colour [ 7 ]. Phylogenetic analysis of this strain, and closely related species synthesising BChl b, led to transferral to the novel Blastochloris genus; the isolate described above being designated Blastochloris (Blc.) viridis [ 8 ].

The reaction centre (RC), the site of charge separation that initiates photosynthetic electron transfer, is encircled by an antenna known as light-harvesting complex 1 (LH1) to form the ‘core’ RC–LH1 supercomplex, the key functional unit for phototrophy in Proteobacteria. RC–LH1 complexes contain five universal components: the L, M, and H subunits of the RC, and the α and β polypeptides of LH1. Most proteobacterial RCs, including that of Blc. viridis, have a bound cytochrome subunit, C, containing four haem cofactors. The RC from Blc. viridis was the first membrane protein to have its structure solved, earning the 1988 Nobel Prize in Chemistry [ 9,10 ]. Some LH1 antennas also contain additional subunits; the Rhodopseudomonas palustris LH1 ring contains a single transmembrane helix known as protein W [ 11,12 ], LH1 from Rhodobacter spp. contains a PufX polypeptide [ 13–15 ], and a further subgroup of these organisms additionally contain PufY [ 16–18 ]. These polypeptides create a channel in the LH1 ring allowing quinone/quinol exchange between the RC and the cytochrome bc1 complex [19,20]. An additional LH1 subunit was also identified in Blc. viridis [21]. However, unlike those mentioned above, this LH1γ polypeptide was found to be in apparent equimolar ratio with the α and β polypeptides [22]. A recent cryo-electron microscopy structure of the RC–LH1 complex from Blc. viridis revealed the location of γ, packing between β polypeptides on the outside of the ring (Figure 1) [ 16 ]. The α : β : γ ratio was found to be 17 : 17 : 16; in this case the ‘missing’ γ subunit creates the channel for quinone diffusion. We proposed that the role of the γ subunit in this complex is to tighten packing of the BChls in LH1, increasing excitonic coupling of these pigments, which results in the extreme ‘red-shift’ of the complex past 1000 nm to the current low-energy limit for photosynthesis on Earth.

Peptide analysis of γ isolated from Blc. viridis indicated that the polypeptide is 36 amino acids in length, and variance at position 34 was detected with threonine and valine residues identified in a 2 : 1 ratio, respectively [22]. This suggested that multiple copies of LH1γ-encoding genes were present in the Blc. viridis genome. Subsequent genome sequencing has revealed that Blc. viridis has four genes encoding putative γ polypeptides, three of which are clustered and share high sequence identity (LH1γ1–3), and the fourth being more divergent (LH1γ4) [23,24] (Figure 2).

In the present study, we constructed mutants of Blc. viridis that do not produce LH1γ, permitting the confirmation of the role of this unusual core complex component. Our results indicate that the loss of γ results in large blue-shifts in the absorption maxima of whole cells and of the isolated RC–LH1 complex to a region of the solar spectrum within which photons poorly transmit through the atmosphere, providing an evolutionary rationale for the recruitment of this additional subunit into the bacterial photosystem.

Deletion of LH1γ-encoding genes prevents absorption of light >1000 nm To determine the role of the LH1γ polypeptides in the Blc. viridis RC–LH1, the open reading frames encoding LH1γ1, LH1γ2, and LH1γ3, which are clustered together in the genome, were replaced with a spectinomycin resistance cassette, generating the mutant strain ΔLH1γ1–3 (Figure 3A). Subsequently, the more divergent paralog LH1γ4 was replaced in the ΔLH1γ1–3 background with an ampicillin/carbenicillin resistance cassette, generating strain ΔLH1γ1–4 lacking all copies of LH1γ-encoding genes (Figure 3B). The presence of LH1γ4 has not been detected in the RC–LH1 of Blc. viridis, thus was not removed from the genome of the WT strain.

BChl-dependent anoxygenic phototrophs have classically been cultured with illumination provided by incandescent bulbs, which emit a broad spectrum of light from the visible deep into the infrared [26]. Since incandescent light bulbs have been banned in many parts of the world (e.g. phased out in the European Union by 2012), many researchers working on these organisms switched to halogen bulbs [27,28], which have now also been banned for production but can still be sourced. Therefore, our standard laboratory conditions for culturing anoxygenic phototrophic proteobacteria is illumination at 100 mmol photons m−2 s−1 provided by halogen bulbs (see Supplementary Figure S1 for bulb emission spectra). The ΔLH1γ1–3 and ΔLH1γ1–4 mutants, along with the WT, were cultured under these conditions, and their whole-cell absorption spectra were measured (Figure 4). The WT displayed the characteristic absorption band with a maximum at 1018 nm. Interestingly, the spectra obtained for the ΔLH1γ1–3 and ΔLH1γ1–4 strains displayed a prominent absorption peak at 825 nm, and a small absorption feature at ∼970 nm, consistently observed through multiple passages. This absorption profile is similar to that of the purified RC from Blc. viridis [29], and to membranes isolated from a mutant of BChl a-synthesising Rhodobacter sphaeroides containing RCs but lacking any antenna complexes, albeit with absorption maxima blue-shifted from those of the BChl b-containing mutants described here [30]. In the context of these studies, our results indicate that mutants ΔLH1γ1–3 and ΔLH1γ1–4 are unable to assemble LH1 in the absence of LH1γ polypeptides, when grown under our standard laboratory conditions. This suggested that loss of LH1γ may severely affect the stability of the LH1 ring, or prevents its assembly altogether.

The absorption features in the 400–550 nm range of the whole-cell spectra of the ΔLH1γ1–3 and ΔLH1γ1–4 mutants grown under halogen light indicate that the cells accumulate carotenoids, accessory pigments that absorb in this range and also play roles in pigment–protein complex stability and quenching of triplet excited states of (B)Chls [31]. The ΔLH1γ1–3 and ΔLH1γ1–4 mutants, along with the WT, were cultured under white light from fluorescent tubes in order to provide carotenoid-specific wavelengths that would not be captured by the BChls in the RC. Under these conditions the strains were phototrophically viable and the whole-cell absorption spectra were measured (Figure 4). As under halogen light, the WT displayed characteristic absorption with a maximum at 1018 nm. Surprisingly, ΔLH1γ1–3 and ΔLH1γ1–4 both displayed spectra with absorption maxima at 972 nm, 46 nm blue-shifted with respect to the WT maximum. These mutant spectra displayed an absorption profile more similar to the WT and unlike those of the RC alone, suggesting that these strains assemble the LH1 ring around the RC, albeit without LH1γ. Interestingly, strain ΔLH1γ1–3, which still contains LH1γ4, displays the same absorption spectra as the mutant lacking all encoding genes; the ΔLH1γ1–4 strain completely lacking LH1γ was subsequently used for all experiments.

LH1 assembles around the RC without LH1γ only under white light To analyse the pigment–protein complexes assembled in the described strains, they were cultured under halogen light and white light, the membranes from these cells were solubilised with detergent, and the resulting membrane complexes were subjected to rate-zonal centrifugation on continuous sucrose density gradients (Figure 5). The WT grown under both halogen light and white light accumulates a sole pigmented complex, each displaying identical densities and absorption spectra. The mutant lacking LH1γ grown under halogen light accumulates a single complex at greatly reduced density compared with that of the WT; the absorption spectrum of this band displays the characteristic profile of an isolated BChl b-containing RC. When this mutant is grown under white light, RCs can also be identified in the sucrose gradients, but a second pigmented D o w n l o a d e d fr o m h tt:// p p o ltr a n d p r e s s . c o m i/ b o c h e m lijt/r a c e p d /f 4 7 9 / 2 4 / 2 4 4 9 / 9 4 1 2 3 5 / b c j 2 0 2 2 0 5 0 8 . p d f b y g u e s t o n 2 6 O c t o b e r 2 0 2 3 complex, with a density slightly less than that of the WT RC–LH1, is also present as the major band. The absorption spectrum of this band displays a maximum at 965 nm, slightly blue shifted from its whole-cell absorbance maximum, as is common for isolated complexes.

The major pigmented band from each of the gradients was isolated and buffer-exchanged in a spin concentrator to remove sucrose. The complexes were denatured in SDS sample buffer and the subunits separated via Tris–tricine gel electrophoresis (Supplementary Figure S2). The stained gel indicates that the major complex from the LH1γ-lacking mutant grown under halogen light does not contain any LH1 components, including LH1β and LH1α. However, the major complex isolated from this strain grown under white light contains LH1β and LH1α, but as expected lacks LH1γ, confirming that this blue-shifted complex is an intact RC–LH1. Loss of LH1γ slows the phototrophic growth rate under halogen light To assess the effect of deletion of the genes encoding LH1γ, the WT and ΔLH1γ1–4 strains were cultured under white light conditions to mid exponential phase, standardised by OD700 and used to inoculate biological triplicates in fresh medium incubated under halogen light and white light. When grown under halogen light, the doubling time of the mutant was almost twice that of the WT (Table 1). However, the growth rates measured under white light, when both strains assemble intact RC–LH1, were comparable, albeit much slower than under halogen light, and each strain displayed inconsistent growth across the triplicate cultures.

The vast majority of anoxygenic phototrophs in diverse bacterial phyla discovered to date use BChl a as the RC pigment for phototrophy [37]. Only a few purple bacteria use BChl b [ 4 ]. Of these, only Blastochloris spp. have orthologs of LH1γ. In each case, the common components of the photosystem are encoded by single ORFs within the photosynthesis gene cluster (PGC), which also contains genes encoding pigment biosynthesis enzymes and photosystem assembly factors [38]. However, each sequenced strain contains multiple LH1γ genes that are located distant from the PGC [24,25,39,40]. This suggests that these genes evolved separately from the other PGC genes, and that assembly of RC–LH1 without LH1γ is safeguarded against by the presence of multiple paralogs in the genome. The evolutionary rationale for this is provided by our results; our mutant strains assembling RC–LH1 lacking LH1γ absorb maximally at 972 nm, which sits in a range of the solar spectrum that poorly transmits through the atmosphere due to absorption by water vapour (Figure 8). In this range, few photons reach the surface of Earth, and those that do also poorly transmit through liquid water at the surface [41]. Therefore, it is likely that Blastochloris spp. evolved or recruited an additional LH1 subunit to influence absorption of the complex, shifting to a region of the spectrum where photons are of lower energy but are more abundant.

LH1γ polypeptides 1 and 2 are identical, and LH1γ3 differs by a single amino acid (see Figure 2); their encoding genes are clustered in the genome. The original proteomic analysis of Blc. viridis demonstrated that highly similar polypeptides containing threonine or valine at position 34 could be isolated from its photosystem and are present at a 2 : 1 ratio, while a polypeptide equivalent to LH1γ4 was not detected [22]. This indicates that the genes encoding LH1γ1–3 are transcribed together, and that LH1γ4 is not produced when the cells are grown under full-spectrum light from incandescent bulbs. Similarly, we could not detect any phenotypic change in the cells of the ΔLH1γ1–3 mutant when compared with the mutant lacking all γ-encoding genes under any condition tested that we could ascribe to the production of LH1γ4. Our finding that expression of the LH1γ4 gene from a plasmid in the ΔLH1γ1–4 mutant leads to the cells having an absorption maximum of 1003 nm indicates that incorporation of γ4 into RC–LH1 induces a moderate red-shift, which may provide more favourable absorption under certain environmental conditions (e.g. when in competition with other BChl b phototrophs such as Halorodospira spp. displaying similar absorption at 1018 nm, see below). It may be that incorporation of LH1γ4 into the WT RC–LH1 can be induced by growth under specific LED regimes in the laboratory, which we intend to explore further.

We found that the ΔLH1γ1–3 and ΔLH1γ1–4 mutants grown under our standard laboratory conditions, with illumination provided by halogen bulbs, resulted in the assembly of RCs, but no intact RC–LH1 complexes. However, growth under white-light fluorescent tubes that would be routinely used for cultivation of oxygenic cyanobacteria permitted assembly of RC–LH1, with an absorption maximum far outside the range of emission from the light source. This appears counterintuitive, but it is possible that the emission from the halogen bulbs, which reduces sharply at ∼850 nm (see Supplementary Figure S1), provides wavelengths that efficiently excite the BChls contributing the major absorption feature of the RC at 831 nm (see RC band spectrum in Figure 5). This RC contains only a single 15-cis-1,2-dihydroneurosporene carotenoid, providing limited absorption in the visible range of the spectrum. The observation that RC–LH1 lacking γ only assembles when the mutants are grown under white light might be explained by the carotenoid content of LH1; the 17 all-trans pigments bound in the ring provide strong absorption in the 400–550 nm range, which overlaps with the narrow, intense emission band of the fluorescent tubes in this range (see Supplementary Figure S1). Taking this observation together with the very slow growth rate of WT and ΔLH1γ1–4 under white light (see Table 1), an alternative explanation might be that Blc. viridis senses this regime as very low light and so may compensate by increasing expression of the LH1 α and β genes. A complete elucidation of this phenomenon will form the basis of a future study. Our results will be noteworthy to those studying anoxygenic phototrophy; although it is possible to purchase LED panels that mimic ‘photosynthetically active radiation’ for oxygenic phototrophs, panels emitting wavelengths to at least 1200 nm that more accurately mimic full-spectrum natural sunlight do not yet exist. As it becomes more difficult to source incandescent and halogen bulbs there may be a need for researchers to assemble broad-spectrum LED regimes themselves, to accurately study phototrophic bacteria under conditions similar to those experienced in nature.

The mutants lacking LH1γ assembling RC–LH1 in white light also accumulate RCs, as can be seen in the sucrose density gradients, while the WT does not (Figure 5). This suggests that the presence of the 16 LH1γ subunits provide a stabilising role; the interaction with the adjacent β polypeptides may fix the complex in place in addition to imparting the red-shift in absorption, and the presence of γ may orient the RC in such a position that efficient quinone/quinol shuttling can occur at the ‘missing’ 17th γ position [ 16 ]. This stabilising effect may be analogous to the role that PufX plays in the primarily dimeric Rhodobacter sphaeroides RC–LH1 [43]. Deletion of the pufX gene leads to formation of solely monomeric RC–LH1, and the recent cryo-EM structure of this complex was solved at much lower resolution than the PufX-containing dimer or monomer, attributed to PufX’s role in orienting the RC within LH1 [ 43,18 ]. It is interesting that we do not observe assembly of RCs without LH1 in the WT under either white light or halogen light conditions. This could be due to the stability imparted by γ; it is likely that the formation of RC–LH1 is a highly co-ordinated process and that γ is incorporated into LH1 at the same time as the α and β polypeptides by the assembly machinery, and this may not be easily reversed once γ has stabilised the complex. Turnover and recycling of light-harvesting and RC complexes is a well-studied process; photosystem II of oxygenic phototrophs is damaged by the reaction it catalyses so is constantly reassembled [44], and the metabolically diverse purple bacteria using BChl a can assemble and disassemble RC–LH1 depending on whether conditions suit aerobic respiration [45]. However, Blastochloris spp. are unable to grow in the presence of O2 and they generally inhabit anoxic ecological niches such as the bottom of microbial mats, where rapid turnover of the photosystem is not required [46]. Here, extremely red-shifted absorption provided by the γ-stabilised complex affords a competitive advantage over the abundant BChl a-containing phototrophs in the mat where wavelengths are highly filtered, and thus a route to survival and growth.

Halorhodospira (Hrs.) halochloris and Hrs. abdelmalekii (formerly Ectothiorhodospira halochloris and Ectothiorhodospira abdelmalekii) are additional BChl b-containing purple bacteria isolated from hypersaline and alkaline soda lakes in Wadi El Naturn, Egypt, that also display whole-cell absorption maxima at 1018 nm [47,48]. The isolated RC–LH1 complexes from these organisms were found to contain three low molecular weight components, and the LH1 absorption at 1018 nm could be reversibly blue-shifted to 964 nm when titrated with acid in the pH range of 7.5–5.7 [49]. The small polypeptide of a similar size to Blc. viridis LH1γ was isolated from the RC–LH1 complex of Hrs. halochloris and was assigned to be analogous, although its sequence was not presented in the publication [50]. The recent complete genome sequence of Hrs. halochloris has revealed than an ortholog of a Blastochloris spp. gene encoding LH1γ is absent [51]. The inability to find LH1γ orthologs in the genome of Hrs. halochloris — as well as in other non-Blastochloris BChl b-containing organisms — via BLAST search is unsurprising since the polypeptide is a small, single transmembrane helix (e.g. the sequences of the single transmembrane PufX polypeptides from Rhodobacter spp. display low sequence identity across the genus [52]). Hrs. halochloris was found to have an unusual complement of genes encoding RC–LH1 subunits [51]. This organism has 2 pufBALMC operons, and an additional two pairs of pufBA genes found elsewhere in the genome, one pair of which ( pufB3A3) is more divergent, sharing only 41% and 42% identity with LH1β1 and LH1α1, respectively, which may lead to assembly of LH1 containing multiple forms of α- and/or β-polypeptides, as was found in the RC–LH1 of BChl a-containing Thiorhodovibrio strain 970 [53]. A recent study demonstrates that desalting the purified Hrs. halochloris RC–LH1 complex results in the same blue-shift described previously with a decrease in pH, which can be reversed by supplementation with salts [54]. Taking all of these results together, it is likely that Halorhodospira spp. have evolved an alternative approach to accessing photons of wavelengths >1000 nm to that taken by members of the Blastochloris genus; the red-shift displayed by these organisms appears to be strongly influenced by electrostatic charge.

This study elucidates the role of the γ polypeptide in the RC–LH1 of Blc. viridis. The presence of the highly conserved γ isoform stabilises the complex and shifts its absorption 46 nm further into the near-infrared. This extreme red-shift allows Blastochloris spp. to access the more abundant photons >1000 nm that are not absorbed by water in the atmosphere. The discovery that plasmid-borne expression of a divergent isoform of LH1γ not found in the WT complex results in only a moderate red-shift of the complex upon assembly provides a route to explore the extent to which polypeptide sequence influences the absorption shift, and whether the low energy limit of natural photosynthesis can be surpassed by rational design of the LH1γ subunit.

Growth of described strains Blc. viridis DSM 133 and mutant strains were grown phototrophically at 30°C in sodium succinate medium 27 (N27 medium) [55] under illumination (100 mmol photons m−2 s−1) provided by 100W Bellight halogen bulbs or a bank of 24W fluorescent tubes as previously described [56]. Immediately after autoclaving, bottles containing hot growth medium were tightly sealed and transferred to an anoxic chamber (Coy Laboratories) maintained with an atmosphere of 95% N2, 2.5% H2, 2.5% CO2. Once inside, lids were loosened and incubated for at least 48 h prior to use to ensure complete deoxygenation. All culture transfers were performed in this chamber to prevent introduction of O2. Where required, medium was supplemented with antibiotics at the following concentrations: kanamycin (30 mg ml−1), spectinomycin (30 mg ml−1), carbenicillin (100 mg ml−1). E. coli strains JM109 [57] and ST18 (DSM 22074) [58] transformed with pK18mobsacB [59] and pBBRBB-derived plasmids [34] were grown in a rotary shaker at 37°C in LB medium supplemented with 30 mg ml−1 kanamycin. ST18 cells were supplemented with 50 mg ml−1 5-aminolevulinic acid (ALA). All strains and plasmids used in the present study are listed in Supplementary Table S1.