Changes in aggregation states of light-harvesting complexes as a mechanism for modulating energy transfer in desert crust cyanobacteria

Significance

All photosynthetic organisms regulate efficiency of excitation energy transfer (EET) to fit energy supply to biochemical demands. For desiccation-tolerant desert crust cyanobacteria this ability is of the utmost importance. These organisms spend most of the daytime in the desiccated state whence absorbed energy is completely quenched. Based on our observations we propose a model where in the hydrated state the organized rod structure of the light-harvesting phycobilisome supports directional EET to reaction centers with minimal losses to thermal dissipation. In the desiccated state this structure gives way to more random aggregates. The resulting EET path exhibits increased coupling to the environment and enhanced quenching. This energy dissipation mechanism is completely reliant on changes in aggregation state of phycobilisome components.

Abstract

In this paper we propose an energy dissipation mechanism that is completely reliant on changes in the aggregation state of the phycobilisome light-harvesting antenna components. All photosynthetic organisms regulate the efficiency of excitation energy transfer (EET) to fit light energy supply to biochemical demands. Not many do this to the extent required of desert crust cyanobacteria. Following predawn dew deposition, they harvest light energy with maximum efficiency until desiccating in the early morning hours. In the desiccated state, absorbed energy is completely quenched. Time and spectrally resolved fluorescence emission measurements of the desiccated desert crust Leptolyngbya ohadii strain identified (i) reduced EET between phycobilisome components, (ii) shorter fluorescence lifetimes, and (iii) red shift in the emission spectra, compared with the hydrated state. These changes coincide with a loss of the ordered phycobilisome structure, evident from small-angle neutron and X-ray scattering and cryo-transmission electron microscopy data. Based on these observations we propose a model where in the hydrated state the organized rod structure of the phycobilisome supports directional EET to reaction centers with minimal losses due to thermal dissipation. In the desiccated state this structure is lost, giving way to more random aggregates. The resulting EET path will exhibit increased coupling to the environment and enhanced quenching.

Deserts cover almost half of the Earth’s terrestrial surface, and although desert conditions may seem unfavorable, they are home for diverse ecosystems. Many of these ecosystems are founded on biological desert crusts, which play an essential role in stabilizing shifting sands and enriching them with nutrients (1, 2). Cyanobacteria are among the first microorganisms to inhabit these crusts where one of the major sources of water is often dew deposited before dawn (3, 4). However, as temperatures elevate, water quickly evaporates. Such conditions can be extremely harmful for photosynthetic organisms and require adaptations on all cellular levels (3–9). These include shifts in metabolic profiles and the accumulation of compatible solutes. A key issue is the adaptation of the photosynthetic apparatus because continued photosynthetic activity under high light, and especially in combination with desiccation, may lead to the production of reactive oxygen species that will cause damage to the entire cell (10–12). The cyanobacteria that colonize sand crusts evolved strategies for coping with these daily cycles of hydration using mechanisms that enable extensive quenching of absorbed light energy. The extent of quenching in these organisms far exceeds that of common laboratory model organisms (13, 14).

Our studies focused on Leptolyngbya ohadii, a crust cyanobacterium isolated from the Nizzana region of the NW Negev desert in Israel (3, 6). This is a keystone organism in this environment (4). To maintain L. ohadii cells in a viable state the desiccation process must be gradual (3). Recovery of photosynthetic activity, on the other hand, can occur immediately upon the addition of water (7). The desiccation/hydration transition does not require the breakdown and renewed production of photosynthetic protein complexes or pigments and allows L. ohadii to use the short time of morning hydration for photosynthesis and avoid photodamage during the remainder of the day (4).

A study of this process indicated similar absorption cross-sections for desiccated and hydrated cyanobacteria; that is, L. ohadii absorbs a similar amount of light energy per wavelength when hydrated and when desiccated (7). Nevertheless, in the desiccated state electron transport is shut down, and fluorescence yield decreases by ∼70%. This requires efficient dissipation of excess energy to heat during desiccation together with a mechanism for fast energy distribution. We found evidence for a block in electron transport and dissipation of excitation energy through the accumulation of a stably oxidized P₇₀₀ (primary photosystem I photochemical center). Oxidized chlorophylls are strong quenchers. Shrinkage of the luminal space, observed in transmission electron microscopy (TEM), and restriction of plastocyanin diffusion provide a possible mechanism for the accumulation of P₇₀₀⁺. However, this is not the only protective mechanism at play in the desiccated state. Our data also indicated a strong quenching reaction in the phycobilisome (PBS) light-harvesting antenna (7).

PBSs are assembled by the association of phycobiliproteins (PBPs) to which linear tetrapyrroles are covalently linked. Typically, PBS cores contain cylinders made up of hexamers of the lowest energy absorbing PBP variant, allophycocyanin (APC; λ_max= 650 nm). PBS rods can be composed of a number of different PBPs (15). In the case of L. ohadii the rods contain phycocyanin (PC; λ_max= 620nm).

In this study we dissected PBS fluorescence quenching processes in L. ohadii. We performed spectroscopic measurements on both hydrated and desiccated samples of L. ohadii, to uncover energy transfer pathways between components of the PBS system in the different states. Structural organization of the interthylakoidal space was examined by neutron and X-ray diffraction techniques. The results suggest a different excitation energy transfer (EET) path within the PBS antenna in hydrated and desiccated samples. The desiccated samples’ EET path exhibits an efficient widespread quenching effect, resulting from interactions of PBS pigments.

Results

Under laboratory conditions L. ohadii is able to retain viability for prolonged periods and regain photosynthetic activity immediately after rehydration. We have demonstrated this in the past for electron transfer activity (7). Fig. 1 demonstrates this remarkable ability for PBS function. A desiccated culture, in which PBS fluorescence was strongly quenched, regained its high fluorescence yield as soon as 2 min after rehydration. The hydration/desiccation process was completely reversible. After ∼2 h, at ambient temperature, the culture was desiccated, and fluorescence was quenched. Rehydration resulted in immediate recovery (Fig. 1).

Fig. 1.

Reversibility of fluorescence yield during desiccation and rehydration. A desiccated L. ohadii culture was exposed to two cycles of desiccation and rehydration. Fluorescence spectra were measured with an excitation wavelength of 580 ± 2 nm. The traces are baseline shifted. The fluorescence trace of the desiccated sample did not exhibit any discernible peaks (0-min curve). However, immediately after rehydration, PBS fluorescence was observed (2 min). The peak is dominated by PC fluorescence at 650 nm and contains a 670-nm APC shoulder. The filter was then desiccated for 180 min (2 to 180 min), before a second rehydration round. Again, an immediate recovery of PBS fluorescence was observed (182 min).

To understand the energy transfer dynamics controlling this process we conducted measurements using a streak camera system that provides picosecond time resolution as well as spectral information. Measurements were performed on L. ohadii samples in the desiccated and the hydrated states. The excitation wavelength was set to 580 nm, exciting mainly the PC component of the PBSs (15, 16). Fig. 2 presents an example of the raw data collected at room temperature (Fig. 2 A and B) and at 77 K (Fig. 2 C and D) from hydrated and desiccated samples. Measurements conducted at room temperature best represent natural conditions. Measurements at 77 K allow better spectral resolution of the various fluorescence peaks.

Fig. 2.

Propagation of photosynthetic fluorescence following 580 nm excitation with picosecond time resolution and spectral information. (A and B) Data for room temperature measurements. (C and D) The 77-K measurements. The fluorescence intensity was normalized (from 0 to 1), and the scale is presented on the right-hand side. In the desiccated samples we observe a shortening of the lifetime and a redshift of the fluorescence, compared with the hydrated sample. The differences are highlighted by white dashed arrows. In the hydrated state at 77 K, three spectral features emerge: ∼650–670, ∼670–700, and ∼720–750 nm. These are normally ascribed to PBS, PSII, and PSI fluorescence, respectively. However, we observed fluorescence at wavelengths >700 nm in isolated L. ohadii PC (Fig. S2).

Several differences appear very clearly when comparing the raw room temperature data of the two states (highlighted by the arrows in Fig. 2). The first is the shortening of the desiccated sample’s florescence lifetime across all wavelengths. The second prominent difference is the wavelength broadening of the desiccated sample fluorescence spectrum toward the red, compared with the hydrated sample. These observations are in line with the lower-resolution time-correlated single-photon counting data in ref. 7. The temporal and spectral resolutions of the measurements reported here provide an opportunity to examine faster processes, occurring before thermal equilibration.

To resolve both spectral and temporal components, global analysis (17) was conducted on the data [Fig. 3; analysis by evolution associated spectra (EAS) (17) is presented in Fig. S1]. Table S1 summarizes the lifetimes of the various components. The best minimal fits for the data were obtained with four fluorescence decay lifetimes.

Fig. 3.

Global analysis of desiccated and hydrated states spectrally resolved picosecond fluorescence. DAS and corresponding fluorescence lifetimes of (A) hydrated and (B) desiccated samples. (C–F) A comparison between the corresponding normalized DAS of hydrated and desiccated components. Each DAS is marked by its corresponding lifetime. The DAS of desiccated cells are dashed. A summary for the observed lifetime ranges in three to four repeats is included in Table S1. In addition to DAS, the data were analyzed using a sequential model, and the obtained EAS is presented in Fig. S1.

Fig. S1.

EAS analysis of spectrally resolved picosecond fluorescence at RT. EAS and corresponding fluorescence lifetimes of (A) hydrated and (B) desiccated samples. (C–F) Comparison between the corresponding hydrated and desiccated EAS components.

Table S1.

Global analysis lifetimes

Component	Desiccated [average (range)]	Hydrated [average (range)]
Room temperature
n	3	4
First	4 (3.2–4) ps	5 (4.5–5.5) ps
Second	16 (15–18) ps	34 (30–37) ps
Third	64 (60–66) ps	117 (111–128) ps
Fourth	233 (221–240) ps	306 (294–315) ps
77 K
n	2	3
First	10 (10–10.5) ps	12 (10.5–12.5) ps
Second	37 (37–37) ps	45 (41–47.5) ps
Third	127 (113–141) ps	143 (130–152) ps
Fourth	443 (437–449) ps	498 (490–510) ps

Lifetimes of DAS components presented in Fig. 2. The average lifetime and the range between the maximum and minimum values recorded for the biological repeats are presented.

The 5- and 3.5-ps decay-associated spectra (DAS) of hydrated and desiccated samples, respectively (Fig. 3 A and B), reflect excitation equilibration following absorption at 580 nm. The 5-ps DAS obtained for the hydrated sample are negative at all wavelengths. One of the peaks of these 5-ps DAS is located at 650 nm, and the other is located around 700–720 nm (Fig. 3C). The lifetime (τ) and the DAS shape are rather similar to those observed in the laboratory model organism Synechocystis sp. PCC 6803 (16). In the desiccated sample the resolved lifetime of the above component is somewhat shorter (Table S1). The difference in the lifetimes of the 3.5- and 5-ps DAS is insignificant because the full width at half maximum of the instrument response function (IRF) is around 3.5–5 ps, and this limits resolving a clear difference between these two short lifetimes. The shapes of the 3.5- and 5-ps DAS of the desiccated and hydrated states show significant differences. In the desiccated state the ∼650-nm peak is broader, and instead of a distinct peak at ∼720 nm, a broad shoulder extending from 680 to 740 nm is observed (Fig. 3).

In steady-state measurements, isolated PC from L. ohadii exhibited two peaks: one centered at ∼650 nm and another broader peak at ∼685 nm. The latter extends well beyond 700 nm (Fig. S2). Similar steady-state spectra were observed in PC from other cyanobacterial species (see, for example, refs. 18, 19 and many others). Based on the similarities with the in vitro data we attribute the first DAS component to PC EET. However, these are in vivo measurements, and the possibility of emission from photosystem I (PSI) cannot be excluded. A significant contribution of PSI is, however, unlikely considering the 580-nm excitation wavelength.

Fig. S2.

Steady state fluorescence spectra of phycocyanin isolated from Leptolyngbya ohadii. (A) The emission spectrum (blue) was measured following excitation at 580 nm, whereas the red Gaussian fitted to the spectrum shows two emission peaks: one at 649 nm and another at 685 nm. (B) The excitation spectra of the 650- and 700-nm emission bands. Inset represents the corresponding absorption spectra of the sample.

The 34-ps DAS in the hydrated state shows a positive peak at ∼640 nm and a negative peak at ∼660 nm, indicating energy transfer from ∼640 to ∼660 nm (Fig. 3D). This DAS was previously attributed, in a laboratory model cyanobacterium, to internal PBS energy transfer, i.e., from PC to PC and from PC to APC₆₆₀ (16, 20). The corresponding DAS in the desiccated state has a lifetime of 16 ps (Fig. 3D), which is markedly different from the 34-ps DAS in the hydrated state. Although the 34-ps DAS of the hydrated state shows EET, the 16-ps DAS of the desiccated state lacks a clear negative amplitude, indicating a major reduction of the corresponding EET process.

The 117-ps DAS in the hydrated state partly reflects EET from APC₆₆₀ to the terminal emitter APC₆₈₀ or to chlorophylls fluorescing at 680 nm (16). The 64-ps DAS of the desiccated state lacks a negative peak, indicating the absence of EET compared with the hydrated state. The EAS of this component show a prominent red shift and broadening of the desiccated state spectrum (Fig. S1).

The 306-ps DAS of the hydrated state and the 233-ps DAS of the desiccated state represent the disappearance of an equilibrated excitation distribution over all of the components of the system, namely, PC, APC₆₆₀, APC₆₈₀, and chlorophylls. It is faster for the desiccated state, and both DAS and EAS indicate a stronger contribution on the long-wavelength side.

The comparison between the desiccated and the hydrated fluorescence kinetics indicates (i) significantly reduced downhill EET (Fig. 3 D and E) toward the RCs and (ii) a substantial decrease of the fluorescence lifetimes, meaning that an additional nonradiative decay pathway has been created.

To test this further we performed measurements at 77 K (Fig. 2). A detailed description of the DAS decomposition is included in Fig. S3. At 77 K the fluorescence lifetimes of both hydrated and desiccated states are longer compared with room temperature. Furthermore, the differences between the desiccated and hydrated samples are smaller. Global analysis results indicate the convergence of both DAS shapes and lifetimes (Fig. S3). At 77 K, PSI and photosystem II (PSII) fluorescence can be resolved. Nevertheless, in the desiccated samples the DAS decomposition indicates reduced EET from the PBSs to both photosystems.

Fig. S3.

DAS decomposition of 77-K measurements. DAS and corresponding fluorescence lifetimes of (A) hydrated and (B) desiccated samples. (C–F) Comparison between the corresponding hydrated and desiccated normalized DAS components. At 77 K the contribution of PSII, PSI, and PBS components can be better resolved spectrally (5). Both the 11.5-ps DAS of the hydrated state and the 10.5-ps DAS of the desiccated state indicate EET from 640 to 660 nm. They also show EET from 640 or 690 to 730 nm. This wavelength range corresponds to PSI fluorescence in the steady-state spectra (5). Based on the lower intensity of the negative peak at ∼660 nm, the relative proportion of the PBS EET process is smaller in the desiccated sample. The 46-ps DAS of the hydrated state and the 36-ps DAS of the desiccated state have a positive peak at 650 nm accompanied by a negative feature at 670–690 nm that can be attributed to EET to APC₆₈₀ and PSII. This feature is smaller in the desiccated state and not as well resolved. The 150-ps DAS of the hydrated state and the 141-ps DAS of the desiccated state look similar between 610 and 700 nm, but they are different around 730 nm, where a negative peak appears only in the hydrated sample. We can speculate that the initial fast EET to 730 nm observed in the 11.5-ps DAS of the hydrated state is a result of chlorophyll excitation and chl-chl EET. Although 580-nm excitation is far from the chlorophyll absorption peak, the extinction coefficient at this wavelength is, nevertheless, not zero. The EET pathway from PC to PSI is longer and observed only in the 150-ps DAS of the hydrated state. This transfer is blocked in the desiccated state. The 500-ps DAS of the hydrated state exhibits three distinct peaks that can be associated with longer decay times of PBSs, PSII, and PSI. The 660-nm PBS peak, which is prominent in the 500-ps DAS of the hydrated state, is nearly absent in the 437-ps DAS of the desiccated state.

The analysis of the measurements performed at room temperature and 77 K support the existence of a fast heat dissipation pathway in the desiccated state. This channel is apparently far more efficient at room temperature than at 77 K. Quenching of excitation energy might be explained by the function of the orange carotenoid protein (OCP). OCP is photoactivated by high blue–green irradiance, binds to APC, and leads to the dissipation of excess energy absorbed in the PBSs under high-light conditions (21, 22). However, OCP is not a likely candidate for the nonradiative dissipation observed in the desiccated state. The extent of fluorescence quenching measured in desiccated L. ohadii is much larger (Fig. 1) than that induced by OCP in other cyanobacteria (e.g., refs. 16, 22–24). Moreover, OCP binds and facilitates energy dissipation by interaction with the APC core (25). However, in the desiccated state we observed only little, if any, energy transfer from PC to APC. Furthermore, the overall effect of OCP knockout or overexpression on results of streak camera measurements is much smaller and focused to APC (16).

We can consider two additional alternatives for the mechanism underlying the activity of the dissipation pathway. The first is a change within the structure of PBS pigment–protein complexes. Such a change can, for example, enhance charge transfer reactions between a bilin pigment and an adjacent amino acid, as demonstrated in the FMO (green sulfur bacteria antenna protein) and isiA (cyanobacterial iron stress induced antenna protein) chlorophyll binding proteins (26, 27). The second is a change in the aggregation state of the PBS proteins leading to the generation of intermediate-coupling interactions between them.

The visible absorption spectra of hydrated and desiccated samples did not show any significant difference in cross-section or in line shape (7). Additional circular dichroism (CD) measurements, in the visible range, failed to detect significant differences as well (Fig. S4). These data do not provide evidence for a major change in the energetic properties of the individual pigments in the hydrated and desiccated states. Furthermore, EPR spectroscopy did not detect a radical that can be associated with a charge transfer reaction in the desiccated state (7).

Fig. S4.

CD spectra. CD spectra of hydrated and desiccated samples (as indicated in the graph). The data are averaged over 4–20 repeats, baseline subtracted, and normalized to the absorption of the red chlorophyll peak. The most characteristic peaks include the chlorophyll-associated peaks at ∼445 and the split band in the red with the positive counterpart. In cyanobacteria these are predominantly associated with PSI. The PBS peak is at ∼620, and the ∼510-nm band represents a carotenoid signal. All of these features can be observed both in the hydrated and in the desiccated sample. The experiment was performed as described in ref. 63.

On the other hand, previous cryo-TEM studies of the ultrastructure of the interthylakoidal space harboring the PBSs did indicate differences between the two states (7). In the hydrated state the interthylakoidal space was well organized with uniform repeat distances and apparent rod structures traversing it. Such PBS architecture was also observed in Acaryochloris marina (28). Both species effectively grow as biofilms. The well-organized architecture is lost in the desiccated state (an example of the TEM observations is included in Fig. S5). Additional ultrastructural information was gained through small-angle X-ray and neutron scattering experiments (SAXS and SANS; Fig. 4). These techniques were used in a number of studies to measure the repeat distances (RDs) in different thylakoid membrane systems in vivo and in vitro (29–31). While providing 1D information, they do not require freezing or fixation.

Fig. S5.

Examples of TEM images of hydrated and desiccated samples. The repeated structures in the space between thylakoid envelopes that are clearly evident in the hydrated state are lost in the desiccated state (for methods and a thorough analysis of the TEM data, see ref. 7).

Fig. 4.

Small-angle X-ray and neutron scattering measurements. Hydrated and desiccated samples were studied by solution (A) SAXS and (B) SANS. The data presented were azimuthally integrated from the 2D raw data and plotted as intensity as a function of the magnitude of the scattering vector, q. In both cases, several weak intensity peaks were observed (indicated by arrows). In B, azimuthally integrated SANS scattering curves were measured at 20 m (hydrated sample) and overlapped 20 and 8 m (desiccated sample) sample–detector distances. The line represents the fitted curve (linear combination of a constant, and a power function and three Gaussians). No peak could be detected between 0.35 and 3.5 nm⁻¹ on SANS curves. The first-order q value of the features observed in the measurements can be translated into a repeat distance RD = 2π/q. Prominent features are marked directly on the graphs.

Both SANS and SAXS scattering profiles carry information about key structural parameters, such as RDs of periodically organized multilamellar membrane systems, but with different sensitivities. The SAXS signal arises from differences in the electron density distribution, whereas SANS is sensitive to protons. In the presence of D₂O, SANS allows distinguishing scattering length density between the aqueous regions and the lipid membrane (32). Also, SANS can usually measure in smaller q ranges than SAXS. Accordingly, SAXS data provide information on relatively smaller intrathylakoidal RDs between electron densities defined by proteins and the lipid bilayer, e.g., in the luminal space, whereas SANS data provide information on the RD of thylakoid membranes.

In the hydrated sample, SAXS features are observed at q = 0.46, 0.57 nm⁻¹ (Fig. 3A). These features are replaced by ones with higher q values in the desiccated state (0.71 and 1.04 nm⁻¹; q = 1.38 nm⁻¹ might be the second harmonics of the principal peak at q = 0.71 nm⁻¹). These q values can be interpreted as decreased width of the thylakoid envelope in the desiccated state, which shrinks from 11–14 nm to 6–9 nm.

SANS data reported a feature at q = 0.072 nm⁻¹ with two additional higher harmonic features at 0.139 and 0.221 nm⁻¹. These values can be interpreted as the RD ≈ 87 nm of the interthylakoidal space. This is slightly larger than reported for Synechocystis 6803 (80 nm in ref. 32 and 60 nm in ref. 31) and might indicate larger light-harvesting antennae. In the desiccated state these characteristic features are almost completely lost. The scattering intensity decreased dramatically due to the loss of water. No Bragg peak could be discerned between 0.05 and 0.1 nm⁻¹; instead, a broad shoulder emerged between 0.10 and 0.15 nm⁻¹, which can be assigned to a broader range of interthylakoidal RDs.

The SAXS and SANS data fit well with cryo-TEM of hydrated and desiccated L. ohadii (Fig. S5; the complete analysis is in ref. 7). In the desiccated state, TEM images indicated an ∼50% decrease in luminal space, a value that is in line with the reduced RD observed in SAXS. This shrinkage was suggested to contribute to the PSI-based quenching mechanism (7, 33). The TEM images also indicated a loss of the organized thylakoid membrane space and of the apparent rod structures in the interthylakoidal space, which is in line with the SANS observations. Taken together, cryo-TEM, SANS, and SAXS support a change in the organization of the interthylakoidal space and in the aggregation state of the PBSs that populate it.

Discussion

In the desiccated state we observed (i) reduced EET between PBS components, (ii) extensive heat dissipation, and (iii) fluorescence red shift (Figs. 2 and 3 and Figs. S1–S3). These changes coincide with a loss of the organized PBS structure in the interthylakoidal space (Fig. 4 and Figs. S4 and S5).

How could a change in the aggregation state of PBSs support a transition between highly efficient EET and quenching? We can draw corollaries from in vitro studies in which we controlled the aggregation state of PC and APC artificially (34). Aggregation of PC in solution leads to shortening of their lifetime, red shift, and a lower fluorescence yield. This was a reversible process that occurred in the absence of any additional auxiliary quenching proteins, such as OCP. The effect of aggregation on APC was much smaller.

It is important to note that aggregates discussed here are not chromophore aggregates but rather aggregates of chromophore-containing proteins. Unlike artificial cyanine and porphyrin aggregates, phycobilisomes represent a different scenario. Pigment distances within PC and APC subunits and between them dictate intermediate coupling strength (35, 36).

The underlying mechanism proposed is based on the intermediate-length (1–2 nm) coupling interactions between PC units through the external β155 pigments. A random network of intermediate coupling through the PC aggregates will generate long and convoluted EET paths and a band gap structure with many localized states that couple well to the wide frequency of environmental thermal noise (37, 38). In an intermediate coupling regime the band structure strongly depends on the level of disorder of the system, the splitting between bands is smaller, and the gap is less protected. Such a structure will support decay to the lowest-energy pigments together with quenching through an increased probability for interactions with the thermal bath. In APC, which lacks the external β155 pigments, the chances for generating a network of intermediately coupled trimers are smaller and so are the effects of its aggregation state on EET.

PC is known to be sensitive to the ionic or dielectric properties of its environment (34). However, changes in these properties during desiccation are not expected to lead to the dramatic fluorescence quenching observed here (Fig. 1). Our data indicate small changes in the aggregation state with no detectable difference in the optical properties of the PC units themselves. We propose an in vivo model where in the hydrated state the organized rod structure supports directional EET to the RCs with minimal losses due to thermal dissipation (Fig. 5). In the desiccated state, this structure is lost, giving way to less organized aggregates. These less organized in vivo aggregates will exhibit EET properties similar to those of in vitro PC aggregates, with better coupling to wide-frequency environmental noise (37).

Fig. 5.

Model for EET processes in the hydrated and desiccated states. A schematic representation of the aggregation state of the PBS antenna in both states. The components are drawn roughly to scale. Green, photosystems; cyan, APC; blue, PC; gray arrows, directional EET; dashed arrows, heat dissipation. In the crowded interthylakoidal space, only small changes in aggregation state are possible; these are exaggerated in the figure for demonstration purposes.

This type of mechanism can also explain the temperature effect. In tightly coupled pigment networks [in Light Harvesting Complex II or FMO, for example, (39–42)] the energetic differences between 300 and 77 K are too small to affect EET. Under an intermediate-coupling regime, however, these temperature differences are expected to have a larger effect (43, 44). A mechanism based on small and localized changes in aggregation of PBS components in the crowded interthylakoidal space may explain the very fast rate of response to rehydration (Fig. 1).

The mechanism suggested here is akin to that proposed by Tamary and coworkers for high light-induced quenching in Synechocystis 6803 where local heat transients generated in the PBSs by nonradiative energy dissipation were suggested to be the source of the nonphotochemical quenching (NPQ) phenomenon (45). Furthermore, the idea that changes in the aggregation state can control the efficiency of EET in the photosynthetic antenna was considered as a mechanism for NPQ in plants. In vitro studies demonstrated that aggregation of light-harvesting complex II (LHCII) antenna complexes leads to efficient energy dissipation (46–51), although the significance of changes in LHCII aggregation state for energy dissipation in vivo is still debated (52, 53). It is important to note a number of key differences between LHCII and PC. (i) LHCII is trapped in the membrane; its range of possible aggregation geometries is limited to the membrane plane. (ii) The pigment-pigment distances within LHCII are, on average, shorter than within PC, supporting a tighter excitonic coupling regime. (iii) LHCII contains carotenoids, dedicated to quenching; these are missing from PC. An aggregation-based mechanism creating a network of intermediately coupled pigments and, henceforth, increased coupling to the thermal noise is much more likely in cyanobacteria than in vascular plant antennae.

Desiccation and rehydration in photosynthetic organisms are complex processes that require protection of many cellular processes, first and foremost, of the photosynthetic apparatus. The rate of recovery from desiccation is highly variable ranging from days in the case of certain vascular plants (54) to a few seconds in the case of desert crust cyanobacteria. So far, we were able to uncover two mechanisms for blocking photosynthetic activity in the desiccated state: constriction of the luminal space leading to a block of plastocyanin diffusion, and accumulation of P₇₀₀⁺ (7) and changes in the aggregation state of PBS components, as discussed here. In both cases the quenching mechanism relies on the rearrangement of existing components of the photosynthetic apparatus. In the mechanism proposed here we also suggest a method for distributing the energy through a PC aggregate. It is the combination of all features of this desiccation mechanism that allows the very rapid response to rehydration.

Evidence for EET processes at the quantum/classical border is emerging from several studies of the photosynthetic apparatus and is highly relevant for new technologies. Here small changes in configuration of an intermediately coupled pigment network support a switch from highly efficient energy transfer to quenching (Fig. 5). In most cases, nanoscale technologies focus on energy transfer devices with tight coupling and low environmental noise (low temperatures) (55). The design principles of the desiccating PBSs suggest an approach that will enable room temperature devices that can be switched to efficient energy transfer. By using these principles it will be possible to develop a coupling controlled nanotoolbox for various future technologies including computers, materials, and sensors.

Materials and Methods

Culture Growth and Preparation for Measurement.

Leptolyngbya ohadii cultures were grown (as described in ref. 7) in YBG11 medium in shaking flasks at 30 °C and 60 μmol photons⋅m⁻²⋅s⁻¹ provided by fluorescent lamps. Cells were homogenized and filtered on 0.45-μm nitrocellulose (NC) filters. For measurement of dry samples, artificial crust filters were left to air dry at room temperature for at least 2 h.

Steady-State Fluorescence Measurements.

Samples were desiccated on the surface of an NC filter. The filter was mounted on a glass slide and fitted into the measurement chamber of a Fluoromax3 spectrofluorometer (Jobin Yvon). Hydration was performed in situ.

Time-Resolved Fluorescence Measurements.

Room temperature time-resolved fluorescence spectra were measured using a picosecond streak camera setup (56) as described in ref. 16. The excitation wavelength was set to 580 nm, to preferentially excite the PBSs. The 77 K measurements were performed as described in ref. 57. Measurements were made on either desiccated or hydrated filters of Leptolyngbya ohadii, and each measurement was repeated two to four times, measuring a different spot or sample each time. The samples were measured on NC filters at both room temperature and 77 K. Global analysis (17) of the data was performed using the TIMP-based Glotaran program (58), to identify the spectral components in the fluorescence signal (SI Text).

Solution Small-Angle X-Ray and Neutron Scattering.

High-resolution SAXS was performed as described in ref. 59. SANS was performed at the KWS-2 instrument operated by Jülich Centre for Neutron Science (JCNS) at the Heinz Maier-Leibnitz Zentrum, Garching, Germany (60, 61) (SI Text). The scattering intensity, I, as a function of the magnitude of the momentum transfer vector, q, was then plotted and analyzed to determine the typical RDs between membranes (62).

Time-Resolved Fluorescence Measurements and Global Analysis

A time window of either 800 or 2,000 ps was used for the time-resolved fluorescence measurements. The repetition rate of the laser was 4 MHz. The intensity of the 580-nm excitation light was ∼100 nW.

The IRF was modeled as a single Gaussian function. The time-resolved spectra were fitted as a superposition of four exponentially decaying components, in the form $S\left(t,\lambda \right)={\displaystyle \sum _{i=1}^n}{c}_i\left(t\right){\epsilon }_i\left(\lambda \right)$. The resulting DAS components are presented after reduction of the contribution of the IRF and time 0 correction. For a sequential model, in which each component decays to the next one, EAS components were obtained (Fig. S1).

Solution Small-Angle Neutron Scattering

A wavelength of 5.15 Å was used as well as a sample to detector of 2, 8, and 20 m to cover a q range from 0.03 to 3.6 nm⁻¹. The samples were held in 2-mm-thick quartz cells. The reduction of raw data was performed by the routine qtiKWS (iffwww.iff.kfa-juelich.de/∼pipich/dokuwiki/doku.php/qtikws) including corrections for detector sensitivity, background noise and empty cell signal. Absolute determination of scattering cross sections I(q) = dΣ/dΩ per unit sample volume in cm⁻¹ was obtained thanks to a calibration with a 1.5-mm-thick polymethylmethacrylate sample. In both cases the resulting 2D scattering patterns were azimuthally integrated (average intensities belonging to identical q values).