Capturing the photo‐signaling state of a photoreceptor in a steady‐state fashion by binding a transition metal complex

Abstract

Binding a small molecule to proteins causes conformational changes, but often to a limited extent. Here, we demonstrate that the interaction of a CO‐releasing molecule (CORM3) with a photoreceptor photoactive yellow protein (PYP) drives large structural changes in the latter. The interaction of CORM3 and a mutant of PYP, Met100Ala, not only trigger the isomerization of its chromophore, p‐coumaric acid, from its anionic trans configuration to a protonated cis configuration, but also increases the content of β‐sheet at the cost of α‐helix and random coil in the secondary structure of the protein. The CORM3 derived Met100Ala is found to highly resemble the signaling state, which is one of the key photo‐intermediates of this photoactive protein, in both protein local conformation and chromophore configuration. The organometallic reagents hold promise as protein engineering tools. This work highlights a novel approach to structurally accessing short lived intermediates of proteins in a steady‐state fashion.

Introdution

In recent years, novel extrinsic spectroscopic probes composed of small molecules, either genetically engineered1, 2, 3 or chemically introduced4, 5, 6 into proteins, have been used to gain valuable insights into functions2, 7 and local structural dynamics and solvent electrostatics of proteins.5, 8, 9, 10, 11, 12 Water‐soluble tricarbonylchloro‐glycinato‐ruthenium (CORM3, a ruthenium carbonyl compound [RC]) is known to be capable of releasing one CO molecule under physiological conditions,13, 14 holding promises as a potential drug.15, 16, 17, 18 Using ligand exchange reactions,19 CORM3 binds specifically to the side chain of histidine, aspartic acid and/or glutamic acid, thus can be used as a probe of protein local structure,10, 20 because its carbonyl stretching vibration is a strong infrared (IR) absorber with vibrational frequency nonoverlapping with protein vibrations. Ideal candidates of such extrinsic probes should, as is often the case, introduce minimal or very limited perturbations to protein local structure.4, 5 However, protein conformational change upon binding extrinsic molecules is usually inevitable21, 22, 23 and can be critically involved in protein functions.24, 25, 26, 27 More interestingly, CORM3 can act as a novel initiator for global protein conformational change, as is shown in this work.

The novel function of CORM3 is demonstrated upon its interaction with photoactive yellow protein (PYP). PYP is a small (14 kDa) water‐soluble photoreceptor isolated from Ectothiorhodospira halophila that is the prototype of the large class of Per‐Arnt‐Sim domain signal protein. The chromophore in PYP is a deprotonated p‐coumaric acid (pCA) that connects to Cys69 through a thiol‐ester bond. Upon photoexcitation (e.g., at 440 nm), pCA undergoes a trans to cis isomerization [Fig. 1(A)], promoting the system from the ground state (pG, with λ_max = 448 nm) to a series of photo‐intermediates on the time scale of milliseconds. The most important photo‐intermediate is a blue‐shifted signaling state (pB, with λ_max = 355 nm),28, 29, 30 which has a cis and protonated pCA (by intramolecular proton transfer from the side chain of Glu4631), and exhibits significant conformational change with respect to the pG state. For the wide‐type PYP, the life time of the pB state is only about 200 ms, while in the Met100Ala (M100A) mutant, a quasisteady pB state29 is reached under continuous light illumination, therefore offering a good opportunity to examine the structure and function of this intermediate.32 Met100 is believed to participate in providing an electron‐donating environment for pCA with a long pair of electrons that stabilizes the pG state,29, 33 M100A mutant eliminates such electron‐donating power, thus slows down the recovery process of the pG state. Further, the pB state is known to be largely unstructured, showing structural signatures of unfolding intermediate.34 Herein, we report our findings on the interaction of CORM3 and PYP M100A mutant. The resultant CORM3‐bound PYP M100A is simply denoted as M100A‐RC. To our surprise, ultraviolet/visible (UV/Vis), IR, and UV circular dichroism (CD) spectroscopic evidences together point to a conclusion that a pB‐like state of PYP is formed and structurally stabilized in the M100A‐RC complex.

Figure 1.

Chromophore isomerization and structural change in the M100A mutant of PYP. (A) pCA isomerization between deprotonated trans‐isomer and protonated cis‐isomer. (B) UV/Vis absorption spectra of PYP M100A (pG and pB states in dashed and solid red lines, respectively) and RC‐bound M100A (black line). A small shoulder peaked at 365 nm for [M100A, pG] is due to partial photo‐conversion of the sample during spectral measurement (red dashed line), in which a weak band at 448 nm for M100A‐RC is due to residual unbound PYP. (C) IR difference spectrum of the pB minus pG of PYP M100A (red), and that of M100A‐RC minus the pG of M100A (black). Inset in (C) shows a comparison of the second‐derivative spectra of the corresponding IR difference spectra near 1300 cm⁻¹.

Results and disussion

Figure 1(B) depicts the UV/Vis spectra of PYP M100A with and without CORM3 binding. Unbound‐M100A in the dark state shows a main absorption peak at 448 nm due to the deprotonated pCA in the trans configuration (dashed line).31 After 1‐min light excitation at 440 nm, the pB state forms with a cis and protonated pCA, as indicated by the disappearance of the 448‐nm band and the appearance of a 354‐nm band (red solid line). Further, the pG state can be restored by light excitation at 365 nm for a few minutes (see Supporting Information, Fig. S1), indicating the photoactivity of the mutant M100A.

For M100A‐RC, the dark state shows a main absorption band peaked at about 352 nm [Fig. 1(B), black line], suggesting a chromophore configuration that highly resembles that of the M100A pB state. Further, upon 365‐nm light irradiation, only insignificant change occurs in the UV/Vis spectrum of M100A‐RC (Fig. S1). Additional light‐ or dark‐adaptation treatment does not alter its UV/Vis spectral signature. These results suggest that M100A‐RC becomes photo‐inactive and most likely is in a pB‐like state in terms of chromophore configuration.

To examine the conformational change of PYP upon RC binding, we focus on the IR spectra in the amide‐I band, which is known to be very sensitive to conformational change of proteins. The IR difference spectrum between M100A‐RC and the pG state of M100A is shown in Figure 1(C) (black curve).

For comparison, the IR difference spectrum of M100A between the pB and pG states is also shown Figure 1(C) (red curve). Clearly, key signature of the amide‐I spectral change from M100A to M100A‐RC resembles that from the pG to pB state of M100A. A decreased peak at 1641 cm⁻¹ is due to a reduced content of the α‐helical and random coil conformations, and the increased peaks at 1626 cm⁻¹ and partially at 1679 cm⁻¹ are likely due to an increased content of the anti‐parallel β‐sheet.35 For the unbound‐M100A, the spectral change from the pG to pB states is estimated to be about 2.7% of the total amide‐I peak intensity, in agreement with a previously reported value for the wild‐type PYP (2.5%36). However, this value is found to be about 8.1% from M100A to M100A‐RC. This indicates that more significant structural change occurs in protein backbone by binding M100A with the RC complex.

The conformational change was also examined by CD measurements in the UV region. The CD spectra are given in Figure S2, and the estimated secondary structures using the CDpro37, 38 with the CONTIN method,37 are summarized in Table 1. For M100A in the pG state, 21.7% of the secondary structure is α‐helix and 22.2% is β‐strand, which is in good agreement with previous CD39 and crystal structural40 results of native PYP. Upon forming the pB state, the total content of the α‐helical conformation decreases by 18.4%, while that of the β‐sheet conformation increases by 13.9%. The binding of PYP M100A by CORM3 causes more structural change: 36.4% decrease in the α‐helical content and 28.8% increase in the β‐sheet content, which can be seen in Table 1. In addition, the content of random coil increases somewhat in M100A‐RC, suggesting the formation of a slightly more disordered structure than photo‐induced pB state. Overall the CD analysis indicates the increase of β‐sheet content at the cost of α‐helix and random coil. These results show that the conformational change appears to be more significant upon CORM3 binding, which is in good agreement with the IR results shown in Figure 1(C).

Table 1.

Secondary Structure Content (in %) of PYP M100A in the pG and pB States, and That of M100A‐RC, All Solvated in H₂O, Evaluated from CD Measurement. α_R, Regular α‐Helix; α_D, Distorted α‐Helix; β_R, Regular β‐Strand; β_D, Distorted β‐Strand. Arrow Indicates the Trend of Change with Respect to the Results in the pG State of M100A

	α‐helix		β‐sheet		Turn	Random coil
	α R	α D	β R	β D
M100A (pG)	11.3	10.4	12.4	8.8	21.0	36.1
M100A (pB)	8.8↓	8.9↓	15.8↑	9.5↑	19.7	37.3↑
M100A‐RC	6.4↓	7.4↓	18.4↑	10.2↑	18.7	38.9↑

Further, the configurational change of pCA is also shown in the IR difference spectra. The negative peak at 1539 cm⁻¹ shown in Figure 1(C) is known to be due to the vibration of the trans‐chromophore in the pG state.41 The positive peaks at about 1570 cm⁻¹ and partially 1602 cm⁻¹ are due to coupled C–C and C=C stretching vibrations of chromophore's aromatic ring and vinyl group caused by the protonation of chromophore.41 The positive/negative pair of peaks at 1679 cm⁻¹ (partially)/1688 cm⁻¹ can be used as the evidence for vinyl vibration changing from the trans to cis form in pCA.42 These changes are observed to be more significantly going from unbound to RC‐bound M100A [Fig. 1(C)]. In addition, the second derivative spectra of difference IR spectra of the two cases in the 1275–1350 cm⁻¹ region was shown in Figure 1(C) (inset), where a negative peak at 1288 cm⁻¹ and a positive peak at 1301 cm⁻¹ are identified, as another piece of evidence of a cis conformation of pCA in both pB state of M100A42 and in RC‐bound M100A.

Furthermore, the prohibited photoactivity of M100A‐RC observed in the UV/Vis spectra is confirmed in the IR spectral region of pCA: 440‐ and 365‐nm light excitation caused IR spectral change in the unbound‐M100A but not in the M100A‐RC. Results are shown in Figure S3. Moreover, the negative peak at 1725 cm⁻¹ in Figure 1(C) is the signature of the disappearance of protonated Glu46 side chain36, 41, 43, 44 suggesting a deprotonated Glu46 in M100A‐RC. On the other hand, part of the broad positive peak in the region of 1570–1580 cm⁻¹ could be due the deprotonated carboxylate side chain of Glu46. Such a signature of COO⁻ is in agreement with the result of a recent work.45 Taking the above results together, we believe that M100A‐RC is in a photo‐insensitive state with both chromophore configuration and protein local environment (particularly the deprotonated Glu46) very similar to the key signatures of the signaling state of native PYP.

To identify the exact residues in PYP that were derived by the RC group we performed liquid chromatography tandem mass spectrometry (LC‐MS‐MS) analysis of CORM3‐treated PYP. The results are given in Table 2. Even though Ru(CO)₂ was found to covalently bind to five distinct amino acid side chains (one His, two Glu, and two Asp), 93% of the bound side chains are limited to two residues: His3 (31.5% binding) and Glu81 (61.5%). Thus, of the two His, 12 Asp, and 7 Glu side chains in PYP, RC‐binding was detected in only 24% of the residues. Glu46 and His108 are buried inside the protein interior, explaining the absence of binding of these residues. Our result indicates that only a subset of the remaining 19 solvent exposed Asp/Glu and His side chains are specifically bound by the RC complex.

Table 2.

Ru(CO)₂ Modified Positions and Ratio in the Sample of PYP M100A Determined from Tryptic Peptides Measurement

Glu2	1.9%
His3	31.5%
Asp20	7.3%
Asp24	4.8%
Glu81	61.5%

We now turned to investigate the IR signature of the carbonyl stretching mode of CORM3. The results are shown in Figure 2. Four peaks in two groups are shown in this region, suggesting two different binding sites in PYP for the RC complex and each complex has two CO groups remaining. A previous report19 showed that in protein environment CORM3 is in the form of Ru^II(CO)₂(H₂O)₃ after releasing one CO group, one glycine chelating ligand and one chloride ion. Protein side chains act as new ligand for the RC complex.

Figure 2.

IR spectrum of PYP M100A‐RC in the CO stretching region. Four components by Voigt function fitting (red and green) and their sum (dashed) at two proposed binding sites of the RC complex are shown: as ₁ and ss ₁ for the Glu81 site and as ₂ and ss ₂ for the His3 site. as, asymmetric stretching mode; ss, symmetric stretching mode.

Two strong peaks at 1952 and 2032 cm⁻¹ can now be assigned to the asymmetric stretching (as) and symmetric stretching (ss) modes of the two CO groups of the RC complex covalently bonded mainly to the side chain of Glu81, and the two weak peaks at 1983 and 2058 cm⁻¹ to the as and ss modes of the two CO groups of RC covalently bonded mainly to the side chain of His3. From the integrated peak area the relative binding population of Ru^II(CO)₂(H₂O)₃ to Glu81 and His3 was determined to be about 2:1. Further, using the integrated peak areas for the amide‐I and CO stretching bands, with a procedure described in Supporting Information, the molar ratio of RC to PYP was determined to be 0.93:1 for RC:PYP at the Glu81 site, and 0.48:1 for RC:PYP at the His108 site, indicating a total binding molar number of about 1.35 with a higher binding affinity to Glu81.

Fluorescence quenching method is known to be very effective in revealing the accessibility of small molecule to specific binding sites.46, 47, 48 It was found that the CORM3 resulted RC complex in PYP quenches the fluorescence emission (excited at 295 or 280 nm) resulting from Trp119, also allowing us to determine the binding affinity of the RC complex to PYP. The fluorescence spectra of PYP in the presence of mole equivalents of CORM3 from 0 to 15 were measured (Fig. S4). The binding constant, K _a, and the number of binding sites, n, between the protein and the transition metal complex, were calculated using Eq. (a)49

$$\log \frac{F_0-F}{F}=\log K_a+n\log [Q]$$ (a)

Here, F ₀ and F correspond to the fluorescence intensities of the protein in the absence and presence of a quencher (i.e., the RC‐complex), whose concentration is [Q]. Using the double logarithm regression of Eq. (a), the binding constant K _a = 2.96 × 10⁶ and number of binding sites n = 1.29 were determined by linear fitting. This suggests that RC complex binds to PYP specifically, with statistically one and one‐third binding sites, which is in reasonable agreement with the IR result shown above (n = 1.41).

It has been proposed that the pB state have a partially unfolded conformation called the “open folded” signaling state.50 The structural change of the wild‐type PYP from the pG to pB states can be assessed on the basis of available PYP crystal structures at the pG (PDB code 2PHY40) and pB (PDB code 3UME51) states, and nuclear magnetic resonance (NMR) structures of the pG (PDB code 3PHY52) and pB (PDB code 2KX653) states. The total atomic displacements from the pG to pB states of each residue (see Supporting Information for details) were summarized and averaged over the total number of atoms in each residue. The obtained averaged displacement is plotted as a function of residue index in Figure 3. In both crystal structure and NMR structure, large conformational change occurs in the region of residue 41–56 (α₃–α₄ region). In the NMR structure, however, the region of residue 1–31 (of the N‐terminal side) shows even more significant structural change, which is helpful in understanding the observed high binding affinity at His3 but not at His108. This is in agreement with previous findings that His108 is solvent inaccessible in the pG state.40, 52

Figure 3.

Residue‐based protein conformational change of PYP from the pG state to pB state. Peptide segments and corresponding secondary structures are also marked in gray streaks. (A) Computed from crystal structures of PYP at the pB state (PDB code 3UME51) and pG state (PDB code 2PHY40). (B) Computed from NMR structures of PYP at the pB state (PDB code 2KX653) and pG state (PDB code 3PHY52).

Histidine has been confirmed as a solid binding site for CORM3 in other protein.19, 20 However, there are two His residues in PYP, His3, and His108. His3 is located in the N‐terminal domain composed of random‐coil and is highly solvent exposed54 and exhibit large structural change upon pB formation (Fig. 3), which, from a kinetic perspective, is also a better binding site than His108, because the latter is in a buried site for the pG state.40, 54 Further, single and resolvable peak with reasonable spectral line width is obtained for both the as ₂ and ss ₂ modes (Fig. 2, Table S1), suggesting only a single histidine binding site present in M100A for the RC complex. Considering these results, His3 as a binding site can also be reasonably assigned even without the LC‐MS/MS result. In addition, the obtained occupation fraction of histidine from Figure 2 is lower than that of glutamic acid (0.48 vs. 0.93), suggesting an even stronger binding affinity for the Glu81 site. This is because of the strong electrostatic interaction between the positively charged [Ru(CO)₂]²⁺ and negatively charged Glu81 side chain.

Further, there are 18 solvent‐exposed Asp and Glu residues in PYP. Most of Asp and Glu residues are located in the random coil region in the pB state. Only Asp20, Asp24, and Glu81 are located in the middle of α‐helices, namely α‐helix 2 and 6, respectively. Glu81 is located in a region of α‐helix 6 that remains a more or less unchanged conformation with respect to the case in the pG state.52 However, Asp20 and Asp24 are located in the α‐helix 2 that is very short and unfolds in the pB state.53 As can be seen from Table 2, lower binding populations are seen for Asp20 and Asp24. Based on these results, we believe the binding reaction is most likely controlled by kinetics. This is why Glu81 has the largest RC occupancy among all the Asp and Glu residues. A proposed structural change from PYP M100A to M100A‐RC, with binding sites of RC complex at His3 and Glu81 are suggested in Figure 4.

Figure 4.

Proposed structural change from PYP M100A to M100A‐RC, with binding sites of RC complex in PYP M100A based on our experimental results. The NMR structure of the pB state (PDB code 2KX653) is used. Ru(CO)₂ binding to His3 and Glu81 are suggested.

The pG state of M100A mutant of PYP is less stable than that of the wild‐type PYP,29 so that a slight external perturbation (CORM3 in this case) can trigger the structural change. CORM3 then releases its original ligands (one CO, glycine and chloride) and takes new ligands (water, side chain groups of Glu81, His3, and nearby residues). As a consequence of the RC binding, protein conformation changes, and Glu46 releases a proton from the carboxylic acid side chain to the anionic pCA through hydrogen bond, allowing the isomerization to occur. This suggests that the isomerization barrier is not too high and can be overcome by protein conformational reorganization. The reported isomerization barrier is 4.4 kcal mol⁻¹,31 in agreement with this argument. The isomerization in this case is chemically driven instead of light activated, which is why the formation of M100A‐RC works well under the dark‐adapted condition (see Methods). Further, in order to recover the pG state, a hydrogen bond between the protonated Glu46 side chain (COOH) and deprotonated pCA has to be reestablished,31 which, does not occur because the RC‐resulted structural change seems to be irreversible. Thus, pCA remains in the cis and protonated configuration in the M100A‐RC complex, causing PYP to be at a pB‐like state. A refoldable pG structure is critically needed for PYP to be functional,36, 55 which is not the case for M100A when having a bound RC complex.

In addition, it seems possible for the protonated pCA to serve as a binding ligand for the RC complex. However, such a possibility can be ruled out easily because this would show a third pair of as and ss modes in the CO stretching region shown in Figure 2, because the local chemical environment of pCA significantly differs from that of either the case of Glu81 or His3.

Further, His3 has broader linewidth than Glu81 in both as and ss modes, suggesting more structural inhomogeneity in the former. This is in agreement with the known fact that the N‐terminus (where His3 is located) is more structurally disordered.

The line width is a sensitive measure of the nature of the chemical environment near a binding site. The obtained spectral line widths of the CO stretching mode for the Glu81 and His3 sites (Table S1) are much broader than those of CORM3 in water, and also broader than those reported in literature (ca. 16 cm⁻¹),20 indicating more structural inhomogeneity (solvent and protein as a whole) for the neighborhood of the RC complex. Because in the M100A‐RC complex the bandwidth of the CO stretching mode for the Glu‐site is slightly narrower than that of the His‐site (e.g., take the as ₁ and as ₂ modes), we believe the latter is less solvent exposed than the former. Under such circumstances, the bound RC complex can work as a probe of local structural as well as hydration dynamics.5

Further, it should be mentioned that crystallization of the M100A‐RC was attempted under various crystallization conditions in order to determine the RC binding site and the new conformation of PYP; however, no diffracting crystals could be obtained so far, agreeing with a more “unstructured” conformation of the M100A‐RC complex.

In summary, a novel role of a CO‐releasing molecule binding to PYP is reported. Such a binding caused a global structural change in PYP and resulted in a structure that highly mimics the signaling state of PYP in various structural and spectroscopic aspects. Our work provides an excellent example for better understanding the interaction between transition metal complexes and biomolecules. Our experimental study highlights a possibility to structurally stabilize intermediates of proteins so that their properties can be examined using desired spectroscopic methods even in a steady‐state fashion.

Materials and methods

Sample

CORM3 was purchased from Sigma and used without further purification. PYP M100A solution was prepared in 20 mM Dulbecco's phosphate‐buffered saline (DPBS) at a concentration of 2.5 mg mL⁻¹ (pH = 7.0), then CORM3 was added to the solution at a concentration of 15‐fold to protein. The mixture was stirred under the dark‐adapted condition for 2 h at room temperature. Superfluous CORM3 was removed by ultrafiltration. The final product, denoted as M100A‐RC, is in light‐yellow color because of residual unreacted PYP M100A. The concentration is about 50 mg mL⁻¹.

UV/vis measurement

UV/Vis spectra were recorded using a Shimadzu UV‐2600 UV/Vis spectrophotometer. M100A or M100A‐RC sample was solvated in 20 mM DPBS and was filled into a 1‐mm thick quartz cuvette. UV/Vis spectra in the range of 600–190 nm were collected with a resolution of 1 nm. The samples were then exposed to a 440‐nm light‐emitting‐diode (LED) source (0.72 W) for 1 min till the sample solution became colorless, immediately followed by the second UV/Vis spectral measurement for the pB state of PYP. After that, photo‐illumination at 365 nm (4 W UV light source) for 2 min was carried out till the color of sample solution turned from colorless to yellow, and immediately followed by the third UV/Vis spectral measurement for the recovered pG state.

CD measurement

CD spectra were measured using a Jasco J‐815 spectropolarimeter with wavelengths ranging from 180 to 260 nm, and using a quartz cell with 1‐mm path length, at room temperature. In order to obtain data at wavelength below 200 nm for M100A‐RC, we replaced the buffer by H₂O by ultrafiltration. Lyophilized powder of unbound‐M100A was directly solvated into H₂O. Each spectrum was accumulated three runs at a scanning speed of 1000 nm min⁻¹ with a bandwidth of 2 nm. M100A sample was dark adapted for the pG state measurement and then illuminated by the 440 nm LED light for 2 min for the pB state measurement.

FTIR measurement

Fourier‐transform infrared spectroscopy (FTIR) experiments were carried out using a Nicolet 6700 spectrometer at 1 cm⁻¹ resolution. The FTIR spectra of M100A and M100A‐RC at pH 7 in 20 mM DPBS were measured in the frequency range of 1000–4000 cm⁻¹ using a CaF₂ IR sample holder with 50‐μm Teflon spacer. The IR spectra of different samples were area normalized in the 1520–1750 cm⁻¹ region, followed by IR difference spectrum calculation.

Fluorescence measurement

Fluorescence spectra were measured using an F‐4500 fluorescence spectrometer (Hitachi) with a four‐way quartz cuvette. PYP M100A samples were excited at 280 nm and the emissions spectra were collected from 300 to 400 nm with a scan speed of 240 nm min⁻¹. PYP M100A was dissolved in H₂O at concentration of 1.9 × 10⁻⁶ M, with CORM3 added at molar ratios to PYP M100A from 0 to 15. Fluorescence spectra were collected 18 h after the addition of CORM3 to the protein to ensure a maximal extent of binding reaction between CORM3 and PYP.

Further details about the experimental methods and more results are provided in the Supporting Information.

Supporting information

Supporting Information