|Home | About | Journals | Submit | Contact Us | Français|
The incorporation of unnatural amino acids into proteins that act as spectroscopic probes can be used to study protein structure and function. One such probe is 4-cyanophenylalanine (PheCN), the nitrile group of which has a stretching mode that occurs in a region of the vibrational spectrum that does not contain any modes from the usual components of proteins and the wavenumber is sensitive to the polarity of its environment. In this work we evaluate the potential of UV resonance Raman spectroscopy for monitoring the sensitivity of the νCN band of PheCN incorporated into proteins to the protein environment. Measurement of the Raman excitation profile of PheCN showed that considerable resonance enhancement of the Raman signal was obtained using UV excitation and the best signal-to-noise ratios were obtained with excitation wavelengths of 229 and 244 nm. The detection limit for PheCN in proteins was ~10 μM, approximately a hundred-fold lower than the concentrations used in IR studies, which increases the potential applications of PheCN as a vibrational probe. The wavenumber of the PheCN νCN band was strongly dependent on the polarity of its environment, when the solvent was changed from H2O to THF it decreased by 8 cm−1. The presence of liposomes caused a similar though smaller decrease in νCN for a peptide, mastoparan X, modified to contain PheCN. The selectivity and sensitivity of resonance Raman spectroscopy of PheCN mean that it can be a useful probe of intra- and intermolecular interactions in proteins and opens the door to its application in the study of protein dynamics using time-resolved resonance Raman spectroscopy.
There is considerable interest in spectroscopic probes that can be used to study the structure, properties and dynamics of proteins. Techniques such as electronic, vibrational, fluorescence and NMR spectroscopy have been used extensively to study both naturally occurring and artificially introduced chromophores in proteins. The application of vibrational spectroscopic techniques to the study of proteins has benefited from the presence of multiple native chromophores in proteins. Resonance Raman spectroscopy is especially useful for selectively probing chromophores (e.g. aromatic side chains, amides, hemes) with high sensitivity. However, when information is needed on a single residue or region of the protein and there are multiple copies of the relevant chromophore in other parts of the protein the important information can be obscured by the overlapping signals. Thus we are interested in the selective incorporation of non-native chromophores which will give strong resonance Raman signals separate from the other chromophores in proteins and also be sensitive to the local environment in the protein.
A promising candidate is 4-cyanophenylalanine (PheCN). Boxer and coworkers have used vibrational Stark spectroscopy to show that the νCN frequency of nitriles, especially aromatic nitriles, is highly sensitive to the electric field in its environment1–4 and the νCN mode of organic nitriles, which falls in the 2150–2250 cm−1 region, does not overlap with any of the common vibrational bands of proteins. Additional IR spectroscopic studies of PheCN have shown that there is a strong dependence of the νCN wavenumber on solvent polarity.5 Reimers and Hall have used theoretical calculations to shed light on the effect of solvation on the νCN frequency of acetonitrile in a wide variety of solvents.6 They noted that specific solvation effects can have a strong impact on the νCN wavenumber and caution that this makes acetonitrile an unreliable probe of the local electric field. While the formation of specific intermolecular interactions by the nitrile group (e.g. hydrogen bonding) may cause difficulties in using the νCN wavenumber to map the local electric field of a system, these specific interactions can still be used to advantage to detect changes in the nitrile’s environment. Whether the change in the νCN wavenumber is caused by specific intermolecular interactions, a general alteration of the local electric field or a combination of the two can serve as a marker for changes in the interaction between molecules. Shifts in the νCN wavenumber have been used to detect the binding of PheCN containing peptides to the hydrophobic surface of another protein5 or their incorporation into a phospholipid bilayer.7 Since the PheCN side chain is similar in size to natural amino acid side chains and is amphiphilic it is not likely to cause significant changes to the structure of a protein when it is substituted for a native amino acid residue.5
To date the studies on PheCN in peptides have used samples prepared using standard solid-phase peptide synthesis methods.5, 7–9 The development of in vivo expression systems that can site-specifically incorporate an unnatural amino acid into a protein10–13 means that it is feasible to introduce PheCN into larger proteins which can not be synthesized using solid-phase methods. A limitation on IR spectroscopy for PheCN monitoring is that it requires solutions of ~1 mM concentration5, 7 and a method with increased sensitivity is desirable. Thus we were interested in determining whether there were conditions under which the resonance enhancement of the νCN Raman signal would give improved sensitivity compared to IR measurements while still providing information on the changes in the PheCN environment. Additionally, the monitoring of PheCN by resonance Raman spectroscopy would open up the possibility of studying protein dynamics by using time-resolved resonance Raman to investigate changes in proteins initiated by photolysis14–17 or T-jump pulses.18–21
Mastoparan X peptide with PheCN substituted for the 5th residue (MPX5) 7 and the calmodulin binding domain of myosin light chain kinase (MLCK579–595) with the third residue replaced by PheCN (MLCK3PheCN)5 were prepared as described previously. 4-Cyanophenylalanine (PheCN) from Bachem, horse heart cytochrome c (97%), Tris.HCl, Tris base (99.9%), and CaCl2.2H2O (99%) from Sigma, DMSO-d6 (99.9%) from Cambridge Isotope Laboratories, acetone (99.9%) and tetrahydrofuran (99.5%) from EM Sciences, NaCN (95%) and NaClO4 (98%) from Aldrich, and palmitoyloleoylphosphatidylcholine (POPC) (>99%) from Avanti Polar Lipids were used as received.
Resonance Raman spectra were obtained at 257 nm, 244 nm and 229 nm with excitation from an intracavity frequency-doubled Ar+ laser (Coherent Innova 300 FRED). Excitation pulses (20 ns, 1 kHz) at 197 nm 22 (0.5μJ/pulse) or 212 nm, 220 nm and 229 nm 23, 24 (0.5μJ/pulse) were obtained by frequency quadrupling the output of a Ti:sapphire laser, which was pumped (527 nm, ~10 mJ/pulse, 70 ns, 1 kHz) by an intracavity frequency-doubled Nd:YLF laser (GM30, Photonics International, Inc.).
The spectra of PheCN in H2O/THF were measured using solutions sealed in 5 mm quartz NMR tubes, which were spun to ensure mixing of the solution. The scattered light was collected at 135° geometry and focused onto a 1.26 m spectrograph (Spex 1269, 3600 grooves/mm grating) equipped with a liquid nitrogen-cooled CCD camera (Roper Scientific). The laser power at the sample was 0.5 mW for all excitation wavelengths except 229 nm, which was 0.2 mW. Temperature-controlled experiments were performed using approximately 5 mL of aqueous solution circulated by a peristaltic pump in a wire-guided free-flowing cell. The temperature of the sample cell and the reservoir were controlled by a VWR 1162A thermostated bath containing a 50:50 mixture of glycerol and water. The temperature of the sample solution was monitored using an Omega 450ATT thermocouple. The scattered light was collected at 135° geometry and focused into a 1.26 m spectrograph (Spex 1269), which was equipped with a holographic grating (3600 groove/mm) and an intensified photodiode array detector (IPAD) (Roper Scientific). The 229 nm excitation from the frequency-quadrupled Ti:sapphire laser was used in experiments on the temperature controlled samples in the flow cell since the IPAD required triggering from a pulsed laser. The laser power at the sample was 1 mW.
An aqueous solution containing PheCN (250 μM) and the internal standard NaCN (100 mM) was used to determine the Raman excitation profile of the PheCN νCN band. Spectra of an aqueous solution of NaClO4 (97.5 mM) and NaCN (100 mM) were used to determine the Raman cross section for the CN− band at 2080 cm−1, using the ClO4− band at 932 cm−1 as the internal standard. As the ClO4− and CN− bands did not fit in the same detector window the water band at ~1640 cm−1 was used to normalize the intensities for the calculation of the Raman cross section.
The detection limits for aqueous solutions of PheCN sealed in NMR tubes with 197 nm, 229 nm and 244 nm excitation was determined using solutions of PheCN (2–100 μM) and NaCN (100 mM). The detection limit for aqueous solutions of PheCN in the presence of equimolar cytochrome c was determined using 10–20 μM solutions in pH 7.6 phosphate buffer in the thermostated flow cell with 229 nm excitation. Due to safety considerations the NaCN internal standard was omitted from the solutions in the flow cell.
The solvent dependence of the PheCN νCN band was determined by recording spectra of H2O/THF solutions (0–100% v/v H2O) containing PheCN (100 μM, except in the 100% THF solution where a saturated solution was used due to the low solubility) using 229 nm and 244 nm excitation. Aqueous solutions of PheCN (100 μM) were used to determine the temperature dependence of the Raman spectrum from 5–80 °C.
Solutions containing MPX5 (in ethanol) or POPC (in chloroform) were mixed together (or not for the controls), dried down in a glass vial under a stream of N2 or Ar, then the residual films evaporated prior to reconstitution. The MPX5:POPC ratio was 1:40 in the mixed film. Samples of MPX5, POPC or POPC plus MPX5 were reconstituted in phosphate buffer (pH 7.1, 40 mM phosphate, 100 mM NaCl) by sonicating for 30–40 min with a Branson Sonifier 250. The liposome preparations were briefly centrifuged to remove particulate matter and the clear supernatants used in the Raman experiments. Raman spectra of MPX5 in phosphate buffer pH 7.1 and in the presence of POPC liposomes were recorded using 229 nm excitation (0.5 mW). Concentrations were as listed in the figure caption. The solutions were placed in sealed quartz NMR tubes and a cooling stream of N2 from a dry ice/ethanol bath was gently blown onto the spinning sample. Raman spectra of solutions of MLCK3PheCN in pH 7.1 Tris buffer were recorded using 229 nm excitation. O2 free samples were prepared by keeping the solutions on ice and purging for at least 2 h with Ar.
Calibration and analysis of the spectra were carried out using GRAMS AI software (Thermo Electron Corp.). Acetone and DMSO-d6 were used as the calibration standards.
Spectra were recorded of aqeous solutions in a 0.100 cm quartz cell using a Agilent 8453 diode array spectrometer.
The Raman cross sections were calculated using the formula 22
where S denotes the standard, and N denotes the sample being measured. is the peak height ratio, ν0 is the excitation wavelength (cm−1), νN and νS are the wavenumbers (cm −1) of the sample and standard, respectively, is the concentration ratio, and AN, AS and A0 are the relative absorption at the sample, standard and laser excitation wavelenths. The Raman cross section of the CN− band at 2080 cm−1 was calculated using the Raman cross sections of ClO4 reported by Perno et al.25 as the standard, using interpolated values when necessary. The values determined for the CN− band at 2080 cm−1 (Supplementary Information, Table S1) were used to determine the Raman cross sections of the PheCN νCN mode at 2237 cm−1 and the PheCN ν8a phenyl ring mode at 1611 cm−1.
The Raman excitation profile for PheCN showed a strong correlation with the UV-vis absorption spectrum (Figure 1). The peak in the electronic absorption spectrum at 196 nm is attributed to a benzene-like Ba,b transition and the 233 nm peak to a benzene-like La transition. These electronic transitions are substantially red-shifted in PheCN compared to phenylalanine, where they occur at 188 nm (ε=60 mM−1 cm−1) and 206 nm (ε=10.5 mM−1 cm−1).26 The La transition in PheCN has a similar intensity (ε=13 mM−1 cm−1) to phenylalanine, but the intensity of the Ba,b transition in PheCN (ε=32 mM−1 cm−1) has decreased considerably compared to phenylalanine. The Raman cross section is higher for the ν8a band than the νCN band at all excitation wavelengths, especially at 197 nm excitation where it is more than five times larger (the ratio ranges from 1.6–2.8 for the other excitation wavelengths). It is not surprising that resonance enhancement associated with these electronic transitions is greatest for the phenyl ring mode,ν8a, but the considerable resonance enhancement also observed for the νCN band indicates that there is substantial mixing between the π electrons of the nitrile group and the aromatic ring, which is also reflected in the red-shift of the electronic absorption bands.
The largest Raman cross sections for both PheCN bands are at 197 nm, the wavelength of the most intense electronic transition, while a second peak occurs in the Raman excitation profile of both bands at longer excitation wavelengths. The second peak in the Raman excitation profile of the ν8a band is substantially red-shifted compared to the maximum in the La electronic absorption spectrum, while there is only a small red-shift in the second peak of the Raman excitation profile for the νCN band. Comparing the excitation profile of the ν8a band for PheCN with those previously determined26 for phenylalanine and tyrosine ν8a bands (Supplementary Information, Figure S1) show that the shape of the PheCN excitation profile is very similar to that of phenylalanine, but it is red-shifted due to the red-shift in the electronic absorption spectrum. The peak at longer wavelength in the excitation profile for PheCN is close to the plateau observed in the tyrosine excitation profile, but its maximum is at somewhat lower energy. At 229 nm, the excitation wavelength commonly used to observe the Raman spectra of tyrosine and tryptophan, the νCN band of PheCN has a larger Raman cross section than the phenylalanine ν8a, tyrosine ν8a and tryptophan W3 bands (Supplementary Information, Figure S2). These are the most intense Raman bands for the aromatic amino acids in the high wavenumber region (the low wavenumber tryptophan W16 and W18 bands have larger Raman cross sections than W3 at 229 nm, but they are still smaller than the Raman cross section of PheCN νCN).26 This means that the sensitivity for the PheCN νCN band is better than that of the tyrosine, tryptophan and phenylalanine under the experimental conditions most often used for recording the Raman spectra of aromatic amino acids.
Though the Raman cross section was highest at 197 nm the detection limit, ~10 μM (Supplementary Information, Figure S3), improved at 229 nm (Figure 2) and 244 nm (Supplementary Information, Figure S4), ≤5 μM, due to the higher noise level at 197 nm. Spectra were also recorded using solutions of PheCN with cytochrome c added to test the effect of the presence of protein, which also absorbs in the UV region, on the detection limits. The addition of the cytochrome c caused a small decrease in signal-to-noise ratio for the PheCN νCN band but the increase in the detection limit was small. The detection limits were ≤10 μM, and the signal- to-noise ratio indicated that the sensitivity for 229 nm excitation using the free flowing thermostated cell and the IPAD (Figure 3) was similar to that obtained using solutions sealed in quartz tubes and the CCD detector. The detection limit for PheCN using Raman spectroscopy is two orders of magnitude lower than the concentrations used in previous IR experiments on PheCN and PheCN containing peptides.5, 7, 27
The Raman spectra of a series of H2O/THF solutions of PheCN with systematically varied H2O/THF ratios were measured using 244 nm and 229 nm excitation (Figure 4 and Supplementary Information, Figure S5) to determine the sensitivity of the PheCN νCN wavenumber to the change from polar H2O to the less polar THF solvent. The shifts in the wavenumber of the PheCN νCN band were quite large, ~8 cm−1 (from 2237.9 cm−1 in 100% H2O to 2229.5 cm H2O to 2229.5 cm−1 in 100% THF for λex = 229 nm), consistent with a previous IR spectroscopic study of PheCN in H2O/THF solutions.5 The νCN shifts were much larger than those observed for the PheCN ν8a and the internal standard CN− bands, which were only ~2 cm−1 (Supplementary Information, Table S2). Even though PheCN is only slightly soluble in pure THF the sensitivity of the Raman technique was sufficient to obtain a good quality spectrum using a saturated THF solution (which was not possible with IR spectroscopy). There was a systematic decrease in wavenumber as the proportion of THF increased, leveling out at high THF concentrations. Unexpectedly, however, the wavenumber increased again as the solvent composition went from 10% H2O to 0% H2O. The peak width increased as the proportion of THF increased, reaching a maximum at 60% H2O and then decreasing. The variation in the width of the νCN band and its asymmetric shape in some of the spectra indicated that it was composed of more than one peak.
In order to determine the parameters for fitting multiple peaks to the PheCN νCN band a series of preliminary fits was carried out using two peaks to fit each spectrum. As the solvent composition changed the wavenumbers of the peaks remained fairly constant over several compositions, then as the proportion of THF increased the higher wavenumber peak would disappear and be replaced by a new band at lower wavenumber (Supplementary Information, Figure S6). Examination of the plot of the wavenumbers of all the peaks with significant intensity in the Raman spectra of the various H2O/THF solutions showed that there was a total of four species present. The wavenumbers and peak widths found for the sets of peaks in these preliminary two peak fits were used to determine average values for the wavenumbers and peak widths of these four main species (Supplementary Information, Table S3) that were then fixed and used to fit all the spectra (Figure 5). Attempts to fit all the spectra using only three species were unsuccessful. There was very little difference in the fitting parameters for the 229 nm and 244 nm excitation, and the speciation diagrams were also very similar (Figure 6 and Supplementary Information, Figure S7). The differences between the two speciation diagrams were probably due to differences in the relative Raman cross sections for the different species at 229 nm and 244 nm (the speciation diagrams show the relative intensity of the bands, not concentrations). The relative Raman cross-sections have not been determined because the total Raman intensity is affected by the change in the solvent absorption as the proportion of THF changes, as well as the difference in the cross sections between the different species, and at high THF compositions (>70% v/v THF) the solvents became immiscible when the internal standard, NaCN, was added.
The 2237.3 cm−1 and 2234.8 cm−1 bands were the major species in the predominately H2O solutions. The 2237.3 cm−1 band was the only significant species in 100% H2O, and it decreased as the % H2O decreased. The 2234.8 cm −1 band initially increased as the % H2O decreased, reached a maximum between 60–70% H2O, then decreased as the proportion of THF increased further. The 2229.4cm−1 band first became significant at 80% H2O; it increased initially, then plateaued between 10–40% H2O, with a final sharp increase in the 0% H2O solution, where it was the only significant band. The 2227.3 cm−1 band did not become significant till the % H2O had decreased to 50%, then it steadily increased as the % H2O decreased to 10%, but was not present in the 0% H2O solution. The absence of the 2227.3 cm−1 band from the 0% H2O solution explains why there was a small increase in the Raman shift for the νCN band as a whole going from the 10% H2O solution to the 0% H2O solution. This was unexpected, and it indicates that the 2227.3 cm−1 band arises from PheCN in a microenvironment that contains predominately THF but where a small amount H2O also has an important effect.
We speculate that it could be due to an arrangement such as a PheCN with the hydrophobic portions of THF surrounding molecules oriented toward the nitrile group while their O atoms are hydrogen bonded to H2O molecules in the second solvation shell. The previous IR study on PheCN in H2O/THF (10–50% H2O) solutions by Getahun et al.5 only identified three components to the νCN band (at 2235.9 cm−1, 2231.4 cm−1, and 2227.9 cm−1) due to the smaller range of solvent compositions studied. The parameters of the species in the IR study do not exactly match the parameters determined for species 1–3 in the fitting of the Raman spectra (Supplementary Information, Table S3), which may be due in part to systematic differences between the two techniques as the three species in the IR spectra all occur at slightly higher wavenumber and have narrower peak widths. However, the trends in the speciation for species 1–3 from the fitting of the Raman spectra are quite similar with the trends in the speciation from the IR measurements,5 though not surprisingly the relative magnitudes of the peak intensities differ between the two techniques.
The temperature dependence of the PheCN νCN band was determined by recording spectra of an aqueous solution from 5–80 °C. As the temperature increased there was a linear decrease in the wavenumber (Figure 7). The wavenumber decreased by 2.7 cm−1 as the temperature was increased 5 °C to 80 °C. The peak width also tended to decrease as the temperature increased (Supplementary Information, Figure S8), but the correlation with temperature was weaker. This was similar to the temperature dependence reported in the IR spectrum of PheCN by Getahun, et al.5 who found a linear shift in the nitrile stretching band position from 2238.2 cm−1 at 2.2 °C to 2234.3 cm−1 at 82.2 °C. They also found a decrease in the νCN peak width, which was attributed to a more homogeneous environment at higher temperatures. The magnitude of the temperature dependence was slightly lower in the Raman spectra (−0.034 cm−1/°C) than in the IR study (−0.048 cm−1/°C).5 The PheCN was fairly stable when heated; a single sample was used to record the spectra at all temperatures and though there was some decrease in the signal intensity as the temperature increased even at 80 °C there was still a strong signal. The shift of the νCN wavenumber with temperature was much smaller than the shift due to changes in the polarity of its environment. This means that it should be possible to use the changes in the position of this band to follow changes in protein conformation induced by temperature changes, such as T-jump initiated protein unfolding. The linearity of the temperature dependence means that this component of the shift in wavenumber can readily be controlled for when investigating changes in protein environments.
The 5-PheCN substituted mastoparan X peptide (MPX5) showed a shift in the νCN wavenumber from 2235 cm−1 in pH 7.1 phosphate buffer to 2232 cm−1 in the presence of POPC liposomes (Figure 8). This was smaller than the shift from 2235.7 cm−1 to 2229.6 cm−1 observed by IR spectroscopy7, but still readily detected. In the studies on the MPX5 peptide, pronounced decreases in the intensity of the PheCN νCN band were noted over time (Supplementary Information, Figures S9 and S10), which had not been seen in the experiments with PheCN alone. The decrease in the νCN band intensity was linear for MPX5 in pH 7.1 buffer, decreasing by 50% in ~60 min. Incorporation of the MPX5 peptide into the POPC vesicles caused the νCN band to decrease more rapidly and the kinetic plot was no longer linear, instead it was fitted with first order exponential decay model that had a time constant of 21±1 min. MPX5 adopts an αhelical conformation when it is associated with lipid membranes,7, 28 so it is possible that changes in intramolecular interactions associated with going from the less ordered solution structure to an α-helical structure in the POPC vesicles are involved in the different degradation kinetics for MPX5 in buffer and in POPC vesicles. However, the mechanism responsible for the degradation of the PheCN signal is unclear at present. Similar decreases in the intensity of the νCN band were also observed in PheCN substituted MLCK3PheCN peptide and exclusion of oxygen from the solution did not prevent the loss of intensity for the νCN band (Supplementary Information, Figure S11). Interestingly, the W3 band from tryptophan, residue 3 in MPX5, disappeared even faster than the νCN band (Supplementary Information, Figure S12); the time constant for its first order exponential decay in the POPC liposomes was 8.5±0.4 min, compared to 21±1 min for the νCN band. This degradation of the PheCN νCN and Trp W3 bands requires further study, not only as it relates to the applicability of resonance Raman spectroscopy of PheCN to monitor changes in protein environments but also because PheCN-Trp has been used as a donor-acceptor pair in a fluorescence resonance energy transfer study8 where the excitation wavelength used, 240 nm, was close to the wavelengths used to record these Raman spectra.
The strong resonance enhancement of the PheCN νCN band made it possible to record Raman spectra at concentrations a hundred-fold lower than those used previously to record the IR spectra. This greater sensitivity of Raman spectroscopy compared to IR spectroscopy for monitoring the νCN vibration of PheCN improves the flexibility of this technique. Not only is the potential number of proteins which can be studied by incorporating this non-natural amino acid as a probe increased but the types of experiments that can be done are also widened. Opportunities to study dynamical processes in proteins by the application of time-resolved Raman spectroscopy techniques, such as T-jump experiments on protein folding have been opened up. The sensitivity of the νCN band to its environment, both the polarity and specific intermolecular interactions, means that it has a wide potential to be utilized in monitoring important changes within a protein or interactions between biological molecules. Besides the demonstrated examples of the binding of a peptide to a protein5 and incorporation into a phospholipid membrane7 PheCN could be used to study changes in protein conformation that expose a buried side chain to the surrounding solvent and vice versa.
This work was supported by NIH grants GM 25158 to T.G.S. and GM54616 and GM60610 to W.F.D.