Aromatic amino acid biosynthesis impacts root hair development and symbiotic associations in Lotus japonicus

Abstract Legume roots can be symbiotically colonized by arbuscular mycorrhizal (AM) fungi and nitrogen-fixing bacteria. In Lotus japonicus, the latter occurs intracellularly by the cognate rhizobial partner Mesorhizobium loti or intercellularly with the Agrobacterium pusense strain IRBG74. Although these symbiotic programs show distinctive cellular and transcriptome signatures, some molecular components are shared. In this study, we demonstrate that 3-deoxy-d-arabino-heptulosonate 7-phosphate synthase 1 (DAHPS1), the first enzyme in the biosynthetic pathway of aromatic amino acids (AAAs), plays a critical role in root hair development and for AM and rhizobial symbioses in Lotus. Two homozygous DAHPS1 mutants (dahps1-1 and dahps1-2) showed drastic alterations in root hair morphology, associated with alterations in cell wall dynamics and a progressive disruption of the actin cytoskeleton. The altered root hair structure was prevented by pharmacological and genetic complementation. dahps1-1 and dahps1-2 showed significant reductions in rhizobial infection (intracellular and intercellular) and nodule organogenesis and a delay in AM colonization. RNAseq analysis of dahps1-2 roots suggested that these phenotypes are associated with downregulation of several cell wall–related genes, and with an attenuated signaling response. Interestingly, the dahps1 mutants showed no detectable pleiotropic effects, suggesting a more selective recruitment of this gene in certain biological processes. This work provides robust evidence linking AAA metabolism to root hair development and successful symbiotic associations.


Introduction
Legume roots establish interactions with beneficial microorganisms in the rhizosphere through complex chemical dialogs. Arbuscular mycorrhizal symbiosis (AMS) and root nodule symbiosis (RNS) are 2 well-known examples, which provide phosphorus and nitrogen sources, respectively, to the host (Oldroyd 2013). In the model legume Lotus (Lotus japonicus), 2 modalities of root colonization exist in RNS, depending on the microorganism partner (Montiel et al. 2021;Quilbé et al. 2022). The cognate rhizobial partner Mesorhizobium loti synthesizes and releases signaling molecules called lipochito-oligosaccharide nodulation factors (NFs), after perception of flavonoid compounds secreted by roots (Bek et al. 2010;Shimamura et al. 2022). Recognition of these compounds activates the symbiotic signaling pathway that allows rhizobial infection and nodule development. M. loti attaches to the root hair tip, inducing it to curl, thereby trapping the bacteria within infection pockets, where a tubular structure called an infection thread (IT) is formed. In this process, the rhizobial partner colonizes the root cell layers intracellularly, through root hair ITs, which advance toward the cortex, where a nodule primordium, generated by reactivation of the cell division in the root cortex, is formed. The bacteria are released from the ITs into the nodule cells as organelle-like structures, called symbiosomes, differentiating into nitrogen-fixing bacteroids (Roy et al. 2020). However, it is estimated that approximately 25% of all legume genera employ an alternative modality of rhizobial colonization, called intercellular infection (Sprent 2007). Our group recently showed that Lotus can also be infected intercellularly by the Agrobacterium pusense strain IRBG74 that was originally isolated from the flooding-tolerant legume Sesbania cannabina (Cummings et al. 2009;Mitra et al. 2016). In this entry mode, IRBG74 induces massive root hair curling and twisting but no root hair ITs are formed; instead, the bacteria pass between the epidermal root cells, forming cortical infection pockets. The migration proceeds both intra-and intercellularly into the nodule cells, releasing the bacteria from transcellular ITs or intercellular infection structures (Montiel et al. 2021). In the Lotus-IRBG74 symbiosis, root hair ITs are not formed but the bacteria are attached to the root hairs, which probably favors the nodulation program.
Strigolactones secreted by the roots are perceived by AM fungi, promoting spore germination and hyphal branching. These responses favor the physical interaction of the fungal hyphae with the root epidermal cells, giving rise to the formation of the hyphopodium (Bonfante and Genre 2010). The fungal hyphae penetrate the epidermal cells, reaching the cortex, where they differentiate into branched hyphae called arbuscules, where carbon and phosphorous sources are exchanged with the plant host (MacLean et al. 2017). Molecular genetic studies have identified important players involved in the signaling pathway that allow the mutualistic associations. In Lotus, crucial components of the symbiotic route comprise plasma membrane receptors, transcription factors, and a range of proteins that orchestrate the colonization of rhizobia through the root cell layers (Roy et al. 2020). The intercellular colonization of IRBG74 in Lotus roots shows evident cellular and transcriptome differences, with distinct genetic requirements with respect to the root hair IT formation process. However, there is a core of symbiotic genes that are indispensable for any modality of rhizobial infection (Montiel et al. 2021;Quilbé et al. 2022), which belong to the common symbiotic signaling pathway (CSSP) and also play essential roles in AMS (Stracke et al. 2002;Lévy et al. 2004;Parniske 2008;Yano et al. 2008).
The root hairs are extensions of specialized epidermal cells with polarized growth and a tubular shape, which increase the surface area for nutrient acquisition. In Arabidopsis (Arabidopsis thaliana), the characterization of a large collection of mutants affected during root hair development and emergence indicates that these processes are regulated by a sophisticated network that encompasses various molecular functions (Shibata and Sugimoto 2019). Root hairs also play a crucial role during RNS, in the early signaling pathway and in the intracellular colonization of rhizobia (Downie 2014). Actin cytoskeleton rearrangements and cell wall modifications in the root hairs are necessary to facilitate the formation and progression of ITs (Cárdenas et al. 2003;Yokota et al. 2009;Hossain et al. 2012;Zepeda et al. 2014;Qiu et al. 2015;Su et al. 2023). Several regulators of root hair development also participate in the establishment of symbiotic associations with rhizobia and AM fungi (Karas et al. 2005;Lei et al. 2015;Ke et al. 2016;Liu et al. 2020;Karas et al. 2021).
The functioning of the nitrogen-fixing nodule and arbuscules is regulated by a large array of metabolic genes. In these symbiotic organs, several enzymes of diverse metabolic pathways modulate the flux of carbon, phosphorous, and nitrogen sources between the root cells and the microsymbionts (Udvardi and Poole 2013). However, transcriptome analyses of legume roots inoculated with compatible rhizobia and AM fungi show that reprograming of metabolic processes also occurs at early stages of the symbiotic associations (Manthey et al. 2004;Deguchi et al. 2007;Handa et al. 2015;Fonseca-García et al. 2021). Despite the evident activation of several metabolic routes during the colonization and organogenesis programs in the mutualistic interactions of legumes, their role has been poorly explored. In this study, we found that expression of a Lotus 3-deoxy-d-arabino-heptulosonate 7-phosphate synthase (LjDAHPS) gene was associated with several stages of RNS and root hair development. DAHPS1 encodes the first enzyme of the shikimate pathway, which leads to the biosynthesis of the aromatic amino acids (AAAs) phenylalanine, tyrosine, and tryptophan (Lemaître et al. 2014). These amino acids serve as precursors of various metabolites, including phytohormones and cell wall components (Tzin and Galili 2010;Geng et al. 2020;Simpson et al. 2021). The characterization of 2 homozygous mutant alleles disrupted in LjDAHPS1 revealed a pivotal role of this gene in root hair development and in the establishment of symbiotic associations with nitrogen-fixing bacteria and AM fungi.
of Lotus roots (Montiel et al. 2021). Among the regulated genes, LotjaGi1g1v0143000 encoding a 3-deoxy-d-arabinoheptulosonate 7-phosphate synthase (hereafter referred as DAHPS1) was the most abundant transcript ( Fig. 1A and Supplemental Table S1). DAHPS1 is the first enzyme of the shikimate pathway (Supplemental Fig. S1A) involved in the biosynthesis of AAAs (Tzin and Galili 2010), which suggests an unexpected role for amino acid homeostasis in RNS. Supporting this notion, DAHPS1 is highly expressed in NF-treated root hairs and nodule primordia (Supplemental Fig. S1C). A further investigation of the gene in the Lotus database (https://lotus.au.dk/) shows that DAHPS1 contains 5 exons and 4 introns (Fig. 1B). To determine whether DAHPS1 belongs to a gene family, a BLAST was performed in the Lotus genome (Kamal et al. 2020). Two homologous genes were identified, LotjaGi1g1v0239000 and LotjaGi6g1v0351200, which were named LjDAHPS2 and LjDAHPS3, respectively (Fig. 1B). Unlike LjDAHPS1, LjDAHPS2 and LjDAHPS3 were not evidently induced by IRBG74 at any timepoint studied (Supplemental Fig. S1B) and showed lower expression levels in root hairs, nodule primordia, and mycorrhized roots (Supplemental Fig. S1C). The deduced amino acid sequence of DAHPS1 encodes a protein of 539 amino acids, with a predicted plastid signal peptide (Supplemental Fig. S2). Similarly, DAHPS2 and DAHPS3 contain an N-terminal plastid-targeting sequence. The protein sequence of these 3 Lotus enzymes is highly conserved (Supplemental Fig. S2), sharing >80% identities. To investigate the phylogeny of Lotus DAHPS, a phylogenetic tree was constructed using full-length protein sequences (Supplemental Table S2). In this analysis, the Lotus DAHPS proteins were compared with homologous sequences found in the legumes Aeschynomene evenia, chickpea (Cicer arietinum), liquorice (Glycyrrhiza uralensis), Medicago truncatula, and common bean (Phaseolus vulgaris). In addition, their phylogenetic was the most abundant transcript among the upregulated genes in Gifu roots at 5 dpi with IRBG74. The violin boxplots were generated with data calculated from RNAseq information (Montiel et al. 2021). Violin boxplots: center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, individual data points. The P-adjust values are indicated below the violin boxplots. B) Graphic representation of the exon (green box) and intron (gray line) composition in the LjDAHPS1 (LotjaGi1gv0143000), LjDAHPS2 (LotjaGi1g1v0239000), and LjDAHPS3 (LotjaGi6g1v0351200) genes. Blue arrowheads indicate the retrotransposon insertions in the dahps1-1 (30141487) and dahps1-2 (30100225) alleles. C) Phylogenetic relationship of DAHPS in various legumes and nonlegumes (monocots and dicots). The chlorophyte (Volvox carteri) and embryophyte (Marchantia polymorpha) DAHPS were included as roots. Bootstrap values are shown at the nodes. distribution was evaluated with their counterparts in several nonlegumes (monocots and dicots) with fully sequenced genomes, i.e. rice (Oryza sativa), sorghum (Sorghum bicolor), A. thaliana, tomato (Solanum lycopersicum), and potato (Solanum tuberosum). The DAHPS family was composed of 4 members in the monocots O. sativa and S. bicolor, while in the dicots, it ranged between 2 and 4 members (Fig. 1C). Particularly, legumes of the inverted repeat-lacking clade (IRLC) contained only 2 DAHPS homologs, except for G. uralensis that possesses 3 (Fig. 1C). The different DAHPSs clustered in subclades that reflect the phylogeny of the species.

DAHPS1 promoter activity in different root tissues and during RNS
Transcriptome profile sources indicate that DAHPS1 is expressed in roots and potentially plays a role in RNS of Lotus. To explore the cellular expression pattern, we monitored the activity of the DAHPS1 2-kb promoter sequence fused to the triple yellow fluorescent protein (YFP) with a nuclear localization signal (pDAHPS1::tYFP-nls). Strong expression of the fluorescent reporter was detected in the apical region of the root, growing root hairs, and emerging lateral root primordia of uninoculated roots, comprising different root tissues (Fig. 2, A to C). Likewise, after M. loti-DsRed and IRBG74-DsRed inoculation, pDAHPS1::tYFP-nls was clearly detected in the epidermal and cortical cells adjacent to intracellular and intercellular infection sites, respectively (Fig. 2, D to F). An elevated promoter activity remained during intercellular colonization and nodule organogenesis (Fig. 2, G and H), which indicates the participation of DAHPS1 during nodule development.

LjDAHPS1 is dispensable for shoot and root growth but crucial during root hair development
To further investigate the potential role of DAHPS1 in the symbiotic associations of Lotus, 2 homozygous mutant alleles, dahps1-1 and dahps1-2, were obtained from the LORE1 mutant collection (Urbanski et al. 2012;Małolepszy et al. 2016) and these contained retrotransposon insertions in the third and first exon, respectively (Fig. 1B). The evidence obtained on the expression profile of DAHPS1 led us to first explore the root phenotype in the dahps1-1 and dahps1-2 mutants. Both the primary root length and the root apical meristem (RAM) length of plants at 10 d post germination (dpg) were not significantly different in the 2 mutant alleles, when compared to those of wild-type Gifu (hereafter referred as Gifu) plants (Supplemental Fig. S3, A to C). Similarly, the root growth dynamics in the dahps1-1 and dahps1-2 mutants was comparable to that recorded for Gifu roots, at least during a period of 1 to 11 dpg (Supplemental Fig. S3D). Likewise, no significant differences were detected in the shoot length in Gifu and dahps1 plants at 5 wk post germination (wpg), grown in nitrogen-replete conditions (Supplemental Fig. S3, E and F). However, the root hair development was dramatically altered in the dahps1-1 and dahps1-2 mutants, compared to Gifu (Fig. 3,  A and B). Although the root hairs seem to emerge normally Figure 2. Detection of DAHPS1 promoter activity in various Lotus root tissues during RNS. DAHPS1 promoter activity was visualized by confocal microscopy in uninoculated (A to C) and inoculated (D to F) Lotus transgenic roots of comparable developmental stages harboring the pDAHPS1::tYFP-nls construct (2 kb of the DAHPS1 promoter sequence fused to the triple YFP with a nuclear localization signal). tYFP-nls was clearly expressed at the root tip A), root hairs B), and emerging lateral root primordium C). Strong promoter activity in cells surrounding the root hair IT D), the intercellular infection E), and nodule organogenesis at 1 wpi with M. loti-DsRed D) and 2 wpi with IRBG74-DsRed E to H). Arrow and arrowheads indicate the IT and intercellular infection, respectively. Scale, 50 µm D, E), 100 µm B, C), and 200 µm A, F).
in the differentiation zone of the dahps1-1 and dahps1-2 mutants with a typical tubular shape, they underwent a progressive swelling, while developing and aging, forming balloon-like structures that eventually collapsed (Fig. 3C). Unlike the root hairs of Gifu plants that reached up to 1 mm in length, the root hairs of dahps1-1 and dahps1-2 mutants barely reached 400 µm, reflecting the drastic impact of their abnormal development (Fig. 3, A to D). Considering that the mutant and wild-type roots grew at similar growth rates and that changes in root hair length began to be detected at 1.5 to 2 mm from the root tip (Fig. 3D), the data suggest that root hair apical growth in the mutants was arrested soon after root hair bulge formation. This phenotype is consistent with the results obtained on the promoter activity of DAHPS1, linking the regulation of AAA biosynthesis to root hair development.

Actin cytoskeleton and cell wall integrity are compromised in root hairs of the DAHPS1 mutants
The drastic alteration in the root hair morphology of the dahps1-1 and dahps1-2 mutants prompted us to study this phenomenon in more detail by analyzing their actin cytoskeleton dynamics. The actin microfilaments, visualized by Alexa-Phalloidin staining, formed long bundles parallel to the long axis of Gifu root hairs (Fig. 4A). Similarly, these structures were observed in the root hairs of the dahps1-1 and dahps1-2 mutants with an early and mild swelling (Fig. 4, B, C, and F). However, the progressive alterations in the root . Altered morphology and growth profile of root hairs in the dahps1-1 and dahps1-2 mutants. Representative images of the roots in Gifu A) and the dahps1-1 mutant B) (the same phenotype was observed for the dahps1-2 mutant) at 10 dpg. C) Sequential images of the progressive deterioration of root hair structure along the root axis of a dahps1-1 mutant at 10 dpg (bottom to top). Scale 1 mm A, B) and 100 µm C). D) Violin boxplots of the maximal root hair length recorded at different distances from the root tip in Gifu, dahps1-1, and dahps1-2 roots at 10 dpg. Student's t-test of root hair length between Gifu (n = 17) and the dahps1-1 (n = 27) and dahps1-2 (n = 25) mutants. *P < 0.05; ***P < 0.001. Violin boxplots: center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, individual data points. hair morphology were accompanied by a gradual disruption of the actin cytoskeleton ( Fig. 4, D to F). A similar scenario was detected in the epidermal and cortical cells, where a sophisticated network of long and abundant bundles was formed in Gifu, while in the dahps1 mutants, the actin cytoskeleton was composed of fewer and short bundles (Supplemental Fig. S4, A to C). To further understand the dramatic changes in the cytoskeleton organization, the actin polymerization sites were monitored in growing root hairs incubated with fluorescently labeled cytochalasin D (Cyt-Fl). In Gifu root hairs, the major fluorescent signal was detected in the apical region, reflecting the fastgrowing (plus) ends of microfilaments (Fig. 4G). In deformed root hairs of the dahps1-1 and dahps1-2 mutants, the fluorescence was considerably reduced in the apical region, and interestingly, sites of actin polymerization were associated with subapical regions (Fig. 4H). This displacement of actin polymerization sites seems to be linked to the misorientation of actin bundles (Fig. 4F) and would explain the root hair phenotype.
The root hair growth and morphology is also regulated by changes in the cell wall dynamics. To assess the composition of the root hair cell wall in the dahps1 mutants, the roots were incubated with propidium iodide (PI), a fluorescent dye that binds the demethoxylated pectin in the hardened cell walls of living root hairs (Rounds et al. 2011). In Gifu, the PI fluorescence was predominantly associated with the root hair shanks, but almost absent at the tip, where the cell wall is more extensible (Fig. 4I). By contrast, in the dahps1-2 mutant, the dye was detected in the periphery of the swollen root hairs (Fig. 4J) and even localized within intracellular structures (Fig. 4J). Our results suggest that the progressive loss of normal root hair morphology in the dahps1-2 mutants is caused by perturbations in the growth dynamics and organization of actin microfilaments, along with changes in the distribution of cell wall components. . Disruption of actin cytoskeleton organization and cell wall dynamics in the root hairs of dahps1 mutants. Imaging of the actin microfilament organization in Gifu A), dahps1-1 B to E), and dahps1-2 F) root hairs stained with Alexa-Phalloidin at 5 dpg. Visualization of the actin polymerization sites in living root hairs of Gifu G) and the dahps1-1 H) mutant, with fluorescently labeled cytochalasin D at 5 dpg. The arrows indicate the polymerization sites at the root tip and subapical region of root hairs in Gifu and dahps1-1, respectively. Visualization of demethoxylated pectins at the cell wall of Gifu I) and dahps1-2 J) root hairs by PI staining at 5 dpg. Bright field and fluorescence images are shown in the left and right panels, respectively. Scale, 20 µm.

Pharmacological and genetic restoration of root hair morphology
DAHPS1 encodes the first enzyme in the biosynthetic pathway of chorismate, a precursor of AAAs. To evaluate its predicted enzymatic function, a heterologous complementation approach was conducted. The Escherichia coli mutant strain NT1402 (Jayaraman et al. 2022), which lacks all 3 cognate DAHPS enzymes, was transformed with the LjDAHPS1 coding sequence (CDS). NT1402 and the complemented strains, NT1402-LjDAHPS1_1 and NT1402-LjDAHPS1_2, grew very well in media supplemented with shikimate and AAAs with casamino acids, which contains a mixture of essential amino acids except for tryptophan (Supplemental Fig.  S5A). In the absence of casamino acids, the growth of the different mutants was drastically perturbed, but in the plates where phenylalanine and tyrosine were present without tryptophan, only microcolonies developed in the NT1402 mutant, while in the complemented mutants, the growth was substantially increased (Supplemental Fig. S5B). This result supports the predicted role of LjDAHPS1 as a relevant component in the AAA biosynthetic pathway.
To corroborate if the altered root hair morphology and growth in the dahps1-1 and dahps1-2 plants were linked to a deficiency of AAA levels, the root hair development was analyzed in 10 dpg seedlings of the different genotypes, germinated and grown on MS medium supplemented with an equimolar concentration of phenylalanine, tryptophan, and tyrosine. Root hairs with a restored tubular shape were observed in the 2 mutant alleles grown with a 50 and 100 µM mixture of AAAs (Fig. 5,A and B,and Supplemental Fig. S6A). Interestingly, this effect was only present in the root tissue that was in contact with the agar that contained the cocktail of amino acids (Fig. 5B). To elucidate whether phenylalanine, tryptophan, or tyrosine was separately contributing to prevent the altered root hair morphology and growth in the dahps1-1 and dahps1-2 mutants, the individual amino acids were added to the growth medium at 50 and 100 µM. The root hair swelling was prevented in both mutant alleles grown in medium supplemented with each amino acid, although to different extents (Supplemental Fig.  S6A). The percentage of plants with restored tubular root hair structure was 68% to 86% with tyrosine, 36% to 77% with phenylalanine, and 38% to 61% with tryptophan ( Fig. 5C). As described earlier, besides the swelling, the apical root hair growth is also arrested in the DAHPS1 mutants. In this regard, the maximal root hair length in the adl1-1 and dahps1-2 mutants was not restored to the levels in Gifu by adding different combinations and concentrations of AAAs to the medium (Supplemental Fig. S6B). This result suggests that the root hair structure and growth are independently impacted by AAAs. Interestingly, the apical root hair growth of Gifu plants grown on media supplemented with AAAs was also affected (Supplemental Fig. S6, A and B), reinforcing the notion that apical root hair growth is influenced by AAAs.
Using a chemical approach, we found that an imbalance in AAA levels was linked to the root hair phenotype in the dahps1-1 and dahps1-2 mutants. To demonstrate that this deficiency was caused by the disruption of the LjDAHPS1 gene, we made a construct composed of its CDS fused to the YFP transcriptionally regulated by the native promoter (pDAHPS1::DAHPS1-YFP). The root hair morphology and development were restored in dahps1-2 transgenic roots expressing the pDAHPS1::DAHPS1-YFP construct (Fig. 5D). Additionally, a clear fluorescent signal of the reporter was detected throughout root hair development (Supplemental Fig. S6C), supporting our previous findings. The fluorescence was associated with plastid-like structures that match with the presence of a plastid signal peptide in the predicted amino acid sequence of DAHPS1.

Gene expression network associated with DAHPS1
The evidence collected in this study demonstrates that disruption of DAHPS1, a gene involved in the biosynthetic pathway of AAAs, has drastic consequences on root hair development. The data suggest that an insufficiency of these amino acids likely impacts the expression profile of genes and biological processes intrinsically connected to these molecules. We addressed this hypothesis by analyzing the transcriptome of dahps1-2 roots by RNAseq in 5-dpg seedlings. Compared with the root expression profile in Gifu plants of the same age, a total of 416 differentially expressed genes (DEGs; P-adjust < 0.5 and Log2FC ≥ 2) were detected (Supplemental Fig. S7A and Table S3). From this list, 163 sequences were upregulated and 253 were repressed in dahps1-2. Nineteen downregulated genes were related to the regulation of cell wall biomechanics. Interestingly, the Cystathionine-β-Synthase-like1 (CBS) and Vapyrin genes, crucial for rhizobial infection in M. truncatula (Murray et al. 2011;Sinharoy et al. 2016), were upregulated in the dahps1-2 roots compared to Gifu (Supplemental Fig. S7B). Similarly, the expression of Nodule inception (NIN), a master transcriptional regulator of rhizobial colonization and nodule organogenesis (Schauser et al. 1999), was induced in the uninoculated dahps1-2 roots (Supplemental Fig. S7B).
We previously described that DAHPS1 was induced in Gifu roots at 5 dpi with IRBG74, when the major transcriptome response is observed in the intercellular infection (DEG Log2FC ≥ 2). To explore the contribution of DAHPS 1 to the symbiotic signaling in the Lotus-IRBG74 interaction, the expression pattern of dahps1-2 roots was analyzed at 5 dpi with IRBG74. Using low stringency parameters (P-adjust < 0.5), we detected 2,840 DEGs in the inoculated Gifu roots and only 1,475 in dahps1-2 (Supplemental Fig. S7C and Table S4). However, the number of DEGs with Log2FC ≥ 2 was somewhat similar, 177 and 201 in dahps1-2 and Gifu, respectively (Fig. 6A). Interestingly, less than 40% of these sequences overlapped (72) and none of the downregulated genes overlapped (Fig. 6, B and C). Despite the low similarity in the expression profiles between Gifu and dahps1-2 roots (Fig. 6, A to C), a core set of early symbiotic genes displayed comparable expression levels after IRBG74 inoculation (Fig. 6D)

DAHPS1 contributes to different modalities of rhizobial infection and nodule organogenesis
The RNAseq analysis in the dahps1-2 roots inoculated with IRBG74 indicates that disruption of LjDAHPS1 affects the symbiotic transcriptome response. To evaluate the relevance of this finding, a nodulation kinetics analysis was conducted on plates for the dahps1-1 and dahps1-2 mutants, recording the number of pink nodules and the total number of nodules at 1 to 6 wpi with M. loti and IRBG74. Nodule formation (pink and total) was significantly reduced in both mutant alleles at most of the timepoints analyzed, compared to Gifu (Fig. 7, A to D, and Supplemental Fig. S8, A and B). The delayed and reduced nodulation of dahps1-1 and dahps1-2 mutants had a negative influence on plant growth, since the length of the aerial parts was significantly shorter compared to that of Gifu at 6 wpi with any of the inocula used (Supplemental Fig. S8, C and D). These findings reveal that DAHPS1 makes an important contribution to the nodulation capacity of Lotus. Since the root hair and nodulation phenotypes were similar in the 2 mutant alleles, we decided to perform a detailed symbiotic characterization only for the dahps1-2 allele.
First, the intracellular invasion of M. loti-DsRed was monitored on Gifu and dahps1-2 roots by confocal microscopy.
Root hair ITs were observed both in Gifu and in dahps1-2 at 1 wpi. In the latter, the infection events only occurred in root hairs with mild morphological alterations; however, the typical root hair curling was not detected; instead, the IT initiation took place in a subapical region (Fig. 7, E and F). Moreover, at this timepoint, the number of epidermal and cortical ITs per root was significantly lower in the dahps1-2 mutant compared to Gifu (Fig. 7G), which might be an indirect consequence of the defect in root hair development. Since the intercellular infection of IRBG74 in Lotus is technically unsuitable for a quantitative evaluation through microscopy techniques, we followed a root endosphere approach; the NodA gene abundance was estimated by qPCR in DNA samples extracted from Lotus roots at 3 wpi with IRBG74 (Montiel et al. 2021). This analysis indicated that dahps1-2 roots had a significant reduction of approximately 50% in the relative abundance of the IRBG74-nodA gene, compared to Gifu (Fig. 7H). These approaches revealed that both intra-and intercellular infections were negatively impacted in the dahps1-2 mutant; however, the effect on the intracellular colonization was more dramatic, with a 70% reduction.
Taken together, these results show that an efficient intraand intercellular symbiotic infection in Lotus depends on the function of DAHPS1. These findings prompted us to investigate the bacteroid colonization in the nodules formed by M. loti and IRBG74 in the dahps1-2 mutant. Infected nodule cells were detected in histological slides, stained with toluidine blue, of 3-wk-old nodules inoculated with M. loti and Figure 6. Attenuated gene expression response in dahps1-2 roots after IRBG74 inoculation. Heatmap A) and Venn diagrams B, C) show a reduced number of DEGs in dahps1-2 roots compared to Gifu at 5 dpi with IRBG74, but several early symbiotic genes were similarly induced D) (P-adjust < 0.5). The values were obtained from RNAseq analysis and are calculated with respect to uninoculated roots of equivalent age on the same genotype. IRBG74 in Gifu (Fig. 8, A and E) and dahps1-2 (Fig. 8, B and F). In Gifu nodules, the infected cells were densely packed with M. loti (Fig. 8A) and IRBG74 (Fig. 8E) bacteroids. By contrast, the infected cells in dahps1-2 nodules were vacuolized with both inocula, showing a deficient filling of the cells by bacteroids (Fig. 8, B and F). Based on these findings, we decided to investigate the symbiosome structure by transmission electron microscopy (TEM). In Gifu nodules colonized by M. loti, the infected cells contained transcellular ITs with round-shaped symbiosomes hosting 1 to 2 bacteroids (Fig. 8C). Although the nodules in dahps1-2 also contained 1 or 2 M. loti bacteroids per symbiosome, these organelle-like structures had irregular shapes (Fig. 8D). These abnormalities were also detected in the dahps1-2 nodules formed by IRBG74, which were accompanied by an evident disruption in the space of the symbiosome (Fig. 8H), contrasting with the integrity of this structure observed in Gifu (Fig. 8G). Altogether, these analyses thus showed that DAHPS1 contributes to rhizobial infection and nodule organogenesis.

AM colonization is delayed in dahps1-2 roots
The expression profile of DAHPS1 during RNS reflects its relevant role in this mutualistic association. Likewise, the data collected from the L. japonicus gene expression atlas (LjGEA) show that DAHPS1 is highly expressed in mycorrhized roots (Supplemental Fig. S1B). This prompted us to evaluate the AM phenotype of dahps1-2. For this purpose, 5-dpg seedlings of Gifu and dahps1-2 were inoculated with Rhizophagus intraradices spores in Magenta boxes. Root fragments of Gifu and dahps1-2 roots were collected at 4 wpi and stained with WGA-Alexa Fluor to visualize AM colonization by confocal microscopy. Fully branched hyphae with arbuscules within the cortical cells were observed in both genotypes, but the rate of arbuscule formation was apparently lower in the dahps1-2 mutant compared to Gifu (Fig. 9, A   Figure 7. Nodulation phenotype of dahps1-1 and dahps1-2 mutants after M. loti and IRBG74 inoculation. Total number of nodules recorded on Gifu (n ≥ 29), dahps1-1 (n ≥ 43), and dahps1-2 (n ≥ 29) at 1 to 6 wpi with M. loti A) and IRBG74 B). Mann-Whitney U test of the total number of nodules (asterisks below the violin graphs indicate significant difference: *P < 0.05; **P < 0.01; ***P< 0.001) between Gifu and the DAHPS1 mutant alleles. Representative images of nodules formed on different Lotus genotypes at 3 wpi with M. loti C) and IRBG74 D). Scale, 1 mm. Root hair infection in Gifu E) and dahps1-2 F) visualized by confocal microscopy at 1 wpi with M. loti-DsRed. The arrows indicate the ITs. Scale bar, 20 µm. G) Number of root hair ITs found in the epidermis and cortex at 1 wpi with M. loti-DsRed on Gifu (n = 19) and dahps1-2 (n = 19). H) Abundance of IRBG74-nodA by qPCR in genomic DNA isolated from Gifu (n = 10) and dahps1-2 (n = 9) roots at 3 wpi with IRBG74 and normalized to the LotjaGi1g1v0152000 gene accumulation. Student's t-test of nodA abundance between roots of Gifu and the mutants. *P < 0.05 and ***P < 0.001. Violin boxplots: center line, median; box limits, upper and lower quartiles; whiskers, 1.5× interquartile range; points, individual data points. and B). Consequently, we conducted a quantitative analysis of different fungal colonization structures by optical microscopy in root segments stained with Trypan blue at 4 and 6 wpi (Trouvelot et al. 1986). Gifu and dahps1-2 roots showed a similar percentage of mycorrhizal frequency at 4 and 6 wpi, a parameter that reflects the presence of AM fungi in the roots (Fig. 9C). However, other indicators that reveal the degree of fungal colonization were significantly reduced at 4 wpi in dahps1-2: cortex mycorrhizal intensity, intensity of mycorrhiza in the root fragments, arbuscule abundance in mycorrhizal parts of root fragments, and arbuscule abundance in the root system (Fig. 9C). These results demonstrate a delay in the establishment of AMS in dahps1-2 plants, since the values obtained for all these parameters were similar to those in Gifu at 6 wpi (Fig. 9C). These results suggest that DAHPS1 is required at the early stages of AM root colonization.

Root hair development is AAA dependent
Root hair biology is relevant due to the role played by these tubular extensions of epidermal cells on nutrient acquisition and its particularly polarized growth. The study of mutants defective in root hair emergence and growth shows that this process is highly complex and dynamic, involving transcription factors, phytohormones, cytoskeleton rearrangement, cell wall modifications, and secondary messengers (Shibata and Sugimoto 2019). Our study demonstrates that perturbance of AAA biosynthesis in the dahps1-1 and dahps1-2 mutants influences the mechanism of root hair development. The observed progressive alteration in the root hair morphology of dahps1-1 and dahps1-2 mutants was reverted by genetic complementation with the DAHPS1 sequence, the first enzyme in the biosynthesis of AAAs.
Besides the genetic complementation, we demonstrated that root hair swelling in dahps1-1 and dahps1-2 can be partially restored in 80% to 90% of the plants by adding a mixture of AAAs in a dose-dependent manner. This result indicates that in the DAHPS1 mutant alleles, the root hair phenotype was caused by the lack or deficiency of these aromatic molecules or their synthesis, and not by the absence of the intermediate compounds produced in the shikimate pathway. Interestingly, the tubular shape of root hairs was recovered in 68% to 86% of the mutants grown in a medium supplemented only with 100 µM of tyrosine. This finding suggests that the root hair swelling in the dahps1-1 and dahps1-2 mutants was mostly provoked by the absence of tyrosine or its derived compounds. Phosphorylation of tyrosine residues occurs in actin-related proteins (Guillen et al. 1999), affecting the cytoskeleton dynamics in diverse biological processes such as plant bending and pollen tube growth (Kameyama et al. 2000;Zi et al. 2007). We observed that both root hair morphology and the actin cytoskeleton progressively deteriorated in dahps1-1 and dahps1-2. These effects were apparently preceded by changes in the distribution of F-actin plus ends. In healthy growing root hairs, the F-actin plus ends are localized at the root tip, paving the way for its growth (Zepeda et al. 2014). By contrast, these structures were detected in subapical regions of the root hairs in the dahps1-1 and dahps1-2 mutants. The critical role played by the actin cytoskeleton during root hair development has been further supported by genetic evidence. The maintenance of the tip growth is affected in the root hairs of the Arabidopsis mutant deformed root hairs 1, affected in the major actin of the vegetative tissue (Ringli et al. 2002). In Lotus, the actin cytoskeleton is severely compromised in the actin-related protein component (arpc1), nap1 (for Nck-associated protein 1), and pir1 (for 121F-specific p53 inducible RNA) mutants (Hossain et al. 2012). Our data, along with previous reports, indicate that tyrosine has a strong influence on the cytoskeletal dynamics of plant cells and that the interference of its metabolism could impact the cell architecture. The wide bulbous swelling of root hairs in the dahps1-1 and dahps1-2 mutants resembles the phenotype of the root hair deficient 1 mutant in Arabidopsis, disrupted in the UDP-D-glucose 4-epimerase (UGE) enzyme, necessary for the galactosylation of cell wall components (Seifert et al. 2002). Cell wall composition and flexibility are certainly crucial to sustain the root hair shape and growth, since the list of cell wall-related mutants with defective root hairs includes the leucine-rich repeat extensins Atlrx1 and Atlrx2 (Baumberger et al. 2003), and the cellulose synthases-like D (Wang et al. 2001;Karas et al. 2021). L-Phenylalanine and L-tyrosine are products of the shikimate pathway and also serve as precursors for phenylpropanoid metabolism. These secondary metabolites include flavonoids and cell wall-associated phenolics (Vogt 2010). It was recently shown that flavonols modulate the reactive oxygen species (ROS) levels that drive root hair development in A. thaliana. Mutants affected in synthesis of flavonols exhibit a greater frequency of trichoblast cells forming root hairs and raised epidermal ROS levels (Gayomba and Muday 2020). Our RNAseq analysis of dahps1-2 roots suggests that the root hair phenotype in dahps1-2 could be related to a misregulation of several genes related to cell wall biomechanics such as Expansin, Pectate lyase, Xyloglucan endoglucanase, CASP-like protein, Endoglucanase, and Pectinesterase. This hypothesis is further supported by the perturbance observed in the cell wall dynamics of dahps1-2 root hairs stained with PI, a fluorescent dye that binds to demethoxylated pectins (Rounds et al. 2011). The absence of PI labeling at the root tip of growing root hairs is presumably related to greater extensibility properties, allowing the polar root hair growth (Rounds et al. 2011). Interestingly, the dahps1-2 root hairs lacked this fluorescent pattern and the PI was rather detected on the entire periphery, which likely reflects a reduced extensibility of the cell wall that prevents apical root hair growth.

Lotus-rhizobia symbiosis
Evidence collected by comprehensive approaches revealed the participation of DAHPS1 at different stages of the Lotus-rhizobia symbiosis, through the interplay with various biological processes. DAHPS1 expression is clearly linked to nodule formation, since its promoter was strongly activated in developing nodules and the number of these structures was significantly reduced in the dahps1-1 and dahps1-2 mutants after M. loti or IRBG74 inoculation. This symbiotic phenotype is probably caused by insufficient levels of AAAs and their derived compounds, in a high-demanding organogenesis program (Mergaert et al. 2020). DAHPS1 is apparently an integral component of the genetic program governing developmental processes, since its promoter was also highly active in the RAM and emerging lateral roots. In M. truncatula, it was recently shown that a large proportion of the transcriptome changes in lateral root primordia also occur in developing nodules (Schiessl et al. 2019). Additionally, the delay in the nodulation kinetics observed in the DAHPS1 mutant alleles could be provoked by deficient reprograming of the transcriptome. The RNAseq analysis of dahps1-2 roots inoculated with IRBG74 showed that, although the major symbiotic genes were induced, the total number of DEGs was considerably lower compared to the transcriptome response in Gifu wild type and only a minor proportion of DEGs overlapped between these 2 genotypes. DAHPS1 is the major 3-deoxy-d-arabino-heptulosonate 7-phosphate synthase isoform expressed in Lotus roots; therefore, it is likely that disruption of this gene has a negative impact on the flavonoid levels derived from phenylalanine. Flavonoids are essential compounds produced by legume roots in the rhizosphere for their chemical crosstalk with rhizobia (Liu and Murray 2016), and the silencing of flavonoid biosynthesis components inhibits nodulation in soybean (Glycine max) and M. truncatula transgenic roots (Subramanian et al. 2006;Wasson et al. 2006).
In addition, disruption of DAHPS1 also interferes with rhizobial colonization both intra-and intercellularly. The presence of IRBG74 was reduced by 50% in dahps1-2 roots compared to Gifu, at 3 wpi. However, the impact on the intracellular infection by M. loti was more dramatic, as the number of epidermal and cortical root hair ITs diminished by >70% in dahps1-2 at 1 wpi compared to Gifu. Additionally, root hair curling, a key structure to trap rhizobia and initiate IT formation, was not observed in dahps1-2 plants. The compromised rhizobial colonization is likely influenced by the abnormal and collapsed root hairs in the DAHPS1 mutants, caused by the perturbed cytoskeletal organization in the root hairs. Cytoskeleton rearrangements are necessary for the root hair rhizobial infection (Yokota et al. 2009;Hossain et al. 2012;Qiu et al. 2015), and apparently, the intercellular invasion also depends on this response, since Lotus mutants disrupted in cytoskeleton-related genes show a severe nodulation phenotype with IRBG74 (Copeland 2021;Montiel et al. 2021).
The analysis of legume mutants indicates that the genetic requirements for RNS and AMS are partially overlapping, leading to the concept of a common symbiosis signaling pathway (CSSP) (Parniske 2008). Recent phylogenomic studies confirm that certain common symbiosis genes have been retained in different plant lineages that engage in intracellular symbiotic associations (Radhakrishnan et al. 2020). However, several transcriptome analyses in legumes show that a larger set of genes is induced in mycorrhized and nodulated roots (Manthey et al. 2004;Deguchi et al. 2007;Handa et al. 2015;Nanjareddy et al. 2017). For instance, mycorrhization and nodulation are affected in Lotus roots silenced or disrupted in the Lectin Nucleotide Phosphohydrolases (LNP), and the SNARE genes LjVAMP72a and LjVAMP72b (Roberts et al. 2013;Sogawa et al. 2018). We found that although the different fungal structures were formed in the dahps1-2 roots at 4 wpi, their numbers were significantly lower than those in Gifu. However, the proportion of the AM components in the root system was comparable to that of Gifu at 6 wpi, reflecting a delay in the colonization process. Such delay in the dahps1-2 mutant is probably linked to a deficiency of strigolactone levels, since these signaling molecules in the AM symbiosis are synthesized from carotenoids, compounds produced through the shikimate pathway. Additionally, the defects observed in the organization of the actin cytoskeleton of the cortical cells in the dahps1 mutant could negatively impact the accommodation of arbuscules.

Conclusion and perspectives
AAAs are the building blocks of indispensable metabolites and proteins for cell functioning. Interestingly, no pleiotropic effects were observed in the dahps1-1 and dahps1-2 mutants, but instead specific phenotypes in root hair development and mutualistic associations. The other 2 DAHPS expressed in Lotus roots could supply the minimal requirements of the cell but were still insufficient for the aforementioned processes. The linking of some AAAs and derived compounds with the cell wall and actin cytoskeleton is consistent with developmental and symbiotic defects observed in the DAHPS1 mutant alleles. However, further research is needed to fully understand the specific metabolites and proteins involved in these different processes.

Germination, nodulation kinetics, and genotyping
The genotypes used in this study belong to the genetic background of L. japonicus accession Gifu (Handberg and Stougaard 1992). For germination, the seedcoat was mechanically removed with sandpaper and the surface sterilized with sodium hypochlorite (3%, w/v) for 10 to 15 min and washed 3 to 5 times with sterile distilled water to remove traces of chlorine. The imbibed seeds were transferred to square Petri dishes with damp paper and incubated at 21 °C for germination. For nodulation assays, 3-to-5-d-old seedlings were transferred to square Petri dishes with a 1.4% (w/v) agar slant supplemented with ¼ B&D solution (Broughton and Dilworth 1971), which provides minimal requirements for plant growth without a nitrogen source to favor the RNS. To test the plant growth in nitrogen-replete conditions, the medium with agar was supplemented with ½ Gamborg's B-5 basal medium (Sigma-Aldrich, G5893). The agar was covered with autoclaved filter paper and inoculated with M. loti R7A or with A. pusense IRBG74 (1 mL of bacterial culture per plate: OD600 = 0.05). The experiment was conducted in a temperature-controlled growth room (21 °C) with photoperiod (16/8 h). Using a stereomicroscope, the number of white and pink nodules was recorded weekly at 1 to 6 wk post inoculation (wpi). The LORE1 lines 30100225 and 30141487, affected in the LjDAHPS1 gene, were obtained from the LORE1 mutant collection (Urbanski et al. 2012;Małolepszy et al. 2016) and genotyped to obtain homozygous mutants with allele-specific primers, following the database guidelines (Mun et al. 2016).

Root phenotyping and pharmacological complementation
Gifu, dahps1-1, and dahps1-2 seeds were surface sterilized as mentioned above and transferred to square Petri dishes with an 0.8% (w/v) agar slant supplemented with 0.2× MS medium and 1% (w/v) sucrose, which allows optimal growth and contains a nitrogen source. For the chemical complementation, 50 and 100 µM of L-phenylalanine, L-tryptophan, and L-tyrosine were added to the agar individually or in a mixture (50 and 100 µM each AAA). The experiment was conducted in a growth chamber with controlled temperature and photoperiod. The root growth dynamics was analyzed by measuring the root length daily in the different genotypes, from the radicle emergence of up to 11 dpg. The lengths of the apical meristem and root hairs were measured from images obtained with a stereomicroscope on 10-dpg plants.
The actin cytoskeleton of 4-dpg seedlings of different genotypes was visualized with epifluorescence microscopy on root segments fixed with Alexa-Phalloidin (Yokota et al. 2009). The dynamics of filamentous actin plus ends was monitored in live root hairs of Gifu, dahps1-1, and dahps1-2 seedlings, mounted carefully in adapted Petri dishes with 1 mL of Fahraeus medium containing 4 µL of Cyt-Fl probe (Molecular Probes; 2.5 µM) (Zepeda et al. 2014). Cells incubated with Cyt-Fl were excited at 484 nm, and emission was collected at 530 nm (20-nm band pass). Similarly, live root hairs were incubated for 20 min with PI (20 to 40 µM) to evaluate the distribution pattern of demethoxylated pectin, following the protocol described by Rounds et al. (2011). All filters used were from Chroma Technology, and image acquisition and analysis were carried out using MetaMorph/MetaFluor software (Universal Imaging, Molecular Devices).

In silico analyses and RNAseq
The nucleotide and peptide sequences of LjDAHPS1, LjADAHPS2, and LjDAHPS3 were extracted from the Lotus genome browser (Mun et al. 2016; https://lotus.au.dk/ genome/). The presence of the signal peptide in the amino acid sequences was predicted with TargetP-2.0 (https:// services.healthtech.dtu.dk/service.php?TargetP-2.0). DAHPS from other plant species were obtained through protein BLAST in Phytozome (Goodstein et al. 2012), using as a query the 3 Lotus DAHPS. The amino acid sequences were aligned and bootstrapped (NJ tree, 1,000 iterations) by ClustalX 2, and the tree was visualized with Dendroscope (Huson and Scornavacca 2012). The accession numbers and annotations of DAHPS sequences are indicated in Supplemental Table S2.
The dahps1-2 seeds were germinated, and the seedlings were inoculated on plates with IRBG74 or mock treated (with water), following the protocol mentioned above. The root segments susceptible to intercellular infection (elongation and maturation zone) were cut and frozen in liquid nitrogen at 5 dpi for total RNA isolation. The RNA integrity and concentration were determined by gel electrophoresis and NanoDrop, respectively. DNA contamination was removed through DNAse treatment. Library preparations using randomly fragmented mRNA were performed by IMGM laboratories (Martinsried, Germany) and sequenced in paired-end 150-bp mode on an Illumina NovaSeq 6000 instrument. A decoy-aware index was constructed for Gifu transcripts using default Salmon parameters, and reads were quantified using the validate Mappings flag (Salmon version 0.14.1 (Patro et al. 2017)). Normalized expression levels and differential expression testing were calculated with the R-package DESeq2 version 1.20 (Love et al. 2014) after summarizing gene level abundance using the R-package tximport (version 1.8.0).

Root infection phenotyping and nodule histology
The different genotypes were inoculated with the M. loti-LacZ strain, and the roots were harvested at 1 wpi for histochemical staining with X-Gal and recording the IT progression within the root tissues. The intercellular infection was analyzed as previously described (Montiel et al. 2021). Gifu and dahps1-2 roots were collected at 3 wpi with IRBG74, incubated for 1 min in a solution for surface disinfection (0.3% w/v of sodium chloride and 70% v/v EtOH), and then washed 5 times with distilled water. The total DNA was extracted from individual roots, adjusted to 10 ng µL −1 , and used as template for qPCR to evaluate the IRBG74 NodA abundance with the primers; forward: GAACTGCAAGTTGACGATCACGC and reverse: AAACGTCGTAACAAGCCCATGTGG. The expression values were normalized to the abundance of the L. japonicus gene LotjaGi1g1v0152000.1 with the oligonucleotides; forward: GAAGGACCCAGAGGATCACA and reverse: CGGTCT TCGTACTTCTTCGC using the delta Ct method (Pfaffl 2001).
Three-wk-old nodules were detached from Gifu and dahps1-2 plants inoculated with M. loti or IRBG74 and preserved in a fixative solution (0.1 M sodium cacodylate pH 7, 2.5% v/v glutaraldehyde). Fixed nodule slices were embedded in acrylic resin and sectioned for light microscopy and for TEM. For light microscopy analysis, the nodule semithin sections (1 µm thickness) were stained with toluidine blue, while for TEM, the ultrathin sections (80 nm thickness) were stained with uranyl acetate (James and Sprent 1999;Madsen et al. 2010).

Constructs for promoter activity and subcellular localization
The predicted promoter (2 kb upstream from the start codon) and CDS (1,617 bp) of LjDAHPS1 were obtained from the L. japonicus Gifu genome (Kamal et al. 2020) in the Lotus database (Mun et al. 2016) and synthetized with 5 ´overghangs for goldengate cloning with BsaI and BpiI restriction sites replaced by silent mutations in the CDS. Promoter, CDS, fluorescent reporters, and terminator modules (Supplemental Table S5) were assembled and cloned into a pIV10 vector (Stougaard 1995), suitable for goldengate technology. Agrobacterium rhizogenes strain AR1193 was transformed with the pDAHPS1::tYFP-nls and pDAHPS1:: DAHPS1-YFP constructs used to induce transgenic hairy roots in Gifu and dahps1-2 at 6 dpg, respectively, following a standardized protocol (Hansen et al. 1989). Three wk after infection, the main root was removed and the plants with hairy roots were transferred to square Petri dishes with ¼ B&D agar or plastic magenta boxes containing LECA substrate. Depending on the construct, the transgenic roots were mock treated or inoculated with either M. loti-DsRed (Kelly et al. 2013) or IRBG74-DsRed (Montiel et al. 2021) for inspection by confocal microscopy.

Confocal microscopy of fluorescent reporters and symbiotic colonization
The confocal microscopy analysis was carried out in a Zeiss LSM780 microscope with excitation laser/emission filter (nm) settings adjusted to the fluorescent markers: autofluorescence, 405/408 to 498 nm; YFP, 514/517 to 560 nm; GFP, 514/517 to 540; and DsRed, 561/517 to 635 nm. Gifu and dahps1-2 plants were inoculated onto plates with the fluorescent-labeled M. loti-DsRed and IRBG74-DsRed strains. For AM analysis, the tissue was incubated for 4 h in EtOH (70% v/v) at room temperature, then transferred to KOH (20% w/v) for 2 to 3 d, and washed 3 times with water. Later, the root segments were treated with HCl (0.1 M) for 1 to 2 h and the solution was replaced by PBS containing 1 µg mL −1 of WGA-Alexa Fluor 488 (Torabi et al. 2021). The root segments inoculated with the fluorescent bacteria, the transgenic roots harboring the pDAHPS1:: tYFP-nls and pDAHPS1::DAHPS1-YFP constructs, and the tissue stained with WGA-Alexa Fluor 488 were mounted onto microscope slides for observation.
The analysis of root hairs after pharmacological treatments was performed on root segments from plants germinated and grown in media supplemented with AAAs, as mentioned above. The tissue was cleared using the ClearSee-adapted protocol (Kurihara et al. 2015). For cell wall visualization, the last step of the protocol consisted of 40-min incubation in ClearSee solution supplemented with 0.1% (w/v) of Calcofluor White, for which Fluorescent Brightener 28 disodium salt (Sigma-Aldrich F3397) was used. The roots were analyzed under a confocal laser scanning microscopy setup built around a Zeiss Axiovert 200M microscope (Oberkochen, Germany) that consisted of a high-speed galvo-resonant scanner for visible wavelengths (SCANVIS), a 405-nm laser OBIS 405nm LX 100mW laser system, 495-nm longpass dichroic, 440/40-nm bandpass emission filter, photomultiplier tube (PMT) modules (standard sensitivity), a Z-Axis piezo stage with controllers, and ThorImageLS 4.0 software. All parts were from Thorlabs Inc. (Newton, NJ, USA).

Arbuscular mycorrhization quantification
Four-d-old seedlings of Gifu and dahps1-2 were placed between 2 discs of cellulose membrane filters (0.22 µm pore size) with 50 to 100 spores of R. intraradices (Symbiom), previously resuspended in Long Ashton solution (Hewitt and Smith 1975) and the carrier substrate provided by the manufacturer. The filters with the inoculated plants were covered with autoclaved sand within Magenta boxes. At 4 and 6 wpi, the root system was detached from the plants and cleared as follows: 2% KOH w/v (1 h at 90 °C), 3 to 5 washes with distilled water, 2% HCl v/v (30 to 60 min), staining with trypan blue solution (1:1:1, lactic acid, glycerol, and water; 15 to 60 min at 90 °C) and washed with 50% (v/ v) glycerol. Stained root segments were mounted onto microscope slides, and different fungal structures were recorded with the help of an optical microscope, following the protocol described by Trouvelot et al. (1986).

Heterologous expression of LjDAHPS1 in the E. coli NT1402 strain
The complementation plasmid pFAJ1708::LjDAHPS1 was constructed by cloning the coding sequence of the LjDAHPS1 transcriptionally fused to the constitutively active nptII promoter. Briefly, oligonucleotides LjDAHPS1_pFAJF (ATCTGATCAAGAGACAGGATATGGCTATCTCTTCCACT-GCCA) and LjDAHPS1_pFAJR (ACGCGGGCCGCGGCGCG CCGGATCCTCACAGTCCTAAAGGGGCAAGAG) were used to PCR amplify LjDAHPS1 from plasmid pL0M-SC3-DAHPS1, generating a DNA fragment with 20-bp overhangs at each end that facilitated Gibson assembly with Xbal/ BamHI-digested pFAJ1708. Plasmid pFAJ1708::DAHPS1 was transformed into chemically competent E. coli ST18 and selected on LB agar containing 50 μg mL −1 5-aminolevulinic acid and 15 μg mL −1 tetracycline. pFAJ1708::DAHPS1 was then isolated from ST18, confirmed by sequencing and transformed directly into NT1402 by electroporation. Two clones were isolated, E. coli NT1402-LjDAHPS1_1 and E. coli NT1402-LjDAHPS1_2, which were utilized in all subsequent growth complementation assays.

Statistical analyses
For multiple comparisons, analysis of variance (ANOVA) followed by Tukey's post-hoc test was conducted. Individual comparisons were done by t-test or Mann-Whitney U test. P-values and the number of samples are shown in the figure legends.

Accession numbers
IDs, sequences, and accession numbers of the genes analyzed in this study are shown in Supplemental Table S2. The RNAseq reads associated with this study are available in the SRA under bioproject accession number PRJNA632725.

Acknowledgments
The E. coli strain NT1402 was kindly provided by Prof. Georg A Sprenger.

Supplemental data
The following materials are available in the online version of this article.
Supplemental Figure S1. Expression profile of Lotus DAHPS in different tissues and conditions. Supplemental Figure S2. Multiple sequence alignment of Lotus DAHPS.
Supplemental Figure S4. Actin cytoskeleton organization in epidermal and cortical cells of dahps1 mutants.
Supplemental Figure S5. Partial heterologous complementation of an E. coli dahps triple mutant by LjDAHPS1.
Supplemental Figure S6. Chemical complementation of dahps1-2 mutant and subcellular localization of DAHPS1-YFP in Lotus transgenic roots.
Supplemental Table S1. List of DEGs in Lotus roots after M. loti or IRBG7 inoculation.
Supplemental Table S2. Peptide sequences of DAHPS in various plant species.
Supplemental Table S5. Sequences used for goldengate cloning.

Funding
This work was supported by the Bill & Melinda Gates Foundation (OPP11772165; grant Engineering the Nitrogen Symbiosis for Africa made to the University of Cambridge), the H2020 European Research Council, research and innovation programme (grant agreement no. 834221), the Consejo Nacional de Ciencia y Tecnología (CONACyT, grant A1-S-9236), and Dirección General de Asuntos del Personal Académico-Universidad Nacional Autónoma de México (UNAM)-Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica (PAPIIT, grant IA200723 and IN204221). I.G.-S. was granted with a PhD scholarship of CONACyT (856458).
Conflict of interest statement. The authors declare no competing financial interest or other conflict of interest.

Data availability
The calculated expression values and statistics of the RNAseq data are included as Supplemental Tables S3 and S4.