A peptide is usually drawn as one definite object: a helix, a ring, a chain folded a particular way. For a folded protein that picture is often close to the truth. For a short peptide in water it usually is not. The chain keeps changing shape, and the shape that matters biologically may appear only when the peptide meets a receptor, a membrane, or a constraint a chemist has built into it.
This page is about peptide conformation in that sense: what a short chain is doing in solution, where its secondary structure comes from, how chemists hold it in place, and how any of it is measured. The rigid peptide bond and the backbone angles that set the limits are covered on the peptide bond page; the four levels of protein structure are on the proteins and polypeptides page.
Does a peptide have one shape?
The honest answer for most short peptides is no, not one shape. In the words of a review of the field, most if not all short linear peptides exist in solution as an ensemble of rapidly interconverting conformations[5]. The chain is not frozen and not random in the sense of having no preferences; it visits many conformations, and some are visited more often than others.
That has a practical consequence for how results are reported. NMR measurements are averaged over the conformations a peptide samples, weighted by how much time it spends in each[5]. A structural model built from those measurements therefore has to be read with care, because a single model need not represent the whole ensemble. Calculating one structure is meaningful when a peptide really does adopt one dominant conformation, for example when a disulfide bond or a bound metal ion restricts it[5].
The useful way to think about it is populations rather than a single answer. A peptide reported as 20% helical has, on average, about a fifth of its residues in a helical conformation under those conditions. That is not the same as saying that a fifth of the molecules are fully helical and the rest are unfolded: more often every molecule is partly helical, in stretches that form and melt at different places along the chain. Change the temperature, the salt, the solvent or the binding partner and those populations shift.
Why short peptides are flexible
The reason is competition for hydrogen bonds. A helix or a sheet is held together by hydrogen bonds between backbone C=O and N-H groups. In water those same groups can hydrogen bond to the solvent instead, and water is present in enormous excess, so intramolecular hydrogen bonds compete poorly[5].
A folded protein wins that competition by burying a hydrophobic core: many side chains pack together, away from water, and the structure pays for itself. A short peptide usually cannot. Its hydrophobic side chains are interspersed with hydrophilic ones along a short chain, which generally prevents the formation of a specific hydrophobic cluster[5].
Early calculations predicted exactly this, with most of the chain extended and the backbone angles in the beta region[5], and early experiments agreed: fragments cut from myoglobin showed little helix in water even though the same sequences are helical inside the intact protein[5].
But the rule is a tendency, not a law, and two exceptions are worth knowing. Designed 16-residue peptides built almost entirely from alanine, with a few lysines or glutamates added for solubility, form stable helices in water, up to about 80% helix at 1 degree Celsius, which overturned the view that a chain shorter than about 20 residues could not show measurable helix in water at all[6]. And a designed 20-residue peptide called the Trp-cage, built only from standard L-amino acids, is more than 95% folded in water at physiological pH, with a melting point of 42 degrees Celsius and cooperative unfolding, the behaviour of a miniature protein rather than a floppy chain[7]. Its core is a tryptophan side chain sheathed by proline rings; the sequence came from truncating and mutating the C-terminal part of exendin-4, the lizard peptide behind exenatide[7].
So the line between peptide and protein is not sharp. Short peptides often sample many conformations in water rather than adopting a single stable fold, and some short peptides, especially cyclic ones, disulfide-bonded ones and carefully designed ones, are genuinely structured.
The shapes a backbone can take
When a peptide does adopt a regular local structure, it is almost always one of a small number of forms. They were worked out for proteins, and the same geometry applies to peptides.
The alpha helix is a coil of about 3.6 residues per turn, held by a hydrogen bond from the C=O of one residue to the N-H four residues further along. It was predicted from the geometry of the planar amide group by Linus Pauling, Robert Corey and Herman Branson in 1951, who gave 3.7 residues per turn for it[8]. The 3-10 helix is a tighter coil with hydrogen bonds to the residue three along; it turns up in short peptides and at the ends of helices. The beta sheet is built from extended strands hydrogen bonded side by side, and the reverse turn is a short motif, often four residues, that reverses the direction of the chain.
Two cautions matter for short peptides. Sheets generally require several strands, so in a peptide they usually mean that molecules have associated with each other rather than that one chain has folded; at the time of the 1991 review there was still no well-characterised case of a single short peptide forming a monomeric hairpin in water[5]. And the turn, not the helix, is the simplest structure a short peptide forms on its own[5]. There is even a name, nascent helix, for the intermediate case: an ensemble of turn-like structures over several adjacent residues, which is not yet a helix but readily becomes one[5].
When binding gives a peptide a structure
The clearest example on this site is GLP-1, the gut hormone behind the GLP-1 weight loss drugs. Its behaviour has been followed from free solution all the way onto the receptor, and the three steps make the general point.
In water the free hormone is largely disordered. In an NMR study at pH 2.5 only a few NOE signals were seen and every backbone amide proton exchanged rapidly with the solvent, both signs of a flexible chain without persistent structure[9].
In a helix-promoting solvent a helix appears, and it appears in a particular order. Adding trifluoroethanol step by step, the helix forms first at the C-terminal end and extends towards the N-terminus as more solvent is added; above about 35% nothing further changes, and the helix then runs from Thr13 to Lys34 with a less well defined region around Gly22[9]. Even then the structure is loose: all the amide protons exchange within 12 hours[9]. Trifluoroethanol is a useful probe rather than a physiological condition, and helicity induced by it does not necessarily match the native state[9].
On the receptor the helix is held. In the crystal structure of GLP-1 bound to the extracellular domain of its receptor, the peptide is a continuous but kinked alpha helix from Thr13 to Val33, and the residues from Ala24 to Val33 are the ones that touch the receptor[1]. The first few residues, His7 to Glu9, are not visible at all, which is the usual signature of a flexible segment[1]. Interestingly, the stretch Thr13 to Glu21 is helical even though it makes no contact with that domain, which suggests that holding the C-terminal part in place stabilises a helix further along the chain[1].
In the complete receptor the peptide goes further. A cryo-electron microscopy structure of the activated receptor with its G protein shows GLP-1 clasped between the extracellular domain and the transmembrane core, with the N-terminus reaching into that core and forming at least one additional helical turn compared with the structure bound to the extracellular domain alone[2].
It is tempting to say the receptor creates the helix, but the mechanism is more interesting than that. In molecular recognition generally, two pictures compete: induced fit, where binding drives a partner into a new conformation, and conformational selection, where the bound conformation already exists in the ensemble and the partner selects it, shifting the populations[10]. Modern evidence supports both, often in sequence: an initial selection followed by induced-fit adjustment of side chains and backbone as the interface is optimised[10]. For GLP-1 the safe statement is that receptor binding stabilises a helical conformation the free peptide can already sample, not that the helix did not exist before.
Binding also does not abolish flexibility. In crystal structures of the angiotensin II receptor, the C-terminal phenylalanine of the bound hormone is poorly resolved and has high temperature factors, indicating that it remains conformationally heterogeneous even inside the binding pocket[11].
When membranes change peptide shape
A lipid surface is a different kind of environment from water: one face of it is oily, and a peptide that can present an oily face of its own is rewarded for folding. Peptides built this way are called amphipathic, meaning that when they coil up, the water-loving and oil-loving side chains end up on opposite sides of the coil.
The human antimicrobial peptide LL-37 shows the effect well. At micromolar concentration in plain water its circular dichroism spectrum is that of a disordered chain[12]. Adding 15 mM of bicarbonate, sulfate or trifluoroacetate brings the helix content to about 40%, while 160 mM chloride is much less effective, so the change is driven by the specific anions and not by ionic strength alone[12]. The transition also depends on peptide concentration in a cooperative way, which points to the helix forming as the molecules associate into oligomers[12]. Below pH 5 the helix is lost and at pH 2 the peptide is entirely disordered[12]. In solutions with the ion composition of plasma, interstitial fluid or the inside of a cell, it is helical[12].
On an actual lipid surface the structure has been solved. In micelles LL-37 is a curved amphipathic helix running from residue 2 to residue 31, with a bend between Gly14 and Glu16 and a disordered C-terminal tail[3]. The same structure is found in micelles whose head groups match the anionic lipids of bacterial membranes, and contacts between the peptide and the lipid were observed directly: the rings of Phe5, Phe6, Phe17 and Phe27, which lie along the concave face, and the arginine side chains all show cross peaks with the lipid[3]. A 12-residue fragment of the middle region, KR-12, keeps a short three-turn amphipathic helix and remains active against bacteria while sparing human cells[3].
The bend is not a defect. GLP-1 has one too: in a membrane-like environment it forms two helical segments with a loose linker around Gly16 in the older numbering, which corresponds to Gly22 in the numbering used for the hormone today[13]. A single straight helix would not work, because the oily face of the first segment would point in almost the opposite direction from the oily face of the second[13]. The distortion shifts the two segments relative to each other so that nearly all the oily residues end up on one side, facing the lipid[13].
A caution the authors of that work stated themselves: a micelle is a model of a membrane, and whether a conformation on a membrane resembles the one the receptor recognises is a separate question[13]. For GLP-1 we can now compare, because the receptor-bound structures came later; for many peptides we cannot.
How chemists constrain peptide shape
If a peptide has to be helical to work, and is helical only part of the time, one strategy is to build the helix into the molecule. Several chemical tools shift the conformational equilibrium, each with its own limits.
Aib and other alpha,alpha-disubstituted residues
Aib, alpha-aminoisobutyric acid, is a noncanonical amino acid in which the hydrogen on the alpha carbon of alanine is replaced by a second methyl group. That extra group severely restricts rotation about the backbone bonds: the allowed angles collapse into two small regions, corresponding to a right-handed or a left-handed helix of the alpha or 3-10 type[14]. Aib has no stereocentre, so which handedness a peptide adopts is decided by the L or D residues around it[14]. Peptides rich in Aib occur naturally in fungi and bacteria, among them alamethicin, which forms ion channels in lipid membranes[14].
The effect is real but it is narrowing, not fixing. Which helix actually forms depends on chain length: in crystal structures, short Aib peptides of eight residues or fewer tend to the 3-10 helix and longer ones to the alpha helix, and the same peptide crystallised under different conditions can switch between them[14]. Aib homopeptides form regular 3-10 helices from the tripeptide upwards, while in an alternating Aib and L-alanine series the alpha helix appears only at the octapeptide[15]. Nor does the trick generalise to every crowded residue: closely related amino acids with two ethyl or propyl groups frequently adopt a fully extended conformation instead of a helix[15]. Aib is also a poor partner for beta sheets, which has made it useful in the opposite role, as a sheet breaker[15].
One more caveat is worth stating plainly, because Aib appears in several modern drugs. In semaglutide, where it sits at position 8 of the GLP-1 sequence, and in tirzepatide, where it sits at positions 2 and 13 of its own chain, Aib is there to block the enzyme DPP-4, which cuts the natural hormone near its N-terminus; its presence in those molecules should not be read as evidence that a helix was deliberately locked in place. Most of the structural data on Aib peptides, moreover, come from crystals of water-insoluble sequences grown from organic solvents, where folding is governed by different forces than in water[14].
Staples, lactam bridges and disulfides
A second approach is to tie two turns of a helix together with a covalent bridge. Earlier bridges used disulfides or lactam links; the developers of the hydrocarbon staple set out to replace them because such cross-links are polar and, in their assessment, pharmacologically labile[16]. The hydrocarbon staple replaced them with a greasy, stable link: two noncanonical amino acids carrying olefin side chains are placed on the same face of the helix and joined by olefin metathesis[16].
The details decide the outcome. Testing a series on a model peptide, bridges spanning one turn gave no benefit, while across two turns the result swung with length: links of 9 and 10 carbons destabilised the helix by 21% and 12%, and a link of 11 carbons stabilised it, raising helix content by 44% over the unmodified control[16]. The same peptides were 41 times more resistant to trypsin; the authors attribute this to the helix, since proteases bind their substrates in an extended rather than a helical conformation[16].
Applied to a biologically active sequence, the gains compound. A 23-residue BH3 peptide is only 16% helical in solution, essentially a random coil; stapled versions reached 35% to 87% helix[4]. Along with the structure came a six-fold tighter binding to its protein partner, a serum half-life of 29.4 hours against 3.1 hours for the parent peptide, and the ability to enter cells[4].
Closing the chain into a ring does the same job by a different route; that is the subject of the cyclic peptides page. In GLP-1, lactam bridges between side chains at positions 16 and 20 or 18 and 22 were well tolerated, and one at 11 to 15 improved potency, which supports the idea that a helix in that part of the chain is what the receptor wants[1].
How we know what shape a peptide has
Two methods do most of the work, and each answers a different question. Knowing their limits explains why a tidy helix in a figure does not mean the peptide is always that shape in water.
Circular dichroism measures the difference in absorption of left- and right-handed circularly polarised light, and it is fast, needs little material and works in ordinary buffers[17]. Regular structures have characteristic signatures: an alpha helix gives negative bands at 222 and 208 nm and a positive one near 193 nm, an antiparallel beta sheet a negative band at 218 nm, and a disordered chain very little signal above 210 nm with a negative band near 195 nm[17]. What CD cannot do is say which residues are involved, since the spectrum is a sum over the whole molecule[17].
It can also miss structure that is there. The carbonyl groups must be well aligned for the helical signal to appear, so a small population of helix with frayed ends can be hard to detect[5]. Estimates of the minimum helix length needed for a recognisable band at 222 nm range from three or four residues to seven to eleven[5], and aromatic side chains can add to or mask the signal[5].
NMR gives the opposite kind of information: site-specific data, residue by residue, from which conformations can be identified in detail[5]. Its limitation is averaging. Because the measurement is slow compared with how fast the chain moves, every NMR parameter is a population-weighted average over all the conformations present, cis-trans isomerisation of a peptide bond, especially at proline, is the classic exception, slow enough for two separate species to appear in the spectrum[5].
That is why the two techniques sometimes disagree, and why the disagreement is not a contradiction: CD reports the average helicity of the whole molecule, NMR reports structure residue by residue[5]. A peptide from myoglobin showed NMR signals over much of the chain indicating a significant population of helical conformations while CD suggested an average of only about 10% helix[5].
One experimental trap applies to both. Peptides associate, and a small amount of aggregate can dominate the measurement. Larger assemblies tumble slowly and often give broadened signals, and, more seriously, contacts between protons on different molecules can be mistaken for contacts within a single chain, producing a misleading picture of what one peptide is doing[5]. The standard control is to check that the result does not change with concentration[5].
Crystallography and cryo-electron microscopy are the methods behind the receptor-bound structures above. They show a peptide held by a partner, which is exactly what is wanted in that case, but packing in the crystal can itself impose a shape. The authors of the GLP-1 structure noted that the kink they observed could be a consequence of crystal packing rather than a property of the hormone, pointing out that replacing the glycine at the kink with alanine did not change binding or activity[1].
Computation is the fourth route, and short peptides are hard for it. Methods that search for low-energy conformations have to cope with chains that populate many structures of similar energy, so a useful answer is an ensemble rather than one model[18]. Deep-learning structure prediction, which has transformed protein modelling, is less accurate for short peptides, in part because peptides are a small fraction of the structural databases these models learn from[18]. The same work illustrates the underlying point with angiotensin II, whose experimental structures differ substantially between solution and its various bound states[18].
Why conformation matters
Four practical consequences follow, and all of them have appeared already on this page.
Binding. A peptide that is already in the right shape pays a smaller entropic price on binding than one that must be ordered from a flexible ensemble. The exendin-4 comparison is a neat illustration: the isolated extracellular domain of the GLP-1 receptor binds exendin-4 with an IC50 of 6 nM but GLP-1 with an IC50 above 500 nM, and the explanation offered is the higher helical propensity of exendin-4 in solution together with extra contacts, with a positive correlation between helicity in solution and affinity for that domain[1]. On the complete receptor the difference largely disappears[1].
Selectivity and signalling. Shape is a property of the complex, not just of the peptide. Small changes at the C-terminus of angiotensin II change the conformation the receptor adopts and shift signalling towards one pathway rather than another[11].
Resistance to enzymes. Stabilising a helix can make the backbone a poorer substrate for the enzymes that cut peptides; the stapled peptides above were 41 times more resistant to trypsin[16]. This connects directly to peptide half-life.
Getting into cells. Folding a chain so that its backbone hydrogen bonds to itself hides polar groups that would otherwise have to be dragged through a membrane, which is one reason stapled peptides entered cells where their parent peptides did not[16][4].
Common questions
What is peptide conformation?
The three-dimensional arrangement a peptide chain takes up at a given moment, set by rotation about the backbone bonds. For a short peptide in water it is better described as an ensemble: many conformations in rapid exchange, with populations that depend on the sequence and the environment[5].
Do short peptides have secondary structure?
Often only in part. Most short linear peptides show conformational preferences rather than a stable fold, and the simplest structure they form on their own is the reverse turn rather than the helix[5]. Designed sequences are the exception: alanine-rich 16-residue peptides reach about 80% helix at 1 degree Celsius[6], and the 20-residue Trp-cage is more than 95% folded in water[7].
Does a peptide fold when it binds its receptor?
Binding stabilises a particular conformation, which is not quite the same statement. The bound shape may already be present in the free ensemble and simply be selected, with further adjustment on binding; both mechanisms operate in molecular recognition[10].
Does Aib lock a peptide into an alpha helix?
No. It narrows the backbone angles to two small regions and favours helical structures, but whether an alpha or a 3-10 helix results depends on chain length, sequence and conditions[14][15]. In semaglutide and tirzepatide, Aib is there to block cleavage by DPP-4.
Why do papers add trifluoroethanol to peptide samples?
Because it promotes helix formation and so reveals a peptide's helical propensity. The structure it induces is not necessarily the one present under physiological conditions, which is why results in trifluoroethanol are reported as such[9].
In short
- Most short linear peptides in water exist as an ensemble of rapidly interconverting conformations rather than one fixed shape; the populations depend on sequence and environment.
- The reason is competition: backbone hydrogen bonds to water compete with the intramolecular hydrogen bonds that hold a helix or sheet together, and a short chain rarely has enough hydrophobic burial to pay for folding.
- Exceptions exist. Designed alanine-rich peptides and the 20-residue Trp-cage are genuinely structured in water, and cyclic or disulfide-bonded peptides are constrained by covalent bonds.
- GLP-1 illustrates the sequence: largely disordered in water, helical in trifluoroethanol starting from the C-terminus, held as a kinked helix from Thr13 to Val33 on the receptor extracellular domain, and extending a further turn into the receptor core in the activated complex.
- A membrane surface selects amphipathic structures. LL-37 is disordered in water but forms a curved helix from residue 2 to 31 on lipid micelles, with its oily side chains facing the lipid.
- Chemists shift the equilibrium with Aib, lactam bridges, disulfides and hydrocarbon staples. These narrow the available conformations rather than guaranteeing a single structure, and the details of length and stereochemistry decide whether a helix is stabilised or destabilised.
- Circular dichroism gives an average over the molecule and can miss small or frayed helices; NMR gives residue-level detail but averages over all conformations present. Aggregation can mislead both.
- Conformation matters for binding affinity, for selectivity between signalling pathways, for resistance to proteases and for crossing membranes.
References
- Underwood CR, Garibay P, Knudsen LB, Hastrup S, Peters GH, Rudolph R, Reedtz-Runge S (2010). Crystal structure of glucagon-like peptide-1 in complex with the extracellular domain of the glucagon-like peptide-1 receptor. Journal of Biological Chemistry 285(1):723–730. DOI: 10.1074/jbc.M109.033829
- Zhang Y, Sun B, Feng D, Hu H, Chu M, Qu Q, Tarrasch JT, Li S, Kobilka TS, Kobilka BK, Skiniotis G (2017). Cryo-EM structure of the activated GLP-1 receptor in complex with a G protein. Nature 546(7657):248–253. DOI: 10.1038/nature22394
- Wang G (2008). Structures of human host defense cathelicidin LL-37 and its smallest antimicrobial peptide KR-12 in lipid micelles. Journal of Biological Chemistry 283(47):32637–32643. DOI: 10.1074/jbc.M805533200
- Walensky LD, Kung AL, Escher I, Malia TJ, Barbuto S, Wright RD, Wagner G, Verdine GL, Korsmeyer SJ (2004). Activation of apoptosis in vivo by a hydrocarbon-stapled BH3 helix. Science 305(5689):1466–1470. DOI: 10.1126/science.1099191
- Dyson HJ, Wright PE (1991). Defining solution conformations of small linear peptides. Annual Review of Biophysics and Biophysical Chemistry 20:519–538. DOI: 10.1146/annurev.bb.20.060191.002511
- Marqusee S, Robbins VH, Baldwin RL (1989). Unusually stable helix formation in short alanine-based peptides. Proceedings of the National Academy of Sciences USA 86(14):5286–5290. DOI: 10.1073/pnas.86.14.5286
- Neidigh JW, Fesinmeyer RM, Andersen NH (2002). Designing a 20-residue protein. Nature Structural Biology 9(6):425–430. DOI: 10.1038/nsb798
- Pauling L, Corey RB, Branson HR (1951). The structure of proteins: two hydrogen-bonded helical configurations of the polypeptide chain. Proceedings of the National Academy of Sciences USA 37(4):205–211. DOI: 10.1073/pnas.37.4.205
- Chang X, Keller D, Bjørn S, Led JJ (2001). Structure and folding of glucagon-like peptide-1-(7-36)-amide in aqueous trifluoroethanol studied by NMR spectroscopy. Magnetic Resonance in Chemistry 39(8):477–483. DOI: 10.1002/mrc.880
- Boehr DD, Nussinov R, Wright PE (2009). The role of dynamic conformational ensembles in biomolecular recognition. Nature Chemical Biology 5(11):789–796. DOI: 10.1038/nchembio.232
- Wingler LM, Skiba MA, McMahon C, Staus DP, Kleinhenz ALW, Suomivuori CM, Latorraca NR, Dror RO, Lefkowitz RJ, Kruse AC (2020). Angiotensin and biased analogs induce structurally distinct active conformations within a GPCR. Science 367(6480):888–892. DOI: 10.1126/science.aay9813
- Johansson J, Gudmundsson GH, Rottenberg ME, Berndt KD, Agerberth B (1998). Conformation-dependent antibacterial activity of the naturally occurring human peptide LL-37. Journal of Biological Chemistry 273(6):3718–3724. DOI: 10.1074/jbc.273.6.3718
- Thornton K, Gorenstein DG (1994). Structure of glucagon-like peptide(7-36) amide in a dodecylphosphocholine micelle as determined by 2D NMR. Biochemistry 33(12):3532–3539. DOI: 10.1021/bi00178a009
- Karle IL, Balaram P (1990). Structural characteristics of α-helical peptide molecules containing Aib residues. Biochemistry 29(29):6747–6756. DOI: 10.1021/bi00481a001
- Toniolo C, Crisma M, Formaggio F, Peggion C (2001). Control of peptide conformation by the Thorpe-Ingold effect (Cα-tetrasubstitution). Biopolymers 60(6):396–419. DOI: 10.1002/1097-0282(2001)60:6<396::AID-BIP10184>3.0.CO;2-7
- Schafmeister CE, Po J, Verdine GL (2000). An all-hydrocarbon cross-linking system for enhancing the helicity and metabolic stability of peptides. Journal of the American Chemical Society 122(24):5891–5892. DOI: 10.1021/ja000563a
- Greenfield NJ (2006). Using circular dichroism spectra to estimate protein secondary structure. Nature Protocols 1(6):2876–2890. DOI: 10.1038/nprot.2006.202
- Li Z, Ru X, Lin Z (2025). Peptide conformation search using fragment splicing and tiered energy models. Computational and Theoretical Chemistry 1245:115108. DOI: 10.1016/j.comptc.2025.115108