Protein Expression and Purification

Horse heart cyt c (HRC, Sigma C2506) was dissolved in a 10 mM sodium phosphate buffer at pH 7.0, oxidized by the addition of solid potassium ferricyanide, and purified on a HiTrap^TM SP HP cation exchange column (GE Healthcare) connected to an ÄKTA purifier fast protein liquid chromatography (FPLC) system as described.¹⁸ The Soret band extinction coefficient (ε₄₁₀ = 106,100 M⁻¹ cm⁻¹) was used to determine the protein concentration of HRC solutions.⁴⁰

The cyc-2.1 (NCBI accession {"type":"entrez-protein","attrs":{"text":"NP_500629","term_id":"17539604"}}NP_500629) or cyc-2.2 (NCBI accession {"type":"entrez-protein","attrs":{"text":"NP_506156","term_id":"17566554"}}NP_506156) genes were cloned into the pET-20b(+) plasmid (Amp^R, Novagen) following the pelB leader sequence for periplasmic protein localization. With these plasmids, coexpression of E. coli cyt c maturation proteins CcmA-H from pEC86 (Chl^R) was intended to assist with attachment of the c-type heme group.⁴¹ While the dual-plasmid expression system yielded large quantities of CYC-2.1, it was not successful for CYC-2.2. Therefore, the cyc-2.2 gene was instead inserted into the pBTR plasmid, which has been a successful system for HRC expression.⁴² The c-type heme insertion in pBTR is accomplished with the help of yeast heme lyase, encoded in the same plasmid.^{42, 43} Site-directed mutations (insertion of stop codons, adjustments of cloning sites, and introduction of Cys mutations at the proteins’ C-termini) were performed using protocols outlined in the QuikChange kit manual with XL1-Blue competent cells (Agilent Technologies). All plasmid sequences were verified at the Molecular Biology and Proteomics Core Facility (Dartmouth College).

Protein expression and purification were initially performed as described,⁴⁴ and subsequently optimized as follows. The pET-20b(+) plasmid containing the cyc-2.1 gene was co-transformed with pEC86⁴¹ into BL21 Star^TM (DE3) E. coli cells (Invitrogen), and colonies were grown on LB agar culture plates with 150–250 µg/mL of carbenicillin and 34–68 µg/mL of chloramphenicol at 37 °C. Colonies were screened for CYC-2.1 expression by inoculating each of twenty 7-mL cultures of Terrific Broth (TB) media (BD Company), containing 150 µg/mL of carbenicillin and 68 µg/mL of chloramphenicol, with a single isolated colony and growing at 37 °C and 220 rpm. Culture pellets were visually inspected after six hours for the appearance of a red color, indicating recombinantly-expressed heme protein. The culture with the highest expression was used to create a frozen stock for future use. Frozen cells were streaked onto an LB agar culture plate with 250 µg/mL of carbenicillin and 68 µg/mL of chloramphenicol and incubated overnight at 37 °C. Two single isolated colonies were picked into each of two 7-mL starter cultures of TB media containing 150 µg/mL of carbenicillin and 68 µg/mL of chloramphenicol, and grown at 37 °C and 220 rpm for six hours. The culture pellet exhibiting the darkest red color was used to inoculate six liters of TB media, containing 150 µg/mL of carbenicillin and 68 µg/mL of chloramphenicol, for large-scale growth. The cultures were shaken at 37 °C and 220 rpm until a red pellet was observed in centrifuged 1-mL culture test samples. The cells were harvested by centrifugation (25 minutes at 4 °C and 4,500 × g). Cell pellets were resuspended in a 50 mM Tris HCl buffer at pH 7.5, and then frozen at −80 °C for future lysis.

The pBTR plasmid containing the cyc-2.2 gene (Amp^R) was transformed into BL21 Star^TM (DE3) E. coli cells (Invitrogen), and colonies were grown on LB agar culture plates with 100 µg/mL of ampicillin at 37 °C. Frozen stocks and starter cultures for CYC-2.2 were unsuccessful. Instead, six isolated colonies were picked for direct inoculation of six 1-liter cultures of TB media, containing 100 µg/mL of ampicillin, for large-scale growth. Flasks were shaken at 37 °C and 220 rpm for approximately 24 hours until a red pellet was observed in centrifuged 1-mL culture test samples, and then the cells were harvested by centrifugation, resuspended in a 50 mM Tris HCl buffer at pH 7.5, and frozen.

Frozen cell pellets were thawed on a rocker for 45 minutes with 1 mM PMSF (MP Biomedicals), 0.033 mg/mL DNAse (Sigma-Aldrich), and approximately 0.1 mg/mL lysozyme (MP Biomedicals). The cells were lysed with a FRENCH® Press at 1500 psi and 4 °C to release the protein, and then the lysate was centrifuged (40 minutes at 4 °C and 43,500 × g). The supernatant was treated with solid ammonium sulfate (351 g salt per L of supernatant, added slowly while stirring the solution at 4 °C) to precipitate non-target proteins,⁴⁵ and then the solution was centrifuged (30 minutes at 4 °C and 43,500 × g). The supernatant was dialyzed into distilled water, followed by dialysis into a 10 mM sodium 2463</rec-number><foreign-keys><key app="EN" db-id="d0t9fpv9owefrless than 40 mL, oxidized by the addition of solid potassium ferricyanide, and purified on a HiLoad 26/10 SP Sepharose HP column (GE Healthcare) with FPLC.

An absorption band at ~650 nm (Figure S1 in the Supporting Information) was present in some preparations of the recombinant CYC-2.1 and CYC-2.2 proteins. The contamination giving rise to this band was successfully removed by acidic acetone treatment.⁴⁶ After addition of acidic acetone, the protein solution was gently swirled for five minutes. A 100 mM sodium borate buffer at pH 10.5 was used to resuspend the precipitated protein. The resuspended protein was dialyzed overnight against a 10 mM sodium phosphate buffer at pH 7.0 and repurified on a HiTrap^TM SP HP cation exchange column.

The C-terminal cysteine mutants of CYC-2.1 (Leu105Cys) and CYC-2.2 (Ala107Cys)^I were labeled with bimane iodoacetamide (Sigma-Aldrich) as previously described.³⁵ To prevent overlabeling, the reaction was quenched after five hours by adding an excess of the reducing agent dithiothreitol. The labeled variants were purified on a 5 mL HiTrap SP HP column using the same buffers as for wild-type proteins.

Matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry measurements (Dartmouth College Molecular Biology Core Facility, Applied Biosystems Voyager-DE^TM PRO BioSpectrometry^TM Workstation) routinely confirmed expected masses of purified wild-type proteins and labeled variants. Protein samples were diluted to about 1 pmol/µL in 1% trifluoroacetic acid, and analyzed after co-crystallization with a sinapinic-acid matrix. Several samples were also examined by electrospray and Fourier transform ion cyclotron resonance (FT-ICR) mass spectrometry at the WM Keck Foundation Biotechnology Resource Laboratory (Yale University).

Spectroscopic Measurements

All spectroscopic experiments were carried out at 21 ± 2 °C except the low-temperature electron paramagnetic resonance (EPR) measurements. Absorption spectra associated with the 695-nm charge-transfer band were measured with a JASCO V-630 scanning spectrophotometer. All other absorption and circular dichroism (CD) spectra were recorded with an Agilent 8453 diode-array spectrophotometer and a JASCO J-715 spectropolarimeter, respectively. Fluorescence spectra were recorded with a HORIBA Jobin Yvon Fluorolog®-3 spectrofluorometer; the intrinsic Trp and extrinsic bimane fluorophores were excited at 295 and 386 nm, respectively. All absorption and CD spectra were referenced against corresponding blanks, and Raman scattering in all fluorescence spectra was removed by subtracting spectra of the appropriate blanks.

Protein concentrations were determined spectroscopically using extinction coefficients recovered with pyridine hemochrome assays.⁴⁷ The assays were done in triplicate, and the concentration C (in mM) of heme in the pyridine solution was determined according to eq. 1, which included a correction for baseline drift:⁴⁸

where A is absorbance; II and III refer to the ferrous and ferric pyridine spectra at the specified wavelength, respectively; and ε is the pyridine hemochrome c double-difference extinction coefficient of 23.97 mM⁻¹ cm⁻¹ used to calculate the protein concentration.⁴⁹ Calculated protein concentrations were used to assign extinction coefficients to the spectra taken in buffer only.

Resonance Raman spectra were acquired on a WITec confocal Raman microscope 200 with linear polarization using a 514 nm laser (spectral center = 1430, grating = 1800 grooves/mm). A 5-µL sample (concentrations between 0.5 and 1 mM, in 100 mM sodium phosphate buffer at pH 7.4) was suspended upside-down from a coverslip over a 1.75-mm deep depression slide to create a “hanging drop” for measurements. Spectra were referenced to the spectrum of a saturated sodium salicylate⁵⁰ standard by assigning sequential pixel numbers to the output wavenumbers of the salicylate, and then generating a third-order polynomial fit to the literature-reported wavenumbers at the peak intensities as a function of the pixel numbers assigned to the experimental intensity peaks. The fit equation was applied to all assigned pixel numbers of the spectrum, thereby generating referenced wavenumbers for all intensities that were used for comparison of the data.

EPR spectra were collected on a Bruker EMX 300 X-band EPR spectrometer equipped with an Air Products low temperature liquid helium system. To obtain spectra of the heme iron, samples of ferric cyt c in a 100 mM sodium phosphate buffer at pH 7.4 were cooled to 10 ± 2 K prior to measurements. Instrument parameters were: microwave frequency of 9.477 ± 0.0014 GHz, microwave power of 3.204 mW, and eight scans per sample.

TR-FRET Measurements and Analyses

Fluorescence lifetimes were measured by time-correlated single photon counting (TCSPC) at 5,000–10,000 counts using a NanoLED-375L diode laser (λ_ex= 375 nm, <70 ps pulsewidth) as the excitation source and a fast TBX-04 detector. All measurements were done under magic angle conditions. Bimane emission was monitored at 480 nm.

TR-FRET data were analyzed in MATLAB (MathWorks) by performing numerical inversion of the Laplace transform based on a set of logarithmically spaced fluorescence-decay rate constants k, as previously described.¹⁸ The Förster relation (eq. 2) allowed for the transformation of the distributions of rate constants k, P(k) , into distributions of donor (D)-acceptor (A) distances r, P(r):⁵¹

In eq. 2, k₀ = 10.8×10⁷ s⁻¹ is the fluorescence decay rate of the bimane model compound, bimane-labeled N-acetyl Cys, and R₀ = 35 Å is the Förster distance for the bimane-heme pair.³⁵

Redox Titrations

Reduction potentials were determined following a standard spectrophotometric procedure in which cyt c was oxidized or reduced with ferri- or ferrocyanide and the absorbance at 550 nm was measured as a function of different ratios of ferri- to ferrocyanide.⁵² Slopes of the linear fit of these dependencies (Figure S2 in the Supporting Information) were used to calculate the cyt c reduction potentials.

Liposome Preparation Centrifugation Binding Assays

The lipids 1,1’,2,2’–tetraoleoyl cardiolipin (TOCL) and 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) (Avanti Polar Lipids, Inc.) were used to prepare a 2.2 mM liposome solution by extrusion.³⁵ Briefly, chloroform-suspended DOPC/TOCL in a 1:1 molar ratio was desiccated under a stream of nitrogen gas and resuspended in a 25 mM HEPES buffer at pH 7.4 to a final lipid concentration of 2.2 mM. The solution was first incubated in a shaker for 30 minutes at 37 °C and 220 rpm, then placed in a sonication water bath for no more than 60 minutes, and finally passed 11 times through a 0.1 µm (SPI) membrane filter at 50 °C using a liposome extruder (Eastern Scientific) to yield 40-nm radii unilamellar vesicles. Vesicle sizes were analyzed at room temperature by dynamic light scattering measurements with a DynaPro NanoStar^TM (Wyatt Technology Europe GmbH).

Centrifugation Binding Assays

Protein-liposome ultracentrifugation binding assays were performed as previously described.^{18, 35}

Peroxidase Assays

Intrinsic peroxidase activities (without liposomes) were measured as oxidation of guaiacol (Sigma-Aldrich) to tetraguaiacol, which absorbs strongly at 470 nm. A volume of 900 µL of cyt c protein in a 25 mM HEPES buffer at pH 7.4 was added to a 1-mL cuvette containing solutions of 80 µL of freshly-prepared guaiacol and 20 µL of hydrogen peroxide to final concentrations of 0.94 ± 0.05 µM protein and 10 mM guaiacol. Hydrogen peroxide (H₂O₂) concentrations were varied from 0 to 75 mM and verified spectroscopically (ε₂₄₀ = 43.6 M⁻¹ cm⁻¹).⁵³ The solution was quickly mixed by pipeting and the measurement was immediately begun. Tetraguaiacol (ε₄₇₀ = 26,600 M⁻¹ cm⁻¹) is a product of four oxidation reactions;⁵⁴ and its formation was monitored every second for 10 minutes by absorbance measurements.

Another peroxidase activity assay, at much lower concentrations of H₂O₂ (100 µM), was used as a control to verify that the differences in intrinsic activities of the three proteins were not related to possible artifacts from high concentrations of H₂O₂ in the guaiacol assays. The formation of etoposide phenoxyl radicals⁵⁵ catalyzed by cyt c was monitored by EPR spectroscopy. Protein solutions (3 µM) in a 25 mM HEPES buffer at pH 7.4 were incubated with etoposide (100 µM) for 1 hour before the reaction was initiated by the addition of H₂O₂ (100 µM). No reaction was noted in the absence of cyt c. Samples were transferred quickly to capillary tubes (1.5 × 90 mm) and the first spectrum was taken immediately after. Instrument parameters were: 3520 G center field; 100 kHz modulation frequency, and 10.11 mW microwave power. After baseline correction, the first derivative output of the spectrometer was integrated to give the absorbance signal and then integrated again to give the area representative of the amount of etoposide phenoxyl radical.

Peroxidase activity assays in the presence of liposomes were performed with guaiacol, as above, except that H₂O₂ concentrations were 5 mM and lipid concentrations were varied from 0 to 200 µM. Cyt c solutions were mixed with freshly-prepared liposome solutions in a 1:9 (v/v) ratio to avoid potential protein aggregation upon contact with lipids, and the mixtures were incubated for at least 30 minutes at room temperature prior to use. Fluorescence of the native Trp residue was measured in these samples (λ_ex = 295 nm) to probe for unfolding behavior before the samples were added to the guaiacol/hydrogen peroxide solutions; protein concentrations were 1.07 ± 0.03 µM and 0.96 ± 0.03 µM during fluorescence measurements and peroxidase activity assays, respectively. Peroxidase activities were measured by the absorbance at 470 nm every second for 30 minutes. Linear portions of the peroxidase activity curves, indicative of the steady-state phase of tetraguaiacol formation, were fit to a first-order polynomial equation and slopes were used to calculate the rate of tetraguaiacol formation. The dependence of the reaction rate on H₂O₂ concentration was fit to the Michaelis-Menten equation (eq. 3).

Expression and Characterization of Recombinant Proteins

Protein expression and purification protocols developed in this study yielded 4–5 mg of pure C. elegans cyt c proteins per liter of E. coli culture. Mass-spectrometry analyses revealed the expected masses of the proteins (Table 1). A previous report of CYC-2.1 isolated from C. elegans found the N-terminal Ser residue to be acetylated,³⁴ similarly to HRC.⁴² Not surprising for E. coli expression systems,⁴² neither the recombinant CYC-2.1 nor CYC-2.2 possessed this post-translational modification. This lack of an acetyl group is not expected to have a significant impact on the proteins’ properties presented in this paper.⁴²

Despite numerous trials, expression of CYC-2.2 with a generally more promiscuous E. coli heme maturation cassette (genes ccmA-H within pEC86)⁶⁰ was not successful. However, insertion of the cyc-2.2 gene into the expression vector pBTR yielded the desired protein product. The inability to utilize starter cultures or maintain frozen stocks of recombinant E. coli that consistently produced CYC-2.2 suggested that this protein might be toxic to the bacteria.

Ammonium sulfate precipitation and ion exchange chromatography have been sufficient methods to purify functional Saccharomyces cerevisiae and HRC cyt c proteins from E. coli recombinant expression.^{18, 44} In some, but not all, preparations of C. elegans cyt c proteins, an unexpected absorption band was detected at ~650 nm after these purification steps (Figure S1 in the Supporting Information). Electrospray and FT-ICR measurements on these crude samples did not reveal the presence of additional components besides the target protein. A similar absorption band has been seen by others with a different cyt c,⁶¹ and its spectral characteristics are consistent with that of verdoheme cyt c, a species that differs only by 3 Da in mass from the native protein.^{62, 63} The 650-nm band did not disappear after the protein was further purified by reverse-phase chromatography but was successfully removed (Figure S1 in the Supporting Information) by acidic acetone treatment, in which cyt c protein was precipitated in an acidic acetone solution. Analyses of supernatants from acidic acetone treatments revealed a number of molecules consistent with products of protein degradation. The precipitated protein was resuspended in a buffer at basic pH and purified again on an ion-exchange column. Acidic acetone treatment resulted in a loss of 59–76% of the total protein. Conversion of verdoheme to an open biliverdin structure in acidic acetone solution could be a potential mechanism of verdocyt c removal.⁶² Rapid degradation of the isolated material hampered our further investigations of the contaminant. Because the 650-nm absorption feature was only present in some preparations, it does not appear to be an intrinsic feature of C. elegans cyt c.

A variety of tools was used to investigate the native structural characteristics of the recombinantly-expressed CYC-2.1 and CYC-2.2 to determine if they are generally consistent with those of the cyt c family of proteins, particularly the well-characterized mammalian analogue HRC. Overlap of the HRC crystal structure with models of CYC-2.1 and CYC-2.2 from sequence threading revealed similar placement of the conserved α-helical segments (Figure 1B). In accord with these predictions, far-UV CD spectra of CYC-2.1 and CYC-2.2 (Figure S3 in the Supporting Information) showed signals consistent with the presence of α-helices. Relative to HRC and CYC-2.1, CYC-2.2 exhibited a greater helical content. A secondary structure prediction algorithm⁶⁴ yielded the following percentages of protein helical structure based on the full-length primary amino acid sequences: 30, 26, and 33% for HRC, CYC-2.1, and CYC-2.2, respectively. The predicted trend reproduced the increased propensity of CYC-2.2 for forming α-helices. However, the predicted value for HRC differed significantly from data retrieved from the protein crystal structure (48% helical),⁶⁵ reflecting the complexity of real proteins, particularly ones with large cofactor groups such as heme, for structure prediction algorithms. The increased helical signals in C. elegans proteins likely arise from the extended protein termini.

The UV-visible absorption spectra of the ferric and ferrous C. elegans proteins CYC-2.1 and CYC-2.2 (Figure 2 and Table 1) are consistent with those reported for CYC-2.1 protein previously isolated from C. elegans.³⁴ These spectral features closely resemble those of other c-type cyt proteins with His/Met heme ligation.²

Open in a separate window

Figure 2

UV-visible absorption spectra of ferric (black) and ferrous (red) (A) CYC-2.1 and (B) CYC-2.2 at pH 7.4. Spectra of ferrous proteins include contributions to absorption signals below 350 nm from dithionite. (C) Near-IR (600–725 nm) absorption spectra of HRC, CYC-2.1, and CYC-2.2.

Heme Ligands

Inspection of the near-IR absorption region (Figure 2C) of the ferric proteins revealed a 695-nm charge-transfer band arising from Met-heme axial coordination. The intensity of this band has been suggested to correlate with the strength of Met-iron bond.⁶⁶ The ε₆₉₅ values for HRC and CYC-2.1 are nearly identical (Figure 1C), while the value for CYC-2.2 protein is about 10% lower. EPR spectra of ferric CYC-2.1 and CYC-2.2 (Figure 3A) revealed features of a low-spin state of the heme iron, consistent with His/Met coordination.²

Open in a separate window

Figure 3

(A) Electron paramagnetic resonance spectra of ferric HRC (black), CYC-2.1 (cyan) and CYC-2.2 (red) proteins ([cyt c]=450–550 µM) in a 100 mM sodium phosphate buffer at pH 7.4 and 10 ± 2 K. (B) Resonance Raman spectra (λ_ex=514 nm) and vibrational assignments of ferric HRC (black), CYC-2.1 (cyan), and CYC-2.2 (red) proteins ([cyt c]=0.5–1.0 mM) in a 100 mM sodium phosphate buffer at pH 7.4 and room temperature.

To confirm similarities of the heme environments in cyt c proteins from C. elegans and mammals, the ferric proteins were further analyzed by resonance Raman spectroscopy. The high-frequency region of the resonance Raman spectrum provides information about heme coordination and spin state and is a sensitive reporter of heme environment. The spectra of the three proteins (Figure 3B) are nearly identical. With 514-nm excitation, ν10 and ν30 are particularly strong markers of the heme spin state. These peaks are observed at 1638 and 1172 cm⁻¹ for CYC-2.1 and 1637 and 1172 cm⁻¹ for CYC-2.2, positions consistent with a low-spin heme, with His/Met ligation.^{67, 68}

Spectrophotometric titrations (Figure S2 in the Supporting Information) yielded reduction potentials of 251 ± 4 mV and 254 ± 2 mV for CYC-2.1 and CYC-2.2, respectively. These values are near that of HRC (262 ± 2 mV)⁵² and are comparable to other cyt c proteins.^69–71 These findings are, once again, support His/Met ligation to the heme iron.

Intrinsic Peroxidase Activities

To further probe the heme environment in the C. elegans cyt c proteins, peroxidase assays were conducted using hydrogen peroxide (H₂O₂) and guaiacol. The guaiacol assay is a well-established procedure for investigations of peroxidase activity of cyt c in its native and denatured states.⁷² All four previously-reported phases of guaiacol oxidation were observed for HRC and C. elegans cyt c proteins over the duration of the experiment: an activation or lag phase (I), a steady-state linear phase (II), a leveling-off of tetraguaiacol formation (III), and finally a degradation phase (IV).⁶³ Phases III and IV have been attributed to suicide inactivation of the heme catalyst and degradation of the inherently unstable tetraguaiacol product by H₂O₂.⁶³ HRC showed a minor peroxidase activity in the native state, and the two C. elegans cyts c had greater activities than HRC at a given H₂O₂ concentration (e.g., 50 mM; Figure 4A). Analyses of the H₂O₂ concentration dependence on the rates of guaiacol formation (Figure 4) yielded the kinetic parameters in Table 2. The recovered values for HRC are consistent with previous reports.⁷³

Open in a separate window

Figure 4

Peroxidase activities of HRC (black), CYC-2.1 (cyan), and CYC-2.2 (red) at pH 7.4 in a 25 mM HEPES buffer. Protein and guaiacol concentrations were 0.96 ± 0.03 µM and 10 mM, respectively. (A) Representative curves of tetraguaiacol formation (A₄₇₀ versus time) at 50 mM H₂O₂. (B) Rates of tetraguaiacol product formation over a range of H₂O₂ concentrations. Lines are fits to eq. 3 with parameters in Table 2.

The guaiacol assay has been applied to a number different cyt c proteins and mutants of varying thermodynamic stabilities^{72, 74} and thus serves as an excellent method for making direct comparisons in this work. However, this assay of a weak peroxidase cyt c (native protein) required high concentrations of H₂O₂, introducing the possibility of protein degradation under these conditions. A control experiment, performed by monitoring the formation of etoposide phenoxyl radicals at much lower (100 µM) concentrations of H₂O₂ catalyzed by the three cyt c proteins (Figure S4 in the Supporting Information), recovered the same trends as seen with guaiacol in Figure 4, arguing against artifacts from high H₂O₂ concentrations in our results.

Protein Interactions with CL-Containing Membranes

Interactions of CYC-2.1 and CYC-2.2 proteins with CL were tested with ultracentrifugation pelleting assays and the fluorescence response from the single Trp residue in each of the two C. elegans proteins. Ultracentrifugation of protein-liposome solutions with increasing lipid concentrations revealed similar binding behaviors for the C. elegans cyt c proteins and HRC (Figure 5),¹⁸ indicating no significant differences among these proteins in their binding interactions with CL-containing liposomes.

Open in a separate window

Figure 5

Percentage of cyt c bound to TOCL/DOPC small unilamellar vesicles (50 mol% CL, radii = 40 nm) after ultracentrifugation of protein-liposome solutions. Protein concentrations were 5 µM. The smooth curve is to guide the eye only.

Fluorescence from the single Trp residue (Trp59 in HRC, Trp58 in CYC-2.1 and CYC-2.2) was used to probe for changes in the proteins’ tertiary structures in the presence of CL (Figure 6). Trp fluorescence is quenched in the native protein by FRET to the heme.^{59, 75} Progressively larger amounts of CL-containing liposomes in cyt c solutions increased Trp fluorescence signals for all three proteins (Figure 7A), suggesting that the cyt c-CL interactions disrupt the proteins’ tertiary structure, resulting in an increased intramolecular Trp-to-heme distance. However, even at high CL concentrations (total lipid-to-protein molar ratios of 200 to 1), the Trp signals were still smaller and their emission maxima were ~ 20 nm blue-shifted, compared to signals of the GuHCl-denatured proteins. While the Trp signals in the native HRC and two C. elegans proteins were comparable, the Trp signal of CYC-2.1 was greater than that of HRC, and CYC-2.2 was greater than that of CYC-2.1 at each concentration of CL (Figure 7).

Open in a separate window

Figure 6

Fluorescence spectra (λ_ex = 295 nm) of 1.07 ± 0.03 µM HRC, CYC-2.1, and CYC-2.2 at pH 7.4 in a 25 mM HEPES buffer (native, black), with 6 M guanidine hydrochloride (GuHCl, blue), and with TOCL/DOPC (200 µM total lipid, 50 mol% CL) liposomes (CL, red).

Open in a separate window

Figure 7

(A) Trp fluorescence intensities at 337 nm (λ_ex = 295 nm) and (B) rate constants of the linear phases of tetraguaiacol formation for HRC, CYC-2.1, and CYC-2.2 at pH 7.4 in a 25 mM HEPES buffer with various total lipid concentrations (TOCL/DOPC liposomes, 50 mol% CL). Protein concentrations were 1.07 ± 0.03 µM and 0.96 ± 0.03 µM during fluorescence measurements and peroxidase activity assays, respectively. In peroxidase activity assays, H₂O₂ and guaiacol concentrations were 5 mM and 10 mM, respectively.

With the addition of CL-containing liposomes, the peroxidase activities of all three proteins increased (Figure 7B). The activity of CYC-2.2 was notably higher than the activity of either HRC or CYC-2.1. For all three proteins, no increase in peroxidase activity was detected upon addition of CL-free liposomes (Figure S5 in the Supporting Information).

Analyses of Cyt c Conformers by TR-FRET

C-terminal Cys mutants of CYC-2.1 and CYC-2.2 were prepared and labeled with the small fluorophore bimane (Figure 8A). Steady-state fluorescence spectra of bimane-labeled proteins in solutions of GuHCl showed a dramatic increase in bimane fluorescence signals. Interactions of labeled CYC-2.1 and CYC-2.2 with CL-containing liposomes also increased bimane fluorescence, but to a lesser degree (Figure 8B). These results are consistent with quenching of bimane fluorescence by the heme under native conditions, when the dye and the heme are close, and a relief of the quenching when the two groups move apart upon protein unfolding.³⁵ No observable shifts in the bimane emission maxima from those of the native proteins were detected upon addition of CL-containing liposomes. Similar to previous results with HRC,³⁵ no changes in the bimane fluorescence spectra of CYC-2.1 and CYC-2.2 from those of the native proteins were observed in the presence of CL-free liposomes (Figure S5 in the Supporting Information).

Open in a separate window

Figure 8

(A) Positions of labeling sites in CYC-2.1 and CYC-2.2 and structure of the bimane label. (B) Fluorescence spectra (λ_ex=386 nm); (C) decays (λ_ex=375 nm and λ_em=480 nm); and (D) distributions of rate constants P (k) (left) and D-A distances P (r) (right) for bimane-labeled (at Cys placed at the last residue position in each protein) CYC-2.1 and CYC-2.2 in a 25 mM HEPES buffer at pH 7.4 (native (N), gray); with TOCL/DOPC liposomes (50 mol% CL, 400 µM total lipid; CL-bound (CL), red); and in 6 M GuHCl solution at pH 7.4 (GuHCl-unfolded (U_GuHCl), blue). At distances longer than 43 Å, energy transfer rate constants and D-A distances cannot be determined reliably; the structures with r ≥ 43 Å are represented by a single bar.

TR-FRET kinetics in bimane-labeled proteins offered additional insights about the conformations of CYC-2.1 and CYC-2.2 in their native, CL-bound, and GuHCl-denatured states. These measurements revealed distinct bimane fluorescence decays for each of these three experimental conditions (Figure 8C). Analyses of these decays yielded distributions of rate constants P (k) and bimane-heme distances P (r) (Figure 8D). The modes of the bimane-heme distance distributions P (r) for the native CYC-2.1 and CYC-2.2 were 24 ± 0.3 Å. These distances are 1–2 Å longer than bimane- and dansyl- heme distances in C-terminal labeled variants of Saccharomyces cerevisiae iso-1 cyt c and HRC,^{76, 77} findings that are consistent with the slightly longer C-terminal helices in both C. elegans proteins (Figure 1A).

The decays of the GuHCl-denatured samples showed no evidence of bimane quenching, suggesting a large (≥ 43 Å) separation of the polypeptide N- (carrying the heme group) and C-termini (carrying the bimane group) upon unfolding of both proteins. Addition of CL-containing liposomes resulted in protein ensembles with two distinct populations of cyt c conformers with short (22–28 Å) and long (> 30 Å) distances between bimane and heme chromophores. With the same concentration of CL-containing liposomes (400 µM total lipid, 50 mol% CL), the populations of extended conformers in CYC-2.1 and CYC-2.2 ensembles were comparable (29–43%).

Thermodynamic Stability and Intrinsic Peroxidase Activity

The equilibrium unfolded studies uncovered a lower thermodynamic stability of CYC-2.2, compared to HRC.¹⁸ The change in the folding Gibbs free energies ΔGf0 is the same, within error, for HRC and CYC-2.1, but the midpoint of the unfolding transition for CYC-2.1 is at a distinctly lower GuHCl concentration (Table 3). The errors in the calculated values of ΔGf0 arise from uncertainties in both [GuHCl]_½ and m_D. Considering the much smaller error in the measured [GuHCl]_½, than in m_D, the CYC-2.1 protein is likely less stable than HRC.

Two non-native His ligands, His26 and His33, misligate to the heme in HRC and contribute to stabilization of its denatured state at neutral pH.^{8, 59} The two C. elegans cyt c proteins both lack His26, and CYC-2.2 also has Asn in place of His32 (His33 in HRC)^I. The N-terminal amino group could also misligate to the heme in unfolded proteins;⁸⁰ however, the blue-shifted heme absorption spectrum of CYC-2.2 in the GuHCl-denatured state (Figure S6 in the Supporting Information) suggests that this group is not a major ligand and likely does not contribute to substantial stabilization of the denatured state. With a gap in the protein sequence at position 26, misligation of the heme by His32 in CYC-2.1 yields a tighter, more destabilized polypeptide loop compared to that in the denatured HRC. This analysis argues against increased stabilization of the denatured state for CYC-2.1 and CYC-2.2, compared to HRC, and instead suggests a destabilization of the native state for the C. elegans proteins.

Analyses of structural models for the two C. elegans proteins identified a number of notable differences in regions that line the heme crevice, compared to those in HRC. In addition to the sequence gap at position 26, the two bulky amino acid residues Leu35 and Ile85 (HRC numbering) are replaced by smaller Val (ΔV= −27 ų) in C. elegans proteins. The shorter 20’s–30’s Ω-loop and less well-packed protein core could lead to destabilization of the native CYC-2.1 and CYC-2.2, compared to HRC. Interestingly, mutations of Ile85 in S. cerevisiae iso-1 cyt c have been shown to dramatically affect stability of this protein.⁸¹ In CYC-2.2, possible displacement of helices associated with the longer N- and C- termini could contribute to additional destabilization of the protein tertiary structure, despite a higher α-helical content.

Peroxidase activity of cyt c is a sensitive indicator of local polypeptide unfolding, which is connected to the dissociation of a fairly weak Met ligand and opening of the heme crevice. Similar to hydrogen-deuterium isotope exchange experiments, the peroxidase assay reports on the conformational fluctuations in the protein native state and offers insights about low free energy unfolding intermediates.⁷² The strong correlation between thermodynamic stability and a small (but detectable) peroxidase activity for native HRC, CYC-2.1, and CYC-2.2 (Tables 2 and ​and3)3) in this study, as well as other cyt c proteins,^{72, 74} suggests that the ease of the local polypeptide unfolding near the cyt c heme is related to the global stability of the protein fold.

Cyt c Unfolding and Amplified Peroxidase Activity upon Protein Interactions with CL

Interactions of cyt c with the mitochondria-specific lipid CL are central to protein function in both oxidative phosphorylation and apoptosis. Recent reports suggest that CL levels strongly affect C. elegans, where CL depletion causes a reduction in the mitochondrial membrane potential and abnormal mitochondrial function in germ cells.²⁹ There is abundant literature proposing different binding models for cyt c–CL interactions, with a general consensus on the importance of electrostatic interactions between the positively-charged protein and the negatively-charged lipid, in particular involving the Lys residues 72, 73, 86, and 87 in cyt c.^{82, 83} These four Lys residues are completely conserved in both C. elegans cyt c proteins. The same binding behavior of HRC and the two C. elegans proteins to CL liposomes in ultracentrifugation experiments suggests the high likelihood of a similar mode of liposome binding for all three proteins. The similar Trp emission maxima for the CL-bound proteins further support this hypothesis. Interestingly, red-shifted emission maxima of the environmentally-sensitive bimane fluorophore at the C-termini of CYC-2.1 and CYC-2.2 suggest that the very end of the C-helix in these proteins is solvent-accessible and does not insert into the membrane. These findings support a predominantly peripheral binding of proteins on the liposome surface, a mechanism recently discussed for HRC.³⁵

Interactions of cyt c with CL-rich liposome surfaces promote destabilization of the proteins’ native structure and polypeptide unfolding.^{16, 18} Increased Trp signal intensities in the presence of CL (Figures 6 and ​and7A)7A) reflect diminished Trp-to-heme FRET, owing to longer distances between the two chromophores. The similar shapes of the three curves (Figure 7A) are consistent with similar binding affinities of the three proteins to CL-containing liposomes. On the other hand, the magnitudes of the Trp fluorescence signal over the entire range of studied CL concentrations clearly differ among the three proteins. Analyses of the Trp fluorescence at high lipid-to-protein ratios, when the majority of cyt c is bound to CL-containing liposomes, suggest longer on average Trp-to-heme distances in the ensemble of CYC-2.1, compared to HRC, and in the ensemble of CYC-2.2, compared to CYC-2.1. These trends parallel the trends in stabilities of the three proteins ([GuHCl]_½) but also the trend observed for the Trp signals in denatured proteins in 6 M GuHCl at pH 7.4 (Figure 7A). Thus, the differences in the magnitude of the Trp signal could arise from differences in either the population of the unfolded species in the three protein ensembles or the degree of protein unfolding.

Measurements of TR-FRET in dye-labeled HRC revealed distinct populations of compact and extended structures for the CL-bound protein.^{18, 35} Herein, we identify the same behavior with bimane-labeled C. elegans cyt c proteins. Notably, these results demonstrate that under identical conditions, the CYC-2.1 and CYC-2.2 proteins have populations of extended conformers comparable to each other and to HRC. Scaling the signals in Figure 7A based on the magnitude of the Trp signal in GuHCl-denatured proteins yields identical Trp curves for all three proteins (Figure S7 in the Supporting Information), further supporting the similarities in the populations of the unfolded species.

While the gross features of the three ensembles of CL-bound proteins (relative populations of compact and extended conformers) appear to be the same, there are subtle conformational differences among the conformers. The magnitude of the Trp fluorescence signal is sensitive to His misligation in denatured cyt c, and the signals of the denatured HRC, CYC-2.1, and CYC-2.2 at pH 7.4 (Figure 6) reflect the propensities for such an interaction in these proteins. With the sixth coordination site of the heme iron occupied by His33 in HRC and His32 in CYC-2.1 (a preferred non-native His ligand in HRC⁵⁹ and the only one in CYC-2.1), the peroxidase activities of these protein are less than that of CYC-2.2. Evidently, His33 (His32) ligation, together with the surrounding polypeptide, offers similar protection to the heme in CL-bound HRC and CYC-2.1. These results corroborate earlier findings of the importance of His misligation in inhibiting peroxidase function of denatured cyt c.⁷²

Recent work from our laboratory has revealed interconversions between compact and extended conformers in the ensemble of CL-bound HRC.³⁵ These dynamics, reflected in partitioning between the two conformer types, depend on protein interactions with the CL-containing liposomes. In this study, we find not only similar binding behaviors to CL-containing liposomes for all three cyt c proteins, but also similar distributions of compact and extended conformers from the proteins labeled with dyes at their C-termini. These findings are particularly intriguing because the proteins differ in their thermodynamic stabilities. It appears that the three cyt c proteins bind to the lipid surface in the same manner, with similar contacts likely provided by conserved residues. Even though cyt c does substantially unfold on the liposome surface, the denaturing mechanism seems to involve specific protein-lipid interactions.

The largely open, extended conformers are of particular functional importance since they are likely to exhibit the greatest peroxidase activity.¹⁸ Distributions of distances between the heme and the protein C-terminus from TR-FRET experiments probe general features of the polypeptide ensemble. While these global features are similar for CYC-2.1 and CYC-2.2, distinct peroxidase activities highlight differences in the heme local environment among the protein conformers. Both global and local characteristics of the protein conformers are thus important considerations in the analysis of functional implications of cyt c unfolding.

We thank Michael Q. Zhu and Philipp B. Rentzsch for their help with protein expression and characterization of the protein contaminant, Anna M. Morenz for her preliminary work on peroxidase assays, and Kara L. Bren for the pBTR plasmid.

This work was funded by Dartmouth College startup funds (E.V.P.), CIPSM (B.C.), funds from the Ludwig Maximilians University Munich (B.C.) and National Institutes of Health grants GM069950 (B.C.), {"type":"entrez-nucleotide","attrs":{"text":"GM076651","term_id":"221303763"}}GM076651 (B.C.), and {"type":"entrez-nucleotide","attrs":{"text":"GM098502","term_id":"221405262"}}GM098502 (E.V.P.). A Zabriskie Fellowship supported Senior Honors research (D.S.P.).