Structure-Correlation NMR Spectroscopy for Macromolecules Using

Article pubs.acs.org/ac

Structure-Correlation NMR Spectroscopy for Macromolecules Using Repeated Bidirectional Photoisomerization of Azobenzene Toshio Nagashima,† Keisuke Ueda,† Chiaki Nishimura,‡ and Toshio Yamazaki*,† †

RIKEN Center for Life Science Technologies, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama 230-0045, Japan Faculty of Pharmaceutical Sciences, Teikyo Heisei University, 4-21-2 Nakano, Nakano-ku, Tokyo 164-8530, Japan

‡

S Supporting Information *

ABSTRACT: Control over macromolecular structure offers bright potentials for manipulation of macromolecular functions. We here present structure-correlation NMR spectroscopy to analyze the correlation between polymorphic macromolecular structures driven by photoisomerization of azobenzene. The structural conversion of azobenzene was induced within the mixing time of a NOESY experiment using a colored light source, and the reverse structural conversion was induced during the relaxation delay using a light source of another color. The correlation spectrum between trans- and cis-azobenzene was then obtained. To maximize the efficiency of the bidirectional photoisomerization of azobenzenecontaining macromolecules, we developed a novel light-irradiation NMR sample tube and method for irradiating target molecules in an NMR radio frequency (rf) coil. When this sample tube was used for photoisomerization of an azobenzene derivative at a concentration of 0.2 mM, data collection with reasonable sensitivity applicable to macromolecules was achieved. We performed isomerization of an azobenzene-cross-linked peptide within the mixing time of a NOESY experiment that produced cross-peaks between helix and random-coil forms of the peptide. Thus, these results indicate that macromolecular structure manipulation can be incorporated into an NMR pulse sequence using an azobenzene derivative and irradiation with light of two types of wavelengths, providing a new method for structural analysis of metastable states of macromolecules.

A

changes in the properties of macromolecule sample solutions or manipulation of macromolecule structures into NMR pulse sequences.7,8,10−12 Azobenzene is widely used as a light-trigger structural manipulator, especially as a functional regulator of protein,13−15 a controller of polymer structures,16,17 and a molecular machine.18 Two-color light irradiation reversibly isomerizes azobenzene between trans and cis within a short time and without severe photodegradation. As a result of the isomerization, the 4- and 4′-carbons are 9 Å apart in the trans-form and 6 Å apart in the cis-form. Furthermore, in the case where 4,4′-diaminoazobenzene is substituted with chloroacetyl groups at its amino groups and is cross-linked with cysteine residues on a peptide through alkylation at chlorine, the average distance between the two sulfur atoms of the cysteine residues is 16.5− 18 and 11−16 Å in the trans- and cis-forms, respectively.19 Furthermore, photoisomerization of azobenzene, even azobenzene-cross-linked peptide in water, is achieved on the picoseconds-order time scale;20−24 thus, azobenzene-crosslinked proteins undergo a structural conversion during NMR acquisition in the same manner as the P- and T-jump methods.

n effective approach to elucidate the mechanism of structural transition, including protein folding, biological macromolecular complex formation, and thermal or photochemical structural conversion, is probing the correlation between structures, thermodynamics, and dynamics. Many methods to observe NMR signal containing the confirmation on structural transition are reported. The analysis of nuclear spin relaxation is applicable for dynamics analysis at an equilibrium state for macromolecules;1−3 furthermore, sample mixing,4−6 pressure jump (P-jump),7−9 and temperature jump (T-jump)10−12 are well-known methods of changing the properties of macromolecule sample solutions and can be used to detect a nonequilibrium state. The typical method of sample mixing, known as the stopped-flow method, requires time resolution of the order of several dozen milliseconds.5,6 However, the jump of the properties of sample solutions proceeds in one direction and is irreversible. To repeat sample mixing and NMR acquisition for higher dimensional experiments, a large amount of sample solution and flow-type experiments are required. Conversely, P- and T-jump processes are reversible within a short time, and the experimental system is enclosed in an NMR sample tube. Consequently, physical property jumps such as P- and T-jumps are suited to NMR data acquisition. The time resolution of these methods is less than the time of nuclear spin relaxation, including longitudinal or transversal relaxation. Thus, the P- and T-jump methods enable © XXXX American Chemical Society

Received: September 9, 2015 Accepted: October 19, 2015

A

DOI: 10.1021/acs.analchem.5b03427 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry

The H-helix mimic peptide of sperm whale myoglobin30 (H2N-GADAQGAMNKALCLFAADIAAKYCEL-COOH) was designed for azobenzene cross-linking on cysteine residues on the α-helix.28 The studies on the peptide fragments derived from whale myoglobin revealed that the helical content of Hhelix was highest among the helices.30 This peptide was mutated for the azobenzene-cross-linked experiments to cysteine at positions i and i + 11 (underscored) to conjugate azobenzene and to alanine at positions i + 3 and i + 4 to avoid contact with azobenzene. In the H-helix of myoglobin, two alanine residues exist at i + 7 and i + 8 positions in the sequence, which is convenient for the design of the peptide without steric effects. The peptide used in this work was produced by solid-phase synthesis by Toray Research Center, Inc. BSBCA22 cross-linked peptide (BSBCA22-peptide) was prepared according to a previously reported method.28 The cross-linked peptide was purified using a ZOBAX 300SB-C18 column in a gradient elution of water and acetonitrile. MALDITOF MS (m/z) calcd for C131H196N34O44S5 [M + H]+, 3110.2821; found, 3110.2646. Finally, the peptide sample was prepared at a concentration of 0.1 mM in 20 mM phosphate buffer (pH 6.5) containing 20% trifluoroethanol-d3 (v/v). To investigate the reaction rate of photoisomerization and the distribution of the photoreaction in the sample tube, BSBCA22 coupled with two cysteine molecules through its chloroacetyl group (BSBCA22-Cys) was prepared in the same manner without HPLC purification. Initially, 2.2 mM cysteine in 20 mM TrisHCl (pH 8.0) was treated with 1 mM tris(2carboxyethyl)phosphine (TCEP) as a reducing agent. Subsequently, 1 mM BSBCA22 was reacted with the reduced cysteine molecules. After the solution was incubated in a sealed tube at 40 °C for 5 h and then allowed to stand overnight under exposure to air to oxidize the remaining TCEP and cysteine, the solvent was exchanged with D2O using the freeze−drying protocol. Moreover, 0.1 mM gadodiamide, a Gd3+ chelate complex, was added to the solution to decrease the T1 relaxation time of BSBCA22-Cys. Finally, the BSBCA22-Cys sample was prepared at a concentration of 0.2 mM by dilution with D2O. NMR Spectroscopy. All NMR spectra were recorded on Bruker Avance III 800 and 900 MHz spectrometers at 25 °C. Measurements were performed in a 5 mm triple-resonance probe (TCI CryoProbe) with pulse sequences (Figure 1) using a homemade sample tube for photoreaction. All NMR data were processed using the NMRPipe31 or TopSpin software (Bruker) and were analyzed using the program NMRViewJ.32 NMR signal integrations to analyze the reaction rate were performed using IGOR Pro (WaveMetrics). Light Sources and Optics. Two optical fibers connected to two light sources were placed in an NMR tube. One fiber was connected to a UV LED (365 nm center wavelength with 7.5 nm bandwidth, Thorlabs, Inc.) and another fiber was connected to an LD (450 nm center wavelength with 2 nm bandwidth, Photon R&D, Inc.) or to a blue LED (455 nm center wavelength with 18 nm bandwidth, Thorlabs, Inc.). The core diameters of the optical fibers connected to the LED and LD were 1000 and 100 μm, respectively. The numerical aperture (NA) of the optical fibers for the LED and LD were 0.48 and 0.22, respectively. The output powers of the UV LED, blue LED, and LD (450 nm) were 104, 53, and 1000 mW at the end of 3-m-long optical fibers, respectively. The timing of light irradiation was controlled by TTL triggers from the NMR console and was synchronized with the NMR rf pulse sequence.

Kaptein and co-workers reported a method for detecting a chemical reaction by NMR; their method is known as spin coherence transfer in chemical reactions (SCOTCH).25,26 In this method, a pulsed laser with a duration of 10−15 ns is used to preserve transverse or longitudinal magnetization through a chemical reaction and cross-peaks are obtained between the reactant and product of a photoreaction. In particular, the use of a pulsed laser provided an advantage with respect to the point of transverse magnetization transfer because spin coherence in the xy-plane was not expected to dephase during the short duration of the pulsed laser. In another case, Akasaka and co-workers demonstrated that a T-jump induced by microwave heating could be performed sufficiently fast for synchronization with an NMR pulse sequence and that statecorrelated 2D NMR (SC-2D NMR) was capable of correlating distinct states of temperature-dependent protein folding.27 They successfully obtained correlation signals between the native and denatured state of ribonuclease A. In the same manner, we use the structural conversion induced by photoisomerization of azobenzene to acquire a structure-correlation spectrum between the isomers of an azobenzene-cross-linked peptide; this method is similar to the SCOTCH experiment but in a closed sample system. Although a pulsed laser was used in the SCOTCH experiment, a continuous wave (CW) light source such as a light-emitting diode (LED) or a laser diode (LD) provides a selectivity of various wavelengths and a small chassis with no moving parts, making them compatible with an NMR spectrometer’s magnetic field. To detect correlation signals, the time for structural conversion must be substantially shorter than the longitudinal relaxation time. In the case of 1H chemical shift correlation between two isomers, the structural conversion is induced during a mixing time of the NOESY pulse sequence. Therefore, the conversion time must be shorter than the T1 relaxation time and should preferably be shorter than the cross relaxation time. Here we present preliminary results indicating that crosspeaks between isomers of unsubstituted azobenzene in DMSOd6 solution were obtained in structure-correlation spectra, where bidirectional photoisomerization was repeated using the light of two different wavelengths, which was controlled by the NMR pulse sequence. Moreover, to maximize the efficiency of photoisomerization of azobenzene within the NMR rf coil and to reduce the time of structural conversion toward a protein application, we developed a novel light-irradiation NMR sample tube to promote rapid and uniform structural conversion. Finally, we acquired a structure-correlation spectrum of an azobenzene-cross-linked peptide to observe the propagation of structural change to the peptide.

■

EXPERIMENTAL SECTION NMR Sample Preparation. 2,2′-Bis(sulfonato)-4,4′-bis(chloroacetamido)azobenzene (BSBCA22) was synthesized via an alternative pathway on the basis of previous reports.28,29 The procedure used to prepare BSBCA22 is described in the Supporting Information. To perform straightforward analysis of NMR spectra, it is required to elevate the population of the converted isomer in the acquisition time of NMR. The advantage of BSBCA22 over 3,3′-bis(sulfonato)-4,4′-bis(acetamido)azobenzene (BSBAA33) lies in the high yield of trans-form under blue-light exposure used in our experiments (Supporting Information Table S1 and Figure S3). B


Article


50 or 100 ms, respectively. For aqueous solutions, a double pulsed-field-gradient spin-echo (DPFGSE) pulse train was adopted to suppress residual HOD or H2O signals. The size of the time-domain data for unsubstituted azobenzene was 4096 (R + I) × 256 (R + I) points in the F2 and F1 dimensions, respectively. Four FIDs were summed per time increment, leading to a total experiment time of 1 h 43 min. The size of the time-domain data for BSBCA22-peptide was 2048 (R + I) × 1024 (R + I) points in the F2 and F1 dimensions, respectively. Sixteen FIDs were summed per time increment, leading to a total experimental time of 10 h 10 min. During this experiment, BSBCA22-peptide was bidirectionally isomerized between its trans- and cis-forms 16 384 (1024 × 16) times.

■

RESULTS AND DISCUSSION Structural Conversion of Azobenzene in a Conventional NMR Sample Tube by Irradiation from the Top End of the rf Coil. Light irradiation isomerizes azobenzene from trans to cis under UV-light exposure and from cis to trans under blue-light exposure (Figure 2a). The structurecorrelation spectra of azobenzene in DMSO-d6 are shown in Figure 2b,c. UV and blue light are emitted from LEDs with center wavelengths of 365 and 455 nm, respectively. In the schematic of the pulse sequence in Figure 1a, UV light is illuminated only during the relaxation delay and blue light is illuminated only during the mixing time. Therefore, at the end of the relaxation delay, the population of the cis-form of azobenzene is elevated. However, at the end of the mixing time, the population of the trans-form is elevated. The converted trans-form retains the magnetization encoding the chemical shift of the cis-form. Therefore, cross-peaks between the trans (F2) and cis (F1) forms appear asymmetrically, and these peaks have the same sign as the diagonal peaks (Figure 2b). Crosspeaks smaller than the cis-to-trans cross-peaks appear on the opposite side as well. This result indicates that blue-LED irradiation during the mixing time induced not only the cis-totrans conversion but also a small occurrence of the trans-to-cis conversion. Similarly, in the correlation spectrum under reversed light-irradiation, strong cross-peaks from trans-to-cis were generated (Figure 2c). cis-to-trans conversion occurred even under UV irradiation during the mixing time, as indicated by small cross-peaks at the opposite side. The possibility of opposite photoisomerization has been described in the literature.33−35 In these experiments, light irradiation was performed using a standard 5 mm sample tube, where two lights guided through optical fibers from LEDs are illuminated in the sample solution at the top end of the NMR rf coil, similar to the conventional CIDNP experiment.36 Light is absorbed at the upper side of solution and cannot convert the structure of azobenzene with spatial uniformity. Furthermore, in these preliminary experiments, the relatively low power of the blue LED was insufficient to induce cis-to-trans conversion of the total amount of azobenzene. The conversion ratio was only 20% even with a long mixing time of 1 s (Figure 2b slice spectrum). Development of NMR Sample Tube for Light Irradiation. To apply the previously described method to structural conversion for peptides or more general proteins, a higher conversion ratio is preferable. In general, in a NOESY experiment for a peptide or protein, the mixing time is set to less than a couple of hundred milliseconds. Although the dilution of the dye to a far lower concentration is one way to shorten the time required to deliver photons to all molecules, a

Figure 1. NMR pulse sequences synchronized with UV- and blue-light irradiation. In the 1H rf row, empty and filled boxes indicate 90° and 180° of the 1H rf pulse, respectively, and smaller filled boxes indicate selective soft 180° pulses. In the pulse field gradient (PFG) row, the size of the box indicates the relative strength. Purple and blue boxes indicate irradiation of UV and blue light, respectively. (a) NOESY pulse sequence with irradiation by UV light during the relaxation delay and by blue light during the mixing time. (b) NOESY with DPFGSE water suppression. The selective 180° pulses are centered on the H2O signal. (c) Imaging pulse sequence used to detect the distribution of the photoreaction in the NMR tube. The selective rf pulses (rectangular shape of 4 ms) are centered on the selected single peak. The strength of the PFG used to dephase and rephase for imaging is 0.27 G cm−1.

The laser and LED heads were placed directly on the NMR magnet and were fixed with adhesive tape; their power supplies were positioned away from the NMR magnet. NMR Measurements. To investigate the reaction efficiency with BSBCA22-Cys under light irradiation, we initially saturated the reverse reaction by prolonged irradiation with the first light source. We then irradiated the sample using the second light source and subsequently acquired the 1D 1H NMR spectrum repeatedly. The conversion efficiency was plotted against the total duration of the second irradiation. To obtain z-axis imaging of the reaction distribution in the sample tube, the reverse reaction was initially saturated using the first light and the second light was subsequently used to irradiate the sample for the duration of each step. Then, a frequency-selective rf pulse excited 1H at position 3 (Azo3) of the azobenzene moiety of BSBCA22-Cys in trans or cis forms, and spin-echo-type imaging was performed. A total of 768 FIDs were summed with a relaxation delay of 3.7 s leading to a total experiment time of 50 min. The 2D structure-correlation spectrum was recorded using a NOESY pulse sequence with synchronized light irradiation. First, during the relaxation delay, the sample solution was irradiated with light from the UV LED. Second, during the mixing time, light from the LD (λ = 450 nm) in the case of the BSBCA22-peptide or from the blue LED in the case of the unsubstituted azobenzene was illuminated. For the unsubstituted azobenzene in DMSO-d6, the relaxation delay and the mixing time used in the experiments were 5 and 1 s, respectively. For the BSBCA22-peptide at a concentration of 0.1 mM, the relaxation delay and the mixing time were 2 s and C


Article


coil. As a result, according to the Beer−Lambert law, the decay of light intensity in a solution of azobenzene becomes smaller, resulting in uniform irradiation of the whole sample solution; therefore, a greater concentration of dye is available. For instance, in the case of a 0.2 mM solution of trans-BSBCA22 (ε365 = 24 000 M−1 cm−1, Supporting Information Figure S3) and a loss of 90% of the intensity of 365 nm light, the light path length is 2.1 mm, which is close to the radius of NMR tube. Here we adopted this technique of irradiation from the center of an NMR tube within the NMR rf coil region (Figure 3a). A roughened side surface of a quartz rod scattered the light, which entered from the top end, toward side direction and irradiated the sample solution. We preferred uniform light irradiation for the entire volume of the sample in our experiments, although lack of uniformity is not a serious problem in CIDNP experiments because the signal from the dark regions is subtracted. To solve this problem, we used upper and lower spacers that are similar to the design of a magnetic-susceptibility matched NMR tube (Figure 3a). To pierce the optical fiber with a coaxial glass tube (outer diameter, 2 mm), we made a hole in the center of both the upper and lower spacers, which were fabricated from magnetic-susceptibility matched glass obtained from Shigemi Co., Ltd. The length of the light path was 1.1 mm between the outer NMR tube (inner diameter, 4.2 mm) and inner tube 2 (outer diameter, 2 mm). Consequently, the detectable sample volume within 16 mm length of the rf coil was decreased from 220 to 170 μL; thus, sensitivity with this sample tube is 0.77 times smaller than that with a standard 5 mm NMR sample tube. The optical fiber’s roughened surface scattered light with a wider angle (relative to the axis of the optical fiber) compared to the angle of total reflection at the air−glass boundary of the outer NMR sample tube (Figure 3c). For this reason, a 2.65μm-thick dielectric mirror coating was applied to the NMR sample tube. This coating provides greater than 95% reflectivity for wavelengths from 330 to 630 nm (angle of incidence (AOI) = 5°) and from 310 to 560 nm (AOI = 45°) (Figure 3b). The dielectric multilayer coating comprised tantalum pentoxide and silicon dioxide, which are neither ferromagnetic nor electrically conductive. Therefore, no undesired effects on the NMR magnetic field were observed. However, the capacitor-like dielectric property of tantalum pentoxide adversely affected the rf tuning. The π/2 pulse of the 1H rf pulse was slightly extended to 10 μs with the dielectric-coated NMR sample tube compared to 9.6 μs in the absence of the coating. Thus, the dielectric mirror coating exhibited no remarkable effect on the NMR spectrum. Reaction Efficiency and Distribution in NMR Sample Tube. To determine the apparent reaction rate constant, we measured the dependence of the population of each isomer on the irradiation duration time (Figure 4a,b). The integral of the trans-Azo3 signal of BSBCA22-Cys was fitted as a singleexponential curve (Figure 4c,d). Finally, the time constants of the exponential decay curves were obtained; the results are summarized in Table 1. The time constant of our novel NMR tube reached 36 ms under blue-light exposure and 2.3 s under UV light exposure. According to these values, for 90% achievement of the upper limit for the isomerization of an azobenzene derivative at a concentration of 0.2 mM, irradiation times of approximately 80 ms and 5 s for exposure under blue and UV light, respectively, were calculated. In the cases where the NMR sample tube coating, tip mirror, or spacers were

Figure 2. Structure-correlation spectra of a 0.55 mM concentration of azobenzene in DMSO-d6 were acquired by light irradiation synchronized with a NOESY pulse sequence. (a) Photoisomerization of azobenzene is reversibly induced by different wavelengths of light. (b) The structure-correlation spectrum acquired using the pulse sequence shown in Figure 1a. The 1D spectrum is extracted for the c2 proton at 6.79 ppm (F1). (c) The spectrum obtained by same pulse sequence used in part b but with UV and blue light exchanged in the pulse sequence shown in Figure 1a. The 1D spectrum is extracted for the t2 proton at 7.86 ppm (F1). The relaxation delay and mixing time were set to 5 and 1 s, respectively. Red signals indicate negative and are generated by NOE.

higher output power of light and a more efficient design of NMR tube are preferred for data collection with reasonable sensitivity. Achieving such an efficient photoreaction requires a novel NMR sample tube. The method of conducting photoreactions in NMR tubes has attracted increased attention for the use of CIDNP.37,38 In a recently developed technique for light irradiation of a sample solution in an NMR tube, light is diffused from a coaxially placed optical fiber in the center of an NMR tube. Therefore, the optical fiber pierces the detection volume surrounded by the NMR rf coil.37,38 This technique provides a shorter light path length than irradiation from the top end of an NMR rf D


Article


Figure 3. (a) Three-layered NMR tube developed for light irradiation in an NMR rf coil. The outer tube is a standard 5 mm NMR tube (gray). The spacers (red) are positioned at the bottom and top of the NMR rf coil in the outer NMR tube. The upper spacer glass is fixed with inner glass tube 1 (orange). Inner glass tube 2 (blue) is positioned inside inner tube 1 and holds optical fibers and a light-scattering quartz rod that was roughened by being sandblasted with alumina abrasive. Two kinds of light are guided into the NMR tube via the optical fibers (core diameters, 100 and 1000 μm) and are introduced into the light-scattering quartz rod (diameter, 1.5 mm). Two light-guide optical fibers were bundled with a brass tube (diameter, 1.5 mm) at the end of the optical fibers. A silver mirror at the bottom of the light-scattering quartz rod was homemade using a silver-mirror reaction between AgCl and glucose as a reductant. The distance between the spacer glasses was set to 16 mm, which is the length of the NMR rf coil. (b) Dismantled NMR tube parts. The grid size in the background is 1 mm. (c) Pictures of the NMR tube for light irradiation without the dielectric mirror coating are shown in the dark state (left) and under blue-light irradiation (right).

Table 1. Dependence of the Apparent Photoreaction Rate of BSBCA22-Cys on NMR Tube Designa coating

tip mirror

spacer

+ − + +

+ + − +

+ + + −

blue/ms 36 57 46 36

± ± ± ±

3.8 3.5 2.6 4.1

UV/s 2.3 2.7 2.5 4.2

± ± ± ±

0.10 0.078 0.15 0.31

a

All experiments were performed using 0.2 mM BSBCA22-Cys. Time constants are reported as the average of the fitting results for three protons on azobenzene and both isomers.

resulted in a reaction 1.6 times faster under blue light and 1.2 times faster under UV light. At a trans-BSBCA22-Cys concentration of 0.2 mM, 28% of UV light was passed through a 1.1 mm-long light path of sample solution, as calculated from an extinction coefficient of ε365 = 24 000 M−1 cm−1 for transBSBCA22 (Supporting Information Figure S3). In contrast, 86% of blue light reached the wall of the outer NMR sample tube, as calculated from an extinction coefficient of ε450 = 2 900 M−1 cm−1 in the UV-adapted state (Supporting Information Figure S3). For this reason, the NMR sample tube coating was more effective under blue-light irradiation. However, if all the photons of emitted blue light were used for excitation of cisBSBCA22-Cys, much better improvement would be observed. The reason for this unconformity is still unknown. A silver mirror at the end of a scattering quartz rod with 1.5 mm diameter reflected residual light that reached the bottom. This improvement provided 1.3-times faster reaction under blue light and 1.1-times faster reaction under UV light. This smaller effect under UV light may be a consequence of the difference in the NA of the optical fiber. Larger emission angles lead to smaller amounts of light reaching the silver mirror.

Figure 4. Time course of the NMR spectra and buildup curve of the photoreaction under blue-light (a, c) and UV-light (b, d) irradiation. Blue circles and magenta squares represent the integrated areas of the NMR signals of trans-Azo3 and cis-Azo3 of BSBCA22-Cys (a, b), respectively. The values of the NMR integrated peak areas are normalized to the sum of isomers. Solid lines are single exponential curves fit to the experimental data.

removed, time constants were obtained under the same conditions to validate effects of these components. The NMR sample tube coating effectively reflected transmitted light through the sample solution. This improvement E


Article


light emitted from the end of the optical fiber exhibited wider divergence than that emitted from the LD (450 nm). This wider divergence explains the UV scattering at the upper region of the roughened quartz rod. However, the reaction efficiency by UV light is acceptable for isomerization of azobenzene from trans to cis during the relaxation delay. Manipulating a Peptide Structure during NMR Measurement. For the BSBCA22-peptide sample, NOESY (mixing time, 250 ms) and TOCSY (70 ms) spectra were separately acquired to assign NMR signals at the trans-enriched state (dark state) and at the cis-enriched state (UV-adapted state). To maintain the equilibrium of the UV-adapted state while thermal conversion from the cis- to the trans-form was almost inhibited (Figure S4), UV light was illuminated for 50 ms prior to the relaxation delay in each cycle of the NMR pulse sequence. HNi−HNi+1 and HNi+3−Hαi NOE signals were observed for the residues between two cysteine residues in the NOESY spectrum of BSBCA22-peptide in the dark state. Therefore, the peptide adopted an α-helix structure between azobenzene-cross-linked cysteine residues (Supporting Information Figure S5).40 This result corresponds to a previous report of a peptide linked to BSBCA through two cysteine residues in an i, i + 11 spacing.41 NOE signals were observed between HNi−HNi+1 sequential NOE of the sample in the UVadapted state, as well. However, obvious NOE signals between HNi+3−Hαi were not observed. Therefore, the cis-form peptide could not maintain the α-helix (Supporting Information Figure S5). In addition, the difference in the Hα chemical shift between the trans- and cis-forms in the cross-linked region was larger than that between the trans- and cis-forms outside the cross-linked region (Figure 6a). Furthermore, the difference in

Finally, spacers were used to remove the sample from the area with insufficient light irradiation. The spacers that were used to reduce the sample volume affected the reaction rate more strongly under UV light compared to blue light. Because blue light transmits longer distances, an excess of blue light could excite azobenzene outside the NMR rf coil. However, the power of UV light was insufficient to excite the whole BSBCA22-Cys sample including the portion outside the coil. These results indicate that no difference was observed under blue-light exposure and that a substantial improvement was observed under UV light exposure with spacers. Transient z-axis images of BSBCA22-Cys were acquired to investigate the distribution of photoreaction in the NMR tube. After blue-light irradiation, the transient distribution of cisBSBCA22-Cys is relatively flat along the z-axis of the NMR sample tube (Figure 5c) and simultaneously reaches equili-

Figure 5. Position-dependent apparent reaction curves observed for the Azo3 proton of the cis-form of BSBCA22-Cys in the photoreaction NMR tube are shown for the case of the reaction pattern of blue irradiation after 6 s of UV irradiation (a) and of UV irradiation after 200 ms of blue irradiation (b). In parts a and b, green squares and red circles represent the NMR intensity at position of the green (−5 mm) and red (+5 mm) filled bars in the z-axis imaging of NMR tubes shown in parts c and d, respectively; solid lines are the single exponential fitting curves. In part c, blue, cyan, green, orange, and red lines indicate transient distribution after blue-light irradiation for 0, 40, 80, 120, and 200 ms, respectively. In the same manner, in part d, blue, cyan, green, orange, and red lines indicate UV irradiation durations of 0, 1, 2, 4, and 6 s, respectively. Pictures (right) with z-imaging spectra show the appearance of irradiated NMR tubes without the optical coating.

Figure 6. (a) Chemical shift difference in which the cis-form is subtracted from the trans-form indicates a structural change of BSBCA22-peptide according to the isomerization of azobenzene. Model structures of peptide in cross-linking with (b) trans- and (c) cisBSBCA22 are shown. Model structures were constructed and optimized using the Avogadro software, without the use of experimental data.39

brium at the upper (−5 mm from coil center) and lower (+5 mm) positions (Figure 5a). However, after UV irradiation, the transient distribution of the cis-form was skewed along z-axis (Figure 5d) and reached an equilibrium state at the upper position (−5 mm) more quickly than at the lower position (+5 mm) (Figure 5b). This difference is evident from the appearance of light irradiation in Figure 5c,d. Because the UV light source is an LED, a high-NA lens was required to couple the LED light source to an optical fiber. Therefore, the LED

chemical shifts, which was calculated by subtracting the cis chemical shift from the trans chemical shift, was predominantly negative. Therefore, according to the theory of chemical shift index, the trans-form was more likely to be an α-helix (Figure 6b).42 Consequently, these results indicate sufficient conformational change of the peptide structure between cross-linked F


Article


Figure 7. Asymmetric structure-correlation spectrum (mixing time of 50 ms) of BSBCA22-peptide accompanied by azobenzene photoisomerization from cis to trans. The amino acid assignments indicate a correlation of the backbone amide and the aromatic proton between two isomers. Assignments of “Azo” indicate correlation in BSBCA22 and the suffix indicates the connection to each cysteine. 1H NMR spectra at the cis-enriched state (UV-adapted state) and at the trans-enriched state (blue-adapted or dark state) are shown along axes F1 and F2, respectively. Dotted blue and magenta lines indicate cross-peaks of structure-correlation spectrum between the two isomers. Appended 1D slice spectra were produced along the dotted black lines.

With this irradiation protocol, the chemical shift of the cisform (cis-enriched state) and the trans-form (trans-enriched state) appeared along axes F1 and F2, respectively; therefore, cross-peaks between the two isomers are asymmetrically shown in Figure 7 (amide and aromatic region) and in Supporting Information Figure S7 (aliphatic region). For example, the amide proton (HN) of D18 appeared in its expected position, as reflected by their chemical shifts in the spectra of both the trans- and cis-forms, which were determined from steady-state NOESY and TOCSY spectra, even though the small shift was due to a temperature increase of approximately 3 °C induced by light irradiation. Large changes due to structural conversion were observed for peaks associated with BSBCA22, which were assigned as Azo3, 5, and 6, and the amide proton of amino acid residues between cross-linked cysteine residues C13 and C24. In particular, cross-peaks from the amide proton in the C13 residue and protons at positions 3 and 5 in azobenzene connected to sulfur atom in C13 were broader along the F1 axis (cis-form chemical shift) than along the F2 axis (trans-form chemical shift). These results indicate that the structure around the C13 residue in the cis-form may fluctuate. It should be noted that the amide proton of the C13 residue in the cis-form was missing in the NOESY and TOCSY spectra of the cisenriched state. Even though the signals in the cis-form spectra

cysteine residues. The chemical shift assignments are presented in the Supporting Information Tables S2 and S3. To obtain the structure-correlation spectrum between the two isomers, blue light was illuminated during the mixing time and UV light was illuminated during the relaxation delay (Figure 1b). Since the structural conversion and NOE simultaneously build up during the mixing time, the correlation spectrum is complicated by the NOE build up in both the trans- and cis-forms. However, cross-peaks between the two isomers were stronger than diagonal peaks, in contrast to weaker cross-peaks generated by NOE. As seen in the appended spectra in Figure 7, the cross-peak from the cis-totrans conversion is larger than the diagonal peak from the cisform and the conversion ratio is 65%. This ratio is substantially better than that in the case of unsubstituted azobenzene irradiated with a low-power blue LED. The conversion time being 100 ms under blue irradiation results in a greater difference between the cross-peak and diagonal peak and the ratio is 80%. Furthermore, severe photodegradation was not observed during these experiments, even though the BSBCA22peptide underwent isomerization 16 384 times and was subjected to a total UV irradiation time longer than 9 h and an LD (450 nm) irradiation time longer than 13 min (Supporting Information Figure S6). G


Article


(8) Kremer, W.; Arnold, M.; Munte, C. E.; Hartl, R.; Erlach, M. B.; Koehler, J.; Meier, A.; Kalbitzer, H. R. J. Am. Chem. Soc. 2011, 133, 13646−13651. (9) Inoue, K.; Yamada, H.; Akasaka, K.; Herrmann, C.; Kremer, W.; Maurer, T.; Döker, R.; Kalbitzer, H. R. Nat. Struct. Biol. 2000, 7, 547− 550. (10) Gal, M.; Zibzener, K.; Frydman, L. Magn. Reson. Chem. 2010, 48, 842−847. (11) Naito, A.; Nakatani, H.; Imanari, M.; Akasaka, K. J. Magn. Reson. 1990, 87, 429−432. (12) Rinnenthal, J.; Wagner, D.; Marquardsen, T.; Krahn, A.; Engelke, F.; Schwalbe, H. J. Magn. Reson. 2015, 251, 84−93. (13) Beharry, A. A.; Woolley, G. A. Chem. Soc. Rev. 2011, 40, 4422− 4437. (14) Banghart, M.; Borges, K.; Isacoff, E.; Trauner, D.; Kramer, R. H. Nat. Neurosci. 2004, 7, 1381−1386. (15) Volgraf, M.; Gorostiza, P.; Numano, R.; Kramer, R. H.; Isacoff, E. Y.; Trauner, D. Nat. Chem. Biol. 2006, 2, 47−52. (16) Kumar, G.; Neckers, D. Chem. Rev. 1989, 89, 1915−1925. (17) Ichimura, K. Chem. Rev. 2000, 100, 1847−1874. (18) Merino, E.; Ribagorda, M. Beilstein J. Org. Chem. 2012, 8, 1071− 1090. (19) Zhang, F.; Zarrine-Afsar, A.; Al-Abdul-Wahid, M. S.; Prosser, R. S.; Davidson, A. R.; Woolley, G. A. J. Am. Chem. Soc. 2009, 131, 2283− 2289. (20) Lednev, I. K.; Ye, T. Q.; Hester, R. E.; Moore, J. N. J. Phys. Chem. 1996, 100, 13338−13341. (21) Nägele, T.; Hoche, R.; Zinth, W.; Wachtveitl, J. Chem. Phys. Lett. 1997, 272, 489−495. (22) Quick, M.; Dobryakov, A. L.; Gerecke, M.; Richter, C.; Berndt, F.; Ioffe, I. N.; Granovsky, A. A.; Mahrwald, R.; Ernsting, N. P.; Kovalenko, S. A. J. Phys. Chem. B 2014, 118, 8756−8771. (23) Satzger, H.; Spörlein, S.; Root, C.; Wachtveitl, J.; Zinth, W.; Gilch, P. Chem. Phys. Lett. 2003, 372, 216−223. (24) Satzger, H.; Root, C.; Renner, C.; Behrendt, R.; Moroder, L.; Wachtveitl, J.; Zinth, W. Chem. Phys. Lett. 2004, 396, 191−197. (25) Pouwels, P.; Kaptein, R. J. Magn. Reson., Ser. A 1993, 101, 337− 341. (26) Kemmink, J.; Vuister, G. W.; Boelens, R.; Dijkstra, K.; Kaptein, R. J. Am. Chem. Soc. 1986, 108, 5631−5633. (27) Akasaka, K.; Naito, A.; Imanari, M. J. Am. Chem. Soc. 1991, 113, 4688−4689. (28) Burns, D. C.; Zhang, F.; Woolley, G. A. Nat. Protoc. 2007, 2, 251−258. (29) Khien, L. T.; Shirling, B.; Riazanova, A. I.; Zatsepin, T. S.; Volkov, E. M.; Kubareva, E. A.; Velichko, T. I.; Pingoud, A.; Oretskaia, T. S. Bioorg. Khim. 2009, 35, 610−617. (30) Reymond, M. T.; Merutka, G.; Dyson, H. J.; Wright, P. E. Protein Sci. 1997, 6, 706−716. (31) Delaglio, F.; Grzesiek, S.; Vuister, G. W.; Zhu, G.; Pfeifer, J.; Bax, A. J. Biomol. NMR 1995, 6, 277−293. (32) Johnson, B. A.; Blevins, R. A. J. Biomol. NMR 1994, 4, 603−614. (33) Zimmerman, G.; Paik, U.; Chow, L.-Y. J. Am. Chem. Soc. 1958, 80, 3528−3531. (34) Gegiou, D.; Muszkat, K. A.; Fischer, E. J. Am. Chem. Soc. 1968, 90, 3907−3918. (35) Rau, H.; Lüddecke, E. J. Am. Chem. Soc. 1982, 104, 1616−1620. (36) Scheffler, J. E.; Cottrell, C. E.; Berliner, L. J. J. Magn. Reson. 1985, 63, 199−201. (37) Kuprov, I.; Hore, P. J. J. Magn. Reson. 2004, 171, 171−175. (38) Feldmeier, C.; Bartling, H.; Riedle, E.; Gschwind, R. M. J. Magn. Reson. 2013, 232, 39−44. (39) Hanwell, M. D.; Curtis, D. E.; Lonie, D. C.; Vandermeerschd, T.; Zurek, E.; Hutchison, G. R. J. Cheminf. 2012, 4, 17. (40) Wüthrich, K.; Billeter, M.; Braun, W. J. Mol. Biol. 1984, 180, 715−740. (41) Guerrero, L.; Smart, O. S.; Woolley, G. A.; Allemann, R. K. J. Am. Chem. Soc. 2005, 127, 15624−15629.

were missing, the signal was traceable from the trans-form through the structure-correlation spectrum.

■

CONCLUSION We fabricated a novel light-irradiation NMR sample tube to promote a rapid and uniform structural conversion of azobenzene and demonstrate structure-correlation NMR spectroscopy. The light-irradiation NMR sample tube with a highpower LD achieved up to 90% conversion from the cis- to the trans-form of an azobenzene derivative at a concentration of 0.2 mM within 80 ms. Therefore, an azobenzene-cross-linked peptide in 20% trifluoroethanol was converted according to an azobenzene structure change from cis to trans within the mixing time of a NOESY experiment. Consequently, correlation signals between the isomers appeared in the structure-correlation spectrum. The apparatus was constructed from simple optical components, required no modification of the spectrometer, required only a small modification of the existing pulse sequences to illuminate light, and enabled manipulation of macromolecular structure including proteins into the NMR pulse sequence. Other applications to a wide variety of structural analyses of macromolecules in a manner similar to P- and T-jump NMR methodologies can be envisaged.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.5b03427. BSBCA22 synthesis procedure and NMR spectra and assignments of BSBCA22-peptide (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was supported by MEXT. We thank Mr. Mitsuo Seki, Shigemi, Inc., for developing the novel NMR sample tube. We thank Dr. Jun Narusawa, Photon R&D, Inc., for his advice concerning the method of light irradiation and design of the optics. We also thank Mr. Hirohisa Kato, Taisyou Optical, Inc., for designing the dielectric optical coating. We thank Enago (www.enago.jp) for the English language review.

■

REFERENCES

(1) Palmer, A. G.; Kroenke, C. D.; Loria, J. P. Methods Enzymol. 2001, 339, 204−238. (2) Sugase, K.; Dyson, H. J.; Wright, P. E. Nature 2007, 447, 1021− 1025. (3) Neudecker, P.; Robustelli, P.; Cavalli, A.; Walsh, P.; Lundström, P.; Zarrine-Afsar, A.; Sharpe, S.; Vendruscolo, M.; Kay, L. E. Science 2012, 336, 362−366. (4) Grimaldi, J.; Baldo, J.; McMurray, C.; Sykes, B. D. J. Am. Chem. Soc. 1972, 94, 7641−7645. (5) Hamang, M.; Sanson, A.; Liagre, L.; Forge, V.; Berthault, P. Rev. Sci. Instrum. 2000, 71, 2180−2183. (6) Yushmanov, P. V.; Furó, I. J. Magn. Reson. 2005, 175, 264−270. (7) Heuert, U.; Krumova, M.; Hempel, G.; Schiewek, M.; Blume, A. Rev. Sci. Instrum. 2010, 81, 105102. H


Article

Analytical Chemistry (42) Wishart, D. S.; Sykes, B. D.; Richards, F. M. J. Mol. Biol. 1991, 222, 311−333.

I


Structure-Correlation NMR Spectroscopy for Macromolecules Using

Recommend Documents