Beta-Peptides and Foldamers: Helices Indexed by Ring Size

Adding one methylene group to every residue of a peptide backbone does more than lengthen the chain. A β-amino acid places that extra carbon either between the carbonyl and the side-chain-bearing carbon — a β³ residue, whose side chain then sits adjacent to the backbone nitrogen — or between the side-chain carbon and the nitrogen, giving a β² residue whose side chain sits adjacent to the carbonyl ^[1]. An α-residue contributes three backbone atoms — nitrogen, Cα and the carbonyl carbon — and a β-residue contributes four, so the torsional space to be searched grows with every residue and the naive expectation is that folding gets harder. It does not: β-peptides form helices, turns and sheets at chain lengths as short as four residues ^[1]. Which helix you get, how you prove it, and what the backbone costs to assemble all follow from the geometry that extra carbon imposes.

Helices indexed by hydrogen-bond ring size

The α-peptide vocabulary does not transfer. Names such as α-helix and 3₁₀ are anchored to that three-atom repeat; for β-peptides the convention is the number of atoms in the ring closed by each backbone hydrogen bond. The motifs characterised for β-amino acid oligomers are the 14-, 12-, 10-, 12/10- and 8-helices, alongside hairpin turns, extended structures, stacks and sheets, with assignments drawn from solution NMR and solid-state X-ray work and with backbone dihedral angles from molecular dynamics used as diagnostics of a particular fold ^[2].

Acyclic β³-peptides — oligomers of the singly homologated proteinogenic amino acids — are a well-studied foldamer class whose side-chain-to-14-helicity relationship has been analysed completely and systematically in water ^[3]. The first solution structure in this family, the β-hexapeptide (H-β-HVal-β-HAla-β-HLeu)₂-OH, was assigned from COSY, ROESY, TOCSY and NOE-restrained modelling as a compact left-handed 3₁ helix of 5 Å pitch, while the same work found β-sheet-type structures in the solid state ^[4]. That divergence is the first practical lesson: for β-peptides, phase and solvent are part of the structural claim rather than background to it.

The 12-helix is defined by a series of interwoven 12-membered hydrogen-bonded rings, each linking a carbonyl oxygen to an amide proton three residues toward the C-terminus, and was established by crystallography on a hexamer and an octamer of (R,R)-trans-2-aminocyclopentanecarboxylic acid (trans-ACPC) ^[5]. The 10/12 pattern is different in kind. A β-hexapeptide built from alternating β² and β³ building blocks was examined by CD and NMR with molecular-dynamics simulation, and its minimum-energy conformation was one the authors said might be described as a novel irregular helix containing 10- and 12-membered hydrogen-bonded rings ^[6]. An irregular helix is still a defined structure, but it is not indexed by a single ring size, and its CD should not be read against 14- or 12-helix reference spectra.

What decides which helix forms

Helix identity here is residue-based. The demonstration that β-amino acids could be chosen to give a 12-helix, rather than the 14-helix reported earlier for other β-peptides, was explicitly framed as showing that altering the nature of the residues exerts rational control over the secondary structure ^[7]. Three levers do most of the work.

Ring constraint. A carbocycle fused across the residue's two backbone carbons is the strongest lever, and the size of that ring selects the family. A five-membered ring, trans-ACPC, gives the 12-helix ^[5]. A six-membered ring, trans-2-aminocyclohexanecarboxylic acid (ACHC), promotes the 14-helix: in a direct comparison of β-peptide heptamers in water, ACHC-containing sequences showed the highest 14-helical backbone stability, with hardly any sensitivity to pH or ionic strength ^[8]. One carbon in the constraining ring moves you between helix families.

β² versus β³ substitution. Alternating the substitution position along the chain is what produced the mixed 10/12 conformation, and with one exception the secondary structures formed by these mixed sequences differed from those of the isomers studied previously ^[6].

Side-chain identity, but not by α-peptide rules. In a systematic analysis of β³-peptide 14-helicity in water, residues branched at the first side-chain carbon were found to be 14-helix-stabilising, macrodipole stabilisation was confirmed as important, and β³-homoserine and β³-homothreonine were suggested to contribute side-chain hydrogen bonding as an additional stabilising force ^[3]. The same study reports that 14-helix propensities of β³-amino acids differ starkly from the α-helix propensities of the analogous α-amino acids, and argues that 14-helix folding is governed by different biophysical forces ^[3]. Helix-propensity tables built for α-peptides are not a design tool on this backbone. It also demonstrates that diverse functionality can be incorporated into a stable 14-helix ^[3], so the constraint bears on which residues stabilise the fold, not on how much chemistry the fold can carry.

Solvent is the variable most often under-reported. Many short β-peptides adopt well-defined conformations in organic solvents while requiring specialised stabilising elements to fold in aqueous solution ^[8]. In the heptamer comparison, a side-chain salt bridge did not generate significant 14-helicity in water by CD or 2D NMR, whereas a side-chain-to-side-chain lactam bridge or cyclic residues did; the lactam-bridged series showed lower backbone stability and greater pH and ionic-strength sensitivity than the ACHC series, but increased order between the side chains ^[8]. A CD trace recorded in methanol or TFE leaves aqueous behaviour an open question.

The economical version of the constraint strategy is dilution. A small proportion of highly preorganised residues can impart high stability to a specific helix in water even when most residues in the chain are intrinsically flexible, so interspersing a few copies of an unfunctionalised rigid residue among a larger number of flexible residues with diverse side chains gives β-peptides that adopt the 14-helix and display those side chains in a predictable spatial arrangement ^[9].

How short a chain still folds

This is where β-peptides diverge most sharply from the backbone they were derived from. Discrete secondary structure appears at four residues ^[1], and the 12-helix was solved crystallographically on a hexamer ^[5]. A short chain also buys more turns here than the residue count suggests: the solution helix assigned to the β³-hexapeptide above was a 3₁ helix ^[4], three residues per turn, so six residues close two full turns and present the periodic side-chain array a helix is wanted for. The α-peptide baseline runs the other way: stable α-helix formation by 16-residue alanine-based peptides in water was reported as surprising when set against a classical view that treated the α-helix as marginally stable in water and short helices as unstable ^[10].

Folding is nonetheless cooperative and length-dependent. CD data for trans-ACPC oligomers in methanol were taken to suggest a strongly length-dependent conformational preference, implying that 12-helix formation is cooperative, as it is for the α-helix in conventional peptides ^[5]. So the short-chain advantage is a statement about where the transition begins, not a licence to assume that a tetramer is fully folded. Working at the short end, measure a length series rather than one chain length, and report the fraction folded you can actually defend.

α/β mixed backbones

A pure β backbone presents side chains on a helix whose rise and residues per turn are not those of an α-helix, which is a problem when the design target is a surface that a natural α-helix displays. Mixed α/β backbones are the response: retain enough α residues that the side-chain display stays recognisable, insert enough β residues to change what an enzyme sees.

Two design classes are in use. Block strategies join an α-peptide segment to a β-peptide segment within one hybrid oligomer; interspersed strategies distribute α and β monomers in a regular pattern throughout the sequence, with some β residues carrying protein-like side chains and others cyclic structures that confer conformational rigidity ^[11]. One α/β helical secondary structure contains C=O(i)···H-N(i+4) hydrogen bonds analogous to those of the α-helix and has been shown by crystallography to form helix-bundle quaternary structures ^[11].

The residue pattern does not uniquely determine the fold. For oligopeptides with regular 1:2 and 2:1 α:β residue patterns, two distinct helical conformations were detected by 2D NMR in methanol for each backbone, and only one helix per backbone was characterised by X-ray crystallography ^[12]. A single crystal structure of an α/β-peptide therefore reports one conformation the backbone can adopt, not the only one it populates in solution.

The practical route into a mixed backbone is a replacement scan. Starting from a 19-residue α-peptide that binds the receptor-recognition surface of vascular endothelial growth factor, systematic α→β replacement throughout the sequence identified homologues containing up to roughly 30% β residues that retained significant affinity for that surface and displayed substantial resistance to proteolysis ^[13]. The scan itself is the method: vary one substitution position at a time and measure binding and proteolysis on the same series.

What a CD trace licenses

Across every study above, CD carries the length series and the stability ranking while the helix assignment itself comes from somewhere else. The 12-helix was named from crystal structures of a trans-ACPC hexamer and octamer, with CD contributing only the length dependence that implies cooperativity ^[5]. The first β³-hexapeptide solution structure came from COSY, ROESY, TOCSY and NOE-restrained modelling, and the X-ray from the same study reported a different, sheet-type arrangement ^[4]. The heptamer comparison of stabilising elements read CD and 2D NMR together, and it was the pair that separated a salt bridge that failed from a lactam bridge and cyclic residues that worked ^[8]. The motif catalogue surveyed at the top of this piece rests on the same two-technique base ^[2].

The consequence for your own data is narrow and worth stating. A CD spectrum supports statements about relative order within a series measured under identical conditions; it does not by itself assign a ring size. Naming a helix requires medium-range NOEs or a crystal, and the record has to name the solvent and the phase that produced them. The α/β backbones above sharpen the point, since there the two techniques did not even agree on how many helices there were to name ^[12].

Protease resistance, and what the assay licenses

For pure β backbones the stability claim is unusually well documented. Thirty-six linear and cyclic β- and γ-peptides of as few as two and as many as fifteen residues were offered as substrates to fifteen commercially available proteases of bacterial, fungal and eukaryotic origin, including a β-lactamase, amidases and the 20S proteasome from human erythrocytes, with an α-eicosapeptide and each enzyme's standard substrate run for comparison. Under conditions giving complete cleavage of the α-peptide within 15 min, the β- and γ-peptides were stable for at least 48 h. An inhibition arm using seven of the analogues against α-chymotrypsin found residual activity toward succinyl-Ala-Ala-Pro-Phe-p-nitroanilide unchanged within experimental error after 15 min of incubation ^[14]. An earlier same-sequence comparison makes the point at a single enzyme: the β-hexapeptide was stable to cleavage by pepsin at pH 2 in water for at least 60 h at 37 °C, while H-(Val-Ala-Leu)₂-OH was cleaved instantaneously under the same conditions ^[4].

The design of that experiment is the part worth copying. The α-peptide control sets the incubation window — the 48 h figure means something only because the α-peptide is fully consumed inside 15 min of it — and the inhibition arm rules out the trivial alternative that the β-peptide simply shut the enzyme down ^[14]. Homologation helps the readout as well as the substrate: every β residue carries one methylene more than its α counterpart ^[1], so the analogue and its α control are resolved by mass as well as by retention time, and a single co-incubation can be quenched at intervals and quantified for both species at once rather than run as two separate endpoint experiments.

Resistance does not transfer automatically to mixed backbones. α/β-peptides manifest decreased susceptibility to proteolytic degradation rather than the near-immunity of an all-β chain; in one series built on a hydrocarbon-stapled α-peptide, the α/β analogue was nearly 100-fold more resistant to proteolysis than the parent stapled α-peptide ^[15]. That is a ratio against a specific stapled parent under one assay, not an absolute, and it is a number to reproduce in your own system rather than to inherit.

A retained-activity claim needs its own analytics. Proteolytic stability and target engagement are independent measurements, and the α→β scan pairs them deliberately on the same compounds ^[13]. Reporting only the stability half licenses nothing about function; reporting only the binding half licenses nothing about the backbone's advantage.

Monomers are the practical ceiling

β³-residues are the cheap case. Arndt–Eistert homologation converts α-amino acids into the corresponding β-amino acids, and the reactions leading to the assembled β-peptides run smoothly and in high yield ^[4]. β²-residues are harder. Routes to achiral β²-amino acids are numerous, but enantiomerically pure preparation has been the subject of many separate investigations — diastereoselective alkylation and Mannich-type reactions of cyclic or acyclic β-homoglycine derivatives bearing chiral auxiliaries, Curtius degradation, enantioselective hydrogenations, reductions, C-H insertions and Michael-type additions under transition-metal catalysis, and resolution of racemates ^[16]. Any design that specifies β² residues, including the alternating β²/β³ sequences that give the 10/12 helix, inherits that synthetic burden. The reason designs pay it is structural: the most intriguing effects on β-peptide secondary structure have been observed when β²-amino acids are present in the backbone ^[16].

Ring-constrained residues that also carry side chains are scarcer still: preorganised residues with side chains are rare and challenging to synthesise, which is exactly why the dilution strategy was developed in the first place ^[9]. In practice this couples structure to cost — the residues that most reliably enforce a helix are the ones least available in functionalised form.

Coupling is the second bottleneck. Sequences designed to adopt the 14-helix, especially those containing ACHC, suffer from poor synthetic efficiency under standard solid-phase conditions; microwave and conventional heating both give excellent results for shorter sequences, but a clear benefit from microwave irradiation emerges for longer β-peptides ^[17]. Budget for a messier crude and more forcing coupling conditions than an α-peptide of the same length would need, and check the crude by LC-MS before committing to a long chain.

Positioning against the other backbone strategies

Side-chain crosslinking leaves the α backbone intact and adds a constraint across it; β and α/β backbones change the repeat unit itself, which is why they alter enzyme recognition rather than only helix population. The two are not exclusive — the 100-fold proteolysis ratio quoted above came from an α/β analogue of an already-stapled α-peptide, that is, backbone modification layered onto side-chain crosslinking ^[15]. The choice is a question of what has to be held constant. If the side-chain display must reproduce an α-helix closely, a stapled α-peptide or a sparsely substituted α/β backbone is the conservative option, and the α→β scan tells you how far substitution can go before affinity falls away ^[13]. If enzyme exposure dominates and the display can be re-optimised, an all-β backbone gives the largest change in enzyme recognition of the options covered here ^[14] and gives you a defined structure to re-optimise on at chain lengths far below those at which stable α-helix formation was itself reported as a surprise ^{[1, 10]}.