CRISPR/Cas9 editing of two adenine phosphoribosyl transferase coding genes reveals the functional specialization of adenine salvage proteins in common bean

Abstract

Adenine metabolism is important for common bean (Phaseolus vulgaris L.) productivity since this legume uses ureides derived from the oxidation of purine nucleotides as its primary nitrogen storage molecules. Purine nucleotides are produced from de novo synthesis or through salvage pathways. Adenine phosphoribosyl transferase (APRT) is the enzyme dedicated to adenine nucleobase salvage for nucleotide synthesis, but it can also convert active cytokinin bases into their inactive nucleotide forms. In common bean, APRT is encoded by four genes. Gene expression analysis, biochemical properties, and subcellular location indicated functional differences among the common bean APRT isoforms. CRISPR/Cas9 targeted down-regulation of two of the four PvAPRTs followed by metabolomic and physiological analyses of targeted hairy roots revealed that, although the two proteins have redundant functions, PvAPRT1 mostly participated in the salvage of adenine, whereas PvAPRT5 was the predominant form in the regulation of cytokinin homeostasis and stress responses with a high impact in root and nodule growth.

Editing of two adenine phosphoribosyl transferase genes in common bean reveals that PvAPRT1 acts in purine nucleotide salvage, while PvAPRT5 is involved in cytokinin regulation and root and nodule development.

Introduction

Common bean (Phaseolus vulgaris L.) is an important legume for human consumption (Mukankusi et al., 2019), and by its symbiotic association with rhizobia it can use atmospheric nitrogen (N₂), thus contributing to a more sustainable agriculture by reducing fertilization requirements. Common bean uses the nitrogen fixed in nodules for the de novo synthesis of purine nucleotides, subsequently oxidized to ureides (allantoin and allantoate), which are the main transport nitrogen molecules in the ureidic legumes (Atkins, 1991; Werner and Witte, 2011; Díaz-Leal et al., 2012). Thus, purine nucleotides are not only constituents of nucleic acids and participate in energy metabolism, but are also the precursors of ureides and hence important for nitrogen metabolism (Zrenner et al., 2006; Zrenner and Ashihara, 2011).

The cell pool of purine nucleotides depends on the regulation of the de novo synthesis, salvage, and degradation pathways. De novo synthesis occurs in most organisms through 10 enzymatic reactions, starting with the formation of phosphoribosylamine, from phosphoribosyl pyrophosphate (PRPP) and glutamine, and ending with the production of adenosine-5ʹ-monophosphate (AMP) and guanosine-5ʹ-monophosphate (GMP) nucleotides (Berens et al., 1995; Zrenner et al., 2006). Degradation of purine nucleotides leads to the synthesis of ureides, whereas salvage pathways recycle purine bases to their respective nucleotides, thus reducing ureide synthesis from these bases.

Salvage of purine nucleobases to nucleotides depends on the activity of only two enzymes, hypoxanthine/guanine phosphoribosyltransferse (HGPRT; 2.4.2.8) and adenine phosphoribosyltransferase (APRT; EC 2.4.2.7). Moreover, salvage reactions are more energetically favourable than de novo synthesis, and they are also essential in cell metabolism (Moffatt and Ashihara, 2002; Smith and Atkins, 2002; Ashihara et al., 2018; Witte and Herde, 2020). APRTs are responsible for adenine salvage (Fig. 1A), which is the main AMP supply in many organisms (Kornberg et al., 1955; Henderson and Paterson, 1973). In humans, deficiency in APRT produces accumulation of 2,8-dihydroxyadenine, generating kidney stones (Bollée et al., 2012), while deficiency in guanine salvage by HGPRT causes Lesch–Nyhan syndrome, with uric acid overproduction and neurological disorders (Torres and Puig, 2007). HGPRT and APRT proteins are crucial for ATP and GTP production and the proper functioning of neural connections in animals (Frenguelli, 2019). In plants, adenine is present at much higher levels than guanine (Ashihara et al., 2018; Witte and Herde, 2020). Thus, although HGPRT is highly relevant in mammalian tissues, the physiological importance of APRT is greater than that of HGPRT in plants (Deng and Ashihara, 2010).

Fig. 1.

Introduction to the adenine phosphoribosyl transferase (APRT) proteins. (A) Schematic representation of enzymatic reaction catalysed by the APRT protein. PRPP, phosphoribosyl pyrophosphate; AMP, adenosine-5ʹ-monophosphate; SAM, S-adenosyl-l-methionine. Plant cell sources of adenine are indicated. (B) Phylogenetic distances of APRTs. Sequences aligned using Clustal W and evolutionary history inferred by neighbour-joining. Optimal tree and percentage of replicate trees clustering together in bootstrap test (1000 replicates) are shown next to the branches. The tree is drawn to scale; branch length units show evolutionary distance. Phylogenetic distances show amino acid differences per site by the p-distance method. Homo sapiens (isoforms a and b), Phaseolus vulgaris, Glycine max, Medicago truncatula, Oryza sativa, Arabidopsis, and Chlamydomonas reinhardtii APRT amino acid sequences were compared. Evolutionary analyses were conducted in MEGA11. Green circle shows prediction of chloroplast signal peptide according to Psort software. (C) Organization of peptide domains in P. vulgaris APRT proteins. Red and purple letters denote invariable residues at the active site.

APRT enzymes were early studied in yeast (Kornberg et al., 1955) and other organisms, including plants (Adams and Harkness, 1976; Moffatt et al., 1994). APRT catalyses the reaction between adenine and PRPP to form AMP (Fig. 1A). This reaction is carried out by a single protein in mammals (Tischfield and Ruddle, 1974), whereas five APRT isoforms have been described in Arabidopsis (Allen et al., 2002). Moreover, plant APRT activity can also catalyse the conversion of active cytokinin bases to their respective inactive nucleotides (Mok and Mok, 2001; Allen et al., 2002; Moffatt and Ashihara, 2002; Zhang et al., 2013; Men et al., 2021). However, despite their relevance in the metabolism of nucleotides, as a source of energy molecules, in regulation of plant growth and development, and in nitrogen storage in ureidic legumes, APRTs have been poorly investigated in these plants.

This study uncovers the molecular and functional characterization of common bean APRT isoforms. Differential metabolomic and physiological effects of CRISPR/Cas9 targeted down-regulation of the two more highly expressed common bean APRTs revealed that PvAPRT1 mostly participates in the purine nucleotide salvage pathway, while PvAPRT5 is more involved in the regulation of cytokinin levels.

Materials and methods

Plant material and culture conditions

Phaseolus vulgaris Great Northern ‘Matterhorn’ (PMB-0220) seeds were sown in pots with vermiculite/perlite (2/1) mix and inoculated with Rhizobium leguminosarum ISP14. Plants were irrigated with nutrient solution (Rigaud and Puppo, 1975) without nitrogen, or with 10 mM NO₃⁻, and cultured under 300 µmol m⁻² s⁻¹ lighting for 16 h at 26 °C and 8 h darkness at 20 °C at 70% relative humidity for 28 d. Plant material was collected, frozen with liquid nitrogen, and stored at −80 °C.

Identification, cloning, and sequence analysis of common bean adenine phosphoribosyl transferases

Arabidosis APRT peptide sequences were used for BLAST search in the P. vulgaris v2.1 genome in Phytozome v13 (Goodstein et al., 2012; Schmutz et al., 2014). Full-length APRT coding sequences from P. vulgaris (PMB-0220) genotype were amplified using gene specific primers (Supplementary Table S1). The cDNAs were cloned and PvAPRT sequences were confirmed by sequencing.

Phylogenetic analysis, physicochemical properties, and subcellular location prediction

Phaseolus vulgaris (PMB-0220) APRT peptide sequences were compared with those from several legume and non-legume plants, algae, and human, and with various APRT sequences from P. vulgaris genotypes available in Phytozome v13. Alignment of sequences was done using DNASTAR Lasergene, and phylogenetic analysis was made using the neighbour joining method (MEGA 11).

Isoelectric point, molecular mass (kDa), stability, and aliphatic index of deduced common bean APRT proteins were calculated using the ProtParam tool. Subcellular location was searched using TargetP software.

Analysis of cis-regulatory motives in the 1500 bp proximal PvAPRT promoter sequences from P. vulgaris UI111 v1.1 (Phytozome v13) was performed by PlanCare prediction software (Lescot et al., 2002).

Expression of recombinant PvAPRT-HA-Strep and PvAPRT-YFP tagged proteins in Nicotiana benthamiana

Full-length PvAPRTs cDNAs were cloned into the expression vectors pXCSHAStrepII or pXCS-YFP to generate C-terminal hemagglutinin (HA)-Strep-tagged proteins (Witte et al., 2004) and C-terminal yellow fluorescent protein (YFP)-tagged proteins (Chen and Witte, 2020), using ClaI and SmaI cloning sites. Constructs were verified by sequencing and introduced by electroporation with Agrobacterium tumefaciens (GV3101-pMP90RK) cells.

Agrobacterium tumefaciens carrying pXCSHAStrep-PvAPRT and pXCHAStrep-YFP-PvAPRT constructs, and A. tumefaciens (C58C1-pCH32) with Pbin61-P19, expressing the P19 silencing suppressor, were cultured overnight at 28 °C, collected by centrifugation at 1100 g for 10 min and resuspended in induction medium (10 mM MgCl₂, 10 mM MES, pH 5.6, and 150 µM acetosyringone). After 2 h incubation at room temperature, each cell suspension was diluted until 0.5 OD_600nm, mixed 1:1 with the Pbin61-P19 and infiltrated into 3-week-old N. benthamiana leaves. Tissue was collected 4 d after agroinfiltration. Purification of recombinant PvAPRT-HA-Strep proteins was done by affinity binding to StrepTactin Sepharose (GE Healthcare) according to Witte et al. (2004).

Protein localization by confocal microscopy

Nicotiana benthamiana leaves expressing the PvAPRT-YFP isoforms were used for protein localization by confocal microscopy, using an Axioskop 2 MOT microscope (Carl Zeiss) equipped with a krypton and an argon laser, controlled by the Carl Zeiss Laser Scanning System LSM5 PASCAL software (Carl Zeiss).

Gel electrophoresis and western blot analysis

Proteins were separated by 12% SDS-PAGE (Laemmli, 1970) and transferred to polyvinylidene fluoride membranes (Sigma-Aldrich). Membranes were incubated overnight at 4 °C with anti-HA (Thermo Fisher Scientific) or anti-GFP (Sigma-Aldrich) monoclonal antibodies at 1:2500 dilution. Alkaline phosphatase conjugated anti-mouse IG (Sigma-Aldrich) at 1:12 000 was used as secondary antibody. Immunodetection was developed by phosphatase activity (0.1 mM Tris–HCl pH 9.5, 0.15 mg ml⁻¹ 5-bromo-4-chloro-3-indolyl-phosphate, 5 mM MgCl₂, and 0.30 mg ml⁻¹ nitro blue tetrazolium).

RNA isolation and gene expression analysis

RNA was isolated from 50–100 mg frozen tissue. Genomic DNA was removed with DNase I (New England Biolabs) and 2.5 µg DNase-treated RNA was used for cDNA synthesis using iScript (Bio-Rad) reverse transcriptase. Gene expression was analysed by qRT-PCR using specific primers (Supplementary Table S1) and iQ SYBR-Green supermix in an iCycler iQ System (Bio-Rad). Relative expression of APRT genes was normalized using Actin-2 or Ubiquitin constitutive genes and estimated according to Livak and Schmittgen (2001).

Adenine phosphoribosyl transferase activity assay

Adenine conversion to AMP by PvAPRT proteins was determined by two-step coupled reactions according to Kojima et al. (1991), with some modifications. Activity was assayed using crude extracts from plant tissues or a known amount of affinity purified recombinant APRTs. Extracts were prepared homogenizing 60 mg tissue with 180 µl extraction buffer (100 mM TEA–NaOH (pH 8), 5 mM DTT, and 1 mM MgCl₂) and centrifuged at 4 °C for 10 min. Supernatants were dialysed through a SpinTrap G25 column and centrifuged for 30 s at 800 g. In the first reaction, to transform adenine to AMP, 50 µl dialysed extract was mixed with 500 µl reaction buffer I (50 mM Tris–HCl, pH 7.5, 5 mM MgCl₂, 1 mM PRPP, and 1 mM adenine) and incubated for 1 h at 37 °C. The reaction mix was boiled at 100 °C for 5 min, centrifuged for 10 min at 3000 g at 4 °C, and the supernatant was collected. For the second reaction, to transform AMP to inosine, 300 µl of the first reaction supernatant was mixed with 500 µl reaction buffer II (100 mM Tris–HCl, pH 9, 10 mM MgCl₂), and 1 U ml⁻¹ of Calf Intestinal Alkaline Phosphatase (Canvax) and 1.5 U ml⁻¹ adenosine deaminase (Sigma/Roche) were added. The mix was incubated 1 h at 37 °C, followed by protein precipitation with 50 µl 3.6 M HClO₄ and centrifugation at 3000 g at 4 °C for 10 min. The resulting inosine was determined by HPLC (Jasco), using AS-2055 Plus autosampler, PU 2089 pump, Jasco MD-2010 Plus Diode Array, and 5 µm, 250 mm×4.6 mm Extrasil ODS2 (C18) column. Samples (20 µl) were resolved with 50 mM sodium phosphate (pH 2.3) and 2% methanol isocratic eluent for 55 min at 0.5 ml min⁻¹ flow rate.

An APRT assay using zeatin as substrate, to transform zeatin nucleobase to zeatin nucleotide, was performed using 0.3 µg affinity purified recombinant APRT1 or APRT5 and 500 µl of reaction buffer I (50 mM Tris–HCl, pH 7.5, 5 mM MgCl₂, 1 mM PRPP, and 50 µM zeatin) incubated for 1 h at 37 °C. Reactions were boiled at 100 °C for 5 min, centrifuged for 10 min at 3000 g, 4 °C, and the supernatants lyophilized and used for hormone metabolite analysis as explained below.

Generation of CRISPR/Cas9 targeted PvAPRT1 and PvAPRT5 mutants

Target CRISPR/Cas9 editing sites in PvAPRT1 and PvAPRT5 genomic sequences were searched using Breaking-Cas software (Oliveros et al., 2016). Sequences with low off-target predictions were selected to design each single guide RNA (sgRNA), and the pUC119-gRNA and pFGC-pcoCas9 vectors (Addgene) were used to generate guide RNAs (gRNAs) and CRISPR/Cas9 gene editing constructs, according to Li et al. (2013). Two sgRNAs were generated by recombinant PCRs using pUC119-gRNA as template and specific primers (Supplementary Table S1) targeting the first and third exonic sequences in PvAPRT1 and the second and third ones in PvAPRT5, and Phusion Hot Start II High-Fidelity Master Mix (Thermo Fisher Scientific). The two guides were purified and inserted in pFGC-pcoCas9 vector and used to transform Escherichia coli Stellar Competent Cells (TaKaRa). The resulting constructs were verified by sequencing using specific primers (Supplementary Table S1) and introduced into Rhizobium rhizogenes K599 by electroporation.

PvAPRT1 and PvAPRT5 RNAi silencing construct

To generate silencing hairpin double-stranded RNA constructs, 353 and 331 bp fragments from the 3ʹ-region of PvAPRT1 and PvAPRT5 cDNAs, respectively, were amplified adding SwaI and AscI restriction sites for the direct fragment, and XbaI and BamHI for the reverse sequences (Supplementary Table S1). The resulting PCR products were cloned into pFGC5941 vector (AY310901) (Kerschen et al., 2004) in sense and antisense orientation, upstream and downstream of an intron, using the respective restriction sites. The resulting constructs were verified by sequencing and used to transform K599 R. rhizogenes by electroporation.

Generation of common bean transformed hairy roots

Hairy root transformation with CRISPR/Cas9-APRTs or with RNAi-APRTs constructs was done according to Estrada-Navarrete et al. (2007). At 10 d after stem wound inoculation with R. rhizogenes, the original roots were removed, and plants with hairy roots at infection sites were sown, inoculated with R. leguminosarum (CIAT-899), and grown under nitrogen fixation conditions. After 28 d, hairy roots and nodules were collected. Expression levels of APRTs was analysed by qRT-PCR and roots with reduced expression of the targeted genes were chosen for further analysis.

Metabolite identification and quantification

Metabolomic analysis was performed according to Salem et al. (2020). Samples (50 mg pulverized freeze-dried tissue) from six independent hairy roots per condition were used. Metabolites were extracted with 1ml of methyl-tert-butyl-ester:methanol (3:1). Hydrophobic and polar fractions were obtained through H₂O:MeOH (3:1 v/v) phase separation. For hormone analysis, 230 µl of the hydrophobic phase was used, and 130 µl and 260 µl from polar fraction were taken for primary and secondary metabolites, respectively. Fractions were concentrated in a speed-vac and pellets were kept at −80 °C until use.

Gas chromatography coupled with mass spectrometry

For primary metabolites, pellets from 130 µl polar fraction were derivatized according to Lisec et al. (2006). Samples were injected into a gas chromatograph coupled to a time-of-flight mass spectrometer (GC-MS) system (Pegasus HT TOF-MS, LECO), using Gerstel Multi-Purpose system autosampler (Gerstel GmbH & Co. KG). A 30 m DB-35 column was used for gas chromatography with helium as carrier at 2 ml s⁻¹ constant flow rate. Mass spectra were recorded at 20 scans s⁻¹ with m/z 70–600 scanning range. Mass chromatograms were evaluated using Chroma TOF 4.5 (LECO) and TagFinder 4.2 software.

Liquid chromatography coupled with mass spectrometry

For secondary metabolites, 260 µl polar fraction were resuspended in 400 µl 50% MeOH, mixing for 5 min. Samples were sonicated for 3 min, centrifuged for 5 min, and supernatants loaded on an UPLC-LC-MS equipped with an HSS T3 C18 reverse-phase column (100 mm×2.1 mm diameter, 1.8 μm particle size; Waters) at 40 °C. Sample of volume 2 μl was loaded per injection with 0.1% formic acid/water and 0.1% formic acid/acetonitrile as mobile phases at 400 μl min⁻¹ flow rate. The UPLC was connected to an exactive focus Orbitrap via a heated electrospray source (Thermo Fisher Scientific). The spectra were recorded using full-scan positive and negative ion-detection mode, covering a mass range from 100 to 1500 m/z, resolution 70 000, and 250 ms maximum scan time. Sheath gas was 60 and auxiliary gas 35. Transfer capillary temperature was 150 °C and heater temperature was 300 °C. Spray voltage was 3 kV, with 25 V capillary voltage and 15 V skimmer voltage. MS spectra were recorded from 0–19 min UPLC gradient. Chromatograms, peak detection, and integration analysis were performed using RefinerMS (version 5.3; GeneData).

Metabolomic data processing was done using MetabolAnalist 5.0 (Pang et al., 2021) and Expressionist Analyst 14.0.5 (Genedata). Metabolite annotation (Supplementary Table S2) was as in Alseekh et al. (2021). An in-house reference compound library allowed for 10 ppm mass error, and a dynamic retention-time shift of 0.1.

Statistical analysis

Analysis of expression and activity in common bean tissues was done using three to four independent replicates from a pool of three plants per replicate (nine plants for each condition). Student’s t-test was performed using the Holm-Šidák method.

For CRISPR/Cas9 PvAPRT1 or PvAPRT5 transgenic roots, a total of 16 plants per condition were tested. From each plant, the expression levels of individual hairy roots were analysed (n≥30). APRT activity and metabolomic analysis were determined using APRT down-regulated transgenic roots samples (n≥6). Statistical significance and fold changes were determined by two-way ANOVA with Tukey’s test using GraphPad Prism 8 (GraphPad Software).

Results

Identification of common bean adenine phosphoribosyl transferases

Using the APRT sequences from Arabidopsis as a query, four APRT coding genes were identified in the P. vulgaris v2.1 genome: Phvul.009G070900 (PvAPRT1), Phvul.001G050500 (PvAPRT2), Phvul.002G061600 (PvAPRT4), and Phvul.009G092200 (PvAPRT5), named by their similarity to four of the five Arabidopsis APRTs, AT1G27450 (AtAPT1), AT1G80050 (AtAPT2), AT4G12440 (AtAPT4), and AT5G11160 (AtAPT5). Phaseolus vulgaris Great Northern ‘Matterhorn’ (PMB-0220) full-length APRT coding sequences were cloned and sequenced. The four APRT protein sequences shared a high degree of similarity. Phylogenetic analysis classified them into two groups (Fig. 1B; Supplementary Figs S1A, S2A) with the conserved (PRT)-type I phosphoribosyl transferase domain characteristic of these enzymes (Fig. 1C). PvAPRT1 protein was closer to PvAPRT4, both carrying a possible chloroplast transfer peptide, than to PvAPRT2 and PvAPRT5, lacking the signal peptide (Fig. 1C; Supplementary Figs S1B, S2B). Comparison of the APRT sequences from three common bean genotypes revealed that PvAPRT1 and 4 proteins appeared in the same cluster, even though the P. vulgaris UI111 v1.1 genotype APRT4 isoform (PvUI111.02G061000) does not have the signal peptide (Supplementary Fig. S2). Four APRT sequences were also found in the amidic legume Medicago truncatula, three homologous to P. vulgaris APRT1, APRT4, and APRT5, while one MtAPRT appeared closer to the Arabidopsis APT3 isoform, which was absent in P. vulgaris (Fig. 1B). In contrast, 13 sequences encoding APRT isoforms were found in soybean (Glycine max) and five APRT sequences were found in rice, whereas only two copies were found in the green alga Chlamydomonas reinhardtii. In contrast, a single gene, although with two transcript variants, encodes APRT in humans (Fig. 1B). Surprisingly, three of the 13 soybean APRT sequences appeared closer to the human and algal proteins than to any other plant isoforms (Fig. 1B).

Biochemical characterization and subcellular location of PvAPRT proteins

Full-length PvAPRT1, PvAPRT2, PvAPRT4, and PvAPRT5 sequences were 246, 193, 230, and 201 amino acids long, with 26.93, 20.9, 25.37, and 21.84 kDa molecular mass, respectively (Supplementary Table S3). Removal of predicted signal peptides led to APRT1 and APRT4 mature proteins of 186 and 184 amino acids, and 20.44 and 20.36 kDa, respectively (Supplementary Table S3), suggesting that the four mature APRT isoforms have similar molecular masses, although two could be cytosolic and two targeted to chloroplasts. To ascertain PvAPRTs cellular location, APRTs–YFP fusions were overexpressed in N. benthamina. Bands of approximately 50 kDa were found in the agroinfiltrated leaves, confirming that APRTs (~20 kDa) were bound to the fluorescent YFP (~27 kDa) (Supplementary Fig. S3A). Confocal microscopy images of leaves expressing APRTs–YFP identified APRT1 and APRT4 in chloroplasts, although a small amount of APRT4 was also observed in the cytosol (Fig. 2A). On the other hand, APRT2 and APRT5, lacking targeting signal peptides, were found in the cytoplasm (Fig. 2A; Supplementary Fig. S3C).

Fig. 2.

Characterization of recombinant Phaseolus vulgaris adenine phosphoribosyl transferase (APRT) isoforms produced in Nicotiana benthamina leaves. (A) Subcellular localization of APRT isoforms. Confocal microscopy images of N. benthamiana leaves expressing the four PvAPRTs fused to yellow fluorescent protein (YFP). Scale bar indicates 20 µm. Bottom panels represent closer images of the same protein. (B) Western-blot detection of recombinant HA–PvAPRTs using anti-HA antibodies in crude extracts from control, empty vector, or HA-Strep-tagged PvAPRT N. benthamina agroinfiltrated leaves. (C) Detection of affinity purified HA-Strep-tagged PvAPRT proteins with anti-HA antibodies. (D) Determination of APRT activity in crude extracts from leaves overexpressing each APRT isoform. (E) Activity of purified HA-Strep-tagged PvAPRT proteins. Measurements were done in three to four biological replicates. Statistical significance was determined by Student’s unpaired t-test comparing control values and overexpressing tissues (**P<0.005 and ***P<0.0005).

C-terminal HA-Strep-tagged PvAPRT1, PvAPRT2, PvAPRT4, and PvAPRT5 were also expressed in N. benthamiana leaves. Western blot immunodetection of the HA epitope confirmed the expression of ~23 kDa proteins, corresponding to the expected size of the recombinant PvAPRT isoforms in crude extracts and after affinity purification (Fig. 2B, C). Crude extracts from leaves expressing APRT-Ha-Strep isoforms were used to optimize the activity assay, with adenine as substrate, for plant tissues. Activity was significantly higher in the leaves overexpressing PvAPRT1 than in PvAPRT4 and PvAPRT5 ones, whereas it was not detected in extracts overexpressing PvAPRT2 (Fig. 2D). The same pattern was found with the purified recombinant APRTs (Fig. 2E). Michaelis–Menten constants (K_m) of the purified proteins were for adenine 0.359, 0.071, and 0.062 mM and for PRPP 0.94, 0.23, and 0.28 mM for PvAPRT1, PvAPRT4, and PvAPRT5, respectively (Table 1; Supplementary Fig. S4). Despite the lower affinity for adenine of APRT1, this isoform presented the highest V_max and catalytic constant, followed by APRT5, which showed higher affinity for adenine but lower V_max and catalytic constant values than the APRT1 isoform (Table 1). By contrast, APRT4 showed even lower V_max and catalytic constant than APRT5, whereas lack of PvAPRT2 activity prevented us from determining its kinetic parameters.

Table 1.

Kinetic parameters of APRTs from P. vulgaris

	Adenine			PRPP
Protein	K m (mM)	V max (µmol mg −1 min −1 )	K cat (min −1 )	K m (mM)	V max (µmol mg −1 min −1 )	K cat (min −1 )
APRT1	0.36 ± 0.061	9.8 ± 0.66	0.426	0.94 ± 0.28	11.59 ± 2.9	0.507
APRT2	—	—	—	—	—	—
APRT4	0.071 ± 0.018	0.67 ± 0.04	0.0291	0.23 ± 0.05	0.73 ± 0.052	0.0023
APRT5	0.062 ± 0.032	1.49 ± 0.2	0.0612	0.28 ± 0.037	1.082 ± 0.06	0.0444

APRT, adenine phosphoribosyl transferase; PRPP, phosphoribosyl pyrophosphate.

APRT expression and activity in Phaseolus vulgaris tissues

Expression of PvAPRTs was measured by qRT-PCR in tissues from 28-day-old plants grown under nitrogen fixation or fertilized with NO₃⁻. Transcript levels of PvAPRT2 and PvAPRT4 were lower than those of PvAPRT1 and PvAPRT5 in every tissue. PvAPRT5 showed the highest expression in nodules, followed by PvAPRT1, which was highly expressed in most tissues (Fig. 3A). Moreover, the nitrogen source did not significantly affect the expression of PvAPRT1, 2, and 5 genes, although expression of PvAPRT4 was significantly improved in roots and stem under nitrate fertilization. Using the optimized APRT assay, enzyme activity was measured in tissues from nitrogen-fixing or NO₃⁻ fertilized plants. APRT activity was higher in nodules, roots, stems, and pods than in the leaves. In addition, APRT activity levels were similar in tissues from plants under the two nitrogen conditions (Fig. 3B).

Fig. 3.

Expression and activity levels of adenine phosphoribosyl transferase (APRT) in Phaseolus vulgaris tissues. (A) Relative expression levels of PvAPRT1, PvAPRT2, PvAPRT4, and PvAPRT5 transcripts in nodules, roots, stems, first leaf, fourth trifoliate leaves, and pods of 28-day-old P. vulgaris plants cultivated under conditions of nitrogen fixation or nitrate fertilization. (B) Enzymatic activity in nodules, roots, stems, first leaf, fourth trifoliate leaves, and pods from common bean plants grown for 28 d under N₂ fixation or fertilized with NO₃⁻. (C) Relative expression of PvAPRT1, PvAPRT2, PvAPRT4, and PvAPRT5 genes along nodule development. (D) APRT enzymatic activity in nodules at several developmental stages. Expression and activity were analysed in four independent samples. Statistical significance was determined by Student’s t-test of nitrate and nitrogen fixing plants using the Holm–Šidák method (***P<0.0005).

Expression of PvAPRT genes in nodules at different developmental stages (Fig. 3C) confirmed the highest level of PvAPRT5 expression along development of the nodules. Moreover, PvAPRT5 expression was higher in nodules at 21 d old than with nodule ageing. In contrast, PvAPRT1 expression was maintained at constant levels in nodules of different age, whereas PvAPRT2 and 4 were expressed at much lower levels, indicating that PvAPRT5 was the predominant gene expressed, particularly in early development of nodules. However, despite the higher PvAPRT4 and PvAPRT5 expression at early nodulation, we found similar enzymatic activity along nodule development (Fig. 3D). Altogether, APRT expression and activity were similar in tissues of plants grown under the two nitrogen sources, although high APRT levels were surprisingly found in nodules, where the de novo synthesis of purine nucleotides is highly prevalent.

The in silico analysis of the 1500 bp proximal promoter of PvAPRT genes showed cis-regulatory motifs related to plant development, and a surprisingly wide variety of motifs associated with stress and defence responses (Supplementary Fig. S5A, B). In the four genes there were MYC and MYB binding site motifs, abscisic acid (ABRE) and drought-related MBS motifs, salicylic acid-related TCA-elements, ethylene and JA-responsive elements, as well as P-box and ERE, associated with gibberellin and ethylene responses, and several motifs associated to light. The large differences in the regulatory motifs found in the four gene promoters suggest that they may not have totally redundant functions and point to a role of PvAPRTs in defence and stress responses.

Targeted CRISPR/Cas9 PvAPRT1 and PvAPRT5 gene knockout

To elucidate the role of APRT proteins in roots and nodules of common bean plants, down-regulation of the two most highly expressed genes, PvAPRT1 and PvAPRT5, was achieved using CRISPR/Cas9 gene editing and RNAi silencing. CRISPR/Cas9 constructs expressing Cas9 nuclease together with two gRNAs targeting first and third, and second and third exonic sequences in PvAPRT1 and PvAPRT5, respectively, were used (Supplementary Fig. S6A, B). Expression levels of the targeted genes were chosen to screen the efficacy of gene targeting (Supplementary Fig. S7A, B), since indels generated by the repair at the target sites would lead to aberrant transcripts, usually prone to degradation. Sequence analysis of genomic DNA amplicons from hairy roots exhibiting reduced expression of APRT 1 or 5 revealed several mutations in the regions near the gRNA targeted sequences. Short insertions, deletions, and substitutions were found, some causing premature transcript termination (Supplementary Fig. S6A, B).

Transgenic CRISPR/Cas9-PvAPRT1 roots showed a significant decrease in PvAPRT1 transcript abundance compared with that in the control roots, transformed with the pFGCpcoCas9 empty vector lacking sgRNAs vector (Fig. 4A, C). Furthermore, the expression levels of the other three APRT genes did not change in these roots. In the same way, in the targeted PvAPRT5 roots, the relative expression of PvAPRT5 was reduced, while the expression of the other APRTs was maintained without significant change (Fig. 4B). These results confirmed specific CRISPR/Cas9 editing of the targeted gene.

Fig. 4.

Relative gene expression and enzymatic activity of adenine phosphoribosyl transferases (APRTs) in APRT CRISPR/Cas9 transformed hairy roots. (A) Relative expression of PvAPRT1, PvAPRT2, PvAPRT4, and PvAPRT5 genes in PvAPRT1 CRISPR/Cas9 targeted transgenic roots. (B) Relative expression in PvAPRT5 CRISPR/Cas9 targeted hairy roots measured in 37 and 48 samples corresponding to three to six independent roots from six to nine PvAPRT1 and PvAPRT5 CRISPR/Cas9 targeted plants. (C) Total APRT enzymatic activity in CRISPR/Cas9 PvAPRT1 and PvAPRT5 targeted roots determined in six root samples (biological replicates), each from an independent CRISPR/Cas9 targeted plant. Statistical significance determined by Student’s t-test of control, empty vector values and those from CRISPR/Cas9 APRT targeted roots (*P<0.05 and **P<0.005 and ***P<0.0005).

Accordingly, APRT activity was lower in the transgenic roots than in those transformed with the empty vector lacking sgRNAs. The decrease in adenine salvage activity was higher in roots with mutated PvAPRT1 than in PvAPRT5 targeted ones. Specifically, targeted APRT1 showed a 47% activity with adenine as substrate, whereas it was 70.4% in the APRT5 transgenic roots compared with controls (Fig. 4C, D).

Transgenic roots with down-regulation of PvAPRT1 and PvAPRT5 were also obtained using RNAi. However, RNAi-mediated silencing was found to be less specific and less efficient than CRISPR/Cas9 editing. PvAPRT5 siRNA not only reduced its expression level, but also significantly decreased PvAPRT1 expression (Supplementary Fig. S8A, B). Moreover, we found lower decrease in total APRT activity in the RNAi roots than in the CRISPR/Cas9 targeted ones (Supplementary Fig. S8C, D). Therefore, CRISPR/Cas9 targeted PvAPRT1 and PvAPRT5 were used for subsequent functional analysis.

Metabolomic changes in PvAPRT1 and PvAPRT5 edited roots

To further investigate PvAPRT1 and PvAPRT5 functional relevance, changes in primary and secondary metabolites in the CRISPR/Cas9-PvAPRT1 and PvAPRT5 targeted roots were determined by GC/MS and LC/MS. Metabolomic analysis in samples with down-regulated levels of the respective transcripts from six independent hairy roots revealed that roots with targeted PvAPRT1 and PvAPRT5 exhibited differential quantitative and qualitative metabolite changes, indicating differences in the functions of these two proteins (Fig. 5A; Supplementary Table S4). A total of 28 metabolites exhibited significant concentration differences in roots with mutated PvAPRT1 and 39 metabolites in those targeting PvAPRT5, in comparison with control roots transformed with empty vector (Fig. 5B).

Fig. 5.

Metabolomic analysis from transgenic roots. (A) Heat map of the primary and secondary metabolites with significant differences (P<0.05) after comparison of CRISPR/Cas 9 hairy root (PvAPRT1 or PvAPRT5) with control tissues. (B) Venn diagrams with the total number of metabolites with significant differences (P<0.05) when hairy root with PvAPRT1 (red) and PvAPRT5 (green) genes down-regulated are compared with control tissues determined in six biological replicates from each group. (C) APRT activity upon zeatin nucleobase. Zeatin concentration decay determined after enzymatic assay with purified PvAPRT1 and PvAPRT5. Statistical significance determined by two-way ANOVA (**P<0.005 and ***P<0.0005).

Transgenic roots with mutated PvAPRT1 showed a significant accumulation of adenine, which was not found in PvAPRT5 targeted roots (Fig. 5A), agreeing with a prevalent role of APRT1 in adenine salvage in common bean roots. The concentration of inosine and adenosine monophosphate purine derivatives also increased in the targeted roots, although increments were lower in PvAPRT1 than in PvAPRT5 samples. Moreover, targeted PvAPRT1 and PvAPRT5 roots exhibited accumulation of the cytokinin N⁶-benzyladenine, but this was only significant in those with down-regulated PvAPRT5 (Fig. 5A; Supplementary Table S4). Transgenic hairy roots also showed differential levels of amino acids; alanine, glutamic acid, and aspartylphenylalanine accumulated in PvAPRT1 mutant, whereas serine, isoleucine, isoleucylproline, and glutamylmethionine accumulated in the targeted PvAPRT5, but not in PvAPRT1. Differential increments in carbohydrates (raffinose, erythrose, mannose), and in glycolysis and tricarboxylic acid cycle metabolites (glucose-6-phosphate and 2-oxo-glutaric, isocitric, pyruvic, succinic, and threonic acids) were also found in the two transgenic roots (Supplementary Table S4). Moreover, salicylic acid and abscisic acid, as well as several flavonoids and saponins, significantly accumulate in transgenic PvAPRT5 targeted roots (Supplementary Table S4).

Since these results strongly suggested some functional specialization, activity catalysing the conversion of zeatin nucleobase to zeatin nucleotide was assayed using APRT1 and APRT5 purified enzymes. We found that less than 30% zeatin was transformed by APRT1, whereas APRT5 was able to transform over 75% of the zeatin nucleobase substrate (Fig. 5C). Therefore, although both isoenzymes could transform zeatin nucleobase to the nucleotide, APRT5 exhibits considerably higher activity on cytokinin substrate than APRT1.

Physiological analysis of APRT transgenic hairy roots

Determination of transcript levels in individual hairy roots showed that 65% and 80% of the roots formed at the respective transformation wounds had lower expression of PvAPRT1 and PvAPRT5 than the control roots transformed with empty vector (Fig. 6A). Roots with control expression levels in the targeted plants was not surprising since induction of hairy roots by R. rhizogenes infection could appear both in transformed and untransformed cells at the wounding site.

Fig. 6.

Physiological analysis from CRISPR/Cas9 PvAPRT1 and PvAPRT5 targeted roots and nodules. (A) Percentage of hairy roots with down-regulated (silenced) or wildtype (non-silenced) expression levels of the PvAPRT1 and PvAPRT5 genes. (B) Nodule number in CRISPR/Cas9 PvAPRT1 and PvAPRT5 targeted hairy roots. (C) Percentage of CRISPR/Cas9 PvAPRT1 and PvAPRT5 targeted hairy roots with and without nodules. (D) Nodule weight from control and transgenic hairy roots. (E) Size of nodules from transgenic root of P. vulgaris with CRISPR/Cas9 PvAPRT1 and PvAPRT5 down-regulated genes. Scale bar, 0.25 cm. (F) Depth (root length) of control (empty vector) and transgenic APRT-targeted P. vulgaris roots. (G) Relative expression of genes related with cytokinin metabolism in control (empty vector) and mutated PvAPRT1 and PvAPRT5 hairy root tissue. Statistical significance determined by multiple t-tests using the Holm–Šidák method (*P<0.05; **P<0.005; ***P<0.0005).

To determine the effect of APRT down-regulation in nodulation, the nodules appearing in each targeted or empty vector transformed plant were collected. Number of nodules per plant was similar in control and mutant roots (Fig. 6B) and they appeared both in roots with down-regulated expression of the PvAPRT genes (silenced) and controls (non-silenced). However, the number of PvAPRT silenced roots without nodules was higher than that with nodules (Fig. 6C), indicating that nodulation occurred preferably in roots with control APRTs levels. Accordingly, expression of the two genes was similar in nodules from PvAPRT1 and PvAPRT5 hairy roots than in control ones (Supplementary Fig. S9A, B), and APRT activity was also at control levels in the nodules of APRT targeted roots (Supplementary Fig. S9C, D). However, size and biomass of the nodules in the APRT mutant roots were significantly larger than that of control hairy roots, transformed with the empty vector (Fig. 6D). Moreover, nodules were larger in PvAPRT5 than in PvAPRT1 edited roots (Fig. 6D, E). Interestingly, the length of transgenic roots was significantly lower than that of the controls, empty vector transformed hairy roots, and this effect was more pronounced in PvAPRT5 down-regulated roots than in those with targeted PvAPRT1 (Fig. 6F).

The accumulation of cytokinin in PvAPRT5 transgenic roots and the effects on nodule size and root length prompted us to determine the expression of genes related with cytokinin signalling (Müller and Sheen, 2007). Interestingly, expression of the Phaseolus homologue to cytokinin-activated signalling ARR (Arabidopsis Response Regulator) A-Type, PvARR15, was induced in both PvAPRT1 and PvAPRT5 transgenic roots. Instead, there were no changes in the expression of B-Type regulator PvARR2 (Fig. 6G). In addition, the expression of LONELY GUY (LOG) gene, encoding cytokinin riboside phosphoribohydrolase, which releases active free-base cytokinin from cytokinin ribotide (Kurakawa et al., 2007; Kuroha et al., 2009), was reduced in the mutant PvAPRT1 and PvAPRT5 roots (Fig. 6G). The decrease in PvLOG expression was significantly higher in mutant PvAPRT5 than in PvAPRT1, thus supporting the previous results of a higher implication for APRT5 than APRT1 in cytokinin inactivation.

Discussion

Conversion of adenine to AMP by APRT (Fig. 1A) is the main form of adenine nucleotide supply in animals and many organisms (Kornberg et al., 1955; Henderson and Paterson, 1973). In plants, purine nucleotides are produced by de novo synthesis in the chloroplast, but the salvage of adenine is also relevant for plant growth and development. Adenine bases and nucleosides are produced by turnover of nucleic acids, but also from the S-adenosyl-l-methionine cycle in biosynthetic ethylene, nicotianamine, and polyamine pathways, and by the activity of methyltransferases in caffeine and glycine betaine production (Ashihara et al., 2018). Besides the recovery of adenine, the plant APRT enzyme is also important in cytokinin metabolism (Allen et al., 2002; Zhang et al., 2013). In ureidic legumes, such as common bean, purine nucleotides synthesized in nodules are the precursors of ureides, which are their main nitrogen storage molecules, playing also important roles in stress responses (Alamillo et al., 2010). Nonetheless, despite their functional relevance, characterization of plant purine salvage enzymes is scarce. Therefore, understanding the role of adenine salvage by the APRT protein is important, because it could affect not only plant development, but also ureide synthesis and tolerance to stress.

In P. vulgaris four genes encode APRT proteins. The four common bean APRT proteins are classified into two clusters, PvAPRT1 and PvAPRT4, and PvAPRT2 and PvAPRT5 (Fig. 1B). PvAPRT1 and 4 contain an N-terminal chloroplast signal peptide, which is absent in the PvAPRT2 and 5 isoforms (Fig. 1C). These subcellular locations were confirmed by confocal microscopy using YFP-tagged proteins. (Fig. 2A). The chloroplast location of PvAPRT1 and 4 differed from early data from Arabidopsis APRTs, where the corresponding homologues AtAPT1, 2, and 3 were found in the cytosol (Allen et al., 2002). Nonetheless, AtAPT1 also contains a possible chloroplast targeting peptide (Fig. 1B), and it has been found in a chloroplast proteome analysis (Zybailov et al., 2008), indicating also a chloroplast location for Arabidopsis APT1. A chloroplast targeting sequence is also present in soybean APRT1 homologue, although not in APRT1 from Medicago and algae (Fig. 1C). The surprising location of APRTs in chloroplast and cytosol could be related to PRPP availability, since PRPP synthetase (ATP d-ribose-5-phosphate diphosphotransferase), which generates PRPP substrate for APRT activity, is present in chloroplasts and cytosol. Free adenine base, however, is not synthesized directly, but as AMP in the de novo purine biosynthetic pathway that takes place in plastids (Zrenner et al., 2006; Coleto et al., 2016). However, adenine itself arises from degradative metabolism of purines, including nucleic acid breakdown, methionine metabolism, and ethylene synthesis, in reactions occurring in the cytoplasm (Moffatt and Ashihara, 2002). On the other hand, the different subcellular locations agrees with a differential functionality of the APRT isoforms, as previously suggested (Ashihara, 2016; Andriotis and Smith, 2019). Moreover, it is well known that cytokinin activity regulates chloroplast development and functioning and prevents chlorophyll degradation (Cortleven and Schmülling 2015). It is conceivable that the isoforms more highly dedicated to cytokinin inactivation will be excluded from plastids, as we found for the PvAPRT5.

Expression of recombinant HA-tagged PvAPRTs demonstrated that the four genes encode mature proteins of the expected molecular mass (Fig. 2B; Supplementary Fig. S3). Moreover, recombinant PvAPRT1, PvAPRT4, and PvAPRT5 isoforms were active using adenine as substrate, and PvAPRT1 was the isoform with highest activity towards adenine (Fig. 2D, E). Accordingly, purified PvAPRT1 exhibited the highest V_max, although it also showed the lowest affinity for adenine, whereas APRT4 and 5 isoforms showed higher substrate affinity, but lower V_max. A similar opposite trend among K_m and V_max was reported for Arabidopsis APRTs, although K_m values for common bean APRTs where higher than those in Arabidopsis (Allen et al., 2002), perhaps reflecting negative effects caused by the larger HA-Strep-tag than the 6×His tag or by the different procedures and buffers used to purify the recombinant proteins. Nevertheless, further biochemical characterization will be required to explain the high, above physiologically expected, K_m values observed with the recombinant PvAPRT proteins. Arabidopsis APT1 also showed the highest activity with adenine, whereas APT5 activity was lower than APT1 with adenine, but similar when using zeatin (Zhang et al., 2013). Moreover, AtAPT4 exhibited low activity and AtAPT2 failed to show any catalytic activity using adenine. Therefore, as in its Arabidopsis counterpart, the common bean APRT1 isoform makes the highest potential metabolic contribution to adenine formation, estimated as V_max/K_m, while the APRT5 isoform also has higher potential than APRT4, although lower than APRT1, and APRT2 was inactive, as found for Arabidopsis APT2 (Table 1; Supplementary Fig. S4).

Transcript levels of PvAPRTs in plants grown under conditions of symbiotic nitrogen fixation or fertilized with NO₃⁻ revealed that PvAPRT1, 2, and 5 expression was unaffected by the nitrogen source, whereas PvAPRT4 exhibited higher expression in plants with nitrate than under nitrogen fixation (Fig. 3A), although without a significant effect on the total APRT activity (Fig. 3B). However, the nitrogen source significantly impacts purine nucleotide metabolism in common bean. Nitrate inhibits the activity of glutamine phosphoribosyl aminotransferase (PvPRAT), the initial enzyme of de novo synthesis of purine nucleotides (Coleto et al., 2016). Therefore, while nitrate reduces the de novo synthesis pathway, it maintains or increases the expression APRTs. Consistent APRT activity in tissues supplied with both nitrogen sources (Fig. 3B) suggests that adenine salvage, generating AMP from adenine, is maintained despite the activity of de novo synthesis. This observation is not surprising, as besides providing AMP, salvage activity also eliminates adenine and adenosine, which could have detrimental effects for plant development (Witte and Herde, 2020). In Arabidopsis, APT1 mutant exhibited male sterility (Gaillard et al., 1998), and down-regulation of APRT increases resistance to oxidative stress (Sukrong et al., 2012).

Common bean APRTs exhibited high expression in roots, stems, young pods, and nodules (Fig. 3A, C). Meristems in these tissues require abundant nucleotides to support cell division, as well as the regulation of cytokinin levels (Mok and Mok, 1994; Dewitte et al., 1999; McAtee et al., 2013). Accordingly, the promoter regions of common bean APRT genes contain several cis-acting motifs associated with phytohormones and stress responses (Supplementary Fig. S5). Besides adenine salvage, plant APRTs catalyse the conversion of cytokinin nucleobases (active forms) to cytokinin ribotides (inactive forms) (Lee and Moffatt, 1993; Zhang et al., 2013; Ashihara et al., 2018). Therefore, expression of PvAPRT genes in these tissues could be linked to a regulatory role in cytokinin activity. Moreover, the different motifs related to stress responses in the four APRT genes also suggest differential roles of the APRT isoforms. Interestingly, PvAPRT2, with several motifs involved in abscisic acid responsiveness, appeared up-regulated in a recent transcriptomic analysis in response to drought stress in common bean (López et al., 2023).

In nodules of ureidic legumes, the de novo synthesis of purine nucleotides is highly induced (Atkins, 1991; Atkins and Smith, 2000; Coleto et al., 2016), suggesting that the salvage pathway involving APRTs could be less necessary. However, PvAPRT genes, particularly PvAPRT5, were highly expressed in nodules. Furthermore, PvAPRT4 and PvAPRT5 exhibited their highest expression at early nodule development. During initial nodulation, de novo synthesis of nucleotides has not yet reached its maximum activity (Coleto et al., 2016) and purine salvage enzymes could be required for synthesizing nucleotides for nodule development. However, high PvAPRT4 and PvAPRT5 expression during early nodulation was not followed by elevated activity levels, at least with adenine as a substrate. Notably, PvAPRT1, showing constant expression throughout nodule ontogeny, also displayed the highest activity towards adenine. Thus, while PvAPRT1 may catalyse adenine salvage in nodules, the expression of PvAPRT5 at early nodulation suggests a more specific role for this isoform in nodule development.

The precise targeting of sequences by CRISPR/Cas9 editing (Zhang et al., 2017; Zhang et al., 2019) together with hairy roots transformation could be used to investigate the specific roles of gene family members in legume roots (Voß et al., 2022; Alamillo et al., 2023). CRISPR/Cas9 targeted mutations in PvAPRT1 and PvAPRT5 resulted in specific down-regulation of the corresponding genes, whereas the post-transcriptional RNAi silencing of the same sequences also affected the expression of the other PvAPRTs, highlighting the higher specificity of CRISPR/Cas9 gene editing compared with RNAi silencing (Fig. 4; Supplementary Fig. S8).

PvAPRT1 knockout roots exhibited a significant accumulation of adenine (Fig. 5A). This coincided with a decrease (up to 50%) in adenine salvage by APRT activity (Fig. 4C). However, although PvAPRT5 editing resulted in a greater reduction in its expression compared with targeted PvAPRT1, the accumulation of adenine was minimal in the mutant PvAPRT5 roots. Furthermore, activity with adenine in the PvAPRT5 edited roots was less reduced than in the mutant PvAPRT1, suggesting that the APRT1 isoform might maintain the adenine salvage in PvAPRT5 mutants. These results confirm a more significant role of the APRT1 than the APRT5 isoform in adenine salvage to AMP.

Additionally, specific down-regulation of the two isoforms resulted in distinct metabolomic changes in the mutant roots, indicating functional differences of APRT1 and APRT5 proteins. Studies with Arabidopsis APT1 showed a role in the conversion of active cytokinin nucleobases to inactive cytokinin nucleotides (Moffatt et al., 1991; Allen et al., 2002; Zhang et al., 2013). Our results show that transgenic roots with targeted PvAPRT5 accumulated more N⁶-benzyladenine and zeatin-glucoside cytokinins than control roots (Fig. 5; Supplementary Table S4). Moreover, PvAPRT5 showed higher activity with zeatin than PvAPRT1 (Fig. 5C), confirming that, although both isoforms can act on adenine and cytokinin as substrates, they do so with differential affinities.

Differences in the metabolomic profiles in PvAPRT1 and PvAPRT5 mutant roots agree with functional specialization of the two isoforms. Besides adenine, mutant APRT1 roots accumulated amino acids, hydroxylamine, urea, and caffeic acid, whereas mutant APRT5 roots accumulated cytokinin derivatives and stress-related compounds, including abscisic acid, salicylic acid, and secondary metabolites such as flavonoids, isoflavones, glycosides, and the carbohydrates raffinose, mannose, and erythrose. Moreover, mutant APRT5 roots accumulated AMP and inosine, whereas adenine level was as in controls (Fig. 5A), thus suggesting that adenine salvage is not the main functional role of the PvAPRT5 isoform.

Regulation of cytokinin levels is crucial for proper nodulation (Frugier et al., 2008). The overexpression of cytokinin receptor histidine kinase 1 (LHK1) induced nodule primordium formation in Lotus japonicus even in the absence of rhizobia (Bauer et al., 1996; Tirichine et al., 2007). Conversely, cytokinin oxidase/dehydrogenase 3 (CKX3) mutants exhibited cytokinin accumulation and inhibition of nodulation (Reid et al., 2016). In our study, a significant proportion of roots with decreased expression of APRTs failed to develop nodules (Fig. 6C). Conversely, nodules formed in the PvAPRT1 and PvAPRT5 targeted roots were larger than those on the control hairy roots, with a more pronounced effect in PvAPRT5 mutants (Fig. 6D, E). These findings, along with PvAPRT5 expression at early nodulation, suggest that this isoform plays a role in nodule development by regulating cytokinin levels.

Cytokinin levels are also crucial for root development (Skoog and Miller, 1957). Mutants of Arabidopsis APT1 accumulated cytokinin N-glucosides, resulting in inhibited root growth (Moffatt et al., 1991; Zhang et al., 2013). In our study, transgenic PvAPRT1, and particularly PvAPRT5, roots were shorter than the control ones (Fig. 6F). These effects may be due to inappropriate cytokinin/auxin levels (Cary et al., 1995; Ruzicka et al., 2009). Elevated cytokinin levels disrupt the expression of PIN genes and inhibit the formation of the auxin gradient in Arabidopsis (Laplaze et al., 2007). Our metabolomic analysis revealed lower levels of the auxin indole acetic acid in the transgenic roots (Supplementary Table S4). This effect is consistent with the higher accumulation of cytokinins and could explain the observed reduction in root length in the PvAPRT mutant roots.

Changes in the expression of cytokinin response regulators (ARRs) and cytokinin activation lonely guy (LOG) genes in the transgenic APRT roots (Fig. 6G) further support that the effects on nodule size and root length are related to the accumulation of cytokinins in these roots. ARRs constitute a large family of cytokinin signalling proteins, in which A-type ARRs function as negative regulators of cytokinin signalling and B-type ARRs function as transcriptional activators of cytokinin responses (Hwang and Sheen, 2001; Kiba et al., 2002; Müller and Sheen, 2007). In the targeted APRT roots, the expression of B-type ARR2 remained unchanged, while A-type ARR15 expression increased, consistent with their role as negative regulators that restrict the effects of cytokinin accumulation in these roots. Conversely, the expression of LOG was lower in transgenic roots. The LOG enzyme catalyses the direct activation of cytokinins (Kuroha et al., 2009), opposite to the reaction catalysed by APRT. Therefore, lower LOG expression agrees with the reduced cytokinin inactivation by mutant APRT activity.

Conclusion

In summary, CRISPR/Cas9 editing and metabolite changes in the mutant hairy roots provided evidence that, while PvAPRT1 and PvAPRT5 have partially redundant functions catalysing the conversion of adenine to AMP and inactivating cytokinins, they also have a high degree of specialization. APRT1 appears more involved in adenine salvage, whereas APRT5 plays a greater role in regulating cytokinin activity and root and nodule development. Furthermore, our findings suggest that APRTs may have unexplored functions in plant stress responses.

Supplementary data

The following supplementary data are available at JXB online.

Fig S1. Phylogenetic distances of PvAPRT proteins.

Fig S2. Phylogenetic comparison of APRT proteins from three P. vulgaris genotypes.

Fig S3. Overexpression of HA-Strep and YFP-PvAPRT proteins in Nicotiana benthamiana.

Fig S4. Determination of the kinetic parameters of P. vulgaris APRT isoforms.

Fig S5. Schematic representation of the cis-regulatory elements in the promoter sequences of PvAPRT1, 2, 4, and 5 genes.

Fig S6. Targeted mutations detected in common bean roots and induced by CRISPR/Cas9 technology.

Fig S7. Analysis of PvAPRT1 and PvAPRT5 expression in individual P. vulgaris hairy roots.

Fig S8. PvAPRTs relative expression and enzymatic activity.

Fig S9. PvAPRTs relative expression in nodules.

Table S1. List of primers used in this study.

Table S2. Metabolite annotation and documentation for GC- and LC-MS data.

Table S3. Physicochemical properties of APRT proteins from P. vulgaris cv. Great Northern ‘Matterhorn’ (PMB-0220).

Table S4. Metabolomic changes from P. vulgaris CRSIPR/Cas9-APRTs transgenic roots.

Author contributions

JMA and CMLV designed and performed the research, and wrote the MS. SA, FJMR, and ARF carried out the metabolomic analysis and contributed to MS editing. PP performed the confocal microscopy analysis. JMA and ARF, obtained financial support for the project. All authors approved the last version of the MS.

Conflict of interest

Authors declare no conflicts of interest.

Funding

This work was supported by Grants PID2020-117966RB-100. Ref. /AEI/10.13039/501100011033 (Ministerio de Ciencia e Innovación, Spain), 1380769-R and P20_00440 (Consejería de Transformación Económica, Industria, Conocimiento y Universidades, Junta de Andalucía), BIO-115 (UCO Programa Propio) and Donacion_351/20 (Fundación Torres Gutiérrez, Spain). ARF and SA are supported by the European Union’s Horizon 2020 research and innovation program, under PlantaSYST (SGA-CSA no. 739582 under FPA no. 664620) and project INCREASE (GA 862862). CMLV was supported by fellowships from UCO Programa Propio and EMBO Scientific Exchange Grant. FJMR was supported by Deutsche Forschungsgemeinschaft Project FE552/39-1and by a ‘Margarita Salas’ post-doctoral fellowship (UCOR02MS) from the University of Córdoba (Requalification of the Spanish university system) from the Ministry of Universities financed by the European Union (NexGenerationEU). Funding for open access was provided from Universidad de Córdoba/CBUA.

Data availability

Data supporting the findings of this study are available within the manuscript and supplementary data. JMA (bv1munaj@uco.es) is the author responsible for the distribution of materials integral to the findings presented in this article.