Research Article

Structure of a symmetric photosynthetic reaction center–photosystem

See allHide authors and affiliations

Science  08 Sep 2017:
Vol. 357, Issue 6355, pp. 1021-1025
DOI: 10.1126/science.aan5611

A homodimeric complex for anaerobic photosynthesis

In plants, algae, and cyanobacteria, large molecular complexes—photosystems I and II—convert light energy into chemical energy, releasing oxygen as a by-product. This oxygenic photosynthesis is critical for maintaining Earth's atmospheric oxygen. At their cores, photosystems I and II contain a heterodimeric reaction center. Reaction centers evolved in an atmosphere lacking oxygen, and the ancestral complex was likely homodimeric, encoded by a single gene. Gisriel et al. describe the structure of a homodimeric reaction center from an anoxygenic photosynthetic bacterium. The structure shows perfect symmetry of the light-collecting antennae and elucidates the electron transfer chain.

Science, this issue p. 1021


Reaction centers are pigment-protein complexes that drive photosynthesis by converting light into chemical energy. It is believed that they arose once from a homodimeric protein. The symmetry of a homodimer is broken in heterodimeric reaction-center structures, such as those reported previously. The 2.2-angstrom resolution x-ray structure of the homodimeric reaction center–photosystem from the phototroph Heliobacterium modesticaldum exhibits perfect C2 symmetry. The core polypeptide dimer and two small subunits coordinate 54 bacteriochlorophylls and 2 carotenoids that capture and transfer energy to the electron transfer chain at the center, which performs charge separation and consists of 6 (bacterio)chlorophylls and an iron-sulfur cluster; unlike other reaction centers, it lacks a bound quinone. This structure preserves characteristics of the ancestral reaction center, providing insight into the evolution of photosynthesis.

Photosynthesis has changed the atmosphere of our planet and is the single most important energy conversion process driving the biosphere (1). The light reactions of photosynthesis involve the conversion of light energy to chemical energy and are catalyzed by specialized membrane proteins called reaction centers (RCs). Photons are absorbed by antenna molecules, and the excitation energy is transferred to the core of the reaction center, where the excited state initiates primary charge separation, in which an electron is transferred from one bacteriochlorophyll (BChl), chlorophyll (Chl), or pheophytin to another. The electron is then transferred through a chain of cofactors to drive downstream metabolism. RCs are typically classified by their terminal electron acceptor: either a [4Fe-4S] cluster (type I) or a quinone (type II) (2).

Oxygenic photosynthesis in higher plants, green algae, and cyanobacteria makes use of photosystem I (PSI), which is a type I RC, and photosystem II (PSII), which is a type II RC. These work together to transfer electrons from water to ferredoxin. By contrast, anoxygenic phototrophic bacteria, such as Heliobacterium modesticaldum, use a single RC to drive a cyclic electron transfer (ET) pathway that creates a proton-motive force across the membrane, which is used to drive adenosine 5′-triphosphate (ATP) synthesis (3). The discovery of heliobacteria (4) led to the identification of unique characteristics of its RC (5, 6), now visualized in the structure herein. The heliobacterial RC has its antenna domain fused to the ET domain like PSI. For historic reasons, we maintain the term RC for the entire complex and distinguish its two domains: the ET domain and the antenna domain.

Despite the low sequence homology of core RC polypeptides from different organisms, reflecting billions of years of evolutionary divergence, all of these proteins share structural similarities within their central ET domain (7). This region is at the interface of the two core subunits, each of which contributes five transmembrane helices (TMHs) that enclose the ET cofactors like a cage. The type II RC from anoxygenic purple proteobacteria (PbRC) is heterodimeric and relies on peripheral antenna subunits to harvest photons. The core polypeptides of PSI (PsaA and PsaB) have an additional six TMHs at the N-terminal side that form a domain coordinating the bulk of the antenna pigments. The core polypeptides of anoxygenic phototrophs that use a type I RC—heliobacteria, chlorobia, and chloroacidobacteria—have a similar arrangement: an N-terminal antenna domain (six TMHs) and a C-terminal ET domain (five TMHs). However, as the genomes of these organisms encode a single core RC polypeptide, their RCs should be homodimeric (810). A homodimeric RC almost certainly preceded heterodimeric RCs in evolution (11). Duplication of the core RC–subunit gene followed by divergence of the two genes would allow for conversion of a homodimeric RC to a heterodimeric RC. This likely occurred on at least three separate occasions, leading to the creation of PSI, PSII, and the RCs of purple bacteria and Chloroflexi (7). High-resolution structures have been obtained from multiple heterodimeric RCs [PbRC, PSI, and PSII (1214)], but no homodimeric RC structures have been solved until now.

Here we present the first structure of a homodimeric type I RC, an x-ray crystal structure at 2.2 Å resolution (crystallography statistics are presented in table S1). It was purified from H. modesticaldum, a thermophilic anaerobe isolated from volcanic soil in Iceland (15). Heliobacteria are the only phototrophic family in the Firmicutes (i.e., low–guanine-cytosine (GC) Gram-positive bacteria) and the only organisms that use bacteriochlorophyll g (BChl g), an isomer of chlorophyll a (Chl a). Despite the low-light habitat of heliobacteria, peripheral antenna complexes have yet to be identified. The RC used by heliobacteria is the only RC that does not seem to require a tightly bound quinone in ET (16), a characteristic that is supported by the structure presented herein. Most commonly referred to as the heliobacterial RC (HbRC), this complex is the simplest known RC, and it has been proposed to be the closest homolog to the common ancestor of all photosynthetic RCs (17, 18).

Overall structure of the HbRC

The HbRC is elliptical in cross section, with major and minor axes of ~114 and 58 Å, respectively (Fig. 1B). The average thickness of the hydrophobic membrane core is predicted to be ~27 Å [(19); see supplementary materials]. The average height of the HbRC is ~50 Å (Fig. 1A). The complex contains 24 TMHs (Fig. 2A): 22 TMHs from the PshA homodimer and 2 more from the newly identified PshX subunits (see supplementary materials and fig. S1 for details of its discovery). The HbRC binds 54 BChl g, 4 BChl g′, 2 81-hydroxychlorophyll a (81-OH–Chl aF), 2 carotenoids (4,4′-diaponeurosporene), 2 lipids, and 1 [4Fe-4S] cluster; four of these cofactors had not previously been described in the Protein Data Bank (see fig. S2). The PshA and PshX polypeptides account for ~140 kDa of the homodimeric RC. The total molecular mass is ~200 kDa, owing to the high cofactor content.

Fig. 1 Overall structure of the HbRC.

Structure as viewed from (A) the N-side or (B) within the membrane. The two PshA polypeptides are colored in red and pink. PshX subunits are colored in orange. Cofactor molecules are shown as stick models and colored teal (ET), blue (antenna), and lime (carotenoids). The [4Fe-4S] cluster is shown as red (Fe) and yellow (S) spheres. BChl and Chl tails have been truncated for clarity.

Fig. 2 Arrangement of the TMHs and pigments in the HbRC compared with PSI.

(A) TMH arrangement of the HbRC and its superposition with the TMHs of PSI (N-side view). TMHs of the two PshA subunits are colored in red and pink and labeled 1 to 11, from N to C terminus. PshX helices are colored in orange and are labeled “X.” Transparent gray helices are from the heterodimeric core of PSI [Protein Data Bank (PDB) ID 1JB0]. (B) N-side view of the cofactor organization in the HbRC (colored) superimposed with the cofactors associated with the PsaA-PsaB heterodimeric core of PSI (transparent gray, PDB ID 1JB0). ET BChls and Chls are colored red, bulk antenna pigments are colored blue, and the three antenna BChl g that flank the ET chain are colored teal. BChl and Chl tails have been truncated for clarity.

As a consequence of the homodimeric core, the ET cofactors are arranged into two identical branches about the C2 symmetry axis (Fig. 3). The first ET cofactor (P800) is a pair of BChl g′ molecules, the stereoisomer at the 132 position of ring E of BChl g (20), and is located toward the positive side (P-side, equivalent to the outside) of the membrane. On the negative side (N-side, equivalent to the cytoplasm), the segment between TMH8 and 9 of each PshA has two conserved Cys residues that ligate FX, the terminal [4Fe-4S] cluster. Between P800 and FX on each symmetric branch are a BChl g and an 81-OH–Chl aF.

Fig. 3 Cofactor arrangement of the ET chain.

Coordinating residues belonging to one PshA of the homodimer are labeled and colored dark red, and those from the other PshA are colored pink. Cofactor carbons are colored differently to ease viewing: P800 (orange), Acc (green), and A0 (blue). The water molecule serving as an axial ligand to A0 is shown as a red ball, and the unidentified molecule of high electron density ligating Acc is shown as a purple ball. Center-to-center distances are labeled on the right side of the image. BChl and Chl tails have been truncated for clarity.


The antenna BChls form two layers within the RC (dotted gray boxes in fig. S3A). Most are relatively close to one another (i.e., within 6 Å), which would allow rapid excitation-energy transfer. The bulk of this population (blue in Fig. 2B) is ≥14 Å away from the ET chain (red in Fig. 2B). There are only three antenna pigments on each side that are within 13 Å of any BChls and Chl in the ET chain (teal in Fig. 2B). Energy transferred to the ET chain from the antenna pool probably arrives via one of these six BChl g molecules.

Most of the antenna BChls (50 of 54) are coordinated by the two PshA polypeptides, primarily from the six TMHs of the antenna domain (Figs. 1 and 2); the other four BChl g are coordinated by the PshX subunits. Of the 27 BChls and 1 Chl associated with each PshA polypeptide, 19 are coordinated by His side chains, 1 by Asn, 1 by Glu, 1 by Gln, 5 by water molecules, and 1 by an unidentified atom that has higher electron density than a water molecule, hypothetically a chloride ion (purple sphere in Figs. 3 and 4B). Both BChls associated with PshX are His coordinated. Surprisingly, there is a BChl g′ associated with each PshA within the center of an antenna pigment cluster (green in fig. S3B).

Fig. 4 Details of the ET cofactors and comparison with PSI.

Magnified views of the interactions of PshA with the ET cofactors are shown in panels (A) P800 and Acc, (B) Acc and A0, and (C) A0 and FX. Center-to-center distances and ligands are labeled (see also table S2). Coordinating residues belonging to one PshA of the homodimer are labeled and colored dark red, and those from the other PshA are labeled and colored pink. Cofactor carbons are colored to facilitate viewing: P800 (orange), Acc (green), and A0 (blue). The axial ligands to A0 (water) and Acc (unknown) are shown as a red or purple ball, respectively. The center of the [4Fe-4S] cluster is represented by a black sphere. The corresponding cofactors of PSI are shown in transparent gray. BChl, Chl, and quinone tails have been truncated for clarity.

ET chain

A notable feature of the ET chain is that it contains three chemically distinct BChls. It begins on the P-side of the HbRC with P800, a BChl g′ dimer that serves as the primary electron donor, and it ends on the N-side with FX, a [4Fe-4S] cluster that serves as the terminal acceptor (Fig. 3). Each BChl g′ in P800 is coordinated by His537 from TMH10, whereas FX is ligated by Cys432 and Cys441 from the loop between TMH8 and 9 of each PshA. The primary donor (P800) and terminal acceptor (FX) are located on the C2 symmetry axis. There are two pairs of BChls on each side of the symmetry axis that link P800 to FX. The one closer to P800 is a BChl g, whereas the one closer to FX is an 81-OH–Chl aF, in agreement with modeling of the electrochromic bandshift of the 81-OH–Chl aF spectrum in the presence of oxidized P800+ (21).

On the basis of biochemical and biophysical measurements (2123), it was expected that the primary acceptor in the HbRC, called A0 by analogy with the corresponding cofactor in PSI, was an 81-OH–Chl aF molecule. The cofactor between the special pair and primary acceptor is often referred to as the “accessory” BChl and Chl (Acc). We will use that term here, but it should not be taken as a functional designation, as the analogous cofactor has been suggested to serve as the primary electron donor in PSI (24, 25), PSII (26), and the PbRC (2729). Acc is coordinated by a small molecule approximately the size of water, and A0 is coordinated by a water molecule. These are H bonded by the side chains of Gln458 and Ser545, respectively (see discussion later in text). As in the other RCs, within a branch of the ET chain, the special pair (P800) and primary acceptor (A0) are coordinated by one polypeptide, whereas the accessory BChl is coordinated by the other polypeptide of the dimer.

The distances between cofactors (table S2) can be used to explain the differences observed in ET rates between PSI and the HbRC. Pump-probe spectroscopic measurements previously estimated the lifetimes of the first two ET steps (RC*→P800+A0→P800+FX) in the HbRC of H. modesticaldum to be ~25 and ~700 ps, respectively (21, 30), in line with measurements in HbRCs from other species (31, 32). The distance between the special pair and primary acceptor is too far to allow charge separation to occur in a single step on the picosecond timescale, necessitating the involvement of the Acc cofactor, which is <3.5 Å from both P800 and A0. Two different models for charge separation can be proposed: one in which Acc serves as the initial acceptor (HbRC*→P800+Acc→P800+A0→P800+FX) and one in which Acc serves as the initial donor (HbRC*→Acc+A0→P800+A0→P800+FX). Future studies informed by this structure will allow testing of these two models.

The role of the menaquinone (MQ) in the HbRC has been controversial. The HbRC that was crystallized in this study lacked MQ (fig. S4), as it is not bound tightly by the protein and thus is not present in this structure. However, the edge-to-edge distance between A0 and FX is ~10.2 Å (table S2), which is much shorter than the distance between A0 and FX in PSI (~14.3 Å) and close enough to allow a maximal ET rate of ~8 × 108 s−1, which is close to the experimentally determined rate of ~1.4 × 109 s−1 (21). Thus, the structure supports the hypothesis that ET from A0 to FX in the HbRC does not require an intermediate cofactor. The lack of a quinone as an intermediate cofactor in forward ET is a striking divergence from other RCs, and the fact that it is not bound tightly by the HbRC may indicate that it can be used as a mobile electron carrier.

Comparison with other RCs

Because of their descent from a common ancestral protein, all RCs are expected to share the same overall architecture (2), but differences in fine detail may lead to distinctions in function. Overall sequence identity between PshA and the analogous subunits of other RCs is low (<30%), but the structure of the HbRC demonstrates that most of the core structural features are well conserved. Figures S5 and S6 show a comparison of the TMH arrangement of the antenna domain and ET domain, respectively, of PshA with that of PsaA and PsaB from PSI, and the CP43 and CP47 chlorophyll-binding subunits of PSII. The antenna domain of PshA (TMH1 to 6) displays structural similarity to the antenna domains of PsaA and PsaB and to CP43 and CP47 (fig. S5). This is consistent with the phylogenetic analysis of the antenna domain of type I RCs and CP43 and CP47, which indicated that the CP43 and CP47 subunits share a common ancestor with the antenna domain of the type I RC core polypeptide (33). The conserved structural arrangements of the TMHs did not extend to the surface helices and loops, however, which are considerably different between PSI and the HbRC. This may reflect different interactions with electron donors and acceptors, as well as with PSI subunits that have no analog in the HbRC. The structural similarity of the ET domain of PshA was much higher with that of PsaA and PsaB than with the core polypeptides of PSII and the PbRC (fig. S6), consistent with the hypothesis that key differences between the structural arrangements of the TMHs in the cores of type I and II RCs were established before the divergence of PshA from the ancestor of PsaA and PsaB.

Of the 58 BChls and 2 Chls in the HbRC, 42 of them are in similar positions in PSI (green in fig. S3C). The most peripheral antenna pigments from PSI are lacking in the HbRC, probably because PSI uses extra subunits that stabilize their binding. The most notable differences are in the BChls that link the bulk of the antenna to the ET chain. Three antenna BChl g are potentially creating this linkage (teal in Fig. 2B and fig. S3A): (i) the one closest to P800 is not found in PSI, (ii) the one closest to Acc is in the same position as an analogous Chl a in PSI, and (iii) the one closest to A0 is shifted toward the P-side by ~5.8 Å relative to a Chl a found in a similar orientation in PSI. This makes sense, as the major bridging pigment in PSI is closest to A0, which is a Chl a, like all the Chls of the antenna and ET domains. In the HbRC, energy transfer from an antenna BChl g (absorbing at 780 to 800 nm) to A0 (81-OH–Chl aF absorbing at 670 nm) would be energetically unfavorable, implying that efficient energy transfer from the antenna to the ET domain should proceed via the Acc or P800 cofactors.

The superposition of the HbRC and PSI ET cofactors are displayed in fig. S7, and the distances between the HbRC ET chain cofactors are listed in table S2. The distance between Chls or BChls in the special pair is closer than in other RCs, with a center-to-center distance of only 5.7 Å in the HbRC, compared to 6.3 Å in PSI, 8.2 Å in PSII, and 7.9 Å in the PbRC. The larger overlap of the BChl g′ macrocycles in P800 may explain the high charge delocalization across the BChls in P800+ observed by Fourier transform infrared (FTIR) spectroscopy experiments (34). Both BChl g′ of P800 are coordinated by His side chains near the beginning of TMH10 (Figs. 3 and 4A), which are conserved with PSI. As predicted by FTIR spectroscopy of P800/P800+ (34), the δ-N of the His ligand interacts with the Cys601 thiol and the peptide carbonyl of Leu533. In PSI, the special pair (P700) is asymmetric in that (i) the Chl on the PsaA side is Chl a′, whereas the other is a Chl a; and (ii) there is a H-bonding network near the Chl a′ that is absent on the other side (35, 36). Notably, the protein environment of each BChl g′ in P800 is more similar to that of the Chl a on the PsaB side of P700—there is no H-bonding network near the 131-keto oxygen, nor are any of the residues involved in the H-bonding network near the Chl a′ in PSI conserved in PshA. Thus, in the HbRC, the stereochemistry of P800 is not correlated with a H-bonding network, leaving the origin and functional importance of this stereochemistry an open question. However, careful examination of the structure suggests that a BChl g could not be accommodated in this site, and thus, that the alternate stereochemistry may be driven mainly by steric considerations; the same is true of the lone BChl g′ in the antenna domain.

The Chl a in the Acc position is not coordinated by the polypeptide. Instead, it is coordinated by a water molecule that is H bonded to an Asn residue (12). In the HbRC, the BChl g in the Acc position is coordinated by an unknown atom (potentially a chloride ion), which seems to be H bonded to a Gln residue in the same position as the Asn of PSI (Figs. 3 and 4, A and B). Thus, this cofactor seems to be bound by the protein in a somewhat similar fashion in PSI and the HbRC.

In PSI, a Tyr from a membrane-parallel helix donates a H bond to the 131-keto oxygen of A0 (12), which is important for tuning the reduction potential of this cofactor (25, 37). This Tyr is replaced in PshA by Ser553, whose hydroxyl group is positioned to donate a H bond to A0 (Fig. 4C). The shorter distance between the parallel helix and the hydroxyl of the H-bond donor places A0 closer to the edge of the membrane and thus to FX. An Arg (PshA-Arg406 from the other subunit) may also be able to H bond to this oxygen; in addition, the proximity of a positive charge could stabilize the A0 radical anion during charge separation. This may explain the recent conclusion of Ferlez et al. (30) that the reduction potential of the A0/A0 couple is ~150 mV more positive in the HbRC than it is in PSI. It was suggested that this change could also be explained by the replacement of the Met that serves as an unusual axial ligand to the Mg(II) of A0 in PSI by Ser in PshA. The structure reveals that the axial ligand of A0 in the HbRC is actually a water molecule that is coordinated by the side chain of Ser545 (Figs. 3 and 4, B and C). Both of these differences are expected to influence the reduction potential of the A0/A0 couple.

The quinone-binding site on the A branch of PSI is lined by the indole of PsaA-Trp697 on one side of the naphthoquinone headgroup and by the side chains of PsaA-Phe689 and PsaA-Leu722 residues on the other side; the homologous residues are also present in PsaB. Only the Leu is conserved in PshA. In PshA, the Phe and Trp are replaced by Met546 and Arg554, respectively; the latter’s side chain occupies the space where the phylloquinone headgroup is located in PSI (fig. S8A). Thus, we can say with structural evidence that MQ cannot be bound by the HbRC in the same location as phylloquinone is bound by PSI. Adjacent to A0 is an extended blob of unassigned electron density, perhaps representing a molecule with a long isoprenoid tail (fig. S8B). Although it does not resemble a quinone, it may be in the MQ-binding site; if so, the MQ would not be placed to be an intermediate in ET from A0 to FX (fig. S8C). In the HbRC, the shorter distance between A0 and FX would allow forward ET in the absence of a quinone as an intermediate cofactor.

The FX cluster–binding site was the only immediately recognizable motif in the PshA sequence when it was first identified (8). Accordingly, the binding of FX to the two conserved Cys in each such motif is similar to FX binding of PSI (Fig. 4C). However, this cofactor is unusual in that it has a ground spin state of S = 3/2 when the cluster is reduced, in contrast to reduced FX in PSI, which has the usual ground spin state of S = 1/2. The symmetric binding site may relieve strain on the cluster, causing this unusual spin state; other known iron-sulfur clusters with a natural ground spin state of S = 3/2 include the Fe protein of nitrogenase and 2-hydroxyglutaryl–coenzyme A (CoA) dehydratase, both of which are homodimers with the cluster bound at the dimeric interface (38, 39). The observation that the FX cluster is such a symmetric cubane that only half of it is seen in the asymmetric unit is consistent with this hypothesis.

In PSI, the PsaC subunit binds two [4Fe-4S] clusters and reduces soluble electron acceptors; it resembles a bacterial-type dicluster ferredoxin and binds tightly to the PsaA-PsaB core because of additional sequence elements (12). The PshB1 polypeptide, originally considered the analog of PsaC in the HbRC, lacks these elements and is removed from the PshA2 core early during purification (22), consistent with its absence in the HbRC structure. PshB1 and its homolog, PshB2 (40), are now thought to serve as ferredoxin electron acceptors rather than subunits (41). An electrostatic surface analysis of the HbRC reveals that the N-side surface is mostly positively charged, with two positive patches close to FX and an extended Lys on either side (Fig. 5). Homology models generated for PshB1 (see supplementary materials) predict a negatively charged surface that may interact with these patches, facilitating reduction by FX.

Fig. 5 Surface electrostatics models of the HbRC.

Positive and negative charge are colored in blue and red, respectively (see scale at left). N-side (top) and P-side (bottom) surfaces of the HbRC are shown. The profile view (middle) reveals an asymmetric charge distribution creating an electrostatic field that would stabilize the charge-separated state (48); membrane boundaries are indicated by dashed lines. The PshB1 ferredoxin and cyt c553 structures shown to the right are multitemplate models based on sequence identity (see supplementary materials). Lys584 and Lys587 form a positively charged surface patch on each side of FX, whereas the side chains of Lys423 (marked) extend out from the HbRC further than any other residue.

Heliobacteria use cytochrome c553 (cyt c553) as an electron donor to the HbRC. This is similar in structure to cyt c6 (56) but is attached to the membrane through covalent linkage to a diacylglycerol (42, 43), thus restricting its diffusion to two dimensions. The surface closest to P800 is fairly neutral (Fig. 5). There are two surface helices at the P-side of P800 that would serve as the possible cyt-binding site. Although similar to that of cyanobacterial PSI (12), the helix in the HbRC is shorter and has mainly polar uncharged side chains (Ser, Thr, Asn, Gln) pointing away from P800, and the Trp that is important for cyt binding in PSI (44, 45) is absent.

The lower surface charge of the HbRC relative to PSI may provide a rationale for asymmetry in the oxygen-tolerant RCs. The HbRC appears to interact with electron acceptors and donors through hydrophobic and polar uncharged surfaces rather than ionic interactions, which would require forming ion bridges between the symmetric surface of the HbRC and the monomeric—and thus inherently asymmetric—electron-carrier proteins. The evolution of asymmetry may have provided an advantage in binding electron donors and acceptors more tightly, based on specific ionic interactions between complementary surfaces.

The positions occupied by the PshX subunits are similar to the sites where PsaI (near PsaB) and PsaJ (near PsaA) are found in PSI (Fig. 2A). Both of these are single-TMH subunits, like PshX. PshX and PsaJ both coordinate two antenna BChls, but only one of these is in the same location, and even in this case, the axial ligand is not conserved in the sequence between the HbRC and PSI. PsaI does not coordinate any antenna chlorophylls. Further, there is very low sequence identity shared between PshX and either PsaI (9%) or PsaJ (10%). Thus, it seems likely that the appearance of PshX is a case of convergent evolution driven by a functional requirement. As with PsaI and PsaJ, a single carotenoid is wedged between the single-TMH subunit and the periphery of the core dimer (lime in Fig. 1A). Carotenoids are found in PSI between the TMH of all transmembrane peripheral subunits and the TMH of PsaA-PsaB, implying that, in addition to their functional role, they play a structural role; and this may be the case in the HbRC as well. However, there are only two carotenoids in the HbRC, in contrast to the 22 carotenoids in PSI, and they are shorter as well (30 instead of 40 carbons). This likely reflects the environmental niche of heliobacteria, which are strict anaerobes and therefore do not need to prevent the formation of potentially harmful singlet oxygen.

One might expect the HbRC to display characteristics more similar to the common ancestor of all RCs because of its overall simplicity: its homodimeric core, the low number of antenna BChls relative to its oxygenic homolog, and its lack of peripheral antenna complexes or other subunits (other than PshX) (46). The conjecture that a homodimeric RC should have preceded a heterodimeric RC in the evolutionary trajectory is also consistent with this idea. However, the HbRC has had a long time to evolve from the ancestral RC and has certainly acquired unique features that are advantageous to the organism. It has been proposed that the Firmicutes, within which the family Heliobacteriaceae resides, branched early in bacterial evolution (7). Despite this, the ancestor of heliobacteria probably did not originate the first RC but instead acquired it through horizontal gene transfer, consistent with the colocation of the pshA gene in one gene cluster along with all the genes required for pigment synthesis (47). However, some traits of the last common ancestor of all RCs may be preserved in the HbRC as a result of its host’s anoxic niche, which has similarities with early Earth.

Supplementary Materials

Materials and Methods

Figs. S1 to S8

Tables S1 to S4

References (4959)

References and Notes

Acknowledgments: We are grateful to J. Whitelegge for his validation of the PshX sequence by mass spectrometry and to Y. Mazor for modeling and graphical assistance. This work was funded by the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences, of the U.S. Department of Energy (DOE) through grant DE-SC0010575 to K.E.R., R.F., and J.H.G. and supported with x-ray crystallographic equipment and infrastructure provided by P. Fromme of the Biodesign Center for Applied Structural Discovery at Arizona State University. The Berkeley Center for Structural Biology is supported in part by the National Institutes of Health, National Institute of General Medical Sciences, and the Howard Hughes Medical Institute. The Advanced Light Source is a DOE Scientific User Facility supported by the Director, Office of Science, Office of Basic Energy Sciences and operated for the DOE Office of Science by Lawrence Berkeley National Laboratory. Results shown in this report are derived from work performed at Argonne National Laboratory, Structural Biology Center at the Advanced Photon Source. The Structural Biology Center is funded by the DOE Office of Science, Office of Biological and Environmental Research. Argonne is operated by UChicago Argonne, LLC, for the DOE Office of Science under contract DE-AC02-06CH11357. K.E.R., J.H.G., and R.F. designed the project. I.S. and C.G. performed purifications, their subsequent optimization, and characterization. C.G., B.F., and R.F. optimized crystallization conditions. C.G. crystallized HbRC that was used in x-ray diffraction experiments. R.F. and C.G. collected x-ray diffraction data. R.F. and C.G. built models for molecular replacement phasing of diffraction data. R.F. and C.G. performed model building and refinement. C.G., K.E.R., and R.F. wrote the manuscript with input from all other authors. The HbRC structure has been deposited into the Protein Data Bank with accession code 5V8K. The authors declare no competing financial interests.

Stay Connected to Science

Navigate This Article