Insights into the Excitonic States of Individual Chlorosomes from Chlorobaculum tepidum

Abstract

Green-sulfur bacteria have evolved a unique light-harvesting apparatus, the chlorosome, by which it is perfectly adapted to thrive photosynthetically under extremely low light conditions. We have used single-particle, optical spectroscopy to study the structure-function relationship of chlorosomes each of which incorporates hundreds of thousands of self-assembled bacteriochlorophyll (BChl) molecules. The electronically excited states of these molecular assemblies are described as Frenkel excitons whose photophysical properties depend crucially on the mutual arrangement of the pigments. The signature of these Frenkel excitons and its relation to the supramolecular organization of the chlorosome becomes accessible by optical spectroscopy. Because subtle spectral features get obscured by ensemble averaging, we have studied individual chlorosomes from wild-type Chlorobaculum tepidum by polarization-resolved fluorescence-excitation spectroscopy. This approach minimizes the inherent sample heterogeneity and allows us to reveal properties of the exciton states without ensemble averaging. The results are compared with predictions from computer simulations of various models of the supramolecular organization of the BChl monomers. We find that the photophysical properties of individual chlorosomes from wild-type Chlorobaculum tepidum are consistent with a (multiwall) helical arrangement of syn-anti stacked BChl molecules in cylinders and/or spirals of different size.

Introduction

The success of photosynthesis has inspired many researchers to study organic matter for solar energy conversion. However, a simple order-of-magnitude estimate reveals that under optimum conditions a typical organic molecule would absorb only a few photons per second (1,2). Hence, employing organic matter for any kind of solar-driven energy production requires an efficient light-harvesting apparatus—an antenna—for collecting as many photons as possible. One of the most efficient antenna systems found in nature is that of the green-sulfur bacteria, so-called chlorosomes. It enables these organisms to perform photosynthesis at extremely low light flux, even using the very dim radiation provided from deep sea thermal vents (3). A chlorosome is considered as a sack formed by a lipid monolayer that is densely packed with hundreds of thousands of bacteriochlorophyll (BChl) c, d, or e molecules (4–7). Within the chlorosome the chromophores are organized in stacked structures that are formed by self-assembly and stabilized by hydrogen-bonding interactions. The intermolecular interactions lead to the formation of delocalized electronically excited states, i.e., excitons, with a strong absorption that is significantly red-shifted with respect to the Q_y transitions of the BChl monomers (6,8). Upon optical excitation, the absorbed energy relaxes to the lowest exciton states on an ultrafast timescale and is then transferred within some ten picoseconds to the baseplate—a construct that accommodates BChl a-containing pigment-protein complexes, is an integral part of the monolipid envelope of the chlorosome, and mediates energy transfer to the photosynthetic reaction center (9–13). It has been found that the exact pigment composition and the morphology of the chlorosomes depends on the growth conditions and features a large degree of heterogeneity. As a consequence of this, structural information with atomic resolution about the supramolecular organization of the BChl molecules within the chlorosome is not available to date. This is therefore the subject of a long-standing debate and from various studies (8,14–19) it was concluded that the interior of the chlorosomes is filled with closely packed rod-shaped structures of 5–10 nm in diameter, but also a lamellar morphology of the BChl aggregates has been suggested (8,14,15,17–22). Recent work, which combined cryo-electron microscopy and NMR came to the conclusion that for wild-type (WT) chlorosomes from the green sulfur bacterium Chlorobaculum tepidum the BChl molecules form syn-anti concentric helical nanotubes that are oriented parallel to the long axis of the chlorosome (16,23).

Because the nature of the electronically excited states is dictated by the supramolecular organization of the monomers, optical spectroscopy has been widely used to study structure-function relationships of supramolecular aggregates (21,24–27). However, the great heterogeneity of the samples leads to inhomogeneous broadening of the spectra and subtle features are masked by ensemble averaging. This large degree of structural disorder has been confirmed by single-molecule studies (24,25,27–32). Unfortunately, most of this work was dedicated to the emission properties of the aggregates, providing information about the electronic ground state, or this research dealt with polarization properties of the excited states at one or a very few fixed excitation wavelengths. Here, we report about polarization-resolved fluorescence-excitation spectroscopy on individual WT chlorosomes from C. tepidum. The spectral information from these measurements is compared with results from computer simulations that were carried out as a function of the supramolecular organization of the aggregate. We find that the optical properties of the chlorosomes are consistent with an arrangement of the BChl molecules in syn-anti stacked cylinders and/or scrolled 2d lamellae (spirals) of different sizes.

Materials and Methods

Chlorosomes from C. tepidum WT strain TLS (ATCC 49652) were obtained by the procedure described in (32). The stock solution (OD₇₃₃ = 10 cm⁻¹) was stored in the dark at −20°C in a buffer (50 mM Tris/HCl, 10 mM Na-Ascorbat, pH 8.3 at room temperature). For experiments on ensembles (see the Supporting Material) the stock solution was diluted in the same buffer to an optical density of 0.1 at 733 nm, which corresponds to 1.1 × 10⁻³ M. The integrity of the samples was controlled by absorption spectroscopy using a commercial spectrometer (Perkin Elmer, Lambda 750 UV/VIS Spectrometer). For single-molecule experiments the stock solution was diluted to 5 × 10⁻⁹ M. For either experiment a drop from the sample solution (∼10 μl) was adsorbed onto a SiO₂ glass substrate under nitrogen atmosphere for 30 min and mounted in a helium cryostat. Measurements were performed at a temperature of 1.5 K.

For the experiments on single chlorosomes, we used a homebuilt widefield/confocal microscope as described in (32,33). The sample was illuminated with linearly polarized light from a tunable Titanium:Sapphire laser (Coherent, 899-01) (Coherent, Santa Clara, CA) that was pumped by a frequency doubled continuous-wave Neodynium-Yttrium-Vanadat (Nd:YVO₄) laser (Coherent Verdi V10). First, we recorded a wide-field image from the sample. Therefore, the excitation was set to 735 nm and the emission from the sample was collected by a microscope objective (Mikrothek, NA = 0.85) that was mounted inside the cryostat. After passing band-pass filters (BP 850/80; AHF Analysetechnik, AG, Tübingen, Germany) that block residual laser light, the signal was registered with a charge-coupled device camera (Andor LUCA^EM R 604)(Andor Technology Ltd., Belfast, UK). Next, a spatially well-isolated chlorosome was selected from the wide-field image, and the microscope was switched to confocal mode. The fluorescence-excitation spectra were obtained by scanning the laser wavelength from 715 to 795 nm with a scanning rate of 2.8 nm/s (≈47 cm⁻¹/s). This was accomplished by a motorized micrometer screw (Melles Griot Nanomover). The accuracy as well as the reproducibility of the wavelength variation were 1 cm⁻¹ and have been verified with a wavemeter (WaveMaster, Coherent). The excitation intensity was adjusted to ∼3 W/cm². Between successive scans of the fluorescence-excitation spectra the polarization of the excitation light was rotated in steps of 3° by means of a waveplate. The emission from the sample was transmitted through a band-pass filter (BP 850/80, AHF) and focused onto a single-photon counting avalanche photodiode (SPCM-AQR-15, Perkin Elmer, Waltham, MA). For reference purposes, we also conducted experiments on ensembles at both room temperature and 1.5 K. To do so, we used the same setup as described previously but replaced the microscope objective by a lens (f = 100 mm) in front of the cryostat.

Results and Discussion

Fig. 1, a, c, and e, show a stack of 250 repetitively recorded fluorescence-excitation spectra from three different individual chlorosomes, where the horizontal axis corresponds to photon energy, the vertical axis corresponds to the polarization of the incident radiation, and the intensity of the spectrum is given by the color code. We note that the emission stems solely from the baseplate, which emits around 820 nm. The black traces in Fig. 1, b, d, and f, are the spectra that result from averaging all 250 individual spectra for each case. The spectra consist of broad asymmetric bands with spectral peaks (mean, width full width half-maximum (FWHM)) at b), 13614 cm⁻¹ (13449 cm⁻¹, 772 cm⁻¹), d), 13,628 cm⁻¹ (13445 cm⁻¹, 563 cm⁻¹), and f), 13369 cm⁻¹ (13377 cm⁻¹, 642 cm⁻¹), respectively. Rotating the polarization of the excitation light between two successive scans by 3° reveals a clear modulation of the detected intensity as a function of the polarization of the incident radiation, see Fig. 1, a, c, and e.

Figure 1.

(a, c, and e) Stack of 250 fluorescence-excitation spectra from three different individual chlorosomes as a function of the polarization of the excitation light in a two-dimensional representation. The fluorescence intensity is indicated by the color code. (b, d, and f) Average of 250 individual spectra (black line), and polarization associated spectra (red, blue lines). The spectra are displayed on the same relative intensity scale that has been normalized for the average spectrum.

To analyze the modulation as a function of the polarization in more detail, we performed a global analysis of the polarization-resolved fluorescence-excitation spectra. Therefore, the spectra were binned on the excitation energy axis into 63 intervals with a width of ∼25 cm⁻¹ each. Next, we averaged the fluorescence intensity within each bin and obtained 63 intensity traces as a function of the polarization angle that were (globally) fitted with the following function:

$$A\left(\nu ,\varphi ,\delta \right)=A_0+A^{\varphi }\left(\nu \right)\cos \left(\varphi \right)+A^{\varphi +\delta }\left(\nu \right)\cos \left(\varphi +\delta \right).$$ (1)

Here, ν refers to the center excitation energy of the binned interval, A^ϕ(ν) and A^ϕ+δ(ν) are the amplitudes associated with the phase angles ϕ and (ϕ + δ), respectively, where δ corresponds to the mutual angle between the transition-dipole moments of the two spectral components, and A₀ is introduced as an offset. We have restricted the analysis to two spectral components, because i), it turned out that for more than two components the result of the fitting dependent sensitively on the initial conditions, and ii), to reduce the number of free parameters for the fitting. During the fitting process, the ϕ and (ϕ + δ) phases were treated as global parameters, i.e., kept constant as a function of the excitation energy, whereas the amplitudes were fitted separately for each excitation energy interval. In analogy to the decay-associated spectra obtained from similar procedures in time-resolved spectroscopy (34–37), we refer to the functions A^ϕ(ν) and A^ϕ+δ(ν) as polarization-resolved spectra (PAS). For the example shown in Fig. 1 a(c, e) the resulting phase angles are ϕ = 38° (122°, 47°) and δ = 78° (78°, 85°). The corresponding PAS are shown in Fig. 1 b(d, f) by the colored lines. For A^ϕ+δ(ν) (red line) we find broad bands with spectral means (FWHM) of b), 13438 cm⁻¹ (790 cm⁻¹), d), 13364 cm⁻¹ (485 cm⁻¹), and f), 13462 cm⁻¹ (620 cm⁻¹), respectively. For the other PASs, A^ϕ(ν) (blue line), the spectral means (FWHM) are b), 13505 cm⁻¹ (698 cm⁻¹), d), 13545 cm⁻¹ (496 cm⁻¹), and f), 13426 cm⁻¹ (687 cm⁻¹). This procedure uncovers unambiguously the contribution of (at least) two spectral components to the fluorescence-excitation spectra presented in Fig. 1. In the following these contributions will be referred to as higher (blue) and lower (red) energetic component according to the position of their spectral means.

Following this procedure for 102 individual chlorosomes we extract from the results the spectral means of the observed bands, the spectral separations between the means of the two PAS, and the mutual orientation of the transition-dipole moments that are associated with these bands, as shown in Fig. 2. The distribution for the spectral means of the total spectra (black bars in Fig. 2 a) is centered at 13,486 cm⁻¹ (mean) and has a width of 59 cm⁻¹ (sdev). The spectral means for the lower (red bars in Fig. 2 a) and the higher (blue bars in Fig. 2 a) spectral components are clearly shifted with respect to each other and can be characterized by (13452 ± 81) cm⁻¹ (mean ± sdev) for the lower component and (13517 ± 72) cm⁻¹ for the higher component. From the difference of the mean values the averaged separation of the two bands can be deduced and amounts to 65 cm⁻¹. However, for individual chlorosomes this splitting can vary over a range from 0 cm⁻¹ to 225 cm⁻¹ as shown in Fig. 2 b. The mean (sdev) of this distribution is 65 cm⁻¹ (50 cm⁻¹). Finally, Fig. 2 c, shows the distribution of the mutual angle δ between the transition-dipole moments of the two spectral components. It varies between 73° and 90° and the distribution is characterized by (85 ± 4)°.

Figure 2.

(a) Distributions of the spectral means for the averaged spectrum (gray), and the lower (red) and higher (blue) energetic PAS component. (b) Distribution of the energetic separations between the means of the PAS from an individual chlorosome. (c) Distribution of the phase difference between the PAS from individual chlorosomes. To see this figure in color, go online.

To associate the experimental data with the supramolecular organization of the BChl molecules within the chlorosomes, we performed numerical simulations as a function of the geometry of the BChl assembly. As a starting point for the arrangement of the BChl molecules with respect to each other, we used the unit cell reported in (23). Accordingly, the basic structural element that determines the short-range order of the chlorosomes from WT C. tepidum is a two-dimensional oblique lattice with two basis vectors with lengths of a = 1.25 nm (a axis) and b = 0.98 nm (b axis), respectively, that are oriented under an angle of 122° with respect to each other, Fig. 3 a top. Each unit cell contains two BChl molecules, one with the Mg atom at the lattice points and the other one shifted by a/2 along the a axis. The molecules are represented in Fig. 3 a by an arrow that corresponds to their Q_y transition-dipole moment. Folding this two-dimensional lattice onto the surface of a cylinder such that the a axis and the symmetry axis of the cylinder coincide, yields the structural model of syn-anti stacked BChl molecules as reported in (23). Effectively, the structure can be envisaged as a set of helices having a distance a/2 that are wrapped around a cylinder, where for each helix the transition-dipole moments of the BChl molecules point alternatingly inward or outward of the cylinder, Fig. 3 b.

Figure 3.

(a) Basic structural element used for the BChl arrangement within a chlorosome. The arrows refer to the transition-dipole moments of the BChl molecules and are arranged under an angle of β = 37°, with respect to the a axis. Each unit cell contains two molecules (green, red), with transition-dipole moments that are aligned under an angle of α = +4°, −4° with respect to the a-b plane on the two sublattices; see also top view, from which the structure has been adapted. (b) Schematic sketch of wrapping the lattice shown in (a) onto a cylinder such that the a axis is parallel to the cylinder axis. For clarity only two helices are shown. (c) Example of a simulated fluorescence-excitation spectrum for a closed cylinder with a diameter of 5.32 nm and a length of 50 nm. The diagonal disorder for the simulation was ΔE_FWHM = 575 cm⁻¹, and a Gaussian with a width of 200 cm⁻¹ (FWHM) was used for dressing the calculated stick spectrum. For more details see text and Supporting Material.

To keep things simple, the electronic excitations of the chlorosomes were described by a Heitler-London type Frenkel exciton Hamiltonian (38–42), considering the mutual interactions between the BChl molecules in dipole-dipole approximation. In particular, we have not considered a distribution of BChl c, d, or e molecules, although it is known that the site energies of these BChl derivatives may differ significantly (43). Yet, rather than introducing additional free parameters, we assigned the same site energy of E₀ = 14993 cm⁻¹ (44) to all the pigments and account for the heterogeneity in the transition energies in terms of diagonal disorder. For the magnitude of the transition-dipole moment of an individual BChl molecule we used a value of μ = 5.5 D (17,18). Diagonalization of the model Hamiltonian yields the eigenstates (exciton states) and the corresponding transition-dipole moments from which we obtain a stick spectrum (for an example see Fig.S2). It is known that the chlorosomes are oriented with their long axis parallel to the substrate surface (30), and therefore we projected the calculated transition-dipole moments into this plane and dressed each stick with a Gaussian profile to mimic a finite linewidth. Because the timescale for the relaxation of the exciton states depends on the exciton energy (9), it would be justified to vary the width of the Gaussians across the absorption band. Such an effect has been observed on single J-aggregates and was also predicted theoretically (33,45). However, for simplicity we use a single fixed value of 200 cm⁻¹ for the width of the Gaussian for dressing the stick spectra. This results in a broad spectrum that reflects the contribution of many exciton states. Subsequently, we analyzed the resulting simulated spectra with the same global fitting algorithm as the experimental spectra. This is a pragmatic approach that allows us to facilitate comparison between simulation and experiment. More details are given in the Supporting Material. An example for a calculated polarization-resolved fluorescence-excitation spectrum for a cylindrical BChl arrangement is shown in Fig. 3 c. The simulated spectra differ from the experimentally observed ones with respect to linewidths and do not reproduce the observed asymmetry of the experimental spectra. This reflects that we used a single width for dressing the stick spectra. Taking into account the underlying simplifications, the calculated pattern and the resulting PAS are in reasonable agreement with the experimental data. These simulations have been performed for five different supramolecular arrangements of the BChl molecules, namely two cylinders with diameters of 5.32 and 9.28 nm, a double wall structure from cylinders with the same radii, and two lamellae with dimensions of 50 × 19 nm² and 50 × 29 nm² that were rolled up to form scrolled 2d-lamellae that will be referred to as 2d-spirals hereafter with overall diameters of 6.57 and 8.66 nm, respectively (see Fig. 4, right column). The sizes of the structures were chosen such that they were compatible with the findings from cryo transmission electron microscopy (16) and yielded a commensurate structure when using the unit cell dimensions reported in (23). The detailed parameterization of these structures is given in the Supporting Material. For each supramolecular structure we generated 1000 spectra by Monte Carlo simulations, where we used the site energy as a random variable. This was chosen from a Gaussian distribution with a width (FWHM) of 575 cm⁻¹ (diagonal disorder), see also the Supporting Material. The resulting spectra were analyzed with the same global fitting procedure as described previously for the experimental data. The outcome of the simulations is shown in Fig. 4 where the first row reproduces the experimental data for better comparison. The underlying supramolecular arrangements of the BChl molecules are indicated by the symbols on the right side.

Figure 4.

The rows refer to different structural models for the supramolecular BChl organization as indicated by the symbols on the right side. From top to bottom: Experimental data for comparison, closed cylinder with 5.32 nm diameter, closed cylinder with 9.28 nm diameter, double-wall cylinder with 5.32 and 9.28 nm, respectively, 2d-spiral covering 426° (6.57 nm diameter), 2d-spiral covering 564° (8.66 nm diameter). The columns refer to different parameters extracted from the spectra. The first column shows the spectral means of the average spectrum (gray), the blue PAS (blue), and the red PAS (red), respectively. The second column corresponds to the spectral separation of the means of the two PAS, and the third column refers to the distributions of the relative phase between the two PAS. The numbers in the figures refer to the mean and standard deviation of the respective distribution. To see this figure in color, go online.

For all simulated spectra the distributions of the spectral means are shifted to lower energies and appear clearly narrower than the experimentally observed distributions. This is a direct consequence of the previously mentioned simplification concerning the site energies of the BChl molecules and neglecting the influence of the environment (dielectric constant ε = 1). Most important for the comparison between simulation and experiment are the spectral separations between the PAS and their mutual phase angles: these parameters are most sensitive to intermolecular interactions, i.e., the distances between and relative orientations of transition moments. In contrast to the experimental data, the model distributions of the spectral separations between the PAS are rather symmetric, and the means shift toward lower values for the larger structures, see Fig. 4, e, n vs. Fig. 4, h, q. Given the multiwall tubular assembly that was found by cryo-electron microscopy (16), and considering the large sample heterogeneity as it was found in (25,29) then the experimental histogram presented in Fig. 4 b would correspond to the superposition of distributions from several supramolecular arrangements similar to those shown in Fig. 4, e, h, k, n, and q. Given that for all these structures the phase angles are narrowly distributed around a value close to 90° in good agreement with the experimental distribution, the simulations do not allow to discriminate between cylinders or 2d-spirals as the structure determining element in the chlorosomes.

In summary, although not a proof, we find that the fluorescence-excitation spectra from individual chlorosomes of WT C. tepidum are compatible with the structural model proposed in (23). For the long-range order of the BChl arrangement it is conceivable that the interior of a chlorosome is filled with a mixture of (multiwall) cylinders and/or 2d-spirals of different size.