Soil samples were collected from agricultural soil in the northeast of China (126°49′40″ E, 45°40′60” N) and were sieved to remove plant residues and stones for further study (The main reason we selected this area was that it had long grown corn, which had also led to the widespread use of atrazine). Atrazine was purchased from Shanghai Macklin Biochemical Co., Ltd. Cyanuric acid was purchased from Heowns Biochem Technologies, LLC, Tianjin. The liquid minimal salt media (MSM) amended with glucose as a carbon source was consisted of 2.4 g K₂HPO₄, 0.2 g MgSO₄·7H₂O, 1.2 g KH₂PO₄, 0.025 g CaCl₂·2H₂O, 0.008 g Fe₂ (SO₄)₃, and 1.0 g glucose in 1,000 ml of deionized water at pH 7.0 (marked as MSMG in this study). The liquid LB media contained 10 g tryptone, 5 g yeast extract, and 10 g NaCl in 1,000 ml of deionized water at pH 7.0. Solid media was formed by adding 17 g L⁻¹ agar into liquid media, and all media were sterilized at 121°C for 30 min for further used.

Soil samples (10 g) were added to 100 ml of MSMG (contained 100 mg L⁻¹ atrazine) and incubated under anaerobic conditions at 30°C and shaken at 160 rpm for 7 days. Then, the enrichment culture (10 ml) was transferred into a fresh MSMG (contained 200 mg L⁻¹ atrazine) and incubated for 7 days at the same conditions. The inoculation process was repeated five times until the final atrazine concentration reached 500 mg L⁻¹. Subsequently, the culture was collected and plated onto MSMG agar plates contained 100 mg L⁻¹ of atrazine and incubated at 30°C in the darkness. After 4 days of incubation, single colonies were selected aseptically and explored their degradation ability for atrazine. A strain designated as ZY with atrazine degradation activity was obtained for further analysis.

The identified experiments of strain ZY, including morphological observation, 16S rRNA gene sequence analysis, and Biology Gene III MicroPlate, were performed. Bacteria Genomic DNA Kit (TransGene Biotech Co., Ltd., Beijing, China) was used to extract the total genomic DNA of strain ZY. Strain ZY was identified via 16S rRNA gene sequencing, the oligonucleotide primers and the PCR cycle parameters used as previously described by Yang et al. (2018). Then the 16S rRNA was sequenced by Shihe Biotech Co., Ltd. (Harbin, China). The 16S rRNA of strain ZY was compared to those 16S rRNA available in the National Center for Biotechnology Information (NCBI) via Nucleotide BLAST. MEGA version 8.0 was used to construct a phylogenetic tree. BIOLOG Gen III MicroPlate (BIOLOG Inc., Hay ward, USA) was carried out to test carbon substrates utilization patterns of strain ZY by using automated tetrazolium-based microbial identification system produced by BIOLOG Inc. (Hay ward, USA; Mekonnen et al., 2019).

The strain ZY was cultured overnight (18 h) in LB media at 30°C and shaken at 160 rpm. After centrifugation, the strain cells were collected and suspended three times in MSMG media. A final cells density of 1 × 10⁸ CFU mL⁻¹ (OD_600nm = 1.0) determined by a UV-spectrophotometer (Shanghai Jinghua Technology Instrument Co., Ltd) was obtained for subsequent degradation experiments.

Atrazine degradation in liquid MSMG media was tested at pH 3.0, 5.0, 7.0, 9.0, and 11.0. The effect of temperature on atrazine degradation was tested at 20, 25, 30, 35, and 40°C. The degradation effects of the initial atrazine concentration were tested at 10, 30, 50, 70, and 100 mg L⁻¹. Only one parameter was changed during each culture. The MSMG suspension with OD_600nm 1.0 were added into the fresh MSMG media at 1:50 (v:v) dosage of inoculation and cultured at 30°C. The OD_600nm and atrazine in the culture were detected every 2 h. Atrazine and cyanuric acid were determined by ultra-performance liquid chromatography (UPLC) analysis (Yang et al., 2018). The media without inoculation was used as controls, and the atrazine degradation was less than 1.0% in the controls.

The strain ZY was analyzed by PCR for the presence of atrazine-degrading genes (i.e., atzA, atzB, atzC, atzD, atzE, atzF, trzN, and trzD). The primer pairs used in this study were described in Supplementary Table S2, and the reaction conditions referred to Sebaï et al. (2011). Heshi Biotech Co., Ltd. (Harbin, China) performed the sequencing of atrazine-degrading genes, and the sequence analysis was done via BLAST on NCBI databases.

To identify the atrazine metabolites produced by P. ureafaciens ZY, the MSMG culture collected at 6 h was analyzed using a rapid and reliable ultra-high performance liquid chromatography-quadrupole-time-of-flight tandem mass spectrometry (UPLC-QTOF-MS, 6545 Q-TOF, Agilent Technologies, U.S.) technique. Specific UPLC-Q-TOF-MS experimental parameters and the mass spectrometry methods were performed as described by previous studies (Tonelli Fernandes et al., 2018; Zhao et al., 2018).

The 200 g soil sample (dry weight) was added with methanol-dissolved atrazine to reach final concentrations of 10, 30, 50, 70, and100 mg kg⁻¹ (dry weight) and then was placed in a fume hood for 4 h to volatilize methanol. The soils were inoculated with strain ZY to reach the initial concentration of 4 × 10⁶ CFU g⁻¹, then passed through a mesh (2 mm) to achieve symmetrical. The treatments with ZY inoculums at 0, 10, 30, 50, 70, and 100 mg kg⁻¹ atrazine were marked as P0, P10, P30, P50, P70 and P100, respectively. Soil samples containing atrazine at 0, 10, 30, 50, 70, and 100 mg kg⁻¹ but without ZY inoculums were labeled as A0, A10, A30, A50, A70, A100, respectively. Then, soil samples were transferred to a plastic bowl (Φ 9 cm) and incubated in darkness at 25°C with 30% water-holding content adjusted by sterile distilled water. After inoculation for 1, 3, 5, and 7 d, soil samples were collected to analyze residual atrazine. In brief, 3 g of soil samples were added with 15 ml methanol for ultrasonic for 20 min, followed by centrifugation for supernatant collection, and the above operations were repeated three times (Luo et al., 2021). The received supernatant was concentrated by a rotary evaporator and dissolved by chromatographic methanol, followed by 0.22 μm filter membrane filtration for UPLC analysis.

The PLFAs analysis was used to assess soil microbial community structures. The PLFAs were extracted and analyzed using the methods reported by Frostegård et al. (1993) and Chaudhary et al. (2020). The Sherlock Microbial Identification System (SMIS), developed by MIDI Corporation (MIDI, Newark, Delaware, USA), was used to identify PLFAs in soil. All PLFAs classifications were shown in Supplementary Table S4 provided by MSIS.

The Illumina MiSeq of PCR-amplified 16S rRNA was used to analysis the structure of soil microbial community. Total DNA of soil was extracted by using the Cetyltrimethyl Ammonium Bromide (CTAB) according to manufacturer’s instructions. The 16S rRNA gene fragment was amplified using a universal primer set 341F (5’-CCTACGGGNGGCWGCAG-3′) and 805R (5’-GACTACHVGGGTATCTAATCC-3′; Logue et al., 2016). Sequencing was performed by LC-Bio Technology Co., Ltd., Hang Zhou, Zhejiang Province, China. The raw reads have been submitted to NCBI with the SRA database accession of PRJNA909019.

All experiments were performed in triplicate, and the results were presented as arithmetic averages and standard deviations. The datas were analyzed with the SPSS 19.0 and Origin 18.0 software. The Ecological Structure Activity Relationships (ECOSAR) model was used to predict the toxicity of atrazine metabolites. Principal component analysis (PCA) was performed by CANOCO 5.0 software. Heat map analysis and network analysis were performed using the R package (v3.5.2).

A strain ZY with a strong atrazine-degrading ability was selected for further study. A light yellow, smooth, and round colony of strain ZY was observed in the LB plate (Supplementary Figure S1A). Its morphological characteristics indicated that strain ZY was a Gram positive, and rod-shaped bacteria (Supplementary Figures S1B, S2).

The 16S rRNA gene sequence of strain ZY was 1,376 bp (accession no. ON878081), very similar to those of accessible strains, e.g., P. ureafaciens NC (99.19%, accession no. NR_029281.1), P. nicotinovorans DSM 420 (98.37%, accession no. NR_026194.1), and P. histidinolovorans DSM 20115 (98.21%, accession no. NR_026234.1). In a phylogenetic tree constructed based on the similar 16S rRNA gene sequences of ZY (Figure 1), the strain ZY clustered with members of the genus Paenarthrobacter sp. and was preliminarily identified as a P. ureafaciens due to the closest relative with P. ureafaciens NC.

Carbon utilization pattern of P. ureafaciens ZY was investigated via GEN III microplate after 48 h of incubation. The results have shown in Supplementary Table S3, Figure S3. Accordingly, P. ureafaciens ZY can metabolize D-fructose, D-mannose, D-galactose as carbon sources, but cannot assimilate dserine, troleandomycin, rifamycin, and minocycline (Supplementary Table S3).

In the liquid MSMG, the effects of pH, temperature, and atrazine concentration on the growth and atrazine degradation of P. ureafaciens ZY were investigated.

At present, atrazine degradation genes have been extensively investigated and proven to be highly conserved in many bacteria (de Souza et al., 1998; Singh and Jauhari, 2017). According to the different genetic compositions, atrazine-degrading bacteria are divided into two types. One type only contains atzA (trzN), atzB, and atzC, which could catalyze atrazine to produce cyanuric acid (Zhao et al., 2018). And the other type used extra genes atzD (trzD), atzE, and atzF to mineralize atrazine into CO₂ and NH₃ (Ma et al., 2017).

In this work, according to the primers in Supplementary Table S1, three genes including trzN, atzB, and atzC were obtained in P. ureafaciens ZY (Supplementary Figure S6). The amplified gene sequences were compared with those in NCBI database. The trzN (438 bp, ON954325) exhibited 100% identity with the trzN gene of Arthrobacter sp. AK-YN10 (HE716868.1) and Pseudomonas sp. ADP (FJ161692.1). The atzB (1,391 bp, ON911740) showed 100% homologous with the atzB gene of Arthrobacter sp. ZXY-2 (CP017421.1) and P. ureafaciens DnL1-1 (CP014574.1). The atzC (616 bp, ON954326) showed 99.67 and 99.67% homologous with the atzC gene of Pseudomonas sp. ADP (U66917.2) and Nocardioides sp. SP12 (AF537329.1). During the degradation process of atrazine by P. ureafaciens ZY, three metabolites of atrazine, were identified as hydroxyatrazine (m/z of 198.0), N-isopropylammelide (m/z of 171.0), and cyanuric acid (m/z of 128.0) by UPLC-Q-TOF-MS (Supplementary Figure S4). In Figure 3A, a putative atrazine degradation pathway was proposed: In the first step, hydrolytic dechlorination was induced by triazine hydrolase which endoded by trzN to yield hydroxyatrazine. Subsequently, hydroxyatrazine ethylaminohydrolase which encoded by atzB catalyzes hydroxyatrazine to form N-isopropylammelide. Finally, the N-isopropylammelide was eliminated from the s-triazine ring which induced by N-isopropylammelide isopropylamidohydrolase (encoded by atzC) to form cyanuric acid. Further determination of cyanuric acid content can also confirm the degradation path of atrazine (Figure 3B). The atrazine degradation pathway of P. ureafaciens ZY is similar to that in Arthrobacter sp. ZXY-2 (Zhao et al., 2018) and Citricoccus sp. strain TT3 (Yang et al., 2018), and different from other atrazine degraders which can mineralize atrazine, such as Ensifer sp. CX-T (Ma et al., 2017) and Pseudomonas sp. ADP (Martinez et al., 2001). By using the ECOSAR model, we also predicted the biotoxicity of the metabolites of atrazine, and the results were shown in Supplementary Table S5. Hydroxyatrazine, N-isopropylammelide, and cyanuric acid showed lower biotoxicity than its parent atrazine. Kolekar et al. (2019) also evaluated cytotoxicity tests by cell viability assay on HepG2 and indicated significant decrease in the toxicity of atrazine by the atrazine-degrading strains inoculation due to its effective degradation and formation of simpler and less/nontoxic metabolites (Hydroxyatrazine and N-isopropyllamilide). Intracellular reactive oxygen species levels and malondialdehyde content exhibited a significantly decreasing tendency after Escherichia coli exposure to intermediates (Hydroxyatrazine, N-isopropyllamilide, and cyanuric acid) of atrazine compared to parent atrazine (Yu et al., 2023). In conjunction, these results indicated that P. ureafaciens ZY had practical application value.

The biodegradation of different concentrations of atrazine was further investigated in soil with or without P. ureafaciens ZY inoculation. Without P. ureafaciens ZY inoculation, the highest degradation of atrazine (17.66%) was found in A30 after 7 days. In the treatments of A10, A50, A70, and A100, the degradation of atrazine was 12.71%, 12.87%, 13.03%, and 7.44%, respectively (Figure 4A). Atrazine was not detected in A0 and P0. Complete degradation of atrazine at 10 and 30 mg kg⁻¹ occurred within 5 days and 7 days in soil inoculated with P. ureafaciens ZY. In P50, P70, and P100, the degradation of atrazine at 7 days was 95.96%, 95.21%, and 85.31%, respectively (Figure 4B). In recent years, Paenarthrobacter has been screened and demonstrated strong atrazine-degrading ability. Jia et al. (2021) found that 95.9% of 5 mg kg⁻¹ atrazine was removed from the soils when inoculated with Paenarthrobacter strain AT-5 with 7 days. Chen et al. (2021) also reported that 96.0% of atrazine had been degraded with Paenarthrobacter sp. W11 inoculation after 49 days of incubation in soils (50 mg kg⁻¹ atrazine). These results indicated that P. ureafaciens had great potential for remediation of atrazine-contaminated soil.

In this work, the freshly added soils were used, and the addition of P. ureafaciens ZY showed a high degradation rate for atrazine in soil. Zhao et al. (2003) revealed that the degradation percentage of atrazine in the freshly added soils was significantly higher than that in the aged soils. The degradation efficiency decreased with the extension of inoculation time, and the half-life of atrazine inoculated in aged soil was longer than that in the freshly added soil. This can be explained by the fact that the bioavailability of atrazine in the freshly added soils is obviously increased compared to that in the aged soils (Yanze-Kontchou and Gschwind, 1995; Radosevich et al., 1997; Zhao et al., 2003).

Figure 5 exhibited the abundance of PLFAs (i.e., the biomass of microorganisms) in different treatments. In non-inoculation treatments, the total PLFAs of A0 (123.026 nmol g⁻¹) is higher than those in other treatments, especially in A70 (37.948 nmol g⁻¹, p < 0.01) and A100 (14.579 nmol g⁻¹, p < 0.01). Among them, the Gram-positive (Figure 5A), Gram-nagative (Figure 5B), Methanotroph (Figure 5C) and Actinomycetes (Figure 5E) also showed the same decrease trend. Previous studies have demonstrated that a specific inhibitory effect on soil microbes occurred by atrazine addition (Ros et al., 2006; Gao et al., 2018; Chen et al., 2021). Gao et al. (2018) also indicated that the utilization of carbon sources by microbial communities was decreased in atrazine-contaminated soil. The decline of soil microbial biomass might be caused by the accumulated metabolites, which might be more toxic to the microbes than the parent in the soil (Cheng et al., 2022). Seghers et al. (2003) revealed that atrazine significantly affected the structure and function of soil microbial community. Besides, the inhibition effects on the soil community and nitrification process were also reported (Chen et al., 2015).

At relatively low concentrations of 10–30 mg kg⁻¹ of atrazine, the total PLFAs of the non-inoculated groups were higher than that in the inoculated groups, the decrease of PLFAs of 32.687 nmol g⁻¹ and 19.858 nmol g⁻¹ were found in A10 and A30 compared to P10 and P30. And the same decrease trend was observed in Gram-negative (Figure 5A), Methanotroph (Figure 5B), Gram-positive (Figure 5C), Anaerobe (Figure 5D), Actinomycetes (Figure 5E) and Eukaryote (Figure 5F), which might be caused by competition between P. ureafaciens ZY and indigenous microorganisms (Abtahi et al., 2020; Koolivand et al., 2022). On the contrary, the biostimulation of P. ureafaciens ZY led to a significant increase of PLFAs at atrazine concentration of 50–100 mg kg⁻¹, especially for Gram-negative (Figure 5A), Gram-positive (Figure 5C), Anaerobe (Figure 5D), Actinomycetes (Figure 5E) in 100 mg kg⁻¹ atrazine treatments. This result was consistent with Gao et al. (2018) and Chen et al. (2021), which was possibly related with two reasons First, with the continuous removal of atrazine after inoculation of P. ureafaciens ZY, the inhibition effect of atrazine on soil microorganisms gradually decreased (Ros et al., 2006). Secondly, most microorganisms in the soil could not have the abllity to metabolize atrazine and cannot directly absorb atrazine as a carbon source and nitrogen source (Singh and Jauhari, 2017). After the decomposition of atrazine by P. ureafaciens ZY, atrazine metabolites could be acted as nitrogen and carbon source by soil microorganisms to meet their growth and proliferation (Chen et al., 2021).

The high-throughput sequencing was applied to investigate native microbial community dynamics at 7 days in different treatments. The OTU (Supplementary Figure S5A), Chao1 (Supplementary Figure S5B), Simpson (Supplementary Figure S5C) and Shannon (Supplementary Figure S5D) indexes of the bacterial community in A0 treatment were higher than that in other treatments at 7 days. These results indicated that atrazine and inoculation of P. ureafaciens ZY reduced the soil community richness and diversity. Jia et al. (2021) reported that the bioaugmentation of atrazine-contaminated soil with strain AT5 significantly decreased the bacterial richness and diversity, which might be due to the persistence and niche occupation of inoculation strain. In addition, P. ureafaciens ZY inoculation was also associated with interte production and nutrient consumption during the metabolism of atrazine, leading to the changes of soil micro-environment. Yang et al. (2021) have shown that the changes in the micro-environment could result in a significant decrease in bacterial richness and diversity. PCA analysis (Figure 6A) showed that both the bacterial community structure and the relative abundance of bacteria at the phylum level in A0 is similar to those in atrazine alone addition, and different from those in P. ureafaciens ZY inoculation treatments. It indicated that inoculums rather than atrazine was the main factor affecting the bacterial community structure (Jia et al., 2021).

In our study, Proteobacteria, Actinobacteriota, Firmicutes, Acidobacteriota, Planctomycetota were the primary bacterial phyla in the soil samples (Figure 6B). Previous study found that Proteobacteria are the dominant microorganisms in atrazine-contaminated soils (Jia et al., 2021). The members of phylum, including Actinobacteria, Firmicutes, and Proteobacteria were reported to be the best sources for degradation of pesticides (Kumar et al., 2021). The reported atrazine degrading bacteria were mainly distributed in Actinobacteria, Firmicutes, and Proteobacteria, such as, Achromobacter (Proteobacteria), Arthrobacter/Citricoccus (Actinobacteria), Bacillus (Firmicutes), respectively (Tonelli Fernandes et al., 2018; Yang et al., 2018; Zhao et al., 2018; Jakinala et al., 2019).

Figure 7A shows the shift in the relative abundances of bacterial genera. The inoculated genus Paenarthrobacter was not been detected in the atrazine alone treatments due to its low abundance. The relative abundances of Paenarthrobacter were 1.08% (P0), 1.11% (P10), 2.56% (P30), 2.75% (P50), 4.39% (P70) and 3.30% (P100) at 7 days, respectively. These results showed that the inoculated P. ureafaciens ZY could survive and proliferate in atrazine-contaminated soils. In our study, the atrazine degradation rate was positively correlated with Frateuria (p < 0.01), Dyella (p < 0.05), Burkholderia-Caballeronia-Paraburkholderia (p < 0.05), and Paenarthrobacter (p < 0.01), respectively. Compared to A0, an increase of relative abundance of 16.71%–312.53% (Frateuria) and 7.09%–66.93% (Dyella) were found in other groups. Atrazine alone addition decreased the relative abundance by 1.49%–26.73% in Burkholderia-Caballeronia-Paraburkholderia, however, P. ureafaciens ZY increased the relative abundance of Burkholderia-Caballeronia-Paraburkholderia by 31.68%–221.78% compared to those in non-inoculation treatments. The previous studies had reported that Frateuria species could metabolize amine (Aoki et al., 1984) and Paraburkholderia lycopersici sp. nov., was involved in soil nitrogen metabolism (Tapia-García et al., 2020). Dyella also showed excellent ability to degrade pollutants (Li et al., 2009; Cai et al., 2022). In conclusion, Dyella, Frateuria and Burkholderia-Caballeronia-Paraburkholderia may be important genera involved in the metabolism of atrazine and its metabolites. In addition, as shown in Figure 7B. Paenarthrobacter ureafaciens ZY increased the abundance genus Streptomyces and Bacillus, and there is a positive correlation between P. ureafaciens ZY, Streptomyces and Bacillus. In the previous study, Bacillus showed atrazine degradation ability (Jakinala et al., 2019) and the inoculation of biomixtures with Streptomyces sp. M7 increased atrazine removal (Saez et al., 2022). These results suggested that there might be a synergistic relationship between P. ureafaciens ZY, Bacillus and Streptomyces during atrazine degradation.

In addition, to further characterize the effects of P. ureafaciens ZY inoculation on soil native microbial community. The co-occurrence patterns of bacterial communities were assessed (Supplementary Figure S7). The more complex and well-connected co-occurrence networks were observed in atrazine alone treatments indicated by edge numbers and density with the value of 500 and 0.025 (Supplementary Table S8). This occurrence might be due to the stimulation of atrazine on the soil microbial diversity. The bacterial taxonomic composition of “hub nodes” in the network differed between atrazine alone treatments and P. ureafaciens ZY inoculation treatments. OTUs belonging to Sphingomonas in P. ureafaciens ZY inoculation treatments, and OTUs belonging to Gaiellales and Sericytochromatia in atrazine alone treatments were more connected with others (Supplementary Figure S7). Guo et al. (2022) found that hub species in microbial networks are recognized as mediators among associated microbiome. Trough network hubs, it follows that P. ureafaciens ZY inoculation may influence the assembly of soil native microbial community by regulating microbe-microbe interactions (Trivedi et al., 2020).