Strategies of organic phosphorus recycling by soil bacteria: acquisition, metabolism, and regulation.

Critical to meeting cellular phosphorus (P) demand, soil bacteria deploy a number of strategies to overcome limitation in inorganic P (Pi ) in soils. As a significant contributor to P recycling, soil bacteria secrete extracellular enzymes to degrade organic P (Po ) in soils into the readily bioavailable Pi . In addition, several Po compounds can be transported directly via specific transporters and subsequently enter intracellular metabolic pathways. In this review, we highlight the strategies that soil bacteria employ to recycle Po from the soil environment. We discuss the diversity of extracellular phosphatases in soils, the selectivity of these enzymes towards various Po biomolecules and the influence of the soil environmental conditions on the enzyme's activities. Moreover, we outline the intracellular metabolic pathways for Po biosynthesis and transporter-assisted Po and Pi uptake at different Pi availabilities. We further highlight the regulatory mechanisms that govern the production of phosphatases, the expression of Po transporters and the key metabolic changes in P metabolism in response to environmental Pi availability. Due to the depletion of natural resources for Pi , we propose future studies needed to leverage bacteria-mediated P recycling from the large pools of Po in soils or organic wastes to benefit agricultural productivity.

Introduction

Phosphorus (P), an essential element for all living organisms, is required for cell membranes, genetic materials, energy carrier molecules, carbon metabolism and metabolic signalling (Westheimer, 1987; Bünemann et al., 2010). Due to limited bioavailable P in soils, soil microorganisms often live under P starvation conditions and thus rely on various strategies to access bioavailable P, by improving P assimilation and optimizing intracellular and extracellular P metabolism (Miller et al., 2010; Richardson and Simpson, 2011; Yang and Post, 2011; Grafe et al., 2018). The concentration of bioavailable P as inorganic P (Pi) is typically low in soil solution (<0.01–1 mg L−1 in highly fertile soils). Of the total P in surface soils, which generally ranges from 20 to 1000 mg P kg−1, organic P (Po) forms comprise 20%–80% (Anderson, 1980; Bünemann et al., 2010; Yang and Post, 2011). Low‐molecular‐weight phosphomonoesters, such as phytic acid, sugar phosphates and ribonucleotides, and phosphoester polymers, such as nucleic acids and phospholipids, are the predominant forms of Po in soils that originate from microbial cell death and plant debris (Turner et al., 2003; Vestergren et al., 2012; Zhang et al., 2012; Dodd and Sharpley, 2015; Hou et al., 2015; Sirois and Buckley, 2019). Synthetic phosphotriesters (also called organophosphates) in soils are derived from anthropogenic sources such as herbicides, insecticides, flame retardants and plasticizers (Gao et al., 2016; Yadav et al., 2018; Wang et al., 2019). Beyond natural and synthetic phosphoester compounds with P‐O bonds, phosphonates with P‐C bonds found in cell membrane lipids, exopolysaccharides and glycoproteins are also present in soils (Table 1) (Metcalf and Van Der Donk, 2009). Besides their natural abundance (Clark et al., 1999; Yu et al., 2013), these compounds have been widely used in agriculture and medicine contemporarily (Galezowska and Gumienna‐Kontecka, 2012; Martinez et al., 2018). There is increased recognition of the importance of soil microorganisms in playing an integral role in recycling P from the array of Po compounds found in soils (Oberson and Joner, 2005; Bünemann et al., 2010; Zhou et al., 2018). Here we present a review of the strategies employed by soil bacteria to synthesize different Po compounds as well as obtaining bioavailable P for their cellular metabolism (Fig. 1): (i) biosynthesis of Po compounds, (ii) diversity and selectivity of phosphatase enzymes, (iii) acquisition of Po compounds, (iv) regulation of phosphatases, Po acquisition and Po metabolism and (v) future research directions to be considered towards harnessing these bacterial capabilities for sustainable P management and engineered P recycling.

Table 1

Classification of bacterial phosphatases based on the type of Po substrates.

Category name	Type of bond	Substrate structure (Neutral)	EC No.	Product	Examples of enzymes	Examples of soil bacteria that produce phosphatases
Phosphoric monoester hydrolase (Phosphomonoesterase)	P‐O		3.1.3.‐	Pi	Alkaline phosphatase (ALP) Acid phosphatase (AP) Nucleotidase Phytase Sugar phosphatase Polynucleotide ‐ phosphatase phosphoserine phosphatase	ALP: γ‐proteobacteria (e.g. Pseudomonas) α‐proteobacteria (e.g. Rhizobia) Actinobacteria, Firmicutes (e.g. Bacillus), Planctomycetes, Cyanobacteria (e.g. Gloeobacter) AP: γ‐proteobacteria (e.g. Pseudomonas, Stenotrophomonas, Xanthomonas, Enterobacter, Erwinia, Lysobacter, Serratia), α‐proteobacteria (e.g. Caulobacter, Methylobacterium, Sphingomonas, Rhodobacter), Firmicutes (Bacillus, Clostridium), Bacteroidetes (Pedobacter) Phytase: α‐proteobacteria, δ‐proteobacteria, γ‐proteobacteria (e.g. Pseudomonas), Bacteroidetes, Cyanobacteria, Actinobacteria, Firmicutes (Bacillus)
Phosphoric diester hydrolase (Phosphodiesterase)	P‐O		3.1.4.‐	P‐R	Exonuclease Phospholipase	γ‐proteobacteria Actinobacteria
Phosphoric triester hydrolase	P‐O		3.1.8.‐	R‐P‐R	Paraoxonase	γ‐proteobacteria (e.g. Pseudomonas, Firmicutes (e.g. Bacillus)
Triphosphoric acid monoester hydrolase	P‐O		3.1.5.‐	P–P–P	dGTPase	Common in all bacteria
Phospho‐anhydrides hydrolase	P‐O		3.6.1.‐	Pi	Adenosine‐triphosphatase Inorganic diphosphatase Nucleoside diphosphate‐phosphatase Nucleoside triphosphate‐phosphatase	Common in all bacteria
Ribonuclease	P‐O		3.1.11.‐ to 3.1.31.‐	P‐R	Exodeoxyribonuclease Exoribonuclease Exonuclease Endodeoxyribonuclease Endoribonuclease	Common in all bacteria
Enzymes hydrolyzing P‐N bonds	P‐N		3.9.1.‐	Pi	Phosphoamidase Protein arginine‐phosphatase	Common in all bacteria Firmicutes (e.g. Bacillus)
Enzymes hydrolyzing P‐C bonds	P‐C		3.11.1.‐	Pi	Phosphonoacetaldehyde hydrolase Phosphonoacetate hydrolase Phosphonopyruvate hydrolase	γ‐proteobacteria (e.g. Pseudomonas)

Enzyme Commission numbers (EC No.), shown in the table, represent the classification of enzymes based on the chemical reactions they catalyse.

Fig 1

Schematic overview of Po recycling processes employed by soil bacteria. Decomposition of plant and microbial biomass results in an accumulation of organic matter, including Po compounds, in soil. In this review, we explain the strategies of soil bacteria in recycling of P from Po coming from soil organic assemblages. As shown in the schematic, these recycling strategies include the degradation of Po compounds by intracellular and extracellular phosphatases, the upregulation of genes encoding transporters for the uptake of extracellular P and the incorporation of the scavenged P into the Po biosynthesis pathways.

Biosynthesis of common Po compounds

Cellular Pi is required for phosphorylation reactions within the central carbon metabolism and the eventual generation of precursors for the biosynthesis of Po biopolymers, namely, nucleic acids and phospholipids, as illustrated in Fig. 2. Here we follow the incorporation of Pi starting with the initial catabolism of glucose, a common carbohydrate in soils and the derivative of cellulose and starch, two common biopolymers in plant matter. Glucose is phosphorylated to produce glucose‐6‐phosphate (G6P) by a hexokinase or via the phosphoenolpyruvate (PEP)‐dependent phosphotransferase system (PTS). Hexokinases transfer the phosphoryl group of adenosine triphosphate (ATP) to glucose or other hexose sugars, such as mannose, fructose, or galactose (Cohen, 2014; Prakasham and Kumar, 2019). In the PTS system, PEP is the phosphoryl donor for substrate phosphorylation (Deutscher et al., 2014; Jeckelmann and Erni, 2020). Following the phosphorylation of the sugar, the phosphorylated metabolite, such as G6P, subsequently enters the pentose phosphate pathway or proceeds to glycolysis. Alternatively, glucose can also enter the gluconate pathway through which glucose is oxidated to gluconate and subsequently phosphorylated to 6‐phosphogluconate. 6‐phosphogluconate can then enter either the pentose phosphate pathway directly or glycolysis through the Entner–Doudoroff pathway (Conway, 1992; Nikel et al., 2015).

Fig 2

Overview of the different metabolic pathways involved in Po biosynthesis in bacterial cells. The metabolic pathways for the biosynthesis of precursors to essential Po biopolymers, nucleotides and phospholipids, starting from glucose are shown. Abbreviations: G6P, glucose‐6‐phosphate; R5P, ribose‐5‐phosphate; PRPP, 5‐phospho‐d‐ribosyl‐α‐1‐diphosphate; IMP/UMP, inosine/uridine monophosphate; CTP, cytidine triphosphate; GAP, glyceraldehyde‐3‐phosphate; G3P, glycerol‐3‐phosphate.

Within the pentose phosphate pathway, following oxidation and decarboxylation reactions, G6P generates ribose‐5‐phosphate (R5P). After phosphorylation with pyrophosphate derived from ATP, R5P becomes 5‐phospho‐d‐ribosyl‐α‐1‐diphosphate (PRPP), which is an important precursor to the biosynthesis of purine and pyrimidine nucleotides, the amino acids histidine and tryptophan, and the cofactors nicotinamide adenine dinucleotide (NAD) and NAD phosphate (NADP) (Cohen, 2014; Hove‐Jensen et al., 2017). Both purine and pyrimidine nucleotides are essential to biomass growth as they serve as building blocks for RNA, DNA and the energy carrier molecules (ATP, NAD and NADP).

Purine nucleotides, such as adenosine‐5′‐monophosphate (AMP) and guanosine‐5′‐monophosphate, can be synthesized either via de novo or salvage pathways. Inosine monophosphate (IMP) is an intermediate of the de novo purine nucleotide biosynthesis and is synthesized by gradually assembling the purine ring on the C‐1 of the R5P moiety in PRPP (Kilstrup et al., 2005; Hove‐Jensen et al., 2017). In the salvage pathway, several enzymes are involved in adding the phosphoribose R5P to recycled nucleobases to produce monoribonucleotides (e.g. synthesis of AMP by phosphoribosylating adenine) (Hove‐Jensen et al., 2017). Further phosphorylation steps on the monoribonucleotide would then produce triphosphorylated ribonucleotides such as ATP, which serves as a phosphoryl or nucleotidyl donor within the cell, or guanosine triphosphate, which also acts as a phosphoryl donor (Walsh et al., 2018).

Pyrimidine nucleotides, such as uridine‐5′‐triphosphate (UTP) and cytidine‐5′‐triphosphate (CTP), are also synthesized via de novo and salvage pathways. De novo synthesis of pyrimidine begins with carbamoyl phosphate, a metabolite generated by phosphorylating and aminating dissolved CO2. Follow‐up reactions involve amino acid addition to and subsequent cyclization of carbamoyl phosphate to eventually generate the pyrimidine ring dihydroorotate (Cohen, 2014). Orotate, which is generated from dihydroorotate, reacts with PRPP to produce orotidine 5′‐monophosphate and, after decarboxylation, uridine‐5′‐monophosphate (UMP) (Cohen, 2014). After two phosphorylation steps, UMP yields UTP, which is the precursor to the synthesis of CTP and subsequent synthesis of pyrimidine compounds (Kilstrup et al., 2005). The pyrimidine salvage pathway, unlike the purine salvage pathway, involves a single phosphoribosyltransferase, uracil phosphoribosyltransferase, which utilizes uracil from nucleic acid catabolism or supplied from the environment (Hove‐Jensen et al., 2017).

The biosynthesis of phospholipids relies on the combination of CTP, a pyrimidine ribonucleotide, with glyceraldehyde‐3‐phosphate (GAP), a glycolytic metabolite. The transfer of the acyl group from GAP to the sn‐1 and sn‐2 positions of glycerol‐3‐phosphate (G3P) by the G3P acyltransferase yields 1,2‐diacylglycerol‐sn‐G3P (Sohlenkamp and Geiger, 2015). This latter metabolite further reacts with CTP to form cytidine diphosphate‐diacylglycerol, which is a common precursor to various phospholipids with an assortment of different headgroups that compose the cell membrane (Parsons and Rock, 2013; Sohlenkamp and Geiger, 2015). In the following section, we discuss the diversity and selectivity of phosphatase enzymes employed by soil bacteria to hydrolyze different Po compounds to achieve Pi acquisition.

Diversity and selectivity of phosphatases for Po hydrolysis

During Pi deficiency, soil bacteria produce phosphatase enzymes to recycle Pi from heterogenous Po mixtures for their own usage and to help the plant growth (Fitriatin et al., 2011; Kumar et al., 2016). Phosphatases catalyse the hydrolytic cleavage of a bond between P and another atom (Table 1). The majority of phosphatases are involved in hydrolyzing the P‐O bonds present in most Po compounds found in soils. Phosphomonoesterases (Enzyme Commission number, EC 3.1.3.‐) represents the most abundant class of extracellular phosphatases released by soil bacteria to release Pi from phosphomonoesters (Sharma et al., 2013; Margalef et al., 2017). Alkaline phosphatases, i.e. ALP, (EC 3.1.3.1), acid phosphatases, i.e. AP, (EC 3.1.3.2), and phytases (EC 3.1.3.8 and 3.1.3.26) are the predominant forms of extracellular phosphatases in soils (Eivazi and Tabatabai, 1977; Hui et al., 2013; Sharma et al., 2013). ALPs, the most well‐studied group, are encoded by phoA, phoD and phoX genes (Neal et al., 2018). The PhoD and PhoX ALPs were reported to be more abundant than PhoA in marine and terrestrial ecosystems (Luo et al., 2009; Sebastian and Ammerman, 2011; Ragot et al., 2015), with PhoDs having the highest abundance in soils (Luo et al., 2009; Tan et al., 2013; Ragot et al., 2015). In addition to their phosphomonoesterase activities, most PhoDs exhibit phosphodiesterase activity, e.g. PhoDAP from Aphanothece halophytica (Kageyama et al., 2011). Based on the signal peptide attached to the protein sequences, it was concluded that PhoD and PhoX ALPs are translocated to the cell membranes (Pop et al., 2002; Monds et al., 2006), whereas the PhoAs are secreted extracellularly or to the periplasm (Kim and Wyckoff, 1991; Monds et al., 2006). A metagenomics study (Luo et al., 2009) reported a 40% higher abundance of cytoplasmic ALPs than extracellular ones in marine bacterial communities. It remains to be determined how the cellular localization of ALPs varies in soil ecosystems.

A soil metagenome analysis for bacterial phosphatase genes showed a higher abundance of ALPs compared to APs in soil (including rhizosphere and bulk soil) (Lidbury et al., 2017a), whereas when considering only the rhizosphere region, the number of gene copies for APs was found to be 10‐ to 100‐fold higher than ALP genes (Fraser et al., 2017). Bacterial APs, known as non‐specific acid phosphatases (NSAPs), are active towards a broad range of substrates harbouring different structures (Rossolini et al., 1998). These APs are categorized into three classes A, B and C; the abundance of classes A and C APs was shown to be several times higher than class B APs in different soil types (Neal et al., 2018). While class A and B APs are mostly periplasmic enzymes, class C APs are associated with the bacterial outer membrane, providing higher accessibility to the Po present in the soil and thus making class C APs more advantageous for soil bacteria than the other classes (Neal et al., 2018).

Phytases are phosphatases capable of degrading phytic acid, the primary P storage in plants and grains (Table 1) (Mullaney and Ullah, 2003; Lei et al., 2013); phytases can also hydrolyze other Po compounds including ribonucleotides and sugars phosphates (Solhtalab et al., 2020, Sariyska et al., 2005, Casey and Walsh, 2003, George et al., 2007, Greiner et al., 2009). Based on their mechanism in the sequential hydrolysis of phytic acid, phytases are grouped into 3‐phytases (EC 3.1.3.8) and 6‐phytases (EC 3.1.3.26). Only a few bacterial phytases are grouped as 6‐phytases, including the periplasmic phytase of Escherichia coli, one of the most studied bacterial phytases (Lim et al., 2000; Menezes‐Blackburn et al., 2013). Most bacterial phytases are classified as 3‐phytases, a group of enzymes whose gene abundance was shown to be 4.5‐fold higher in Pi‐deficient soil relative to Pi‐rich soil (Yao et al., 2018). Considering their structural differences, bacterial phytases can be classified into four subcategories, namely, histidine acid phosphatases, β‐propeller phytases (BPPs), protein‐tyrosine phosphatase‐like phytases (cysteine phosphatase) and purple acid phosphatases (PAP) (Lei et al., 2013). Among those, BPPs are reported to be the predominant bacterial phytases secreted in the soil near the rhizosphere (Lim et al., 2007; Jorquera et al., 2013). To obtain an accurate evaluation of the bacterial phytase diversity in soil ecosystems, new metagenomic studies of the diversity of phytases based on the updated genomic databases of soil bacteria are still warranted (Lim et al., 2007).

In addition to the three main bacterial phosphomonoesterases mentioned above, PAP‐like sequences were also found in genomes of 43 bacteria and four cyanobacteria species (Yeung et al., 2009; Bhadouria and Giri, 2021). Based on homology models, bacterial PAPs are similar structurally to mammalian PAPs in the vicinity of their active sites (Schenk et al., 2000). However, the structures of bacterial PAPs have not been determined yet, and only a few have been characterized for their physiological roles and biochemical properties (Yeung et al., 2009; Zhu et al., 2019; Forrellad et al., 2020). Interestingly, bacterial PAPs from Burkholderia pyrrocinia (PAP9), B. cenocepacia (BcPAP) and Mycobacterium tuberculosis (Rv2577) have alkaline pH optima unlike previously characterized PAPs from plants and mammals, which are optimally active at the pH range of 4–7 (Yeung et al., 2009; Zhu et al., 2019; Forrellad et al., 2020).

The hydrolysis of phosphodiesters requires the activity of a group of enzymes called phosphodiesterase (EC 3.1.4.‐), such as nuclease and phospholipases (Table 1). Phosphomonoesters are the product of these reactions and will subsequently be hydrolyzed to Pi by phosphomonoesterase activities mentioned earlier (Table 1) (Turner and Haygarth, 2005; Bünemann, 2008). Similar to phosphomonoesterases, elevated gene abundance of phosphodiesterases such as exoribonuclease and sphingomyelin phosphodiesterase is found in Pi‐deficient soils (Yao et al., 2018). In response to the presence of the toxic phosphotriesters (such as in paraoxon, a metabolite of the insecticide parathion), soil bacteria have evolved enzymes, called phosphotriesterase or organophosphate hydrolase (encoded by opd gene) (Table 1) (Ali et al., 2012; Khalid et al., 2016), to detoxify these compounds by breaking the P‐O bond (Theriot and Grunden, 2011). Other P‐O hydrolases are intracellular enzymes, which are responsible for catalysing the dephosphorylation processes related to regulations, signalling pathways and motility (Zuo and Deutscher, 2001; García‐Caballero et al., 2017). The C‐P lyases were historically known as the only enzymes that could break down the P‐C bonds in phosphonates, but other hydrolase enzymes, such as phosphonopyruvate hydrolase and phosphonoacetate hydrolase (Agarwal et al., 2011), can also target the stable P‐C bonds (Quinn et al., 2007; Agarwal et al., 2011). The P‐N bonds, which are mostly found in histidine‐ or arginine‐phosphorylated proteins, can be lysed by intracellular phosphatases primarily involved in protein dephosphorylation (Chou et al., 2005), thus not likely relevant to the Pi acquisition of soil bacteria from the extracellular region. The diverse categories of extracellular, periplasmic and membrane‐associated phosphatases mentioned here are produced by various soil bacterial species present to facilitate the Pi recycling from Po‐containing natural organic assemblages.

In relation to the presence of high microbial activity in the rhizosphere, phosphatase activities are reported to be higher in the rhizosphere than the bulk soil (Spohn and Kuzyakov, 2013). There is a higher abundance of AP genes in rhizosphere bacteria than in bulk soil (Spohn and Kuzyakov, 2013; Fraser et al., 2017). Among bacterial NSAPs, only class A and C are present in the environment, whereas class B enzymes are from bacteria associated with eukaryotic hosts (Neal et al., 2018; Udaondo et al., 2020). While class A NSAPs are mainly produced by Gram‐negative bacteria (including Pseudomonas, Caulobacter, Stenotrophomonas, Methylobacterium, Sphingomonas and Xanthomonas species (Neal et al., 2018), class C NSAPs are found in a broad range of soil bacteria including Bacillus, Clostridium, Enterobacter, Erwinia, Lysobacter, Pedobacter, Pseudomonas, Rhodobacter and Serratia species (Neal et al., 2018). Unlike APs, ALP genes are mostly found in bacterial communities in the bulk soil instead of the rhizosphere (Spohn and Kuzyakov, 2013). Among the diverse range of phyla producing ALPs (Sebastian and Ammerman, 2009; Ragot et al., 2017), Proteobacteria and Actinobacteria were the most abundant communities harbouring both phoD and phoX genes, followed by Firmicutes for phoD and Planctomycetes for phoX genes (Ragot et al., 2017; Lidbury et al., 2021). Particularly, phoD (the most relevant ALP genes in soil bacteria) was found in the orders Actinomycetales, Bacillales, Gloeobacterales, Rhizobiales and Pseudomonadales (Tan et al., 2013; Ragot et al., 2015; Wan et al., 2020).

A diverse range of bacteria including α‐proteobacteria, δ‐proteobacteria, γ‐proteobacteria, Bacteroidetes, Cyanobacteria and Actinobacteria were found to possess the phytase‐like genes (Jorquera et al., 2008). As the most abundant bacterial phytase, BPP genes were detected in the strains of Pseudomonas and Bacillus from plant roots (Jorquera et al., 2012), including phytases from Bacillus amyloliquefaciens (Shim and Oh, 2012; Boukhris et al., 2015), B. laevolacticus (Gulati et al., 2007), B. subtilis (Lim et al., 2007) and Pseudomonas sp. FB15 (Jang et al., 2018). In addition to BPPs, the genes and the phytase activity related to histidine acid phosphatases and PAPs were found in Acinetobacter and Buttiauxella sp. of soil samples (Rix et al., 2020).

Although there is little information about phosphodiesterase enzymes related to soil bacteria, a community proteogenomics study found that archaea and γ‐proteobacteria are the major producers of phosphodiesterases particularly in response to Pi deficiency (Yao et al., 2018). Moreover, it was shown that a considerable number of soil bacteria mainly from Actinobacteria and proteobacteria phyla harbour the genes for phospholipase enzymes. For instance, P. fluorescens can catalyse all the degradation steps of glycerophosphodiesters extracellularly using their glycerolphosphodiesterases, GlpQII (Lidbury et al., 2017b). Due to their relevance to biotechnological applications, phosphotriesterases were characterized from several soil bacteria, such as P. diminuta, OPH (Gorla et al., 2009), P. putida (Mulbry and Karns, 1989; Khalid et al., 2016), Agrobacterium, OpdA (Horne et al., 2002), Flavobacterium, OPH (Mulbry and Karns, 1989; Horne et al., 2002) and Burkholderia, OphB (Taesung et al., 2007). With respect to phosphonatase enzymes for breaking the P‐C bonds, Pseudomonas species such as P. fluorescens are the notable soil bacterial species shown to produce phosphonatases, PhnA (Panas et al., 2006).

Soil physicochemical properties play an important role in influencing both the bacterial communities producing phosphatases (Marschner et al., 2004; Margalef et al., 2017; Ragot et al., 2017) and influencing the stability and activity of phosphatases (Fig. 3) (Allison, 2006; Schimel et al., 2017). For instance, expectedly, ALP abundance was higher in grassland soils with alkaline pH, than in forest soils with acidic pH (Ragot et al., 2017). Based on a soil metagenome analysis (Lidbury et al., 2017a), the number of bacterial AP genes remained the same in both acidic and alkaline pH values, but the number of ALP genes decreased significantly at low pH. Similar to APs, the number of phytase and phosphodiesterase genes has been shown to remain constant in response to high and low pH (Lidbury et al., 2017a). In addition to pH, the type of crops is one of the main factors determining the type of phosphatases and their bacterial producers in agricultural soils. For example, α‐proteobacteria Sphingopyxis and Asticcacaulis and the β‐proteobacteria Ralstonia and Cupriavidus that harbour phoD genes were more associated with sorghum rhizosphere, whereas α‐proteobacterium Bosea and β‐proteobacterium Achromobacter with NSAP genes were found more in maize rhizosphere (Neal et al., 2021). Moreover, it was shown that bioavailability of enzyme cofactors can determine the abundance of phosphatase genes in soil. For instance, presence of Ca2+ can facilitate high abundance of ALP (phoD and phoX) and BPP genes, whereas the absence of Ca2+ is correlated with the abundance of NSAPs in soil (Neal and Glendining, 2019). Interestingly, while the relative abundance of Po compounds was negatively correlated with the abundance of phoX‐harbouring Actinobacteria, it had no effect on proteobacteria (Ragot et al., 2017) and was correlated positively with the abundance of all phoD‐harbouring communities (Ragot et al., 2017; Luo et al., 2019). Furthermore, it was shown that introducing high Po and organic matter via long‐term manure supplementation enhanced the gene abundance of ALPs, phytases, exopolyphosphatases and pyrophosphatases in soil more than inorganic fertilization (Ma et al., 2020). Among ALP genes, phoX genes were found in fewer ecosystems than phoD genes (Ragot et al., 2015). Moreover, the bacterial communities producing PhoD are different in different types of soil. For instance, in vertisol and planosol (both rich in clay minerals), proteobacteria are the dominant phoD‐associated bacteria, whereas the Deinococcus‐Thermus and Firmicutes were predominant in leptosol (extremely gravelly and/or stony soil) (Ragot et al., 2017). Therefore, depending on the type of soil in the rhizosphere of a specific crop, a combination of different environmental factors such as land use, pH and nutrient concentrations (Po, Pi, total carbon and nitrogen) can determine the type of bacterial communities harbouring phosphatase genes in the rhizosphere (Margalef et al., 2017; Ragot et al., 2017; Sun et al., 2020).

Fig 3

Overview of the diversity of bacterial phosphatases in soils and the influence of soil characteristics on enzyme activity.

A. Soil bacteria can produce different categories of phosphatases shown by different colours.

B. Bacterial phosphatases can be transmembrane, periplasmic, or extracellular.

C. Different physiochemical properties of the soil can affect the activity and stability of bacterial phosphatases, including pH of the soil, concentration of present Pi and Po compounds, type of the soil which is determined by presence of different minerals and organic matter.

Kinetic parameters of phosphatases vary depending on the type of phosphatase (Hui et al., 2013), the origin of the enzyme (Hui et al., 2013), soil type (Giaveno et al., 2010), soil depth (Stone and Plante, 2014) and the extent of P amendments (Zhang et al., 2018). While similar half‐saturation constants (K m) were found for ALPs and APs across different types of soil, higher maximum initial rates (V max) were obtained for APs. Moreover, phosphatases from soil bacteria had lower V max than plant phosphatases. Compared to the purified enzymes from soil microorganisms, a lower substrate affinity (higher K m) and lower catalytic activity (lower V max) were observed for soil‐associated phosphatases (Hui et al., 2013). In deep soils compared to surface soil, kinetic parameters of soil‐associated phosphatases (V max and K m) decreased. However, while phosphatase catalytic efficiency was maintained in the first meter of soil profiles, it decreased by 50% at 140 cm soil depth (Stone and Plante, 2014).

The catalytic efficiency of soil‐associated phosphatases is dependent on their structural stability, which is influenced by soil environmental factors such as pH, temperature, adsorption to minerals and organic matter. The optimal activity of APs was found at pH 5.2 and ALPs at pH 9.2 (Hui et al., 2013). Among different minerals, montmorillonite was shown to be destructive to the catalytic activity of ALPs (Zhu et al., 2016). Binding to goethite (an iron‐oxyhydroxide mineral) also led to a decrease in the activity of ALP (Zhu et al., 2016), but the catalytic activity of AP remained unaffected (Olsson et al., 2012). Phytase was also adsorbed greatly to different minerals surfaces (Giaveno et al., 2010). Among those minerals, kaolinite and montmorillonite caused the greatest activity loss upon adsorbing phytases. However, the clays with heterogeneous surfaces, such as oxisol clays, could enhance the phytase stability in soil (Giaveno et al., 2010). Consistent with the positive effect of heterogeneous mixtures, it was shown that the addition of organic matter, such as biochar amendments, also increased ALP activity up to fourfold (Khadem and Raiesi, 2019). Furthermore, ALP immobilization on montmorillonite and goethite was shown to protect the enzyme against heavy metal inhibition (Wang et al., 2017; Tan et al., 2018).

In addition to gene expressions in the bacterial community, the abundance of different forms of P in the soil can affect the catalytic activity of the phosphatases (Zhang et al., 2018). The addition of Pi inputs to the Pi‐deficient forest soil led to a decrease in soil phosphatase catalytic efficiency (i.e. V max/K m) due to inhibition of enzymes by Pi. However, in Pi‐rich soils where the initial catalytic efficiency of phosphatases was higher, the extra addition of Pi inputs to the soil further increased the phosphatase efficiency (Zhang et al., 2018). The strategies evolved by bacteria for the uptake of the released Pi from Po after extracellular phosphatase activity as well as the strategies developed for the direct uptake of Po compounds will be discussed in the following section.

Acquisition of Pi and Po compounds by soil bacteria

Upon hydrolysis of Po compounds by the activity of phosphatases discussed in the previous section, the resulting Pi is assimilated by soil bacteria primarily via phosphate inorganic transporters (Pit) and the phosphate‐specific transport (Pst) system (Wanner, 1996). Primarily studied in E. coli, the low‐affinity Pit transporters assimilate Pi under Pi‐replete conditions and the high‐affinity Pst system is induced under Pi‐depleting conditions (Hoffer et al., 2001a; Hsieh and Wanner, 2010; McCleary, 2017). Similar trends of transitioning to the Pst system with reduced Pi availability were observed in other soil bacteria such as Bacillus species, Burkholderia cenocepacia, Pseudomonas aeruginosa, Sinorhizobium meliloti, Streptomyces coelicolor and Staphylococcus aureus (Qi et al., 1997; Voegele et al., 1997; Hoshino, 1998; Antelmann et al., 2000; Saier Jr et al., 2002; Hoi et al., 2006; Voigt et al., 2006; Martín et al., 2011; Kelliher et al., 2018; Barreiro and Martínez‐Castro, 2019; Shropshire et al., 2021). Though Pi is the most favourable form of P to bacteria, soil bacteria are also capable of assimilating Po compounds such as organophosphates and phosphonates (Table 1).

The organophosphate:phosphate antiporters (OPA) family allows the transport of various phosphorylated metabolites, such as G3P, G6P, 2‐ or 3‐phosphoglycerate, or PEP (Elvin et al., 1985; Lemieux et al., 2004a; Albermann et al., 2014). These Po metabolites represent intermediates in the glycolytic pathway in cellular carbon metabolism. The G3P transporter (GlpT) and hexose‐6‐phosphate transporter:antiporter (UhpT), both of which are members of the OPA family, are well‐studied in E. coli (Winkler, 1966; Ambudkar et al., 1990; Brzoska et al., 1994; Fann and Maloney, 1998; Auer et al., 2001; Huang et al., 2003; Lemieux et al., 2004a). Homologues to the glpT gene have been identified in B. subtilis and P. aeruginosa as well (Castañeda‐García et al., 2009). In Salmonella typhimurium, the phosphoglycerate transporter protein (PgtP), another member of the OPA, catalyses the transport of 2PGA, 3PGA and PEP (Varadhachary and Maloney, 1991). In addition, E. coli has the Ugp (uptake of glycerol phosphate) system, an alternative transport system for glycerophosphoryl diesters, encoded by the ugpBAECQ operon (Schweizer et al., 1982; Brzoska and Boos, 1988). Corynebacterium glutamicum lacks genes homologous to those encoding GlpT, UhpT and PgtP but contains homologues of the genes encoding the Ugp system (Ishige et al., 2003). Reports of C. glutamicum utilizing G3P as a sole P source led to the conclusion that G3P is taken up by the Ugp system encoded by its ugpAEBC operon (Eggeling and Bott, 2005; Wendisch and Bott, 2005; Lindner et al., 2012). In addition to the aforementioned transporters in the OPA family, other novel bacterial transporters that may be involved in Po utilization have been reported. For instance, a recent study that performed a comparative genomics analysis on eight Flavobacterium strains detected the expression of six distinct phosphate utilization system transporter complexes (PusCD) and proposed that these complexes may target phosphodiesters in soil, most likely lipid headgroups and nucleic acids (Lidbury et al., 2021).

Phosphonates (with C‐P bonds), found in soil from both biogenic and anthropogenic (e.g. herbicides) sources, can be a potential P source to soil bacteria subjected to Pi deficiency (Hilderbrand and Henderson, 1983; McGrath et al., 2013; Hove‐Jensen et al., 2014). The C‐P lyase pathway involving the phosphonate transporter (Phn) system is considered a major phosphonate catabolism route in bacteria and is well‐studied in E. coli (McGrath et al., 2013; Hove‐Jensen et al., 2014). The 14‐gene operon, phnCDEFGHIJKLMNOP, in E. coli encodes the proteins required for the uptake and assimilation of phosphonates via the C‐P lyase pathway with the first three genes (phnCDE) encoding an ABC‐type phosphonate transport system (Metcalf and Wanner, 1993; White and Metcalf, 2004; Stasi et al., 2019). Via the C‐P lyase pathway, alkylphosphonates are converted to the corresponding alkanes and Pi (McGrath et al., 2013). A survey of bacterial genes revealed that the genes encoding the phosphonate transport polypeptides are found as either phnCDE or phnCDEE, in which there are two distinct transmembrane polypeptides (Hove‐Jensen et al., 2014). The survey revealed that E. coli, Burkholderia pseudomallei and P. stutzeri have phnCDE for the Phn system, whereas S. meliloti, Ochrobactrum anthropic, Agrobacterium radiobacter, Mesorhizobium loti and Bradyrhizobium japonicum have phnCDEE (Hove‐Jensen et al., 2014). Moreover, P. stutzeri has an additional operon relevant to the C‐P lyase pathway, htxABCDEFGHIJKLMN, where htxBCD encodes the ABC transport system for phosphonate (White and Metcalf, 2004; Hove‐Jensen et al., 2014). Orthologous genes to phnCDE were identified in C. glutamicum and S. aureus (Ishige et al., 2003). Unlike the Phn system activated under Pi limitation, it has been reported that Pi‐insensitive 2‐aminoethylphosphonic acid (2EAP) degradation in P. putida NG2 and BIRD‐1 (Ternan and Quinn, 1998; Murphy et al., 2021). A recent study identified the proteins relevant to 2EAP assimilation in P. putida BIRD‐1, whereby 2‐aminoethylphosphonate XVW (Aep XVW) was the primary Pi‐sensitive 2AEP transporter with AepSTU serving an auxiliary role and 2AEP permease (AepP) was essential for Pi‐insensitive growth on 2AEP (Murphy et al., 2021).

A metagenomics study on the soil microbial community grown with legume and grass found that microorganisms involved in Po mineralization were more abundant in soil grown with legume, whereas microorganisms involved in Pi solubilization (from inorganic P minerals) were more abundant in the soil grown with grass (Zhou et al., 2018). This study identified the Al and P contents as well as Po composition in the soil as the major factors of the reported differences (Zhou et al., 2018). Therefore, in addition to P availability, this study suggests that different environmental factors may influence the activation of P acquisition strategies in soil bacteria. In the following section, we discuss the regulation involved in Po acquisition and metabolism depending on the P availability. Though the acquisition of Po compounds by soil bacteria has been well studied, there is a lack of understanding on how these compounds are secreted into the soil environment beyond cell leakage or cell lysis. The secretion of GAP, dihydroxyacetone phosphate and PEP by E. coli had been reported but could not be explained due to the lack of relevant annotated transport reactions (Paczia et al., 2012; Pinu et al., 2018).

Regulation of phosphatases, Po acquisition and Po metabolism

As discussed above, bacteria can scavenge Pi from Po compounds in their environment through enzymatic action using phosphatases, including ALPs, APs, phytases, phospholipases and nucleases. We also highlighted that, in addition to Pi uptake, many bacteria can directly transport Po compounds such as organophosphates and phosphonates as alternative P sources. When the environmental Pi level is limited, bacteria increase the expression of Pi‐scavenging enzymes and several P transporter systems to increase Po acquisition (Wanner, 1996; Santos‐beneit, 2015). In addition, soil bacteria can modify their metabolic network related to the biosynthesis of Po compounds such as phospholipids, nucleotides and nucleic acids. In this section, we discuss the mechanisms employed by soil bacteria to regulate the following processes: enzyme production for Pi‐scavenging from Po, expression of Po transporters and metabolic changes in Po biosynthesis under Pi‐limiting conditions. The overview schematic of bacterial regulation of these processes is illustrated in Fig. 4.

Fig 4

Overview of bacterial regulation of the production of phosphatases, the expression of Po transporters and the metabolic changes in Po biosynthesis in response to Pi deficiency. Under Pi deficiency, the sensor protein PhoR undergoes autophosphorylation and subsequently transfers the phosphate group to phosphorylate and activate the response regulator PhoB, which regulates a set of genes in the pho regulon (relevant genes are presented in the blue box). The pho regulon controls the expression of the phosphate‐specific transport (Pst) system, the phosphonate transporter (Phn) system and the uptake glycerol phosphates (Ugp) system which respectively mediate the uptake of Pi, phosphonates (Pn) and glycerol phosphates including glycerolphosphodiesters (GPD) and glycerol‐3‐phosphate (G3P). The phosphate inorganic transporter (Pit) system is the low‐affinity transporter that primarily transport Pi under replete Pi conditions. Other Po transporters including the G3P transporter (GlpT), the hexose‐6‐phosphate transporter (UhpT) and the phosphoglycerate transporter protein (PgtP) belong to the OPA family, transporting their corresponding Po in exchange with internal Pi. The corresponding Po molecules are G3P for the GlpT system; glucose‐6‐phosphate (G6P), fructose‐6‐phosphate (F6P), mannose‐6‐phosphate (M6P) for the UhpT system; phosphoenolpyruvate (PEP), 2‐phosphoglycerate (2PGA), and 3‐phosphoglycerate (3PGA) for the PgtP system. Each of these Po transporters is controlled by a distinct regulatory system, not regulated by the pho regulon. The transcription of glpT and ptgP additionally require the complex of cyclic adenosine monophosphate (cAMP) and the cAMP receptor protein (CRP). Other relevant non‐pho regulated genes are listed in the green box. Besides the pho regulon, Pi deficiency also triggers the stringent response, signalled by the accumulation of guanosine tetraphosphate and guanosine pentaphosphate [collectively as (p)ppGpp] which inhibit many metabolic processes such as amino acid metabolism, carbohydrate metabolism, nucleotide synthesis, RNA synthesis, DNA replication, transcription and translation.

Many of the genes involved in the production of Pi‐scavenging enzymes and Po transporters are controlled by a global regulatory mechanism called the phosphate (pho) regulon (Santos‐beneit, 2015; Martín and Liras, 2021).

The pho regulon is inactive under replete Pi condition but, during Pi scarcity, the pho regulon is activated to regulate a set of genes towards maintaining an adequate supply of Pi (Wanner, 1990; Santos‐beneit, 2015; Morimoto et al., 2016; Martín and Liras, 2021). Mediated by the pho regulon, Pi is transported primarily through the Pst system upon depletion in Pi, whereas the Pit system is the primary Pi transporter when Pi is replete (Rosenberg et al., 1977; Willsky and Malamy, 1980; Qi et al., 2016; Grafe et al., 2018). Specifically, the pho regulon upregulates pstSCAB, the operon that encodes the proteins of the Pst transporter system (Wanner, 1990; Vuppada et al., 2018). Importantly, the regulation of the pho regulon is critical to the subsequent regulation of enzymes involved in Pi acquisition from Po compounds.

In bacteria, the extracellular Pi level is the responsible signal that regulates the pho regulon (Rao et al., 1993; Vuppada et al., 2018) which is mediated by a two‐component regulatory system composed of an inner‐membrane sensor protein and a transcriptional response regulator (Wanner, 1990; Santos‐beneit, 2015). First characterized in E. coli, this regulatory system of the pho regulon consists of two proteins: PhoR for the sensor protein and PhoB for the response regulator protein (Tommassen et al., 1982). These proteins are called by different names in other bacteria, for instance, the corresponding proteins in B. subtilis are PhoR and PhoP (Hulett et al., 1994; Novak et al., 1999; Kočan et al., 2006; Glover et al., 2007). When environmental Pi is abundant, PhoR remains in its repressor form, which functions as a phosphatase that keeps PhoB inactive (Wanner, 1990; Carmany et al., 2003). Under Pi limitation, PhoR undergoes autophosphorylation to be converted to the activator form, which functions as a kinase that activates PhoB via phosphorylation (Wanner, 1990; Gao and Stock, 2013). The active PhoB (i.e. phosphorylated PhoB) subsequently binds to promoter regions of the genes controlled by the pho regulon to upregulate or suppress targeted gene transcription by interacting with the RNA polymerase sigma factor RpoD in the promoter regions of those genes (Makino et al., 1993; Blanco et al., 2002). The specific DNA sequences, which are targeted by phosphorylated PhoB, are two 11‐base pair direct‐repeat sequences known as Pho boxes (Blanco et al., 2002). The consensus DNA sequences in the Pho boxes differ in different bacteria (Blanco et al., 2002; Yoshida et al., 2012; Sola‐Landa et al., 2013). Beyond Pi assimilation, the pho regulon may additionally control several genes involved in nitrogen, carbon and iron metabolism (Santos‐beneit, 2015; Millan‐oropeza et al., 2020). Bioinformatic genome analysis together with transcriptional studies on the response to Pi depletion has been used to identify novel members of the pho regulon (Baek and Lee, 2006; Lidbury et al., 2016; Ordóñez‐Robles et al., 2017; Martínez‐Castro et al., 2018). The heterogeneity in the occurrence of pho regulon members was observed at the species‐specific level within the same genus. (Lidbury et al., 2016; Ordóñez‐Robles et al., 2017; Lidbury et al., 2021).

In most cases, the following genes encoding phosphatase‐type enzymes (responsible for the hydrolytic release of Pi from P‐O containing Po compounds) are upregulated in a pho regulon‐dependent manner during Pi starvation: phoA, phoX and phoD for ALPs; phoN for AP (in S. typhimurium); phoD for phospholipase; appA and phyC for phytases; glpQ and upgQ for glycerophosphodiester phosphodiesterases, and ushA for 5′‐nucleotidase (Wanner, 1990; Kasahara et al., 1991; Russell D. Monds et al., 2006; Santos‐beneit, 2015; Lidbury et al., 2016; Martín et al., 2017; Grafe et al., 2018; Lidbury et al., 2017b). Exceptions include Streptomyces tsukubaensis whose phoA and phoD genes are not controlled by the pho regulon (Martínez‐Castro et al., 2018). The pho regulon also regulates the phnCDEFGHIJKLMNOP operon, which encodes several C‐P lyase enzymes that catalyse the release of Pi by breaking C‐P bonds (Stasi et al., 2019). In addition to phosphatase regulation by the pho regulon, a few non‐pho regulated phosphatases have been reported. For instance, the phoC gene, which encodes a class A NSAP in Morganella morganii (Thaller et al., 1994), and at least three C‐P lyases [phosphonoacetaldehyde (encoded by phnX), phosphonoacetate hydrolase (encoded by phnA), and phosphonopyruvate hydrolase (encoded by palA)] are pho regulon‐independent (Quinn et al., 2007). The regulation of the phoC gene of M. morganii has not yet been elucidated. The phnX, phnA and palA genes are regulated by distinct transcriptional regulators encoded by lysR, phnR and palR genes respectively (Quinn et al., 2007).

Several Po transporter genes that are regulated positively by the pho regulon include the Phn system for phosphonate uptake (phnCDE) and the Ugp system for the uptake of glycerophosphodiesters and G3P (ugpBAECQ) (Wanner, 1990; Santos‐beneit, 2015). Unlike the Phn system, the recently identified aminoethylphosphonate transporter AepP is Pi‐insensitive (Murphy et al., 2021) and thus is not regulated by the pho regulon. Other Po transporters such as GlpT, UhpT and PgtP systems, all of which belong to the OPA family by transporting their respective organophosphate via anion exchange with internal Pi, are generally not regulated by the pho regulon (Elvin et al., 1985; Wanner, 1996; Lemieux et al., 2004b). Despite the similar transport mechanism of these OPA‐type transporters, each is regulated by a distinct regulatory system.

In E. coli, the GlpT system, which is a G3P transporter encoded by the glpT gene is not controlled by the pho regulon (Larson et al., 1982; Wanner, 1996; Law et al., 2009). The glpT gene is a member of the glp regulon, which regulates the catabolism of glycerol, G3P and glycerophosphodiesters (Larson et al., 1992). The transcription of glpT and other genes in the glp regulon are regulated by the G3P regulon repressor (GlpR), encoded by the glpEGR operon in E. coli (Yang and Larson, 1998; Lemieux et al., 2004a; Escapa et al., 2013). GlpR repressor decreases the expression of the genes in the glp regulon by binding to the operators which are close to or overlapping with their promotor regions (Yang and Larson, 1998). The expression of GlpT is induced by extracellular G3P, which binds to the GlpR repressor to lower its binding affinity, alleviate the GlpR repression of the glpT gene and consequently increase the GlpT expression (Cozzarelli et al., 1968; Law et al., 2009). The expression of GlpT requires the presence of cyclic adenosine monophosphate (cAMP), which forms a complex with the cAMP receptor protein (CRP) and binds to the promoter region to activate the transcription of the glpT gene (Castañeda‐García et al., 2013). The glpT gene as well as other genes in the glp regulon are additionally governed by catabolite repression through products of the glycerolipid biosynthetic pathway (Lemieux et al., 2004a). Under Pi starvation conditions, G3P is mainly transported by the pho regulon‐dependent Ugp system, leading to the increase in internal Pi concentration (Xavier et al., 1995). A high level of internal Pi will stimulate the uptake of G3P by the GlpT system, which takes up G3P by exchanging out Pi, thus balancing the intracellular Pi level (Xavier et al., 1995). Unlike in E. coli, the GlpT system in B. subtilis is encoded by the glpQT operon which is induced by Pi depletion as well as glycerol (Antelmann et al., 2000; Lidbury et al., 2017b).

The UhpT transporter, which is encoded by the uhpT gene, mediates the uptake of hexose‐6‐phosphate (Sonna et al., 1988; Cattoir et al., 2020). The synthesis of UhpT transporter is tightly regulated by the products of the uhpABC regulatory genes: a response regulator UhpA, a histidine kinase UhpB and a receptor/transporter protein UhpC (Islandt and Kadner, 1993; Schwöppe et al., 2002; Västermark and Saier, 2014; Cattoir et al., 2020). The UhpB protein functions as an activator of UhpA, which regulates uhpT transcription (Weston and Kadner, 1988; Wright and Kadner, 2001). The UhpC protein has a high sequence similarity to the UhpT transporter (Islandt and Kadner, 1993) and can serve as a G6P‐specific transporter with a relatively low transport rate compared to UhpT (Västermark and Saier, 2014). The UhpC protein functions as a high‐affinity G6P receptor that signals UhpB to undergo autophosphorylation and activate the UhpA regulator in the presence of external G6P (Wright and Kadner, 2001; Västermark and Saier, 2014; Cattoir et al., 2020). Thus, the expression of the UhpT transporter is induced by extracellular G6P (Dietz and Heppel, 1971; Hoffer et al., 2001b; Castañeda‐García et al., 2013, Cattoir et al., 2020). Similar to the GlpT system, the expression of the UhpT transporter requires the cAMP‐CRP complex to bind to the promoter site of the uhpT gene (Castañeda‐García et al., 2013; Cattoir et al., 2020). The UhpT transporter has been suggested to mediate the uptake of G6P as a carbon source, not a phosphate source (Hoffer et al., 2001b). Under Pi limitation, the pho regulon upregulates the production of phosphatases which can degrade G6P in the periplasm. Consequently, much higher concentrations of the extracellular G6P are required to induce the expression of UhpT (Hoffer et al., 2001b).

In addition to transporting Po, both GlpT and UhpT have also received considerable attention due to their ability to transport fosfomycin, an antibiotic commonly used to treat urinary tract and gastrointestinal infections (Santoro et al., 2011; Kurabayashi et al., 2015; Ballestero‐Téllez et al., 2017; Hirakawa et al., 2018; Aghamali et. al., 2019; Cattoir et al., 2020). Studies on fosfomycin susceptibility in enterohaemorrhagic E. coli found that the expression of glpT and uhpT genes are positively controlled by both CRP‐cAMP and a global transcriptional regulator called FNR, which is activated under anoxic conditions (Kurabayashi et al., 2015; Kurabayashi et al., 2017).

In S. typhimurium, the PgtP protein, which is encoded by the pgtP gene, is a specific transporter for PEP, 2PGA and 3PGA (Milton H Saier Jr et al., 1975; Varadhachary and Maloney, 1991; Albermann et al., 2014; Jiang et al., 2021). The transcription of pgtP is regulated by three regulatory proteins encoded by the pgtA, pgtB and pgtC genes: a response regulator PgtA, a sensor kinase PgtB that activates PgtA and a protein PgtC that modulates the kinase activity of PgtB (Yu and Hong, 1986; Jiang et al., 1988; Niu et al., 1995). The expression of PgtP is induced by extracellular PEP, 2PGA and 3PGA (Saier et al., 1975). The two proteins PgtB and PgtC are involved in the induction of the pgtP transcription by modulating the activity of PgtA in response to inducer binding (Jiang et al., 1988; Yang et al., 1988). The PgtA regulator is thought to function as a derepressor which becomes active under the inducing condition and subsequently facilitates the transcription of the pgtP gene (Jiang et al., 1988). Recently, Jiang et al. (2021) reported a newly identified protein named VrpA as a potential positive regulator of pgtP. In response to low glucose availability, the cAMP‐CRP induces the transcription of vrpA and activates the expression of PgtP to upregulate 3PGA uptake (Jiang et al., 2021).

Besides regulation of Pi or Po acquisition, bacteria alter several intracellular processes to manage P allocation in response to different environmental P availability. Several soil bacteria remodel their lipid membranes to reduce the proportion of P‐containing lipids by replacing phospholipid with glycolipid (Geiger et al., 1999; Lidbury et al., 2016). For instance, S. meliloti was reported to replace phospholipids with phosphate‐free lipids (Geiger et al., 1999). Using proteomics analysis of soil Pseudomonas species, Lidbury et al. (2016) found that proteins involved in lipid remodelling (PlcP, DagK, OlsA, OlsB, Cfa and TauD) in the pho regulon had a higher expression when these species (P. putida, P. fluorescens and P. stutzeri) were grown under Pi limitation (Lidbury et al., 2016). In B. subtilis, Pi starvation initiates the synthesis of teichuronic acid, which is a non‐phosphate containing anionic polymer that replaces teichoic acid, a phosphate‐rich anionic polymer in the cell wall (Allenby et al., 2005; Botella et al., 2011; Fritz and Mascher, 2014; Devine, 2018). Under this condition, the active PhoP represses the genes involved in teichoic acid synthesis such as tagAB and tagDEF but upregulates the genes involved in teichuronic acid (tuaA‐H operon) (Devine, 2018). The biosynthesis of nucleotides and nucleic acids is also affected by P availability. Phosphorus starvation leads to the accumulation of guanosine tetraphosphate (ppGpp) and guanosine pentaphosphate (pppGpp), collectively called (p)ppGpp (Spira and Yagil, 1998; Sivapragasam et al., 2017). The (p)ppGpp molecules are the key signalling molecules modulating general metabolism and mediating the stringent response, the bacterial metabolic response to nutrient starvation and other stress (Hauryliuk et al., 2015). When Pi is depleted, the level of (p)ppGpp increases and triggers the stringent response to inhibit primary bacterial metabolism, including amino acid metabolism, carbohydrate metabolism, ion transport systems, and the synthesis of ribosomal RNA, and DNA replication, transcription and translation (Hauryliuk et al., 2015; Wu et al., 2020). In B. subtilis, the (p)ppGpp molecules directly inhibit DNA replication by inhibiting primase, an essential enzyme in the replication process (Wang et al., 2007). While the stringent response generally downregulates DNA and nucleotide synthesis (Traxler et al., 2008), it has been shown to promote the purine salvage pathway, which recycles purines, in S. coelicolor (Sivapragasam and Grove, 2016) and Agrobacterium fabrum (Sivapragasam et al., 2017). The functions of (p)ppGpp and the molecular mechanisms of (p)ppGpp‐mediated regulation are complex and beyond the scope of this article. These topics are presented in detail in the review article by Hauryliuk et al. (2015).

Future research directions

In the previous sections, we have presented what is known regarding Po cycling by soil‐associated or relevant bacteria, from the biosynthesis of Po compounds to the acquisition of P from Po either via the activity of phosphatase enzymes in releasing Pi or by direct Po assimilation with specific transporter systems, and the regulation involved in these processes depending on the Pi availability (Fig. 1). Despite the important role of bacteria in soil P cycling, there is limited understanding on species‐specific diversity of phosphatases, Po transporter systems and their regulation. For instance, bacterial orders such as Solibacterales, Acidobacterales and Actinomycetales, which showed significant roles in P cycling, remain poorly characterized (Bergkemper et al., 2016). Such lack of species‐specific information on soil bacteria stems from the well‐established fact that less than 1% of soil microorganisms are considered cultivable by standard cultivation methods (Rappé and Giovannoni, 2003; Vartoukian et al., 2010; Giagnoni et al., 2018). Thus, to this date, research on soil bacteria often relies on referencing model organisms such as E. coli and B. subtilis. However, some metabolic processes and regulatory mechanisms differ from species to species. For instance, the details of how the pho regulon senses Pi availability and regulates responses vary among different organisms (Santos‐beneit, 2015). The PgtP transporters, which are found in S. typhimurium are absent in the commonly studied E. coli K12 (Goldrick et al., 1988; Albermann et al., 2014). The regulation of phosphatase genes in S. tsukubaensis is considerably different from the model organism S. coelicolor (Martínez‐Castro et al., 2018). Furthermore, few studies have addressed diesterase enzymes (Yao et al., 2018), despite the importance of these enzymes in the initial degradation of P polymers such as nucleic acids. The diversity of diesterases, the bacterial communities involved in their secretion and the soil conditions that can affect their production or reactivity remain to be elucidated. To establish a detailed understanding of bacterial processes in soil P cycling, more species‐specific studies beyond the model organisms are needed. Metagenomics represents an important approach employed to gain information on uncultivated bacteria (Vartoukian et al., 2010). However, due to the small size of genomic libraries related to the genomes of the soil bacteria, there is still limited information regarding the species‐dependent diversity of APs and phytases in soil bacterial communities. Therefore, new metagenomics studies on the updated genes in the database can be useful in acquiring information about the P‐scavenging capabilities of the microbial communities present in soil environments, albeit resolving any insights on regulatory mechanisms would still be challenging with this approach. We identified the following research topics, which are imminent to improve our understanding of the role of soil bacteria in cycling Po and provide insights into the potential biotechnological applications of these bacteria in sustainable agriculture:

Study of Po metabolism and regulations of P metabolism and P acquisition in representative pure cultures of soil bacteria to obtain information beyond those provided by the studies with E. coli and B. subtilis.

Genomic, proteomic and exoproteomic studies of single and mixed cultures of soil bacteria under different P availability and P sources.

Metagenomics studies on soil bacteria and Po utilization under different soil environments, i.e. different pH, climate, land usage, P amendments, and so on.

Studies on Po metabolism and regulation by soil bacterial communities.

Determination of stability and lifespan of different classes of phosphatases in response to different soil characteristics, such as different minerals and organic matter.

Determination of kinetic parameters of soil‐associated phosphatases and the influencing factors (metal cations, mineral surface chemistry, pH).

Establishing curated databases for kinetics and stability parameters of soil phosphatases to establish accurate predictive models.

Molecular modelling and atomic‐scale studies on the interactions and reactivity of phosphatases with different mineral surfaces and organic matter complexes.

Studies on the possible mechanistic role of bacterial chemotaxis towards Pi/Po resources in the soil.