C-capping and helix stability: the pro C-capping motif

J. Mol. Biol. (1997) 274, 276±288

C-capping and Helix Stability: The Pro C-capping Motif

JesuÂ s Prieto and Luis Serrano*

EMBL, Meyehofstrasse 169117 Heidelberg, Germany

Abbreviations used: TSP, sodiu(2,2,3,3,-2H4)propionate; HPLC, hchromatography; NOE, nuclear Otwo-dimensional; ppm, parts perscalar correlated spectroscopy; TOspectroscopy; NOESY, nuclear Ospectroscopy; CD, circular dichro

0022±2836/97/470276±13 $25.00/0/m

Here we have performed a statistical analysis of the protein database to®nd new putative local C-terminal motifs in a-helices. Our analysisshows that certain combinations of X-Pro pairs (Asn, Cys, His, Phe, Tyr,Trp, Ile, Val and Leu), in which residue X is the C-cap and the Pro is atposition C0, are more abundant than expected. In those pairs, except forthe aliphatic residues, the presence of the Pro residue at C0 tends torestrict the f and c dihedral angles of the residue at position C-cap,around ÿ130�, 70�, respectively. For the aromatic residues as well as forHis, the w1 angle is around ÿ60� and the edge of the His and aromaticrings are close to the carbonyl group of the residue i ÿ 4. In all the pairshaving the above dihedral angles for residue C-cap, the main-chainamino group of Pro at C0 is close to the last three main-chain carbonylsof the a-helix. The above structural arrangements suggests the existenceof a stabilising electrostatic interaction of the residues at positions C-capand C0 with the helix macrodipole. We have denominated this putativelocal motif, the Pro-capping motif. To asses its importance in helix stab-ility we have analysed by nuclear magnetic resonance (NMR) and far-UVcircular dichroism (CD) a set of polyalanine-based peptides containingtwo of the above pairs: His-Pro and Phe-Pro, as well as the correspond-ing controls. In the case of the His-Pro pair we have found NMR evi-dence for the formation of the Pro-capping motif in aqueous solution. CDanalysis shows that the presence of a Pro residue alters the C-cap proper-ties of the preceding amino acids in the case of His and Phe makes themmore favourable. The Pro-capping motif with the appropriate sequence,determines the location of the C terminus of a-helices and stabilises thehelical conformation having Pro as the C0 residue.

# 1997 Academic Press Limited

Keywords: a-helix; circular dichroism; nuclear magnetic resonance; proteinstability; helix/coil
*Corresponding author
Introduction

The identi®cation of local structure motifs whichare suf®ciently stable to reduce the conformational¯exibility at certain points of the polypeptidechain, is quite important to simplify and under-stand the way in which a protein folds. In a-helicestwo local motifs have been identi®ed at the N ter-minus. The capping-box motif which involves tworeciprocal hydrogen-bonds involving the main-chain residue N-cap and the side-chain of a Glu, orAsp, at position N3 and the N-cap side-chain with

m 3-trimethylsilyligh performance liquidverhauser effect; 2D,million; COSY, 2DCSY, total correlation

verhauser enhancementism.

b971322

the main-chain of residue N3 (Harper & Rose,1993; Richardson & Richardson nomenclature,1988). The second is the result of a hydrophobicinteraction between the side-chains of the two resi-dues located at positions N0 and N4, plus a cap-ping box (Seale et al., 1994), or a good cappingresidue (Ser, Thr, Asn, Asp: hydrophobic staple;MunÄ oz et al., 1995; MunÄ oz & Serrano, 1995a). Atthe C terminus of a-helices the presence of a resi-due with positive angles at position C0 and twomain-chain/main-chain hydrogen-bonds betweenresidues C3 with C00 and C2 with C0, has beendenominated the Schellman motif (Schellman,1980). In the case of the Schellman motif thesequence ®ngerprint consists of a Glycine residueat position i, a residue with a hydrophobic side-chain at positions i ÿ 4 and i � 1 and a polar orAla residue at position i ÿ 2 (Aurora et al., 1995).Experimental analysis of polyalanine-based pep-tides with moderate helical content, has shown

# 1997 Academic Press Limited

Helix Termination Signals 277

that the Schellman motif is not signi®cantly formedin aqueous solution and seems to contribute littleto helix stability. However, it is formed in 30% TFE(Viguera & Serrano, 1995a). The same is observedwith a natural fragment corresponding to the a-helix of the B1 domain of protein G (Blanco &Serrano, 1995). A similar study has been done in adesigned heteropolypeptide with a high helicalcontent (Gong et al., 1995). The authors concludethat the motif was formed in aqueous solutionbased on a structure calculation in vacuo, but as inthe previous cases there was no NOE evidence forit. These results suggest that the Schellman motif isformed as a consequence of helix formation anddoes not strongly nucleate a-helices, opposite towhat happens with the local motifs at the N termi-nus. If this is the case, it could be that other localinteractions at the C terminus of a-helices stabiliseand nucleate these secondary structure elements.Here we have performed a statistical analysis ofthe database, to detect other possible C-cap localmotifs. Based on this analysis we have synthesizedand analysed a set of peptides by NMR and CD, todetermine the role of a new putative motif, thePro-capping, in helix stability.

Results

Through the following text we will refer fre-quently to the different positions at the C-terminalof the a-helix. We have chosen to follow thenomenclature described by Aurora et al. (1995).

Table 1. Statistical preferences for the different amino acidposition C0, compared to those when any amino acid occupi

Sec.Straa

H/H/H/STC/STCX-ProH

aH/H/H/STC

X-Xb

Ala 0.095 (14) 0.10 (16Cys 0.054 (8) 0.02 (38Asp 0.027 (4) 0.05 (81Glu 0.014 (2) 0.04 (62Phe 0.068 (10) 0.03 (51Gly 0.041 (6) 0.19 (31His 0.061 (9) 0.03 (52Ile 0.054 (8) 0.02 (39Lys 0.068 (10) 0.07 (11Leu 0.120 (18) 0.08 (12Met 0.007 (1) 0.02 (28Asn 0.101 (15) 0.08 (12Pro 0.000 (0) 0.00 (0)Gln 0.041 (6) 0.04 (70Arg 0.041 (6) 0.04 (72Ser 0.054 (8) 0.07 (107Thr 0.034 (5) 0.03 (51)Val 0.047 (7) 0.03 (46)Trp 0.014 (2) 0.00 (8)Tyr 0.061 (9) 0.03 (53)

All the values have been normalised to 1. In brackets we show thpro®le used in the database search, where H is helix, S is extended,

aX-ProH, probability of ®nding any amino acid at position C-capbX-X, probability of ®nding any amino acid at position C-cap of acX-Pro/X-X, Ratio of the two probabilities.dX-ProC, probability of ®nding any amino acid before a Pro in theeX-ProH/X-ProC, Ratio of the two probabilities.

The nomenclature for a-helices and the C terminus¯anking residues is as follows:

C4 C3 C2 C1 C-cap C0 C00

where C4 to C1 are helical residues, C-cap, C0 andC00 are non-helical residues. The criteria for a resi-due to be helical or not is based on the Kabsch &Sander (1983), de®nition of secondary structure.

Statistical and structural analysis

Apart from the high frequency of Gly and therelative abundance of His, Lys and Arg residues atthe last helical turn, Pro immediately following theC-terminus of a-helices (position C0), is also remark-ably abundant (Richardson & Richardson, 1988;Dasgupta & Bell, 1993). Consideration of the resi-due at position C-cap when Pro is at position C0,shows that certain pairs are more frequent thanexpected from the normal distribution at the C-capof a-helices (Table 1). Interestingly enough, the ali-phatic and aromatic residues, together with Asn,His and Cys are the most favoured residues accom-panying Pro at position C0. These preferences couldbe the result of some favourable interactionbetween the two residues and not be related tohelix termination signals. To eliminate this possi-bility we searched the database for the 20 X-Propairs, without looking for any secondary structurein particular. Comparison of the relative probabil-ities of the 20 amino acids when accompanying aPro residue at the end of helices or in any secondary

s at position C-cap of an a-helix, when there is a Pro ates that position

/STCX-Pro/X-Xc

*/*/*/*/*X-ProC

d X-ProH/X-ProCe

1) 0.9 0.08 (228) 1.2) 2.3 0.02 (61) 2.6) 0.5 0.06 (172) 0.5) 0.3 0.05 (141) 0.3) 2.1 0.04 (128) 1.60) 0.2 0.06 (186) 0.7) 1.9 0.03 (94) 1.9) 2.2 0.06 (192) 0.83) 1.0 0.06 (182) 1.15) 1.6 0.09 (262) 1.4) 0.4 0.02 (62) 0.39) 1.3 0.06 (181) 1.7

± 0.04 (111) 0.0) 0.9 0.04 (107) 1.1) 0.9 0.04 (117) 1.0) 0.8 0.07 (196) 0.8

1.1 0.07 (200) 0.51.6 0.07 (220) 0.62.7 0.01 (34) 1.21.8 0.04 (110) 1.6

e actual number of hits for each case. Sec.Str., Secondary structureT is turn, C is coil and * means any secondary structure.

before a Pro when this residue is at position C0 of an a-helixn a-helix.

database.

278 Helix Termination Signals

structure conformation, supports the idea that thepreferences observed are helix dependent (Table 1).

We have also analysed the dihedral angles andthe w1 rotamer of the 20 amino acids at position C-cap of an a-helix (Figure 1A to F). We have dividedthe 20 amino acids according to their relative fre-quencies in the X-Pro pairs shown in Table 1, aswell as their chemical nature. In Figure 1A, C and Ewe show the f and c dihedral angles of Gly, Ala,Ser, Thr, Asp, Glu, Gln, Met, Lys and Arg(Figure 1A), Leu, Ile and Val (Figure 1C) and Phe,Tyr, Trp, His, Cys and Asn (Figure 1E), at positionC-cap of a-helices when there is a Pro residue atposition C0. The same is displayed in Figure 1B, Dand F, but in this case we have considered everyamino acid at position C0. At a ®rst glance it isimmediately clear that the presence of a Pro residue

tends to favour f angles at position C-cap in therange ÿ150� to ÿ120�. This effect is much moreimportant for the last group of amino acids (Phe,Tyr, Trp, His, Cys and Asn; 48 out of 55 cases havethese dihedral angles, while in the other two groupsthey correspond to around 50% of the cases). Thesame applies to the c angle which is quite restrictedaround 70� in this group (Figure 1E) and correlatesvery well with the f angle values. Regarding the w1

angle we observe that it is mainly restricted aroundÿ60� for the last two groups, but this preference isnot related to the presence of a Pro at position C0(data not shown). These results indicate that a Proat position C0 results in strong restrictions in the fand c angles of Phe, Tyr, Trp, His, Cys and Asnresidues at position C-cap (we observe for all ofthem, approximately the same relative number of

Figure 1. Dihedral angle distri-bution of the 20 amino acids atposition C-cap of an a-helix. Theprotein database was searched(see Materials and Methods), forthe following secondary structureconformation: H/H/H/STC/STC,where H is helix and STC, meansnon-helix. The amino acids at pos-ition C-cap were divided in threegroups according to Table 1 andtheir chemical properties. We havesorted the different hits accordingto the value of the f angle in anascending manner. In the plot weshow in the x-axis all the casesfound in the database for eachquery. In the y-axis we show the f(circles) and c (squares) angles foreach particular hit. A, Gly, Ala, Ser,Thr, Asp, Glu, Gln, Met, Lys andArg at position C-cap, when Pro isat position C0. B, Gly, Ala, Ser, Thr,Asp, Glu, Gln, Met, Lys and Arg atposition C-cap, when any aminoacid can be found at position C0. C,Leu, Ile and Val at position C-capwhen Pro if at position C0. D, Leu,Ile and Val at position C-cap whenany amino acid can be found atposition C0. E, Phe, Tyr, Trp, His,Cys and Asn at position C-cap,when Pro is at position C0. F, Phe,Tyr, Trp, His, Cys and Asn at pos-ition C-cap, when any amino acidcan be found at position C0.

Table 2. Dihedral angles found for the Asn-Pro, Cys-Pro, His-Pro, Phe-Pro, Tyr-Pro and Trp-Pro pairs at the C termi-nus of a-helices

Helix PositionC-cap C0

aa Prot. Res. f c w1 w2 f c

Asn1xyz* 271 ÿ105.76 119.57 ÿ69.47 ÿ77.34 ÿ62.25 ÿ22.271mla* 251 ÿ133.00 157.05 ÿ72.93 ÿ77.84 ÿ57.78 143.662pgd 382 ÿ154.61 94.08 178.54 7.31 ÿ71.07 ÿ8.252abk 17 ÿ149.93 90.60 ÿ172.92 ÿ179.34 ÿ58.48 ÿ32.681pnk 108 ÿ143.78 64.12 ÿ175.14 173.38 ÿ73.39 ÿ4.171php 184 ÿ147.91 62.91 ÿ166.14 43.29 ÿ66.03 150.001dad 72 ÿ138.41 85.24 175.22 ÿ141.14 ÿ63.52 ÿ34.821ido 61 ÿ144.01 76.14 173.01 32.85 ÿ72.93 48.791cot 65 ÿ134.47 77.30 ÿ58.75 ÿ85.25 ÿ59.12 ÿ28.491tml 20 ÿ129.29 57.56 ÿ67.95 ÿ84.86 ÿ62.34 ÿ21.691dpe 178 ÿ124.58 84.73 ÿ50.43 104.35 ÿ63.64 140.781oyc 353 ÿ135.51 95.71 ÿ70.11 ÿ88.02 ÿ62.86 ÿ28.321clc 304 ÿ131.18 79.90 ÿ56.57 ÿ103.03 ÿ64.30 ÿ8.281lki 97 ÿ130.59 70.56 ÿ71.61 ÿ46.51 ÿ63.94 ÿ16.201cdk 25 ÿ123.62 77.13 ÿ51.96 ÿ58.51 ÿ76.94 135.99

Cys1fxd* 50 ÿ82.99 120.91 175.93 ÿ67.92 ÿ26.831thv* 134 ÿ52.77 128.85 ÿ169.59 ÿ58.80 146.682erl* 36 ÿ111.56 148.30 ÿ69.85 ÿ64.66 147.531fxd 18 ÿ143.19 71.08 ÿ153.50 ÿ63.10 ÿ19.871cus 93 ÿ132.30 78.13 ÿ55.03 ÿ61.24 ÿ21.841thv 177 ÿ138.15 85.63 ÿ73.89 ÿ63.39 ÿ24.242cmd 109 ÿ137.73 65.48 53.79 ÿ64.15 ÿ16.821rec 31 ÿ127.63 81.41 ÿ56.87 ÿ50.11 ÿ29.32

His1smd* 330 ÿ78.20 138.56 ÿ163.52 ÿ83.58 ÿ64.88 ÿ21.492ohx* 105 ÿ71.00 139.37 ÿ173.69 ÿ82.77 ÿ63.94 ÿ29.272hbg 38 ÿ134.34 77.42 ÿ48.43 84.73 ÿ49.42 ÿ39.481nar 75 ÿ136.93 73.79 ÿ56.16 89.24 ÿ57.85 ÿ28.77

1mml 191 ÿ129.21 68.39 ÿ53.10 81.34 ÿ60.87 ÿ20.081cse 237 ÿ135.33 73.09 ÿ50.25 ÿ83.93 ÿ66.18 ÿ17.872ctc 29 ÿ137.28 60.13 ÿ60.78 ÿ83.65 ÿ57.45 ÿ27.072dri 180 ÿ136.89 64.63 ÿ55.16 ÿ76.09 ÿ65.94 ÿ13.611pda 29 ÿ135.50 79.69 ÿ48.10 ÿ73.84 ÿ63.29 ÿ11.84

Phe1csh* 129 ÿ70.26 146.59 ÿ89.77 ÿ29.27 ÿ74.60 157.293dfr 49 ÿ62.84 137.63 ÿ55.87 ÿ34.93 ÿ73.20 ÿ22.49

1gsa* 297 ÿ135.68 158.61 ÿ71.74 ÿ75.36 ÿ78.36 66.771xyz 293 ÿ122.23 80.63 ÿ54.32 ÿ82.04 ÿ53.80 134.73153l 131 ÿ136.68 73.56 ÿ64.86 ÿ72.48 ÿ72.83 ÿ4.931cpc 63 ÿ124.57 71.12 ÿ60.19 ÿ64.73 ÿ56.00 ÿ23.322cyp 98 ÿ129.22 73.80 ÿ60.29 ÿ81.72 ÿ69.79 ÿ15.231mml 6 ÿ136.26 63.86 ÿ52.79 ÿ78.29 ÿ63.72 ÿ25.261rec 49 ÿ131.93 71.49 ÿ75.83 ÿ87.69 ÿ61.68 ÿ35.591frp 218 ÿ124.06 82.90 ÿ58.22 80.64 ÿ59.67 123.16

Tyr2abk* 55 ÿ64.88 ÿ29.51 ÿ69.05 ÿ8.18 ÿ63.27 ÿ19.334enl* 289 ÿ124.40 152.99 ÿ56.49 ÿ77.65 ÿ75.24 55.531cse 22 ÿ131.52 59.66 ÿ66.22 ÿ81.33 ÿ64.85 ÿ8.404ptp 152 ÿ133.19 72.02 ÿ64.98 ÿ77.37 ÿ54.26 132.381thg 393 ÿ119.82 91.07 ÿ74.01 80.46 ÿ78.75 169.413tgl 129 ÿ131.09 66.31 ÿ59.83 ÿ90.22 ÿ59.96 ÿ19.031lau 108 ÿ127.40 86.05 ÿ70.53 ÿ78.16 ÿ58.77 ÿ26.931gky 25 ÿ127.12 65.11 ÿ58.10 ÿ82.99 ÿ65.26 ÿ26.963rub 94 ÿ123.51 76.63 ÿ60.95 ÿ75.90 ÿ65.34 ÿ13.57

Trp1vhh 80 ÿ133.31 85.30 ÿ53.86 ÿ71.71 ÿ52.01 133.412hpd 325 ÿ130.03 76.50 ÿ73.36 113.58 ÿ59.83 126.80

The secondary structure ®ngerprint used for searching the protein database was HHH/STC/STC, where H is helical and STC isnon-helical. Position ®ve was always a Pro and position 4, Asn, Cys, His, Phe, Tyr and Trp. The proteins found are shown in the®rst column (Prot), together with the sequence number of the fourth residue (Res). The programme WHATIF (Vriend, 1990) hasbeen used for data search. Those cases in which the f and/or c angles of residue C-cap deviate by more than one standard devia-tion from the total average, are marked with asterisks.



cases in which the f and c angles are different bymore than one standard deviation from the average;(Table 2). Statistical analysis results in an averagevalue of ÿ124(�23)� for the f angle and 87(�33)�for the c angle. If we eliminate those cases separ-ated by more than one standard deviation from theaverage, we obtain values of ÿ133(�9)� and 75(�9)�for the f and c angles, respectively. Interestinglyenough, a search of a-helices having a C-cap residue

with the above dihedral angles ®nds 54 cases, ofwhich 50 have a Pro at position C0. The relativeprobabilities of ®nding Asn, Cys, His, Phe, Tyr andTrp at C-cap, in this group, is double of what weshowed in Table 1 (data not shown). In all thesecases the proteins found to carry these amino acidpairs, did not have any signi®cant sequence or func-tional similarity (data not shown).

The above data suggests that the combination ofPro at position C0 of a-helices with Asn, Cys, His,Phe, Tyr and Trp at position C-cap, is a local motifthat could stabilise the helical conformation, deter-mine the end of the helix and the orientation of thepolypeptide chain exiting this secondary structureelement. We have looked for the common charac-teristics of all these pairs when residue C-capadopts the dihedral angles mentioned above, to®nd any structural reasoning behind these prefer-ences. In all cases the main-chain NH group ofresidue C-cap makes a bifurcated hydrogen-bonded to the main-chain CO groups of residuesi ÿ 4 and i ÿ 3. The Pro ring sits almost perpen-dicular to the helix C-terminal one turn with itsmain-chain amino group facing the last three heli-cal carbonyl groups (average distances of 3.4(�0.3),4.4(�0.4) and 4.2(�0.1) AÊ , to residues i ÿ 2, i ÿ 3and i ÿ 4, respectively; Figure 2 Top and Bottom).In the case of His and the three aromatic aminoacids, their side-chains adopt a w1 angle of � ÿ 60�,which results in the edge of the ring being closer tothe CO group of residue i ÿ 4. For His, its side-chain Nd1, or Ne2, groups are pointing towards theCO group of residue i ÿ 4, and the distance andgeometry could correspond to a weak hydrogen-bond, or to a favourable dipole-dipole interaction(average distance � 4.3(�0.7) AÊ (Figure 2 Top).The aromatic rings of Phe and Tyr point in thesame direction (Figure 2 Bottom), but in this caseone of the Cd groups is closer to the CO groupof residue i ÿ 4. For Asn and Cys the results arenot so clear cut since the w1 angle is not sorestricted (Table 2).

We have chosen two pairs as representative ofthis group of amino acids, Phe-Pro and His-Pro, tofurther investigate the importance of having a Proat position C0 of an a-helix. The pair Ala-Pro has

been also selected as a control (the relative prob-ability of being at position C-cap associated or notwith a Pro at C0, is around 1, Table 1).

Peptide design

To assess experimentally the contribution of thePro-capping motif, we have designed a series ofpeptides with the following template sequence:

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18N00 N0 Nc N1 N2 N3

Ac-Tyr-Gly-Ser-Ala-Ala-Glu-Ala-Ala-Ala-Arg-Ala-Ala-Ala-Lys±Y- -Z-Gly-Gly-Am

where Ac means acetylated, Am, amydated andNc means the N-cap. The C-cap position variesdepending of the residue located at position Z.When Z is a Gly, the C-cap will be at this position.When Z is a Pro the C-cap will be at position Y.This temple peptide has a capping-box motif (Ser3and Glu6) to increase the helical content due to itsfavourable nucleating activity (Lyu et al., 1992,1993). There is a Tyr residue at the N terminus todetermine peptide concentration, which is separ-ated by one Gly residue from the rest of the pep-tide to minimise the contribution of the aromaticring to the ellipticity at 222 nm (Chakrabartty et al.,1993). At the C terminus there are two Gly resi-dues to prevent any effects of the C-terminal groupon the helical content (MunÄ oz & Serrano, 1995b).There is an Arg and a Lys ar positions 10 and 14,respectively, to favour peptide solubility. Atposition Y we have placed Ala, Phe or His and atposition Z, Pro or Gly.

Ac-YGSAAEAAARAAAKAGGG-Am C-AAGAc-YGSAAEAAARAAAKHGGG-Am C-AHGAc-YGSAAEAAARAAAKFGGG-Am C-AFGAc-YGSAAEAAARAAAKAPGG-Am C-AAPAc-YGSAAEAAARAAAKHPGG-Am C-AHPAc-YGSAAEAAARAAAKFPGG-Am C-AFP

The peptides C-AAG, C-AFG and C-AHG havebeen analysed as control peptides to calibrate theC-cap contribution of Ala, Phe and His and also inthe case of C-AHG to determine the favourableinteraction of His� with the helix dipole(Armstrong & Baldwin, 1993).

NMR analysis in aqueous solution

To determine if the Pro-capping motif is formedindepently of packing contacts against the rest ofthe protein we have analysed by nuclear magneticresonance peptides C-AAP, C-AFP, C-AHP and C-AHG. Peptide C-AHG has been analysed as a con-trol since it is well known that a His at the C termi-nus of an a-helix stabilises the helical conformation(Armstrong & Baldwin, 1993). By comparing pep-tides C-AHG and C-AHP, we could distinguish ifany NOE of the His with the rest of the helix is

Figure 2. Top, Stereo Figure of the C-terminal of an a-helix from protein 2hbg (residues 31 to 39), with His-Prosequence at positions C-cap and C0. Bottom, Stereo Figure of the C-terminal of an a-helix from protein 2cyp (residues92 to 100), with a Phe-Pro sequence at positions C-cap and C0. The side-chains of the rest of the residues have beenremoved. The distance between the groups mentioned in Results are shown by broken lines.


due to its favourable C-cap contribution or if it isinduced by the presence of a Pro residue at pos-ition C0.

Figure 3 shows the CaH conformational shifts ofthe four peptides. Except for peptide C-AAP wecan observe negative values for the CaH confor-mational shifts between residues Ser3 and Lys14,which suggest the existence of a signi®cant helicalpopulation spanning these residues. In the case ofpeptide C-AAP, the helical conformation stops oneresidue before at position Ala13. The most conspic-uous difference between these peptides, is thelarge down®eld conformational shift of the Ca pro-ton of residue 15 when there is a Pro at position 16(Figure 3). A similar effect has been described inrandom coil peptides of the type G-G-X-P-G(Wishart et al., 1995) and it is probably due to theconformational restriction imposed on residue i ÿ 1by the side-chain of Pro at position i. Comparisonof the three peptides with Pro at position 16,shows that the main differences are located aroundresidues Ala13, Lys14, as well as position 15. Thelarger absolute value of the CaH conformationalshift at position 14 in peptide C-AFP could be

partly due to ring current effects of the aromaticring of Phe and not necessarily to a larger helicalpopulation. Figure 4 shows the NOE summary forpeptides C-AFP, C-AHP and C-AHG. Due to thehigh number of Ala residues there is an extensivesignal overlap and consequently only some of thenon -sequential NOEs can be detected. We ®nd inall the cases the expected NOEs in a capping-box:between the Cb and Cg protons of Glu6 and theamide protein of Ser3 and between the Cb protonsof Ser3 and the amide proton of Glu6. In the His-Pro peptide we ®nd weak NOEs between the side-chain Cd group of Pro16 and the Ca proton of resi-dues Lys14 and Ala13, as well as between the Cd

and Cb groups of His15, and the Ca of Ala12 or 13(they are overlapped). These NOEs are expected ifthe His-Pro pair adopts the conformation shown inFigure 2 Top, indicating that a conformation simi-lar to that found in the protein database is formedin aqueous solution, in equilibrium with other con-formations. In the C-AFP peptide only the side-chain NOE with the Ca of Ala12 or 13 (they areoverlapped), is found.

Figure 3. Conformational shifts of the CaH protons.(Empty circles) C-AAP, (®lled circles) C-AHP, (emptytriangles) C-AFP, (®lled triangles) C-AHG.

Figure 5. Far-UV CD spectra. Empty squares, C-AAG;®lled squares, C-AFG; empty circles, C-AAP; ®lledcircles, C-AHP; empty triangles, C-AFP; ®lled triangles,C-AHG.


CD analysis

We can analyse the relative contribution of thePro-capping motif, to a-helix stability through thecircular dichroism (CD) analysis of the helical con-tent of the peptides mentioned above (Figure 5).The error in the estimation of the helical content ofpeptides in the absence of aromatic residues (ifthey are not separated by at least two Gly residuesfrom the helical region, Chakrabartty et al., 1993),or alternative secondary structure conformations,is mainly dependent on the imprecision in thedetermination of peptide concentration. The pre-sence of a Tyr residue allows us to obtain an accu-racy of �1 to 2% in the estimation of peptideconcentration (Shoemaker et al., 1990). In the caseof the peptides containing Phe residue in thehelical region, the aromatic ring has a signi®cantpositive contribution around 222 nm, preventingus from using the method of Chen et al. (1974), to

Figure 4. NOE summary of the different peptides analysechain of residue i and the side-chain of residue i � n. dschbdschdsch, indicates NOEs between the side-chain of residucate an NOE assignment that is dubious due to signal overl

calculate the helical content (Chakrabartty et al.,1993). Previously, we found that the aromatic con-tribution of Phe at 193 nm, is nil and that it is poss-ible to correlate the ellipticity at this wavelengthwith the helical content, when similar peptides areavailable without a Phe (Viguera & Serrano,1995b). In Table 3 we show the quanti®cation ofthe helical content for the different peptides, usingthe ellipticity at 222 and 193 nm. We also presenttwo other independent parameters related to thehelical content, the position of the minimum andthe ratio R1 (see Materials and Methods), whichcorrelate very well with the estimation of the heli-cal population (correlation coef®cient 0.89 and0.996, with the values obtained from the ellipticityat 222 and 193 nm, respectively).

Mutation of Gly16 to Pro results in a very sig-ni®cant destabilization of the helical conformation(C-AAG: 39% versus C-AAP: 26%, in agreementwith previous studies which showed that Pro is aworse C-cap than Gly (Doig & Baldwin, 1995).This result correlates very well with the NMR data

d in this work, dbsch, indicates NOEs between the main-, indicates NOEs between the side-chain of residue i � n.e i and the side-chain of residue i � n. Broken lines indi-apping.

Table 3. CD analysis

%Helixe %Helixf

Peptide 193 nma 222 nmb Min(nm)c R1d (222 nm) (193 nm)

C-AAG 24575 ÿ13929 205.6 ÿ1.49 39 40C-AFG 21476 ÿ11453 205.5 ÿ1.42 32 37C-AHG 24653 ÿ13295 205.5 ÿ1.54 40 40C-AAP 10524 ÿ8607 204.6 ÿ0.77 26 26C-AFP 15898 ÿ9430 205.2 ÿ1.13 28 31C-AHP 19063 ÿ11499 205.2 ÿ1.26 34 34

aMean residue ellipticity at 193 nm.bMean residue ellipticity at 222 nm.cPosition of the ellipticity minimum.dRatio R1, which is proportional to the helical content and independent of protein concentration (Bruch

et al., 1991).eHelical content determined by the method of Chen et al. (1974).fHelical content calculated from the ellipticity at 193 nm (see Materials and Methods).

Table 4. Quanti®cation of the free energy contributionof the Pro-capping motif

Peptide %Helixexp %HelixAgadir ��Gcapping(Ala-X)

C-AAG 40 40C-AFG 37 36 ÿ0.4 � 0.2C-AHG 40 40 ÿ0.8 � 0.2a

C-AAP 26 26C-AFP 31 31 ÿ1.0 � 0.2C-AHP 34 34 ÿ1.2 � 0.2a

The error was determined by ®tting AGADIR1s (MunÄ on &Serrano, 1997) with the CD data assuming an experimentalerror in the determination of the helical content of �2%.

aIncluding the electrostatic interaction with the helix dipoleat C-cap (ÿ0.6 kcal/mol). %Helixexp, experimental helical con-tent measured by CD and after correction for the aromatic con-tribution of Phe. %HelixAgadir, helical content predicted byAGADIR1s after ®tting the CD data and using the parametersdescribed in Materials and Methods. ��Gcapping(Ala-X), C-cap-ping contribution of Phe and His compared to Ala, when thereis a Gly or a Pro residue at position C0.


which showed that the length of the helical confor-mation is one residue shorter in peptide C-AAP,than in peptide C-AAG. Substitution of Ala15 byPhe in the peptide with a Gly at position 16 (C-AFG), results in a decrease on the helical confor-mation. This is expected since the intrinsic helicalpropensity of Phe is far worse than Ala (O'Neil &DeGrado, 1990; Lyu et al., 1990; Horovitz et al.,1992; Blaber et al., 1993; Chakrabartty et al., 1994;MunÄ oz & Serrano, 1994, 1995b). However, this isnot the case for His, since peptide C-AHG has thesame helical content as peptide C-AAG. His intrin-sic helical propensity is even worse than that ofPhe (O'Neil & DeGrado, 1990; Lyu et al., 1990;Horovitz et al., 1992; Blaber et al., 1993;Chakrabartty et al., 1994; MunÄ oz & Serrano,1995b). Therefore, the only explanation must bethat His� is a far better C-cap than Ala (Armstrong& Baldwin, 1993). Opposite to this, Doig &Baldwin (1995) found that His� at the C terminusof a peptide with non-protected ends decreased thehelical conformation with respect to Ala. It couldbe that His is a good C-cap when its carbonylgroup is part of a peptide bond. Interestinglyenough, the behaviour of the peptides containing aPro at position 16 is very different. Peptides C-AFPand C-AHP have higher helical contents than thecontrol peptide C-AAP. Since we know by NMRthat the helical conformation in those peptidesstops at position 15, this suggests that the C-capcontribution of His and Phe when followed by aPro residue is more favourable than when followedby a Gly residue.

Free energies of interaction

We have estimated the free energies of inter-action by using the helix/coil transition theoryincorporated in the algorithm AGADIR1s (MunÄ oz& Serrano, 1997) (Table 4, see Materials andMethods). A Pro residue tends to restrict the pre-ceding residue into the extended region of theRamachandran plot due to the cyclic nature of thePro side-chain (Flory, 1988). In agreement with thisthe statistical analysis of the protein databaseshows that whenever there if a Pro residue the

probability of the preceding residue to be inextended angles is 2.5 times larger than expectedfrom a random distribution. Moreover, when a Proresidue is at the C-cap position it cannot contributea hydrogen bond to residue i ÿ 4. As a result Ala,His and/or Phe residues in the helical confor-mation followed by a Pro at position C-cap, havean extra energy term corresponding to the lack of ahydrogen bond and the higher cost in ®xing themin helical angles (see Materials and Methods). Infact as we mentioned above and as the NMR anal-ysis indicates, in all the peptides containing a Proresidue at position 16 the helical conformationstops at Lys14. Under these considerations, the C-cap contribution of Phe and His residues withrespect to Ala, when residue C0 is a Pro, is morefavourable than when residue C0 is a Gly (Table 4).

pH titration

It has been previously described that the pre-sence of a His residue close to the C terminus of apeptide stabilises the helical conformation when itis charged (Armstrong & Baldwin, 1993). Figure 6shows the pH titration for peptides C-AHG and

Figure 6. pH titration of peptides C-AHP and C-AHG.In this Figure we plot the helical content, determinedfrom the ellipticity at 222 nm, versus the pH. Filledtriangles, Peptide C-AHG. Filled circles, Peptide C-AHP.


C-AHP. The helical content of the peptidesincreases when going from PH 3 to PH 5, due tothe titration of Glu6. The side-chain of Glu6 in thecapping box conformation if quite close to the Nterminus of the helix and therefore it will have afavourable interaction with the helix dipole whenit is charged. Also, there will be a favourable elec-trostatic interaction with Arg10. However, fromPH 5 to pH 9, the helix decreases as expected if theHis side-chain is having a favourable interactionwith the helix dipole at the C terminus of the helix.This occurs for both peptides, although in the caseof the C-AHP peptide the decrease in helical con-tent when the His residue is neutral is slightly lar-ger, suggesting that the interaction with the helixdipole is slightly stronger than in peptide C-AHG.

Discussion

In proteins there are three factors that coulddetermine the statistically favourable presence of aparticular residue at a certain position of a second-ary structure element: (i) The residue could stabil-ise the particular secondary structure element. (ii)It could prevent its elongation due to its poor ten-dency to adopt that particular conformation. (iii)Because tertiary structure considerations. The ®rsttwo features can be associated like in the case ofN-capping by a Ser residue (where Ser is a badhelix former and a very good N-capping residue).At the N terminus of an a-helix, the presence inphase of a hydrophobic staple and a capping boxmotif ®ngerprint, stabilises strongly the helical con-formation and it is rarely found in the middle ofan a-helix (MunÄ oz et al., 1995). In proteins, the pre-sence of a Gly residue alone, or in the context ofthe Schellman sequence ®ngerprint (Aurora et al.,1995), does not guarantee to the same extent, that

the helix will be terminated before the Gly(Viguera & Serrano, 1995a).

The presence of a Pro residue in a sequence pro-duces a more drastic helix destabilization than aGly (O'Neil & DeGrado, 1990; Lyu et al., 1990;Horovitz et al., 1992; Blaber et al., 1993;Chakrabartty et al., 1994; MunÄ oz & Serrano, 1995a,b), and therefore the probability of the helix termi-nating one or two residues before this residue ismuch larger. Due to steric constraints the residuepreceding a Pro tends to adopt extended f and cdihedral angles. This together with the fact thatPro cannot donate a proton to make a main-chain/main-chain hydrogen bond, prevents its appear-ance as a C-cap residue in a-helices (Table 1).However, Pro is quite often found at position C0 ofa-helices (Richardson & Richardson, 1988). A Proresidue could be selected at C0 to prevent helixelongation but also because it could have a favour-able interaction with the a-helix and/or positionthe preceding residue in a favourable confor-mation. Our analysis of the protein databaseshows in fact that certain combinations of X-Propairs, with Pro as the C0 residue, are more abun-dant than expected in the protein database.

Structural analysis of all proteins in the databasecontaining the above pairs shows that in the caseof Asn, Cys, Phe, Tyr, Trp and His, the dihedralangles are relatively ®xed, In those cases the main-chain amino group of Pro is very close to the lastthree main-chain carbonyls of the a-helix. In thecase of aromatic and His-Pro pairs, that are theones analysed here in more detail, the edge of theirside-chain rings are close to the carbonyl of residuei ÿ 4 and one of the two side-chain N atoms of Hisis at hydrogen-bond distance of this main-chaingroup. For His this side-chain orientation clearlyargues in favour of an electrostatic interaction withthe helix dipole as well as the carbonyl group ofresidue i ÿ 4. For the Phe side-chain a similarexplanation can be produced based on theoreticalcalculations (Bodner et al., 1980; Nemethy &Scheraga, 1981) and for the fact that in the veryfavourable interaction of an aromatic ring and asulphur containing molecule (i.e. Met and Cys), theedge of the side-chain interacts with the sulphuratom (Viguera & Serrano, 1995b; Stapley et al.,1995). The most interesting result is the ®xed orien-tation of the main-chain amino nitrogen atom ofPro, with respect to the carbonyl groups of resi-dues i ÿ 2, i ÿ 3 and i ÿ 4. The main-chain nitrogenatom of Pro cannot make a hydrogen bond, how-ever, this N group has a partial positive chargeand it can establish favourable dipolar interactionswith the C-terminal helix dipole. These interactionscannot be very strong since in the Ala-Pro pair theside-chain of Pro rarely makes these contacts.Therefore, it seems that there could be a coopera-tive effect involving the interaction of residue C-cap with the helix as well as the dipole-dipoleinteraction of the Pro amino group at position C0with the helix macrodipole.


In the case of the Schellman motif we found thatthe statistical preferences in the protein databasewere not translated in strong helix stabilisingeffects (Viguera & Serrano, 1995a). In the case ofthe Pro-capping motif this does not seem the casesince the experimental analysis of poly-alaninebased peptides shows clearly that the pairs His-Proand Phe-Pro are helix stabilising in comparisonwith the Ala-Pro pairs. Moreover, the NMR anal-ysis shows that in the case of the His-Pro pair aconformation similar to that found in the databaseis present in aqueous solution, although probablyin equilibrium with other conformations. In thecase of the Phe-Pro pair, we have found less evi-dence for this conformation (only one NOE),although we have seen by CD that it contributesigni®cantly to helix stability with respect to otherPhe-X pairs. We must consider that, although Pheseems to be favourable as a C-cap residue whenPro is at position C0, His at the C-cap positioncould contribute up to ÿ0.6 kcal/mol to helix stab-ility when it is charged MunÄ oz & Serrano, 1995b).Therefore, the C terminus of the a-helix with Hisshould be less frayed and the helix content larger,than when there is a Phe residue. This could meanthat the weak NOEs detected in the His-Pro pair,are no longer detected in peptide C-AFP. A similarsituation was found when analysing the formationof the Schellman motif in peptides with low helicalcontent, in water and TFE (Viguera & Serrano,1995a). In the structural analysis of the N and Ctermini in a peptide series, Kallenbach and co-workers (Lyu et al., 1993), analysed several pep-tides having different X-Pro pairs (Asn-Pro, Gln-Pro, Ala-Pro, Ser-Pro, Thr-Pro and Gly-Pro) at theC terminus. They found that the pair Asn-Pro wasthe most favourable as expected from the statisticalanalysis shown in Table 1, although they did notprovide any reasoning behind those preferences.More interestingly the NMR analysis of the peptidecontaining the sequence Asn-Pro, exhibited NOEswhich were compatible with the Asn side-chainmaking a hydrogen bond back to the carbonylgroup of residue i ÿ 4 (Zhou et al., 1994), as wehave found for the His-Pro peptide. Therefore, itseems that the statistical preferences found in theprotein database roughly correlate with the ther-modynamic preferences found in model peptidesas it occurs with the capping box and hydrophobicstaple motifs.

In the above discussion there is an apparent con-tradiction. If the C-capping contribution of His andPhe when there is a Pro at C0 is larger than whenthere is a Gly, why is the helical content of the cor-responding Pro-peptides smaller than in the Gly-peptides? The reason is that with a Gly at positionC-cap the residue before it can be helical. In thecase of Pro at C-cap, the energy cost of ®xing thepreceding residue in helical angles is much largerand therefore the corresponding segments will con-tribute little to the partition function. In fact that iswhat we see from the NMR analysis of corre-sponding peptides, the residue before the Pro resi-

due is not signi®cantly adopting a helicalconformation. However, if now we look at a par-ticular helix segment having His at position C-capand Gly at position C0 and compare it with thesame segment having His at C-cap and Pro at C0,the stability of this helix segment will be larger inthe second case. This is a very important distinc-tion. To compare peptides and proteins we need tolook at the free energy of a particular sequence cor-responding to the length of the protein helix.Therefore, we can conclude that Pro not only actsas a strong helix stop signal, but in combinationwith certain residues it could stabilise a helical con-formation while determining the way in which thepolypeptide chain exits the helix.

Materials and Methods

Peptide synthesis

The peptides were synthesised by the EMBL peptidesynthesis service using Fmoc chemistry and PyBOP acti-vation at a 0.025 mmol scale. Peptide homogeneity andidentity were analysed by analytical high performanceliquid chromatography, amino acid analysis and matrix-assisted laser desorption time-of-¯ight mass spec-troscopy. The concentrations of the peptide sampleswere determined by ultraviolet absorbance (Gill & vonHippel, 1989) and amino acid analysis.

Far-UV circular dichroism (CD) spectroscopy

CD spectra were recorded on a Jasco-710 dictographcalibrated with (1S)-(�)-10-camphorsulphonic acid. CDspectra were obtained in the continuous mode by takingpoint measurements every 0.1 nm, with 100 nm/minscan rate, a response of one second and a 1 nm bandwidth. Twenty consecutive scans were averaged. Cellswith path lengths of 0.01 cm and 0.5 cm were used forthe analysis of the samples. The peptides were analysedat 4�C and pH 3. The pH was adjusted with HCl.

Peptide aggregation

To determine if the peptides aggregated, we recordedfor-UV CD spectra in the 5 mM to 1 mM concentrationrange at pH 3. At pH 3 we did not ®nd any concen-tration dependence of the mean residue ellipticity in pep-tides: C-AAG, C-AAF, C-AHP, C-AFP and C-AHG.However, at pH 3 between 20 m M and 1 mM peptideconcentration there was a clear difference in the shape ofthe CD spectra of peptide C-AAP. In this peptide weexplored the concentration range between 5 and 50 mM(5, 10, 20 and 50 mM). The spectra between 5 and 20 mMremained the same (data not shown). Therefore, we usedthe CD data obtained at 20 mM. Comparison of NMRmonodimensional spectra obtained at pH 3 and peptideconcentrations of 0.1 and 2.5 mM, did not show anydifference in the chemical shift values, or in the linewidth in any of the peptides.

Quantification of the helical population by CD

The helical content of the peptides was determined bydifferent methods. The ®st one is based on the empiricalequation developed by Chen et al. (1974):


% helical content � 100�Ellipticityobs222=

�ÿ39;500� �1ÿ 2:57=n�� 1�where n is the number of peptide bonds and Ellipticityobs

222

is the mean residue ellipticity of the peptides at 222 nm.In the second method the helical content was estimatedfrom the ellipticity at 193 nm using equation:

% helix � 14:769� 0:0010261� Ellipticityobs193 �2�

where Ellipticityobs193 is the mean residue ellipticity at

193 nm. This equation was obtained by correlating theellipticity changes at 193 nm with the helical contentsobtained from the method of Chen et al. (1974) for thepeptides without Phe. Using this equation it is possibleto obtain the corrected helical contents for the aromaticcontribution in the peptides with Phe (Viguera &Serrano, 1995b). Since the ellipticity at 222 nm is extre-mely dependent on the accuracy in the determination ofthe peptide concentration, we have used two other cri-teria that are independent of peptide concentration. Oneis the position of the ellipticity minimum and the other isthe ratio R1 (Bruch et al., 1991). The R1 ratio is obtainedby dividing the ellipticity at 193 nm by the ellipticityminimum in the range 200 to 210 nm.

Nuclear magnetic resonance

NMR samples were prepared by dissolving the lyo-philised peptides in 0.5 ml of 2H2O with 10% 2H2O (byvol.), using milli Q water from a Millipore water systemand 2H2O from Cambridge Isotope Chem. At a concen-tration of around 2.5 mM and 0.1 mM. Minute amountsof HCl and NaOH were added in order to adjust the pHof the samples to 3; this was measured with an Ingoldcombination electrode (Wilmad) inside the NMR tubeand isotope effects were not corrected. Sodium 3-tri-methylsilyl (2,2,3,3,-2H4) propionate (TSP) was used asan internal reference at 0.00 ppm. NMR experimentswere performed an a Bruker AMX-500 spectrometer at281 K. The data were processed with the programUXNMR from Bruker in an Aspect X32 computer. DOF-COSY (Piantini et al., 1982), NOESY (Kumar et al., 1980),and ROESY (Bothner-By et al., 1984) spectra wereacquired using standard procedures. NOESY andROESY spectra were routinely recorded for every pep-tide and the spectra were jointly analysed to discard arti-factual NOEs as those arising from spin diffusion. Themixing time used was 200 ms in the NOESY spectra.TOCSY (Rance, 1987) spectra acquired using the stan-dard MLEV17 spin lock sequence and z-®lter with 80 msmixing time. The spectral width was 5555.55 Hz and thewater signal was presaturated during the relaxationdelay (one second) and also during the mixing time ofNOESY spectra. The size of the acquisition data matrixwas 2048 � 512 words in f2 and f1, respectively. Eightscans were acquired for the COSY and TOCSY exper-iment and 64 for the NOESY and ROESY. Fourier trans-formation of the two-dimensional data matrix wasmultiplied by a phase-shifted sine-bell or square-sine-bellwindow function in both dimensions. The correspondingshift was optimised in every experiment. Cross-peakintensities were evaluated by visual inspection of thecontour levels. The conformational shifts (conformation-dependent chemical shifts dispersion) of the Ca protonswere obtained by subtracting the random coil values(Merutka et al., 1995) from the measured one for eachresidue.

1H NMR assignment

Complete assignment of the 1H NMR spectra ofeach peptide in aqueous solution was made using thestandard two-dimensional sequence speci®c methods(WuÈ thrich, 1986). Firstly, spin systems were identi®ed byjoint analysis of phase-sensitive COSY and TOCSY spec-tra. The second stage involves the assignment of thesespin systems to speci®c residues in the peptide sequenceon the basis of sequential bN, aN or/and NN spatialconnectivities observed in the NOESY spectra. The tableswith the assignment of the peptides are available uponrequest.

Protein database search

The database of 3-D structures has been obtained fol-lowing the principles described by Hobohm et al. (1992).This database has been ®ltered for quality of the dataand consists of 279 proteins with less than 50% sequencehomology for a total of 59,117 amino acidic residues (thelist of proteins is shown in MunÄ oz & Serrano, 1995d),and it is currently included in the program WHATIF(Vriend, 1990). The search was conducted using theScan3D option and the search query was either H/H/H/STC/STC or */*/*/*/*, where H is helix, S is strand,C is coil, T is turn and * is any type of secondary struc-ture.

Calculation of free energies of interactionwith AGADIRls

The change in free energy for a-helix formation uponmutation of each a-helix cannot be precisely calculatedusing a standard two-state model. A more precise esti-mation can be obtained by ®tting a helix/coil transitionalgorithm to the changes in helical content detected byfar-UV CD (MunÄoz & Serrano, 1995b, c). The differencein free energy for a-helix formation (�G) between thepseudo-wild-type and mutated peptides was calculatedfrom their differences in helical content as measured byfar-UV CD, after aromatic correction. The modi®ed ver-sion of the helix/coil transition algorithm AGADIRls;MunÄ oz & Serrano, 1997), was used to ®t the far-UV CDdata of the peptides, to a 1% difference of the experimen-tal value. AGADIR1s overpredicts the helical content ofthe control peptides (C-AAG; 51% predicted versus 40%;C-AHG 48% versus 40% and C-AFG, 42% versus 37%).Therefore, before ®tting the peptides containing thedifferent mutations, we proceeded to modify some of theparameters in the algorithm in order to be able to repro-duce the experimental data. We found that the best com-bination required a decrease in the helical intrinsicpropensities for the different amino acids (free energycost required to ®x an amino acid in helical angles), of0.06 kcal/mol and an increase in the C-capping contri-bution of Gly (from ÿ0.4 to ÿ0.7 kcal/mol). The rest ofthe parameters: capping effects, side-chain/side-chaininteractions within the helix, and electrostatic inter-actions with the helix dipole as well as between chargedresidues, were the same as previously described MunÄ oz& Serrano, 1995b,c). Once we could reproduce the helicalcontent of the control peptides we proceeded to deter-mine the contribution of the X-Pro pairs to helix stability.The way we did it was to consider that residue X couldnot adopt the helical conformation (we have evidencefrom the NMR analysis that this is the case) and then wemodi®ed the C-capping contributions of Ala, Phe andHis to reproduce the experimental data. The error in the


estimations was obtained by determining the C-cappingcontribution that will give a helical content 2% higherthan the experimental one. The difference between thetwo C-capping values is the error in the energy esti-mation.

Acknowledgements

J.P. was a fellowship holder of the Human Capitaland Mobility programme. We are grateful to Dr F.Blanco for critical reading of the manuscript. A newversion of AGADIR1s (AGADIR2s), containing thehydrophobic staple, Schellman and Pro-capping motifsis available on the WWW (http://www.embl-heidelberg.de/Services/index.html#5).

References

Armstrong, K. M. & Baldwin, R. L. (1993). Charged his-tidine affects a-helix stability at all positions in thehelix by interacting with the backbone charges.Proc. Natl Acad. Sci. USA, 90, 11337±11340.

Aurora, R., Srinivasan, R. & Rose, G. D. (1995). Rulesfor a-helix termination by glycine. Science, 264,1126±1129.

Blaber, M., Zang, X. & Mathews, B. (1993). Structuralbasis of amino acid a-helix propensity. Science, 260,1637±1640.

Blanco, F. J. & Serrano, L. (1995). Folding of protein GB1 domain studied by the conformational character-ization of fragments comprising its secondary struc-ture elements. Eur. J. Biochem. 230, 634±649.

Bodner, B. L., Jackman, L. M. & Morgan, R. S. (1980).NMR study of 1:1 complexes between divalent sul-fur and aromatic compounds: a model for inter-actions in globular proteins. Biochem. Biophys. Res.Commun. 94, 807±813.

Bothner-By, A. A., Stephens, R. L., Lee, J. M., Warren,C. D. & Jeanloz, R. W. (1984). Structure determi-nation of a tetrasacharide: transient nuclear Over-hauser effect in the rotating frame. J. Am. Chem. Soc.106, 811±813.

Bruch, M. D., Dhingra, M. M. & Gierasch, L. M. (1991).Side chain backbone hydrogen bonding contributesto helix stability in peptides derived from an a-heli-cal region of carboxypeptidase A. Proteins: Struct.Funct. Genet. 10, 130±139.

Chakrabartty, A., Kortemme, T., Padmanabhan, S. &Baldwin, R. L. (1993). Aromatic side-chain contri-bution of far-ultraviolet circular dichroism of helicalpeptides and its effect on measurement of helixpropensities. Biochemistry, 32, 5560±5565.

Chakrabartty, A., Kortemme, T. & Baldwin, R. L. (1994).Helix propensities of the amino-acids measured inalanine-based peptides without helix-stabilizingside chain interactions. Protein Sci. 3, 843±852.

Chen, Y. H., Yang, J. T. & Chow, K. H. (1974). Determi-nation of the helix and b form of proteins inaqueous solution by Circular dichroism. Biochem-istry, 13, 3350±3359.

Dasgupta, S. & Bell, J. A. (1993). Design of helix ends.Int. J. Pept. Res. 41, 499±511.

Doig, A. J. & Baldwin, R. L. (1995). N- and C-cappingpreferences fro all 20 amino acids in a-helicalpeptides. Protein Sci. 4, 1325±1336.

Flory, P. J. (1988). Statistical Mechanics of Chain Molecules,Oxford University Press.

Gill, S. C. & Hippel, P. H. (1989). Calculation of proteinextinction coef®cients from amino acid sequencedata. Anal. Biochem. 182, 319±326.

Gong, Y., Zhou, H. X., Guo, M. & Kallenbach, N. R.(1995). Structural analysis of the N- and C-terminiin a peptide with consensus sequence. Protein Sci. 4,1446±1456.

Harper, E. T. & Rose, G. D. (1993). Helix stop signals inproteins and peptides: the capping box. Biochemis-try, 32, 7605±7609.

Hobohm, U., Scharf, M., Schneider, R. & Sander, C.(1992). Selection of representative protein data sets.Protein Sci. 1, 409±417.

Horovitz, A., Matthews, J. & Fersht, A. R. (1992). a-helixstability in proteins. II. Factors that in¯uence stab-ility at an internal position. J. Mol. Biol. 227, 560±568.

Kabsch, W. & Sander, C. (1983). Dictionary of proteinsecondary structure: pattern of recognition ofhydrogen-bonded and geometrical features. Biopoly-mers, 22, 2577.

Kumar, A., Ernst, R. R. & WuÈ thrich, K. (1980). A two-dimensional nuclear Overhauser enhancement (2DNOE) experiment for the elucidation of completeproton-proton cross relaxation networks in biologi-cal macromolecules. Biochem. Biophys. Res. Commun.95, 1±6.

Lyu, P. C., Liff, M. I., Marky, L. A. & Kallenbach, N. R.(1990). Side chain contributions to the stability ofalpha-helical structures in peptides. Science, 250,669±673.

Lyu, P. C., Zhou, H. X., Jelveh, N., Wemmer, D. E. &Kallenbach, N. R. (1992). Position-dependent stabi-lizing effect in a-helices: N-terminal capping in syn-thetic model peptides. J. Am. Chem. Soc. 114, 6560±6562.

Lyu, P. C., Wemmer, D. E., Hongxing, X. Z., Pinker,R. J. & Kallenbach, N. R. (1993). Capping inter-actions in isolated a-helices: Position-dependentsubstitution effects and structure of a Serine-cappedpeptide helix. Biochemistry, 32, 421±425.

Merutka, G., Dyson, H. J. & Wright, P. E. (1995). `Ran-dom coil' 1H chemical shifts obtained as a functionof temperature and tri¯uoroethanol concentrationfor the peptide series GGXGG. J. Biomol. NMR, 5,14±24.

MunÄ oz, V. & Serrano, L. (1994). Elucidating the foldingproblem of helical peptides using empiricalparameters. Nature: Struct. Biol. 1, 399±409.

MunÄ oz, V. & Serrano, L. (1995a). Analysis of i,i � 5 andi,i � 8 hydrophobic interactions in a helical modelpeptide bearing the hydrophobic staple motif. Bio-chemistry, 34, 15301±15306.

MunÄ oz, V. & Serrano, L. (1995b). Elucidating the foldingproblem of helical peptides using empirical par-ameters. II: Helix macrodipole effects and rationalmodi®cation of the helical content of naturalpeptides. J. Mol. Biol. 245, 275±296.

MunÄ oz, V. & Serrano, L. (1995c). Elucidating the foldingproblem of helical peptides using empirical par-ameters. III: Temperature and pH dependence.J. Mol. Biol. 245, 297±308.

MunÄ oz, V. & Serrano, L. (1995d). Intrinsic secondarystructure propensities of the amino acids, using stat-istical f and c matrices: comparison with the exper-imental scales. Proteins: Struct. Funct. Genet. 20,301±311.

http://www.emblheidelberg.de/Services/index.html#5

http://www.emblheidelberg.de/Services/index.html#5


MunÄ oz, V. & Serrano, L. (1997). Development of themultiple sequence approximation within the AGA-DIR model of a-helix formation. Comparison withZimm-Bragg and Lifson-Roig formalisms. Biopoly-mers, 41, 495±509.

MunÄ oz, V., Blanco, F. & Serrano, L. (1995). The hydro-phobic-staple motif and a role for loop-residues ina-helix stability and protein folding. Nature Struct.Biol. 2, 380±385.

Nemethy, G. & Scheraga, H. A. (1981). Strong inter-action between disul®de derivatives and aromaticgroups in peptides and proteins. Biochem. Biophys.Res. Commun. 94, 807±813.

O'Neil, K. T. & DeGrado, W. F. (1990). A thermodyn-amic scale for the helix-forming tendencies of thecommonly occurring amino acids. Science, 250, 646±650.

Padmanaban, S. & Baldwin, R. L. (1994a). Helix-stabiliz-ing interaction between Tyr and Leu of Val whenthe spacing is I,I � 4. J. Mol. Biol. 241, 706±713.

Piantini, U., Sùrensen, O. W. & Ernst, R. R. (1982). Mul-tiple quantum ®lters for elucidating NMR couplingnetworks. J. Am. Chem. Soc. 104, 6800±6801.

Rance, M. (1987). Improved techniques for homonuclearrotating-frame and isotropic mixing experiments.J. Magn. Reson. 74, 557±564.

Richardson, J. S. & Richardson, D. C. (1988). Amino acidpreferences for speci®c locations at the ends ofa-helices. Science, 240, 1648±1652.

Schellman, C. (1980). In Protein Folding (Jaenicke, R. ed.),Elsevier/North-Holland, New York.

Seale, J. W., Srinivasan, R. & Rose, G. D. (1994).Sequence determinants of the capping-box, a stabi-

lizing motif at the N-termini of a-helices. ProteinSci. 3, 1741±1745.

Shoemaker, K. R., Fairman, R., Schultz, D. A.,Robertson, A. D., York, E. J., Stewart, J. M. &Baldwin, R. L. (1990). Side-chain interactions inthe C-peptide helix: Phe8 . . . His12�. Biopolymers, 29,1±11.

Stapley, B. J., Rohl, C. A. & Doig, A. J. (1995). Additionof side chain interactions to modi®ed Lifson-Roigtheory application of energetics of phenylalanine-methionine interactions. Protein Sci. 4, 2383±2391.

Viguera, A. R. & Serrano, L. (1995a). Experimental anal-ysis of the Schellman motif. J. Mol. Biol. 251, 150±160.

Viguera, A. R. & Serrano, L. (1995b). Side-chain inter-actions between sulphur containing amino acidsand phenyalanine, in a-helices. Biochemistry, 34,8771±8779.

Vriend, G. (1990). WHATIF: a molecular modeling anddrug design program. Mol. Graph. 8, 52.

Wishart, D. S., Bigam, C. G., Holm, A., Hodges, R. S. &Sykes, B. D. (1995). H, C and N random coil NMRchemical shifts of the common amino acids. I.Investigations of nearest neighbor effects. J. Biomol.NMR, 5, 67±81.

WuÈ thrich, K. (1986). NMR of Proteins and Nucleic Acids,John Wiley and Sons, New York.

Zhou, H. X., Lyu, P. C., Wemmer, D. E. & Kallenbach,N. R. (1994). Structure of a C-terminal a-helix Capin a synthetic peptide. J. Am. Chem. Soc. 116, 1139±1140.

Edited by J. Thornton

(Received 21 May 1997; received in revised form 29 July 1997; accepted 30 July 1997)

C-capping and helix stability: the pro C-capping motif

Documents

Transcript of C-capping and helix stability: the pro C-capping motif