The PufX quinone channel enables the light‐harvesting 1 antenna to bind more carotenoids for light collection and photoprotection

Photosynthesis in some phototrophic bacteria requires the PufX component of the reaction centre–light‐harvesting 1–PufX (RC‐LH1‐PufX) complex, which creates a pore for quinone/quinol (Q/QH 2) exchange across the LH1 barrier surrounding the RC. However, photosynthetic bacteria such as Thermochromatium (T.) tepidum do not require PufX because there are fewer carotenoid binding sites, which creates multiple pores in the LH1 ring for Q/QH 2 exchange. We show that an αTrp‐24→Phe alteration of the Rhodobacter (Rba.) sphaeroides LH1 antenna impairs carotenoid binding and allows photosynthetic growth in the absence of PufX. We propose that acquisition of PufX and confining Q/QH 2 traffic to a pore adjacent to the RC QB site is an evolutionary upgrade that allows increased LH1 carotenoid content for enhanced light absorption and photoprotection.

Photosynthesis in some phototrophic bacteria requires the PufX component of the reaction centre-light-harvesting 1-PufX (RC-LH1-PufX) complex, which creates a pore for quinone/quinol (Q/QH 2 ) exchange across the LH1 barrier surrounding the RC. However, photosynthetic bacteria such as Thermochromatium (T.) tepidum do not require PufX because there are fewer carotenoid binding sites, which creates multiple pores in the LH1 ring for Q/QH 2 exchange. We show that an aTrp -24 ?Phe alteration of the Rhodobacter (Rba.) sphaeroides LH1 antenna impairs carotenoid binding and allows photosynthetic growth in the absence of PufX. We propose that acquisition of PufX and confining Q/QH 2 traffic to a pore adjacent to the RC Q B site is an evolutionary upgrade that allows increased LH1 carotenoid content for enhanced light absorption and photoprotection.
Keywords: bacterial photosynthesis; carotenoid; light-harvesting; membrane protein; quinones, reaction centre The reaction centre-light-harvesting 1 (RC-LH1) complexes of purple phototrophic bacteria collect energy from sunlight, and in many bacteria, they also receive energy from LH2 antenna complexes, thereby increasing the light-absorbing capacity of the photosystem. The RC traps this energy by initiating a series of electron transfers that eventually produce a quinol molecule, destined to leave the RC-LH1 complex and diffuse to the cytochrome bc 1 (cytbc 1 ) complex [1]. Here, the original excitation energy is converted to a proton motive force. In some bacteria such as Rhodospirillum (Rsp.) rubrum and Thermochromatium (T.) tepidum, the LH1 antenna ring completely surrounds the reaction centre [2][3][4], but in others, such as Rhodobacter (Rba.) capsulatus, Rba. sphaeroides and Rhodopseudomonas (Rps.) palustris, the LH1 ring is interrupted by a PufX/PufW polypeptide [5][6][7]. The structure of the dimeric Rba. sphaeroides RC-LH1-PufX complex [8] shows that PufX opens a small channel in the LH1 ring that allows quinols/quinones (Q/QH 2 ) to cross the LH1 barrier and gain access to the RC Q B site [8]. This structure accounts for the abolished photosynthetic growth of PufX-minus mutants [9][10][11][12][13][14], as it appears that the space in the LH1 ring occupied by PufX is filled by two extra LH1 subunits that block access to the RC Q B site [15]. However, many bacteria do not have a PufX yet they still photosynthesise; the recent structure of the T. tepidum RC-LH1 complex shows how this is possible, and inspection of the LH1 ring structure revealed a series of small pores that were suggested to allow Q/ QH 2 exchange in the absence of PufX [4]; a similar situation could apply to Rsp. rubrum, as proposed earlier [16].
If, in principle, Q/QH 2 can cross the LH1 barrier directly in T. tepidum and Rsp. rubrum, why do bacteria such as Rba. sphaeroides need to incorporate PufX into their LH1 rings to create a quinone channel? The explanation is proposed to originate in the carotenoid (Crt) contents of bacteria with and without PufX; for example, the T. tepidum and Rsp. rubrum LH1 complexes have a carotenoid: bacteriochlorophyll (BChl) ratio of 1 : 2 [4,17] and the LH1 complex is composed of a 1 b 1 BChl 2 Crt 1 units, whereas this ratio is 1 : 1 for the Rba. sphaeroides LH1 [18] and the LH1 complex is assembled from a 1 b 1 BChl 2 Crt 2 units [8]. We suggest that these 'extra' LH1 carotenoids confer significant advantages in terms of light harvesting and photoprotection and, although they would likely block the quinone pores identified in the 'low-carotenoid' T. tepidum LH1 complex [4], PufX alleviates this potential problem by preventing LH1 from fully encircling the RC. Thus, the pore created by PufX allows Q/QH 2 exchange [8], and the efficiency and robustness of the antenna are increased by increasing the carotenoid content.
In order to test this proposition, we constructed a genomic point mutation that lowers the carotenoid content of LH1, and combined it with genomic deletion of the pufX gene. The PufX À strain DM1R, which harbours the LH1 aW -24 F mutation that lowers LH1 carotenoid content, grows photosynthetically, whereas the PufX À mutant control with normal LH1 carotenoids cannot do so. We suggest that partial loss of carotenoid from LH1 a 1 b 1 BChl 2 Crt units encircling the RC in the DM1R mutant opens some pores in LH1 for quinone exchange, mimicking the 'low-carotenoid' LH1 complexes of Rsp. rubrum and T. tepidum to a limited extent, and enabling this strain to tolerate the loss of PufX.

Rba. sphaeroides strains
Genomic mutations were constructed in various combinations in the Rba. sphaeroides 2.4.1 WT using the suicide vector pk18mobsacB [19] as described [20,21]. These mutations were as follows: (a) deletion of the pucBA1 and pucBA2 genes encoding polypeptides of the LH2 complex, (b) deletion of the pufX gene and (c) site-specific mutation of pufA to convert aTrp -24 ?Phe (the numbering of LH1 residues relates to the conserved His 0 residue that binds bacteriochlorophyll). Separately, we constructed a control strain DX13 harbouring a R 49 LR 53 L variant of PufX, which yields monomeric PufX-containing RC-LH1 complexes (Qian et al., in preparation; Table 1).

Rba. sphaeroides strains
The cells were grown in M22+ medium supplemented with vitamins and 0.1% casamino acids. The cells were grown either semi-aerobically in the dark, as 1.5 L cultures in 2-L conical flasks in a shaking incubator at 34°C, or photosynthetically at room temperature in stirred 1-L Roux bottles at~50 lE light intensity.
The photosynthetic growth curves were recorded for cultures in 15-mL flat-bottomed screw cap tubes sealed with Parafilm and filled to capacity, and with a micro stir bar. The cultures were illuminated by halogen light bulbs to give a minimum light intensity of 50 lE. Growth was monitored with a WPA Colourwave CO7000 Medical Colorimeter equipped with a 680-nm filter. Four separate cultures of each strain were grown simultaneously and their optical density values averaged for each time point. Each inoculum, using a semi-aerobically grown culture of each strain, was adjusted so that each culture started with the same number of cells.

Pigment extraction
Cells from 1 mL of an 80 mL semi-aerobic culture of the relevant strains were pelleted in a microcentrifuge and extracted twice into 0.5 mL of acetone : methanol (7 : 2 v/v). Absorption spectra of pooled 1 mL samples of extracts were recorded using a Cary 60 spectrometer. The peak heights at 478 nm (carotenoids) and 770 nm (BChls) were measured; three replicates of three separate semi-aerobic cultures were measured to obtain the final averaged data. All procedures were conducted under dim light conditions.

Membrane preparation and low-temperature spectroscopy
Membranes were prepared as described in Olsen et al. [22] and kept at À20°C prior to use. Low-temperature (80K) absorbance spectra were measured using a DN10 Cryostat (Oxford Instruments, Abingdon, UK) mounted in a Cary 60 spectrophotometer (Agilent, Stockport, UK).

Results
Photosynthetic growth rates of DM1R, and PufX + and PufX À control strains In order to lower the carotenoid content of the Rba. sphaeroides LH1 complex, thereby producing at least a few of the pores observed in the low-carotenoid T. tepidum RC-LH1 complex [4], the LH2 À , PufX À strain DM1R was constructed in which aTrp -24 at the N terminus of the LH1a polypeptide was changed to Phe (LH1aW -24 F). This mutation is similar to the aTrp -24 to Tyr alteration in Rba. capsulatus that lowered the carotenoid content of LH1 by 60% [23]. In order to simplify spectral analyses, all strains harboured genomic deletions of pucBA1, pucBA2 that remove LH2 complexes. DM1R was tested for photosynthetic growth against the two LH2 À control strains: DO10, which is the same as DM1R, that is PufX À but with normal LH1 complexes, and DX13, which is PufX + but with monomeric RC-LH1-PufX core complexes to match the monomeric RC-LH1 core complexes in DM1R and DO10. Absorption of whole cells (not shown) showed that DM1R has 61% of the LH1 content of DO10, which likely arises from some impairment of LH1 assembly by the LH1aW -24 F alteration. DM1R grew more slowly than the positive control PufX + strain DX13, likely because DX13 has an optimised quinone channel created by PufX, but significantly faster than the PufX À LH2 À control strain DO10, which was photosynthetically inactive as expected (Fig. 1A). Equally inactive is the RC-only (LH2-minus, LH1minus) strain RCO1 [24], which shows no photosynthetic growth within the same 24-h period used in Fig. 1, and at the same 50 lE light intensity (results not shown).

Biochemical and spectroscopic analyses of the DM1R and DO10 strains
The ability of DM1R to grow photosynthetically could be explained if the LH1aW -24 F alteration destabilises the formation of a complete LH1 ring; then, missing LH1 a 1 b 1 units could create gaps for Q/QH 2 exchange. In order to investigate this point, we performed biochemical and spectroscopic analyses of the PufX À DM1R and DO10 strains. The low-temperature (77K) absorption spectra in Fig. 2A were normalised for LH1 absorption, to emphasise the equivalence of the core complexes in terms of LH1 : RC stoichiometry (see below), and to reflect the presence of some RCs with no LH1, which appear in fractions B1 and B2 (Fig. 2B,C).
To analyse differences between DM1R and DO10 strains in more detail, comparable amounts of DM1R and the DO10 control membranes, in terms of RC content, were solubilised with b-dodecylmaltoglucoside (b-DDM) and then fractionated by centrifugation through sucrose density step gradients. Figure 2B shows that b-DDM releases some 'free' RCs from DM1R membranes, as shown by the distinctive absorption spectrum of these RC complexes (Fig. 2C, B1,B2, blue). The equivalent fractions from the DO10 control, barely visible in the images of the gradients ( Fig. 2B), have a similar absorption spectrum (Fig. 2C,B1,B2, red). These spectra show that the limited impairment of LH1 assembly by the LH1aW -24 F alteration in DM1R has left some RCs with no encircling LH1 antenna. Another likely consequence of lessened LH1 assembly is seen in the B1 fractions; some RCs are present in each case, but relative to DO10 the B1 fraction in DM1R has more unattached carotenoid and also a small absorption peak at~680 nm (Fig. 2C, arrowed), possibly from an intermediate of BChl biosynthesis.
If the LH1 ring was interrupted in DM1R, due to incomplete assembly of LH1 a 1 b 1 subunits round the RC, there would be space for Q/QH 2 exchange despite the loss of PufX and this would explain the photosynthetic growth of DM1R in Fig. 1A. However, the cores that do assemble in DM1R have a normal LH1 : RC stoichiometry, as shown by the close correspondence between the overlaid spectra of the B3 fractions for DM1R and the DO10 positive control (Fig. 2C, bottom). These two spectra have almost identical 874 : 805 nm (LH1 : RC) absorption ratios so these two PufX À mutants, DM1R and DO10, have a complete ring of LH1 subunits encircling the RC. The only difference between the RC-LH1 complexes of DM1R and DO10 is seen in the 450-to 550-nm carotenoid region where there is clear evidence for lowered carotenoid content, 14% lower when measured at 508 nm, as a consequence of the W -24 F alteration of LH1 in DM1R.
In summary, the lowered carotenoid content of the LH1 antenna (Fig. 2C, bottom spectra), a consequence of the LH1aW -24 F mutation, is the likely explanation for the acquisition of photosynthetic growth in the absence of PufX. The presence of LH1-free RCs cannot account for the photosynthetic growth of the PufX À strain DM1R because the antenna-free, RConly strain RCO1 is photosynthetically inactive under the same light conditions (results not shown). The DM1R mutant has a lowered carotenoid content relative to LH1 BChl pigments The appearance of more carotenoid in the B1 fraction in the DM1R gradient (Fig. 2B,C) shows that some of this pigment, destined for binding to LH1, is unattached to any complex. The presence of some 'free' carotenoid in the B1 fraction of the DO10 control shows that carotenoid biosynthesis is not perfectly coupled to complex assembly. In order to verify the loss of some carotenoid from the LH1 of the DM1R mutant (Fig. 2C), DM1R (LH1aW -24 F, PufX À ) and the DO10 (WT LH1, PufX À ) control were grown semi-aerobically and the carotenoid and bacteriochlorophyll (BChl) pigments were extracted from whole cells with acetone/methanol. Conversely, the presence of more 'free' carotenoids in DM1R (Fig. 2B,C), which are included in the extractions of whole cells, leads to an overestimation of LH1-bound carotenoids in DM1R. Overall, we conclude that the W -24 F alteration of LH1 in DM1R has impaired the binding of carotenoids, as also found in the earlier study [23] on a mutant of Rba. capsulatus harbouring an LH1aW -24 Y alteration.

Discussion
It has long been known that PufX À mutants of Rba. sphaeroides are photosynthetically inactive [9,[11][12][13][14]25] because they assemble an LH1 ring that completely encloses the RC [25,26]. This observation has been difficult to reconcile with bacteria with similarly closed LH1 antenna rings such as Rsp. rubrum, T. tepidum and B. viridis [2,4,27,28], which do grow photosynthetically. Aird et al. [16] conducted a steered molecular dynamics simulation study of the LH1 complex of R. rubrum and proposed that the LH1 ring could allow quinones to move across the complex. Subsequently, the 3D structure of T. tepidum [4] provided a structural basis for this proposal by identifying pores in the LH1 ring for Q/QH 2 exchange. In the present work, we set out to engineer a few such pores in the LH1 ring of Rba. sphaeroides, having first deleted the pufX gene in order to close the normal pore created by the PufX polypeptide [8]. We were guided by the earlier mutagenesis work of Babst et al. [23], who showed that the LH1aW -24 Y mutation results in the loss of 40% of the carotenoids from the LH1 complex of Rba. capsulatus. Here, we show that alteration of LH1aW -24 to F in the DM1R mutant, which lowers the cellular carotenoid content and also the level of carotenoids bound by the DM1R LH1 complex, restores some photosynthetic growth to a PufX À LH2 À strain of Rba. sphaeroides. The structural basis for lowered carotenoid levels in the LH1aW -24 F complex has not been established, as carotenoid binding sites are poorly defined in the Rba. sphaeroides RC-LH1-PufX complex. This mutation also affects the assembly of the LH1 complexes to a limited degree, which results in some 'free' RCs in the membrane with no surrounding LH1 ring; similar effects were observed in previous studies of LH1 assembly mutants of Rba. sphaeroides [21,29]. The PufX-minus RC-LH1 complexes in the DM1R mutant assemble closed LH1 a 16 b 16 rings because co-operative forces between the RC and LH1 a 1 b 1 BChl 2 subunits drive the process of encircling RCs with LH1 a 1 b 1 BChl 2 Crt 2 units to completion, at the expense of some 'free' RCs that have no LH1 subunits. This LH1 ring completion process is observed even under quite extreme circumstances; for example, only~14% of the LH1a polypeptide survives deletion of the LH1 assembly factor LhaA yet complexes still assemble with a normal RC : LH1 stoichiometry, leaving a substantial population of 'free' RCs [21]. In summary, we propose that the LH1 ring of the PufX À LH1aW -24 F mutant DM1R completely surrounds the RC, on the basis of the near-identical RC : LH1 absorption ratios for DM1R and the control strain DO10 (Fig. 2C, bottom). The LH1-free RCs in DM1R require no PufX to facilitate Q/QH 2 turnover at the RC Q B site, but they are unable to function at the low light intensities used because they have no antenna to collect and deliver excitation energy; this is why the control RC-only strain RCO1 shows negligible photosynthetic growth. Thus, the observed growth rate of DM1R likely arises from the lowered carotenoid content of the RC-LH1 complexes and the ensuing pores for Q/QH 2 traffic, which are depicted in Fig. 3.
Phototrophic bacteria such as T. tepidum and Rsp. rubrum possess monomeric LH1 complexes that completely encircle the RC but small pores in each  complex, created by having only a single carotenoid for each a 1 b 1 BChl 2 subunit, likely allow Q/QH 2 exchange and therefore turnover at the RC Q B site [4,16]. Figure 3 shows schematic diagrams of a section of the LH1 ring of the control strain DO10, the reduced carotenoid mutant DM1R, T. tepidum and the core monomer mutant DX13, which in this context is equivalent to the WT. We illustrate the proposed gaps between LH1 a 1 b 1 BChl 2 Crt units that allow Q/QH 2 exchange across the T. tepidum LH1 ring [4], and suggest that in Rba. sphaeroides the second carotenoid within each LH1 a 1 b 1 BChl 2 unit effectively plugs these pores and prevents Q/QH 2 exchange. This problem in the PufX À strain DO10 is partly alleviated by thẽ 14% reduction in the carotenoid content of LH1 in strain DM1R, which creates a few pores for passage of Q/QH 2 . The optimal configuration is found in DX13 and the WT, where Q/QH 2 traffic is confined to a site adjacent to the RC Q B site, and a doubling of carotenoid content provides enhanced light absorption by the LH1 ring.
In relation to monomeric cores lacking PufX, the dimeric RC-LH1-PufX complex of Rba. sphaeroides has several functional upgrades that improve its light- Fig. 3. Schematic representation of part of the RC-LH1 complex of T. tepidum, Rba. sphaeroides control strains DO10 and DM13, as well as the lowered LH1 carotenoid strain DM1R. The schematic depicts a section through the LH1 ring at the level of putative quinone pores between adjacent LH1 a 1 b 1 BChl 2 units. Carotenoids are depicted as foreshortened cylinders, viewed end-on. For simplicity, BChls are not shown. In T. tepidum, only one of the two carotenoid binding sites is occupied for each a 1 b 1 BChl 2 LH1 unit; we propose that in Rba. sphaeroides the carotenoid (shown in dark orange) between the a and b LH1 apoproteins corresponds to the binding site seen in the T. tepidum structure. In Rba. sphaeroides strain DO10 (WT LH1, PufX À ) full occupancy of carotenoid binding sites in LH1 prevents formation of T. tepidum-like pores. Lowering the LH1 carotenoid content in strain DM1R impairs full occupancy of the carotenoid binding sites and passage of quinones through one or more pores is sufficient to permit some photosynthetic growth; 86% of the LH1 carotenoids are retained in DM1R with respect to DO10, so the assumption is made that most carotenoid binding sites are not affected by the alteration of LH1aW -24 to F in the DM1R mutant. In the DX13 and WT strains, the PufX polypeptide creates a quinone pore, which allows the assembly of 14 a 1 b 1 BChl 2 Crt 2 units round the RC. harvesting and quinone exchange functions. First, PufX binds to the RC-H subunit and creates a single, specialised channel for efficient quinone traffic across the LH1 ring at a strategic location, relatively close by the RC Q B site [8]. Second, the LH1 complex is freed from maintaining small pores for quinone exchange across LH1 a 1 b 1 BChl 2 Crt units, so allowing recruitment of more carotenoids that enhance the light-harvesting and photoprotective functions of the LH1 complex. Photoprotective effects are initiated by the oxygen-driven in situ modification of spheroidene to yield spheroidenone [30][31][32], activating an intermolecular charge-transfer state that channels excitation energy to the LH1 BChls whilst optimising the triplet energy for singlet oxygen quenching [33]. This highly effective photoprotective mechanism occurs even in LH1 complexes that have been removed from the native membrane [34]. Third, PufX-promoted dimerisation of the RC-LH1-PufX complex confers additional benefits by allowing excitation and quinone sharing, as originally proposed by Comayras et al. [35] and shown directly in the structure of the dimeric RC-LH1-PufX complex [8].