A novel cis-element enabled bacterial uptake by plant cells
Abstract:
Nitrogen is essential for plant growth and development. A wide phylogenetic variety of land plants ranging from mosses and gymnosperms to angiosperms have evolved symbioses with nitrogen-fixing bacteria that convert atmospheric nitrogen into ammonium. For example, the fern Azolla maintains colonies of nitrogen-fixing cyanobacteria in specialized apoplastic cavities, outside the plant cell wall enclosure. A major biological breakthrough was the evolution of the nitrogen-fixing root nodule symbiosis (RNS) characterized by the intracellular accommodation of bacteria in lateral organs ('nodules') formed on roots. The occurrence of the RNS is restricted to a monophyletic clade, encompassing four angiosperm orders: the Fabales, Fagales, Rosales and Cucurbitales (FaFaCuRo). Because of this phylogenetic restriction and scattered occurrence of RNS within the FaFaCuRo, Soltis and colleagues postulated that the last common ancestor of the FaFaCuRo clade acquired a genetic change, a 'predisposition', which enabled members of this clade to subsequently evolve RNS multiple times independently. The intracellular accommodation of bacteria and root nodule development are two genetically separable and, to this extent, independent features of RNS, which is therefore genetically possible that they did evolve sequentially and not at the same time. The phylogenetic diversity of bacterial symbionts plus the variation of nodule anatomy and development across the RNS-competent FaFaCuRo species, together with the gap of 30 million years between the last common ancestor and the oldest fossil root nodules in this clade, further fuelled the hypothesis that nodule organogenesis evolved several times independently and was not a feature of the last common ancestor. The recent discovery of multiple losses of RNS within the FaFaCuRo clade has initiated a discussion about whether this genetic change in the common ancestor was perhaps sufficient for the formation of RNS. Nonetheless, the precise nature of this key event in the evolution of nodulation has remained a mystery for more than two decades. We asked which evolutionary acquisitions by the last common ancestor, in the form of novel traits and the underlying genetic causes, enabled the evolution of the RNS. From a phylogenetic perspective, such acquisitions should be: (1) exclusively present in the FaFaCuRo clade and absent outside of this clade and (2) conserved throughout the FaFaCuRo clade or at least maintained in RNS-competent (hereafter called 'nodulating') species. The uptake of bacteria into living plant cells is, with one exception (Gunnera), phylogenetically restricted to the FaFaCuRo clade. The uptake of bacteria requires the localized lysis of the plant cell wall, which threatens cell integrity because of the turgor pressure imposed by the protoplast. A systematic comparison of features associated with the RNS across the entire FaFaCuRo clade pinpoints a single unique and shared trait—the uptake of bacteria into living plant cells with intracellular physical support structures—that fulfils both above-mentioned criteria to be acquired by the common ancestor. These structures come in a diversity of shapes (infection threads (ITs) and infection pegs) and in at least two different cell types (epidermal and cortical) but are all characterized by the apposition of matrix material, which is thought to maintain cell integrity during the localized lysis of the plant cell wall. Although this matrix material is a common feature of all analysed successful bacteria uptake events in FaFaCuRo species, only one type, cortical ITs, can be found in almost all nodulating species. Cortical IT formation is an evolutionary breakthrough because it allowed clonal selection of bacteria, specific control of nutrient exchange and increased nitrogen fixation efficiency. By contrast, in Gunnera, cell integrity is maintained by physical closure of a multicellular cavity by extracellular matrix material. This difference, together with the phylogenetic distance of Gunnera from the FaFaCuRo clade, suggests an independent origin of bacterial uptake in this genus. To search for gene gains specific for the FaFaCuRo clade, a genome-wide comparative phylogenomic analysis was performed; however, not a single gene following the aforementioned evolutionary pattern was identified. Here, we tested the hypothesis that the 'predisposition' event involved gain of novel cis-regulatory elements. Changes in gene regulation can be important drivers of functional and morphological evolution. Emergence or loss of even a single cis-regulatory element can lead to dramatic phenotypic consequences, for example, novel organ formation. Phylogeny has dated the common ancestor of the FaFaCuRo clade to approximately 104 million years ago. A long-standing hypothesis states that the evolution of RNS involved co-opting genes from the arbuscular mycorrhiza (AM) symbiosis, which can be traced back to the earliest land plant fossils 410 million years ago. This hypothesis is underpinned by similarities in intracellular accommodation structures and the common requirement of both symbioses for a set of so-called common symbiosis genes that are conserved across land plant species able to form AM and encode symbiotic signal transduction and intracellular restructuring machineries. Results Discovery of PACE The transcription factor-encoding Nodule Inception (NIN) gene is positioned at the top of an RNS-specific transcriptional regulatory cascade and is indispensable for RNS. The promoter of NIN is a potential physical target for such a co-option event, because it defines the molecular interface between common symbiotic signal transduction and the specific transcriptional networks underlying RNS development. We therefore compared the NIN-promoter sequences of 37 angiosperm species including 27 FaFaCuRo members and identified only one motif fulfilling the aforementioned criteria, which we called Predisposition Associated cis-regulatory Element (PACE). The phylogenetic distribution of PACE was further investigated in an expanded search comprising 163 plant species in the promoter of NIN and the entire NIN-like protein (NLP) gene family, including NLP1 from which NIN diverged at the base of the eudicots. PACE was found in all nodulating FaFaCuRo members and four non-nodulating species that have lost RNS but maintained NIN. Importantly, PACE was absent from all the NLP promoters analysed. Thus, within the NIN-like gene family, the phylogenetic distribution of PACE is NIN- and FaFaCuRo-clade specific and is consistent with a model in which PACE was acquired by the NIN promoter of the last common FaFaCuRo ancestor. Intriguingly, the 29-nucleotide-long PACE encompassed and extended beyond the previously identified binding site of the transcription factor Cyclops, which is encoded by a common symbiosis gene required for the development of both AM and RNS. Left: schematic illustration of the phylogenetic relationships between species inside (light-red shade) and outside (light-grey shade) the FaFaCuRo clade and the presence (+) and absence (-) pattern of RNS, NIN and PACE. Centre: PACE sequence alignment of the displayed species, in which grey shadings indicate more than 50% sequence identity. On top of the alignment, the PACE consensus sequence is depicted as a position weight matrix calculated from the displayed RNS-competent species. Right: graphical illustration of how PACE connected NIN to symbiotic transcriptional regulation by CCaMK–Cyclops, enabling IT development in the root cortex. This acquisition coincided with the predisposition event. X and Y represent hypothetical proteins binding to sequences flanking the Cyclops binding site. Given this clade-specific distribution of PACE, we searched for conserved motifs in the promoter sequences of two genes encoding transcriptional regulators, ERF Required for Nodulation 1 (ERN1) and Reduced Arbuscular Mycorrhiza 1 (RAM1) that are also known Cyclops targets. We identified motifs within the promoters of both, ERN1 and RAM1, encompassing the previously identified Cyclops binding sites. In sharp contrast to PACE, their presence extended beyond the FaFaCuRo clade. We tested the functional relevance of these distinct phylogenetic distribution patterns in transcriptional activation assays in Nicotiana benthamiana leaf cells. Transactivation by Cyclops was restricted to NIN promoters from FaFaCuRo species but extended to non-FaFaCuRo species for RAM1 promoters. Importantly, PACE was necessary and sufficient for the activation of the NIN promoter by Cyclops. Together with the exclusive occurrence of PACE in the NIN promoter of the FaFaCuRo clade, these results are in line with the hypothesis that the mechanistic link between Cyclops and the NIN promoter was established in the last common ancestor of this clade. PACE drives the expression of NIN during IT development in the cortex NIN is indispensable for IT development and its precise spatiotemporal expression is essential for this process. Because cis-regulatory elements are master determinants of gene expression patterns, we investigated the effect of PACE on the expression of NIN in physical relation to the bacterial uptake and accommodation stages during nodule development. We used the model legume Lotus japonicus in combination with its compatible nitrogen-fixing bacterium Mesorhizobium loti as experimental system. The process by which L. japonicus promotes the intracellular colonization by and accommodation of M. loti can be subdivided into successive stages: (1) entrapment of bacteria in a pocket formed by a curled root hair, (2) uptake of bacteria into a developing IT within that root hair, (3) IT progression into and through the outer cortical cell layers, (4) IT branching and extension within the nodule primordium and (5) release of bacteria from ITs into plant membrane-enclosed organelle-like structures called symbiosomes leading to (6) mature nodules characterized by infected cells densely packed with symbiosomes and the pink colour of leghemoglobin. To determine the PACE-mediated spatiotemporal expression domain, we introduced a GUS reporter gene driven by PACE fused to a region comprising the NIN minimal promoter and the 5' untranslation regions (UTR) into L. japonicus wild-type roots. The roots were subsequently inoculated with M. loti MAFF 303099 expressing DsRed (M. loti DsRed) facilitating detection of the bacteria through their fluorescence signal in root hairs and nodules. The NIN minimal promoter did not mediate reporter gene expression at any stage of bacterial infection. Intriguingly, the earliest detectable GUS activity mediated by PACE:NINminpro:GUS was clearly restricted to a zone in the nodule primordia that roughly correlated with the site of bacterial infection (indicated by a local accumulation of DsRed signal) and later expanded to the entire central tissue of the nodule. PACE-driven reporter expression was neither detected in root hairs harbouring ITs nor in nodules in which cells from the central tissue were filled with symbiosomes. Importantly, PACE-mediated expression was distinct from that mediated by the LjNIN 3 kb promoter (NINpro) or the NINpro with PACE mutated or deleted (NINpro::mPACE and NINpro::∆PACE, respectively) that conferred reporter expression across the central tissue of the nodule. We concluded on the basis of these observations that the PACE-mediated expression domain is temporally and spatially restricted and possibly accompanies the development of bacterial accommodation structures in the nodule. To further resolve this relationship between PACE-driven gene expression and bacterial accommodation at the cellular level, we compared—simultaneously in the same tissue—the progression of bacterial infection with the expression pattern mediated by PACE fused to the NIN minimal promoter (PACE:NINminpro) and by a NIN promoter with mutated PACE (NINpro::mPACE). A red and a yellow fluorescent protein (mCherry and YFP, respectively) targeted to the nucleus by fusion to a nuclear localization signal (NLS) were used as reporters. The resulting promoter:reporter fusions (PACE:NINminpro:NLS-mCherry and NINpro::mPACE:NLS-YFP) were placed in tandem on the same transfer-DNA (T-DNA) allowing a nucleus-by-nucleus comparison of their relative expression. This T-DNA construct was introduced into L. japonicus wild-type roots that were subsequently inoculated with M. loti R7A expressing the cyan fluorescent protein (CFP) or with M. loti MAFF 303099 expressing the green fluorescent protein (GFP) to facilitate detection. During the first stages of bacterial invasion (stages 2 to 3), PACE-mediated mCherry was expressed specifically in cortical cells carrying ITs and in directly adjacent cells. By contrast, the NINpro::mPACE-driven YFP signal was not detected in those cells. In sections of developing nodules, in which infection had progressed to stage 3 or 4, PACE-mediated mCherry was expressed specifically in a—hereafter called 'IT zone'—comprising cortical cells and primordium cells that carried ITs and in some, but not all, directly adjacent cells. Intriguingly, the expression domains marked by mCherry and YFP fluorescence were distinct from each other: whereas the PACE-driven mCherry signal was consistently marking the IT zone, the NINpro::mPACE-driven YFP signal was observed in primordium cells surrounding this zone. The thin (approximately 1–2-cell-thick) border between the two domains was characterized by nuclei emitting both YFP and mCherry signals. In so-marked cells, ITs were typically not detected. The expression pattern mediated by the NIN promoter (containing PACE) was congruent with the sum of both promoter fragments. On the basis of these clearly distinct and complementary reporter expression domains governed by PACE versus the remaining promoter, we concluded that (1) PACE directs NIN expression to a specific IT zone and that (2) the NIN promoter comprises cis-regulatory elements that drive expression outside the PACE territory that is in root hairs (together with PACE), non-infected cortical and primordium cells and nodule cells filled with symbiosomes. These additional cis-regulatory elements might be addressed by other transcription factors that have been reported to bind to this promoter. These transcription factors might be counteracted by, for example, repression in the IT zone. Mutational dissection of PACE reveals a quantitative effect of sequences flanking the CYC-box on IT development To test the relevance and specific role of PACE in nodule and IT development, we performed complementation experiments using plants homozygous for the nin-2 or nin-15 mutant alleles. The nin-2 mutant allele harbours a frameshift mutation of the NIN gene, leading to a NIN loss-of-function phenotype, which is absence of both IT formation and nodule organogenesis. The nin-15 mutant allele carries a Lotus Retrotransposon 1 insertion within the NIN promoter 143 bp 3' of PACE. We examined the restoration of bacterial infection 21 days post inoculation (dpi) with M. loti DsRed by quantifying the number of root hairs harbouring ITs and the number of infected nodules. Nodule development in the legume Medicago truncatula is dependent on NIN expression mediated by a regulatory region containing several putative cytokinin responsive elements (CE). In L. japonicus, a similar CE region is positioned 45 kb upstream of the NIN transcriptional start site. To enable transgenic complementation experiments, we synthetically fused a 1 kb or 5 kb region encompassing this distant CE to the 5' end of a 3-kb NIN promoter. The NIN gene driven by these promoters (CE1kb:NINpro:NIN and CE5kb:NINpro:NIN) restored the formation of root hair ITs on 78% and 95% and infected nodules on 40% and 88% of nin-2 transgenic root systems, respectively. Importantly, this complementation success relied on the presence of PACE. nin-2 roots transformed with the same fusion design but carrying a mutation of PACE (CE1kb:NINpro::mPACE:NIN and CE5kb:NINpro::mPACE:NIN) did not restore root hair ITs; however, nodule formation was not impaired when using the cytokinin element-containing region of 5 kb (CE5kb:NINpro::mPACE:NIN). We concluded that PACE is indispensable for bacterial infection but not for nodule development. The 29-bp-long PACE sequence revealed by MEME encompasses and extends beyond the previously identified Cyclops binding site (CYC-box). Its degree of conservation may be interpreted as a trace of an ancestral PACE version present in the last common ancestor of the FaFaCuRo clade. Within PACE, the CYC-box is surrounded by less conserved flanking sequences. To dissect the specific contributions of the CYC-box and PACE sequences flanking the CYC-box ('flanking') to PACE function, we mutated the box and the flanking sequences independently (CE:NINpro::mbox:NIN and CE:NINpro::mflanking:NIN, respectively). Mutation of the CYC-box abolished root hair ITs. Interestingly, mutation of the flanking sequences led to a 50% reduction of the number of transgenic root systems carrying infected nodules, whereas the formation of root hair ITs was not impaired. This mutational dissection revealed two separable functions of PACE: whereas the PACE–Cyclops connection is essential for IT development, the flanking sequences significantly promote bacterial infection during nodule development and possibly act as binding sites for additional, yet undefined, transcription factors. Our data suggest that PACE comprises synergistic binding sites for both Cyclops and cooperating transcription factors. We conclude that the high level of conservation of the CYC-box is a consequence of the indispensable nature of this cis-element for the progression of the IT through the cortex. The higher level of diversification of sequences flanking the CYC-box might be a consequence of changes in transcription factors occupancy over evolutionary time scales. Considering this scenario, it is possible that such flanking sequence-occupying transcription factors are not conserved throughout the entire FaFaCuRo clade. PACE-mediated NIN expression defined an infection zone in the nodule cortex. To genetically separate the initiation of nodule development from IT formation and thereby enable a focused analysis of the role of PACE in cortical IT formation, we utilized the nin-15 mutant, which is impaired in IT formation but retains the capacity to form nodules. Most of these nodules were uninfected (92% and 86% plants carrying no root hair ITs and no infected nodules, respectively), and cortical cells filled with symbiosomes were never observed. This mutant therefore provided an ideal background to study the role of PACE in cortical IT formation, circumventing the negative epistatic effect of the inability of nin loss-of-function mutants to initiate cell divisions. PACE insertion into the tomato NIN promoter confers RNS capability To artificially recapitulate the functional consequence of PACE acquisition into a non-FaFaCuRo NIN promoter, we chose tomato (Solanum lycopersicum) which belongs to the Solanaceae, a family phylogenetically distant from the FaFaCuRo clade. Consistent with the absence of PACE, a GUS reporter gene driven by the tomato NIN promoter (S. lycopersicum NIN promoter (
