Introduction

Green photosynthetic bacteria exhibit large light-harvesting complexes called chlorosomes, which are ellipsoidal particles with typical dimensions of 150 × 50 × 20 nm. Chlorosomes are composed of self-assembled bacteriochlorophylls (BChl) c, d, or e (Blankenship and Matsuura 2003; Frigaard and Bryant 2006). In addition, chlorosomes contain BChl a, carotenoids, quinones, and lipids. BChl a is located in the crystalline baseplate connecting the chlorosome to the cytoplasmic membrane (via FMO protein layer in green sulfur bacteria). Proteins are located only in the baseplate and an envelope of the chlorosome (Chung and Bryant 1996; Vassilieva et al. 2002). A lamellar model for arrangement of BChl molecules within chlorosomes was proposed on the basis of X-ray scattering and electron cryomicroscopy (cryo-EM) (Psencik et al. 2004). In this model, BChl aggregates form lamellar structures with lamellar spacing between 2 and 4 nm, depending on the main BChl type. Interdigitated esterifying alcohols hold the lamellar layers together and provide a hydrophobic environment for accumulation of carotenoids and quinones (Psencik et al. 2006). In green sulfur bacterium Chlorobaculum (Cba.) tepidum (formerly known as Chlorobium (Chl.) tepidum) the lamellar system often remains parallel to the main axis of the chlorosome, while it exhibits considerable curvature in the two perpendicular directions. This curvature was initially interpreted as being due to general undulations (Psencik et al. 2004). Later Oostergetel et al. (2007) obtained end-on cryo-EM projections of the chlorosomes from Cba. tepidum which enabled direct visualization of the curved lamellar structures, including lamellar tubules as well as undulations. Cryo-EM and nuclear magnetic resonance experiments on chlorosomes from a Cba. tepidum mutant containing single BChl d homolog led recently to a proposal that BChl aggregates form concentric nanotubes with BChls arranged into helical spirals (Ganapathy et al. 2009). Lamellar arrangement was also observed in the brown-colored green sulfur bacteria, belonging to the same Chlorobi phylum as Cba. tepidum, but containing BChl e as the main pigment. In these chlorosomes bulk of the aggregates is composed of smaller lamellar domains, which exhibit random relative orientations, and parallel lamellar system is only found in the proximity of the baseplate (Psencik et al. 2006). Recently, the presence of lamellar aggregates was demonstrated in BChl c containing chlorosomes from Chloroflexus (Cfl.) aurantiacus (Psencik et al. 2009), a member of a different phylum than the previously studied Chlorobi species. This finding suggests that the lamellar organization is universal among all types of chlorosomes and perhaps an intrinsic property of the chlorosomal bacteriochlorophylls, although the higher-order structure may differ between species and reflect subtle variations in the composition. In addition, comparison of chlorosomes from different species suggests that the lamellar spacing is approximately proportional to the chain length of the esterifying alcohols. Larger spacing also correlates with an increasing number of lipophilic components (i.e., carotenoids and quinones) that partition with the esterifying alcohols into the apolar space of the lamellar system (Psencik et al. 2006).

Self-assembling aggregates with similar properties to those in chlorosome can also be prepared in vitro (for a review see Balaban et al. 2005; Miyatake and Tamiaki 2005). Studies proved that an addition of non-polar compound (lipid, carotenoid, quinone) is necessary for aggregation of BChl c from green sulfur bacteria in aqueous solutions (Hirota et al. 1992; Klinger et al. 2004; Alster et al. 2008). The hydrophobic effect stabilizes the interactions between esterifying alcohols and other lipophilic components and constitutes the driving force for self-assembly under these conditions. Aggregates can also be prepared in non-polar organic solvents such as hexane (Smith et al. 1983). In this case the assembly is most likely driven by the hydrophilic interaction between the polar groups on chlorin rings (Zupcanova et al. 2008). BChl c aggregates from hexane exhibit very similar nuclear magnetic resonance pattern as chlorosomes (van Rossum et al. 1998) suggesting that the short-range arrangement is similar in both cases; however, it is not clear whether also the long-range arrangement (i.e., lamellar organization) is the same as in chlorosomes. This question is addressed in this study, showing that the hexane-induced BChl c aggregates exhibit a structure remarkably similar to that found in chlorosomes. The main objective of this study, however, was to further explore the structural role of the esterifying alcohol in the lamellar arrangement. Using bacteriochlorophyllides exhibiting increasing length of the esterifying alcohol we show that the lamellar spacing is proportional to this length.

Materials and methods

Cells of Cba. tepidum were grown in modified Pfenning’s medium (Wahlund et al. 1991) for 3 days at 48°C under constant illumination from a 60-W tungsten lamp. Pigments were extracted from whole cells using a mixture of acetone and methanol (7:2, vol/vol). BChl c was isolated from the extract by means of reverse-phase HPLC as described in (Klinger et al. 2004). The four main C8 and C12 homologs of BChl c esterified with farnesyl at C173 (Fig. 1) were combined. 8-ethyl-12-ethyl BChl c ([E,E]BChl c F) and 8-propyl-12-ethyl BChl c ([P,E]BChl c F) were the two most abundant homologs used in this study, accounting for 60 and 30% of the total content, respectively. Pigments were dried and stored at −20°C. BChl c aggregates in hexane were prepared by dissolving purified BChl c in dichloromethane (concentration ~1 mM). Aggregates were formed immediately upon addition of at least 10-fold excess of neat hexane followed by vigorous mixing. The resulting aggregates were sedimented and slowly dried with nitrogen at room temperature to prepare sample for powder X-ray scattering. Bacteriochlorophyllide c (BChlide c) derivatives esterified with methyl, butyl, and octyl at the C173 position (Fig. 1) were prepared from BChl c by transesterification under KOH alkaline conditions as described earlier (Zupcanova et al. 2008). Aggregates from all transesterified BChlides c in hexane were prepared in the same way as hexane-induced aggregates from BChl c.

Fig. 1
figure 1

Molecular structure of BChlide c. R stands for the alkyl group of the esterifying alcohol. In this study, BChlides esterified with methyl, butyl, and octyl were used and compared with BChl c esterified with farnesyl. Substituents R8 can be mainly ethyl, propyl, or isobutyl, R12 can be methyl or ethyl

Full size image

The small and medium angle X-ray scattering (SAXS and MAXS) measurements were performed as described earlier (Psencik et al. 2004) using a setup comprising sealed Cu fine-focus X-ray tube, Montel Optics by Incoatec GmbH and HI-STAR 2-D multi-wire proportional counter as a detector. Sample to detector distances were 50 and 18 cm, respectively. The wide-angle X-ray scattering (WAXS) was measured with an in-house X-ray system utilizing Rigaku rotating anode source (0.3 mm × 0.3 mm Cu target), focusing optics based on a curved Cu-plated mirror and a curved asymmetric Si(111) monochromator crystal and MAR345 image plate detector placed on the focal point. The dried samples were cleaved into thin flakes which were subsequently mounted and exposed to X-rays along the cleavage plane. Scattering anisotropy was measured by directing the incident X-ray beam along the surface of the dried thin flakes (ca. 1 × 2 mm). The anisotropic scattering patterns were integrated sector-by-sector to preserve the orientation information as shown in Fig. 3.

Results

BChl c aggregates in hexane

Figure 2 compares X-ray scattering measurements on BChl c aggregates from hexane with scattering curves of BChl c containing chlorosomes from Cba. tepidum. The scattering curves from two different chlorosome preparations illustrate the natural variability of the scattering pattern. The scattering of BChl c aggregates from hexane is very similar to the scattering from chlorosomes. Each of the three peaks (~0.3, 0.52, and 0.66 1/Å) observed for the dried aggregates have a corresponding peak in the chlorosomal diffraction pattern. The agreement between their relative intensities and widths is also excellent. This result thus confirms that the three peaks at q values of~0.3, 0.52, and 0.66 1/Å observed in the scattering from native chlorosomes originate solely from BChl c aggregates, as previously proposed.

Fig. 2
figure 2

X-ray scattering curve from dried BChl c aggregate (solid line) compared with scattering curves from two different Cba. tepidum chlorosome samples (dotted line, data from (Psencik et al. 2004) and dashed line, data from (Ikonen et al. 2007)) to illustrate the natural variability of the scattering patterns

Full size image

In native chlorosomes the q~0.3 1/Å peak was assigned to the lamellar spacing on the basis of cryo-EM and other evidence (Psencik et al. 2004). The X-ray data presented here show that the hexane-induced aggregates are also organized into lamellar structures. The lamellar spacing of the hexane-induced aggregate (20.7 Å) is somewhat smaller than that observed for wild-type Cba. tepidum chlorosomes in Fig. 2 (~21–22 Å). This most likely reflects changes to lamellar packing due to carotenoids, which are absent from the hexane-induced aggregates (see “Discussion” section and Psencik et al. (2006)).

While the position of the lamellar peak is somewhat variable, the positions of the two remaining peaks, i.e., at 0.52–0.54 1/Å and 0.66–0.68 1/Å, respectively, exhibit much less variability in chlorosomes and also in aggregates. We propose that these two peaks reflect the arrangement of the BChl c within individual lamellar layers and reflect distances within and between pigment stacks. Although it is not possible to construct a unique lattice from this data alone the observed Bragg distances are compatible with a dimer in the asymmetric unit as previously proposed (Psencik et al. 2004).

Scattering from native chlorosomes exhibits two additional peaks when compared to that from the hexane-induced aggregate. These two peaks are less intense, but significantly sharper compared to the peaks attributed to BChl c. In addition, their position at 0.195 1/Å (32.2 Å spacing) and 0.76 1/Å (spacing 8.2 Å) does not vary between different wild-type chlorosome preparations. These peaks were affected for carotenoids deficient Cba. tepidum mutants that exhibit defects in chlorosome morphogenesis, and it was argued that these changes reflect structural rearrangements of the baseplate (Ikonen et al. 2007). In addition, both peaks disappeared during treatment of the chlorosomes with hexanol, and their disappearance correlated well with an overall change of the chlorosome shape from ellipsoidal to a round one, which was attributed to the destruction of the baseplate (Arellano et al. 2008). Based on such indirect evidence, these two peaks were suggested to arise from the chlorosome crystalline baseplate. Absence of the two peaks from the scattering pattern of hexane-induced aggregates is consistent with this interpretation.

It is interesting to note that X-ray scattering patterns of both chlorosomes and BChl c aggregates from hexane exhibit a significant and similar anisotropy. Figure 3 shows the orientation of the four X-ray scattering reflections measured from a thin flake of a dried chlorosome sample. Two reflections have dominant contribution in the direction normal to the plane of the flake while the other two are parallel. The lamellar peak reflection (peak I in the Fig. 3) contributes to scattering in the normal direction reflecting the typical lamellar morphology. The ~0.66 1/Å peak (peak III) exhibits anisotropy similar to the lamellar peak despite not being directly related (e.g., second order). Assuming that the chlorosomes are oriented within the flake with their long axis parallel to the surface the data suggest that the lattice vector c (i.e., the lamellar spacing; lattice vectors defined in (Psencik et al. 2004), see also Fig. 3 here) is oriented perpendicular to the long axis of the chlorosome while b (peak II) and the 0.76 1/Å baseplate spacing (peak IV) are oriented along the long axis of the chlorosome. Lattice vector a (peak III) has the dominant contribution in the direction perpendicular to the long axis of the chlorosome. Unfortunately, the 0.195 1/Å baseplate peak was not well resolved in the anisotropic measurements, presumably because of its position on the sharply decaying baseline. After subtraction of the baseline (not shown) the peak was resolved in the parallel scattering indicating that the main baseplate spacing (32.2 Å) is oriented along the long axis of the chlorosome, in agreement with the recent direct cryo-EM observation for Cfl. aurantiacus (Psencik et al. 2009).

Fig. 3
figure 3

Anisotropic X-ray scattering curves from a dried BChl c aggregate (solid line) and Cba. tepidum chlorosome (dotted line). MAXS intensities in the direction normal (red line) and parallel (blue line) to the flake, integrated over 60° sectors in the detector plane (angle ϕ). The inset at the top illustrates the sample-detector configuration used for data acquisition. The inset below shows the orientation of the unit cell with respect to the long axis of the chlorosome or the aggregate flake, respectively. The lattice constant c (lamellar spacing, see also Fig. 6) is perpendicular to the long axis

Full size image

BChlide c aggregates in hexane

Given the proposed aggregate structure, with the esterifying alcohols filling the space between lamellar layers, the lamellar spacing shall be proportional to the length of the esterifying alcohol (Fig. 6). In order to examine this relationship we prepared a series of BChlides c differing in the length of the esterifying alcohol (methyl, butyl, and octyl). All three compounds formed aggregates in hexane which in turn exhibited optical spectra almost identical to those obtained for aggregates from BChl c esterified by farnesyl, as shown previously (Zupcanova et al. 2008). Figure 4 compares X-ray scattering patterns of the BChlide aggregates with that of BChl c esterified with farnesyl. Methyl-BChlide c diffraction is dominated by two peaks at 0.49 and 0.57 1/Å. Together with two weaker peaks at 0.36 and 0.83 1/Å the scattering pattern does not resemble that of BChl c aggregates. This compound may adopt a polycrystalline form distinct from the lamellar aggregates. The two lower angle peaks were found at nearly identical positions (at 0.35 and 0.48 1/Å, respectively) in previous study of methyl-BChlide c aggregates (note that the higher angle peaks were outside the q range used in the study, q < 0.55 1/Å) (Umetsu et al. 1999). In contrast to methyl-BChlide c, both butyl and octyl derivatives yield patterns similar to that of BChl c with the exception of the lamellar peak which progressively shifted from ~0.3 1/Å for farnesyl to 0.36 1/Å (lamellar spacing of 17.5 Å) for octyl, and finally to 0.44 1/Å (lamellar spacing of 14.3 Å) for the butyl substitution. Other peaks reflecting the order within the lamellar layers remained at the same positions, indicating that pigment arrangement within stacks was conserved. This is consistent with their chlorosome-like visible spectra. Consequently, lamellar spacing does not influence the short-range order of assembled pigments and their spectral properties. The wide-angle patterns (q > 1 1/Å) were very similar for all pigments, further confirming their similar short-range packing.

Fig. 4
figure 4

X-ray scattering curves of aggregates prepared in hexane from BChl c esterified with farnesyl and several BChlides c. Data for BChl c are combined from MAXS and WAXS measurements; data from BChlides are combined form SAXS and WAXS measurements. The several sharp peaks observed in the wide-angle region are most likely from a residual dried potassium hydroxide, which was used during the BChlides preparation

Full size image

Discussion

Comparison of X-ray scattering profiles obtained from chlorosomes of Cba. tepidum and BChl c aggregates from hexane confirms that the peaks at q values of ~0.3, 0.52, and 0.66 1/Å originate from BChl c aggregates. The data also demonstrate that the same lamellar structures are formed, including the extent of disorder, no matter whether the aggregates are assembled in cells into native chlorosomes or in vitro in a non-polar organic solvent. Thus, the self-organization is remarkably robust and encoded in the chemical structure of the pigment alone. Lamellar structures are formed also by aggregates of butyl-BChlide c and octyl-BChlide c but not by methyl-BChlide c. The fact that the X-ray scattering from methyl-BChlide c aggregates differs significantly from the other derivatives is consistent with an important role of the esterifying alcohol length in the formation of lamellar structures (Zupcanova et al. 2008).

In this study, a natural mixture of BChl c homologs as found in native chlorosomes was used for hexane-induced assembly. Similar lamellar structures were obtained for synthetic BChl c analogs, and it was noted that racemic mixture supported assembly more readily than isolated epimers alone (Balaban 2005). A similar conclusion has been drawn also for BChls in native chlorosomes (Steensgaard et al. 2000). This may explain why no X-ray scattering from a hexane-induced BChl c aggregate had been observed when a single BChl c homolog R[E,E]BChl c F was used (Umetsu et al. 1999). In the same study (Umetsu et al. 1999) X-ray scattering from methyl-BChlide c, which was a mixture of all naturally occurring homologs, yielded very similar results to those presented here. On the other hand, similar X-ray scattering patterns were obtained for dichloromethane-induced aggregates of either a natural mixture of BChl c homologs used in this study (unpublished data) or a single R[E,E]BChl c F homolog (Umetsu et al. 1999). The presence of well-ordered aggregates in the mutant producing a single BChl d homolog (Oostergetel et al. 2007; Ganapathy et al. 2009) also suggests that the roles of various homologs and epimers in aggregation are far from being understood.

The 0.76 1/Å diffraction was originally assigned to a 110 reflection from the proposed monoclinic lattice of BChl c aggregates (Psencik et al. 2004). The absence of this and another chlorosome peak (q = 0.195 1/Å) from the diffraction of the hexane-induced aggregate demonstrates that these two peaks must originate from a different periodic structure within the chlorosome, most probably the crystalline baseplate. This interpretation is supported by the fact that these peaks are diminished in intensity when the integrity of the baseplate is affected, e.g., as a consequence of mutagenesis (Ikonen et al. 2007) or hexanol treatment (Arellano et al. 2008). In addition, these two peaks are significantly sharper than those originating from BChl aggregates. Finally, the Bragg spacing corresponding to 0.195 1/Å peak (32.2 Å) and its orientation along the long axis of the chlorosome fits very well with the baseplate spacing (33 Å) and orientation observed in chlorosomes from Cfl. aurantiacus by cryo-EM (Psencik et al. 2009). The 0.76 1/Å peak (8.2 Å spacing) most likely reflects packing order within the baseplate, perhaps spacing between α-helices within a dimer of the CsmA protein (Pedersen et al. 2008; Psencik et al. 2009).

The lamellar spacing, observed here for the BChl c derivatives, when plotted as a function of the length of the esterifying alcohol, increases approximately linearly (Fig. 5). Thus, the lamellar spacing is proportional to the length of the esterifying alcohol, as illustrated schematically in Fig. 6. Absorption spectra of aggregates from all studied BChlides and BChl c in hexane are very similar, including position of the Qy band absorption maxima at ~745 nm (Zupcanova et al. 2008). The Qy band of monomeric BChl c exhibits a peak at around 670 nm, and the red shift of the Qy band in aggregates, observed for both the hexane-induced aggregates and native chlorosomes, is caused by excitonic interactions between the transition dipole moments of the monomers. The data presented here demonstrate that aggregates exhibiting the same red shift differ in their lamellar spacing. This means that only interactions within the same lamellar layer contribute significantly to the excitonic coupling, even for a thickness of ~1.5 nm of butyl-BChlide (absorption spectra of methyl-BChlide are also very similar, but for this compound we do not have a clear evidence that it forms lamellar aggregates). As a consequence, only a single pigment layer should be sufficient for excitonic calculations of the spectral properties of BChl aggregates in chlorosomes. It is interesting to note that the excitonic interactions within lamellar layers can be reversibly abolished by treatment with hexanol without disrupting the lamellar order, at least in native chlorosomes (Arellano et al. 2008).

Fig. 5
figure 5

Scaling of lamellar spacing with the chain length. Spacings corresponding to the two main peaks observed for methyl-BChlide c are shown (black). The data points obtained for hexane-induced aggregates of other BChlides and farnesyl-BChl c (red) were used to produce the linear fit (dotted). Blue data points correspond to the lamellar spacing in native chlorosomes (data from Psencik et al. 2004, 2006, 2009). The green arrow depicts the change in lamellar spacing upon carotenoids and quinones removal by a hexane wash (Psencik et al. 2006). The lengths of the esterifying alcohols are based on the HPLC analysis of the same chlorosomes from green sulfur bacteria (Psencik et al. 2006) and Cfl. aurantiacus (Psencik et al. 2009), which were used in the X-ray experiments. Average lengths of the carbon skeletons of alcohols were calculated ignoring their methyl substituents

Full size image
Fig. 6
figure 6

Scheme illustrating the proposed scaling of lamellar spacing with the esterifying alcohol tail length. Polar chlorin rings of BChl molecules are depicted as green plates and the apolar alcohol tails as sticks. In this scheme the lamellar curvature is not apparent

Full size image

In addition to data obtained for aggregates in vitro, Fig. 5 also illustrates how the lamellar spacing varies upon change of the esterifying alcohol length and carotenoid content in chlorosomes from different species. The data for chlorosomes deviate from the extrapolated linear dependence, being systematically larger than the predicted value. However, the data presented in Fig. 5 also suggest that this deviation is caused mainly by the additional amount of carotenoids, and probably also quinones, which are not present in the in vitro aggregates. Both carotenoids and quinones were shown to co-habit the hydrophobic space between the lamellar layers with esterifying alcohols. When majority of these hydrophobic molecules are washed out from the chlorosome by hexane treatment the spacing decreases to a value close to the extrapolated linear relationship. In Fig. 5, this is illustrated by a spacing decreased from 27.3 to 22.4 Å observed for chlorosomes from Chl. phaeovibrioides (Psencik et al. 2006). Similarly, spacings close to the predicted value were observed for Cba. tepidum chlorosomes, which harbor only small amount of carotenoids.

While the spacing clearly increases with the amount of carotenoids, no quantitative relationship can be derived based on present data. This is because the dependence is complicated by the presence of quinones which will certainly contribute to the lamellar spacing. In addition, a certain carotenoid fraction is present in the baseplate, and thus does not contribute to the spacing. Finally, chlorosomes from different species contain different carotenoids, which may occupy more or less lamellar space due to their various sizes, conformations, and possibly also have distinct interactions with BChls. This is illustrated by structural characterization of Cba. tepidum mutants that accumulate different carotenoid variants (Ikonen et al. 2007). The largest spacing (23.7 Å) was observed for a mutant harboring phytoene in the 15-cis conformation, a carotenoid analog that is likely to occupy larger excluded volume within the lamellae than the carotenoids normally present in the wild-type chlorosome. In general, the lamellar spacing is primarily determined by the length of esterifying alcohols and may be further modulated by the excluded volume of carotenoids and quinones.

Now, a question may arise whether the observed correlations between the average length of the esterifying alcohols and the lamellar spacing have any physiological significance. The smallest lamellar spacing (~2 nm) and carotenoid content (carotenoid/BChl molar ratio of ~0.1:1) was found in Cba. tepidum, which lives anaerobically under moderate light conditions (Wahlund et al. 1991; Castenholz and Pierson 1995) and therefore can survive without extra light-harvesting capability or protection by carotenoids. On the other hand, BChl e containing brown-colored species of green sulfur bacteria often live in deeper water layers under very low light conditions. Due to the light scattering and absorption of the water column and phytoplankton, mainly light in the spectral region between 500 and 600 nm reaches their habitat (Vila and Abella 1994). Bacteria adapt to these conditions by an unusual splitting of the Soret absorption band of the aggregated BChl e (Cox et al. 1998; Arellano et al. 2000) and by formation of small domains of lamellar aggregates with different orientations, presumably to capture photons with different polarizations (Psencik et al. 2006). In addition, chlorosomes of brown-colored bacteria contain large amounts of carotenoids, which are able to increase absorption in a broad range around 500 nm. Brown-colored species exhibit on the average longer esterifying alcohols and consequently larger lamellar spacing compared to Cba. tepidum. This enables them to accommodate the additional carotenoids. The important role of carotenoids for light-harvesting was demonstrated by observation of efficient (60–70%) and fast energy transfer (transfer time faster than 100 fs) from the short-lived carotenoid S2 state to BChls in chlorosomes from brown-colored green sulfur bacterium Chl. phaeobacteroides (Psencik et al. 2002).

In contrast to green sulfur bacteria, Cfl. aurantiacus is a facultative aerobic bacterium and therefore protection of photosynthetic apparatus against oxidative damage is required. Carotenoids are known, apart from their light harvesting function, to provide photo-protection to the photosynthetic apparatus. Indeed, a considerably enhanced carotenoid biosynthesis was observed for Cfl. aurantiacus under intense irradiation (Schmidt et al. 1980; Larsen et al. 1994; Ma et al. 1996). This suggests that under these conditions carotenoids have a protective rather than light-harvesting function. As a result, the carotenoid/BChl c molar ratio can be as high as 0.45:1 in Cfl. aurantiacus (Schmidt et al. 1980), which is one of the highest ratios among all green bacteria. As for the brown-colored green sulfur bacteria, large lamellar spacing is required in this case to accommodate the additional carotenoids. Therefore, it is not surprising that BChl c in Cfl. aurantiacus is esterified with the longest esterifying alcohols (dominating alcohol is octadecanol) and consequently exhibits the largest lamellar spacing (~3.3 nm) among all green bacteria studied. Thus, the data presented here suggest that the observed correlation between the length of the esterifying alcohol and the lamellar spacing is a mechanism used by various bacteria to accommodate larger amounts of carotenoids either for light harvesting or photo-protection.

Another consequence of the larger lamellar spacing is an increased observed disorder within the BChl aggregate. This additional disorder is probably the main reason why the two peaks assigned to the lattice within the lamellar layers (0.52 and 0.66 1/Å) were not observed for chlorosomes from BChl e containing green sulfur bacteria and from Cfl. aurantiacus, i.e., the species containing BChls with longer esterifying alcohols.

The three main scattering peaks of BChl c yielded a unit cell for the aggregate arrangement and the corresponding cell dimensions are compatible with a dimer of BChl c molecules in the asymmetric unit (Psencik et al. 2004). Self-assembly experiments with BChl c and BChlide c molecules revealed an important role of the esterifying alcohols, and favor structural models where alcohol tails extend to both sides of the lamellar layer (Klinger et al. 2004; Zupcanova et al. 2008). Such an arrangement facilitates the hydrophobic interaction between esterifying alcohols from the neighboring layers and provides plausible explanation for the formation of lamellae. Two previously proposed models fulfill these two requirements. One of them invokes an antiparallel piggyback dimer as the building block (Nozawa et al. 1994; Egawa et al. 2007) while the other proposes a unit cell containing two parallel monomers with alternating syn and anti configuration (Ganapathy et al. 2009). In that study, syn and anti refers to the orientation of the OH ligation of adjacent BChls with respect to the farnesyl tail. The lattice dimensions of both models are in principle compatible with the X-ray data presented here. A schematic illustration of the arrangement of aggregates shown in Fig. 6 is thus valid for both models. The anisotropic X-ray data allowed to determine the preferred orientation of the ordered lamellar system and its unit cell within the chlorosome. As expected, the lamellar spacing (lattice vector c) is oriented normal to the long axis of the chlorosome, lattice vector a nearly normally. Axis b is parallel to the main axis of the chlorosome. The same conclusion can be drawn also for in vitro aggregates. In this case the role of long axis of the chlorosome is replaced by any axis parallel to the sample flake. The X-ray data are consistent with a presence of curved lamellae and open, partially formed multilayered cylinders as shown by cryo-EM (Oostergetel et al. 2007). The lattice from (Psencik et al. 2004) (a = 9.6 Å, b = 12.0 Å, c = ~20 Å) and its orientation in respect to the long axis of the chlorosome proposed here is similar to the model shown in Fig. 5c of Ganapathy et al. (2009) (note the different lattice constant designations, here we have a < b < c while Ganapathy et al. (2009) uses b = 9.8 Å < a = 12.5 Å).

In summary, the pigment arrangement of BChl c in hexane-induced aggregates is virtually identical to that seen in native chlorosomes, including the lamellar spacing, which scales with the length of esterifying alcohols. The scalable spacing enables native chlorosomes to accommodate variable amounts of carotenoids in response to changing growth conditions. Thus, the BChl molecule is a versatile, self-assembling building block.