The NBRIP medium containing 10 g/L glucose, 0.25 g/L MgSO₄⋅7H₂O, 5 g/L Ca₃(PO₄)₂/sodium phytate/FePO₄/AlPO₄, 5 g/L MgCl₂⋅7H₂O, 0.2 g/L KCl, 0.1 g/L (NH₄)₂SO₄, 2 mL/L trace element solution was applied to acclimate and isolate PSB (Misra et al., 2012). Trace element solution contained (g/L) EDTA, 10; MnSO₄⋅H₂O, 2.2; FeSO₄⋅7H₂O, 1.0; CuSO₄⋅5H₂O, 0.5; CoCl₂⋅6H₂O, 0.3; Na₂MoO₄⋅2H₂O, 0.2; and CaCl₂, 0.1. The initial pH of all mediums was adjusted to 7.0 using 1 mol/L NaOH and 1 mol/L HCl solution.

Five gram of bulk soil collected from Laiyang Experimental Station in Shandong Province, China, were added to 50 mL of sterile water and shaken at 180 rpm for 30 min, and the mixture was kept stand for 10 min. In the first round of acclimation, 10 mL of soil suspension were inoculated to 100 mL of liquid NBRIP medium containing Ca₃(PO4)₂, and incubated at 30°C with shaking of 180 rpm for 7 days. The microbial suspension was collected and centrifuged at 5000 rpm for 10 min, and then microbial suspension was resuspended with 10 mL of sterile water after precipitation was washed twice. Taking the first round acclimation procedure, the microbial suspension was inoculated to liquid NBRIP containing sodium phytate, FePO₄ or AlPO₄, respectively, and incubated at the same condition for another 7 days. The acclimated microbial suspension was designated as A1 (NBRIP containing Ca₃(PO₄)₂), A2 (NBRIP containing sodium phytate), A3 (NBRIP containing FePO₄), and A4 (NBRIP containing AlPO₄), respectively. Additionally, part of the microbial suspension of A1, A2, A3, and A4 were centrifuged at 5000 rpm, and the microbial precipitations were stored at −80°C for subsequent DNA extraction.

Taking stepwise dilution strategy, 0.1 mL of 10^–5 and 10^–8 diluent of microbial suspension A4 was evenly spread on NBRIP containing Ca₃(PO₄)₂ agar medium and incubated at 30°C for 5 days. The screening strategy for isolation of PSB was by picking colony with clear halo zone. Single colonies were sub-cultured by picking and streaking five times to isolate pure colonies. A total of 18 strains (Ap-1, Ap-2, Ap-3, Ap-4, Ap-6, Ap-8, Ap-9, Ap-10, gp-1, gp-2, gp-3, gp-4, gp-6, gp-7, gp-8, gp-11, gp-12, and gp-15) were gained and identified by simple 16S rRNA gene sequencing in Wuhan Qingke innovation Biotechnology Co., Ltd.

The total DNA of microbial precipitations and soil were extracted using a DNA extraction kit (MoBio, Carlsbad, CA, United States) according to the manufacture’s instruction. The DNA concentrations were measured using a NanoDrop 2000 Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, United States). All DNA samples were stored at −80°C.

To determine the bacterial community composition of microbial precipitations (A1, A2, A3, and A4), the V3–V4 region of bacterial 16S rRNA gene was amplified using the primers 338F (5′- ACT CCT ACG GGA GGC AGC A-3′) and 806R (5′- GGA CTA CHV GGG TWT CTA AT-3′) (Mori et al., 2013). The 16S rRNA gene fragments were amplified under the following condition: an initial denaturation at 95°C for 3 min, 30 cycles of 95°C for 40 s, 58°C for 40 s, and 72°C for 50 s, and a final extension at 72°C for 10 min. The triplicate PCR products were pooled and purified by gel electrophoresis, and quantified using a QuantiFluor^TM -ST (Promega, United States). Sequencing was carried out on an Illumina Miseq platform at Majorbio Bio-Pharm Technology Co., Ltd., Shanghai, China. The Miseq raw reads were deposited in the National Center for Biotechnology Information (NCBI¹) Short Read Archive (SRA) database under accession numbers SRR8731888–SRR8731891.

The raw reads were purified following the pathway of Mothur (Schloss et al., 2009). To minimize the effects of random-sequencing errors, we removed (i) sequences that did not exactly match barcodes and primers; (ii) sequences that contained ambiguous bases call; (iii) sequences with an average quality score <20; and (iv) sequences with maximum homopolymers <10 bp. The purified sequences were clustered into operational taxonomic unit (OTU) at 97% identity against the SILVA v128 reference. The community diversity in different acclimation periods was compared using α-diversity indices including Shannon index, ACE, and Chao1.

The 18 strains were separately inoculated to LB medium and incubated at 30°C with shaking of 180 rpm for 16 h. Strains were collected by centrifuging and washed twice with sterile water, and subsequently resuspended. The optical density (600 nm) was adjusted to 1.0 by dilution of the strain pellets with sterile water. These mixtures were designated as seed cultures (OD = 1.0), and separately inoculated to NBRIP solid and liquid medium containing Ca₃(PO₄)₂, sodium phytate, FePO₄, or AlPO₄, and incubated at 30°C for 5 days. After incubation, 1 mL of bacterial suspension was collected to determine soluble P concentration, and the diameter of the halo (HD) and the diameter of colony (CD) on solid medium plates were measured (Liu et al., 2015). The PO₄^3–-P content in bacterial suspension was determined using molybdenum blue method (Waterlot, 2018).

To determine P solubilizing capacities of the 18 isolated PSB, soils used in incubation experiment were collected from an uncultivated field in Wuhan, middle of China (30°28′N, 114°21′E). The original pH, TC, TN, water soluble P (WSP), Olsen P, total P, and total lead were 6.92, 0.52%, 0.68%, 0.09 mg/g, 0.22 mg/g, 0.89 mg/g, and 0.57 mg/g, respectively. Three groups with three replicates were designed: 100 g soil + 10 mL seed cultures (S + B group), 95 g soil + 10 mL seed cultures + 5 g Ca₃(PO4)₂ (S + B + T group), 95 g soil + 10 mL seed cultures + 5 g Ca₃(PO4)₂ + 10 mL NBRIP liquid medium without P sources (S + B + T + N group). Sterile water was added to adjust soil moisture to 80%, and these experimental groups were incubated at room temperature for 30 days. After incubation, soils were collected to determine physicochemical properties, including the content of WSP and available P (AP). The WAP and AP were extracted by ultrapure water and 0.5 mol L^–1 NaHCO₃ (pH 8.5), respectively, at a ratio of 1:20 of soil to solution, and the extracted solutions were then filtered through Whatman Grade No. 42 Quantitative Filter Paper (Olsen et al., 1954), and then were measured using molybdenum blue method.

To evaluate the effect of PSB gp-1 inoculation on heavy metal immobilization, the soils containing weight ratio of 1, 2, 3, 4, 5, and 10% Ca₃(PO4)₂ were spiked with Pb(NO₃)₂ at a concentration of 500 mg Pb/kg soil and incubated at room temperature for 30 days. No strain gp-1 addition but with Ca₃(PO4)₂ addition, and without both strain gp-1 and Ca₃(PO4)₂ addition were used as blank control. Every 5 days, sterile water was added to maintain the moisture value of 80%. After incubation, soils were collected and dried at 60°C, then ground and sieved through a 0.150 mm mesh for available heavy metal evaluation. Heavy metal contaminated soils were extracted with acetic acid (HAc) solution (0.11 mol/L) with a soil to solution ratio of 1:40 and shaken at 180 rpm for 16 h (Yuan et al., 2017). The concentrations of Pb(II) in extraction solutions were measured at an AA240FS atomic-absorption spectrophotometer (Varian Company, United States). Additionally, the content of WSP and AP in heavy metal contaminated soils was determined.

The absolute abundances of P-cycling-related genes in Pb contaminated soil were measured using quantitative PCR (qPCR). Primer sequences for the target genes were applied for qPCR based on previous literatures. Primer bppF (5′–GAC GCA GCC GAY GAY CCN GCN NTN TGG–3′) and primer bppR (5′–CAG GSC GCA NRT CAN CRT TRT T–3′) were employed to amplify bpp gene (Huang et al., 2009); primer ALPS-F730 (5′–CAG TGG GAC GAC CAC GAG GT–3′) and primer ALPS-1101 (5′–GAG GCC GAT CGG CAT GTC G–3′) were applied to amplify phoD gene (Hu et al., 2018); primer gcdF (5′–CGG CGT CAT CCG GGS NTN YRA YRT–3′) and primer gcdR (5′–GGG CAT GTC CAT GTC CCA NAD RTC RTG–3′) were used to amplify gcd gene (Cleton-Jansen et al., 1990). Additionally, primer Eub338 (5′–ACT CCT ACG GGA GGC AGC AG–3′) and primer Eub518 (5′-ATT ACC GCG GCT GG-3′) (Fierer et al., 2005) were selected to amplify 16S rRNA gene. Standard curves were generated using a 10-fold serial dilution of a known amount of recombinant plasmid containing specific gene fragment. Quantitation was performed on three technical replicates with an Applied Biosystems QuantStudio3 Real-time PCR System (Applied Biosystems, Foster City, CA, United States) in a 10 μL reaction system, and was implemented at 95°C for 5 min, followed by 40 cycles of 95°C for 15 s and 55°C for 1 min. The abundances of all genes were expressed as copies per gram of freeze-dried soil. In addition, we employed these primers to amplify bpp, phoD, and gcd from Acinetobacter pittii gp-1.

Genomic DNA- extracted from strain gp-1 was used as template for amplification of the ppk and pqq genes. According to the gene sequences encoding PPK and PQQ in several related bacteria, two conservative DNA fragments were discovered with the DNAMAN program with default parameters for each gene. Primer F1 (5′–ATG ACA GAA GGT GTT GGC CT–3′) and primer R1 (5′–CTA TAT ATC TGT AAT CGT GTG GAC T–3′) were used to amplify the whole pqq gene. While primer F2 (5′–ATG AAT ACA GCG ATT ACA CCA A–3′) and primer R2 (5′–TTA TTT AAT AAT TTC TAA TAG TGC CCT TTG–3′) were employed to amplify the whole ppk gene. The acquired pqq gene had 1155 bp coding a polypeptide of 384 amino acids (accession number: MG820120). The obtained ppk gene had 2079 bp coding a polypeptide of 692 amino acids (accession number: MG820119). The obtained gene sequences were aligned with the closely related sequences extracted by BLAST² from the GenBank database.

In this assay, the strain gp-1 was inoculated to NBRIP with a shaking speed of 150 rpm at 30°C. The cells were collected at the 0, 6, 9, 12, and 24 h for extracting total RNA by using TriZol reagent. The cDNA synthesis was conducted using a CellAmp^TM Direct RT-qPCR Kit (Takara Biotech, Beijing, China). To quantitative analysis of gene expression, primer 16S-F (5′–GCG GAG AGA AGT AGC TTG CT–3′) and primer 16S-R (5′–CCG ACT TAG GCT CAT CTA TTA G–3′) were used to amplify 16S rRNA gene; primer ppk-F (5′–CCA TTT ACC TTA CAT GCT CAG CT–3′) and primer ppk-R (5′–GGT CAA TCT GAA CAC CTG CTT–3′) were applied to amplify ppk gene; primer pqq-F (5′–GAT GCC TTA GCA GGT TCA AA–3′) and primer pqq-R (5′–CTC AAC TGT ATC TGC ATT AAG TT–3′) were employed to amplify pqq gene. The three pairs of primers were designed based on the complete gene sequence data obtained in this assay. The abundance of 16S rRNA gene was used as inner control for copy number estimation. The qPCR was performed using SYBR green PCR Master Mix (Applied Biosystems) and reactions were operated in a real time PCR system (Applied Biosystems 3) in the following steps: denaturation at 94°C for 10 min, followed by 40 cycles of denaturation at 94°C for 15 s, annealing at 60°C for 1 min. Additionally, the concentration of extracellular phosphorus at each time point was also measured. Simultaneously, the granule staining was conducted to investigate the existence of polyphosphate (Poly-P) according to the method described in previous literature (Wan et al., 2017).

Significant differences were obtained by the one-way analysis of variance (one-way ANOVA) with means compared using the Tukey test in IBM SPSS 19. A neighbor-joining phylogenetic tree with 1000 bootstrap replicates was built based on 16S rRNA gene sequence using MEGA6 (Tamura et al., 2013). The immobilized Pb was assessed using the equation: Pb immobilization efficiency = [HAc extractable Pb (Soil + Ca₃(PO₄)₂)−HAc extractable Pb (Soil + Ca₃(PO₄)₂ + PSB)] / HAc extractable Pb (Soil + Ca₃(PO₄)₂) × 100%

During the four acclimation periods, the α-diversity indices including OTU (32–62), ACE (36–64), Chao1 (35–68), and Shannon index (1.30–2.23) decreased (Table 1), suggesting the species diversity gradually decreased. The Good’s coverage of all samples was 99.99%, suggesting the amplicon libraries could represent most of the species in the natural habitat.

The composition of bacterial community at both phylum and genus level presented large differences in four acclimation periods (Figure 1). At the phylum level, Proteobacteria was the dominant phylum in A1 (70.28%) and A2 (94.02%), and Bacteroidetes dominated in A3 (54.74%) and A4 (70.99%) (Figure 1A). while Firmicutes and Deinococcus–Thermus were the secondary level bacteria. At the genus level, Cellvibrio was the first dominant genus in A1 (41.72%) and A2 (52.3%), while Sphingobacterium was the supreme dominant genus in A3 (51.64%) and A4 (70.40%) (Figure 1B). The total relative abundances of Sphingobacterium, Cellvibrio, and Pseudomonas in A3 and A4 could reach to 93.88 and 98.2%, respectively. In these periods, Algoriphagus, Caulobacter, Caulobacteraceae, Cupriavidus, Deinococcus, Devosia, Flavobacterium, Hydrogenophaga, Niveispirillum, Paenibacillus, Phenylobacterium, Pseudomonas, Pseudorhodoferax, Ramlibacter, Rheinheimera, and Sediminibacterium exhibited low abundances. There results indicated that species diversity declined and bacterial community composition changed during long-term acclimation.

These 18 PSB were identified as Acinetobacter (gp-1), Pseudomonas (gp-2 and Ap-3), Massilia (gp-3 and gp-6), Bacillus (gp-4, gp-7, gp-8, gp-11, gp-15, and Ap-10), Arthrobacter (gp-12), Stenotrophomonas (AP-1, Ap-6, and Ap-9), Ochrobactrum (Ap-2), and Cupriavidus (Ap-4 and Ap-8) based on 16S rRNA gene sequence (Table 2). The colony morphology of most strains presented circular, wet texture, and entire edge with the exception of Bacillus strain, and most of them exhibited off-yellow and crateriform (Table 2). A phylogenetic tree was built to reflect the phylogenetic distance between these PSB (Figure 2). The 18 strains were clustered into two distinct parts: Gram-negative bacteria and Gram-positive bacteria. Besides, these Gram-negative bacteria belonged to Proteobacteria, and Gram-positive bacteria belonged to Firmicutes and Actinobacteria.

The 18 PSB presented different abilities to utilize Ca₃(PO₄)₂, phytate, FePO₄, and AlPO₄ (Figure 3). Basically, Ca₃(PO₄)₂ was the optimal P source for these 18 PSB in liquid medium (Figure 3A), followed by AlPO4 (Figure 3D), FePO4 (Figure 3C), and phytate (Figure 3B). The P dissolving level of these 18 strains in NBRIP liquid medium separately containing Ca₃(PO₄)₂, phytate, FePO₄, and AlPO₄, were 47.08–250.77, 1.19–7.91, 7.35–46.10, and 14.99–81.99 mg/L PO₄^3–-P, respectively. Additionally, the P solubilizing level represented by HD/CD ratio of these 18 strains in NBRIP solid medium separately containing Ca₃(PO₄)₂, phytate, FePO₄, and AlPO₄, were 0–4.62 (Figure 3E), 0–6.75 (Figure 3F), 0–13.58 (Figure 3G), and 1.00–8.50 (Figure 3H), respectively. The A. pittii gp-1 presented good performance for utilizing these four types of P sources in both liquid and solid medium, with P dissolving level ranging from 7.91 to 250.77 mg/L and from 3.80 to 7.11 mg/L, respectively. These results suggested that A. pittii gp-1 had great potentials for multiple P sources utilization.

The details of the content of WSP and AP under soil culture conditions after 30 days’ incubation are shown in Table 3. The WSP and AP content achieved the highest value under conditions with A. pittii gp-1 addition in the same soil incubation group. In S + B group, S + B + T group, and S + B + T + N group, the WSP content in soils with A. pittii gp-1 addition were 0.42, 0.52, and 0.80 mg/g, respectively; while AP content in soils with A. pittii gp-1 addition were 0.51, 0.81, and 1.64 mg/g, respectively. Additionally, we found the content of WSP and AP in S + B + T + N group were significantly higher than that in S + B group and S + B + T group (P < 0.05) (Table 3). Most PSB with the addition of Ca₃(PO₄)₂ could also significantly increase the content of WSP and AP. These results indicated that A. pittii gp-1 presented great utilization potential for enriching soil plant-absorbable P in practical condition.

The content of WSP and AP and the Pb immobilization efficiency increased as the addition rate of tricalcium phosphate increased (Figure 4). When the addition ratio of tricalcium phosphate was 10%, the WSP and AP reached the highest value of 0.62 and 1.05 mg/L, respectively. The Pb immobilization efficiency increased from 7.16% at tricalcium phosphate addition ratio of 1 to 22.80% at tricalcium phosphate addition rate of 10%. Additionally, the Pb immobilization efficiency was significantly correlated with WSP (r = 0.987, P < 0.01) and AP (r = 0.990, P < 0.01) (Table 4).

The abundances of 16S rRNA genes, gcd, bpp, and phoD in soils without both strain gp-1 and Ca₃(PO₄)₂ addition were 6.37 × 10⁹, 4.08 × 10⁶, 3.38 × 10⁶, and 1.34 × 10⁷ copies/g, respectively. The addition of tricalcium phosphate and A. pittii gp-1 presented different effects on the abundances of P-cycling-related genes (Figure 5). The 16S rRNA gene abundance (6.76 × 10⁹–9.12 × 10¹¹ copies/g) slightly increased as the Ca₃(PO₄)₂ addition ratio increased in soils without gp-1 addition (P > 0.05; Figure 5A). While 16S rRNA gene abundance in soils with gp-1 addition was significantly higher than that in soils without gp-1 addition and in soils without both gp-1 and Ca₃(PO₄)₂ addition (P < 0.05), and the 16S rRNA gene abundance increased from 1.58 × 10¹¹ copies/g at Ca₃(PO₄)₂ addition ratio of 1% to 4.47 × 10¹¹ copies/g at Ca₃(PO₄)₂ addition ratio of 10%. These results indicated that the significant increase of bacterial abundance might be due to the retained PSB A. pittii gp-1. Besides, gcd gene abundance in gp-1 added soils (ranging from 8.64 × 10⁶ copies/g to 2.19 × 10⁸ copies/g) was significantly higher than that in no gp-1 added soils (ranging from 4.11 × 10⁶ copies/g to 1.76 × 10⁷ copies/g) and in soils without both gp-1 and Ca₃(PO₄)₂ addition (P < 0.05; Figure 5B). At the same ratio of tricalcium phosphate addition, we found that bpp gene abundance in soils with gp-1 addition (3.69 × 10⁶–1.26 × 10⁷ copies/g) was higher than that in soils without gp-1 addition (3.36 × 10⁶–5.80 × 10⁶ copies/g) and in soils without both gp-1 and Ca₃(PO₄)₂ addition (P > 0.05; Figure 5C). However, the phoD gene abundance in with gp-1 added soils (1.61 × 10⁷–2.75 × 10⁷ copies/g) was slightly higher than that in without gp-1 added soils (1.33 × 10⁷–1.72 × 10⁷ copies/g) and in without both gp-1 and Ca₃(PO₄)₂ added soils (P > 0.05; Figure 5D). The Pb immobilization efficiency was significantly correlated with bpp (r = 0.917; P < 0.05) and gcd (r = 0.817; P < 0.05) gene abundance (Table 4). These results indicated that P solubilization was responsible for the immobilization of Pb. In addition, gcd could be amplified from A. pittii gp-1 using the primers described above, while bpp and phoD did not. These results implied that the addition of A. pittii gp-1 could increase the abundances of organic P-cycling-related genes.

The complete sequences of ppk and pqq genes in A. pittii gp-1 presented the highest similarities with complete sequences of ppk and pqq genes in Acinetobacter lactucae strain OTEC-02 (accession number: CP020015.1) (Figure 6), with corresponding similarities of 96.01% (Figure 6A) and 94.98% (Figure 6B), respectively. The expression levels of P-cycling-related genes and concentration of free phosphorus presented dynamic change during strain gp-1 incubation (Figure 7). The relative expression level of pqq gene significantly (P < 0.05) increased 18.18 times at the 6th h, and then declined to 4.99 times at the 24th h (Figure 7A). While the relative expression level of ppk gene dramatically (P < 0.05) increased, and achieved the highest value of 5.23 times at the 24th h. Similarly, the concentration of extracellular phosphorus remarkably increased from 0.29 mg/L at the initial period to 74.58 mg/L at the 24th h (P < 0.05; Figure 7B). The extracellular phosphorus concentration was positively correlated with the relative expression levels of ppk (r = 0.951; P < 0.05) and pqq (r = 0.495; P > 0.05). In addition, granule staining presented positive reaction (Figure 7C), suggesting Poly-P existed in A. pittii gp-1 cells. These results implied that ppk and pqq genes were closely correlated with phosphorus cycling, and the process of extracellular inorganic insoluble phosphorus solubilization was coupled with the process of polyphosphate synthesis.

In this study, we found big shifts in bacterial community composition during four acclimation stages. Some PSB have been identified belonging to Proteobacteria, Bacteroidetes, Firmicutes, and Deinococcus–Thermus (Vyas and Gulati, 2009; Shrivastava et al., 2010; Hanif et al., 2015; Li et al., 2017), which were also the major bacterial phylum in our study. Some genera, such as Sphingobacterium and Cellvibrio, have not been reported as PSB, however, they were the major genus in four acclimation stages. This might be due to they are potential PSB could not be cultured at present, or they are closely related with PSB.

After four rounds of acclimation, we gained 18 PSB belonging to eight genera including Acinetobacter, Arthrobacter, Bacillus, Cupriavidus, Massilia, Ochrobactrum, Pseudomonas, and Stenotrophomonas. Previous literatures have reported that some PSB belong to these genera (Yu et al., 2011; Imran et al., 2014; Liu et al., 2014; Silva et al., 2017), but most studies have not reported PSB possessing multiple P sources utilizing abilities. For instance, strain gp-1 identified as Acinetobacter genus in our study presented good performance for solubilizing Ca₃(PO₄)₂, FePO₄, and AlPO₄, and for digesting phytate, with corresponding P solubilizing levels were 250.77, 46.10, 81.99, and 7.91 mg/L PO₄^3–-P, respectively. The P solubilizing capacity of A. pittii gp-1 is higher than that of other reported bacteria, such as Acinetobacter baylyi W16 for Ca₃(PO₄)₂ (91 mg/L) (Yu et al., 2011), Enterobacter sp. C for Ca₃(PO₄)₂ (100 mg/L) (Melo et al., 2018), Bacillus sp. AB066338 for Ca₃(PO₄)₂ (137 mg/L) (Liu et al., 2014), Burkholderia seminalis. PSB7 for Ca₃(PO₄)₂ (145 mg/L) (Panhwar et al., 2014), Pantoea dispersa Cav.cy3 for Ca₃(PO₄)₂, FePO₄, and AlPO₄ (<50 mg/L) (Chen et al., 2014), Pseudomonas sp. for phytate (0.40 mg/L) (Maougal et al., 2014), and Paenibacillus elgii B56 for AlPO4 (16 mg/L) and Bacillus megaterium B119 for phytate (4.9 mg/g) (Oliveira et al., 2009). Some PSB present higher P solubilizing ability than that of A. pittii gp-1, such as Pseudomonas trivialis BIHB 745 for Ca₃(PO₄)₂ (827 mg/L) (Vyas and Gulati, 2009), Serratia marcescens RP8 for Ca₃(PO₄)₂ (974 mg/L) (Misra et al., 2012), Acinetobacter sp. ASL12 for Ca₃(PO₄)₂ (717 mg/L) (Liu et al., 2014), Bacillus cereus SPC09 for phytate (579.01 mg/L) (Zhang et al., 2015), Saccharomyces cerevisiae CICIMY008 for phytate (93 mg/L) (Chen et al., 2016), Burkholderia cepacia B116 for phytate (52.7 mg/L) (Oliveira et al., 2009). However, the multiple P sources utilizing ability promises gp-1 itself easily get used to environment, and therefore exhibits great potential for phosphate chemical industry and agro-ecosystems. Additionally, the stepwise acclimation by using Ca₃(PO₄)₂, phytate, FePO₄, and AlPO₄ provides a useful approach to obtain PSB possessing multiple P sources utilizing capacity.

Many studies have reported that PSB contribute to the immobilization of heavy metal. For instance, PSB Pantoea sp. CS2-B1 and Enterobacter cloacae SM1-B1 can immobilize Pb (Park et al., 2011); Enterobacter sp. could enhance Pb immobilization (Chen et al., 2019); Serratia marcescens OPDB3-6-1 shows good capacity for the immobilization of Pb, Cd, and Cu (Zhu et al., 2019); Pseudomonas sp. strain PG-12 exhibits good performance for Pb immobilization (Manzoor et al., 2019); Leclercia adecarboxylata B3 and Pseudomonas putida F2-1 can turn soluble Pb(II) into insoluble form (Teng et al., 2019). The promotion mechanism of heavy metal immobilization by PSB has been acknowledged, namely PSB produce soluble phosphorus and metabolite, and then these products could bind with heavy metal ions and turn into insoluble form (Park et al., 2011; Yuan et al., 2017; Teng et al., 2019; Zhu et al., 2019). The Pb immobilization efficiency was significantly correlated with soil available P and WSP in this study, which is similar with the findings that Pb immobilized amount and available phosphorus present significant correlation (Park et al., 2011; Yuan et al., 2017). Additionally, some studies have reported that the addition of inorganic phosphorus can increase the abundance of P-cycling-related gene (Silva et al., 2017; Hu et al., 2018; Wei et al., 2019), and alter bacterial community composition (Lagos et al., 2016; Luo et al., 2017; Wei et al., 2019). However, the effect of PSB addition on indigenous organic P-cycling-related gene abundance is rarely reported. To our knowledge, we first report that the addition of A. pittii gp-1 can significantly increase the abundance of inorganic P-cycling-related gcd-harboring bacteria via direct input of A. pittii gp-1, and slightly increase the abundances of organic P-cycling-related bpp-harboring and phoD-harboring bacterial communities via changing indigenous bacterial community. Additionally, limited study has revealed the relationship between Pb immobilization and P-cycling-related gene abundance in soils with PSB addition. The Pb immobilization efficiency was significantly positively correlated with gcd-harboring bacterial abundance and bpp-harboring bacterial abundance, suggesting gcd-harboring bacteria and bpp-harboring bacteria might be responsible for Pb immobilization. This finding might be due to on the one hand A. pittii gp-1 can survive from harsh condition; on the other hand, A. pittii gp-1 can dissolve inorganic phosphorus for enriching the soil available phosphorus, therefore causing changes in bacterial community composition and abundances of P-cycling-related genes. Previous literatures have reported that bacteria in Acinetobacter genus harbor heavy metal resistance gene Saffarian et al., 2017; Zhou et al., 2019), and phytase encoded by bpp-harboring bacteria presents high resistance to heavy metal damage (Yung et al., 2014).

To explain the process of phosphorus metabolism of A. pittii gp-1, we quantified the expression of polyphosphate kinase gene (ppk) and pyrroloquinoline quinone (pqq). Our results revealed that the transformation of insoluble tricalcium phosphate into soluble phosphorus involved in a strong expression of pqq gene. This finding is in line with previous reports that the expression of pqq gene is closely correlated with the solubilization of inorganic phosphorus (Ogut et al., 2010; Wagh et al., 2014; Oteino et al., 2015). We also observed an increasing expression level of ppk gene, which was closely correlated with the content of extracellular P. In addition, we found the existence of polyphosphate by using granular staining. To our knowledge, this is the first report that the solubilization of inorganic insoluble phosphorus is coupled with the synthesis of polyphosphate. It has been reported that polyphosphate regarded as high-energy compound could be hydrolyzed when phosphorus accumulating bacteria undergo undernourished conditions (Wan et al., 2017; Li and Dittrich, 2019; Zhong et al., 2019). Previous literatures have reported that some bacteria belonging to Acinetobacter genus harbor the ability to synthesize polyphosphate and could be regarded as phosphorus accumulating bacteria (Keating et al., 2016; Han et al., 2018). Therefore, we can conclude that one part of soluble phosphorus can be used for the formation of bacterial substance (e.g., DNA and RNA), and another part of soluble phosphorus can be applied for the synthesis of polyphosphate during inorganic phosphorus solubilization. However, the process of polyphosphate synthesis should be inhibited or blocked when use PSB in agro-ecosystems, since polyphosphate is not plant-absorbable phosphorus.