Modulation of zinc- and cobalt-binding affinities through changes in the stability of the zinc ribbon protein L36
- First Online:
- Cite this article as:
- Kou, W., Kolla, H.S., Ortiz-Acevedo, A. et al. J Biol Inorg Chem (2005) 10: 167. doi:10.1007/s00775-005-0625-9
Cysteine-rich Zn(II)-binding sites in proteins serve two distinct functions: to template or stabilize specific protein folds, and to facilitate chemical reactions such as alkyl transfers. We are interested how the protein environment controls metal site properties, specifically, how naturally occurring tetrahedral Zn(II) sites are affected by the surrounding protein. We have studied the Co(II)- and Zn(II)-binding of a series of derivatives of L36, a small zinc ribbon protein containing a (Cys)3His metal coordination site. UV–vis spectroscopy was used to monitor metal binding by peptides at pH 6.0. For all derivatives, the following trends were observed: (1) Zn(II) binds tighter than Co(II), with an average KAZn/KACo of 2.8(±2.0)×103; (2) mutation of the metal-binding ligand His32 to Cys decreases the affinity of L36 derivatives for both metals; (3) a Tyr24 to Trp mutation in the β-sheet hydrophobic cluster increases KAZn and KACo; (4) mutation in the β-hairpin turn, His20 to Asn generating an Asn-Gly turn, also increases KAZn and KACo; (5) the combination of His20 to Asn and Tyr24 to Trp mutations also increases KAZn and KACo, but the increments versus C3H are less than those of the single mutations. Furthermore, circular dichroism, size-exclusion chromatography, and 1D and 2D 1H NMR experiments show that the mutations do not change the overall fold or association state of the proteins. L36, displaying Co(II)- and Zn(II)-binding sensitivity to various sequence mutations without undergoing a change in protein structure, can therefore serve as a useful model system for future structure/reactivity studies.
KeywordsZinc ribbonMetal binding constantsTotal correlation spectroscopyNuclear Overhauser enhancement spectroscopyThiolate-rich zinc site
Electrospray ionization mass spectrometry
High performance liquid chromatography
Inductively coupled plasma mass spectrometry
Ligand field stabilization energy
Ligand-to-metal charge transfer
Nuclear Overhauser enhancement spectroscopy
Total correlation spectroscopy
Catalytic Zn(II) sites contain carboxylate-rich and/or imidazole-rich ligation environments with an open coordination site for the solvent. The Zn(II) functions as a Lewis acid to activate a nucleophile (typically the coordinated solvent).
Structural Zn(II) sites stabilize a protein’s native fold or a protein–protein interface. These sites are thiolate-rich and are typically coordinately saturated.
We are interested in how the protein environment ultimately controls the properties (i.e., reactivity, metal affinities, etc.) of thiolate-rich Zn(II)-binding sites, causing some to be only structural, while others with similar coordination environments perform chemistry. Previous computational [11, 12] and small molecule model studies [13–16] suggest that both the first ligation shell and the surrounding environment can play important roles in determining these properties. Our complementary approach is to develop a simple model system through reengineering a small, naturally occurring Zn(II)-binding motif, modifying either the zinc ligands or distant sites in the protein, and then characterizing their metal-binding and reactivity properties. Previous Zn(II)-binding proteins studied this way include carbonic anhydrase II , TFIIIA-like zinc fingers [18, 19] and the phage T4 gene 32 protein .
The protein is small (37 residues), simplifying synthesis and characterization.
The solution structure is known (PDB no. 1DFE) . L36 adopts a zinc ribbon folding motif containing a three-stranded β-sheet with a (Cys)3His Zn(II)-binding site formed by two loops at one end of the protein (Fig. 1).
The metal ligands display different solvent accessibilities in the folded protein, providing a potential variance in reactivities.
Only one chromophore (Tyr24) is present, and is part of one of two small hydrophobic clusters that likely enable the protein to fold (Val7, Val16, Val23 and Val25 on one side of the β-sheet; Ile17, Tyr24 and Ile26 on the other side). Replacement of Tyr with Trp could provide a useful future reporter of local environmental changes via fluorescence/absorbance changes.
The Zn(II)-binding site contains several backbone-to-Cys NH–S hydrogen bonds , similar to those observed in other Zn(II)-binding proteins [24–26]. Protein modifications that remove/moderate their strength would allow us to test their importance in determining structural and reactivity properties.
A previous study reported that L36 binds Co(II) preferentially over Zn(II) .
L36 derivatives and amino acid sequences
Materials and methods
Peptide synthesis and purification
Peptides were synthesized at a 0.1 mmol scale with an Applied Biosystems 433A solid-phase peptide synthesizer equipped with an Alltech model 450 UV detector using 9-fluorenylmethoxycarbonyl (Fmoc)-protected amino acids and standard solid-phase peptide synthetic methods . Fmoc-amide resin was used, which produces an amide-protected C-terminus upon peptide cleavage from the resin. The N-terminus of each peptide was acetylated prior to cleavage. Cleavage of the peptide from the resin and side chain deprotection were effected using a 9:0.5:0.3:0.2 trifluoroacetic acid (TFA):thioanisole:ethanedithiol:anisole mixture. Crude peptides were precipitated with cold diethyl ether, lyophilized to dryness, and purified through reversed-phase high-performance liquid chromatography (HPLC) using a Waters Delta 600 preparative HPLC instrument and Vydac C4 columns with doubly distilled deionized (DI) water/acetonitrile/0.1% TFA mobile phase gradients. During purification, excess tris[2-carboxyethyl]phosphine hydrochloride (TCEP·HCl; Pierce) was added to the crude peptide solution, and the fractions containing pure peptide were frozen immediately to limit Cys side chain oxidation. The identities of the peptides were verified using electrospray ionization mass spectrometry (HT Laboratories, San Diego, CA, USA).
Preparation of peptide, Co(II) and Zn(II) solutions
All peptide stock solutions were prepared by resuspending lyophilized peptide solids in doubly distilled DI water under anaerobic conditions. Peptide concentration was determined spectrophotometrically with a Cary 50 UV–vis spectrophotometer using the absorbance of Tyr (ε275=1,420 M−1 cm−1) or Trp (ε280=5,600 M−1 cm−1). Co(II) and Zn(II) stock solutions were prepared by dissolving CoCl2·6H2O (Mallinckrodt) or ZnCl2 (EM Science), respectively, in degassed, doubly distilled DI water. Co(II) and Zn(II) concentrations in the metal stock solutions were measured by flame atomic absorption spectroscopy using a Varian SpectrAA-5 atomic absorption spectrophotometer. For Co(II), a cobalt hollow cathode lamp (Fisher Scientific) was used with a detection wavelength of 240.7 nm and a slit width of 0.2 nm. For Zn(II), a zinc hollow cathode lamp (Fisher Scientific) was used with a detection wavelength of 213.9 nm and a slit width of 1.0 nm. The background concentrations of Co(II), Zn(II) and other metals in the DI water were measured by inductively coupled plasma mass spectrometry (ICP-MS) and were low enough to not influence the results significantly. The concentrations were as follows: Zn, 5.58 ppb (8.5×10−2 μM); Co, below detection limits, or less than 0.10 ppb (less than 1.7×10−3 μM); Cd, 0.23 ppb (2.1×10−3 μM); Fe and Ni, below detection limits.
Free thiol quantitation
The number of free thiols in apo-peptide solutions was determined using a 5,5′-dithiobis(2-nitrobenzoic acid) (DTNB; Acros Organics) colorimetric assay [29, 30]. All solutions were prepared and reactions performed under anaerobic conditions. A homocysteine calibration curve was generated using standards between 20 and 100 μM prepared from a 100 μM stock in 0.1 M NaH2PO4·H2O buffer (pH 8.0). To each standard, 50 μl of a 0.01 M DTNB solution was added and the resulting mixture was incubated for 30 min. The absorbance was measured between 240 and 600 nm and corrected for the buffer absorbance. The free thiol concentration was determined using ε410=13,650 M−1 cm−1 for the aromatic thiolate ion produced from the reaction of DTNB with reduced thiols. Peptide samples were prepared in 0.1 M NaH2PO4·H2O buffer (pH 8.0) with peptide concentrations between 20 and 80 μM, and were then treated with DTNB for 30 min. The resulting absorbance at 410 nm was compared with the calibration curve to determine the free thiol content.
Co(II) titrations were carried out at 25°C in 20 mM N-(2-hydroxyethyl)piperazine-N′-ethanesulfonic acid (HEPES) buffer (pH 6.0) under anaerobic conditions. Three CoCl2 stock solutions with varying metal concentrations were used to reduce dilution effects. For each titration point, an appropriate volume of stock CoCl2 solution was added to an approximately 100 μM peptide solution and allowed to stand for 2 min (control experiments demonstrated no change in absorbance of a mixture after 2 min—data not shown). Absorbance spectra were collected from 250 to 900 nm at a scanning rate of 1 nm s−1 for each sample using a 1.0-cm quartz cuvette. Three titrations were conducted for each peptide. Spectra were corrected for buffer and peptide absorbance and for dilution from the addition of Co(II) stock. For all titrations reported in this paper, the shape of the d–d transition region (500–750 nm) did not change during the course of the experiment, indicative of a single type of tetrahedral environment present.
For each titration, the Aλ versus [Co(II)]tot data were best fit using a model where two species (SP1 and SP2) are present, each of which can bind Co(II) in a 1:1 stoichiometry with a different binding affinity (tight binder SP1 and weak binder SP2; see “Results” section for details). Fits were performed using the program DynaFit (BioKin), which simulates multiple complex equilibria to perform nonlinear least-squares regression. Experimental parameters included the percentage of SP1 and SP2 in the titration (based on DTNB titration results; percent SP1 equated with the percent reduced, percent SP2 equated with percent oxidized). Fitted parameters included KDCo [dissociation constant for Co(II)] for species SP1 and SP2, and the extinction coefficient (ελ) for the weaker binder SP2 (ελ for SP1 could principally be defined from the titration data). KACo was calculated by taking the inverse of KDCo.
Zn(II) binding was monitored using a Zn(II) versus Co(II) competition experiment [20, 31]. First, a saturated Co(II)-bound peptide sample was generated by the addition of 100 equiv or more of Co(II) from a Co(II) stock solution to a peptide solution (between 50 and 100 μM) in 20 mM HEPES buffer (pH 6.0) at 25°C under anaerobic conditions. For each titration point, an appropriate volume of stock Zn(II) solution was added to the Co(II)-peptide solution and allowed to stand for at least 1 min. Bleaching of the Co(II) signal for each sample was monitored by collecting absorbance spectra from 250 to 900 nm at a scanning rate of 1 nm s−1 using a 1.0-cm quartz cuvette. Spectra were corrected for buffer and peptide absorbance and for dilution from the addition of the Zn(II) stock.
CD spectral measurements were made using a sealed 0.1-cm rectangular quartz cuvette and an Aviv Instruments model 202 CD spectrometer equipped with a thermoelectric sample changer to control the temperature. Samples were prepared under anaerobic conditions and contained 100 μM peptide in 20 mM HEPES buffer (pH 6.0). When appropriate, aliquots from a Zn(II) stock were added to yield a final metal concentration between 150 and 200 μM. All CD data were corrected for buffer contributions by subtracting a buffer blank. Wavelength scans were collected from 190 to 320 nm at 25°C with a bandwidth of 1 nm and averaging for 2 s nm−1.
Size-exclusion chromatography (SEC) was used to determine the association state of Zn(II)-bound derivatives of L36. Zinc-bound protein samples (100 μM) were prepared in 200 mM HEPES buffer (pH 6.0) under anaerobic conditions. For each sample, 4 μl was loaded onto a Tosoh Super SW2000 SEC column (4.6 mm×300 mm; 4-μm particle size) attached to an HP1100 analytical HPLC. Chromatography runs were performed under isocratic conditions at a flow rate of 0.35 ml min−1 and monitored at 280 nm (350 nm reference) using a diode-array detector. The column was calibrated using horse heart cytochrome c (12.4 kDa), bovine lung aprotinin (6.5 kDa) and vitamin B12 (1.4 kDa).
NMR spectra were collected on a 500 MHz Varian Unity-INOVA spectrometer equipped with a z-axis gradient. Samples were prepared under anaerobic conditions by dissolving dry peptide in 90% H2O/10% D2O containing 2.0 mM Zn(II) to give a final peptide concentration of 1.0 mM. The sample pH was adjusted to 6.0 with NaOH, with the peptide acting as the sole buffer. NMR experiments were performed at 25°C. The 1D experiments and 2D nuclear Overhauser enhancement spectroscopy (NOESY) experiments employed WATERGATE for suppression of the water signal . Total correlation spectroscopy (TOCSY) experiments used presaturation for water suppression and an MLEV-17 sequence for the spin lock . The 2D NOESY and TOCSY experiments used sweep widths of 6,300.2 Hz in the direct and indirect dimensions, and collected 2,048 complex points (real plus imaginary) in the direct dimension and typically 150 (TOCSY) or 200 (NOESY) complex points in the indirect dimension. The data were processed using FELIX (Accelyrys) and analyzed with NMRView . For the quantitation of NOESY cross peak volumes, NOESY data were processed identically and apodized with sinebell squared window functions that were shifted by 90° in each dimension.
Selection of derivatives
Mutation of the binding site. The C3H group of derivatives contain the native (Cys)3His Zn(II)-binding site, while the members of the C4 group each contain a His32-to-Cys mutation, yielding a (Cys)4 coordination site; these two groups allow us to compare S3N and S4 metal ligation.
Mutation of the β-hairpin turn. L36 contains a single type I′ β-hairpin turn between β-strands 1 (residues 15–19) and 2 (residues 22–26), with His and Gly at positions i+1 and i+2 of the turn (Fig. 1). The derivatives C3H_N and C4_N contain a His20-to-Asn mutation, which is expected to stabilize a type I′ β-turn [36–38]. We expect mutations that stabilize the protein fold to lead to higher metal-binding affinities, as observed for β-sheet mutations in traditional zinc finger domains [39, 40]. A stabler fold should also lead to a decrease in the reactivity of a subset of the Cys ligands that become more effectively buried as the protein folds .
Mutation in the hydrophobic cluster. Tyr24 is part of a hydrophobic cluster on one face of the three-stranded β-sheet in L36. C3H_W and C4_W contain a Tyr24-to-Trp mutation, which introduces a better fluorescent probe for future folding studies. We are interested in the effect this larger aromatic residue will have on the hydrophobic cluster.
For the C4 derivatives, the replacement of His32 with Cys leads to redshifted and broadened d–d transition regions (Fig. 2). The formation of a shoulder at 732 nm is consistent with spectral changes observed in tetrahedral Co(II) compounds when a non-thiolate ligand is replaced with an additional thiolate ligand (Fig. 2b) [20, 46]. The d–d transitions are observed at 614 nm (average ε=230 M−1 cm−1), a maximum at 688 nm (average ε=370 M−1 cm−1), and 732 nm (average ε=300 M−1 cm−1. The more broadened d–d transition envelope is consistent with either a greater distortion from tetrahedral symmetry than for the C3H coordination site, or with an averaging of conformations . A change is also observed in the LMCT band intensities at 305 nm (average ε=4,060 M−1 cm−1) and 350 nm (average ε=1,800 M−1 cm−1) and in their broadness, attributable to an increase in the number of coordinating thiolate ligands from 3 to 4 (Fig. 2a) . The changes in the UV–vis spectra are therefore consistent with the designed change from a (Cys)3His Co(II)-binding site in the C3H derivatives to a (Cys)4 site in the C4 derivatives.
Co(II) versus Zn(II) specificity
To observe Zn(II) binding and measure Zn(II)-binding constants, we used a metal competition assay in which Zn(II) is added to a Co(II)-saturated peptide sample. The Zn(II) is expected to displace the Co(II) and bind to the same coordination site. The addition of Zn(II) led to a bleaching of the Co(II)–peptide absorbance (LMCT and d–d transition regions), with a corresponding increase in the octahedral Co(II) spectrum, consistent with Co(II) being replaced by Zn(II) in the tetrahedral binding site (Fig. 3b).
Co(II) and Zn(II) metal binding constants for L36 derivatives
For Zn(II)-binding titrations, νCo was plotted versus [Zn(II)]tot and the data were fit using a competitive inhibitor model which assumes that all peptide is in 1:1 metal:peptide complexes bound by either Co(II) or Zn(II), with effectively no free peptide in solution (Figs. 4b and S1)1. The presence of two isobestic points in the Zn(II) titrations at approximately 420 and 560 nm (Fig. 3) demonstrates the conversion of Co(II)-peptide to Co(H2O)62+ with no Co(II) intermediates. The Zn(II)-binding constants for all L36 derivatives are shown in Table 2.
Zn(II) binds tighter than Co(II). All L36 derivatives display a significant preference for Zn(II) over Co(II), with an average KAZn/KACo of 3(±2)×103 (Table 2). This selectivity for Zn(II) over Co(II) has been observed for other tetrahedral CxH(4−x) sites in proteins, with literature values for KAZn/KACo ranging from 3.1×102 at pH 7.5 for gp32 (with a HC3 binding site) to 3.2×105 at pH 7.0 for CP-1 (with a C4 binding site) [19, 20, 40, 47, 48].
C3H coordination sites bind both Co(II) and Zn(II) tighter than C4 sites. For each peptide derivative, it can be seen that the C3H metal-binding site has a higher affinity for Co(II) and Zn(II) than the corresponding C4 site. For example, the (KACo)C3H deriv/(KACo)C4 deriv values are 8.6±6.5 for C3H_N versus C4_N, 5.6±4.5 for C3H_W versus C4_W, and 16±13 for C3H_NW versus C4_NW. This trend has been reported previously for carbonic anhydrase-II , the zinc finger domain gp32 , and a consensus zinc finger sequence .
Tyr24 to Trp mutation leads to an increase in metal-binding constants. The replacement of Tyr with the larger and more hydrophobic side chain Trp yields an increase in the metal-binding constants. (KACo)C3H_W/(KACo)C3H is 20±15, and (KAZn)C3H_W/(KAZn)C3H is 24±17. Aromatic residues are also found in analogous positions in the β-sheet of other small β-sheet proteins, including the C-terminal zinc ribbon domain of human transcription elongation factor TFIIS (PDB no. 1TFI)  and the WW domain , suggesting a potential role as aromatic “anchors” or cores to help stabilize hydrophobic clusters.
His20 to Asn mutation in the β-hairpin loop increases the Co(II)- and Zn(II)-binding constants The His-Gly turn in the three-stranded β-sheet was mutated to Asn-Gly to stabilize the β-hairpin turn and potentially increase the metal-binding affinity. This is reflected in the binding constants, with (KACo)C3H_N/(KACo)C3H equal to 6.7±5.0 and (KAZn)C3H_N/(KAZn)C3H equal to 16±12.
The oligomeric state of L36 derivatives was assessed using SEC. We looked specifically at the C3H derivatives, since these peptides bind Zn(II) tighter than the C4 peptides and are therefore expected to be the most resistant towards oxidation . Each derivative was eluted as a single symmetrical peak (data not shown). The apparent molecular masses based on a globular protein calibration curve, with the calculated monomer molecular masses in parentheses, are as follows (all in grams per mole): C3H, 5,270 (4,540); C3H_N, 5,310 (4,504); C3H_W, 4,800 (4,550); C3H_NW, 5,440 (4,493). All Zn(II)-bound derivatives are approximately monomeric under the same pH and concentration conditions used in the metal-binding studies.
Zn(II) binds tighter than Co(II) The lower affinity of the L36 derivatives for Co(II) versus Zn(II), which is demonstrated by the ability of Zn(II) to displace Co(II) in competition titrations, has been assigned in large part to a loss of ligand field stabilization energy (LFSE) for Co(II) upon going from the octahedral Co(H2O)62+ to the tetrahedral Co(II)–peptide complex [32, 59]. In contrast, Zn(II) has no LFSE for any coordination geometry, so there is no change when Zn(II) goes from an octahedral to a tetrahedral complex. The KACo and KAZn values determined for the L36 peptides are slightly smaller than many reported in the literature, but compare favorably to those measured at similar pHs [48, 60]. Interestingly, our measured preference of Zn(II) over Co(II) for L36 is opposite to that observed by Boysen et al.  for a 26-residue derivative of the same protein (residues 9–34). We have tested the same 26-mer used by Boysen et al. and observe no difference in metal specificity between full-length L36 (C3H) and this 26-residue derivative (see C3H_26 in Table 2). Although it is not clear why this discrepancy exists, it should be noted that in their study, they observed Zn(II)-dependent changes in the CD of the 26-mer consistent with our results, yet they saw no significant population of Zn(II)-bound 26-mer under their electrospray ionization mass spectrometry (ESI-MS) conditions used to measure relative metal-binding affinities, even in the absence of competing metals. This calls into question the ability of their ESI-MS conditions to obtain accurate relative metal-binding information.
The C3H coordination site binds both Co(II) and Zn(II) tighter than the C4 site. We observed an increase in the specificity for binding of Zn(II) over Co(II) as the number of thiolate ligands was increased from 3 to 4 (Table 2). This is consistent with differences observed in several other naturally occurring Zn(II)-binding proteins [17, 20, 60] and designed sites , and is attributed to a decrease in the energy split between the e and t2 sets of d orbitals in a tetrahedral field .
Co(II)- and Zn(II)-binding is increased when His20 is replaced with Asn in the β-hairpin turn. It is becoming increasingly evident that turns in proteins can play important roles as nucleation sites for protein folding , and can also affect folding and unfolding rates [63, 64]. It would be interesting to observe how different turn sequences affect the interactions of metal cofactors with metal-binding domains. Blasie and Berg  and Kim and Berg  have demonstrated stabilization of the native protein fold in zinc finger domains through incorporation of stabilizing interstrand side chain/side chain contacts, which also leads to increases in the measured metal-binding constants. The metal-binding and protein-folding processes are thermodynamically coupled, since the zinc finger is only folded in the presence of Zn(II) or Co(II) . In other words, a change in the equilibrium metal-binding constant that results from a change in an amino acid not directly involved in metal chelation is correlated with a change in ΔGfolding (conformational stability of the protein) [39, 40].
In summary, we have studied the Co(II) and Zn(II) binding of a series of derivatives of the zinc ribbon protein L36. CD, SEC, and 1D and 2D 1H NMR experiments have shown that the L36 motif is stable towards mutations introduced into the β-hairpin turn, a hydrophobic cluster on the β-sheet, and the tetrahedral coordination environment of the metal—no changes in the overall fold or association state of the various proteins are observed. The derivatives, however, do display changes in the Co(II)- and Zn(II)-binding constants, with an increase observed when the mutation is expected to stabilize the protein fold. These properties should make L36 a useful model system for future structure/reactivity studies.
Although one can envision a more thorough treatment with two metals [Co(II) and Zn(II)] competitively binding to two different protein forms (SP1 and SP2), the number of variables involved makes extracting useful information impossible. This model can be greatly simplified by the assumption that the SP1 and SP2 protein forms both have a similar ratio of KAZn to KACo, making them indistinguishable in the competitive titration. The validity of this assumption is supported by the quality of fits obtained with Eq. 2, which assumes both SP1 and SP2 have the same ratio of KAZn to KACo. It is further supported by the consistency of the trends obtained both internally and with literature on Zn-binding proteins, and that no change in the shape of the spectrum occurs during the course of the titration, suggesting that both protein forms have the same set of ligands and ligand geometry (possible if SP2 binds metals as an SP2–SP2 dimer).
We wish to thank Torleif Härd for providing us with the original NMR assignments for L36, Piu Zhao for assistance with the NMR studies, and Adam Lyons for assistance with peptide purification. Financial support for this work was provided by the American Chemical Society Petroleum Research Fund (PRF no. 40835-AC3; G.R.D.) and The Robert A. Welch Foundation (AT-1448; G.R.D.).