plMi-

ashi H

ada T,9

I l76,

6488-

") Arch

II R J

,799-

Indian Journal of Biochemistry and Biophysics

Vol. 4l, October 2004, pp. 233-240

Characterization of non-planar peptide groups in protein crystal structuresAnjan Kumar Dasgupta 1, Rabi Majumdar" and Dhananjay Bhattacharyya'<

'Department of Biochemistry, University of Calcutta, 35 Ballygunge Circular Road, Kolkata 700019

2Department of Biophysics, Molecular Biology and Genetics, University of Calcutta92 Acharya Prafulla Chandra Road, Kolkata 700009

3Biophysics Division, Saha Institute of Nuclear Physics, 11AF Bidhan Nagar, Kolkata 700064

Received 1 January 2004; revised 4 August 2004Peptide groups are generally assumed to be planar in protein structure, due to 'rigid' partial double bond character of

peptide bonds, thus the value of peptide torsion angle ro should be restricted to 1800 for the usual/rans form of peptide unit.However, on analyzing the ultra-high resolution protein crystal database, we find that in some cases, ro deviates> l O? fromits usual value of 1800, indicating significant non-planarity of peptide groups. Moreover, the non-planarity for most of theamino acids is found to be 'biased' towards values of co smaller than 1800• Similar trend for ro is confirmed by the neutrondiffraction data for proteins. The neutron diffraction database also reveals that non-planar peptide groups are generallycorrelated to 'pyramidal' structure of the peptide-nitrogen bonds. Consequently, the hydrogen atom of peptide groupdeviates from its planar position, as measured by the 'improper' torsion angle 8. Thus, we find that both the angles (j) and 8point towards a significant amount of non-planarity of peptide groups, which cannot be ignored. The role of peptide non-planarity in protein function is, however, not yet clear.

Keywords: X-ray diffraction, database analysis, neutron diffraction, non-planar peptide, peptide bond, protein crystalstructure.

IPC Code: C 07 K

Protein structure has enormous flexibility due to thepossibility of alteration of the two main chain torsionangles, and ur, as well as the side chain torsionangles X in some cases. This gives rise to its spatialstereo structure and folding, directed by its aminoacid sequence'. Peptide units are believed to be theplanar components of a protein, as a result of thepartial double bond character of peptide bonds. Thisimparts rigidity to the peptide torsion angle 00, thevalue of which is restricted to 1800 for the usual transpeptide unit having staggered geometry. Thishypothesis, originally put forward by Pauling, wasfirst questioned by Ramachandran et al.2 on the basisof semi-empirical quantum chemical calculations.Experimental evidence of non-planar amide groups incyclic peptides also established a similar picture.".Also, in folded proteins, cis peptide units are oftenfound to exist, indicating the possibility of cis-trans

*Author for correspondenceE-mail: bhatta@biop.saha.ernet.inFax: +91-33-2337-4637Tel: +91-33-2337-0379

isomerization for peptide units under folding strain. Itwas first shown experimentally by Watenpaugh et al.9

that 00 in a protein structure can have values differentfrom 180° and it is biased towards values lower than180°. This was later extensively studied usingstructural database analysis!", as well as using abinitio quantum chemical calculations 11. These studiesindicate that the peptide groups are generally non-planar having 00 closer to 175° rather than 185°, andhave recently been reviewed12.

The literature on protein structure analysis,however, does not indicate any cause or effect of suchnon-planar peptides'", It is not even clear whetherthese non-planarities are due to errors in the atomicpositions of the crystal structures. We have, therefore,studied the ultra-high resolution crystal structuresobtained from protein data bank (PDB)13, for severalunrelated proteins to find out if the torsion angle windeed shows a systematic and asymmetric deviationfrom 180°. We have further analyzed all the availableprotein crystal structures solved by using neutrondiffraction data to confirm our results for the peptidetorsion angle 00. In addition, the neutron diffraction

234 INDIAN J. BIOCHEM. BIOPHYS., VOL4I, OCTOBER 2004

database helped us to determine the improper'dihedral angle e (value of which is also 1800 for aplanar peptide unit), as a measure of pyramidalcharacter of the amide nitrogen atom.

MethodIn analyzing ultra-high resolution crystal

structures obtained from protein data bank, aresolution cut-off value of 1 A was first attempted,which gives only 82 structures. In order to obtain areasonably large number of such structures forstatistical analysis, the resolution cut-off criterion wasfinally relaxed to 1.2 A, which gives 236 structures,according to June 2002 update, some of theseconsisting of multiple polypeptide chains. Takingthese multiple chains into account and also removingthe much smaller peptide chains, such as thoseconsisting of 20 residues or less, 244 chains werefinally analyzed altogether. The details are availablein hUp:llwww.saha.ernet.in!cmb_page/htmllNonPlanarPeptides.html.

Only 10 protein crystal structures are currentlyavailable in the protein data bank solved on the basisof neutron diffraction method, a relatively newtechnique in structural biologyl3. In view of such asmall number of protein structures solved by theneutron diffraction technique, so far, we analyzed allthe available structures, without any resolution cut-off. The peptide non-planarities are evaluated fromtorsion angles 00,while the resulting partial pyramidal

Torsion Angle co

s:Pl Ga·~~ I, --G"c~~Ni~\1 P2

H

Improper Angie 9

Fig. l-Schematic representation of a peptide unit describingproper torsion angle (J) and the improper torsion angle on the

nitrogen atom

characters of peptide nitrogen bonds are evaluated interms of improper dihedral angles e, defined by(C'(i-l)--Ca(i)--N(i)--HN(i» as illustrated in Fig. 1.

The software STRIDE'4 was used to analyzehelical and the other secondary structures of eachamino acid in different protein chains. The lessfrequently observed structures, such as the 310helixand some others, were not considered for furtheranalysis. The propensity values shown in Table 1were calculated, following Chou and Fasman 15 byusing the following expression:

N( 00)straiNaaPaa(oo)=----- ...(1)

where N(W)straa'is the number of occurrence of anamino acid 'aa' with large Wdeviation (either 00190°) in a given secondary structure, Nstraa thetotal number of occurrence of the same amino acid inthe same secondary structure, Naa the total number ofoccurrence of the same amino acid in the entiredatabase, and N the total number of all amino acidsoccurring in the database. Thus, the parametersPaaC170-)and Paa(i90+) represent the propensity of anamino acid 'aa' to satisfy our unusual torsion anglecriteria, namely w < 170° and co> 190°, respectively.

All the four different moments of torsion angle00were calculated, following standard statisticaldescription" for large degrees of freedom. Thedimensionless parameters for testing of skew andkurtosis were computed by using the followingexpressions:

...(2)

and

I( OOi -

Table I-Number of Occurrence and propensity values (within parenthesis for each secondary structure type) of different amino acids indifferent secondary structures for adopting non-planar peptide geometry with peptide torsion angle (a) smaller than 170", (b) larger than 190"

(a) Paa(170·)Occurrence

Occurrence and propensity inAmino acid with W

236 INDIAN 1. BIOCHEM. BIOPHYS., VOL 41, OCTOBER 2004

Results and Discussion'Biased' peptide torsion angle deviations in protein structures

In analyzing protein structures on the basis ofultra-high resolution crystal database, we have foundthat a sizable number of amino acids develop 'large'non-planarity, due to peptide torsion angle deviation~w = Iw-180ol = 10° or more (which we call 'large'deviations). This is irrespective of the protein size, itsamino acid content or its secondary structure, asshown in supplementary material available on-line fordetails (http://www.saha.emet.in/cmb_page/htrnl/NonPlanarPeptides.html), which also lists the number ofhighly non-planar pep tides in each of the proteinstrands and the parameters for testing skew andkurtosis. We have calculated the correlationcoefficient between the gl parameter (for testingskew) and the amount of non-planarity bias[N(w1900)] for all 244 proteins chains,as shown in Fig. 2. It is found to be -0.14. This valueof the correlation coefficient is about 98% significantfor 244 degrees of freedom, indicating that the twovariables are highly correlated.

The number of occurrence and the propensity ofeach amino acid with such 'large' deviations in w asparts of proteins having different secondary structuresare shown in Table 1. The results clearly indicate thatthere is a significantly larger number of amino acidswith co< 170° (2395 cases), than those with w>190°(1571 cases), out of a total number of 42764 peptidesin the database. This asymmetry or 'biased' deviationtowards smaller values of co is definitely non-randomand is specially highlighted in Fig. 3, where the

its theTable 2-Statistical analysis of variation of peptide torsion angle (co) for different amino acid residues at C-terminal end

ned in[Four different moments, mean, standard deviation, skew and kurtosis are calculated along with the corresponding dimensionless

es are quantities for skew and kurtosis values]shows

Amino Total Average Standard Skew, reflecting Kurtosis, reflecting Test of Skew Test of Kurtos-erage, acid occurrence co deviation asymmetry spread (gl) (g2)for all Ala 4140 179.60 5,10 -17,22 3468.90 -0.1294 2.1113inctive

Cys 891 179.45 5.58 -11.41 3111.23 -0.0656 0.2002of the

Asp 2611 179.22 6.06 -12.90 5331.67 -0.0578 0.9429

Glu 2299 179.33 5.15 -9.50 3040.91 -0.0696 1,3228I, the Phe 1553 179.47 6,56 -9,59 6382.21 -0.0340 0.4556I atom

Gly 3709 178.90 5.63 21.52 4365,94 0.1191 1.2741ith the

likelyHis 923 179.76 6.61 44.86 7666.05 0.1551 1.0064

sibility Ile 2216 178.62 6.07 -54.62 5378.33 -0.2439 0,9540

ith the Lys 2232 179.255.49 -5.23 4193.44 -0.0316 1.6126

acids.Leu 3117 179.17 5.25 -27,84 2873.25 -0.1923 0.7784

able to Met 758 179.29 4.83 3.27 1993.13 0.0290 0.6603

~ atom. Asn 2097 179,39 5,63 1.54 3623.64 0.0086 0.6192

Fids in Pro 1945 180.02 6.24 -57.07 7344.59 -0.2347 1.8390

ro, Trp GIn 1545 179.18 5.40 -9,94 3156,29 -0.0633 0.7214

lbit the Arg 1735 179.47 5.47 -1.67 3463.58 -0.0102 0.8640

~where Ser 2849 179.33 6.01 -8.12 5652.98 -0.03741.3345

value Thr 2802 178.74 6.29 -8.12 5652.98 -0.0374 1.3345

eptides Val 3018 178,78 5.61 -17.77 3973.29 -0.1005 1.0055

own In Trp 639 179.77 5.43 -4.16 2798.34 -0.0260 0.2238

fOur of Tyr 1595 180.02 6.69 40.20 6671.68 0.1345 0.3393

gard toTable 3-List of all the protein structures obtained from PDB and solved by neutron diffraction technique [(The positions of hydrogen atoms arealso determined in these structures, which allowed us to calculate the improper dihedral angles 8 at the nitrogen atomic positions. The refinement

and structural parameters are also given to highlight the relation between peptide torsion angles and improper torsion angles (8 )]

PDB ID Protein name No. of Resolution Refinement Cases with Cases with Correlatioramino protocol between coacids to < 170 w>190 8 190 and 9

lC57 Concanavalin A 237 2.4 A X-PLOR 0 0 0 0 0.99

ICQ2 Sperm whale myoglobin 153 2.oA X-PLOR 0 0 4 6 -0.47

IGKT Endothapepsin 329 2.1 A SHELXL 3 3 1 1 0.84

II05 Hen egg-white lysozyme 129 2.oA X-PLOR 1 0 35 43 -0.35

1LZN Hen egg-white lysozyme 129 1.7 A X-PLOR 7 5 2 1 0.93

INTP Serine protease 223 1.8 A Constrained 13 5 1 2 0.86difference fourier

2MB5 Carbonmonoxy myoglobin 153 1.8 A PROLSQ 0 0 0 0 0.7

7209 3INS* Zn-insulin 21 2.2A PROLSQ 0 0 1 0 0.6230 0 1 0 3 0.09

protein 21 0 0 1 0 0.56

ar and co 30 0 0 1 0 0.58nature of 5RSA Ribonuclease A 124 2.oA PROLSQ 0 1 2 4 0.831800 and 6RSA Ribonuclease 124 2.oA PROLSQ 0 1 3 4 0.82

equencies*This protein has four chains and their structural properties are given in separate lines.

DASGUPTA et al.: NON-PLANAR PEPTIDE GROUPS IN PROTEIN CRYSTAL STRUCTURES 237

238 INDIAN J. BIOCHEM. BlOPHYS., VOL 41, OCTOBER 2004

A comparison of Tables 1 and 2 indicates thatGly and Pro residues have no preference for directionof non-planarity as their skewness are of oppositesense from the two measures, which is expected. Thelargest residue viz., Trp also has no preference,possibly due to its bulky size, which may lead toseveral unusual contacts with rest of the protein. Theside chain of His most probably forms severalimportant functional contacts, such as those withheme group, and may have unusual strains. It is notclear why Tyr prefers 0) >190°, but this may be due toboth polar and non-polar interactions of its side chainor some other yet unknown factor. There are,however, several other strains and steric factors in afolded protein, which may often dominate the localfactors.

Analysis of the propensity values in Table 1indicates that 0) 190°. Thisconfirms the asymmetry demonstrated in Fig. 3.

The fact that e and 0) are correlated is even bettershown in Fig. 6. This shows deviation of both theangles in few representative proteins. The pair-wisecorrelation coefficients for these angles in the case ofall protein structures are also listed in Table 3, whichindicate high degree of correlation for majority of thestructures. The correlation coefficient value of 0.28 is

40

~50

[,60

proteinlJSE;

globinsneutron~ctively.mctures,ar. The, where

tainedihown.

thesemailerout of

have. This

betterth the-wise

caseofwhichof the.28 is

DASGUPTA et al.: NON-PLANAR PEPTIDE GROUPS IN PROTEIN CRYSTAL STRUCTURES 239

200 A200

180

1600 20 40 60 80 100 120 140

1i:; 200a.

(b)0!....a.E.....-0 180c~s,\lJa.

~ 1600.. 0 50 100 150 200 250

200

180

160 L-~ __ -L__~ __L-~ __-L__~~o 20 40 60 80 100 120. 140 160

Residue number

Fig.6-Variation of peptide torsion angle co (in black) andimproper torsion angle (in green), describing pyramidal nature ofpeptide nitrogen atom are shown as functions of residue number forthree representative proteins solved from neutron diffraction data

over 99.9% significant for about 1700 data points.The overall correlation between the two variables(taking all proteins) is also demonstrated in Fig. 7.

ConclusionIt is seen from our results that the non-planarity

of peptide units is a significant feature of proteins andcannot be ignored. Most of the amino acids at C-terminal side of a peptide tend to induce a biased non-planarity to the peptide. Notable exceptions are Glyand Pro, possibly indicating that these non-planaritybias arise from steric clash between Cfl atom and lonepair electrons at nitrogen. Earlier, we reRorfed similarnon-planarity of amino groups of DNA2 ,21,where wefound that this improves the strength of the bifurcatedcross-strand hydrogen bonds across the successivebase pairs, thereby contributing significantly to therigidity of DNA molecule for some specific basesequences. Similarly, for proteins, the question iswhether the peptide non-planarity can facilitate

150

100 -

50 -

•

~ 0 ~~~~LY~.u~~~~~~~~cQ):J0-

~ 200 -.--------.-~--------.u,

180 182 184 186 188Deviattofls- in w

190 192

50

B

150 -

100 -

o180 183 186 189 192 195 198 201 204

Deviations in e

Fig. 5-Frequency distribution of (A): peptide torsion angle to;and (B): variations in peptide improper dihedral angle e in theprotein structures obtained from neutron diffraction technique

[The convention adopted in Fig. 3 is followed]

205 I

200

195

'" 190IIIenc185co

c0

"@ 180.9Q;

175a.ea.S 170

165

160

•

..,

•

•

• •

•• 1'r •,.-...- ..••••••• t; I•• _'Ii ~

• • ••I••.~. It'. I.,r. ,

.. ~. I•••155 ----,--.-.------.--------.~------r____--- ----------155 165 175 185 195 205

•• •

••

Peptide torsion angle co

Fig. 7-Correlation between the peptide torsion angles coandimproper torsion angles e

240 INDIAN 1. BIOCHEM. BIOPHYS., VOL 41, OCTOBER 2004

hydrogen bonding in DNA-protein recogrutionprocesses. The inadequate number of hydrogen bondsreported, so far (on the basis of planar peptide units),in the crystal structures of DNA-protein and DNA-drug complexes22-24 has led attempts to find outalternative mechanism for specific DNA-ligandrecognition, such as shape-complementarity, water-mediated interactions, etc.24-26. Consideration ofpyramidal character of peptide nitrogen bonds mayindicate importance of hydrogen bonding interactionas the dominant process in protein-DNA recognition.

Acknowledgement

We thank Ms Jhuma Das of the Dept. of Physics,University of Pune, India and Ms Ranjita GhoshMullick of the Dept. of Biochemistry, University ofCalcutta for computational assistance. One of us(RM) thanks the Council of Scientific and IndustrialResearh (CSIR), India for the grants and facilitiesunder the Emeritus Scientist Scheme.

References

Schultz G E & Schirmer R H (1979) Principles of ProteinStructure in Springer Advanced Text in Chemistry (Cantor CR, ed.) Springer Verlag, New York

2 Ramachandran G N, Laksminarayan A V & Kolaskar A S,(1976) Biochim Biophys Acta, 303, 8-13

3 Winkler F K & Dunitz 1 D (1975) Acta Cryst B 31, 268-2694 Winkler F K & Dunitz J D (1975) Acta Cryst B 31, 273-2755 Winkler F K & Dunitz J D (1975) Acta Cryst B 31,276-2786 Winkler F K & Dunitz J D (1975) Acta Cryst B 31, 278-281

7 Winkler F K & Dunitz 1 D (1975) Acta Cryst B 31, 286-2888 Dunitz 1 D & Winkler F K (1975) Acta Cryst B 31, 251-2639 Watenpaugh K D, Sieker L C & Jensen L H (1979) J Mol

Bioi, 131 509-53210 MacArthur M W & Thornton J M (1996) J Mol Bioi 264,

1180-119511 Head-Gordon T, Head-Gordon M, Frisch M J, Brooks III C

L & Pople J A (1991) JAm Chem Soc 113, 5989-599712 Edison A E (2001) Nature Struct Bioi 8, 201-20213 Berman H M, Westbrook J, Feng Z, Gilliland G, Bhat T M,

Weissig H, Shinbyalov N & Bourne P E (2000) Nucl AcidsRes 28, 235-242

14 Frishman D & Argos P (1995) Proteins 23, 566-57915 Chou P Y & Fasman G D (1974) Biochemistry 13,211-22216 Snedecor G W & Cochran W G (1989) Statistical Methods,

8th edn., Iowa State University Press, Ames, Iowa17 Shamovsky I L, Ross G M & Riopelle R J (2000) J Phys

Chem B 104,11296-1130718 Herzberg 0 & Moult J (1991) Proteins: Struct Funct Genet

11,223-22919 Merritt E A, Kuhn P, Sarfaty S, Erbe J L, Holmes R K & Hol

W G J (1998) J Mol Bioi 282, 1043-105920 Bhattacharyya D & Majumdar R (2001) Indian J Biochem

Biophys, 38,16-1921 Bhattacharyya D, Kundu S, Thakur A R & Majumdar R

(1999) J Biomol Struct Dynam 17,289-30022 Werner M H, Gronenborn A M & Clore G M (1996) Science

271,778-78423 Tabernero L, Bella J & Aleman C (1996) Nucl Acids Res 24,

3458-346624 Pjura P E, Grzeskowiak K & Dickerson R E (1987) J Mol

Bioi 197, 257-27125 Wemmer D (1998) Nature Struct Bioi 5, 169-17126 Janin J (1999) Structure 7, R277-R279