**2. Peptide Models of the DSE**

The structural preferences associated with unfolded proteins are often described in terms of a predisposition for specific pairs of backbone dihedral angles, phi (Φ) and psi (Ψ). Visually, this is demonstrated with a Ramachandran plot, shown in Figure 1, where pairs of Φ, Ψ angles that are sterically accessible to a polypeptide chain are mapped [47]. For example, using a representative plot computed for the central residue in a poly-alanine tripeptide, it shows that (Φ, Ψ) = (0◦ , 0◦ ) is found in a disallowed region of the plot because these angles for the central residue place the backbone carbonyl oxygen and backbone nitrogen from the first and third residues, respectively, inside the normal contact limits, creating a steric conflict. In contrast, (Φ, Ψ) = (−90◦ , 90◦ ) for the central alanine has no such contact violations for any of the tripeptide atoms, and thus this angle pair is physically allowed. When an unfolded protein shows preferences for some allowed Φ, Ψ pairs at the expense of others, specifically during the rapid interconversion between states of its conformational ensemble, it is said that the unfolded protein exhibits a conformational bias. Φ Ψ Φ Ψ Φ Ψ Φ Ψ − Φ Ψ

Φ Ψ Φ Ψ Φ Ψ α **Figure 1.** Sterically allowed backbone conformations in polypeptides. (**A**) Peptide backbone dihedral angles, Φ and Ψ. (**B**) Ramachandran plot of allowed Φ, Ψ for the central residue in a poly-alanine tripeptide, calculated from structures generated computationally using a hard sphere collision (HSC) model [48,49] and the "normal" atom pair distances from Ramachandran et al. [47]. Approximately 9000 random structures were generated to find 200 sterically allowed configurations. Highlighted by the circled areas are Φ, Ψ regions corresponding to the PPII and α-helix backbone conformations, as indicated.

The idea that unfolded proteins and polypeptides in water may exhibit intrinsic biases for some backbone conformations at the expense of others began to receive widespread consideration when the observation was made that, for a protein chain to achieve its unique structure in a biologically relevant time frame, a random search of all accessible conformations is not possible [50]. The unfolded chain, accordingly, must search a smaller conformational space to what would be predicted from steric considerations alone. This observation predicted that folding is guided by the structural characteristics of the DSE, and experiments to identify folding intermediates, both kinetic [51,52] and equilibrium [53,54], and measure the intrinsic conformational propensities of the amino acids [5] have been extensively pursued over the many decades since.

Early experimental evidence indicating structural preferences in the DSE was provided by Tiffany and Krimm from studies on short poly-proline and poly-lysine peptides using circular dichroism (CD) and optical rotatory dispersion (ORD) spectroscopies [55–57]. Though these short peptides were unfolded, owing to insufficient chain length for forming compact, globular structures, Tiffany and Krimm found strong preferences for PPII structures. This structural motif at the residue level corresponds to the *trans* isomer of the peptide bond and (Φ, Ψ) of approximately (−75◦ , +145◦ ) [43,55]. Its presence in a polypeptide can be established from positive and negative bands in the spectroscopic readings at ~220 nm and ~200 nm, respectively [55,56]. The predisposition for adopting PPII was linked to a variety of factors, such as low temperatures, steric hindrance between side chains, a lack of internal hydrogen bonding, and protonation [57]. Short peptides of poly-glutamic acid also were observed to transition from α-helix at low pH to PPII at neutral pH and higher, identified from CD and ORD spectroscopies [56], indicating that structural transitions between one region of the Ramachandran plot to others could occur for some sequences owing to simple changes in the peptide charge state. These results, Tiffany and Krimm hypothesized, predict a DSE dominated by backbone interconversions between three main structural states: PPII, α-helix, and unordered, where unordered is represented by the random chain [57]. They also speculated, to some resistance [58–60], that solvation effects may contribute to the observed PPII preferences, since the PPII configuration places the backbone amide and backbone carbonyl oxygen polar groups in favorable positions for contact with water. Intrinsic PPII propensities thus could be helpful for keeping unfolded proteins solvated. Overall, their findings from these peptide-based studies supported the idea that unfolded proteins, though highly dynamic and exhibiting broad structural heterogeneity, nonetheless can show backbone conformational biases that are determined locally by sequence details.

Peptide studies have also made extensive use of poly-alanine, because of the natural abundance of alanine in proteins and its chemically simple side chain (i.e., a methyl group). Using a peptide called XAO, where A is an alanine heptamer and X and O are flanking diaminobutyric acid and ornithine, respectively, Kallenbach and coworkers found strong, temperature-dependent preferences for the PPII conformation [11]. <sup>3</sup> *JHN<sup>α</sup>* coupling constants measured by NMR techniques were used to estimate the Φ angle at each alanine position from the Karplus relationship [61], and it was found that Φ was approximately −70◦ at low temperatures. Because both PPII and α-helix can have Φ angles near this value (Figure 1), the presence of the α-helix was ruled out by a lack of measurable NOEs between successive amides in the peptide chain, which is an indicator for α-helix formation. The CD spectrum of XAO also confirmed PPII content. Increasing temperatures caused gradual reductions in populating the PPII state that coincided with an increasing population of β-strand conformations to approximately 10% at 55 ◦C. The reduction in PPII content at high temperatures implied a favorable enthalpy of PPII formation that was also observed by Tiffany and Krimm [57]. Further studies of XAO by Asher et al. using UV Raman spectroscopy established that XAO is structurally similar to a 21-residue alaninepeptide, AP, that forms α-helix under cold conditions [62]. AP transitions to PPII at higher temperatures, and demonstrates that AP, similarly to XAO, shows temperature-dependent conformational preferences.

Additional studies that examined a single alanine flanked on both sides by two glycines (i.e., Ac-(Gly)2-Ala-(Gly)2-NH2) found intrinsic preferences for PPII and heatinduced shifts toward β-strand backbone conformations [63]. Temperature-dependent transitions that exhibit similar structural characteristics have also been seen in alanine tripeptides, tetrapeptides, and octapeptides [18,64,65].

To explore the determinants of the PPII bias in greater detail, quantitative studies designed to measure its dependence on amino acid type were initially conducted by Creamer and coworkers [12]. Host–guest substitutions at an internal position in a prolinerich peptide (Ac-(Pro)3-X-(Pro)3-Gly-Tyr-NH2, where X is the substitution site) were used to analyze substitution-induced effects on the CD spectrum and measure a scale of relative

PPII propensities for 18 of the 20 common amino acids. Bias estimates for tryptophan and tyrosine were not measured, because the aromatic contribution to the CD spectrum from their side chains overlaps with the region where signal height was used to determine PPII content [66,67], impeding their analysis. These experiments found that amino acids with charged side chains, except for histidine, had relatively high preferences for the PPII conformation in this peptide. The observed biases, measured at 5 ◦C, were mostly insensitive to changes in solution pH from 2 to 12. Residues with small, non-polar side chains, such as alanine and glycine, reported somewhat higher propensities for PPII that, in general, exceeded the biases observed from residues with non-polar and bulky side chains, such as isoleucine and valine. The list of amino acid-specific intrinsic propensities for PPII determined in these studies is given in Table 1.

**Table 1.** Experimental intrinsic propensity for the PPII backbone conformation measured in short peptides.


<sup>a</sup> Measured at the X position in Ac-(Pro)3-X-(Pro)3-Gly-Tyr-NH<sup>2</sup> by Creamer and coworkers, at 5 ◦C, and excluding Tyr and Trp [12]. <sup>b</sup> Measured at the X position in Ac-(Gly)2-X-(Gly)2-NH<sup>2</sup> by Kallenbach and coworkers, at 20 ◦C, and excluding Gly and Pro [16]. <sup>c</sup> Measured at the X position in Ac-Val-(Pro)2-X-Val-(Pro)2-(Arg)3-Tyr-NH<sup>2</sup> by Hilser and coworkers, at 25 ◦C [17].

Similarly, Kallenbach and coworkers extended their NMR- and CD-based structural studies of the short peptides mentioned above to include other amino acid types at the central residue position in Ac-(Gly)2-X-(Gly)2-NH2, where X was the substitution site. Substitution-induced effects on peptide structure were then used to establish a scale of PPII bias in this glycine-rich host [16]. Substantial intrinsic PPII propensities were found, giving additional support to the idea that unfolded states are predisposed to PPII (see Table 1). The magnitude of the PPII bias at the peptide guest position, surrounded by glycine, however, was noticeably different (and typically larger) when compared to the amino acid-specific biases that were measured in the proline-based host by Creamer. This predicts position-specific PPII bias in an unfolded chain that is modulated by the amino acid identity at neighboring sites, which has been subsequently verified [68]. Moreover, the glycine-rich peptides exhibited a heat-induced shift in structure from PPII to nonPPII with a slight bias at high temperatures for strand-like conformations. The intrinsic PPII propensities reported in Table 1 from Kallenbach were measured at 20 ◦C.

A third experimental scale of PPII propensity in peptides was measured calorimetrically by Hilser and coworkers [13,17,69]. Their experiments utilized a peptide host–guest

system in which the *Caenorhabditis elegans* Sem-5 SH3 domain binds a peptide in the PPII conformation [70]. This peptide (Ac-Val-(Pro)3-Val-(Pro)2-(Arg)3-Tyr-NH2) is derived from the recognition sequence of a SH3 binding partner, Sos (Son of Sevenless). A non-interacting residue of this peptide corresponding to its fourth position [13] was substituted for each amino acid before binding was measured by isothermal titration calorimetry. The observed change in binding affinity reflects a change in the conformational equilibrium between binding-incompetent and binding-competent (i.e., PPII) states of the peptide ligand, which can be interpreted as a PPII propensity [13,69]. Once again, a substantial intrinsic bias for PPII was observed, albeit at magnitudes and rank orders that were different when compared to the scales determined by either Creamer or Kallenbach. Elam et al. conclude that there is a general consensus regarding amino acids that are high in PPII propensity (proline, lysine, glutamine, and glutamic acid) and low in PPII propensity (histidine, tryptophan, tyrosine, and phenylalanine), with the other amino acids falling in between [17]. The intrinsic PPII propensities in Table 1 from Hilser's group were measured at 25 ◦C.

There are a number of other studies beyond the few described above, each of which uses their own system to examine the structural propensities of the different amino acids in peptides (reviewed in ref. [71]). While the ranks of relative PPII propensities are often both quantitatively and qualitatively different when compared between studies, possibly owing to the use of different host models, all studies have indicated the same general conclusions that (1) unfolded peptides have structural preferences that are predominantly locally determined [72]; (2) nevertheless, these preferences at individual positions can be modulated by the structural features of neighboring residues [68], and (3) importantly, the unfolded chain does not evenly sample the sterically allowed regions of Ramachandran space [71].

In addition to PPII propensities, alanine-based peptides have been utilized to measure intrinsic α-helix-forming tendencies in a host–guest model that was designed to avoid stabilizing side chain–side chain and side chain–macrodipole interactions [22]. Though cold temperatures were required for this peptide to populate helix at appropriate levels for study, Baldwin and coworkers measured amino acid substitution effects on the CD signal at 222 nm and determined an experimental scale of α-helix intrinsic propensities for each of the 20 common amino acids. At 0 ◦C, most of the amino acids disfavored forming helix at guest positions in the alanine-based host, while leucine and arginine were indifferent to helix-formation. Alanine, however, had a preference for forming helix in this host. The intrinsic propensity for forming α-helix determined by Baldwin and coworkers for each of the common amino acids is provided in Table 2.


**Table 2.** Experimental intrinsic propensity for the α-helix measured in short peptides.

**Table 2.** *Cont.*


<sup>a</sup> Measured in an alanine-rich host at 0 ◦C by Baldwin and coworkers [22]. In the original report, bias for the α-helix was given as a free energy (∆*G*) of helix formation. The values for ASP, GLU, LYS, and ARG represent the charged species; His value is for the neutral species. <sup>b</sup> α-helix propensities were estimated from the free energies as *Kα*/(1 + *Kα*), where *K<sup>α</sup>* = e−∆*G*/*RT* , ∆*G* is from column 2, *R* is the gas constant, and *T* is temperature.
