Abstract
Fluctuating bioavailability of inorganic phosphate (Pi), often caused by complex Pi-metal interactions, guide root tip growth and root system architecture for maximizing the foraged soil volume. Two interacting genes in Arabidopsis thaliana, PDR2 (P5-type ATPase) and LPR1 (multicopper oxidase), are central to external Pi monitoring by root tips, which is modified by iron (Fe) co-occurrence. Upon Pi deficiency, the PDR2-LPR1 module facilitates cell type-specific Fe accumulation and cell wall modifications in root meristems, inhibiting intercellular communication and thus root growth. LPR1 executes local Pi sensing, whereas PDR2 restricts LPR1 function. We show that native LPR1 displays specific ferroxidase activity and requires a conserved acidic triad motif for high-affinity Fe2+ binding and root growth inhibition under limiting Pi. Our data indicate that substrate availability tunes LPR1 function and implicate PDR2 in maintaining Fe homeostasis. LPR1 represents the prototype of an ancient ferroxidase family, which evolved very early upon bacterial colonization of land. During plant terrestrialization, horizontal gene transfer transmitted LPR1-type ferroxidase from soil bacteria to the common ancestor of Zygnematophyceae algae and embryophytes, a hypothesis supported by homology modeling, phylogenomics, and activity assays of bacterial LPR1-type multicopper oxidases.
Introduction
Optimal plant growth exquisitely depends on edaphic resources. The pivotal role of inorganic phosphate (H2PO4- or Pi) in metabolism, paired with its scarce bioavailability, render the mineral nutrient a strongly restrictive factor in terrestrial primary production (Lopez-Arredondo et al., 2014; Crombez et al., 2019). Insolubility of Pi salts and immobility of Pi complexed on clay or metal oxide minerals severely restrict P accessibility. Thus, plants actively seek and mine this crucial macroelement, but must concurrently navigate Pi-associated metal toxicities (Al, Fe) by adjusting root system architecture and modifying rhizosphere chemistry (Kochian et al., 2015; Abel, 2017; Gutierrez-Alanis et al., 2018). When challenged by Pi limitation, most dicotyledonous plants attenuate primary root extension and stimulate lateral root development for increasing the soil volume foraged by multiple root tips, which are the hotspots for Pi capture (Kanno et al., 2016). Root tips monitor heterogeneous Pi distribution (local Pi sensing) for guiding root development (Peret et al., 2014; Abel, 2017). In Arabidopsis thaliana, Pi deprivation rapidly attenuates root cell elongation (<2 h) in the transition zone and progressively inhibits cell division (<2 days) in the root apical meristem (RAM) (Müller et al., 2015; Balzergue et al., 2017). Persistent Pi starvation corrupts the stem-cell niche (SCN), which is followed by root growth arrest (Sanchez-Calderon et al., 2005; Ticconi et al., 2009). Notably, local Pi sensing depends on external Fe availability, which points to antagonistic biologic Fe-Pi interactions (Svistoonoff et al., 2007; Ticconi et al., 2009; Müller et al., 2015; Hoehenwarter et al., 2016; Balzergue et al., 2017; Dong et al., 2017; Godon et al., 2019; Wang et al., 2019).
Genetic approaches identified key components of root Pi-sensing (Abel, 2017; Gutierrez-Alanis et al., 2018; Crombez et al., 2019). A module of functionally interacting genes expressed in overlapping root cell types, PHOSPHATE DEFICIENCY RESPONSE 2 (PDR2), LOW PHOSPHATE RESPONSE 1 (LPR1) and its close paralog LPR2, which plays a minor but additive role in the root response to Pi availability (Svistoonoff et al., 2007), encodes proteins of the secretory pathway and targets both cell elongation and cell division in Pi-deprived root tips. PDR2, the single orphan P5-type ATPase (AtP5A), functions in the endoplasmic reticulum (Ticconi et al., 2004; Jakobsen et al., 2005; Dunkley et al., 2006; Ticconi et al., 2009; Sorensen et al., 2015; Naumann et al., 2019) whereas LPR1, a multicopper oxidase (MCO) with presumed Fe2+-oxidizing activity, is localized to cell walls (Svistoonoff et al., 2007; Müller et al., 2015). On low Pi, the PDR2-LPR1 module mediates LPR1-dependent, root cell type-specific Fe3+ accumulation in the apoplast, which correlates with ROS (reactive oxygen species) generation and callose deposition (Müller et al., 2015). While ROS formation promotes peroxidase-dependent cell wall stiffening in the transition zone (Balzergue et al., 2017), callose deposition interferes with cell-to-cell communication and thus inhibits RAM activity (Müller et al., 2015). LPR1-dependent root cell differentiation likely intersects with peptide and brassinosteroid signaling (Gutierrez-Alanis et al., 2017; Singh et al., 2018).
Current evidence points to LPR1 as a principal component of Fe-dependent Pi sensing. Upon Pi limitation, insensitive lpr1 mutations cause unrestricted primary root extension by preventing Fe accumulation and callose deposition in root tips. Because loss of LPR1 (or external Fe withdrawal) suppresses hypersensitive pdr2 root phenotypes in low Pi, PDR2 restricts LPR1 function; however, the underlying mechanisms are unknown (Müller et al., 2015; Hoehenwarter et al., 2016; Naumann et al., 2019). Thus, the biochemical identity of LPR1 and the mechanism of LPR1 activation upon Pi deprivation need to be established. Here we determine the catalytic properties of purified native and mutant LPR1 variants and monitor LPR1 expression in root tips. We show that LPR1 encodes a prototypical, novel ferroxidase with high substrate (Fe2+) specificity and affinity. While LPR1 is expressed in root meristems and is independent of PDR2 function or Pi and Fe availability, LPR1-dependent root growth inhibition in limiting Pi is highly sensitive to low external Fe concentration. LPR1 substrate availability governs the local Pi deficiency response, whereas PDR2 maintains homeostasis of LPR1 iron reactants. Intriguingly, LPR1-like proteins, characterized by possessing in their active site an acidic triad motif and distinctive Fe2+-binding loop, are present in all extant land plants, in Zygnematophyceae algae, and in soil bacteria. Our phylogenetic and biochemical analyses support the hypothesis that LPR1-type ferroxidases evolved very early during bacterial land colonization and appeared in plants through horizontal gene transfer from Terrabacteria to the common ancestor of Zygnematophyceae and Embryophyta.
Results
LPR1 expression in root meristems is independent of PDR2 and external Pi supply
Analysis of pLPR1Col::eGFP expression in Pi-replete A. thaliana primary and secondary root tips revealed highest pLPR1 promoter activity in the SCN and weaker GFP-derived fluorescence in proximal endodermal and cortical cells (Fig. 1a). We compared cell-specific pLPR1 promoter activity in wild-type and pdr2 roots upon transfer of Pi-replete seedlings (5 d-old) to +Pi or –Pi media for up to 7 d. The pLPR1 expression domain and GFP intensity were similar between both genotypes on +Pi medium and did not change notably in the wild-type upon Pi deprivation. In pdr2 root tips, pLPR1::GFP expression was maintained for at least 1 d on –Pi agar and thereafter ceased because of RAM disorganization (Fig. 1b). We observed a similar genotype-and Pi-independent pLPR1::GFP expression pattern in root tips of plants continuously grown on +Pi or –Pi medium for up to 4 d (Extended Data Fig. 1). Because gene expression and RAM activity rapidly respond to Pi deprivation (within 24 h) (Müller et al., 2015; Hoehenwarter et al., 2016, Balzergue et al., 2017), we analyzed steady-state mRNA levels in excised root tips 1 d after seedling transfer from Pi-replete to +Pi or –Pi media. Our data confirm that expression of LPR1 and its functionally redundant sister gene, LPR2, is independent of external Pi supply or PDR2 function (Fig. 1c).
To monitor LPR1 protein abundance, we generated peptide-specific anti-LPR1 anti-bodies that specifically recognize LPR1 in roots of overexpression (pCaMV 35S::LPR1) plants (Extended Data Fig. 2a, b). However, the antibodies failed to detect LPR1 in wild-type and pdr2 root tips, even when profuse lateral root formation is stimulated to increase the number of root meristems, which express pLPR1::GUS. Likewise, attempts to enrich LPR1 by immuno-or chemical precipitation did not improve detection (Extended Data Fig. 2e-f). Eventually, using Tandem Mass Tag labeling (TMT) Mass Spectrometry, we detected with high confidence LPR1- and LPR2-derived peptides by quantitative proteomics on excised wild-type and pdr2 root tips. Our data indicate that LPR1 and LPR2 abundance in root meristems does not depend on PDR2 function nor on external Pi status (Fig. 1d).
Purified native LPR1 displays specific and high-affinity ferroxidase activity
The lack of evidence for Pi- or PDR2-dependent regulation of LPR1 expression prompted us to purify and characterize the LPR1 MCO enzyme. Because numerous attempts failed to express active, affinity-tagged recombinant LPR1, we purified native LPR1 to near homogeneity (monomeric protein of ∼70 kDa) from leaves of LPR1-overexpression plants (Fig. 2a). The three-step purification procedure involved differential ammonium sulfate precipitation followed by size exclusion and cation exchange chromatography, which yielded 2 μg LPR1 protein per gram leaf material (Extended Data Table 1). Immunoblot analysis, peptide sequencing by mass spectrometry, and ferroxidase activity assays verified the identity of LPR1, which was not detectable in identically prepared fractions from lpr1lpr2 leaves (Extended Data Fig. 3, Extended Data Fig. 4).
The diverse MCO protein family comprises phenol oxidases (laccases), bilirubin oxidases, ascorbate oxidases, and metal oxidases such as ferroxidases from yeast (Fet3p) or humans (ceruloplasmin) (Graff et al., 2020). We previously reported elevated ferroxidase activity in root extracts of LPR1-overexpression plants (Müller et al., 2015). Using purified native LPR1 and the ferrozine assay, we determined kinetic parameters of its ferroxidase activity. LPR1 exhibited a typical Michaelis-Menten saturation kinetics for Fe2+ oxidation, which revealed an apparent Km value of 1.8 μM, a Vmax value of 29 nkat mg-1, and a turnover frequency kcat of 1.9 s-1 (Fig. 2b-d). LPR1 displayed highest ferroxidase activity at pH 5.8, a value consistent with LPR1 function in the apoplast (Müller et al., 2015). In addition to its ferroxidase activity, we tested LPR1 for laccase activity with ABTS (2,2’-azino-bis[3-ethylbenzothiazoline-6-sulfonic acid]) as the substrate, and for ascorbate, bilirubin and manganese oxidase activities (Fig. 2b, Extended Data Fig. 5). The inability to oxidize the four MCO substrates indicates specificity of LPR1 for ferrous iron. Because proteins of the secretory pathway are often N-glycosylated, we treated purified LPR1 with a mixture of deglycosylating enzymes. We did not obtain evidence for LPR1 glycosylation or phosphorylation (Extended Data Fig. 4b-d). Peptide sequencing by mass spectrometry of purified native LPR1 indicated absence of both types of posttranslational modifications; however, ectopic overexpression of LPR1 in leaves may have overwhelmed and masked detectable LPR1 protein modifications.
Local Pi sensing requires the Fe-binding site on LPR1
We previously derived a structural model of LPR1 based on its presumed function in Fe homeostasis and distant similarity to Fet3p, which suggested presence of an Fe2+-binding (Müller et al., 2015). We independently employed the YASARA package to identify by PSI-BLAST iterations and PDB searches high-scoring templates for LPR1 homology modeling. The five top-most ranking templates are all crystal structures of the spore-coat protein A (CotA), a bacterial (Bacillus subtilis) MCO laccase (Enguita et al., 2004). The refined LPR1 model reveals the hallmarks of MCO enzymes, i.e. the mononuclear T1 Cu site and the multinuclear T2/T3 Cu cluster, and it further supports the presence of an acidic triad (E269, D370, and D462) predicted to compose the Fe2+ binding site (Fig. 3a-c).
We generated by site-directed mutagenesis cDNAs for expressing in planta LPR1 variants with single or multiple amino acid substitutions in the presumed Fe2+-binding pocket (E269A, D370A, D462A, and combinations thereof) or proximal T1 Cu site (H464A, H568A, and C563A). LPR1 expression (immunoblot analysis) and specific ferroxidase activity were determined in extracts from transiently transfected tobacco (Nicotiana benthamiana) leaf discs. Expression of wild-type LPR1 (p35S::LPR1WT) resulted in the highest specific ferroxidase activity of all plasmids tested (Fig. 4a). The LPR1WT leaf extract exhibited typical Michaelis-Menten saturation kinetics for Fe2+ oxidation and revealed an apparent Km value of 3.6 μM (Fig. 4a, inset), which is similar to the value obtained with purified wild-type LPR1 (Fig. 2c).
Amino acid substitutions in the acidic triad of the predicted Fe2+ binding site impaired specific ferroxidase activity to varying extents. While LPR1D370A expression and specific activity were comparable to LPR1WT, leaf extracts expressing the LPR1D370A variant showed an almost 4-fold higher Km-value for Fe2+ (13.6 μM) when compared to LPR1WT leaf extracts (Fig. 4a). On the other hand, expression of LPR1E269A showed an ∼80% reduction of the LPR1WT ferroxidase activity measured above background, and expression of LPR1D462A revealed an almost complete loss of ferroxidase activity, which was similar to that of control transfections. Expression of LPR1 variants with multiple amino acid substitutions, LPR1E269A, D462A and LPR1E269A, D370A, D462A did not increase specific ferroxidase activity above the background level, which was also observed for leaf extracts transfected with plasmids encoding LPR1 variants with a disrupted mononuclear T1 Cu site, i.e. LPR1H464A, LPR1H568A, and LPR1C563A (Fig. 4a). However, the T1 Cu site mutant variants were noticeably less abundant, or undetectable (LPR1H464A), based on immunoblot analysis. Low protein abundance suggests LPR1 protein instability because Cu is a co-factor for both MCO activity and proper MCO protein folding.
To verify some of the results by in planta complementation, we generated lpr1 plants expressing LPR1WT, LPR1E269A, LPR1D370A and LPR1D462A variants under control of the pCaMV 35S promotor. We compared primary root extension of the transgenic lines to wild-type and pdr2, as well as to lpr1lpr2 and lpr1 seedlings upon transfer from Pi-replete conditions (5 d) to +Pi or –Pi agar medium for 4 d (Fig. 4b). While overexpression of LPR1 restored primary root growth inhibition of insensitive lpr1 seedlings on –Pi agar, overexpression of variants LPR1E269A and LPR1D462A did not significantly reduce the long root phenotype of lpr1 seedlings on low Pi medium. We noticed a weak, albeit poorly, complementing effect for the LPR1D370A variant (Fig. 4b). Thus, these data are largely consistent with the ferroxidase activity assays using tobacco leaves (Fig. 4a).
Using the LPR1 purification protocol, we prepared LPR1E269A, LPR1D370A and LPR1D462A variants for studying their kinetic parameters (Extended Data Fig. 6). We obtained low amounts of nearly pure LPR1E269A and LPR1D462A variants (detectable by immunoblot analysis), but we failed to prepare LPR1D370A. However, the final fractions of all LPR1 variant preparations did not display any ferroxidase activity, which points to a compromised Fe2+-binding pocket. In summary, our structure-function analysis of LPR1 firmly demonstrates a requirement of the predicted Fe2+-binding site for LPR1 ferroxidase activity (and possibly protein stability), as well as for LPR1-dependent Pi sensing by root meristems.
Fe availability tunes LPR1-dependent Pi sensing
Because LPR1 expression and LPR1 abundance do not noticeably respond to external Pi supply (Fig. 1), we tested in nutrient shift experiments whether LPR1 substrate availability governs local Pi sensing. We transferred Pi-replete seedlings (5-d-old) to +Pi or –Pi agar medium supplemented with increasing Fe concentrations (0-1000 μM) and monitored primary root extension for up to 4 d. Upon transfer to Fe-supplemented +Pi media, wild-type, lpr1lpr2, and pdr2 seedlings displayed a similar daily extension of the primary root, which was not greatly altered by the addition of up to 200 μM Fe. Higher Fe concentrations (500-1000 μM) strongly inhibited primary root growth irrespective of the genotype, which is likely caused by general Fe toxicity (Reyt et al., 2015; Zhang et al., 2018) (Extended Data Fig. 7).
Seedling transfer from +Pi+Fe control agar to –Pi medium without iron (–Pi –Fe) does not significantly inhibit primary root extension of all genotypes tested, which we previously reported (Müller et al., 2015). However, we observed striking genotype-dependent differences in root growth inhibition upon transfer to –Pi media supplemented with increasing Fe (Fig. 5). Intriguingly, wild-type roots displayed a triphasic growth response to increasing Fe supply (Fig. 5a). Low Fe concentrations (2.5-25 μM, phase I) strongly reduced primary root extension (by 60%), whereas intermediate Fe availability (50-100 μM, phase II) was less inhibitory (ca. 30%). Higher Fe supply (>100 μM, phase III) caused a gradual and pronounced inhibition of primary root growth, which was also observed on +Pi medium supplemented with high (>500 μM) Fe concentrations (Extended Data Fig. 7a). Interestingly, lpr1lpr2 seedlings did not display the triphasic Fe dose response in Pi deficiency (Fig. 5b). Primary root extension was insensitive to low Fe for up to 50 μM; however, lpr1lpr2 root growth was gradually inhibited by higher Fe supply (100-1000 μM) in a similar fashion as wild-type roots (Fig. 5a, b). On the other hand, Pi-deprived pdr2 seedlings displayed, as the wild-type, strong inhibition of primary root growth on gradually increasing, low Fe concentrations (2.5-25 μM). While root growth inhibition was maximal at 25 μM Fe (by 85%), higher concentrations (50-1000 μM Fe) neither rescued nor intensified pdr2 root growth inhibition on –Pi media (Fig. 5c, d).
It is important to point out that the apparent Km value of LPR1 (2-3 μM Fe2+) corresponds well with the first inhibition phase of primary root extension on low Fe (0-10 μM) under Pi limitation (Fig. 5e). This suggests that LPR1 ferroxidase activity and function in local Pi sensing are primarily determined by substrate availability. This proposition is supported by unaltered pLPR1 and pPDR2 promoter activities and by stable steady-state LPR1 and PDR2 mRNAs levels in wild-type root tips upon seedling transfer to –Pi agar supplemented with increasing (0-1000 μM) Fe concentration (Fig. 6a, Extended Data Fig. 8).
The PDR2-LPR1 module governs Fe re-distribution in Pi-deprived root meristems
We monitored by Perls/DAB (diaminobenzidine) staining the accumulation and distribution of labile iron (Fe3+) in primary root tips upon transfer of Pi-replete seedlings to various –Pi+Fe agar media (Fig. 6b). Tissues of the RAM that accumulate Fe in the apoplast upon Pi deprivation overlap with the cell-specific LPR1 expression domain (Müller et al., 2015). In Pi-starved wild-type roots, Fe progressively accumulated in the SCN with gradually increasing Fe supply. The intensity of Perls/DAB staining peaked at 10 μM Fe and steadily decreased with higher Fe concentrations (25-1000 μM). The RAM of insensitive lpr1lpr2 seedlings did not stain for iron above background, except for the highest Fe concentrations applied (500 μM and 1000 μM). Interestingly, the cell-type specificity and intensity of Fe staining were evidently similar between Pi-deprived wild-type and pdr2 root meristems for up to 10 μM Fe supply, while at higher Fe concentrations (25-1000 μM) intense Perls/DAB staining was observed in the entire pdr2 root tip (Fig. 6b). Importantly, upon transfer of Pi-replete wild-type, lpr1lpr2, and pdr2 seedlings to Fe-supplemented +Pi control media, gradually increasing low external Fe supply (0-10 µM) did not intensify Perls/DAB staining above background (Extended Data Fig. 9a). Moderate Fe supply (25-200 µM) appreciably increased Perls/DAB staining in the SCN and columella of wild-type and pdr2 root tips, but not in lpr1lpr2 root meristems, whereas excess Fe supply (500 µM and 1000 μM) caused Fe overload in root tips of all genotypes (Extended Data Fig. 9). Thus, Fe overload and root growth inhibition on excessively high Fe concentration are independent of LPR1 function.
Finally, we monitored callose formation by aniline blue staining in root meristems upon seedling transfer to Fe-supplemented –Pi media. Our data reveal that callose deposition in Pi-deprived root tips largely reflects Fe dose-dependent and genotype-specific patterns of Fe accumulation in root tips (Fig. 6c). The similar responses of Pi-deprived wild-type and pdr2 root meristems to low external Fe availability (0-10 μM) in terms of growth inhibitions (Fig. 5), Fe3+ accumulation (Fig. 6b), and callose deposition (Fig. 6c) are consistent with the conclusion that PDR2 function does not restrict LPR1 expression, biogenesis or ferroxidase activity. However, the unrestrained Fe3+ accumulation in pdr2 root tips at moderately elevated Fe availability (>10 μM) and high external Fe supply suggests a role for PDR2 in maintaining Fe homeostasis and regulating Fe pools in root meristems. This conclusion is supported by the analysis of LPR1-overexpression seedlings (OxL1) for iron-dependent root growth inhibition (Fig. 5d), along with the observed Fe3+ accumulation and callose deposition in Pi-deprived root tips (Fig. 6b, c). In both assays, the response of OxL1 roots to increasing external Fe supply mimics the response of pdr2 roots challenged by Pi limitation. The data suggest that root growth inhibition by gradually increasing low a Fe supply (2.5-25 μM) in –Pi condition is independent of the LPR1 protein level. At intermediate and high Fe concentrations (>25 μM), ectopic p35S::LPR1 expression shifts Fe3+ accumulation from the RAM to the columella and epidermis at agar contact sites, indicating that substrate availability determines LPR1 ferroxidase activity (Fig. 6b).
Progenitors to embryophytes acquired LPR1-type ferroxidase from soil bacteria
The substantial amino acid sequence similarity (37% identity) between LPR1 and bacterial CotA, which oxidizes bulky organic substrates such as ABTS or bilirubin (Enguita et al., 2004; Sakasegawa et al., 2006), prompted us to study the phylogenetic relationship between LPR1 and annotated MCO proteins (UniProt Database). We retrieved 189 MCO sequences and generated a midpoint-rooted phylogenetic tree featuring two major branches (Fig. 7). MCO group I, composed of two monophyletic clades, assorts fungal laccases including ferroxidases involved in Fe import (clade Ia), and plant laccases with ascorbate oxidases (clade Ib). Paraphyletic MCO group II includes bacterial, fungal, and mammalian MCO proteins of unknown specificities, or of presumed functions in N assimilation (Cu-dependent nitrite reductases), Fe export (ferroxidases), and hemostasis (blood coagulation factors). CotA and LPR1-like MCOs of Arabidopsis and rice (Ai et al., 2020) form a monophyletic clade within the bacterial paraphyletic segment of group II (Fig. 7).
Comparison of the primary and tertiary structures rationalizes the strikingly different substrate specificities of CotA and LPR1. The amino acid sequence alignment of CotA and LPR1-like MCOs indicates absence of a bona fide Fe2+-binding acidic triad in CotA (Extended Data Fig. 10). While the first and third acidic amino acid residues on LPR1 (E269 and D462) are embedded in two conserved segments, the second residue (D370) is located in a variable linker flanked by two hydrophobic motifs in LPR proteins (Fig. 3j). Although these features, with the exception of the three acidic residues, are conserved in CotA, its variable linker is shorter (aa 321-326) and likely folds into a tight surface loop, permitting access of bulky molecules (e.g., ABTS) to the substrate binding pocket (Fig. 3d,g, Extended Data Fig. 11a). The longer surface loop on LPR1 (aa 363-373) harbors D370 and may provide a flexible lid-like segment for high-affinity Fe2+ binding, in addition to possibly preventing access of bulky substrates (Fig. 3e,h). Notably, despite a similar architecture of the Fe2+-binding and electron-transfer sites near the T1 Cu center, the surface topology of the Fe2+-binding pocket differs between LPR1 and yeast Fet3p (Extended Data Fig. 11b,c).
Sequence similarity searches (NCBI nucleotide collection) in the eukaryal domain indicated restriction of LPR1-like proteins (presence of the acidic triad) to embryophytes (land plants). Unlike MCO laccases, which form large families in plants (Turlapati et al., 2011), bona fide LPR1-like MCO ferroxidases are often encoded by two genes in the bryophytes and tracheophytes (Extended Data Table 2). It is generally accepted that land plants evolved from a small but diverse group of green algae (the charophytes or streptophyte algae), which comprise a paraphyletic assemblage of five classes of mainly freshwater and terrestrial algae (Mesostigmatophyceae including Chlorokybophyceae, Klebsomidiophyceae, Charophyceae, Coleochaetophyceae, and Zygnematophyceae) (de Vries and Archibald, 2018; Furst-Jansen et al., 2020). Phylogenomic analyses increasingly favor the Zygnematophyceae, or alternatively a clade comprising the Zygnematophyceae and Coleochaetophyceae, as the sister group of land plants (Wickett et al., 2014; Zhong et al., 2014). We identified sequences coding for predicted LPR1-type MCOs in the recently published genomes of two Zygnematophyceae (Cheng et al., 2019), Mesotaenium endlicherianum and Spirogloea musicola, but not in the genomes of Mesostigma viride and Chlorokybus atmophyticus (Wang et al., 2020) (Extended Data Table 2). We employed a hidden Markov model (HMM) approach using a profile of full-length LPR1-like MCOs from 14 land plants to analyze data sets of the 1KP project (One Thousand Plant Transcriptomes Initiative 2019) (One Thousand Plant Transcriptomes, 2019). We did not identify LPR1-like sequences featuring an acidic triad among the rhodophytes, glaucophytes or chlorophyte algae; however, half of the analyzed charophyte and most of the bryophyte transcriptomes revealed such MCO sequences. Among the charophyte transcriptomes, 23 LPR1-like sequences are present in the Zygnematophyceae and one in the Coleochaetophyceae (Extended Data Table 3, Extended Data Fig. 14).
While the HMM approach supported restriction of LPR1-related proteins to the streptophytes (embryophytes plus charophytes), it pointed to a widespread occurrence of CotA-like MCOs (often annotated as bilirubin oxidases) in the bacterial and archaeal domains (Extended Data Table 3). Interestingly, if filtered for the presence of the LPR1-type acidic triad and variable linker sequence, we identified at least 35 bacterial LPR1-like MCOs with substantial amino acid sequence identity (35-40%) to Arabidopsis LPR1 (Extended Data Tables 2-4, Extended Data Fig. 12). We selected four such bacterial MCO sequences for homology modeling, which suggests presence and topology of a LPR1-type surface loop for Fe2+ binding (Fig. 3f,i, Extended Data Fig. 13a). Indeed, when expressed in Escherichia coli, recombinant LPR1-like MCOs from Streptomyces clavuligerus and Sulfurifustis variabilis show ferroxidase activity (Extended Data Fig. 13b-c).
Bacterial genera harboring the >35 LPR1-like ferroxidase genes are limited to five phyla (Extended Data Table 2), comprising so-called Terrabacteria (Firmicutes, Actinobacteria, Chloroflexi) as well as members of Bacteroidetes and Proteobacteria isolated from soil habitats (Battistuzzi et al., 2004; Battistuzzi and Hedges, 2009; Marin et al., 2017). A phylogenetic tree of DNA sequences coding for LPR1-like MCOs in bacteria, two streptophyte algae and selected embryphytes reveals a monophyletic clade of streptophyte sequences nested within the radiation of bacterial LPR1-like MCOs (Fig. 8a). Such a tree topology suggests a single horizontal gene transfer (HGT) event from a bacterial donor to a progenitor of the embryophytes, which is consistent with the exon-intron structure of LPR1-like genes (Fig. 8b). The number of introns increased from one (bryophytes) to three (tracheophytes) during evolution. All introns are in phase-0 and thought to partition the gene of bacterial origin into symmetric exons for maintaining its ancient functionality (Mayer et al., 2011; Husnik and McCutcheon, 2018). A cladogram including coding sequences of additional bryophytes and streptophyte algae (1 KP Project), which correspond to the internal polypeptide covering the entire acidic triad segment in Arabidopsis LPR1 (aa 264-465), supports the proposition that streptophyte ancestors of the embryophytes acquired LPR1-type ferroxidase from soil bacteria by HGT (Extended Data Fig. 14).
LPR1-type MCO ferroxidases emerged during bacterial land colonization
To explore the origin of LPR1-type ferroxidases, we searched, using the internal 202-aa LPR1 segment as query, the eubacterial and archaeal domains for putative MCO sequences harboring only partial acidic triad motifs. We additionally identified at least 80 such MCO sequences, which are absent in marine Hydrobacteria but limited to Terrabacteria, Proteobacteria, and Halobacteria (Extended Data Table 3). Most of the predicted MCOs (∼60) presumably lack the second conserved acidic residue (corresponding to D370 on LPR1) in their variable linker sequences, which however is expendable for high-affinity Fe3+-binding (Fig. 4). Predicted MCOs of the Firmicutes show the highest combinatorial variation of acidic triad motifs, suggesting that CotA-type laccases evolved from LPR1-type MCOs by linker contraction and sequential loss of acidic triad residues (Extended Data Table 4). A cladogram comprising bacterial MCO sequences related to LPR1 and CotA suggests that LPR1-type ferroxidases arose early during land colonization and were subject to lateral gene transfer among phyla of Bacteria and Archaea (Extended Data Fig. 15). Because many of the extant bacterial genera identified are known to conduct iron- or sulfur-based anoxygenic photosynthesis and/or to recycle its organic products by dissimilatory iron or sulfate reduction under anaerobic or microaerophilic conditions, LPR1-type MCOs with presumed functions in iron metabolism likely emerged prior to oxygenic photosynthesis by cyanobacteria and the ensuing Great Oxygenation Event (Weber et al., 2006; Sleep and Bird, 2008).
Discussion
Uneven Pi availability guides root development via local adjustment of root tip growth. Upon Pi limitation, root meristem maintenance is under genetic control of the PDR2-LPR1 module and depends on Fe co-occurrence (Svistoonoff et al., 2007; Ticconi et al., 2009; Müller et al., 2015). Here we show that LPR1, a principal determinant of local Pi sensing in A. thaliana, encodes a novel prototypical MCO ferroxidase of high substrate specificity (Fe2+) and affinity (Km ∼ 2 μM). Metal oxidases including ferroxidases related to Fet3p (limited to Fungi) or ceruloplasmin (limited to Animalia) form a major group within the ancient MCO superfamily whose diverse members are widely distributed in all domains of life (Janusz et al., 2020). Although LPR1 and Fet3p share similar catalytic parameters as well as an analogous architecture of the Fe2+-binding and adjoining T1 Cu site (Jones et al., 2020), our study suggests that LPR1-type MCOs displaying ferroxidase activity emerged very early during bacterial land colonization. LPR1-type ferroxidases (or their MCO progenitors) possibly crossed bacterial phyla multiple times to diversify by lateral gene transfer, which is widespread among soil bacteria (Ochman et al., 2000; Klumper et al., 2015). For example, in the genus Bacillus (Firmicutes), LPR1-type MCOs evolved to CotA-type laccases by progressive remodeling of the acidic triad segment. Gram-positive soil bacteria produce endospores, which are fortified with various spore coat proteins including CotA to survive in harsh environments (McKenney et al., 2013). While the precise biochemical function of CotA is not known, its unusually large substrate binding cavity and correspondingly short lid-like loop, which likely derived from an LPR1-type progenitor MCO (Extended Data Table 3, Extended Data Fig. 15), presents a unique structural feature among MCO laccases (Enguita et al., 2003).
The physical proximity between soil bacteria and the terrestrial/subaerial common ancestor of streptophytes likely facilitated our proposed single HGT event of LPR1-type ferroxidases, which occurred before the divergence of Zygnomatophyceae (possibly extending to Coleochaetophyceae) and embryophytes (∼580 mya) (Cheng et al., 2019). Our data corroborate the conjecture that plant terrestrialization was accelerated by substantial HGT from soil bacteria to early land plant progenitors (Yue et al., 2012; Husnik and McCutcheon, 2018; Cheng et al., 2019). Diversification of LPR1-type ferroxidases in land plants was possibly limited to rare whole-genome duplication events (Soltis and Soltis, 2016), which is suggested by the single retained LPR sister gene pair in A. thaliana, LPR1 (At1g23010) and LPR2 (At1g71040) (Abel et al., 2005). The unknown extent of lateral gene transfer among bacteria curtails precise identification of an HGT donor, probably explaining why LPR1-type ferroxidase sequences of four bacterial phyla (Proteobacteria, Chloroflexi, Actinobacteria, Firmicutes) are monophyletic with the streptophyte sequences (Fig. 8a, Extended Data Fig. 15). Nonetheless, the metabolic lifestyle and iron biochemistry of extant bacterial sister genera may allow insight into the function of plant LPR1-like ferroxidases. Members of the four phyla are facultative anaerobic or microaerophilic, spore-forming chemoorganotrophs, which are capable of dissimilatory or fermentative Fe3+ reduction and have been isolated from iron-rich soils or artificial Fe(III) oxide-enriched growth substrates (Lentini et al., 2012; Li et al., 2012; List et al., 2019; Wang et al., 2020). For example, genera of the Geobacteraceae family, including Geobacter and closely related Desulfuromonas species, are the predominant Fe3+ reducers in many anaerobic sediments. Such bacteria chemotactically locate extracellular Fe(III) oxide minerals and transfer electrons via nanowires to its solid-phase surfaces (Childers et al., 2002). Other microbial strategies for accessing insoluble Fe(III) oxides involve production of soluble external electron shuttles such as redox-active antibiotics or chelating ligands, which simultaneously increase Pi bioavailability in soil (Reguera et al., 2005; Weber et al., 2006; Liptzin and Silver, 2009; Glasser et al., 2017; Michelson et al., 2017; McRose and Newman, 2021) (Fig. 9). Although considered as strict anaerobes, Geobacter species can tolerate episodes of dioxygen exposure and code for ROS-scavenging proteins (Methe et al., 2003). Notably, G. metallireducens contains four genes encoding CotA-like proteins, the genes presumably having been acquired from Bacillus by lateral gene transfer (Berini et al., 2018). While two of these genes encode LPR1-type MCOs (see Extended Data Fig. 15), a fifth gene expresses a biochemically characterized ABTS-oxidizing MCO with a very low Km for dioxygen (<10 M) (Berini et al., 2018). If low Km (O2) values also apply to similar MCO enzymes, bacterial LPR1-type ferroxidases may promote Fe-redox cycling to protect against oxidative stress associated with Fe3+ reduction and resultant Fenton chemistry.
LPR1 and related plant ferroxidases likely facilitate analogous processes in root tips. Upon Pi limitation, the STOP1-ALMT1 module, a C2H2 zinc finger transcription factor (SENSITIVE TO PROTON RHIZOTOXICITY 1) and one of its direct target genes (ALUMINUM-ACTIVATED MALATE TRANSPORTER 1), activates malate release into the rhizosphere and apoplast of internal root tissues (Balzergue et al., 2017) (Fig. 9). Malate mobilizes Pi from insoluble metal complexes by Fe3+ chelation (Abel, 2017; Balzergue et al., 2017; Mora-Macias et al., 2017; Gutierrez-Alanis et al., 2018). Ensuing cell wall chemistry (e.g., Fe3+ reduction by ascorbate), augmented by photochemistry under laboratory conditions, generate Fe2+ and ROS (Grillet et al., 2014; Müller et al., 2015; Abel, 2017; Zheng et al., 2019). LPR1-dependent Fe redox cycling (Fe3+ re-formation and cell wall deposition) attenuates ROS production and presumably ROS signaling in the SCN. We propose that the constitutively expressed MCO ferroxidase, LPR1, senses subtle increases in Fe availability as a Pi-dependent cue to adjust root tip growth to Pi deprivation. Regulation of LPR1 activity by substrate availability is supported by our observation that PDR2/AtP5A counteracts LPR1 function by maintaining Fe homeostasis in root meristems (Fig. 9), which points to a novel role of single, ER-resident, orphan P5-type ATPases in plants (Sorensen et al., 2015; Naumann et al., 2019).
Methods
Plant lines and growth conditions
Arabidopsis thaliana accession Columbia (Col-0), Col-0 mutant lines pdr2-1, lpr1lpr2, and transgenic lines pCaMV 35S::LPR1 (p35S::LPR1) and pLPR1::eGFP-GUS were previously described (Svistoonoff et al., 2007; Ticconi et al., 2009; Müller et al., 2015). GATEWAY technology (Invitrogen) and Agrobacterium-mediated transformation were used to generate transgenic Arabidopsis lines expressing p35S::LPR1E269A; p35S::LPR1D370A and p35S::LPR1D462A. Seeds were surface-sterilized and germinated on 1% (w/v) Phyto-Agar (Duchefa) containing 2.5 mM KH2PO4, pH 5.6 (high Pi or +Pi medium) or no Pi supplement (low Pi or –Pi medium), 50 µM Fe3+-EDTA, 5 mM KNO3, 2 mM MgSO4, 2 mM Ca(NO3)2, 2.5 mM MES-KOH, pH 5.6, 70 μM H3BO3, 14 μM MnCl2, 10 μM NaCl, 0.5 µM CuSO4, 1 µM ZnSO4, 0.2 µM Na2MoO4, 0.01 µM CoCl2 and 5 g/l sucrose. The agar was routinely purified by repeated washing in deionized water and subsequent dialysis using DOWEX G-55 anion exchanger (Ticconi et al., 2009). ICP-MS analysis of the washed agar (7.3 μg/g Fe and 5.9 μg/g P) indicated a contribution of 1.3 μM Fe and 1.9 μM P to the solid 1% agar medium. For root length measurements, 27-54 seedlings were transferred to the indicated media and gain of primary root length was marked daily. Photos were analyzed using ImageJ software. Additional lateral root were induced as previously described (Himanen et al., 2002). Hydroponically grown seedlings were germinated under moderate shaking in 200-ml flasks containing 50 ml liquid +Pi medium.
Microscopy
Green fluorescent protein (GFP) fluorescence was visualized using a Zeiss LSM 780 confocal laser-scanning microscope (excitation 488 nm, emission 536 nm) in phosphate-buffered saline. Co-localization of GFP and PI (propidium iodide) was monitored in sequential mode (excitation 561 nm, emission 630 nm). Seedlings were incubated for 2 min in 0.1 mg/ml PI solution. For GUS (β-glucuronidase) staining, seedlings were incubated in 50 mM Na-phosphate (pH 7.2), 0.5 mM K3Fe(CN)6, 0.5 mM K4Fe(CN)6, 2 mM X-Gluc, 10 mM EDTA and 0,1% TritonX at 37°C and subsequently cleared using chloral hydrate solution (7:7:1 chloral hydrate:ddH2O:glycerol) as described before (Wong et al., 1996). Callose was stained for 1 h with 0.1% (w/v) aniline blue (AppliChem) in 100 mM Na-phosphate buffer (pH 7.2) and carefully washed twice. Fluorescence was visualized using a Zeiss LSM 880 confocal laser-scanning microscope (excitation 405 nm, emission 498 nm) in 100 mM Na-phosphate buffer (pH 7.2)(Müller et al., 2015). Histochemical iron staining (Perls/DAB) was performed as previously described (Müller et al., 2015) with minor changes to the protocol. Plants were incubated for 10 min in 4% (v/v) HCl, 4% (w/v) K-ferrocyanide (Perls staining), or K-ferricyanide (Turnbull staining). For DAB intensification, plants were washed twice (ddH2O) and incubated (15 min) in methanol containing 10 mM Na-azide and 0.3% (v/v) H2O2. After washing with 100 mM Na-phosphate buffer (pH 7.4), plants were incubated for 3 min in the same buffer containing 0.025% (w/v) DAB (Sigma-Aldrich) and 0.005% (v/v) H2O2. The reaction was stopped by washing 100 mM Na-phosphate buffer (pH 7.4) and optically clearing with chloral hydrate (1 g/ml, 15% glycerol).
Real-time quantitative PCR
Total RNA was prepared from excised root tips (tip growth gained after seedling transfer) by using the peqGOLD Plant RNA Kit (VWR). One biological replicate represents 40-60 pooled root tips. RNA samples were dsDNase treated (dsDNase, Thermo Scientific, EN0771) and quantified. cDNA was prepared using 1 µg total RNA, which was reverse transcribed by using oligo(dT) with a First-Strand cDNA Synthesis Kit (Fermentas) according to the manufacturer’s protocol. Quantitative real-time PCR was performed with first strand cDNA template on a QuantStudio 5 PCR System with Fast SYBR Green Mix (Applied Biosystems). The reported fold induction was analyzed by the ΔCt-method and normalized to the endogenous UBC9 (ubiquitin-conjugating enzyme 9) control. Gene-specific amplimers are listed in Supplementary Table 1.
Quantitative proteomics
Primary root tips were excised and collected into liquid nitrogen. Tissue lysis, sample preparation, protein labeling, Tandem-Mass-Tag (TMT) spectrometry, MS/MS data analysis, and TMT-quantifications were performed as recently described (Rodriguez et al., 2020; Stephani et al., 2020).
LPR1 homology modeling
YASARA 13.9 (Krieger et al., 2002; Krieger et al., 2009) was used to derive 25 homology models of LPR1 or bacterial MCOs, each based on five PDB (The Protein Data Bank) (Berman et al., 2000) templates (five X-ray structures of laccase CotA from B. subtilis: 2WSD, 2X88, 4AKO, 2X87, and 4AKP). Quality analysis with PROCHECK (Laskowski, 1993) and PROSA II (Sippl, 1990, 1993) identified the best fit for each protein. All Cu+ cations of the templates were adopted and merged into the models. The ferrous iron (Fe2+) was manually added in proximity to residues E269 and D370 of the highly conserved acidic triad on LPR1 or the respective conserved positions in the bacterial ferroxidase models. Subsequently, the model was refined by 20 cycles of simulated annealing refinement with the corresponding tool of YASARA. Molecular surfaces were created with the modeling program MOE (Molecular Operating Environment v2019.0101, Chemical Computing Group Inc., Montreal, QC, Canada, 2019).
Site-directed mutagenesis
For introducing point mutations into plasmid DNAs, site-directed mutagenesis was carried out with the Quick Change II Site-directed mutagenesis Kit (Agilent) according to the manufacturer’s instructions. Briefly, two complementary primers containing the desired mutation of the plasmid were used to amplify two overlapping, complementary strands of the plasmid with staggered nicks. After amplification, the parental DNA was digested with Dpn I and the mutated plasmids were transformed into E. coli Top 10 or XL1 Blue cells. The primers used for generating the different mutations are listed in Supplementary Table 1.
Purification of native LPR1 protein variants
Transgenic A. thaliana lpr1 mutant plants expressing p35S::LPR1, p35S::LPR1E269A; p35S::LPR1D370A and p35S::LPR1D462A were grown for 8 weeks on soil in short-day conditions (8 h light, 16 h darkness, 21°C). Entire plant rosettes were harvested and homogenized in liquid nitrogen. To extract whole proteins, 15 g plant material was vortexed in 40 ml buffer A (50 mM Tris-Cl pH 6.8, 100 mM NaCl, 0.5 mM EDTA, 10% glycerol) containing 1 mM PMSF and 1× protease inhibitor (ROCHE) followed by incubation for 30 min at 4°C (shaking). After clearance of the extract by centrifugation (500 × g, 30 min, 4°C), the supernatant was subjected to 40% (NH4)2SO4 precipitation (1 h at 4°C). The resulting pellet (4,500 × g, 45 min, 4°C) was discarded and the supernatant treated with 80% (NH4)2SO4 for 1 h at 4°C. The resulting protein pellet was solubilized in 2-3 ml buffer A, loaded on a HighLoad Superdex 200 gel filtration column (HL 16/60, GE Healthcare), and eluted with buffer A as the mobile phase. Fractions containing LPR1 (detected by immune blot analysis) were directly applied to cation exchange carboxymethyl-sepharose column (HiTrap CM FF, 1-ml, GE Healthcare) equilibrated with buffer B (20 mM Na2HPO4-NaH2PO4, pH 7). Fractions were eluted using a linear salt gradient (0-1 M NaCl) in buffer B. LPR1 eluted at 350 mM NaCl. LPR1-containing fractions were stored at −20°C until further use and the enzyme was stable for two weeks. LPR1 abundance and activity were confirmed by immunoblot analysis and ferroxidase assays, respectively. For protein silver staining, the gels were incubated twice for 20 min each or overnight in fixing solution (10% acetic acid, 40% methanol). Subsequently, the gels were incubated in 30% methanol, 1.2 mM NaS2O3, 829 mM Na-acetate for 30 min, followed by three washing steps (5 min each) in distilled H2O. Silver staining was performed by incubating the gels in an aqueous AgNO3 solution (2 mg/ml) for 20 min followed by two washing steps with water. The gels were developed (staining of protein bands) in 236 mM NaCO3 containing 0.04% formaldehyde, and the reaction was stopped by incubation in 40 mM Na-EDTA.
Deglycosylation and phosphatase treatments
Purified LPR1 protein was analyzed using Protein Deglycosylation Mix II (New England Biolabs) according to the manufacturer’s instructions. Fetuin was used as a control. Phosphorylation of purified LPR1 protein was tested according to (Maldonado-Bonilla et al., 2014) using λ-protein phosphatase (New England Biolabs). In brief, root material was harvested in phosphatase buffer supplemented with 1 × protease inhibitor (ROCHE). After addition of 1,200 U phosphatase, reactions were carried out for 90 min at room temperature. Samples were inactivated at 95°C for 5 min and analyzed by immunoblotting.
Peptide sequencing
Proteins were in-gel digested with trypsin and further processed as previously described (Majovsky et al., 2014). Dried peptides were dissolved (5% acetonitrile/0.1% trifluoric acid), injected into an EASY-nLC 1000 liquid chromatography system (Thermo Fisher Scientific), and separated by reverse-phase (C18) chromatography. Eluted peptides were electro-sprayed on-line into a QExactive Plus mass spectrometer (Thermo Fisher Scientific). A full MS survey scan was carried out with chromatographic peak width. MS/MS peptide sequencing was performed using a Top10 DDA scan strategy with HCD fragmentation. MS scans with mass to charge ratios (m/z) between 400 and 1300 and MS/MS scans were acquired. Peptides and proteins were identified using the Mascot software v2.5.0 (Matrix Science) linked to Proteome Discoverer v 2.1 (Thermo Fisher Scientific). A precursor ion mass error of 5 ppm and a fragment ion mass error of 0.02 Da were tolerated in searches of the TAIR10 database amended with common contaminants.
Carbamidomethylation of cysteine was set as fixed modification and oxidation of methionine was tolerated as a variable modification. A spectrum (PSM), peptide and protein level false discovery rate (FDR) was calculated for all annotated PSMs, peptide groups and proteins based on the target-decoy database model and the percolator module. PSMs, peptide groups and proteins with q-values beneath the significance threshold α=0.01 were considered identified.
Immunoblot analysis
Polyclonal LPR1 epitope-specific antibodies were raised in rabbits against a mixture of two synthetic peptides (peptide I: 175-PKWTKTTLHYENKQQ-189; peptide II: 222-VESPFQLPTGDEF-234) and affinity-purified (EUROGENTEC, Seraing, Belgium). Total proteins were extracted from frozen plant material in buffer A (50 mM Tris-HCl, pH 6.8, 100 mM NaCl, 0.5 mM EDTA, 10% glycerol) containing 1 × protease inhibitor (ROCHE). After centrifugation (20,000 × g, 10 min, 4 °C), the protein concentration of the supernatant was determined (2D-Quant, GE Healthcare), and proteins were separated by SDS/PAGE on 8-10% (w/v polyacrylamide) gels and transferred to PVDF membranes (Semi-Dry-Blot, GE Healthcare). After transfer, membranes were exposed to blocking buffer (1× TBS, 0.05% w/v Tween, 3% w/v milk powder) at room temperature for 1 h or overnight. To detect LPR1, affinity-purified, peptide-specific anti-LPR1 antibody was used 1:1000 in blocking buffer for 1 h at room temperature or at 4 °C overnight. Horseradish-peroxidase-conjugated goat anti-rabbit IgG (BioRad, 1:5000) was chosen as a secondary antibody, and the ECL Select or Prime Western Blotting Detection Reagent (Thermo Fisher) was used for visualization. The epitope-specific anti-LPR1 antibody detects 100 ng purified, native LPR1 protein and recognizes only one major protein of ca. 70 kDa in extracts of the p35S::LPR1 overexpression line (Extended Data Fig. 2). Plant specific actin-antibody (Sigma-Aldrich) was used (at a dilution of 1:2000) as loading control. To control for recombinant bacterial MCO expression anti-His-HRP (Miltenyi Biotec) was used (at a dilution of 1:10000) in the blocking buffer.
Ferroxidase and other MCO assays
Protein concentration was determined using Qubit Fluorometric Quantification System (Thermo Fisher) according to the manufacturer’s instructions. All reagents except human ceruloplasmin (Athens Research) were purchased from Sigma-Aldrich. Ferroxidase activity was determined as previously described (Müller et al., 2015) using typically 25 µM Fe(NH4)2(SO4)2 × 6H2O as the substrate and 3-(2-pyridyl)-5,6-bis(2-[5-furylsulfonic acid])-1,2,4-triazine (ferrozine) as a specific Fe2+ chelator to scavenge the remaining substrate after the reactions. The rate of Fe2+ oxidation was calculated from the decreased absorbance at 560 nm using a molar extinction coefficient of ε560=25,400 M-1 cm-1 for the Fe2+-ferrozine complex(Hoopes and Dean, 2004). Phenol oxidase (laccase) activity with ABTS (2,2’-azino-bis(3-ethylbenzothiazoline-6-sulfonic acid), ascorbate oxidase activity, and bilirubin activity were measured in 0.1 M NaH2PO4-Na2HPO4 (pH 5.6 – 7.2) as described (Johannes and Majcherczyk, 2000; Sakasegawa et al., 2006; Peng et al., 2015).
Transient expression assays
The transient transformation of Nicotiana benthamiana leaves was carried out using Agrobacterium tumefaciens strains that carried the indicated plasmids and the pCB301-p19 helper plasmid (Burstenbinder et al., 2013). Bacteria were grown overnight to an OD600 = 0.5 – 0.8, harvested (10,000 × g, 4 min, 4 °C) and washed two times with 2 ml transformation buffer (10 mM MES-KOH, pH 5.5, 10 mM MgCl2, 150 µg/ml acetosyringone) and subsequently dissolved in transformation buffer to an OD600 of 1. The bacteria carrying the expression construct were mixed 1:1 with the ones harboring the pCB301-p19 plasmid and incubated for 1 h at 20 °C. Subsequently, the bacteria were injected at the bottom side of leafs of 5-7 week-old plants. Samples were harvested 4 d post infiltration.
Expression of bacterial MCO proteins
Genes encoding potential bacterial ferroxidases from Sulfurifustis variabilis (BAU47383.1) and Streptomyces clavuligerus (QCS10718.1) were codon optimized (Supplementary Table 2), synthesized at the Invitrogen GeneArt Gene Synthesis platform, and cloned into pVp16-Dest vector for IPTG (isopropyl-β-thiogalactopyranoside)-induced expression (3 h at 37°C) in E.coli strain ArcticExpress (Agilent). After sonication, the cleared cell lysates were directly used for ferroxidase activity assays.
Phylogenetic analyses
For MCO sequence alignment and phylogenetic analysis, 193 referenced protein sequences of the annotated “multicopper oxidase family” were obtained from uniProt Knowledgebase (www.uniprot.org) and filtered for fragments. CotA (P07788) was added to the dataset. All phylogenetic trees were calculated by sequence alignment using MAFFT 7 (Katoh et al., 2002) with default settings and created at the CIPRES web-portal with RAxML 8.2.10 (Stamatakis et al., 2008) for maximum likelihood analyses using the JTT PAM matrix for amino acid substitutions in RAxML.
Data base searches
Full-length protein sequences related to Arabidopsis LPR1 (At1g23010) were obtained for select land plant species from NCBI (tblastn searches). To identify LPR1-related genes in the main taxonomic groups of bacteria, archaea, early eukaryotes and basal plants, blastp (version 2.10.1+) (Camacho et al., 2009) and hmmer (version 3.3) (Madera and Gough, 2002) were used (www.hmmer.org) (Camacho et al., 2009; Finn et al., 2011). For bacteria and archaea, representative genomes of the key taxonomic groups were obtained from NCBI assembly. Transcriptome data from the 1KP project (One Thousand Plant Transcriptomes Initiative 2019) (One Thousand Plant Transcriptomes, 2019) were taken for early eukaryotes and basal plant species. The Arabidopsis LPR1 protein sequence was used as a query to search each genome or transcriptome. Only hits with an alignment length of >175 amino acids were considered. In addition, a profile hidden Markov models approach was applied, generating an HMM profile for LPR1-releated sequences from 14 higher plant species for scanning each genome or transcriptome. All hits were scanned for the presence of a LPR1-type Fe2+-binding site, which is composed of an acidic triad with the following three consensus sequence motifs: 1. [WVI]XP[EA][YAF]X[GA]; 2. N[DTS][AG]XXP[YF]PXG[DE]X(5-10)[VI][ML]XF; and 3. NXTX[DEG]XHP. For final validation, candidate sequences were individually aligned with Arabidopsis LPR1 and visually inspected. If applicable, representative contiguous sequences covering all acid triad signature motifs were used as query to interrogate (tblastn searches at NCBI) each bacterial phylum for LPR1-type sequences with incomplete acid triad signatures.
Statistical Analyses
Statistical differences were assessed by one-way ANOVA and Tukey’s HSD posthoc test, using built-in functions of the statistical environment R (R Development Core Team, 2018). Different letters in graphs denote statistical differences at P < 0.05. Graphs were generated using the ggplot2 R package.
Author contributions
C.N. and S.A. conceived and designed the experiments; C.N. and M.H. conducted the major experiments and analyzed the data; C.A., N.T., A.T.N., J.Z. and S.A. contributed to additional experiments or data analysis; W.B. performed the homology modeling; R.I., K.M., Y.D., and W.H. conducted and advised the proteomics analyses; P.J. and M.Q. conducted and advised the phylogenomic analyses; G.S. provided conceptual insight and edited the article; C.N. and S.A. wrote the article.
Acknowledgements
We thank M. Ried for insightful and critical reading of the manuscript, T. Desnos for initial discussions, T. Toev for generating plant lines, G. Durnberger and M. Schutzbier, Vienna BioCenter, for expert technical assistance, and members of the department for suggestions. This work was supported by institutional core funding (Leibniz Association) from the Federal Republic of Germany and the State of Saxony-Anhalt to S.A. and by an EMBO short-term fellowship to C.N. Work in the K.M. group was financially supported by the EPIC-XS, project number 823839, the Horizon 2020 Program of the European Union, and the ERA-CAPS I 3686 project of the Austrian Science Fund.