Design of variants

Alanine (Ala) mutations were constructed for 32 of the 33 non-Ala residues of hPin1 WW (residue 31 is Ala in wild type hPin1 WW). Only the C-terminal residue, Gly39, which is completely solvent exposed, was not mutated. Ala-mutagenesis eliminates the side-chain beyond the β-carbon, generally without significantly changing the φ, Ψ backbone dihedral angle preferences, thus allowing an unbiased and quantitative evaluation of the influence of each side chain to the stability of the protein. This perturbation thermodynamic analysis identifies the residues whose side chains are critical for acquiring and maintaining a three-stranded β-sheet structure, as mutation of those to Ala significantly destabilizes the structure.^14,15 We also conducted a partial Gly-scan within the two solvent-exposed reverse turns, that is, in the loop 1 (Ser16-Arg21) and loop 2 (His27-Asn30) substructures that connect β-strands 1 and 2 and β-strands 2 and 3, respectively. Gly-substitutions are generally expected to be more destabilizing than Ala-mutations, because of the larger conformational freedom of Gly in the unfolded state due to its expanded permitted φ, Ψ backbone dihedral angles.

For the Ala substitutions that resulted in unfolded or substantially destabilized hPin1 WW variants, specific point mutations were prepared at those positions to learn more about the requirements for acquiring and maintaining a stabilized β-sheet structure. For example, Tyr23 was mutated to Phe and Leu, in addition to the destabilizing Ala substitution. Asn26 was mutated to the isosteric but charged residue Asp, and Pro37 was mutated to the unnatural amino acid 3,4-dehydroproline (Δ_3,4P).

Characterization of variants

Equilibrium thermal denaturation curves of hPin1 WW variants were recorded utilizing far-UV CD spectroscopy. Folded wild type hPin1 WW, like other members of the WW domain family, exhibits a characteristic positive ellipticity around 226 nm [Fig. ​[Fig.1(A),1(A), solid black spectrum], which disappears upon thermal [Fig. ​[Fig.1(A),1(A), dashed black spectrum] or chaotrope denaturation and thus serves as a folding probe.^{7,16,18,34–36} Only four (Trp11Ala, Tyr24Ala, Asn26Ala, Pro37Ala) out of the 32 Ala-mutants investigated in this study appeared to be completely [Fig. ​[Fig.1(A),1(A), Trp11Ala, solid green spectrum] or partially unfolded [Fig. ​[Fig.1(A),1(A), Pro37Ala, solid red spectrum] at physiological temperatures, as indicated by a significantly reduced positive ellipticity in the far-UV CD spectrum at 226 nm. Thermal unfolding curves of these variants were either non-cooperative {Trp11Ala [Fig. ​[Fig.1(B),1(B), green symbols], Tyr24Ala, Asn26Ala} or the mutant was so destabilized that the pretransition baseline was not detectable [Pro37Ala; Fig. ​Fig.1(B),1(B), red symbols], even at low temperature (2°C).

Open in a separate window

Figure 1

Thermodynamic analysis of wild type hPin1 WW and variants thereof. (A) Monitoring thermal denaturation of hPin1 WW by far-UV CD spectroscopy: Spectra of folded wild type hPin1 WW domain (2°C, solid black line), denatured wild type hPin1 WW domain (98°C, dashed black line), intrinsically unstructured Trp11Ala hPin1 WW domain (2°C, solid green line) and partially folded Pro37Ala hPin1 WW domain (2°C, solid red line) are shown. (B) Raw thermal denaturation data from wild type hPin1 WW domain (filled black symbols), and the intrinsically unstructured Trp11Ala (filled green symbols) and the partially folded Phe37Ala (filled red symbols) hPin1 WW variants. The hPin1 WW domain concentration in (A) and (B) was ∼15 μM. Measurements were performed in 20 mM sodium phosphate, pH 7.0 using a 2 mm quartz cuvette. The solid line connecting the black symbols is a fit of the data to a two-state model [Eqs. (1) and (2) in Materials and Methods]. The solid lines connecting the red and green symbols are simply guides to the eye and have no theoretical foundation. (C) Representative normalized equilibrium thermal denaturation curves [Eq. (3), Materials and Methods] of hPin 1 WW variants. Shown are data for Y23A-, L7A-, T29A-, P8A-, N30A-, and wild type hPin1 WW domains, in increasing order of stability. (D) Plot of the folding free energy (ΔG, in kJ/mol) at 40°C versus the midpoint of unfolding (T_M) for the hPin1 WW point variants listed in Table ​TableI.I. For clarity, the data point obtained for wildtype hPin1 WW is color coded red.

All other mutants were fully folded at 2°C and exhibited cooperative thermal denaturation curves [representative curves are shown in Fig. ​Fig.1(C)].1(C)]. The midpoint of unfolding (T_M, defined as ΔG(T_M) = 0) and folding free energies (ΔG(T)) for each mutant were determined by fitting the thermal denaturation curves to a two-state unfolding model as described previously (Table ​(TableII).⁷ All folded mutants exhibit similar unfolding cooperativities, as indicated by the excellent correlation between the extracted free energies of folding (ΔG(T)) and the folding midpoints (T_M) estimated from a two-state analysis of the thermal denaturation curves [Fig. ​[Fig.11(D)].

The effect of the various Ala- [Fig. ​[Fig.2(A)]2(A)] and Gly-substitutions [Fig. ​[Fig.2(B)]2(B)] on the stability of hPin1 WW, calculated at near-physiological temperature (40°C), is depicted graphically. Significantly destabilized mutants whose ΔΔG exceed 4 kJ/mol, but are in the measurable range, are depicted by blue bars. Red bars indicate highly destabilized mutants where only a lower estimate of ΔΔG can be offered, as these hPin1 WW variants were unfolded and therefore accurate free energies could not be obtained. Except for the Asn30Gly variant (loop 2), which significantly exceeds the stability of the Asn30Ala analogue, the Gly variants are either less stable than the Ala-mutants, or exhibit T_M's within a couple of degrees of the corresponding Ala variants, reflecting similar stabilities. The higher stability of the N30G mutant is consistent with a preference for Gly at this position among WW domain family members (http://smart.embl-heidelberg.de/smart/do_annotation.pl?DOMAIN = SM00456), suggesting that the gain in stability results from an optimization of the local ϕ/ψ-backbone dihedral angles and improved loop formation predisposition. On average, the six-residue loop 1 substructure (residues Ser16-Arg21) is much more permissive of substitutions by Gly than the four-residue reverse turn 2 (residues His27-Asn30), consistent with loop 1's increased conformational mobility relative to reverse turn 2,^7,19,31,37 and therefore, it is able to accommodate destabilizing Gly mutations without significantly compromising hPin1 WW folding or stability. Conformational disorder in the loop 1 substructure of folded hPin1 WW has been shown directly by NMR backbone relaxation studies³⁸ and seems to be important for the biological activity of hPin1 WW.^31,38

Open in a separate window

Figure 2

Quantitative analysis of the complete Ala- and partial Gly-scans of the hPin WW domain. (A) Effects of the Ala mutations on the equilibrium stability of hPin1 WW. A positive value indicates that the Ala mutation is destabilizing relative to the wild type hPin1 WW domain. Mutations that are destabilizing by more than 4 kJ/mol are labeled (single letter code). Mutations of residues Leu7, Arg14, Tyr23, Phe25, and Thr29 result in destabilized variants that are only fully folded at low temperature (2°C) (blue bars). Accurate free energies cannot be reported for Ala-mutations at residues Trp11, Tyr24, Asn26, and Pro37, as the corresponding Ala-mutations resulted in partially or fully denatured WW domains, even at 2°C. The energies reported therefore are lower estimates (red bars). Residues that contribute to β-strands 1–3 are indicated. (B) Effects of Gly substitutions at 13 positions on the equilibrium stability of hPin1 WW. A positive value indicates that the Gly mutation is destabilizing relative to the wild type protein. Mutations that are destabilizing by more than 2 kJ/mol, as well as mutations that are more stable than wild type, are labeled (single letter code). Mutations of residues Thr29 and Ala31 result in destabilized variants that are only fully folded at low temperature (2°C) (blue bars). Residues that contribute to the loop 1 and loop 2 substructures are indicated. (C) Structural depiction of the hPin1 WW domain. The side chains of residues Trp11, Tyr24, and Pro37 that constitute the hydrophobic stability minicore are shown explicitly in ball-and-stick mode and are color-coded red. (Leu7, colored blue, bottom right, also is part of this stability minicore.) The side chains of residues Arg14, Tyr23, and Phe25 that contribute to the second hydrophobic core involved in ligand binding are shown explicitly in ball-and-stick mode and are color-coded blue (top left). (D) Structure of 3,4-dehydroproline (Δ_3,4P).

Considering the data from the Ala-scan as a whole, the mutations can be classified into 3 groups based on their effect on the folding thermodynamics: (i) mutations that lead to unfolded or partially folded WW domains at all temperatures [Fig. ​[Fig.2(A),2(A), red bars], (ii) mutations that result in variants that are substantially unfolded (>40%) at physiological temperature, but fully folded at low temperature (2°C) [Fig. ​[Fig.2(A),2(A), blue bars], and (iii) mutations that affect hPin1 WW stability only slightly (gray bars).

Ala-mutants that remain unfolded at 2°C

Of the four Ala variants that are completely denatured at physiological temperature and are substantially denatured at low temperature (2°C) [Trp11Ala, Tyr24Ala, Asn26Ala, and Pro37Ala; Fig. ​Fig.2(A),2(A), red bars], three of these side chains (Trp11, Tyr24, and Pro37) form a hydrophobic mini core that is highly conserved in all WW domain family members [Fig. ​[Fig.2(C),2(C), labeled red residues]. These residues are not directly involved in peptide ligand binding.³⁹ No evidence for aggregation of these unfolded mutants is apparent, even upon long-term storage at 4°C, consistent with earlier work.⁴⁰ Preliminary studies indicate that the aggregation resistance is directly related to the presence of the Pro-rich N- and C-terminal extensions, as their removal resulting in a Gly10-Trp34 hPin1 WW domain leads to measurable amyloidogenicity at physiological salt concentrations (M. J. and J. W. K., unpublished data). Pro37 is an invariant residue in WW domain family members that appears to be required for the integrity of the hydrophobic mini core.⁴¹ Only a very conservative, non-natural and near-isosteric substitution of the proline ring bearing a double bond, a 3,4-dehydroproline analogue (Δ_3,4P), yields a cooperatively folded, yet destabilized protein [Fig. ​[Fig.2(D),2(D), Table ​TableII].

Trp11 and Tyr24 are highly conserved amongst WW domain family members.^42,43 Mutagenesis studies demonstrate that only variants containing aromatic residues at position 11 and 24 are folded, consistent with positional frequencies calculated from >200 WW domain sequence entries in the data base.^42,44 Trp is the most frequent amino acid at position 11, and is only replaced by Phe. Tyr is the most abundant residue at position 24, but in some instances it is substituted by Phe and less frequently by Trp. The equivalent of Tyr24, shown to be critical for folding of hPin1 WW in this study, is mutated to a Cys residue in the PQBP1 WW domain (a brain-enriched transcriptional regulator) causing Golabi-Ito-Hall syndrome associated with mental retardation.⁴⁰

Asn26 serves both as a H-bond donor and acceptor, mediating side-chain-side-chain and side-chain-backbone H-bonds, all of which are disrupted in the Asn26Ala and Asn26Leu variants that are unfolded.¹⁰ A near-isosteric Asp residue can compensate for some (but not all) of the lost interactions exhibited by the hydrophobic side chain replacements, and is frequently found in place of Asn at position 26 (http://smart.embl-heidelberg.de/smart/do_annotation.pl?DOMAIN = SM00456), consistent with our data revealing that the Asn26Asp variant is folded. A reverse mutation (Asp30Asn) at the topologically equivalent position in the hYap WW domain³⁶ stabilizes this domain, seemingly accounting almost completely for the loss in stability seen in hPin1 WW upon reverse mutation, suggesting similar side-chain-side-chain and side-chain-backbone H-bonds in both WW domains.

Ala-mutations at the Trp11, Tyr24, Asn26, and Pro37 equivalent positions in the related FBP28 and hYap65 WW domains also yield unfolded proteins,^41,45 suggesting that this hydrophobic minicore is generally important for WW domain structural stability. We envision that the absence of Pro37 would render the hydrophobic side chains of Trp11 and Tyr24 accessible to solvent, decreasing the hydrophobic effect stabilizing the β-sheet structure of the WW domain. Increased solvent exposure of aromatic side chains and/or changes in β-sheet twist may explain why most de novo designed triple-stranded β-sheets that lack the equivalent of a delocalized Trp11-Tyr24-Asn26-Pro37 hydrophobic minicore are only modestly stable, and exhibit a high tendency to aggregate.

Ala-mutants that are folded at 2°C, but significantly destabilized at physiological temperature

Five Ala-variants (Leu7Ala, Arg14Ala, Tyr23Ala, Phe25Ala, and Thr29Ala) are unfolded to a substantial degree (>40%) at physiological temperature, but are fully folded at lower temperature (2°C) and exhibit cooperative unfolding transitions with defined pre- and post-transition baselines [Fig. ​[Fig.2(A),2(A), blue bars].

The side chain of Leu7 [colored blue in Fig. ​Fig.2(C),2(C), bottom right] participates in the formation of the hydrophobic stability core (colored red) mentioned in the previous section [Fig. ​[Fig.2(C)].2(C)]. The two methyl groups of Leu7 seal the otherwise partially solvent-exposed hydrophobic minicore in a snap-lid fashion, similar to Ile17 in the sequence-related hYap65 WW domain.³⁰ The thermodynamic influence of Leu7 on hPin1 WW folding seems to be particularly sensitive to side chain modification, as evidenced by the increasing destabilization observed across the Leu7Ile-Leu7Norval-Leu7Val-Leu7Ala mutant series (Table ​(TableII).

The side chain of Thr29 serves as an H-bond acceptor for the backbone amide of residue Ala31 at the beginning of β-strand 3. Our differential scanning mutagenesis at position 29 reveals that Ser, the only natural residue that can replace Thr29 without eliminating the aforementioned side-chain-backbone H-bond, exhibits a wild type like stability, whereas both Thr29Ala and Thr29Gly variants are considerably more destabilized.

The remaining three side chains in this category, Arg14, Tyr23, and Phe25, are all positioned on the concave side of the β-sheet and form a second hydrophobic minicore [Fig. ​[Fig.2(C),2(C), blue color, top left]. Together with residues Ser16 and Arg17 (loop 1), and Trp34 (β-strand 3), these three residues are important for ligand binding. The side chains of residues Trp34 and Arg17 are energetically most important for ligand binding, and Ala mutations at these positions lead to a 18- (Trp34) and 6-fold decrease (Arg17) in the binding constant.³⁹ Since these side chains only contribute marginally to the thermodynamic stability of hPin1 WW, it must be concluded that residues Arg17 and Trp34 have evolved primarily for function. A slight gain in binding affinity (approximately twofold) is observed upon Ala-mutation of residues Arg14 and Phe25, but this gain of function appears to be outweighed by a strong decrease in thermodynamic stability, rendering Arg14Ala and Phe25Ala mutants substantially denatured (>50%) at physiological temperature. Arg14 and Phe25, therefore, appear to have evolved to enhance stability without compromising peptide binding, for example the F25Y variant exhibits a WT K_d.⁴⁶

Spectroscopic studies

CD-spectroscopic measurements were made in 10 mM sodium phosphate, pH 7.0^7,16 with an AVIV model 202SF circular dichroism spectrometer (Lakewood, NJ), equipped with a Peltier temperature control system. Measurements were made at a protein concentration of 15 μM in 2 mm quartz cuvettes (Hellma, Forest Hills, NY).

The authors wish to thank Dr. Evan Powers for his help with the figures, Dr. Colleen Fearns for her critical reading and editing of the manuscript, as well as Dr. Houbi Nguyen and Dr. Martin Gruebele (Department of Chemistry, UIUC) for helpful discussions.