Effects of D‐Amino Acid Replacements on the Conformational Stability of Miniproteins

Abstract

For many proteins, proper function requires adoption of a specific tertiary structure. This study explores the effects of L‐to‐D amino acid substitutions on tertiary structure stability for two well‐known miniproteins, a single‐site variant of the chicken villin headpiece subdomain (VHP) and the human Pin1 WW domain (WW). For VHP, which features an α‐helix‐rich tertiary structure, substitutions led to significant destabilization, as detected by variable temperature circular dichroism (CD) measurements. For WW, which has a β‐sheet‐rich tertiary structure, most single L‐to‐D changes seemed to cause complete unfolding at room temperature, according to CD measurements. These findings suggest that amino acid residue configuration changes at a single site will often prove to be deleterious in terms of tertiary structure stability, and in some cases dramatically destabilizing.

We probed the effects of L‐to‐D amino acid substitutions on protein stability using far‐UV CD. Comparison of the α‐helix‐rich villin headpiece subdomain and the β‐sheet‐rich WW domain revealed that L‐to‐D substitutions generally destabilize proteins.

Introduction

Proteins are responsible for a wide range of functions that are necessary for life; no other class of biomolecules is so pervasive in terms of critical activities that include catalysis, energy capture and utilization, signal transduction and transport. From a chemical perspective, it seems remarkable that molecules constructed from only 20 amino acid building blocks can display such a broad range of functions. The proteinogenic α‐amino acids are defined not only by their side chains, but also, except for glycine, by their absolute configuration. The natural translational machinery strongly favors incorporation of L‐α‐amino acids into proteins. In contrast, prokaryotic nonribosomal synthetases often incorporate D‐α‐amino acid residues into short peptides.[ ¹ , ² ] Ribosomally produced proteins isolated from natural sources can contain low levels of D residues. ^[3] Some short, bioactive peptides of ribosomal origin contain D residues that are apparently generated as a post‐translational modification. ^[4] Configurational change at a single site can exert profound effects on the biological activity of a peptide hormone. ^[5]

Incorporation of D‐α‐amino acid residues via chemical peptide synthesis is straightforward. Peptides containing D residues have been developed as drugs and drug candidates,[ ⁶ , ⁷ ] and have proven useful as research tools.[ ⁸ , ⁹ , ¹⁰ ] Inclusion of D residues in macrocyclic peptides can influence conformation and promote membrane permeability.[ ¹¹ , ¹² , ¹³ ] Because D residues suppress protease action at nearby positions, ^[14] there is considerable interest in engineering translational machinery to promote D residue incorporation.[ ¹⁵ , ¹⁶ ]

Protein function often depends upon conformation, but our understanding of the impact of L‐to‐D changes on the folding behavior of linear polypeptides is limited. α‐Helix stability is generally diminished when an L residue is replaced by the corresponding D residue. For alanine, this stereochemical change destabilizes the α‐helix by ~1 kcal/mol[ ¹⁷ , ¹⁸ ]; however, energetic consequences of a configuration change appear to vary depending on the side chain. ^[19] Hecht et al. used modified ribosomal machinery to interrogate single L‐to‐D variants of two enzymes, E. coli dihydrofolate reductase (159 residues; three epimers examined) and firefly luciferase (550 residues; two epimers examined). ^[16] The resulting protein diastereomers were evaluated in terms of specific catalytic activity, which provides a very sensitive measure of the polypeptide's ability to achieve a native‐like tertiary structure. Even when the L‐to‐D modification was made at a position known to be important in terms of catalytic function, substantial enzymatic activity was retained (≥10 % of native specific activity). These findings suggest that although L‐to‐D modifications can be disruptive in terms of tertiary structure, the rest of the protein may be able to adjust to compensate at least partially for the disruption in a sufficiently large protein.

Here, we investigate the effects of L‐to‐D modifications on the conformational stability of two tertiary folding units that are small enough to be easily prepared via conventional solid‐phase peptide synthesis (SPPS). The villin headpiece subdomain, 35 residues, forms a tertiary structure containing several α‐helices[ ²⁰ , ²¹ ]; below, “VHP” refers to the sequence from chicken with a single change (Asn27→His) ^[21] (Figure 1A). The human Pin1 WW domain (“WW” below), 34 residues, forms a β‐sheet‐rich tertiary structure ^[22] (Figure 1B). Neither of these “miniproteins” contains a disulfide or requires metal ion coordination for folding. Thus, each is a good vehicle for determining how single L‐to‐D substitutions, implemented at multiple positions, influence tertiary structure stability.

Figure 1.

A) Crystal structure of VHP (1YRF) with the positions of L‐to‐D substitution shown in blue. B) Structure of WW (4GWT) with the positions of L‐to‐D substitution shown in blue. C) Chemical structures of L‐amino acid and D‐amino acid. D) Sequences of VHP (1) and WW (2).

Results and Discussion

Experimental Design

High‐resolution structures have been determined by x‐ray crystallography or NMR for VHP, WW and sequence variants.[ ²¹ , ²² ] In each case, the native structure is associated with a distinctive circular dichroism (CD) signature. ^[23] For VHP, this signature has minima near 208 and 222 nm, as expected given that many residues participate in α‐helical secondary structure. For WW, the CD signature features a highly characteristic maximum near 227 nm, which presumably reflects contributions from the two Trp side chains. Our study employed far UV CD data to compare VHP and WW with multiple diastereomers containing at least one L‐to‐D substitution.

Diastereomers of VHP

We used variable‐temperature CD to evaluate the conformational stability of VHP and a set of diastereomers. ^[24] The protein concentration was set at 50 μM, and all CD measurements were conducted in 20 mM sodium acetate buffer, pH 5.0. Each of the 10 single L‐to‐D VHP variants examined displayed a CD signature at room temperature that suggested a retention of α‐helical secondary structure (Figure 2). However, in every case, the extent of α‐helicity was diminished, with the Trp23 epimer most extreme. For variable‐temperature measurements, we monitored mean residue ellipticity (MRE) at 223 nm. Upon heating, the solution containing VHP itself displays a sigmoidal thermal transition with a midpoint at 55.1±1.2 °C. This transition was reversible, as indicated by the data collected as the sample was cooled (Figure S1). Solutions containing most of the single L‐to‐D variants also displayed a sigmoidal change in MRE at 223 nm upon heating, but in some of these cases the transition was not fully reversible (Figure S2–S11). This observation suggests that heating causes a portion of the protein to aggregate irreversibly. Therefore, the midpoints of the heating transitions are interpreted as an apparent melting temperature (T_m,app).

Figure 2.

Far‐UV CD for VHP and diastereomers containing a single L‐to‐D substitution. Samples contained 50 μM polypeptide, 20 mM sodium acetate buffer, pH 5.0. Measurements at 25 °C. A) Substitutions in helix 1, B) substitutions in helix 2 and loop, C) substitutions in helix 3.

Table 1 summarizes results for the 10 single L‐to‐D VHP variants. L‐to‐D substitution at Gln26, His27, Lys30 or Glu31 in Helix 1 resulted in modest destabilization relative to VHP, with His27 showing the smallest change ($\Delta$ T_m,app=−1.2 °C) and Lys24 the largest ($\Delta$ T_m,app=−28.5 °C). The trend among the Helix 1 substitutions was generally consistent with the trend reported for D‐amino acid substitutions in an isolated α‐helix ^[19] , but the large destabilization observed for the Lys24 epimer ($\Delta$ T_m,app=−28.5 °C) deviated from this trend. The large difference between substitution at Lys24 and Lys30 shows that side chain identity is not the only factor that determines the impact of stereochemical inversion. The crystal structure of VHP reveals that the side chain of Lys24 makes more contacts with other side chains than does the side chain of Lys30 (Table S2). Presumably, greater involvement of the side chain in tertiary packing interactions will correlate with a large negative effect of L‐to‐D substitution on tertiary structure stability.

Table 1.

Apparent melting temperatures derived from CD for VHP analogues with single L‐to‐D substitution.

Types	Variant	Tm,app ( )	$\Delta$ Tm,app ( )
	VHP	55.09 (±1.2)	‐
Helix 1	E4e	45.2 (±0.7)	−9.9 (±1.9)
	A8a	41.2 (±0.7)	−13.9 (±1.9)
Helix 2	S15s	50.2 (±0.7)	−4.9 (±1.9)
Loop	N19n	42.0 (±1.1)	−13.1 (±2.3)
Helix 3	W23w	Not determined	–
	K24k	26.6 (±1.2)	−28.5 (±2.4)
	Q26q	48.1 (±1.5)	−7.0 (±3.7)
	H27h	53.9 (±2.8)	−1.2 (±4.0)
	K30k	50.4 (±0.8)	−4.7 (±2.0)
	E31e	46.8 (±0.9)	−8.3 (±2.1)

L‐to‐D substitution at Ser15 ($\Delta$ T_m,app=−4.9 °C) in Helix 2, Glu4 ($\Delta$ T_m,app=−9.9 °C), and Ala8 ($\Delta$ T_m,app=−13.9 °C) in Helix 3 or Asn19 ($\Delta$ T_m,app=−13.1 °C) in the loop between Helices 2 and 3 exhibited modest destabilizing effects. In each case, CD at room temperature indicated retention of an α‐helical signature.

To further explore the tolerance for D residues in VHP, we designed double and triple L‐to‐D substituted analogs based on the most stable single substitutions: Ala8, Ser15, Asn19, His27, and Lys30. Data in Table 2 suggest that double L‐to‐D substitutions lead to a more significant destabilization than single substitutions. The ΔT_m,app values for the double substitutions are larger than the sum of ΔT_m,app values for the single substitutions. Thus, for example, the VHP stereoisomer with L‐to‐D substitution at His27 and Lys30 exhibited ΔT_m,app=−11.4 °C, while substitution at His27 alone resulted in ΔT_m,app=−1.2 °C, and substitution at Lys30 alone resulted ΔT_m,app=−4.7 °C. These findings suggest that multiple substitutions may cooperatively enhance tertiary structure destabilization.

Table 2.

Apparent melting temperatures derived from CD for VHP analogues with multiple L‐to‐D substitutions.

Variant	Tm,app ( )	$\Delta$ Tm,app ( )
VHP	55.1 (±1.2)	‐
H27h_S15s	44.4 (±0.6)	−10.7 (±1.8)
H27h_K30k	43.5 (±1.4)	−11.6 (±2.6)
H27h_A8a	35.0 (±0.9)	−20.1 (±2.1)
H27h_N19n	34.1 (±2.1)	−21 (±3.3)
A8a_S15s	30.0 (±1.1)	−25.1 (±2.3)
S15s_N19n	28.8 (±1.6)	−26.3 (±2.8)
H27h_K24k	23.3 (±1.7)	−31.8 (±2.9)
A8a_N19n	20.7 (±3.9)	−34.4 (±5.1)
H27h_A8a_S15s	24.5 (±5.8)	−30.6 (±7.0)
H27h_A8a_N19n	16.7 (±3.9)	−38.4 (±5.1)
H27h_A8a_S15s_N19n	Not determined	–

Thermal destabilization trends observed for double substitutions were qualitatively consistent with those identified in single substitutions. Thus, for example, double‐substituted stereoisomer substitution at His27 and Lys24 (ΔT_m,app=−31.8 °C) manifests a greater thermal destabilization relative to the stereoisomer with double substitution at His27 and Lys30 (ΔT_m,app=−11.4 °C), which correlates with the observation that single substitution at Lys24 (ΔT_m,app=−28.5 °C) is more destabilizing than single substitution at Lys30 (ΔT_m,app=−4.7 °C). Most VHP diastereomers with double L‐to‐D substitution maintained some degree of α‐helicity at 25 °C, according to CD (Figure S26). The two VHP diastereomers with three L‐to‐D substitutions were highly destabilized, with T_m,app values near or below room temperature. The diastereomer with four L‐to‐D substitutions exhibited a CD spectrum suggesting random coil at room temperature. Collectively, these results indicate that the VHP tertiary structure is readily destabilized by L‐to‐D substitution at a small number of positions. The destabilizing effect is not simply additive but appears to be exaggerated for multiple substitutions.

Diastereomers of WW

CD measurements involving WW were conducted with 50 μM protein in 10 mM sodium phosphate buffer, pH 7.0 (Figure 3). Comparisons involving eight single L‐to‐D variants revealed that only two of these diastereomers retained the characteristic maximum near 227 nm. One of these variants was the epimer at Arg16, in the loop between β‐strands 1 and 2. The other was the epimer at Gln28, near the C‐terminus of β‐strand 3. The other six L‐to‐D replacements occurred in the core of the β‐sheet that dominates the WW domain fold, and each seemed to disrupt tertiary structure formation at room temperature.

Figure 3.

Far‐UV CD for WW and derivatives containing a single L‐to‐D substitution. Samples contained 50 μM polypeptide, 10 mM sodium phosphate buffer, pH 7.0. Measurements at 25 °C. A) WW and the two diastereomers that retain the native CD maximum at ~227 nm, B) WW and the highly destabilized diastereomers.

Variable‐temperature studies with WW, as monitored by MRE at 227 nm, revealed a sigmoidal transition. However, this thermal transition was not reversible, nor was the thermal transition for the Arg16 epimer or the Gln28 epimer. Table S1 presents the thermal denaturation data for both of these epimers. L‐to‐D substitution at position Arg16 led to ΔT_m,app=−13.4 °C, while L‐to‐D substitution at Gln28 exhibited a larger destabilization, ΔT_m,app=−18.6 °C. These findings show placing a single D residue in WW is highly destabilizing to the tertiary fold.

Conclusions

We have used two miniproteins that adopt a specific tertiary structure to evaluate the impact of stereoisomeric changes on conformational stability. The villin headpiece subdomain and the Pin1 WW domain are complementary in that most residues in VHP participate in α‐helical secondary structure, while most residues in WW participate in β‐sheet secondary structure. It is challenging to coax the ribosome to incorporate D‐α‐amino acid residues, but D incorporation is straightforward with chemical synthesis. The lengths of VHP and WW (35 and 34 residues) make them amenable to synthesis and evaluation of many diastereomers.

Our results indicate that the tertiary structures of both miniproteins are very sensitive to destabilization via L‐to‐D substitution. For VHP, CD data suggested that most of the single‐site modifications we examined did not lead to complete loss of folding, as indicated by retention of some α‐helicity at room temperature, although every substitution site led to some degree of destabilization. Multiple L‐to‐D substitutions resulted in greater destabilization; the one diastereomer we examined with four L‐to‐D substitutions seemed to be fully unfolded at room temperature. WW appeared to be even more sensitive to disruption by L‐to‐D substitution than was VHP, since only two of the eight single substitutions we evaluated led to retention of the characteristic CD maximum near 227 nm. This greater sensitivity may arise because the WW tertiary structure is intrinsically less stable than the VHP tertiary structure (T_m,app=55 °C vs. 44 °C), or because α‐helical secondary structure is more tolerant of L‐to‐D substitution than is β‐sheet secondary structure, or a combination of these factors.

Our finding that L‐to‐D substitution is generally destabilizing in two small tertiary structures offers an interesting perspective on previous studies showing that replacing a native Gly residue with a D residue can occasionally stabilize peptide or protein folding patterns.[ ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ ] In these cases, the native Gly residue displays backbone torsion angles that would be unfavorable for L residues but are favorable for D residues. For Gly, of course, backbone torsion angles preferred by L residues and those preferred by D residues are energetically equivalent. The use of D residues to promote “mirror image” reverse turns has proven useful in the design of β‐sheet‐forming peptides.[ ³¹ , ³² , ³³ ]

Polypeptides containing mixtures of L and D residues can adopt distinctive conformations that enable unique functions. Gramicidin A, for example, features an alternation of L and D residues. This natural antibiotic can form a double‐stranded β‐helix that functions as an ion channel. ^[34] For ribosomally produced polypeptides, however, evolutionary selection has necessarily operated on homochiral molecules. The results reported here are consistent with earlier studies ^[16] in showing that evolved protein tertiary structures are generally destabilized by L‐to‐D modification. This trend may explain why mammals devote energy to sustaining α‐amino acid homochirality. ^[35]

Conflict of Interests

The authors declare no competing interests.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Xu R., Huang J., Kuhn A. J., Gellman S. H., ChemBioChem 2025, 26, e202500085. 10.1002/cbic.202500085

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.