Single-molecule spectroscopy of photosynthetic proteins in solution: exploration of structure – function relationships |

In photosynthetic light harvesting, absorbed photoenergy transfers through networks of pigment – protein complexes to a central location, known as the reaction center, for conversion to chemical energy. This process occurs with remarkable near-unity quantum e ﬃ ciency. Pigment – protein complexes exhibit emergent properties very di ﬀ erent from those of their component molecules, resulting from interactions among the pigments and between the pigments and the surrounding protein environment. Thus, the precise molecular mechanisms of the energy transport process are complex and coupled, obscuring their molecular origin. Furthermore, because these interactions are sensitive to the molecular structure, they vary from complex to complex and vary in time for each individual complex due to ﬂ uctuations of the protein's conformation. To explore the e ﬀ ect of complex-to-complex variation, we study individual photosynthetic proteins, one at a time. We use a solution-phase, single-molecule technique, known as the Anti-Brownian Electrokinetic (ABEL) trap, to elucidate the conformational dynamics of single photosynthetic pigment – protein complexes without introducing additional perturbations from immobilization strategies. After reviewing the principles of the ABEL trap, we demonstrate its application to the study of several photosynthetic pigment – protein complexes. We demonstrate that the ABEL trap approach can lead to an increased understanding of photosynthetic complexes by presenting three examples: (1) analysis of photodegradation pathways, (2) characterization of complex-to-complex heterogeneity, and (3) identi ﬁ cation of distinct functional forms.


Introduction
Photosynthetic systems ourish by simultaneously balancing two tasks: (1) conversion of photoenergy to chemical energy with up to near-unity quantum efficiency; and (2) adaption to varying light intensities by safely dissipating excess photoenergy. 1 These processes occur in networks of pigment-protein complexes that absorb sunlight and transfer the photoenergy from complex to complex to reach a central location, known as the reaction center, where charge separation occurs. Pigmentpigment and pigment-protein couplings within each complex drive these initial absorption and energy transport processes because they determine the excited state energies and dynamics. These couplings are highly sensitive to inter-molecular distances; they vary drastically with small changes in the protein structure. As a result, the excited state energies and dynamics vary from protein to protein due to differences in protein conformation and vary in one protein over time due to protein uctuations. 2,3 In ensemble measurements, the excited state energies and dynamics can be obscured because only the average values are observed. In single-molecule measurements, however, the static and dynamic heterogeneity in these properties can be explored. [4][5][6][7][8][9][10][11] Pigment-protein photosynthetic complexes typically consist of an array of pigments, specically (bacterio)chlorophyll, carotenoids, and phycobilins, surrounded by a protein matrix. One well-studied pigment-protein complex is the primary antenna complex from purple bacteria, light-harvesting complex 2 (LH2). A structural model of LH2 is shown in Fig. 1a. It has a nonameric ring structure, where each monomer (shown in Fig. 1b) contains three bacteriochlorophyll (BChl) molecules and one carotenoid. 12 The protein structure gives rise to two concentric rings of BChl, known as the B800 and B850 rings, based on their absorption maxima as shown in the linear absorption spectrum (Fig. 1c). The linear absorption spectrum consists of broad, well-separated peaks across the visible and near-infrared. Disentangling the molecular origin of the excited states giving rise to these peaks is challenging because the states that absorb and transfer photoenergy are shaped by the transitions of the uncoupled pigments, pigment-pigment interaction, and multiple types of interactions between pigments and the surrounding protein. Finally, inhomogeneity, or differences from protein to protein, further broadens these peaks in ensemble measurements. For example, whether the inhomogeneity arises from ellipticity in the protein ring or from subunit-to-subunit disorder has been debated. 13,14 Overall, however, much has been learned from a variety of previous theoretical, biochemical, steady-state, and transient absorption studies. 2,7,15,16 Single-molecule uorescence spectroscopy has provided an incisive tool to characterize inhomogeneity and reveal information about the molecular origins of the excited states of pigment-protein complexes. 7 Fluorescence can be a background-free measurement, and thus has the sensitivity to report on even single photosynthetic proteins. Both experimental and theoretical work has shown protein-to-protein variation in dynamics at the single-molecule level. In experimental work, single-molecule spectra of LH2 showed heterogeneity of emission energies, [17][18][19] as well as symmetry-destroying uctuations of the protein ring. 13 Single-molecule spectra of the major lightharvesting complex of photosystem II (LHCII) from higher plants also exhibited heterogeneities in emission energy, which were attributed to conformational changes that introduced mixing with charge transfer states. 20 Recent single-molecule ultrafast experiments observed heterogeneity in the excitedstate dynamics, specically in the periods and phases of quantum coherent oscillations in the rst $400 fs aer excitation. 21 In theoretical work, heterogeneity in excited-state dynamics has also been observed, in both coherent oscillations 22 and energy and electron transfer. 23 The previous single-molecule work, however, all relied on various immobilization schemes that can perturb protein conformation. [24][25][26] Because the pigment-protein interaction depends sensitively on the protein structure, this type of perturbation can alter the energies and dynamics of the excited states of pigment-protein complexes. In fact, conicting uorescence emission intensities and spectra have been attributed to differences in immobilization schemes. 14 The Anti-Brownian Electrokinetic (ABEL) trap is a novel singlemolecule tool that enables extended observation of single biomolecules in solution. 4,27-31 By applying this technique to photosynthetic proteins, the perturbation from immobilization or encapsulation is removed, and the native inhomogeneity and dynamics can be explored. Application of the ABEL trap to photosynthetic systems has provided insight into the structure-function relationships that underlie the excited states of these systems. [32][33][34] Here, we review experiments illustrating that the combination of a solution-phase environment and a single-molecule approach reveals hidden excited state characteristics and dynamics.

The principles of the ABEL trap
The ABEL trap holds single, solution-phase uorescent particles in a micron-scale observation region for tens of seconds without perturbative attachment or encapsulation. Electrokinetic forces counter diffusion in the x-y plane. 28 A microuidic cell made of fused silica or PDMS and glass connes the particle to hundreds of nanometers in the z direction, which is less than the focal depth of an optical microscope. 28 The principle of the ABEL trap is a closed-loop feedback system, as illustrated in Fig. 2. Fluorescent particles are maintained within an observation region through three steps: (1) optical detection of position. The laser beam is constantly scanning in a micron-scale grid pattern at $kHz, which denes the observation region. When a uorescent particle diffuses into the observation region and spatially and temporally overlaps with the beam, the particle can emit a photon. When a photon is detected, the position of the particle can be approximately inferred from the known position of the beam. (2) Electric elds are applied in x and y to the highly dilute ($picomolar) sample solution held in a microuidic cell. (3) The electric elds induce electrokinetic forces 35 that move the particle back to the center of the observation region. This cycle is repeated for every detected photon to maintain the position of the uorescent particle within the observation region until it enters a dark state, at which point no position information can be extracted and the particle diffuses away. 28 Of particular note is the fact that the particle is trapped in a region of time-averaged uniform intensity. This means that on time scales longer than the beam motion, the molecule's brightness (intensity), a critical parameter for uorescence analysis, can be precisely measured, as opposed to the case with uorescence correlation spectroscopy (FCS 36 ) where molecules diffuse randomly through a non-uniform Gaussian-shaped focal spot, requiring complex analysis and averaging to extract intensity information. 37 Recent further technological improvements in the ABEL trap use a priori knowledge about particle dynamics to optimally estimate position (i.e. the Kalman lter) and apply feedback. 29,31,38 This has enabled trapping of particles as small as single uorophores, and thus demonstrates the ability to trap any soluble, uorescently-labeled particle. 30,31 This improved ability to trap molecules enables not just smaller particles, but also dimmer ones, to be studied. The ability to maintain dimmer particles within the eld of view is particularly important for studying photosynthetic systems, as these experiments rely on the intrinsic uorescence of the pigments inside the proteins, as opposed to articially attached synthetic dyes specially designed to be stable single-molecule emitters.
In the ABEL trap, the emitted uorescence intensity, lifetime, and spectrum can be simultaneously monitored for each single molecule. 30 Changes in uorescence intensity arise from changes in absorptivity and/or uorescence quantum yield of the single molecule. By using a pulsed laser and analyzing the delay times between uorescence photons and their corresponding excitation pulses, the excited-state lifetime can be determined. Finally, uorescence spectra, which provide essential information about electronic and vibronic energy structure of the system, are recorded by inserting a beam splitter into the uorescence emission path. A fraction ($30%) of the emitted photons can be spectrally dispersed and detected on an EMCCD, while the rest are used for the position estimation required for trapping and recording of uorescence intensity and lifetime. Correlated changes between these three variables on individual proteins provide particularly useful information to study the photodynamics of these systems.

Analysis of photodegradation pathways: allophycocyanin
Single-molecule experiments enable studies of multi-step processes where each step cannot be precisely synchronized.
Indeed, it is oen observed that when individuals are directly measured, discontinuous step-like changes in emission brightness are observed. One such process is photodegradation, where individual proteins can transition to multiple intermediate states in response to light before irreversibly entering a dark state (photobleaching). 39 Monitoring these processes accesses two important types of information: (1) the dynamics and intermediates in photodegradation; and (2) isolation of individual subunits. In photosynthetic systems, multiple pigments, each in its own local environment, produce overlapping spectroscopic signatures. During the photodegradation process, the photo-induced destruction of each pigment removes its spectroscopic contribution. In the case where the photo-destroyed pigment does not quench neighboring pigments, the laser can serve as a spectroscopic knock-out of individual pigments, reducing the number of pigments contributing to the measurements.
Allophycocyanin (APC) is a water-soluble pigment-protein complex located at the base of the primary antenna, the phycobilisome, in cyanobacteria and red algae. 40 It is positioned in the middle of the energy transfer chain as the excitation migrates from the antenna complexes towards the reaction center. APC is a trimer with three-fold symmetry, with three weakly-coupled dimers of phycocyanobilin pigments. The trimeric architecture of APC is not stable in solution but can be maintained using chemical crosslinking (XL). Energy relaxes within the dimer in several hundred femtoseconds, and then between the dimers on $100 ps timescale. 41,42 APC is water soluble, and thus is particularly convenient to study in solution. For this reason, APC was chosen for the rst ABEL trap study of the photodynamics of a photosynthetic antenna protein. 32 Intensity and lifetime traces of two single XL-APC complexes are shown in Fig. 3a. Single XL-APC complexes exhibit clear stepwise decreases in intensity (le axis). The lifetime (right axis) oen, but not always, shows changes correlated with the intensity changes. The correlation between intensity and lifetime is displayed by plotting each intensity level with its concomitant lifetime on a 2D scatter plot in Fig. 3b. The different states of XL-APC can be determined by clustering the points in the scatter plot, as indicated by the colors. 32 In vivo, the trimeric structure of APC is formed by proteinprotein interactions between three monomers in an equilateral triangular geometry. Each monomer protein carries two phycocyanobilin pigments. In the case of APC, in addition to enhancing the light harvesting capability of the complex, trimerization also introduces a $30 nm red shi in absorption and $20 nm red shi in emission spectrum. Taking advantage of these properties, monomer APCs can be distinguished from trimers in the ABEL trap by analyzing the initial brightness and emission spectrum of each probed molecule ( Fig. 3c and d), thus following natural dissociation of wild-type APC at singlemolecule concentrations. Here, the rst two molecules in Fig. 3c are identied as monomers due to the low initial intensity and blue-shied emission spectrum peaked at $646 nm. The last molecule, identied as an undissociated trimer, emits photons peaked at $660 nm and when entering the trap, is $3 times brighter compared to the monomers. Moreover, identied monomers frequently show sequential photobleaching of its two pigments (Fig. 3c, 0-1 seconds and 2-3.2 seconds), demonstrating the resolution of a single pigment in the complex. Full details of monomer behavior are currently under study, which could provide valuable insight in the role of selfaggregation in shaping the light-harvesting function of APC.
This experiment identies the intermediates and timescales of photodegradation of APC. The correlated decreases in uorescence intensity and lifetime of the trimer indicate an increase in quenching. Because the quenching occurs on a nanosecond timescale (based on the reduction in uorescence), the energy transfer from APC to other complexes, which occurs much more rapidly, will not be affected. The quenching, therefore, does not interfere with the light-harvesting apparatus. Importantly, the data shows that APC maintains its functionality, during the photodegradation process. As illustrated here, studying proteins on the single molecule level provides an avenue to characterize the photodegradation process, and more generally, multi-step processes, in which each step cannot be synchronized.

Characterization of heterogeneity: peridinin-chlorophyll protein
Photosynthetic proteins exhibit intrinsic heterogeneity and can also uctuate between conformations at physiological temperatures. Because solution-phase ABEL trap measurements remove any contribution from potentially perturbative immobilization schemes, which may introduce articial heterogeneity, the intrinsic static and dynamic heterogeneity of a protein can be accurately characterized.
Peridinin-chlorophyll protein (PCP) is a water-soluble pigment-protein complex from dinoagellates. It has a trimeric structure. Each monomer exhibits pseudo-twofold symmetry, containing two Chl, each of which is surrounded by four peridinins that rapidly transfer energy to the enclosed Chl. 43 Energy transfers from the Chl to membrane bound complexes to reach the reaction center. As with APC, PCP is a water-soluble protein, and thus it is particularly important to study in an aqueous environment. Although trimeric in vivo, at the low concentrations required for single-molecule studies, $99% of the protein is in the monomeric form. The remaining trimers can be identied spectroscopically, as they appear at triple the brightness of the monomeric protein.
In Fig. 4a, uorescence intensity and lifetime traces for two individual PCP complexes at two different powers are shown. Corresponding spectral dynamics were also recorded by dividing the uorescence emission into two channels (<680 nm, blue and >680 nm, red, with total uorescence intensity in black). Fluorescence intensity changes are generally correlated with lifetime changes, but uncorrelated with spectral changes. 33 The correlations between intensity, lifetime, and spectral emission are shown in the scatter plots in Fig. 4b and c. Periods of constant intensity are plotted with the concomitant lifetime (Fig. 4b) and spectral ratio (Fig. 4c). 33 The diagonal elongation of the distribution in Fig. 4b shows that uorescence intensity and lifetime are correlated. In contrast, the horizontal elongation of the distribution in Fig. 4c shows that uorescence intensity and spectrum are uncorrelated, indicating that spectral changes and lifetime changes arise from distinct molecular mechanisms. The difference between these two plots highlights the power of single-molecule techniques. While there is spread in all measurements, we can distinguish multiple different classes of transitions by simultaneously recording multiple variables on single photosynthetic proteins.
From these results, the static and dynamic heterogeneity can be extracted. Individual PCP complexes pumped with 515 nm light at 0.37 kW cm À2 irradiance exhibit uorescence intensities that range from 1 to 5 cpms (Fig. 4b), lifetimes that range from 0.5 to 6 ns (Fig. 4b), and spectral ratios that range from 0.3 to 0.8 (Fig. 4c), as well as transitions across these ranges (Fig. 4a). The observed heterogeneity and transitions are far larger than what would be expected purely from shot noise. Furthermore, the heterogeneity in these variables increases with excitation intensity. 33 As shown here, single-molecule spectroscopy enables the full distribution of variables to be determined, instead of reporting only the average value, as would be available from an ensemble experiment.
Identification of functional forms: light-harvesting complex 2 (LH2) The behaviors of photosynthetic proteins change in response to light, pH, and temperature. While ensemble structural and functional studies have produced an averaged picture, singlemolecule experiments have the ability to identify functional forms of individual complexes, such as asynchronous changes from one state to another, transient intermediates, or rare events.
LH2 exhibits an order of magnitude more heterogeneity in excited state energies than many other photosynthetic proteins, and so single-molecule experiments have been particularly insightful. Previous experiments found heterogeneity, but relied on perturbative attachment strategies. In fact, in one paper the authors attribute differences from previous results to differences in attachment strategies. 14 With the ABEL trap, we identify the unperturbed functional forms of LH2.
In Fig. 5a, uorescence intensity and lifetime traces from three sets of experiments on LH2 are shown. These three sets correspond to excitations of each pigment group (carotenoids, B800, and B850) to evaluate their individual contributions. Under all three excitation conditions, these complexes show correlated changes in uorescence intensity and lifetime. However, at higher intensities, as shown in Fig. 5a, center, asterisk, similar intensities can show dramatically different concomitant lifetimes. In agreement with previous experiments, individual LH2 complexes also showed spectral changes accompanying intensity changes as illustrated in Fig. 5b. However, spectral changes were only observed for $3% of the 1073 transitions analyzed. 34 Correlations between uorescence lifetime and intensity reveal three states, labeled as A, B, and C, in the 2D intensitylifetime plot in Fig. 5c. As shown in the traces in Fig. 5a, individual complexes switch between these three states. Interestingly, all three states exhibit the same spectra within the $1 nm resolution of the spectrometer. The intensity-lifetime photodynamics, therefore, are independent of the spectral behaviors previously characterized. 34 The intensity dependence of the populations of these three states provides insight into the molecular mechanism behind these states. States A and B appear at all intensities, Fig. 5c, le and right, where state C appears only at high intensities, Fig. 5c, right. The ratio of time in state C increases nonlinearly with excitation uence. Because of this non-linear dependence combined with molecular modeling, we attribute state C to a photobleached pigment. The rate of transition from state A to state B increased linearly with excitation uence. Thus, we assign state B to be a local conformational change, triggered by the $50-100 kT of an optical excitation. In contrast, the rate of transition from state B to state A is independent of intensity, indicating state B returns to the ground state conformation (state A) via thermal uctuations. 34 State B is a reversible, photoactivated conformation that, notably, has a pathway for quenching excess energy, as shown by the shorter uorescence lifetimes. This pathway is activated by high excitation intensities and de-activated by thermal motions, thereby providing photoprotection without interfering with function.
Because of the asynchronous nature of the transitions between states A and B, these dynamics would not be accessible in ensemble measurements. Thus, single-molecule spectroscopy, as shown here, serves as a powerful tool to identify functional conformations of photosynthetic complexes.

Conclusions and future outlook
The ABEL trap offers an unprecedented ability to explore the properties of single uorescent proteins. In a solution-phase environment, the measured heterogeneity and dynamics report on the intrinsic properties of the proteins, instead of on the properties convolved with a perturbation from attachment or encapsulation. As we have described here, this has enabled new insights into the complex structure-function relationships that underlie photosynthetic light harvesting. Specically, we have analyzed photodegradation pathways, characterized heterogeneity, and identied asynchronous switching between conformations.
Current work improving the ABEL trap to enable trapping of dimmer particles will continue to increase the number of photosynthetic proteins that can be studied with this approach. The on-going application of the ABEL trap to study photosynthetic pigment-protein complexes, such as experiments to study higher plants, will offer insight into the variety of protein architectures and dynamics found in photosynthetic organisms.
In future experiments, the impact of the wavelengths of light, pH, temperature, and chemical environment can all be explored. Using the dynamics of single proteins, the ABEL trap has the potential to provide a tool to build up a picture of how photosynthetic systems survive despite the rapid changes in natural conditions. Exploring the microscopic dynamics of single proteins will elucidate the properties of these building blocks that combine to produce the apparatus responsible for fueling most life on earth.