Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy
Spectral shift mechanisms of chlorophylls in liquids and proteins
Graphical abstract
Highlights
► Solvent shift mechanisms were studied in spectra of chlorophylls. ► Specific and non-specific effects are distinguished. ► Transition energies are obtained for pigments in solvent-free state. ► Polarizability increases upon excitation. ► Bacteriochlorophyll is sensitive to aromatic character of solvent.
Introduction
Solvent shifts are well documented in a large number of liquids for chlorophyll a (Chl a) by Seely and Jensen [1], Szalay et al. [2], North et al. [3], and others [4], [5], [6], and for bacteriochlorophyll a (BChl a) by several authors [7], [8], [9], in particular, Limantara et al. [10], Fiedor et al. [11], and this laboratory [12]. Relatively strong influence of solvent refractive index n on spectra has been found, as compared to polarity, or dielectric constant ε. Optical spectra of chlorophylls (see the structures in Fig. 1) depend also on the coordination state of central Mg ion [4], [5], [6], [7], [8], [11], [13], [14], hydrogen bonding [4], [5], [6], [15], and even the phytyl ester group [11], [16]. Analysis of spectral shifts as a function of dielectric properties can yield transition energy in free, non-solvated pigments by extrapolation [17], [18], [19], change of polarizability Δα [18], [19], and a measure of redistribution of charges upon excitation [20], including the dipole moment difference between the ground and the excited state Δμ [21], [22]. These properties, as well as characterization of spectral changes in protic (electrophilic) and nucleophilic solvents are essential for understanding pigment behaviour in proteins.
In functional photosynthetic pigment–protein complexes the absorption of chlorophylls is bathochromically shifted, as compared to solutions. Exciton interaction between closely spaced chromophores may easily cause profound transformation of spectra, such as a 100 nm red shift of BChl a in LH1 antenna B870 of purple bacteria [23]. On the other hand, it has been proposed that 9 BChl a molecules in the LH2 peripheral antenna B800 are not coupled excitonically [24], [25], [26]. Thus, a 50 nm (−800 cm−1) shift with respect to the 0–0 transition of non-solvated pigment at 750 nm [18], [19] must be due to protein environment. The vacuum frequency ν0 was an extrapolation of refractive index dependence in liquids at the room temperature. Direct measurement of fluorescence excitation spectra in cold supersonic beams [27], [28], [29], [30] has not yet been possible for thermally labile chlorophylls. Extrapolated ν0 of solvent-free BChl a and Chl a [18], [19], [31] have been used as benchmarks on a few occasions only [32], [33], and need confirmation.
Structure analysis proves that pigment environments in protein are not identical, initiating computational efforts to find individual site energies [34]. Obviously, switching off excitonic coupling is a “Gedankensexperiment”, and the site energies remain largely hypothetical. The difficult problem has been addressed by reconstitution experiments with modified pigments [35]. The computer-aided modeling and quantum chemical calculations of site energies are promising, but give controversial results [24], [25], [26], [34], [36], [37], [38], [39], [40], [41], [42]. The site energies of Chl a, serving as inputs to exciton calculations extend sometimes over 20 nm [36], [37], [38] (e.g. from 660.4 to 682 nm in CP43 [37] and from 665 to 691 nm in CP47 [38]). Calculated blue shift (655 nm in CP47 [36]) with respect to the least polarizable solvent ether (660.6 nm) [1] is hard to rationalize. On the other hand, many site energies are lower than absorption peak in aniline (674.5 nm) [1], a hydrogen-bonding, bis-ligating, aromatic liquid with high refractive index, having the largest shift observed.
Extensive modeling studies of famous Fenna–Matthews–Olson complex (FMO) of BChl a [34], [39], [40], [41], [42] revealed seven sites spread from 790 to 820 nm, in two different species of bacteria [40], whereas the solvent maxima lie between 770 ± 1 nm (ether) [7], [8], [10], [11] and 790 nm in CS2 [10]. Electrochromic shift in the protein field, acting on the dipole moment difference Δμ of 2 D may contribute to site energy shifts, extending from −135 to 195 cm−1 [40], [42]. As compared to FMO, the calculated variation in site energies is even larger for plant antennas (up to 730 cm−1 in CP47 [36]) that is hardly compatible with the negligible Δμ for Chl a (0.1 ± 0.1 D) [43].
Mutually coupled pigments embedded in protein matrix pose serious challenges as a highly complex system. The quest should start with pigment spectrum in vacuum, proceed with liquid or solid solutions, thereafter deal with monomeric pigments in protein, and finally, multi-chromophore assemblies are to be addressed. The relatively well defined protein pocket may provide some simplification, as compared to chaotic liquid or glass. Moreover, protein is an active matrix, able to structurally deform the pigment [34], [44], as well as bring charges close to it [45], and sustain internal electric fields [41]. By stark contrast, it has been shown recently [20] that a cavity in polar liquid housing a dye molecule effectively lacks even fluctuating fields. The observed moderate spectral displacements in centrosymmetric chromophores are due to self-interaction via the reaction fields created by quadrupoles and bond dipoles of the ground state [20]. In a solid-like protein the restricted mobility of polar groups would suppress the reaction field of a chromophore.
The separation of different types of interactions in liquids is carried out step by step, using sets of apolar (defined as those having ε ⩽ 1.15n2) and polar solvents, as described in our recent works [20], [22], [46]. A set of n-alkanes provides reliable extrapolation to solvent-free state [17]. Monofunctional polar solvents, obeying the point dipole model of Onsager will be used to specify the polarity dependence [46]. Solvent polarity and polarizability effects are distinguished in a bilinear regression applied to combined sets of alkanes and well-behaving polar liquids. A correct procedure was adopted for determination of effective polarity ε′, basing on Ref. [47], instead of using a difference of Lorenz–Lorentz type function ϕ(ε) − ϕ(n2) that is not tenable [22], [46].
Besides dispersive and electrostatic interactions, solvatochromic shifts can be due to formation of complexes with solvent molecules, referred to as specific solvation [4]. A noticeable red shift of the S1 band in Chl a as a result of hydrogen bonding was first demonstrated by multi-parameter regression treatment, using an empirical solvent electrophilicity (H-bond donating) parameter [31]. Later, bathochromism in protic environments was examined in detail by Krawczyk [6], [15], who confirmed hydrogen bonding to isocyclic carbonyl by means of resonant Raman spectra [48]. Separation of the effects of refractive index n, dielectric constant ε, as well as solvent nucleophilicity and electrophilicity, by applying four-parameter linear regression analysis (plus the fifth parameter to account for steric repulsion of nucleophiles), as done by us earlier [31], was not very reliable. Parameter sets subject to statistical treatment must be mutually orthogonal, and show no pair-wise correlations, even incidental ones. This requirement could be hard to satisfy, since, for example, both H-bond donating ability (electrophilicity) and ε depend on the concentration of dipolar hydroxy groups.
We emphasize the importance of using the 0–0 transition frequency of free pigment ν0 in defining the absolute spectral shift. Even combined non-specific and specific solvation cannot fully explain enhanced red shifts occurring in proteins. Linear [40], [41], [42] and quadratic electrochromism (polarization) in very strong electric fields are feasible mechanisms that require charged atoms to be placed very close to pigment [45] and, moreover, a special apolar, or very rigid local environment. The latter should be recognizable by X-ray structure analysis of pigment–proteins.
Our analysis of solvatochromism begins with apolar liquids, whose dielectric constant is solely due to electronic polarizability (ε ⩽ 1.15n2). The Lorenz–Lorentz function ϕ(n2) of n (ϕ(n2) = (n2 − 1)/(n2 + 2)) constitutes an excellent measure of solvent polarizability density. Plotting peak maxima against ϕ(n2) reveals “anomalies” with respect to normal alkanes Cn, as a reference set (Fig. 2). Dispersive shifts are enhanced in CCl4 and CS2, because of high polarizability of Cl and S atoms, for their relatively small van der Waals volumes [17], [22], [49]. Polarizable continuum model is well applicable to aliphatic liquids, consisting of elements H, C, N, and O. Dioxane and benzene are referred to as pseudopolar, because their polarity derives from the quadrupole moment [46] that is not contributing to static dielectric constant ε.
Polar solvents possess ε, exceeding optical permittivity n2 (ε > 1.15n2), owing to permanent molecular dipole moments. A model of polarizable point dipoles [50], [51] is best satisfied in case of monofunctional solvents that are aprotic and aliphatic compounds carrying a single prominent bond dipole, such as (mono)halogenides and nitriles, or several bond dipoles centered at a single atom (ethers, ketones, nitroalkanes, and DMSO), including esters and amides and alicyclic structures, but omitting oligohalogenides (e.g. CHCl3), as C–H acidic and electron accepting [22], [46]. The influence of optical and dipolar (orientational) components of solvent permittivity is postulated to be separable. As emphasized already by Gerhold and Miller [47], and several subsequent publications (see discussions in Refs. [22], [46]), the effective dielectric constant ε′ for hypothetical, non-polarizable system of dipoles cannot be calculated after Debye, as a difference between ϕ(ε) − ϕ(n2). We adopt the correct procedure, proposed by these authors. In order to find ε′, the quadratic equation (Eq. (6)) in Ref. [47] was solved:where d = a/b, a = (ε − 1)/(ε + 2) − (n2 − 1)/(n2 + 2), and b = ε/(ε + 2)(2ε + 1).
Solvent refractive index n, dielectric constant ε, its effective value ε′, and electrophilicity parameter α [52] are collected in Table 1, together with the S1 absorption peak maxima of Chl a [1], [2], [3], [4], [5], [6] and BChl a [7], [8], [9], [10], [11], [12]. The sensitivity of transition maxima ν with respect to fast and slow components of permittivity will be characterised by two-parameter linear regression, using standard Origin® program
Fitting is performed for alkanes joined with monofunctional polar solvents. Parameter sets must be orthogonal to each other [52], so correlation between n and ε′ was checked and found to be negligible. Eq. (2) yields the same intercept ν0 and slope p values, as the single parameter correlations with ϕ(n2) in alkanes (Table 2).
Solvents are grouped as illustrated on a sketch of “olympic rings” (Fig. 3). The left upper circle corresponds to the joint set used in regression, whereas the right one represents rest of apolar solvents. Lower two rings incorporate protic and aromatic liquids. Obviously, phenol belongs to both, whereas aniline is also bis-ligating with respect to Chl a. Methanol is protic and bis-ligating (L2). Benzene is apolar and aromatic, and 1,4-diazine (solid, not studied) should, in addition, be L2. Apolar dioxane and polar monofunctional tetrahydrofuran also form L2. Finally, visual separation of shift mechanisms is achieved in figures by using overlapping symbols of ○, ■, ◊, and * for alkanes and monofunctional, protic, bis-ligating, and aromatic liquids, respectively.
If several Chl a and BChl a S1 band maxima are available for the same solvent, the most outlying data are enclosed in parentheses in Figures and Table 1, as less reliable.
Section snippets
Transition frequency in isolated pigments
Accurate zero-phonon, adiabatic transition energies in simpler tetrapyrroles have been determined in fluorescence excitation spectra of cold molecular expansions [27], [28], [29], [30]. Chlorophylls are thermally less stable and non-volatile, so that ν0 may be obtained from solvatochromism by extrapolation. When spectral maxima in liquid n-alkanes are plotted as a function of ϕ(n2), the intercepts are very close to 0–0 transition energies of free chromophores in vacuum [17] (Table 3). The ν0
Comparison of a liquid to protein environment
Scientific optimists may hope that precise structure data, in conjunction with unlimited computation power would suffice to reproduce spectra and excited state dynamics in any photosynthetic protein complex. Still the detective’s work of careful disentangling interactions can be more rewarding. One should proceed from simple to complex, i.e. from isolated pigment to solution to protein. As the intermediate stepping stones, van der Waals complexes with solvent molecules [26], [30], and monomeric
Conclusions
Although the structure of a number of photosynthetic pigment–protein complexes is known with nearly atomic resolution, the absorption spectra remain a challenge even to most advanced computations. Pigment–pigment coupling is the main factor shaping the spectra in functional complexes, but the role of pigment–protein interactions can be substantial. A detailed description of spectral shifts in homogeneous liquids is indispensable for understanding pigment spectra in protein environment.
A summary
Acknowledgments
This paper is a tribute to Rein Avarmaa, our mentor and pioneer of high resolution spectroscopy of chlorophylls who died 25 years ago, 1987. We are grateful to Dr. Margus Rätsep for stimulating discussions. This work was financed by the Estonian Science Foundation Grant no. 8369.
References (112)
- et al.
Spectrochim. Acta
(1965) Biochim. Biophys. Acta
(1989)- et al.
Biochim. Biophys. Acta
(2008) - et al.
Biochim. Biophys. Acta
(1975) J. Photochem. Photobiol. A: Chem.
(1992)Chem. Phys.
(1992)J. Photochem. Photobiol. A: Chem.
(2012)- et al.
Spectrochim. Acta A
(1985) - et al.
Biophys. J.
(2000) - et al.
Biophys. J.
(2006)
Chem. Phys. Lett.
Biophys. J.
J. Photochem. Photobiol. A: Chem.
J. Photochem. Photobiol. A: Chem.
Biophys. J.
Biochim. Biophys. Acta
Biophys. J.
Biochim. Biophys. Acta
Biochim. Biophys. Acta
J. Lumin.
Chem. Phys. Lett.
Biophys. J.
Chem. Phys.
Chem. Phys.
Spectrochim. Acta A
Acta Phys. Acad. Sci. Hungar.
Acta Phys. Polon.
Photochem. Photobiol.
Biofizika
Photochem. Photobiol.
J. Am. Chem. Soc.
J. Am. Chem. Soc.
Photochem. Photobiol.
J. Am. Chem. Soc.
Annales Universitatis Mariae Curie-Skłodowska
Section AAA
Photosynth. Res.
J. Phys. Chem.
Opt. Spektrosk.
J. Phys. Chem. A
J. Phys. Chem. B
J. Phys. Chem. B
Phys. Chem. Chem. Phys.
Phys. Chem. Chem. Phys.
J. Chem. Phys.
J. Chem. Phys.
J. Chem. Phys.
J. Phys. Chem.
J. Phys. Chem.
J. Phys. Chem. B
Cited by (38)
Tuning antenna function through hydrogen bonds to chlorophyll a
2020, Biochimica et Biophysica Acta - BioenergeticsCitation Excerpt :The natural conclusion is that the Soret transitions of Chls-a with strongly‑hydrogen-bonded keto groups are red-shifted relative to those of Chls-a with free keto groups – consistent with the observed shifts in the Soret transition between the LHC proteins measured here. The literature indicates that isolated Chl-a in vitro exhibits parallel shifts in the Soret and Qy transitions [14,15,42], and so it is tempting to conclude that the presence of hydrogen bonds to the Chl-a carbonyl will also red-shift the position of this Qy band. Comparison of the spectra of LHCII obtained by two different purification methods confirms this conclusion.
Refractive index dependence of solvatochromism
2018, Journal of Photochemistry and Photobiology A: ChemistryCitation Excerpt :A number of contradictions is still pending. As a rule, we have handled the theoretical formulas as (semi)empirical ones [3–5,11,44–48]. The interaction between solute and solvent molecules treated as polarizable dipoles involves four obvious energy contributions: dispersion, polarization of solute by solvent dipole, induction (polarization of solvent by solute dipole), and the interaction between the dipoles [18,22,25].
Exciton states and optical properties of the CP26 photosynthetic protein
2018, Computational Biology and ChemistryExcitation energy transfer in ruthenium (II)-porphyrin conjugates led to enhanced emission quantum yield and <sup>1</sup>O<inf>2</inf> generation
2017, Journal of LuminescenceCitation Excerpt :However, because its molar absorptivity is only one-tenth of that of the Soret band, it would appear at most as a weak shoulder associated with the porphyrin Soret band. The 1MLCT band absorption can be further confirmed by the enhanced absorption in nonpolar solvents, which is consistent with the reported MLCT character of Ru(II)-polypyridyl moiety [19]. The Soret band of the Zn(II) porphyrin-Ru conjugates 3–4, shows substantially diminished intensity when compared to the corresponding free-base porphyrins, due to the Zn(II) metalation of the porphyrin significantly raises its π* energy level, making it the highest-excited state in these conjugates.
High hydrostatic pressure effects on spectral-optical variables of the chlorophyll pool in climacteric fruit
2016, LWTCitation Excerpt :HHP effects on absorption spectrum of chlorophyll have been studied in depth on solvated chlorophyll and isolated membrane-embedded, in wild-type and mutagen chlorophyll-binding light harvesting complexes, both from higher plants and purple photosynthetic bacteria (Ellervee, Linnanto, & Freiberg, 2004; Gall, Ellervee, Tars, Scheer, & Freiberg, 1997; Puusepp, Kangur, & Freiberg, 2015). In solution, response to HHP was characterized most often by red shift, while hypsochromic, blue shift occurred in chlorophyll due to hydrogen bonding as pointed out by Renge and Maurig (2013). Blue shift was described for gas pressure, while responses to hydrostatic pressure has been rarely studied (Renge, 2000).