Thylakoid isolation and chlorophyll determination

Healthy, mature P. abies needles were harvested from all five branches under dim light and pooled on ice. Ten grams (fresh weight) of needles was transferred into an ice-cold homogenizer (2 inch blades and 200 ml stainless steel chamber, Omni-Inc, GA, USA) and 100 ml grinding buffer was added (50 mM Hepes–KOH, pH 7.5; 330 mM sorbitol; 5 mM MgCl₂; 10 mM NaF; 10% (w/v) polyethylene glycol, 6000 kDa; 0.075% (w/v) bovine serum albumin; 0.065% (w/v) Na-ascorbate). All steps were performed in a cold room at 4°C with ice-cold reagents. Needles were homogenized for 90 s at 8000 rpm, filtered through two layers of Miracloth, and centrifuged at 4600 g for 6 min. The pellet was re-suspended with a soft paint brush in 25 ml shock buffer (50 mM Hepes–KOH, pH 7.5; 5 mM MgCl₂; 10 mM NaF), transferred into a new centrifuge tube, and the thylakoids were pelleted at 4600 g for 6 min. Thylakoids were re-suspended in 15 ml storage buffer (50 mM Hepes–KOH, pH 7.5; 5 mM MgCl₂; 100 mM sorbitol; 10 mM NaF), starch and residual PEG were removed by brief centrifugation (2 min, 200 g) and thylakoids were again pelleted at 4600 g for 6 min. Thylakoid pellets were re-suspended with a brush in 200 µl of storage buffer, aliquoted, and immediately frozen in liquid nitrogen for later use. All buffers contained 10 mM NaF to preserve the in vivo state of the thylakoid supercomplexes.

Isolation of thylakoids from Arabidopsis was carried out according to Suorsa et al. (2015). Concentration of chlorophyll extracted from thylakoids in 80% buffered acetone was determined according to Porra et al. (1989).

2D-lpBN-PAGE and protein staining

Thylakoids were solubilized with n-dodecyl β-D-maltoside (β-DM) at 1% and 2% (w/v) final concentration for Arabidopsis and P. abies, respectively. Two per cent (w/v) β-DM was required to effectively solubilize P. abies thylakoids (see Supplementary Fig. S1 at JXB online). The solubilized thylakoids were subjected first to lpBN-PAGE for separation of the protein complexes and subsequently to SDS-PAGE for identification of the protein subunits of each complex as described in Järvi et al. (2011). SYPRO Ruby (Invitrogen) staining of the gels was performed according to the instructions supplied. Silver staining was carried out according to Blum et al. (1987).

Mass spectrometry analysis

Protein spots were excised from stained gels and subjected to in-gel tryptic digestion as described by Suorsa et al. (2015). Eluted peptides were identified by nanoscale liquid chromatography–electrospray ionization MS/MS using either a Q-Exactive or a Q-Exactive-HF mass spectrometer (Thermo Scientific) and applying a gradient from solvent A (0.1% formic acid) to B (80% acetonitrile, 0.1% formic acid) of 8–43% for 10 min followed by a step to 100% for 2 min and 8 min 100% solvent B. The 10 most intense peaks in every full MS scan (range 300–2000 m/z) with a resolution of 120 000 were selected for fragmentation in MS2 with a dynamic exclusion window of 10 s. The acquired spectra were matched against a custom protein database (see below) using Proteome Discoverer 2.2 (Thermo Scientific) with an in-house installation of the Mascot server (v. 2.5.1), allowing Cys carbamidomethylation as fixed modification and Met oxidation, Asn/Gln deamination and protein N-terminal acetylation as dynamic modifications.

The mass spectrometry proteomics data were deposited at the ProteomeXchange Consortium (PXD010071, http://proteomecentral.proteomexchange.org; accessed 15/04/2019) via the PRIDE partner repository (Vizcaíno et al., 2013).

Custom protein database for mass spectrometry analysis

A custom protein database was constructed from three different public databases via procedures shown in Fig. 1. Picea abies peptide sequences were derived from Database 1 (chloroplast-encoded gene models, NCBI, NC_021456.1), Database 2 (nuclear-encoded gene models, ConGenIE, Nystedt et al., 2013) and Database 3 (transcripts, ConGenIE). Only transcripts from thylakoid-associated proteins were included, since the MS analysis focused on this subset of proteins.

Open in a separate window

Fig. 1.

Procedure for generation of the merged protein database for MS/MS analysis of thylakoid proteins in P. abies. Database 1 and Database 2 with chloroplast- and nuclear-encoded protein sequences, respectively, were merged with protein sequences derived from transcripts (Database 3). Transcript nucleotide sequences were selected by tBLASTn searches with known thylakoid associated protein sequences from reference species. These candidate sequences were translated to protein sequences and contaminating as well as truncated sequences were removed. This procedure led to an annotated spruce thylakoid protein database (red box, Supplementary dataset S1), which was combined with Database 1 and Database 2 to form a merged protein database (Supplementary dataset S2) that was used for MS/MS identification of P. abies thylakoid proteins.

Candidate transcript sequences from P. abies were identified using tBLASTn homology searches with Arabidopsis or Physcomitrella patens reference sequence queries (Phytozome V11, Joint Genome Institute). The 10 highest scoring BLAST hits were translated to candidate protein sequences and manually checked for redundancy and possible sequencing errors. Full-length protein sequence candidates were trimmed to the predicted N-terminal methionine, and truncated sequences were only used when no full-length protein sequences were found. To exclude contamination (e.g. from lichens; Delhomme et al., 2015) phylogenies of orthologous sequences of representative model species were constructed (A. thaliana, Selaginella moellendorffii, Ph. patens, Chlamydomonas reinhardtii, Synechocystis sp. PCC 6803 and Anabaena sp. PCC 7120 collected from Phytozome V11 and CyanoBase) and candidate sequences occurring together with algal or cyanobacterial orthologues were removed. In cases where multiple unique P. abies sequences were obtained for a single protein, all sequence variations were considered equally valid and therefore included in the custom database (Supplementary dataset S1). Finally, all three databases were combined into a merged protein database that was used for MS spectrum searches (Supplementary dataset S2).

Identification of LHC family proteins from land plants

LHC protein sequences from land plants were identified by homology searches using reference LHC protein sequences from Arabidopsis (LHCB1, AT1G29920.1; LHCB2, AT2G05100.1; LHCB3, AT5G54270.1; LHCB4.1 (LHCB4), AT5G01530.1; LHCB4.3 (LHCB8), AT2G40100.1; LHCB5, AT4G10340.1; LHCB6, AT1G15820.1; LHCB7, AT1G76570.1; LHCA1, AT3G54890.1; LHCA2, AT3G61470.1; LHCA3, AT1G61520.1; LHCA4, AT3G47470.1; LHCA5, AT1G45474.1; LHCA6, AT1G19150.1) and from Ph. patens (LHCB9, Pp3c5_22920v3.1) against databases from 84 different species (Supplementary Table S1). The 10 highest scoring BLAST hits for each reference LHC sequence were translated to amino acid sequences and aligned with the reference. Redundant and truncated sequences (≤20% of reference sequence) were discarded and candidate sequences were trimmed to the predicted N-terminal methionine.

A multiple sequence alignment containing all reference LHCA and LHCB sequences was constructed and six diagnostic regions were identified (Fig. 2) to clarify the identities of candidate LHC sequence from other species. Candidate sequences were considered true orthologues when they shared >75% sequence identity with a particular reference sequence within at least one diagnostic region. Strong conservation of selected diagnostic regions across species was illustrated by constructing sequence logos from a multiple sequence alignment of each region within each LHC orthologue group (see Results and Supplementary Table S2).

Open in a separate window

Fig. 2.

Diagnostic regions for LHC homologue identification. Multiple sequence alignment of LHCB1–8 and LHCA1–6 proteins from Arabidopsis and LHCB9 from Physcomitrella patens was generated and manually adjusted for maximal sequence overlap. N-terminal protein sequences are not shown. Boxes indicate diagnostic regions 1–6, which were used to classify LHC protein sequences. Each LHC candidate sequence was considered a true orthologue if it shared >75% sequence identity with a corresponding reference sequence. Strong conservation of selected diagnostic regions across species was illustrated by constructing sequence logos from a multiple sequence alignment of each region within each LHC orthologue group (see Results and Supplementary Table S2).

Excluded from this procedure were LHCBM sequences, which severely disrupted the LHC multiple sequence alignment. LHCBM sequences of non-vascular plants were assigned according to homology to LHCBM sequences from Ph. patens. Algal LHCA proteins were assigned to their closest land plant orthologue group (Alboresi et al., 2008; Iwai and Yokono, 2017).

In silico comparison of light-harvesting complex family proteins in gymnosperms and other land plants

While building the custom database, it became evident that deep analysis was required to determine the correctly assigned LHC family protein sequences (e.g. LHCB1 or LHCB2) to be included. To this end, the distribution of LHC family proteins LHCA1–6 and LHCB1–9 were investigated in 84 different species ranging from green algae (Chlorophyta), liverworts (Marchantiophyta), mosses (Bryophyta), hornworts (Anthocerotophyta), Lycophytes and ferns (Monilophytes) to seed plants (gymnosperms and angiosperms). Deep comparison was focused on differences between angiosperm and gymnosperm LHC sequences. In the case of Pinaceae, every genus (Picea, Pinus, Abies, Cedrus, Tsuga, Pseudotsuga, Nothotsuga, Larix, Pseudolarix, Keteleeria, and Cathaya) was represented by at least one species.

Since all LHC proteins are homologous, the LHC proteins were identified by homology searches of various genome and transcriptome databases using reference LHC sequences from Arabidopsis (LHCA1–6, LHCB1–8) and Ph. patens (LHCB9). To this end, six manually determined diagnostic regions of Arabidopsis LHC proteins (see ‘Material and methods’) were defined (Fig. 2). This allowed the identification of truncated LHC sequences that would not have been found by relying only on the best homology hits from BLAST searches. This analysis yielded a total of 1366 LHC protein sequences in 84 species. The designation of orthology between each LHC protein and its corresponding reference sequence was supported by strong homology of selected diagnostic regions within each orthologous group, as illustrated in sequence logos for each region in Supplementary Table S2.

The identified LHC proteins in each species were used to create a broad overview of the distribution of LHC homologues in plants. The comparison revealed large variation in the occurrence of LHC proteins between different evolutionary groups, as described in detail below. The results for 35 representative species are shown in Fig. 3 and for all 84 investigated species are shown in Supplementary Table S3.

Open in a separate window

Fig. 3.

Overview of LHCA1–6 and LHCB1–9 protein distribution in 35 land plant species. Green boxes indicate LHC homologue was identified, white boxes indicate LHC homologue was not identified. Phylogenetic tree drawn after Clarke et al. (2011) with classifications for gymnosperms after Christenhusz et al. (2011) and for angiosperms after APG IV system (The Angiosperm Phylogeny Group et al., 2016). Different hierarchies are indicated by typeface: bold (divisions, class and clade), normal (order) and italic (families). Different evolutionary groups are represented by 35 selected species (for all investigated species see Supplementary Table S3): Acam, Acorus americanus; Agro, Agathis robusta; Amtr, Amborella trichopoda; Aqco, Aquilegia coerulea; Arth, Arabidopsis; Bepe, Betula pendula; Ceas, Centella asiatica; Chre, Chlamydomonas reinhardtii; Cofl, Cornus florida; Crja, Cryptomeria japonica; Cymi, Cycas micholitzii; Died, Dioon edule; Gibi, Ginkgo biloba; Gngn, Gnetum gnemon; Guma, Gunnera manicata; Higr, Hibbertia grossulariifolia; Kala, Kalanchoe laxiflora; Lyja, Lygodium japonicum; Mapo, Marchantia polymorpha; Myfr, Myristica fragrans; Noae, Nothoceros aenigmaticus; Orsa, Oryza sativa; Phpa, Physcomitrella patens; Piab, P. abies; Poru, Podocarpus rubens; Rhto, Rhododendron tomentosum (Ledrum palustre); Sagl, Sarcandra glabra; Scve, Sciadopitys verticillata; Semo, Selaginella moellendorffii; Soly, Solanum lycopersicum; Spol, Spinacia oleracea; Tacu, Taxus cuspidate; Vivi, Vitis vinifera; Wemi, Welwitschia mirabilis; Xiam, Ximenia americana. ‘A’ indicates algae-specific LHCA isoforms adapted to land plant LHCA nomenclature (Alboresi et al., 2008; Iwai and Yokono, 2017); ‘M’ indicates LHCBM orthologues instead of LHCB1 and LHCB2 in Chlorophyta, Marchantiophyta, Bryophyta, and Anthocerotophyta.

Among the LHCB proteins, LHCB4 and LHCB8 appeared to have the most distinct distribution between gymnosperms and angiosperms. There are three isoforms of LHCB4 known in Arabidopsis: LHCB4.1 and LHCB4.2 (hereafter LHCB4, as their amino acid sequences are 89% identical and 92% similar) with a longer C-terminus, and LHCB4.3 (hereafter LHCB8 as earlier suggested by Klimmek et al., 2006) with a shorter C-terminus. In the current study, analysis of angiosperms found that LHCB4 was completely conserved, while LHCB8 occurred only in Eurosids (Malvids and Fabids) and in the order Caryphyllales. In the gymnosperm species investigated, LHCB4 and LHCB8 proteins were also differentially present, with both proteins identified in members of Araucariaceae and Podocarpaceae (Araucariales), Sciadopityaceae, Taxaceae, and Cupressaceae (Cupressales), and Ginkgoaceae (Ginkgoales), while LHCB4 was not found in Pinaceae (Pinales), Gnetaceae (Gnetales), or Welwitschiaceae (Welwitschiales) and LHCB8 was not found in the evolutionarily older Cycadaceae and Zamiacae (Cycadales). Interestingly, LHCB4, but not LHCB8, was found in the evolutionarily older non-seed plants (see Fig. 3).

Isolation of LHCB4 and LHCB8 sequences from 63 and 43 different species, respectively, allowed the characterization of distinct differences in the C-termini of orthologues from different evolutionary groups (Fig. 4). The C-terminus of LHCB4 was found to be highly conserved in both length and amino acid composition across angiosperms, gymnosperms (excepting P. abies and other members of Pinaceae in which LHCB4 was not found) and non-seed plants. The C-terminus of LHCB8 in Eurosids and gymnosperms was approximately 10 amino acids shorter than in LHCB4, while the Caryphyllales-type LHCB8 was more similar in length to LHCB4. LHCB8 C-termini also showed substantial amino acid sequence variation between gymnosperms, Eurosids and Caryphyllales, although they were well conserved within each orthologue group, as illustrated in the sequence logos in Fig. 4. Importantly, a 15-amino-acid motif that was strictly conserved in LHCB4 C-terminus (WxTHLxDPLHTTIxD; residues 271–285 in Lhcb4.1 AT5G01530.1, Arabidopsis), was absent from all LHCB8 sequences isolated here. Another unique feature identified in LHCB4 and LHCB8 was a sequence insertion, relative to other LHC sequences, between amino acids 85 and 134 (Lhcb4.1 AT5G01530.1, Arabidopsis). This region harbored considerable sequence variability, both between LHCB4 and LHCB8 and between orthologues from different species (see diagnostic region 2, Supplementary Table S2).

Open in a separate window

Fig. 4.

Sequence logos of LHCB4 and LHCB8 C-termini in different plant groups. Number of species and sequences used for logos as well as reference sequence with accession number and amino acid position for each plant group were the following: LHCB4 – Angiosperms, 41 species, 52 sequences: Arabidopsis AT5G01530.1, 253–290; LHCB4 – Gymnosperms, 16 species, 16 sequences: Sequoia sempervirens UCSsempervirens_isotig06909, 262–297; LHCB4 – non-seed plants 7 species, 12 sequences: Physcomitrella patens Pp3c4_5680V3.1, 260–296; LHCB8 – Eurosids, 8 species, 9 sequences: Arabidopsis AT2G40100.1, 253–276; LHCB8 – Caryophyllales, 5 species, 5 sequences: Spinacia oleracea XP_021841502, 250–284; LHCB8 – Gymosperms, 15 species, 15 sequences: P. abies comp95233_c3_seq1, 275–300. Sequence logos were generated with WebLogo3.6 (Schneider and Stephens, 1990; Crooks et al., 2004).

LHCB3 and LHCB6 proteins were not identified in Pinaceae, Gnetaceae, and Welwitschiaceae, but were present in other gymnosperms, as reported earlier (Kouřil et al., 2016), and in all angiosperm species investigated, as well as in evolutionarily early non-seed land plants. These LHC homologues were found to be absent from green algae, as previously established (Alboresi et al., 2008). LHCB5 orthologues were found in every species investigated (Supplementary Table S3), while the presence of LHCB1, LHCB2, and LHCB7 was variable. LHCB9 is a specific LHC homologue of the moss Ph. patens (Alboresi et al., 2008) and was identified only in that species, while LHCBM sequences replaced LHCB1 and LHCB2 in Chlorophyta, Marchantiophyta, Bryophyta, and Anthocerotophyta.

In contrast to the LHCB proteins, which showed varying distribution in different evolutionary plant groups, the LHCA proteins LHCA1–4 were found in all species investigated with the known exception of LHCA4 in Ph. patens (Alboresi et al., 2008). LHCA isoforms, which are more abundant in algae than in land plants, were here grouped according to their land plant orthologues (Alboresi et al., 2008; Iwai and Yokono, 2017). LHCA5 and LHCA6 were present in angiosperms, except in Ximenia americana (Olacaceae, Santales) and Lophophora williamsii (Cataceae, Caryophyllales), which lacked both proteins. Generally, LHCA6 was only identified in angiosperm species (Fig. 3; Supplementary Table S3).

LHCA5 was present in most gymnosperm species; however, it was not identified in any members of Pinales, Gnetales, or Welwitschiales, nor in Cycas rhumpii (Supplementary Table S3). Among non-seed plants, LHCA5 was also missing from Equisetum hyemale (Equisetales, Monoliophytes) and Marchantia polymorpha (Marchantiales, Marchantiophyta; Ueda et al., 2012).

Thylakoid protein complexes in Picea abies

The in silico identification of the LHC family proteins was expanded to a biochemical analysis with identification of thylakoid protein complexes isolated from summer needles (see ‘Material and methods’) of P. abies (chlorophyll a/b ratio 3.42 ± 0.09), which were compared with those of Arabidopsis (chlorophyll a/b ratio 3.19 ± 0.01). The isolated thylakoids of both species were solubilized with β-DM, using 2% β-DM for P. abies and 1% β-DM for Arabidopsis as final concentrations, and the photosynthetic protein complexes were separated via lpBN-PAGE (Fig. 5). The entire thylakoid membrane, including the appressed grana partitions, is solubilized by β-DM (Järvi et al., 2011; Suorsa et al., 2015), while stronger protein–protein interactions remain intact, thus allowing the investigation of the basic building blocks of native thylakoid protein complexes (Rantala et al., 2017). The concentration of 2% β-DM for P. abies was based on selecting the most optimal detergent concentration that on one hand solubilized all the protein complexes and on the other hand kept the larger supercomplexes intact (see Supplementary Fig. S1).

Open in a separate window

Fig. 5.

Comparison of thylakoid protein complexes isolated from Arabidopsis and P. abies. After solubilization with n-dodecyl-β-D-maltoside (β-DM), the thylakoid complexes were (A) separated by lpBN-PAGE and (B) subsequently stained with Coomassie. Arabidopsis and P. abies thylakoids were solubilized with 1% and 2% β-DM, respectively. Samples were loaded with 8 µg chlorophyll per lane. dm, dimer; mc, megacomplex; mm, monomer; sc, supercomplex.

The lpBN-PAGE separation showed a typical pattern for thylakoid protein complexes in both species (Fig. 5). The bands were named according to the latest Arabidopsis nomenclature (Rantala et al., 2017). In the high molecular mass region in Arabidopsis and P. abies, four bands of the PSII–LHCII supercomplexes (PSII–LHCII sc) were visible (Caffarri et al., 2009; Suorsa et al., 2015; Rantala et al., 2017). They comprise two PSII core complexes (C2, PSII core dimer) with associated LHCII trimers (LHCB1–3) in different stoichiometric ratios. The LHCII trimers are categorized as strongly bound LHCII trimers (S) or moderately bound LHCII trimers (M), both of which are connected to the PSII core via LHCB5 or LHCB4/8 and LHCB6, respectively (Caffarri et al., 2009). PSII core dimers together with various LHCII species give rise to C2S2M2, C2S2M1, C2S2, and C2S1 supercomplexes in the thylakoid membrane. lpBN-PAGE of solubilized thylakoids from both Arabidopsis and P. abies achieved sufficient separation of lower molecular mass complexes PSI, PSII dimer (PSII dm), and ATP synthase from the smaller PSII monomer (PSII mm) and Cyt-bf complexes, as well as loosely bound L-LHCII, comprising only trimeric LHCII, and LHCII monomers.

Despite apparent similarity of the major protein complexes in the lpBN gel, distinct differences between P. abies and Arabidopsis were found in thylakoid protein complex composition. Picea abies completely lacked the M-LHCII band (Fig. 5A), which in Arabidopsis is composed of the minor antenna proteins LHCB4 and LHCB6 as well as trimeric LHCII (Bassi and Dainese, 1992). The PSI–NDH megacomplexes were also absent from P. abies thylakoids (Fig. 5B), consistent with the loss of all plastid encoded subunits of the NDH-1 complex from P. abies (Nystedt et al., 2013). Notably, a large chlorophyll-free protein complex was also visible in P. abies (Fig. 5A), which was identified as ribulose-1,5-bisphosphate carboxylase/oxygenase (RuBisCo).

Distinct protein composition of the light harvesting antenna of PSII in Pinaceae

Strong support for the different occurrences of the LHCB proteins in P. abies was obtained from 2D lpBN/SDS-PAGE analysis (Fig. 6), where LHCB3, LHCB4, and LHCB6 spots were absent from P. abies thylakoid complexes, while LHCB8 was identified.

The absence of LHCB3 and LHCB6, as well as the presence of LHCB8 (instead of LHCB4) in P. abies thylakoids is likely a reason for the lack of the pentameric M-LHCII band in lpBN-gel (Fig. 5). In Arabidopsis, M-LHCII is composed of the M-trimer (i.e. a moderately bound LHCII trimer of the PSII–LHCII sc) together with LHCB4 and 6 (Bassi and Dainese, 1992). While in Arabidopsis the M-LHCII band is the result of thylakoid solubilization with mild detergents (Caffarri et al., 2009; Rantala et al., 2017), no corresponding stable M-LHCII was obtained from solubilized P. abies thylakoids. This is because in P. abies the M-trimer in the PSII–LHCII sc lacks both LHCB3 and the link provided by LHCB6 (Kouřil et al., 2016), while LHCB4 is replaced with LHCB8 (Figs 5, ​,6).6). Instead, a detached ‘M-trimer’ was found to migrate together with trimeric L-LHCII in the lpBN gel. Interestingly, a significant number of unique peptides from LHCB5 were found in the L-LHCII band (spot 35, Supplementary Table S4), leaving open the possibility that LHCB5 is part of the LHCII trimers. Previous studies have shown that in Arabidopsis lhcb2 antisense lines, which are deficient in the expression of both LHCB2 and LHCB1, LHCB5 forms homo-trimers and LHCB5/3 hetero-trimers in place of the LHCB1/2 subunits (Ruban et al., 2003, 2006), strengthening the idea that LHCB5 could be part of a LHCII trimer in P. abies.

Evolutionary changes in the composition of LHCII subunits suggest that P. abies and other Pinaceae have tuned their LHCII composition to maintain a smaller PSII antenna in comparison with Arabidopsis. The differences between these species can be pinpointed to the structural role of the M-trimer and its binding partners in the PSII–LHCII sc, which in Arabidopsis comprise LHCB4/8 and LHCB6 (van Bezouwen et al., 2017). In particular, new evidence is emerging on the special role of LHCB8, often referred to as LHCB4.3, which has already been shown to have a unique sequence and expression profile despite close homology with LHCB4 (Jansson, 1999; Klimmek et al., 2006; Sawchuk et al., 2008). Detailed studies of Arabidopsis knock-out mutants lacking LHCB members that are missing from P. abies, namely LHCB3 (Damkjær et al., 2009), LHCB4 (de Bianchi et al., 2011), and LHCB6 (Kovács et al., 2006; de Bianchi et al., 2008), revealed smaller PSII antenna size, suggesting that these individual LHC subunits affect each other’s stability in the PSII–LHCII sc (Andersson et al., 2001; Kovács et al., 2006; de Bianchi et al., 2008, 2011). Like in P. abies thylakoids (Fig. 5), the M-LHCII complex has been shown to be missing from Arabidopsis mutants lacking LHCB3, 4, or 6 (de Bianchi et al., 2008, 2011; Betterle et al., 2009; Caffarri et al., 2009). Furthermore, LHCB8 cannot alone compensate for the lack of LHCB4 in Arabidopsis (de Bianchi et al., 2011). Instead, LHCB8 seems to have a specific role in long-term high light acclimation, increasing at both transcript (Arabidopsis; Floris et al., 2013) and protein (Pisum sativum; Albanese et al., 2016, 2018) levels, unlike LHCB4.

In addition, decrease in PSII antenna size is also well described in wild type plants during long-term high-light acclimation (Anderson et al., 1988; Walters and Horton, 1994; Murchie and Horton, 1997). At the level of single subunit stoichiometry, a strong decrease in LHCB3 and LHCB6 proteins was observed in Arabidopsis after high light acclimation, which consequently led to reduced amounts of the M-LHCII complex (Ballottari et al., 2007; Kouřil et al., 2013; Wientjes et al., 2013; Bielczynski et al., 2016). Thus, it is conceivable that P. abies and other Pinaceae have adapted their PSII–LHCII sc structure at the genome level to be similar to that observed in Arabidopsis after long-term high light acclimation. In this respect, the substitution of LHCB8 for LHCB4 in Pinaceae, Gnetaceae, and Welwitschiaceae gives a new focus on the role of the LHCB8 isoform. However, since only south-facing branches were sampled in the current study, further analyses are required to clarify whether this feature is identical in north-facing branches.

On the amino acid level, LHCB8 is distinguished from LHCB4 by its different C-terminal sequence (Fig. 4). According to the recently solved PSII–LHCII sc structure of Arabidopsis (van Bezouwen et al., 2017), the LHCB4 C-terminus is in close contact with LHCB6 and likely involved in stabilizing the attachment of LHCB6 to the PSII–LHCII sc. The permanent replacement of LHCB4 with the C-terminally shorter LHCB8 in Pinaceae, Gnetaceae, and Welwitschiaceae may have evolved in parallel with the loss of LHCB6 in these species. LHCB8 lacks a 15-amino-acid motif that contains the mixed chlorophyll a/b-binding site b3 for chlorophyll b614 (Bassi et al., 1999; Pan et al., 2011), which is conserved in the C-terminus of LHCB4. Absence of this chlorophyll from the LHCB8 C-terminus could have ramifications for energy-transfer routes (Cinque et al., 2000; Salverda et al., 2003) and proposed quenching mechanisms (Ioannidis and Kotzabasis, 2015; van Bezouwen et al., 2017) in the PSII–LHCII sc. Additionally, LHCB4 and LHCB8 have a longer N-terminal domain that is not found in other LHC subunits (see diagnostic region 2, Supplementary Table S2). This sequence overlaps with motif II of the ‘knot’ structure of paired PSII–LHCII sc, recently characterized in Pisum sativum (Albanese et al., 2017). Thus, the absence of LHCB4 and the distinct presence of only LHCB8 with a unique motif II ‘knot’ structure could also be involved in unique photosynthetic adaptation in Pinaceae, Gnetaceae, and Welwitschiaceae.

Because of their close homology, it is likely that LHCB8 is a product of LHCB4 gene duplication, with new functions gained through modification of the C-terminus. Variability in the C-terminus of LHCB8 in gymnosperms, Caryophyllales and Eurosids (Fig. 4) and the lack of LHCB8 sequences found in modern species representing the common ancestor of these groups (Fig. 3; Supplementary Table S3) suggest that LHCB8 orthologues may have evolved through independent LHCB4 gene duplications. It is reasonable to assume that the presence of both LHCB4 and LHCB8 increased the functional flexibility of the PSII antenna system, allowing more efficient adaptation to changes in the environment. In contrast, the loss of LHCB3, 4, and 6 from species of Pinaceae, Gnetaceae, and Welwitschiaceae argues for a highly specialized PSII antenna system that may be more capable of long-term high light acclimation at the expense of functional flexibility. Comparative analysis of Pinaceae, Gnetaceae, and Welwitschiaceae species is needed to resolve how the above-described antenna modifications affect the collection, distribution, and dissipation of excitation energy and the general performance and tolerance of the photosynthetic machinery.

This project was funded by the European Union’s Horizon 2020 research and innovation program under grant agreement no. 675006 and the Academy of Finland (project no. 307335 and 303757). Special thanks go to Roberta Croce for helpful discussions about LHCB8. The authors thank Virpi Paakkarinen, Jyrki Rasmus, and Ville Käpylä for their excellent technical assistance.