The structure of plastocyanin tunes the midpoint potential by restricting axial ligation of the reduced copper ion

Blue copper proteins are models for illustrating how proteins tune metal properties. Nevertheless, the mechanisms by which the protein controls the metal site remain to be fully elucidated. A hindrance is that the closed shell Cu(I) site is inaccessible to most spectroscopic analyses. Carbon deuterium (C-D) bonds used as vibrational probes afford nonperturbative, selective characterization of the key cysteine and methionine copper ligands in both redox states. The structural integrity of Nostoc plastocyanin was perturbed by disrupting potential hydrogen bonds between loops of the cupredoxin fold via mutagenesis (S9A, N33A, N34A), variably raising the midpoint potential. The C-D vibrations show little change to suggest substantial alteration to the Cu(II) coordination in the oxidized state or in the Cu(I) interaction with the cysteine ligand. They rather indicate, along with visible and NMR spectroscopy, that the methionine ligand distinctly interacts more strongly with the Cu(I) ion, in line with the increases in midpoint potential. Here we show that the protein structure determines the redox properties by restricting the interaction between the methionine ligand and Cu(I) in the reduced state.


I. Supplementary Methods
Pc homology S2 Characterization of protein preparations S2 Assessment of Cu(II) content S2 Redox titrations S3 NMR spectroscopy S4 FT IR spectroscopy S4 Freeze-thaw treatment S6 Addition of denaturant S6 Time-dependent appearance of reduced state S7 Fig. S1. Structural homology of Pc S9 Fig. S2. Sequence homology of Pc S10 Fig. S3. SDS-PAGE gel of Pc mutants S11 Fig. S4. Mass spectra S12 Fig. S5. Assessment of Cu(II) content S13 Fig. S6. Nernst plots from chemical redox titrations S14 Fig. S7. HSQC spectra of 15 N Pc S15 Fig. S8. FT IR spectrum of d4-cystine S16 Table S1. Parameters from Gaussian fits to spectra of d2Cys89 S16 Table S2. Parameters from Gaussian fits to spectra of d3Met97 S17 Fig. S9. Correlation of Em to linewidth of d3Met97 S17 Fig. S10. Alternate Gaussian fits to spectra of d3Met97 for reduced Pc S18 Table S3. Parameters from alternate Gaussian fits to spectra of d3Met97 for reduced Pc. S19 Fig. S11. d2Cys89 and d3Met97 FT IR spectra following freeze-thaw S20 Fig. S12. d2Cys89 FT IR spectra of wt Pc in denaturing buffer S20 Fig. S13. d2Cys89 and d3Met97 FT IR spectra of Pc over time S21 Table S4. Parameters from Gaussian fits to spectra   with a liquid nitrogen-cooled mercury-cadmium-telluride detector. Interferograms collected for buffer and each protein sample were zero-filled by a factor of 8, apodized with a Blackman-Harris 4-term function, and Fourier transform was performed using the Mertz algorithm for phase correction, as previously described 7 . Absorption spectra were calculated from the resulting sample and reference transmission spectra. Slowly varying baseline was removed by fitting the spectral region excluding the band area to a polynomial function (MATLAB 2019b). High frequency noise was removed by applying a Fourier filter.

S6
Each baseline-corrected spectrum was fit to a Gaussian function (Table S1, Table S2). For d3Met97 of Cu(I) Pc the linewidth of the fit to the C-D absorption increases corresponding with the protein Em (Fig. S9, Table S2). To assess whether two bands may underlie the broader linewidths, we fit each spectrum to a sum of two Gaussian functions (Fig. S10, Table S3). When none of the parameters (amplitude, linewidth, frequency) of the two Gaussian functions are restricted, the spectra best fit to a band with nearly the same frequency as the single Gaussian fit and a second band of 2-6% relative area. When the frequency and linewidth of one band is fixed to the absorption found for wt, the spectra best with inclusion of a second band of substantial area.
However, the frequency of this band increases among mutants similarly as the single Gaussian fit, showing the same correlation with Em as observed with the fit to a single Gaussian band.
Freeze-thaw treatment. The IR spectra of oxidized Pc that had been previously frozen in liquid nitrogen and subsequently thawed indicates appearance of a reduced population (Fig. S11).
Curiously, the proteins are frozen in a purified, oxidized state in aerobic, aqueous buffer solution; the reducing agent is unclear. Mass spectrometry shows no change to indicate chemical modification (Fig. S4b).
In order to assess whether the absorbance observed results from reduced Pc rather than a denatured or apoprotein state, we collected a spectrum of 50 mM d4-cystine in 500 mM HCl (Fig.   S8). The high concentration is required due to the breadth of the absorption, which is higher than that of Cu-ligated d2Cys89 and makes the absorbance less easily identified. 500 mM HCl is required for solubility. The spectrum shows a single absorbance at ~2244 cm -1 resulting from the C-D2 asymmetric stretch.

S7
Addition of denaturant. Appearance of the reduced state can be induced by addition of subdenaturing concentrations of urea (Fig. S12). Samples in 2 mM potassium phosphate, pH 7 were brought to 3-5 M urea by the addition of a freshly prepared 8 M stock. Upon incubation of the protein for several hours in 3 M urea, a small (5-15%) absorption band for the C-D asymmetric stretch of d2Cys89 characteristic of the reduced state (2210 cm -1 ) appears as a minor species.
Addition of 5M urea leads to 63% conversion to a reduced species after four hours.
Time-dependent appearance of reduced state. In addition to Pc that had been freshly prepared and that which had undergone freeze/thaw treatment, we analyzed freshly prepared protein incubated in sealed tubes at 4°C over ~40 days (Fig. S13). A band at 2210 cm -1 characteristic of reduced Pc appears over time; however, the behavior of the mutant proteins in this analysis was indistinguishable from wt. The integrated areas of the absorption bands indicate ~20% reduced protein after ~10 days, an additional slight increase to ~30% population after 20 days in all except S9A Pc, then no subsequent change until the final analysis (Table S4). Oxidation upon exposure to air during sample loading likely contributes to the oxidized population detected in IR analysis.
Concurrently, whiteish precipitates also appear in the samples. The Ems of all proteins determined at time of FT IR spectroscopic analysis are unchanged from freshly analyzed protein, although we note this analysis is based on the 597 nm band so could be exclusively reporting on a subpopulation of natively folded, oxidized Pc. No consistent changes can be discerned in the 1600-1700 cm -1 region associated with amide I vibrations (Fig. S13).         following addition of 1.5 mM chemical oxidant potassium ferricyanide (orange) and redox S23 mediators as described for chemical redox titrations in 20 mM HEPES, 50 mM sodium chloride, pH 6.8. Protein is at the same concentration in both spectra. Higher intensity of the characteristic 600 nm absorption following addition of the oxidant indicates the transition to the reduced state is reversible. Asymmetry at higher wavelengths (~650 nm) is due to a known artifact associated with poor background matching in our cuvettes.