (52)
`G(13),V,T(12)
`G,D(13),V(12)
`[P]
`[A,N,R]
`I(11),A,V(9)
`G(12),T,P(9) [E]
`I(21),A(14), V,R(10)
`V,I,L,N(14)
`[D,K,P]
`[S,E,K,P]
`K(21)
`I,T,K,D(14)
`L(25),V,Y(19)
`Q(19),G,S(12)
`
`X-Pro (VIb) (116)
`Xnp-Xnp (29)
`
`Pro-Pro (14)
`Gly-Pro (16)
`
`X-Pro (VIb) (116)
`
`C. Preference of groups of residuesc
`X-Pro (VIa) (52)
`Sm(21),Ar(8),
`Bb(32),Sp(18)
`Sm(15),Ar(11),
`Bb(27),Sp(17)
`Sm(21),Ar(7),
`Bb(38),Sp(10)
`Sm(14),Ar(7),
`Bb(14),Sp(14)
`Sm(0),Ar(25),
`Bb(25),Sp(6)
`
`Xnp-Xnp (29)
`
`Pro-Pro (14)
`
`Gly-Pro (16)
`
`D. Most (and least) likely residuesd
`X-Pro (VIa)
`Bb,Sm [P]
`X-Pro (VIb)
`Bb
`Xnp-Xnpe
`Bb,Sm
`Pro-Pro
`-
`Gly-Pro
`Ar,Bb
`
`Sm(13),Ar(16),
`Bb(24),Sp(17)
`Sm(18),Ar(15),
`Bb(18),Sp(13)
`Sm(20),Ar(13),
`Bb(31),Sp(17)
`Sm(7),Ar(7),
`Bb(35),Sp(14)
`Sm(18),Ar(12),
`Bb(18),Sp(18)
`
`Bb,G [A,N,R]
`Bb,Sm [E]
`Bb,Sm
`Bb
`-
`
`P
`
`P
`P
`P
`
`P
`
`P
`A,N(17),T(14),
`D,F(10) [E,K]
`P
`P
`
`P
`
`P
`
`Sm(24),Ar(17),
`Bb(20),Sp(30)
`P
`
`A(19),L,Y(10)
`[G,N]
`N,S,T,Y(9)
`G(17),W,S(14),
`A,D(10) [I,T,K]
`P
`G
`
`Sm(19),Ar(26),
`Bb(10),Sp(10)
`Sm(9),Ar(23),
`Bb(16),Sp(21)
`Sm(27),Ar(24),
`Bb(3),Sp(27)
`P
`
`G
`
`P
`
`Ar,A [G,N]
`Ar,Sp
`Sm,Sp,W
`P
`G
`
`P
`P
`Sp,Sm
`P
`P
`
`G,F(15), T,D,N(10)
`[I,K,P]
`G(46),A(15)
`A(18)
`G(25),A,I(17)
`
`S,I(13), D,N,A(10)
`[L]
`S(23),Q(15)
`P(14)
`Q,R(17)
`
`G(23),F(12),T,
`D(10) [K,P]
`A(17),V,Y(9)
`A(24),G,L(14)
`[V,F,P]
`F(29),T,Q(14)
`V,F(19)
`
`S(15),I(12), P(10)
`[L]
`P(13),T(9)
`G(21),Q(14),
`L,S(10) [T,E]
`L,T(21),P(14)
`G(19)
`
`Sm(29),Ar(24),
`Bb(16),Sp(20)
`Sm(23),Ar(13),
`Bb(20),Sp(11)
`Sm(38),Ar(7),
`Bb(10),Sp(17)
`Sm(0),Ar(43),
`Bb(21),Sp(0)
`Sm(12),Ar(31),
`Bb(31),Sp(12)
`
`Sm,Ar [K,P]
`Sm,Bb
`Sm
`Ar
`Ar,Bb
`
`Sm(12),Ar(10),
`Bb(20),Sp(31)
`Sm(9),Ar(12),
`Bb(20),Sp(20)
`Sm(28),Ar(6),
`Bb(10),Sp(20)
`Sm(7),Ar(7),
`Bb(21),Sp(14)
`Sm(19),Ar(12),
`Bb(18),Sp(12)
`
`Sp,Bb [L]
`Sp,Bb
`Sm
`-
`-
`
`At each position (Figure 1(b)), the percentage residue composition is calculated and the residues having high values are entered
`with the percentage composition given in parentheses (when multiple residues have the same value the number is given after the
`last entry). If the first entry has a distinctly higher value than the next, it is given in bold and underlined. Residues whose average
`occurrence in protein structures is greater than 4 % (Pal & Chakrabarti, 1999a), but are not found at all in a given position, are given
`in italics within square parentheses. The number of cases in each category is given in column 1.
`a Given in Table 3 (sparsely populated types are excluded).
`b X-Pro sequences are broken into two classical VIa and VIb turns.
`c Residues are grouped as: Sm, small (G, A); Ar, aromatic (F, Y, W, H); Bb, b-branched (V, I, T); and Sp, short polar (S, D, N).
`d Indicated by either one-letter amino acid code, or a two-letter group designation.c
`e Less likely to have Pro and Lys all throughout.
`
`around the cis peptide. Because of steric factors,
`VIb-2 type needs to have a small residue (Gly in
`particular) at either position (2) or (20). Based on
`the above, the notable presence (or absence) of var-
`ious residues around the cis peptide moieties are
`summarized in Table 2D. Interestingly, there are
`only two examples of Pro-X cis peptides: 2CTC
`(PDB file) with sequence, Leu-Tyr-Pro-Tyr-Gly-Tyr
`and 1MKA, Pro-Ala-Pro-Asn-Met-Leu.
`
`Possible role of neighbouring residues in
`cis-trans isomerization
`
`Data in Table 2C show a contrast in the relative
`presence of aromatic and b-branched residues
`around the cis peptide units. For X-Pro cases, while
`aromatic residues have a higher presence at pos-
`ition (1), their numbers decline as one moves out
`along the sequence from the cis bond. On the other
`hand, the branched residues, show the opposite
`trend and have the maximum presence at position
`
`(3) (even for Xnp-Xnp cases). This observation is
`suggestive of the steric requirement for the isomer-
`ization of a trans peptide bond into cis. The resi-
`dues with two bulky alkyl groups at Cb (close to
`main-chain) if located at position (1) hinders the
`isomerization process. Support for the steric clash
`having an inhibitory role on the isomerization pro-
`cess also comes from nature of the residue preced-
`ing Pro-Pro cis peptides. In the sequence X-Pro-
`Pro, one may ask what determines the second
`bond to be in the cis peptide conformation rather
`than the first. It appears that a large percentage of
`these cases have a b-branched residues for X (and
`in addition, aromatics at position (20)). Even Xnp-
`Xnp cis peptides have a few such residues at either
`position (the relatively higher number at position
`(10) is due to a large contribution from Thr which,
`due to its polar features, acts in a different way, as
`discussed below). The b-branched residues, how-
`ever, may have a beneficial role when located at
`position (2) or (3). Because of the larger steric clash
`
`
`
`Cis Peptide Bonds
`
`275
`
`Table 3. Types of turns mediated by cis peptide bonds and their geometries
`
`Turn typea
`
`Conf.b
`
`No.
`
`f1
`
`c1
`
`f10
`
`c10
`
`A. Xnp-P
`VIa-1
`VIa-2
`VIb-1
`
`VIb-2
`VIb-3
`VIc
`VId
`
`B. P-P
`VIa-1
`VIb-1
`VIb-2
`
`BA
`BA
`BB
`
`BB
`BB
`RA
`RB
`
`BA
`BB
`BB
`
`39
`13
`100
`
`12
`4
`5
`7
`
`7
`6
`1
`
`(cid:255)74(24)
`(cid:255)131(24)
`(cid:255)117(26)
`(cid:255)134(12)
`(cid:255)100(20)
`104(38)
`102(20)
`
`(cid:255)54(5)
`(cid:255)69(6)
`(cid:255)84
`
`141(9)
`145(16)
`138(16)
`
`98(23)
`183(8)
`188(8)
`186(25)
`
`147(5)
`160(8)
`149
`
`(cid:255)93(9)
`(cid:255)79(9)
`(cid:255)77(10)
`(cid:255)78(12)
`(cid:255)72(10)
`(cid:255)83(9)
`(cid:255)69(8)
`
`(cid:255)81(5)
`(cid:255)77(11)
`(cid:255)96
`
`12(16)
`(cid:255)16(24)
`158(17)
`
`165(9)
`154(2)
`(cid:255)16(7)
`171(23)
`
`9(10)
`149(14)
`115
`
`Dist (A˚ )
`(2)-(20)
`
`Secondary structurec
`
`5.9(6)
`6(1)
`6.3(8)
`
`4.5(7)
`7.7(2)
`8.4(4)
`8.3(3)
`
`5.6(3)
`7.4(7)
`6.3
`
`TT(39)
`SS(5),CS(3),ET(2),BS,ES,II
`SS(50),CS(16),SC(10),ES(8),
`CC(8),BS(4),EC(2),BC,EE
`TT(12)
`SC(2),SS,EC
`CS(4),SS
`CC(3),EE(3),BC
`
`TT(7)
`SS(3),CC,SS,EE
`TT
`
`5
`3
`15
`2
`1
`3
`
`(cid:255)89(21)
`(cid:255)113(41)
`(cid:255)108(29)
`(cid:255)123(6)
`(cid:255)155
`131(30)
`
`134(30)
`149(9)
`121(23)
`121(57)
`176
`174(11)
`
`14(36)
`(cid:255)15(17)
`168(15)
`152(26)
`129
`202(13)
`
`6.4(9)
`7(1)
`8(1)
`6(1)
`8.6
`9.1(6)
`
`TT(5)
`CC,ET,EE
`EC(5),EE(4),SC(3),ES(2),CC
`TT(2)
`EE
`CC(2),EC
`
`C. Xnp-Xnp
`VIa-1
`VIa-2
`VIb-1
`VIb-2
`VIb-3
`VId
`
`BA
`BA
`BB
`BB
`BB
`RB
`
`(cid:255)111(17)
`(cid:255)106(7)
`(cid:255)134(21)
`(cid:255)102(23)
`(cid:255)102
`(cid:255)91(2)
`Data for eight cases are not included in the Table: two Pro-Xnp cases (with conformations BA and BB); one C-terminal cis peptide;
`one Gly-Pro sequence (LB); and four sterically strained non-Gly-Pro sequence (LB(2), AB(1), AA(1)). Representative diagrams are
`given in Figure 7.
`a VIb-3, VIc and VId turns have Gly at position (1). The hydrogen bond (Figure 6) is usually between residues (2) and (20) (provid-
`ing CO and NH groups, respectively) in VIa-1, (3) and (30) in VIb-2, and (1) and (10) (providing CH and CO, respectively) in VIb-3
`and VId.
`b Conformation based on the location of the two residues in the Ramachandran plot (see Materials and Methods and Figure 5).
`c Of positions (1) and (10) as specified by the program DSSP (Kabsch & Sander, 1983): H, a-helix; I, p-helix; E, strand; T, hydrogen
`bonded turn; S, non-hydrogen bonded turn; C, non-regular structure. The number of observations, if more than one, is given in
`parentheses.
`
`between the main and side-chain atoms, the f,c
`angles of
`these residues lie in a limited range
`(Chakrabarti & Pal, 1998), and thus they can act as
`a tether or a wrench to hold the chain in position
`while an adjacent bond is being isomerized.
`A corollary of the above hypothesis is that the
`small residues offering the minimum steric resist-
`ance should facilitate the cis form. Indeed, a large
`number of Gly and Ala residues are found in pos-
`itions (1) of X-Pro, (1) and (10) of Xnp-Xnp, and (20)
`of both. In the case of Xnp-Xnp there may be
`another factor operating during the trans to cis iso-
`merization. Most of these have polar residues, Ser,
`Thr, Asp and Asn at position (10) and their side-
`chains are usually within the hydrogen bonding
`distance of the main-chain NH group at the same
`position (although the angles, in the range 60-120 (cid:14),
`do not fulfill the usual hydrogen bond criterion).
`Even though the geometry may not be optimum, it
`is quite plausible that during isomerization such
`interaction may satisfy the hydrogen bonding
`potential of the NH group, and thus lower the acti-
`vation energy of the process. Participation of a
`nearby residue facilitating the cis-trans isomeriza-
`tion is known (Reimer et al., 1997). Once formed,
`the cis peptides may be stabilized by interactions
`(discussed later) involving aromatic residues which
`are found in large numbers at positions (1) of X-
`Pro and (20) of Pro-Pro and X-Pro (turn type VIa).
`
`Correlation between main-chain and
`side-chain conformations
`
`Recently, we have shown how the side-chain tor-
`sion angle w1
`is correlated with the backbone
`angles f and c of residues held by trans peptide
`linkage, and how the result can be used to classify
`the amino acid residues (Chakrabarti & Pal, 1998).
`The paucity of data for cis peptides does not allow
`one to study the interrelationships of angles for
`individual residues. However, some general trends
`can be deciphered (Figure 3). For example, in Xnp-
`Pro cases, the means of the distributions of the c
`values of Xnp get changed (130 (cid:14) ! 135 (cid:14) ! 148 (cid:14))
`as w1 goes from (cid:255)180 (cid:14) to (cid:255)60 (cid:14) to 60 (cid:14) (confor-
`mational states t, g(cid:135) and g(cid:255), respectively, which
`occur in the ratio (cid:25)3:5:1; Figure 3(a)). For a Pro
`residue in this position, though any value of w1
`from (cid:255)30 to (cid:135)30 (cid:14) is possible, negative values pre-
`dominate (in the ratio (cid:25)2:1). As noted earlier
`(MacArthur & Thornton, 1991), c is above 60 (cid:14) for
`in this position. Considering f
`a
`residue
`(Figure 3(b)), the points are rare below (cid:255)140 (cid:14) in
`the g(cid:135) state, whereas in the other two states,
`although the spread is from ca (cid:255)60 to (cid:255)170 (cid:14), most
`of the points are closer to the latter value.
`Pro in cis X-Pro has a noteworthy dependence of
`w1 on f and c (Figure 3(c) and (d)). Residues pre-
`dominantly have a positive w1 (positive:negative
`(cid:25)6:1). Notably, however, when c is less than 60 (cid:14),
`
`
`
`276
`
`Cis Peptide Bonds
`
`Figure 3. Joint distributions of w1 with f and c for residues at positions (1) and (10). Symbols used: ~, Pro, *,
`non-Pro, and these are open for X-Pro and filled for X-Xnp cases.
`
`a positive value of w1 is the norm, and only when
`c is (cid:24)120 (cid:14) or more a few points are also observed
`in the negative range of w1. Starting at (cid:255)60 (cid:14) the f
`values go up to (cid:255)80 (cid:14) when w1 is negative, whereas
`for positive w1 it can extend upto (cid:255)110 (cid:14).
`Although the residue X in both X-Pro and X-Xnp
`peptide units has similar conformational features,
`those for Pro and Xnp are considerably different.
`The most conspicuous but obvious difference is the
`w1 angles, which are restricted in the range (cid:255)30 to
`(cid:135)40 (cid:14) for Pro, whereas for the non-Pro residues
`
`there are three conformational states. Additionally,
`however, compared to the former, the f values of
`the latter are shifted towards more negative region
`(Figure 3(d)). Without the constraint on f imposed
`by the pyrrolidine ring, non-Pro residues, by
`taking a more extended value of f, reduces the
`steric clash between Ca of position (1) and the car-
`bonyl group of position (10).
`A striking feature of the w10 f10 plot (Figure 3(d))
`is the near linear relationship between the two par-
`ameters (irrespective of whether it is a Pro or a
`
`
`
`Cis Peptide Bonds
`
`277
`
`Figure 4. w1,c and w1,f plots for trans proline residues.
`
`non-Pro residue) when w10 has a positive value. In
`fact, if one excludes the Pro rings with small puck-
`ering (w10 < 10 (cid:14))
`then the correlation coefficient
`between the two parameters is (cid:255)0.65 (equation:
`w10 (cid:136) (cid:255) 0.42 f10 (cid:135) 1.59).
`
`Pyrrolidine ring puckering in cis peptides
`
`The pyrrolidine ring of the Pro residue invari-
`ably occurs in puckered conformations, which are
`essentially of two types, UP (or A or Cg-exo) and
`DOWN (or B or Cg-endo) depending on the place-
`ment of the Cg atom and the CO group of Pro on
`the opposite or the same the side of the plane
`defined by the remaining ring atoms (N, Ca, Cb
`and Cd)
`(Ramachandran et al., 1970; Ashida &
`Kakudo,
`1974; Milner-White
`al.,
`1992;
`et
`Chakrabarti & Chakrabarti, 1998). The UP confor-
`mation is characterized by negative w1 and w3 and
`positive w2 and w4 values, and the opposite holds
`good for the DOWN conformation. Both are isoe-
`nergetic when Pro is involved in a trans X-Pro
`bond. For the cis isomer however, 89 % of the Pro
`residues in proteins exhibit DOWN pucker with
`average values for the four side-chain torsion
`angles being 30, (cid:255)36, 24 and (cid:255)8 (cid:14) (Milner-White
`et al., 1992). To find out the reason for such an
`occurrence we have carried out a conformational
`analysis (in terms of f, c and w1) of Pro residues
`involved in cis (Figure 3(c) and (d)) and trans
`(Figure 4) peptide bonds. MacArthur & Thornton
`(1991) had observed that compared to trans, cis
`proline residues show a displacement to a more
`negative f values in both the A and B regions, and
`a more positive c value in the A region so as to
`reduce the steric clash between the Ca group of the
`preceding residue and Pro carbonyl group. While
`validating the earlier observations our results indi-
`cate the striking dependence of the main-chain tor-
`sion angles on w1. Cis proline residues in the A
`region (c (cid:136) (cid:255) 60 to 0 (cid:14)) with negative value of w1
`are almost non-existent; though rather uncommon,
`residues with c in the range of 10 to 120 (cid:14) are only
`found for trans residues if w1 is positive, whereas
`for cis such points are absent. More remarkable,
`however, is the interdependence of f and w1 as
`both the torsion angles are around bonds in the
`pyrrolidine ring, and can thus contribute to the
`observed puckering of the ring. If residues with
`positive and negative values of w1 are considered
`
`separately, within each group, as f is reduced w1
`tends to increase. As already mentioned,
`the
`unfavourable main-chain contacts around the cis
`bond are reduced by making f more negative as
`compared to the trans bond. As a result, when w1 is
`negative, while f varies from (cid:255)75 to (cid:255)40 (cid:14) for
`trans, the range is (cid:255)80 to (cid:255)60 (cid:14) for the cis proline
`residues. A shortened range of f means only a few
`cis Pro residues can have negative w1 angles. On
`the other hand, a more negative and wider range
`of f ((cid:255)110 to (cid:255)60 (cid:14)) is available when w1 is posi-
`tive, which is thus the predominant state of the
`side-chain
`conformation
`(DOWN puckering)
`observed for Pro residues involved in cis peptide
`bonds. Thus local steric interaction (resulting in
`more negative f, which in turn causes w1 to be
`positive) explain the DOWN puckering of the cis
`Pro residues, whereas in the case of the UP pucker-
`ing observed in Pro residues in the middle of a-
`helices it was a specific C-H(cid:1) (cid:1) (cid:1)O interaction invol-
`ving the CdH groups
`that was
`responsible
`(Chakrabarti & Chakrabarti, 1998).
`
`Conformations delineating cis peptide
`mediated turns
`
`The classic VIa and VIb turns formed by cis pro-
`line residues (Lewis et al., 1973) can be described
`by the two residues (1) and (10), residing in regions
`B and A, respectively, of the Ramachandran plot
`(see Materials and Methods), and both occupying
`the region B, respectively (MacArthur & Thornton,
`1991). Consequently, we have constructed f,c
`plots for pairs of residues forming cis peptides, X-
`Pro and X-Xnp, in three groups corresponding to
`type VIa and VIb turns, and those falling outside
`(Figure 5).
`In Figure 5(a), BA conformation (type VIa) of the
`two residues are shown. When f1 is greater than
`(cid:24) (cid:255) 90 (cid:14) there is a hydrogen bond between resi-
`dues (2) and (20) (sometimes between (2) and (30))
`(Figure 6). As f is decreased (below (cid:255)90 (cid:14)), the CO
`group of position (2) moves away from the NH
`(20) and the hydrogen bond between
`group of
`them is lost (the same thing can also be achieved,
`though the number of cases is not many, by
`decreasing c10 below (cid:255)30 (cid:14) so as to turn away the
`NH group of (20)). Consequently, we have subdi-
`vided type VIa turn type into two groups, VIa-1
`and VIa-2, the former with hydrogen bonding and
`
`
`
`278
`
`Cis Peptide Bonds
`
`Figure 5. f,c plots for residue pairs at positions (1) and (10) (each point is indicated by the one-letter amino acid
`code of the corresponding residue). (a) The first residue is in the region B and the second in A; (b) both are in the
`region B; and (c) the rest (except five cases, the first residue is in R, whereas the second is either in A or B region).
`Only the specified regions of the Ramachandran plot are shown in (a) and (b).
`
`
`
`Cis Peptide Bonds
`
`279
`
`the more
`of
`mational parameters). Because
`extended nature of f, the turn opens up in Xnp-
`Xnp cases (in Figure 7, compare (c) and (h), both
`having
`the
`same
`turn type,
`but different
`sequences), which thus have a longer (2)-(20) dis-
`tance (between Ca atoms) than what is observed in
`the corresponding X-Pro motif. The (2)-(20) distance
`is usually restricted below 7 A˚
`for a b-turn
`(Wilmot & Thornton, 1988). However, type VIc,
`VId and VIb-3 turns have longer distances (and
`may be termed as pseudo turns). This is because
`although the cis peptide causes a sharp turn
`between residues (1) and (10), preceding (1) there is
`also another turn caused by a positive f1 (in VIc
`and VId) and a very extended c1, one nullifying
`the effect of the other and nearly aligning the chain
`directions beyond positions (2) and (20). The rever-
`sal of the chain direction in a turn can be shown
`by using a virtual torsion angle defined in Figure 8.
`As expected, the peak for the distribution of X-Pro
`cis peptides occurs at a small angle (30 (cid:14)), but
`values extending up to 180 (cid:14) are found. The rela-
`tively less restricted nature of the turn in X-Xnp
`case is indicated by a shift of the peak to a higher
`value (60 (cid:14)).
`
`Position relative to the protein centre
`
`The global position of the cis bond is of vital
`interest due to various reasons. The speed of pro-
`tein folding is believed to be controlled kinetically
`by the rate of cis-trans isomerization. Proteases
`have been isolated which are selective for the Tyr-
`Pro bond only after it has been isomerized to the
`trans conformation (Vance et al., 1997). Similarly,
`the membrane-binding conformation of bovine
`prothrombin is generated following the trans ! cis
`isomertization of
`an X-Pro bond (Evans &
`Nelsestuen, 1996). For these proteins the proper
`exposure of the bond concerned should have a
`direct bearing to the function.
`To address the question of the location of the cis
`peptides in the three-dimensional structure we
`have carried out two types of calculations. First is
`the radial distribution of the cis units relative to the
`centre of mass of the polypeptide chain. Such a
`depiction is provided in Figure 9(a) where the pos-
`ition of
`the cis peptide in concentric
`shells
`(obtained by dividing the distance from the centre
`of mass to the outermost atom in the structure into
`ten equal parts) is shown. The X-Pro peak occurs
`at shell number 7, which may indicate a position in
`shallow crevices close to the surface of the protein.
`The distribution has a broad shoulder at 4, which
`is suggestive of a group of cis peptides, possibly
`with functional ro