Evolutionary origins and functions of the carotenoid biosynthetic pathway in marine diatoms

Abstract

Carotenoids are produced by all photosynthetic organisms, where they play essential roles in light harvesting and photoprotection. The carotenoid biosynthetic pathway of diatoms is largely unstudied, but is of particular interest because these organisms have a very different evolutionary history with respect to the Plantae and are thought to be derived from an ancient secondary endosymbiosis between heterotrophic and autotrophic eukaryotes. Furthermore, diatoms have an additional xanthophyll-based cycle for dissipating excess light energy with respect to green algae and higher plants. To explore the origins and functions of the carotenoid pathway in diatoms we searched for genes encoding pathway components in the recently completed genome sequences of two marine diatoms. Consistent with the supplemental xanthophyll cycle in diatoms, we found more copies of the genes encoding violaxanthin de-epoxidase (VDE) and zeaxanthin epoxidase (ZEP) enzymes compared with other photosynthetic eukaryotes. However, the similarity of these enzymes with those of higher plants indicates that they had very probably diversified before the secondary endosymbiosis had occurred, implying that VDE and ZEP represent early eukaryotic innovations in the Plantae. Consequently, the diatom chromist lineage likely obtained all paralogues of ZEP and VDE genes during the process of secondary endosymbiosis by gene transfer from the nucleus of the algal endosymbiont to the host nucleus. Furthermore, the presence of a ZEP gene in Tetrahymena thermophila provides the first evidence for a secondary plastid gene encoded in a heterotrophic ciliate, providing support for the chromalveolate hypothesis. Protein domain structures and expression analyses in the pennate diatom Phaeodactylum tricornutum indicate diverse roles for the different ZEP and VDE isoforms and demonstrate that they are differentially regulated by light. These studies therefore reveal the ancient origins of several components of the carotenoid biosynthesis pathway in photosynthetic eukaryotes and provide information about how they have diversified and acquired new functions in the diatoms.

Figure 1. Hypothesized carotenoid biosynthetic pathway in diatoms.

The genes identified in this study, phytoene synthase (PSY), phytoene desaturase (PDS), ξ-carotene desaturase (ZDS), lycopene β-cyclase (LCYB), β-carotene hydroxylase (BCH), lutein deficient-like (LTL), zeaxanthin epoxidase (ZEP), violaxanthin de-epoxidase (VDE) and violaxanthin de-epoxidase-like (VDL), are indicated. The BCH-encoding gene is absent in the P. tricornutum genome, ZEP3 and VDL2 are absent in the T. pseudonana genome. The two xanthophyll cycles are boxed, A) the violaxanthin cycle and B) the diadinoxanthin cycle. α-carotene and lutein are not produced by diatoms. Dashed arrows indicate hypothetical conversion steps, according to Lohr and Wilhelm (1999, 2001).

Figure 2. Domain structure of violaxanthin de-epoxidases and related proteins.

A) Schematic representation of VDE, VDL and VDR proteins (not to scale). Three different domains are shown; the cysteine-rich domains include the N-terminal targeting sequence. Black and red asterisks indicate the positions of conserved and alternative cysteine residues, respectively. The central lipocalin domain contains the lipocalin binding fold. Conserved and divergent lipocalin motifs (roman numbers) are given in black and red, respectively. The size of the lipocalin motif was determined by sequence alignment of VDE sequences and a representative group of lipocalin proteins. The C-terminal glutamic acid-rich domain indicates the percentage of Glu residues in this domain. B) Alignment of the N-terminal cysteine-rich domains of several plant and diatom VDEs. Also included is a sequence derived from the amoeba Acanthamoeba castellanii. C) Alignment of the lipocalin motifs I, II and III of several different lipocalin VDE, VDL and VDR proteins. The distance (in amino acids) between the three lipocalin motifs is also indicated. The lipocalin motif consensus sequences, as derived from kernel lipocalins (Flower, 1996), are indicated above the alignment and conserved motifs within the alignment are indicated in red. The abbreviations used are: Lip, lipocalin; TIL, temperature induced lipocalin; CHL, chloroplastic lipocalin; PRBR, plasma retinol-binding protein precursor; CC, crustacyanin; At, Arabidopsis thaliana; Cr, Chlamydomonas reinhardtii; Dd, Dictyostelium discoideum; Gv, Gloeobacter violaceus; Hg, Homarus gammarus; Hs, Homo sapiens; Mt, Medicago truncatula; Nt, Nicotiana tabacum; Pt, Phaeodactylum tricornutum; Py, Porphyra yezoensis; Ta, Triticum aestivum; Tp, Thalassiosira pseudonana; Vc, Vibrio cholerae.

Figure 3. Maximum likelihood phylogenetic tree of violaxanthin de-epoxidases and related proteins.

A maximum likelihood phylogenetic tree (loglk = −13981.66253) as inferred from amino acid sequences (141 amino acid characters) of violaxanthin de-epoxidases and related proteins was computed using WAG model for amino acid substitution (selected by PROTTEST) with discrete gamma distribution in four categories. All parameters (gamma shape = 2.158; proportion of invariants = 0.000) were estimated from the dataset. Numbers above branches indicate ML/NJ bootstrap supports. ML bootstraps were computed using the above mentioned model in 300 replicates. An NJ tree was inferred using AsaturA program with cutoff value 0.906 and 1000 replicates. Black stars indicate both bootstraps over 90%. Nodes that display different NJ topology than the one obtained by ML are indicated by “dt”.

Figure 4. Domain structure of zeaxanthin epoxidases.

A) Schematic representation of ZEP proteins (not to scale). Three different domains are discriminated; the N-terminal targeting sequence (note that the TpZEP1 gene model is not complete), the central flavin-containing monooxygenase (FMO) domain containing ADP and FAD binding sites, and the C-terminal domain containing the FHA motif. Conserved and divergent lipocalin motifs (roman numbers) are given in black and red, respectively. The size of the lipocalin motif was determined by sequence alignment of ZEPs and a representative group of FMO proteins. B) Alignment of the lipocalin motifs I, II and III of several different lipocalins and ZEP protein sequences. The lipocalin motif consensus sequence, as derived from kernel lipocalins, is indicated above the alignment and conserved motifs within the alignment are indicated in red. The asterix indicates the Gly304 identified by Baroli et al. (2003). The distance (in amino acids) between the lipocalin motifs are also indicated. The abbreviations used are: Lip, lipocalin; TIL, temperature induced lipocalin; CHL, chloroplastic lipocalin; PRBR, plasma retinol-binding protein precursor; CC, crustacyanin; At, Arabidopsis thaliana; Cr, Chlamydomonas reinhardtii; CW80, Chlamydomonas sp. W80; Dd, Dictyostelium discoideum; Gv, Gloeobacter violaceus; Hg, Homarus gammarus; Hs, Homo sapiens; Np, Nicotiana plumbaginifolia; Pt, Phaeodactylum tricornutum; Py, Porphyra yezoensis; Ta, Triticum aestivum; Tp, Thalassiosira pseudonana; Vc, Vibrio cholerae; Vu, Vigna unguiculata.

Figure 5. Maximum likelihood phylogenetic tree of zeaxanthin epoxidases and related proteins.

A maximum likelihood phylogenetic tree (loglk = −48991.81356) as inferred from zeaxanthin epoxidases and related protein amino acid sequences (312 amino acid positions). The tree was computed using WAG model for amino acid substitution (selected by PROTTEST) with discrete gamma distribution in four categories. All parameters (gamma shape = 1.760; proportion of invariants = 0.010) were estimated from the dataset. Numbers above branches indicate ML/NJ bootstrap supports. NJ tree was inferred using AsaturA program with cutoff value 0.908 and 1000 replicates. Black stars indicate both bootstraps over 90%. The original annotation of ZEP related proteins is indicated. Nodes that display different NJ topology than the one obtained by ML, are indicated by “dt”.

Figure 6. mRNA levels of carotenoid biosynthesis and LHC-related genes upon white, blue or red light stimulation.

48-hour-dark-adapted P. tricornutum cells were exposed to 175 µmol m⁻² s⁻¹ continuous white light, or 25 µmol m⁻² s⁻¹ continuous blue or red light and the relative transcript levels of PSY (A), PDS1 (B), FCPB (C) and ELIP-like (D) were determined after 1, 3, 5, 8 and 12h by qRT-PCR using H4 as a reference gene. The values were normalized to the transcript levels in the dark. Data are averages of triplicate measurements. The error bars represent standard deviation.

Figure 7. mRNA levels of xanthophyll cycle-related genes upon white, blue or red light stimulation.

48-hour-dark-adapted P. tricornutum cells were exposed to 175 µmol m⁻² s⁻¹ continuous white light, or 25 µmol m⁻² s⁻¹ continuous blue or red light and the relative transcript levels of ZEP1 (A), ZEP2 (B), ZEP3 (C), VDE (D), VDL1 (E) and VDL2 (F) were determined after 1, 3, 5, 8 and 12h by qPCR using H4 as a reference gene. The values were normalized to the transcript levels in the dark. Data are averages of triplicate measurements. The error bars represent standard deviation.