PEDS Advance Access published online on January 24, 2008
Protein Engineering Design and Selection, doi:10.1093/protein/gzm070
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Two-dimensional surface display of functional groups on a β-helical antifreeze protein scaffold
1Department of Structural Biology 2Chemical Research Support, Weizmann Institute of Science, Rehovot 76100, Israel
3 To whom correspondence should be addressed. E-mail: deborah.fass{at}weizmann.ac.il
 |
Abstract |
|---|
 Top Abstract Materials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
We tested a disulfide-rich antifreeze protein as a potential scaffold for design or selection of proteins with the capability of binding periodically organized surfaces. The natural antifreeze protein is a β-helix with a strikingly regular two-dimensional grid of threonine side chains on its ice-binding face. Amino acid substitutions were made on this face to replace blocks of native threonines with other amino acids spanning the range of β-sheet propensities. The variants, displaying arrays of distinct functional groups, were studied by mass spectrometry, reversed-phase high performance liquid chromatography, thiol reactivity and circular dichroism and NMR spectroscopies to assess their structures and stabilities relative to wild type. The mutants are well expressed in bacteria, despite the potential for mis-folding inherent in these 84-residue proteins with 16 cysteines. We demonstrate that most of the mutants essentially retain the native fold. This disulfide bonded β-helical scaffold, thermally stable and remarkably tolerant of amino acid substitutions, is therefore useful for design and engineering of macromolecules with the potential to bind various targeted ordered material surfaces.
Keywords: antifreeze protein/beta-helix/circular dichroism/disulfide bonds/protein folding
An aesthetic and captivating example of the relationship between protein structure and function is provided by the ‘antifreeze’ or thermal hysteresis proteins (THPs) (Davies et al., 2002
). In particular, the β-helical insect THPs display a highly ordered two-dimensional (2D) array of threonine residues, in which the positioning of the side chain hydroxyl groups mimics the spacing of oxygen atoms on various facets of ice crystals (Liou et al., 2000b). The THP ice-binding face, evolved for recognition of an inorganic crystalline material, differs from typical protein surfaces engaged in interactions with target molecules. When the targets are aperiodic macromolecules or small ligands, the binding surface presented by the protein is correspondingly rugged and varied. In contrast, the β-helical THP surface is broad, repetitive and flat.
Interactions of proteins with crystalline surfaces have been observed in biomineralization and several pathological conditions (Perl-Treves and Addadi, 1988; Mann et al., 1989; Giachelli, 2005; Gotliv et al., 2005). Although the number of inorganic materials shown to be natural targets for protein binding is growing, proteins known to recognize periodic inorganic structures are still rare. We hypothesize that other surface-binding functionalities are encoded in protein sequence space and that non-natural proteins able to recognize and distinguish repetitive 2D targets can be identified by design or selection. Along these lines, others have selected by phage display and related techniques peptides with purported ability to bind and distinguish non-biological surfaces (Brown, 1992; Whaley et al., 2000). The drawback to a random peptide-based approach is that binding may occur via general association of an ensemble of peptide conformations with the surface rather than due to selection of a particular structure. This conclusion is based on the predominance of certain functional groups but the lack of a true consensus sequence for peptides selected on, for example, semiconductor surfaces (Whaley et al., 2000).
We sought instead to take advantage of the fact that the β-helical THP domain is pre-evolved for recognition of a crystalline lattice. We chose for our study the Tenebrio molitor THP (TmTHP), a compact, right-handed β-helix composed of 12-amino acid repeats that form tight, internally disulfide bonded loops (Liou et al., 2000b) (Fig. 1). A Thr-Cys-Thr sequence from each repeat forms the strands of a parallel β-sheet spanning the length and width of the protein. The cysteines participate in disulfide bonds in the protein core, whereas the threonine side chains project outwards in an orderly array. Disulfide bridging and interstrand hydrogen bonding make the TmTHP protein particularly rigid (Daley et al., 2002; Daley and Sykes, 2004).
|
We probed the tolerance of the TmTHP fold to multiple amino acid substitutions on the ice-binding face. Previously, limited site-directed mutagenesis was done to study the resilience of antifreeze activity to mutation (Marshall et al., 2002
). In our study, we explore the sequence space of the disulfide-bonded β-helical fold of TmTHP regardless of its function as an antifreeze protein, with the expectation that specificity for other surfaces may replace ice-binding activity in certain mutants. We also address the contribution of the threonine grid to the architecture of this repeat protein. Our findings demonstrate that the TmTHP fold tolerates extensive mutagenesis, especially toward amino acids with at least moderate β-sheet propensity, thus providing a good basis for the design of proteins that bind periodic, 2D surfaces. |
Materials and methods |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Construction of mutants
Construction of mutants on the basis of the wild-type MBP-His6-TEV-TmTHP vector (Bar et al., 2006) was facilitated by introducing an MfeI restriction site upstream and an SphI site downstream of the region encoding the four Thr-Cys-Thr repeats in wild-type TmTHP. Synthetic templates were assembled by overlapping-primer extension reactions followed by nested PCR amplification using a mutagenic forward primer spanning the MfeI site and a reverse primer spanning the SphI site. The mutant fragments were reintroduced into the MBP-His6-TEV-TmTHP vector after MfeI/SphI double-digest.
Protein expression and purification
TmTHP mutants were purified essentially as the wild type (Bar et al., 2006). Briefly, fusion proteins were expressed at 15°C for 36–48 h in the Origami B (DE3) plysS Escherichia coli strain (Novagen), used to enhance disulfide bond formation in the cytoplasm. Proteins were purified on Ni-NTA agarose and cleaved with TEV protease (Kapust and Waugh, 2000). Apart from Ha–h and Hc-f, the mutants were reapplied to Ni-NTA after digestion to remove uncleaved material and the cleaved maltose-binding protein (MBP) carrying a C-terminal His6 tag. The Ha–h and Hc-f mutants were separated from MBP using gel filtration. Digested mutants were purified by high performance liquid chromatography (HPLC) on a C18 reversed phase column (218TP1010 from Vydac) with solvent B (90/10 v/v acetonitrile/water, 0.1% trifluoroacetic acid) gradient rates of 0.1 – 1.0% /min, as optimized for each mutant. Proteins eluted at 17 – 46% solvent B according to absorbance at 230 nm.
Protein concentrations were determined by amino acid analysis (Chemical Services, Weizmann Institute of Science) or using the BCA Protein Assay Kit (Pierce) with standard curves referenced to samples of pre-determined concentration. Reported concentrations have an estimated error of ±10%.
Circular dichroism (CD) spectra and thermal denaturation were measured on an Aviv (model 202) spectrophotometer with a 1 mm quartz cuvette at protein concentrations of 10–30 µM. Wavelength scans were performed at 20°C in 10 mM sodium phosphate buffer, pH 6.0. Ha–h samples at pH values between 5 and 9 were obtained by diluting a protein stock solution to 20 µM in 50 mM sodium phosphate buffer. High pH samples of Ha–h, Kc–f, Ka–h and wild type were prepared by dialysis against: (1) 10 mM K2HPO4 at pH 9.5; (2) 50 mM Na2CO3 at pH 10.7 and (3) 10 mM K2HPO4 adjusted to pH 12.5 with NaOH. The final pH was verified using indicator paper. Reduced TmTHP was prepared by incubating the protein for 3 h in 150 mM dithiothreitol (DTT), 20 mM sodium phosphate buffer, pH 8.0. DTT was removed on a PD-10 (GE Healthcare) column. Wavelength scan data were collected every 1 nm with a 3 s data averaging time, and eight scans were averaged. Thermal denaturation was conducted from 4°C to 90°C and back to 4°C in steps of 2°C with a 1 min equilibration and 3 s signal averaging time at each temperature.
NMR samples contained 2–5 mg/ml protein in 90% H2O/10% 2H2O, 15 mM sodium phosphate buffer, pH 6.0. Spectra were recorded at 303 K on a Bruker (Karlsruhe, DE) DMX-500 spectrometer equipped with a 5 mm TXI CryoProbe. 1D 1H NMR spectra were acquired using a spectral width of 14 ppm. 2D homonuclear Hartmann-Hahn spectra (HOHAHA) (Davis and Bax, 1985) were acquired using isotropic mixing time of 54 ms, 4096 complex data points in t2, and 256 increments in t1. Water suppression was achieved using the excitation sculpting technique (Hwang and Shaka, 1995). Data were processed and analyzed using the program TopSpin (Bruker).
|
Results |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Expression and purification of TmTHP mutants
The ice-binding face of TmTHP (Liou et al., 2000b) displays 10 threonine residues, eight of which fall into a 2 by 4 grid (Fig. 1B) that showed virtually no NMR chemical shift changes as a function of temperature within a given range (Daley et al., 2002; Daley and Sykes, 2004). We introduced simultaneous changes at grid positions, substituting either all eight threonines or the central four. In each mutant, the threonines were replaced by a single amino acid type, generating TmTHP variants of the form Xa-h or Xc-f in which X represents N, K, H, Y or V. These particular amino acids span the range of statistical β-sheet propensities (Fig. 1C) (Kyngas and Valjakka, 1998) and provide a diversity of side chain sizes and functional groups.
TmTHP variants were produced recombinantly in E. coli (Bar et al., 2006). The majority of the coding sequence was constructed anew for each variant due to the dispersion of mutation sites throughout the TmTHP gene. Coding sequences were inserted downstream of MBP, a His6 tag, and a TEV protease cleavage site. High yields (
40 mg fusion protein per liter cell culture) were obtained for all variants except for Ya–h (5-20 mg/l).
TmTHP mutants fused to MBP were analyzed by SDS–polyacrylamide gel electrophoresis (SDS–PAGE) (Fig. 2). A decrease in migration rate upon reduction (comparable with 5 kDa increase in molecular weight) was seen for most of the TmTHP mutants, as for wild type (Bar et al., 2006). This shift was ascribed to a change in hydrodynamic radius of the denatured fusion protein that takes place when the disulfide cross-links in the TmTHP portion are removed. Interestingly, the V and Y variants in the oxidized state migrated faster than wild-type TmTHP. These mutants are on the higher end of the β-sheet propensity scale and are more hydrophobic than wild type. Mutants with low β-sheet propensities showed some disulfide-mediated dimerization, as judged by a band around 110 kDa that disappeared upon reduction. Na-h in particular appeared as multiple bands between 55 and 65 kDa, implying a mixed population of isomers.
|
MBP fusion proteins were further processed by TEV protease cleavage. Apart from Ya–h, all TmTHP variants were soluble after removal of MBP; Ya–h precipitated directly upon addition of TEV and was readily re-solublized only in protein denaturants. Following TEV cleavage, the TmTHP variants were purified by reversed phase HPLC. Wild-type TmTHP was previously shown to elute as a single, sharp peak early in an acetonitrile gradient (Liou et al., 2000a
; Bar et al., 2006). The variants, however, showed widely different retention times and peak profiles (Fig. 3 and Supplementary Fig. 1 available at PEDS online). For example, Va–h eluted as a single major peak after MBP at 46% solvent B. In contrast, Na-h eluted as two broad peaks between 19% and 21% B. Both peaks contained oxidized protein according to mass spectrometry (Biological Services Department, Weizmann Institute of Science), but CD and NMR showed the second and smaller of the peaks to contain better folded material (Supplementary Fig. 2 available at PEDS online). In general, most variants eluted in one major peak, but when multiple TmTHP related peaks were obtained, the populations that yielded CD spectra most closely resembling that of wild type were isolated and characterized.
|
HPLC-purified proteins (Supplementary Fig. 3 available at PEDS online) were analyzed by mass spectrometry to confirm their identities and to assess oxidation state (Table I). For most mutants, the observed masses were within 2 Da of the calculated masses for fully oxidized protein. These results were supported by reaction with Ellman’s reagent (Ellman, 1959
). Using this assay, fewer than two thiols were detected per protein molecule, within experimental error of full oxidation (not shown).
|
Folding of TmTHP mutants
The CD spectrum of wild-type TmTHP showed a double minimum at 205 and 222 nm, in agreement with previous reports (Liou et al., 2000a; Bar et al., 2006). In contrast, the spectrum of reduced TmTHP showed a deep minimum at 198 nm, indicating that reduction of disulfides unfolds the protein (Fig. 4A, inset). The CD spectra of all tested TmTHP mutants, with the exception of Ka–h and Kc–f, had minima at 205 and 222 nm. Mutants with β-sheet propensity higher than threonine (Yc–f, Vc–f and Va–h) gave spectra nearly superposable with that of wild-type TmTHP (Fig. 4A). The other mutants differed from wild type in the ratio of the signals at 205 and 222 nm; Hc–f, Nc–f, Ha–h and Na–h had increasingly deep minima at 205 nm (Fig. 4B). The Ka–h and Kc–f spectra showed a single minimum at 200 nm (Fig. 4B, inset), suggesting that these mutants are largely unfolded. We could not reliably quantify the secondary structure composition of the TmTHP mutants using CD analysis software because these programs (Lobley et al., 2002; Whitmore and Wallace, 2004) fail to detect the true secondary structure of wild-type TmTHP, generally interpreting its spectrum as various combinations of 
-helices and unordered structures (not shown). Nevertheless, the CD spectra indicate that most TmTHP variants qualitatively retain the signature of the TmTHP β-helical fold.
|
Structures of Kc–f, Ka–h, and Ha–h as a function of pH
We hypothesized that pH may affect the tertiary structures of the H and the K mutants and that the K mutants might fold poorly due to charge repulsion. We therefore measured the CD spectra of Ha–h at various pH values (Fig. 4C) and of Ka–h at high pH to favor deprotonation of the lysine side chains. Between pH 5 and 6, Ha–h showed a sharp transition from low to high 222/205 signal ratios. As the pH was further increased, the ratio gradually decreased from its maximum at pH 6 or 7. For Kc–f and Ka–h, little difference was seen between the spectra at pH 6.0 and pH 12.5 (not shown). As the lysine side chains in these mutants are expected to be deprotonated at pH 12.5 (the pKa of an isolated lysine side chain is 10.5, and the surrounding positively charged environment on the mutated face should lower the pKa below this value), other factors aside from the concentration of positive charge on one face are likely to be responsible for the poor folding of Kc–f and Ka–h. For reference, the spectra of wild-type TmTHP at pH 6.0 and 10.7 are indistinguishable, and only a slight increase in the intensity of the band at 205 nm was observed at pH 12.5 (not shown).
Thermal stability of TmTHP mutants
The thermal stabilities of TmTHP mutants were measured by loss of CD signal at 222 nm as a function of temperature (Fig. 5). Sigmoidal transitions were observed except for Ka–h, Kc–f and Va–h, which did not show considerable change upon heating up to 90°C. The K mutants were probably unfolded between 4°C and 90°C, whereas Va–h was folded across this temperature range. The slopes of the folded and unfolded baselines of all mutants, when they could be determined, were close to zero before normalization (not shown). Thermal denaturation was largely reversible with virtually no hysteresis under the experimental protocol. Signal recovery at 222 nm was 100% for Vc–f, Yc–f and wild-type TmTHP, >75% for Nc–f and Hc–f, 60% for Na–h and 50% for Ha–h. The midpoint of the thermal transition (Tm) for wild-type TmTHP was 66°C. The Tm of most mutants fell within the narrow range between 56°C and 66°C. Exceptions were Vc–f, with a Tm of 78°C, Yc–f, with an estimated Tm of 80°C and Va–h with Tm >90°C. It should be noted that in a mixed population of well and poorly folded molecules, the thermal denaturation experiment monitors primarily the better folded fraction, which has a stronger signal at 222 nm.
|
NMR analysis
The tertiary structures of the mutants were assessed in more detail using NMR spectroscopy. All soluble mutants were tested by 1D 1H NMR for chemical shift dispersion. Only Ka–h and Kc–f showed poor dispersion and broad peaks between 1.0–4.5 and 7.0–8.5 ppm, indicative of unfolded material or lack of fixed tertiary structure (Fig. 6). All other mutants exhibited wide chemical shift dispersion and were further examined by 2D 1H HOHAHA NMR (Davis and Bax, 1985). Residue numbering and partial assignment of TmTHP mutant 1H chemical shifts were based on Daley et al. (2002).
|
The chemical shift dispersion observed in the NMR spectra of the Xc–f mutants (Yc–f, Hc–f, Nc–f and Vc–f) indicates that they are folded. 2D 1H HOHAHA spectra of these mutants are of excellent quality and typical of high-resolution spectra of <10 kDa, stable proteins. Superposition of the ‘fingerprint’ region of each mutant on that of wild type demonstrates the similarity in chemical shifts (Fig. 7A and Supplementary Fig. 4 available at PEDS online). Most residues show identical 1H chemical shifts, and any differences are usually <0.1 ppm. Even the amide proton resonance frequencies of the mutated residues differ only slightly from the wild type. All mutants in this series showed amide proton chemical shift differences of <0.15 ppm compared with wild type at the four mutated positions, with T38 and T50 showing smaller differences than T40 and T52 (Fig. 7A and Supplementary Fig. 4 available at PEDS online). In particular, Nc–f showed no appreciable change (<0.03 ppm) for all four mutation sites, and Hc–f showed only slight changes (<0.1 ppm). The chemical shifts of residue C39, located between two mutated residues in the third strand, and the chemical shifts of neighboring residues T37 and N53 also hardy changed, despite their proximity to the mutation sites, indicating that the backbone structure is maintained within the mutated strands. The chemical shift changes observed for the amide resonances of C27 and C63, located in the second and fifth β-strands, were slightly larger than the changes on the central strands. Interestingly, the amide proton of V49 was particularly sensitive to mutation of the threonine grid, showing chemical shift differences between 0.05 ppm (Vc–f) and 0.22 ppm (Nc–f) relative to wild type. Furthermore, distinct peaks were observed for the two
methyl groups of V49 in Hc–f and Yc–f, as opposed to the degeneracy previously seen for wild-type TmTHP. Nevertheless, these changes are small enough to suggest that the overall scaffold is retained.
|
The 2D NMR spectra of the Xa–h mutants Ha–h, Na–h and Va–h show wide chemical shift dispersion (Fig. 7B and Supplementary Fig. 4 available at PEDS online). The observed resonance line widths of Ha–h are typical of a well-folded protein of <10 kDa, although some broadening is seen. Furthermore, the individual HN/H
correlations in the Ha–h 2D spectrum are almost identical to wild type. Most changes of amide proton chemical shifts are <0.15 ppm relative to wild type, as observed for the Xc–f mutant series. Again, V49 was an exception, with a chemical shift difference of 0.22 ppm relative to wild type and no degeneracy for the chemical shifts of the two methyl groups. Although Ha–h peaks are generally sharp and well dispersed, there is an underlying broad, undefined region in the narrow chemical shift range defined by 8.1–8.6 ppm in one dimension and 3.9–4.6 ppm in the other.
The Va–h mutant spectrum showed more line broadening than Ha–h, but the individual correlations are still strikingly similar to those of wild-type TmTHP. The HN/H correlations are clearly observed and for the most part overlay those in the wild-type spectrum. However, some correlations between the HN and other aliphatic side chain protons are much weakened due to broadening. Interestingly, the poorly defined region is more limited than that of Ha–h, ranging from 8.15 to 8.4 and 3.9 to 4.3 ppm.
The spectrum of Na–h shows even more severe line broadening than Va–h, such that few of the peaks correlating the amide shifts to the side chain protons are observed. Nevertheless, the HN correlations still detectable in Na–h retain the positions seen in the wild-type spectrum. The poorly defined region extends even into the portion of the spectrum describing interactions between amide and aliphatic side chain protons (8.0–8.6 and 2.6–3.3 ppm).
|
Discussion |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
The sequence requirements for folding of the disulfide-bonded β-helical repeat proteins have not previously been investigated, in part due to the difficulty of producing even wild-type versions recombinantly (Tyshenko et al., 2006
). Motivated by the possibility that the binding target repertoire of the TmTHP fold could be expanded by protein engineering, we asked whether the β-branched threonines are necessary for proper folding of the protein or solely for forming the array of hydroxyls involved in ice mimicry. We investigated semi-systematically the range of amino acid residues that can be accommodated on the ice-binding face while maintaining the overall architecture of the protein.
Typically, disulfide bonding is considered a complicating feature in protein engineering. In fact, repeat proteins used to date for protein design and selection have generally lacked them (Main et al., 2005). We were not dissuaded from using an internally disulfide-bonded scaffold, however, in the expectation that the disulfides would stabilize populations of properly folded molecules. Moreover, thermal denaturation of TmTHP and its mutants was reversible, unlike other parallel β-helix proteins (Kamen et al., 2000; Buchko et al., 2006), and folded subpopulations would not exchange with misfolded species and could be isolated. Finally, disulfide bonding status could be used as an indicator of successful folding.
Disulfide mapping of all TmTHP mutants was prohibitively difficult due to the large number of cysteines and the repetitive amino acid sequence. We thus used alternative techniques to evaluate disulfide bond formation. Gel assays showed that MBP fusion proteins were obtained largely in monomeric form, despite the high protein yields and high cysteine content. Ellman’s assay indicated background levels of free sulfhydryl groups in all mutants within a day after cell lysis, consistent with the proteins being fully oxidized. Although many of the native disulfides are between sequential cysteines and therefore favored over long-range pairings, the 16 cysteines in TmTHP can in principle be paired in more than 2 million ways, only one of which is correct. The observation of dominant peaks by HPLC for most of the TmTHP mutants suggests bias toward a single or limited set of disulfide isomers. Together, these experiments indicate that a discrete, internally disulfide-bonded state is favored for all mutants tested. When a poorly folded population was detected in significant quantities (i.e. in the Na–h preparation), it was separated from the better folded population, which did not re-equilibrate once isolated.
Spectroscopic studies support the conclusion that most TmTHP mutants folded into a structure resembling wild-type TmTHP. CD is particularly powerful for assessing the global structure of the mutants because of the unique signature of TmTHP: a double minimum at 205 and 222 nm reminiscent of spectra of -helices (Li et al., 1998; Liou et al., 2000a; Bar et al., 2006). Such double minima are rarely observed in CD spectra of all-β proteins. Other β-helices give CD spectra (Sieber et al., 1995; Gauthier et al., 1998; Kamen et al., 2000; Buchko et al., 2006) resembling those of typical all-β proteins of the β-I class (Sreerama and Woody, 2003), with a single minimum between 210 and 220 nm. In the study of Sieber et al. (1995), the CD spectrum of each structural component within a β-helical protein was calculated individually. Although the spectrum anticipated solely for the β-helix motif has a double minimum, it is buried under other components such that the experimental spectrum had a single minimum at 216 nm. The double minimum may be evident in TmTHP because this protein has no other structural features apart from a β-helix. However, spruce budworm THP, which is also a nearly pure β-helix but left- rather than right-handed, lacks the double minimum and shows a single minimum at 216 nm (Gauthier et al., 1998). The differences between the CD spectrum of TmTHP and those of other β-helices can thus be explained by the differences in the β-helices themselves; the β-turns, disulfide bridges and extraordinarily tight loops of the TmTHP fold may contribute to its distinguishing CD fingerprint.
All the TmTHP mutants tested in this study, except Kc–f and Ka–h, gave CD spectra with the characteristic double minimum of TmTHP, demonstrating that the TmTHP structure is maintained in at least a significant fraction of the population or in a major part of each protein molecule. In particular, Vc–f and Va–h showed CD spectra almost identical to wild type (Fig. 4A). The Yc–f spectrum is also consistent with a backbone conformation unchanged relative to wild type (R.Woody, personal communication), as electronic transitions in the tyrosine side chains are expected to impact the CD spectrum at 190 and 230 nm (Woody, 1978). Interestingly, the Yc–f, Vc–f and Va–h mutants were extraordinarily thermally stable, which might be considered unusual for a protein containing 5–10% mutated residues. However, a high Tm is often observed for designed proteins (Binz et al., 2003; Kuhlman et al., 2003; Main et al., 2003) and may be useful for future applications of the proteins described herein. In agreement with the CD spectra and thermal melts, the rapid migration of these mutants fused to MBP (Fig. 2) suggests increased compactness and resistance to SDS-induced denaturation.
The CD spectra of Hc–f, Nc–f, Ha–h and Na–h, the mutants with lower β-sheet propensities, differed from wild type mostly in the intensities of the signals, with an increase in the 205 nm band and a decrease in the 222 nm band (Fig. 4B). Notably, the spectrum of the Dendroides canadensis antifreeze protein (Li et al., 1998), a TmTHP homolog, resembles the spectra of Hc–f and Nc–f more than the spectrum of wild-type TmTHP in the intensity of the 205 nm band. A single lower wavelength minimum (200 nm) is characteristic of the β-II class of all-β proteins. The β-I and β-II subclassification is based solely on CD spectra, but unifying features for each class may be sought in their high-resolution structures (Manavalan and Johnson, 1983; Sreerama and Woody, 2003). Structures of β-II proteins have β-sheets with very short, irregular strands or sheets that are extremely twisted. The β-strands of wild-type TmTHP are indeed very short, but the sheet is atypically flat. Accordingly, the increase of negative signal at 205 nm for some TmTHP mutants may be due to increased distortion or twisting in the β-sheets. Support for the retention of β-secondary structure is provided by the 2D-HOHAHA NMR spectra of these mutants, but this experiment does not address the degree of twisting. It is also plausible that the differences between the CD spectra arise from changes in the geometry of disulfide bonds or turns, which are dominant features of the β-helix motif. Another possible explanation for the increase in the negative band at 205 nm for TmTHP mutants is a fraction of unordered polypeptide, as CD spectra of random coils have a deep minimum around 198 nm (Manavalan and Johnson, 1983). This last possibility is supported by Kc–f and Ka–h, which have deep CD minima at 200 nm and were shown to be unfolded by NMR and thermal denaturation. In general, the β-II conformation can be distinguished from random coil by CD in the near-UV range (Wu et al., 2000). However, the signal of TmTHP and its mutants is low in this range due to the paucity of aromatic residues. The interpretations discussed above to explain the differences between the CD spectra of the mutants and wild type may also apply to changes in the CD spectrum of Ha–h at different pH values. It is likely that at lower pH (i.e. pH 5), protonation of the histidine side chains causes unfolding by charge repulsion. Close to neutral pH, Ha–h adopts a conformation most similar to wild-type TmTHP. The slight drop in apparent foldedness at higher pH suggests that the most favorable interactions between histidines on the grid occur with a mixture of protonated and deprotonated side chains.
Complementary to the CD analysis, we used 1H NMR to probe the TmTHP mutant structures on an individual residue basis. Proton chemical shift values are very sensitive to chemical environment and are therefore excellent indicators of subtle changes in structure. The 1D NMR spectra of Ka–h and Kc–f gave clear evidence that these proteins are unfolded. The 2D spectra of all other mutants, despite some broadening of the NMR signals, showed that most of the amide and H proton chemical shift values are very similar to those of the corresponding residues in wild-type TmTHP. The numerous downfield shifted H frequencies suggest that the corresponding residues are within β-sheets. Even Na–h, which shows the most severe line broadening, still seems to retain the overall tertiary structure and β-sheet content of the wild-type protein. Amide proton resonance frequencies are particularly sensitive to their own side chain identity and environment. In Yc–f, Hc–f, Nc–f and Vc-f, the amide proton chemical shifts corresponding to the mutated positions and neighboring residues (C27, T37, C39, N53 and C63) were hardly affected. The chemical environments around the mutation sites in these variants are therefore similar to wild type. In Ha–h, Na-h and Va–h, however, the mutated positions could not be unambiguously assigned, and the broad background suggests a fraction of poorly folded polypeptide. These observations are consistent with the CD spectra obtained for Ha–h and Na–h. However, Va–h is extremely thermally stable, and its CD spectrum is almost identical to that of wild type. The source of the broad background in this variant remains to be determined. Thus, apart from the discrepancy regarding Va–h, the NMR studies agree with the CD results, supporting retention of tertiary structure in all but the K mutants.
Ka–h and Kc–f, which appear to be anomalies according to the β-sheet propensity scale, in fact behave consistently with previous experimental measurements of β-sheet-forming preferences for residues at the edge of a β-sheet (Minor and Kim, 1994). The statistical β-sheet propensity scale lists lysine as more favorable than asparagine. In contrast, the experimentally determined 
G values relative to alanine on an edge strand of a model sheet place lysine/arginine at the bottom of the scale, above only glycine and proline (Minor and Kim, 1994). Although none of the mutated strands are edge strands in the parallel β-sheet, the TmTHP strands are so short that each surface-exposed residue is at an end position along the strand. Residues that project from the sheet have neighbors on only three sides rather than four, similar to an edge position, as opposed to an interior position, of a sheet. It is possible that the entropic price of restricting the side chain rotamers of lysine by forming the ordered array, without the favorable enthalpic interactions obtained by fully burying the aliphatic portion of the side chain, is too great to permit folding of these variants.
This study was conducted to determine if it is possible to diversify the functional groups on the TmTHP ice-binding surface without undermining the protein architecture. Modifying a folded protein scaffold may have advantages over selection of random peptides in the search for biomolecules that recognize organized 2D surfaces. Binding of ice crystals by THPs in a background of liquid water as competitor can be explained by geometrical complementarity between evenly spaced ranks of threonines on the protein and atoms in the ice lattice. Such shape complementarity is an important feature of diverse protein-target interfaces (Lawrence and Colman, 1993) and may provide the co-operativity between individual interactions that stabilizes the association. Studies of TmTHP dynamics by NMR suggest that the protein adopts the ice-matching structure in solution prior to recognition of the ice crystal (Daley et al., 2002). Backbone and side chain rigidity reduces the entropic barrier to ice binding, as the side chains need not be repositioned or restricted conformationally upon binding. A pre-arranged lattice of amino acid side chains may be similarly useful in binding other ordered surfaces. Because proteins with such rigid, repetitive ligand-binding sites are rarely observed in nature, we investigated whether other functional groups can be arranged in grid configuration on the TmTHP scaffold. Our panel of folded mutants includes 2 x 2 or 2 x 4 arrays of imidazyl, amide, phenyl (only 2 x 2) and isopropyl groups. Fixed arrays of primary amines (lysine side chains) were not obtained on the TmTHP backbone in solution. Nevertheless, the lysine mutants may adopt an ordered structure once a relevant surface is introduced. For example, electrostatic interactions between the lysine side chains and a negatively charged surface could help overcome the entropic barrier to folding of the polypeptide chain. Proteins that are highly charged and unfolded in solution are common in biomineralization and may assume restricted conformations upon adsorption on substrates and interaction with mineral (Addadi et al., 2001).
In conclusion, this study provides a panel of proteins with distinct arrays of functional groups restricted by a series of disulfide bridges. These proteins may be useful for systematic structural studies of β-helices and of highly disulfide bonded folds. Furthermore, our engineered proteins have the potential for specific recognition of ordered arrays of particular atoms, molecules or ions. For instance, Ha–h and Hc-f mutants may adsorb onto CdS, CdSe, ZnS or Au, but with higher specificity than observed for cell-surface displayed hexahistidine peptides (Peelle et al., 2005). Va–h and Vc-f may bind to hydrophobic groups of organic crystals and form facets that are usually buried in aqueous solutions. A variety of additional mutants can be engineered on the basis of the TmTHP scaffold and selected for specific molecular recognition of other materials. Of special interest would be technologically useful substances such as semiconductors. Moreover, specific binding can be sought not only for certain materials but also for specific crystal planes. Whaley et al. (2000) searched for peptides that discriminate between different crystallographic orientations of semiconductors. However, short oligomers can adopt various conformations in solution and lose specificity. Proteins, on the other hand, have less conformational freedom and therefore are advantageous for specific recognition of 2D surfaces if the correct amino acids for such interactions are properly positioned. Studies have shown that antibodies or antibody fragments can be selected for specific recognition of particular crystal planes (PerlTreves et al., 1996; Bromberg et al., 1998; Artzy Schnirman et al., 2006). We hypothesize that the TmTHP scaffold is a promising starting point for generation of proteins with high affinity and selectivity to molecularly defined surfaces. Proteins with specific surface binding properties can potentially serve in various micro- and nanotechnological applications, including assembly of nanoparticles and control over crystal growth and shape. Whether any of the TmTHP mutants have acquired the ability to recognize an inorganic material surface remains to be addressed.
|
Funding |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Clore Center for Biological Physics at the Weizmann Institute of Science.
|
Footnotes |
|---|
Edited by Dr Michael Hecht
|
Acknowledgements |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
We thank Robert Woody and Nicholas Price for aid in interpretation of CD results. We also thank Lia Addadi for helpful discussions and Dan Minor for critical reading of the manuscript.
|
References |
|---|
|
TopAbstractMaterials and methodsResultsDiscussionFundingAcknowledgementsReferences |
|---|
Addadi L., Weiner S., Geva M. Z. Kardiol. (2001) 90(Suppl. 3):92–98.
Artzy Schnirman A., Zahavi E., Yeger H., Rosenfeld R., Benhar I., Reiter Y., Sivan U. Nano Lett. (2006) 6:1870–1874.[CrossRef][Web of Science][Medline]
Bar M., Bar-Ziv R., Scherf T., Fass D. Protein Expr. Purif. (2006) 48:243–252.[Web of Science][Medline]
Binz H.K., Stumpp M.T., Forrer P., Amstutz P., Pluckthun A. J. Mol. Biol. (2003) 332:489–503.[CrossRef][Web of Science][Medline]
Bromberg R., Kessler N., Addadi L. J. Cryst. Growth (1998) 193:656–664.[CrossRef]
Brown S. Proc. Natl Acad. Sci. USA (1992) 89:8651–8655.
Buchko G.W., Ni S., Robinson H., Welsh E.A., Pakrasi H.B., Kennedy M.A. Protein Sci. (2006) 15:2579–2595.[CrossRef][Web of Science][Medline]
Daley M.E., Sykes B.D. J. Biomol. NMR (2004) 29:139–150.[CrossRef][Web of Science][Medline]
Daley M.E., Spyracopoulos L., Jia Z., Davies P.L., Sykes B.D. Biochemistry (2002) 41:5515–5525.[CrossRef][Web of Science][Medline]
Davies P.L., Baardsnes J., Kuiper M.J., Walker V.K. Philos. Trans. R. Soc. Lond. B Biol. Sci. (2002) 357:927–935.
Davis D.G., Bax A. J. Am. Chem. Soc. (1985) 107:2820–2821.[CrossRef][Web of Science]
Ellman G.L. Arch. Biochem. Biophys. (1959) 82:70–77.[CrossRef][Web of Science][Medline]
Gauthier S.Y., Kay C.M., Sykes B.D., Walker V.K., Davies P.L. Eur. J. Biochem. (1998) 258:445–453.[Web of Science][Medline]
Giachelli C.M. Orthod. Craniofac. Res. (2005) 8:229–231.[CrossRef][Medline]
Gotliv B.A., Kessler N., Sumerel J.L., Morse D.E., Tuross N., Addadi L., Weiner S. Chembiochem (2005) 6:304–314.[CrossRef][Web of Science][Medline]
Hwang T.L., Shaka A.G. J. Magn. Reson. A (1995) 112:275–279.[CrossRef][Web of Science]
Kamen D.E., Griko Y., Woody R.W. Biochemistry (2000) 39:15932–15943.[CrossRef][Web of Science][Medline]
Kapust R.B., Waugh D.S. Protein Expr. Purif. (2000) 19:312–318.[CrossRef][Web of Science][Medline]
Kuhlman B., Dantas G., Ireton G.C., Varani G., Stoddard B.L., Baker D. Science (2003) 302:1364–1368.
Kyngas J., Valjakka J. Protein Eng. (1998) 11:345–348.
Lawrence M.C., Colman P.M. J. Mol. Biol. (1993) 234:946–950.[CrossRef][Web of Science][Medline]
Li N., Kendrick B.S., Manning M.C., Carpenter J.F., Duman J.G. Arch. Biochem. Biophys. (1998) 360:25–32.[CrossRef][Web of Science][Medline]
Liou Y.C., Daley M.E., Graham L.A., Kay C.M., Walker V.K., Sykes B.D., Davies P.L. Protein Expr. Purif. (2000) a 19:148–157.[CrossRef][Web of Science][Medline]
Liou Y.C., Tocilj A., Davies P.L., Jia Z. Nature (2000) b 406:322–324.[CrossRef][Medline]
Lobley A., Whitmore L., Wallace B.A. Bioinformatics (2002) 18:211–212.
Main E.R., Xiong Y., Cocco M.J., D’Andrea L., Regan L. Structure (2003) 11:497–508.[Medline]
Main E.R., Lowe A.R., Mochrie S.G., Jackson S.E., Regan L. Curr. Opin. Struct. Biol. (2005) 15:464–471.[CrossRef][Web of Science][Medline]
Manavalan P., Johnson W.C. Nature (1983) 305:831–832.[CrossRef]
Mann S., Webb J., Williams R.J. Biomineralization—Chemical and Biochemical Perspectives (1989) New York: Weinheim, Federal Republic of Germany press. 133–156.
Marshall C.B., Daley M.E., Graham L.A., Sykes B.D., Davies P.L. FEBS Lett. (2002) 529:261–267.[CrossRef][Web of Science][Medline]
Minor D.L. Jr, Kim P.S. Nature (1994) 371:264–267.[CrossRef][Medline]
Peelle B.R., Krauland E.M., Wittrup K.D., Belcher A.M. Langmuir (2005) 21:6929–6933.[CrossRef][Web of Science][Medline]
Perl-Treves D., Addadi L. Proc. R. Soc. Lond. B Biol. Sci. (1988) 235:145–159.[Medline]
PerlTreves D., Kessler N., Izhaky D., Addadi L. Chem. Biol. (1996) 3:567–577.[CrossRef][Web of Science][Medline]
Sieber V., Jurnak F., Moe G.R. Proteins (1995) 23:32–37.[CrossRef][Web of Science][Medline]
Sreerama N., Woody R.W. Protein Sci. (2003) 12:384–388.[CrossRef][Web of Science][Medline]
Tyshenko M.G., d’Anjou M., Davies P.L., Daugulis A.J., Walker V.K. Protein Expr. Purif. (2006) 47:152–161.[CrossRef][Web of Science][Medline]
Whaley S.R., English D.S., Hu E.L., Barbara P.F., Belcher A.M. Nature (2000) 405:665–668.[CrossRef][Medline]
Whitmore L., Wallace B.A. Nucleic Acids Res. (2004) 32:W668–W673.
Woody R.W. Biopolymers (1978) 17:1451–1467.[CrossRef][Web of Science]
Wu J., Yang J.T., Wu C.S. Anal. Biochem. (2000) 200:359–364.[CrossRef]
Received September 3, 2007; revised November 1, 2007; accepted November 5, 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||