High diversity and abundance of cultivable tetracycline-resistant bacteria in soil following pig manure application

By performing a microcosm experiment mimicking fertilization, we assessed the dynamic distribution of tetracycline-resistant bacteria (TRB) and corresponding tetracycline resistance genes (TRGs) from pig manure (PM) to the fertilized soil, by culture-dependent methods and PCR detection. Cultivable TRB were most abundant in PM, followed by fertilized soil and unfertilized soil. By restriction fragment length polymorphism (RFLP) analysis, TRB were assigned to 29, 20, and 153 operational taxonomic units (OTUs) in PM, unfertilized soil, and fertilized soil, respectively. After identification, they were further grouped into 19, 12, and 62 species, showing an enhanced diversity of cultivable TRB in the soil following PM application. The proportions of potentially pathogenic TRB in fertilized soil decreased by 69.35% and 41.92% compared with PM and unfertilized soil. Bacillus cereus was likely widely distributed TRB under various environments, and Rhodococcus erythropolis and Acinetobacter sp. probably spread from PM to the soil via fertilization. Meanwhile, tetL was the most common efflux pump gene in both unfertilized and fertilized soils relative to PM; tetB(P) and tet36 were common in PM, whereas tetO was predominant in unfertilized and fertilized soil samples. Sequencing indicated that over 65% of randomly selected TRB in fertilized soil with acquired resistance derived from PM.

Due to broad-spectrum activities against a wide range of pathogenic bacteria in both humans and animals, tetracyclines (TCs) have been used in anti-infective therapy and breeding industry for many years 1 . TCs are more frequently used for treatment and prophylaxis, and even as growth inducers, in livestock than humans 2,3 , which results in the selection of resistant animal pathogens through horizontal gene transfer (HGT) by means of mobile genetic elements [4][5][6] . The average antibiotic consumption per Chinese is nearly 10 times that of American individuals, with markedly elevated consumption by pigs in China 7,8 . Consequently, animal manures possess the highest number of antibiotic resistance genes (ARGs), especially tetracycline resistance genes (TRGs) 9 . In rural China, pig manure is often applied as organic fertilizer directly to the soil without any treatment. As a major source of antibiotic pollution 8 , it leads to large-scale soil and water pollution, harming humans through the food chain [10][11][12][13] . Therefore, how to safely process pig manure before field application is of great interest in China. To achieve this, uncovering the transfer characteristics of ARGs from manure to the fertilized soil and analyzing the shift in hosts harboring TRGs are critical to understanding the vital factors affecting the biosafety of pig manure.
Multiple studies have assessed TRG distribution in various hosts by the metagenome sequencing technology. Zhu et al. found that tetQ, tetW, tetX, tet32, tetO, tetM, tetL, and tetG are most abundant in the soil 14 . Ghosh and LaPara demonstrated that the most common genes are tetL, tetA, tetM, and tetG 15 . Li et al. showed that tetM is central to the TRG network, and could be used as an indicator to quantitatively estimate the abundances of other TRGs 16 . In three populations, tet32, tet40, tetO, tetQ and tetW were found to be prevalent in all gut samples, with tetQ being the most abundant 9 . Jurado-Rabadán et al. revealed that tetM is the most common TRG in enterococci 17 . As for TRG hosts, different results were obtained by researchers. Gao et al. found that Bacillus is the most dominant genus in tetracycline-resistant bacteria (TRB) in aquaculture environment 18 . Huang et al. indicated Microcosm experiment. Sterile Petri dishes (150 mm × 33 mm) containing 50-gram of pig manure, soil, and soil + pig manure, respectively (n = 3 per group), were prepared. Soil was collected from the upper 15 cm layer from barren land in Yangzhou University, with no fertilizer applied for over ten years. The characteristics of the soil samples were: pH 6.41; soil-water ratio, 1:1; organic matter, 11.04 g kg −1 ; cation exchange capacity, 8.96 cmol kg −1 . After pulverization and sieving (2 mm), soil samples were mixed evenly with pig manure specimens in different treatments mentioned above, in Petri dishes at a rate of 0.4% according to the traditional fertilization recommendations. All three treatments were placed at 25 °C and incubated for 30 days, since most organic fertilizers exhibit fertilization efficiency within 15-30 days. The moisture content of each manure sample was adjusted to 55% using sterile ddH 2 O 25,26 . Moisture content was derived according to the following formula: water weight (g)/dry soil weight (g) × 100%, where dry soil weight was determined after drying to constant weight at 110 °C 27 . Counting, screening, and identification of TRB. Ten-gram samples (wet weight) were added to 90 mL of sterile dH 2 O, shaken at 120 rpm, and placed at room temperature for 20 min. The flask was left for 5 min to allow soil particles to settle, followed by a ten-fold serial dilution with sterile dH 2 O. A total of 100 µL of serial tenfold dilutions were plated on Luria-Bertani (LB)-TC agar medium, which comprised 1/10-strength LB 28 agar supplemented with 16 μg ml −1 TC to grow cultivable TRB according to the Clinical and Laboratory Standards Institute (CLSI) document M100-S16 29 . Agar plates were incubated at 28 °C for 24 h, followed by routine counting. From plates with around 300 colonies each, individual colonies were picked, respectively, and streaked for single colony generation on LB-TC agar medium. Bacterial strains were separately stored at −80 °C in LB broth containing 20% glycerol.
Each pure culture was grown on a LB-TC agar plate for 12-48 h depending on growth rate; then, three loops of bacterial lawns were scraped into 200 μL of sterile ddH 2 O, followed by incubation in a water bath at 100 °C for 10 min and centrifugation at 8000 rpm for 3 min. The resulting supernatant was stored at −20 °C as DNA template. Nearly full length 16S rRNA was amplified with primers 27 f (5′-AGAGTTTGATCMTGGCTCAG-3′) and 1492r (5′-TACGGYTACCTTGTTACGACTT-3′) 30 . PCR was carried out in a 50 μL mixture system containing 10 μL DNA template, 0.2 mM of each dNTP, 0.4 μL of each primer, 1.25 U PrimeSTAR ® HS DNA Polymerase (TaKaRa, Dalian, China), and 1 × buffer (including Mg 2+ at 1.5 mM final concentration). Amplification was performed on an Eppendorf Mastercycler (Perkin-Elmer, Inc., Waltham, MA) under the following conditions: initial denaturation at 94 °C for 5 min; 30 cycles of 94 °C for 1 min, 55 °C for 1 min, 72 °C for 1.5 min; final extension at 72 °C for 10 min. Amplification products were assessed by agarose gel electrophoresis (1% w/v agarose in Tris-Borate-EDTA buffer). The resulting PCR products were digested with the restriction enzyme HinfI (TaKaRa, Dalian, China), separately, and distinguished according to different patterns mirrored by agarose gel electrophoresis at 1.2%. Only one randomly selected PCR product within the same HinfI-digested fingerprint pattern was sequenced by Sangon Biotech. Co., Ltd., Shanghai, China. After comparison with the GenBank reference sequences, the obtained sequences for representative strains from different operational taxonomic units (OTUs) were deposited in GenBank using the submission tool Sequin. The accession numbers of TRB were KX981212 -KX981438, and KY048431 -KY048441 (duplicates were discarded, keeping only one representative strain per species). Phylogenetic trees were constructed using the neighbor-joining algorithm in MEGA5 31 .
To further identify each strain at the species level, strains within the same genus based on 16S rRNA gene sequences were respectively subjected to identification through their biochemical and morphological properties according to the Bergey's Manual of Systematic Bacteriology. To confirm the TRGs base on size, five randomly selected bands for a particular TRG were excised from the agarose gel, followed by DNA recovery with a specific kit. The purified DNA was cloned into the pMD18-T vector (Takara Bio Inc.) and transformed into chemically competent E. coli DH5α. The extracted plasmid DNA from a positive clone was sequenced with universal primers by Sangon Biotech. Co., Ltd., Shanghai, China. After sequence comparison with BLAST, the confirmed PCR product was loaded on the gel as the positive control to verify other PCR products obtained for the same TRG.
Correlation between the TRG percentage in cultivable TRB and TRG abundance obtained with the culture-independent approach. To assess if the culture-dependent method adopted in this study was reliable, the percentage of TRGs in TRB and TRG abundance obtained by real-time quantitative PCR (q-PCR) approach were assessed.
Total microbial DNA was extracted from manure, soil, and soil + pig manure samples with Power-Soil TM DNA Isolation Kit (MO BIO Laboratories Inc., CA, USA) according to the manufacturer's instructions. Six TRGs, namely tetB, tetC, tetM, tetO, tetT, and tetZ genes, were amplified with primers described in our previous work 23 . PCR was performed on a Bio-Rad Miniopticon (Bio-Rad Laboratories, CA, USA) with SYBR Green I for estimating the copy numbers of TRGs. A total of 20 μL reaction system containing 10 μL of iTaq Universal SYBR Green Supermix, 0.4 mM of each primer, and 10 ng of template DNA was set up. The amplification procedure consisted of 95 °C for 1 min, followed by 40 cycles of 94 °C for 10 s, 61 °C, 68 °C, 55 °C, 60 °C, 46 °C, and 61 °C for 45 s (corresponding to the tetB, tetC, tetM, tetO, tetT, and tetZ genes, respectively), and the subsequent disassociation curve generation. Data were analyzed for target genes from soil and/or manure samples as previously described 32 . Amplification efficiency (E) was estimated from the slope of the standard curve with the following formula: E = (10 −1/slope ) − 1 33 . PCR efficiency between 95% and 105% was adopted for further analysis 34 .
Data analysis. Raw data were imported into Excel for analysis. Network visualization was performed on the interactive platform Cytoscape (version 3.2.0). Other graphs were obtained using Sigma Plot for Windows Version 10.0 (Systat Software, San Jose, CA, USA).

Correlation between the percentage of TRGs in cultivable TRB and TRG abundance obtained by the q-PCR approach.
It is necessary to assess whether the culture-based method adopted in this study is feasible. We therefore evaluated the correlation between the percentages of six randomly selected TRGs in cultivable TRB and their abundance levels obtained by the q-PCR approach. To avoid large differences in TRG abundance levels, the data obtained by the two methods were ranked and shown in Fig. 1. A good linear relationship was observed, indicating the reliability of the method used in this work.
Cultivable TRB. Although cultivable TRB in pig manure were about four and three log units higher than those in unfertilized soil and soil + pig manure samples, OTU and species numbers were lower than those of soil + pig manure treatment (Table 2). Among the three treatments, the indices of cultivable TRB in the unfertilized soil were ranked lowest. These findings indicated that (1) relatively high abundance and low diversity of cultivable TRB were found in pig manure, and (2) cultivable TRB in the soil could be greatly enhanced by pig manure application.  The succession in cultivable TRB at the species level from pig manure to fertilized soil is shown in Fig. 2. Specific species in pig manure, soil, and soil + pig manure accounted for 19.5%, 4.9%, and 52% of all species, respectively. Bacillus cereus was present in all three samples, and represented relatively abundant TRB in the environment. Most species in pig manure were not present in the fertilized soil, which indicated that other factors such as nutrients played stimulatory roles in the enhancement of bacterial species. Seven species, including Chryseobacterium lathyri, Rhodococcus equi, Microbacterium sp., and Pseudomonas fragi, were found in both unfertilized and fertilized soils, suggesting that they may be stubborn soil species which are hard to control. R. erythropolis and Acinetobacter sp. were probably spread from pig manure to the soil via fertilization, and more attention should to be paid to these species.
Frequency of the detected tetracycline resistance determinants. In cultivable TRB derived from the three samples, except tetY, tet38, tet45, tet44, and tet34, the remaining 39 TRGs were all found at different frequencies. In general, the detected species possessed efflux pump genes in all three samples, with most of  them having multiple efflux pump genes (Fig. 3). For example, Arthrobacter protophormiae (accession number KY048441), Stenotrophomonas koreensis (accession number KY048438), and Acinetobacter sp. (accession number KY048432) had 13 such genes. The frequencies of RPP and enzymatic modification genes were similar in each sample, and these two TRG groups in the fertilized soil were about 50% lower than in pig manure and unfertilized soil samples. The TRG with unknown function (tetU) showed highest frequency in pig manure, followed by soil and fertilized soil samples.
Of the efflux pump genes, tetB, tetL, and tetZ were the most common TRGs in pig manure, with frequencies 94.74%, 84.21%, and 68.42%, respectively; tetL (83.33%), tetB (75.00%), and tetA (66.67%) showed the highest frequencies in soil sample, and the top three efflux pump genes in the fertilized soil were tetL (82.26%), tetA  (Fig. 4). As for RPP genes, tetB(P), tet36, tetM, and tetO were found at more than 30%, while in unfertilized and fertilized soil samples tetO absolutely had the highest frequency. Meanwhile, tet37 and tetX were both detected in pig manure at frequencies of 47.37% and 36.84%, respectively, while only the tetX gene was found in unfertilized and fertilized soil samples at frequencies 75.00% and 50.00%, respectively. Preferential hosts for different TRG groups. The networks of efflux pump genes and their hosts are shown in Fig. 5. The most complex network of TRGs and their hosts was obtained in the fertilized soil, followed   by pig manure and normal soil. This finding indicated that pig manure application promoted the expression of efflux pump genes among diverse bacterial hosts. From pig manure and normal soil to fertilized soil, preferential hosts for efflux pump genes were changed from Stenotrophomonas koreensis (9 efflux pump genes), Providencia vermicola (9), A. protophomiae (8), Acinetobacter sp. (8), Paenibacillus lautus (10), Sphingobacterium anhuiense (10), P. fragi (7), and Rhodococcus equi (6) to Variovorax paradoxus (9), Achromobacter mucicolens (9), Acinetobacter sp. (8), P. frederiksbergensis (9), Bacillus sp., et al. (7). This finding suggested that Acinetobacter sp. was probably spread with fertilization, and changes in other preferential hosts for efflux pump genes in the fertilized soil might be stimulated by pig manure.
The preferential hosts for RPP genes were also changed obviously with pig manure application (Fig. 6). In unfertilized soil samples, all species had only tetO as RPP gene, while two groups of networks were distinguished   Table 3. GC-contents of genomic tetL in different hosts. tetL was amplified using the primer pair tetL-FW (5′-GTMGTTGCGCGCTATATTCC-3′) and tetL-RV (5′-GTGAAMGRWAGCCCACCTAA-3′). Acinetobacter sp. may also be the host for tetracycline-modifying enzyme genes introduced by pig manure (Fig. 7). B. cereus was a common host for genes in all three treatments. Except for Microbacterium sp. and R. equi, other species were possibly stimulated by pig manure. Besides, the hosts of the unknown TRG tetU seemed to be also induced by pig manure (Fig. 8).

Discussion
The negative effects of pig manure on TRG spread to the soil require special attention. Multiple studies have reported the high abundance and diversity of TRGs and/or TRB in pig manure and commercial organic fertilizers 14,22,35,36 . However, this study found that cultivable TRB in fertilized soil were three times more diverse than in pig manure and soil ( Table 2), indicating that pig manure application does not only enhance TRG abundance but also, more importantly, could increase the diversity of cultivable TRB in the soil. This undoubtedly intensifies the negative effects of pig manure on the spread of TRGs to the soil. Using a metagenomics approach, Udikovic-Kolic et al. showed that manure-treated soil has less phylogenetic diversity of bacteria compared with NPK-treated soil 37 . Although bacterial diversity in fertilized and unfertilized soils was not assessed in the current study, it can be inferred that larger proportions of bacteria were tetracycline resistant in the fertilized soil compared with the untreated soil, with many of them harboring proto-resistance or silent resistance genes 38 , which change into the tetracycline-resistant type following manure application.
Intriguingly, pig manure application did not increase the percentage of pathogenic TRB in the soil ( Table 2, Fig. 2); meanwhile, the diversity of pathogenic TRB decreased from 47.37% (pig manure) and 25.00% (untreated soil) to 14.52% (fertilized soil). On the one hand, some TRB with antagonistic effects became predominant following pig manure application, and may be capable of inhibiting sensitive pathogenic bacteria. For example, B. amyloliquefaciens 39  Myroides odoratimimus and Alcaligenes faecalis, in pig manure may be more adapted to the environment than to untreated or fertilized soil, since they are common in the gut environment 45,46 .
It can be inferred that R. erythropolis and Acinetobacter sp. were probably spread from pig manure to soil via fertilization, and more attention should be paid to these species. R. erythropolis can cause bloodstream infection 47 , and was firstly detected in pig manure. The high adaptability in distinct and even extreme environments of this TRB has been reported by many studies 48,49 ; this may be the reason for its wide distribution. Acinetobacter sp. in this study was not accurately identified at the species level, but the relatively high amount of TRGs as well as the wide distribution traits in this species also requires attention. Besides, B. cereus, which possessed around 5 TRGs, was found in all three samples. B. cereus is an opportunistic pathogen capable of causing food poisoning 50 ; however, it is often isolated for its potential to promote plant growth, and has been developed for commercial use 51,52 . Therefore, attention should be paid when using bio-agents containing this bacterium.
TRGs have diverse and distinct hosts between pig manure and the fertilized soil, with a high risk of spreading TRGs via pig manure application. To date, little is known about the changes of TRGs from pig manure to the soil. As shown above, tetL was the most common efflux pump gene in both untreated and fertilized soils versus pig manure, which is partly consistent with Peng et al. 53 . Besides, tet42 and tetK were most common in pig manure followed by fertilized soil and untreated soil, suggesting that they could be introduced into the soil via fertilization. Differences of tetA and tet33 in the three samples were also obvious, indicating that the four genes tet42, tetK, tetA, and tet33 could be used as indicators for monitoring efflux pump genes in TRGs among various treatments. However, further investigation is required since (1) the above data were obtained by culture-dependent methods with possible predilection for TRB growth on specified media, and (2) bacterial cell numbers were not taken into account in this study.
For RPP genes, pig manure increased tetM and tetQ in the soil, in part corroborating our previous study using the PCR detection approach 23 . As shown above, tetB(P) and tet36 were common in pig manure but undetected in both untreated and fertilized soils, in disagreemnt with a previous study 53 . The tet36 gene was firstly indentified in swine manure pits 54 , and is seldom used as an indicator in soil environments, suggesting that it may be only common in pig manure, with reduced risk of spreading. A similar result was obtained for tet37, an enzymatic modification gene which is rarely found in TRB 55 . We firstly reported that Zimmermannella faecalis, Acinetobacter sp., A. johnsonii, Wautersiella falsenii, et al. isolated from pig manure had the latter TRG. Another such gene, tetX, was increased in soil after manure application, and may have potential roles in degrading TCs in soil environments. In addition, most TRB had more than one TRG in this study, and many of them may acquire mobile genetic elements (MGEs) for fitness in the presence of TCs in the fertilized soil. This process can, on the other hand, impose a metabolic burden on bacterial hosts 56 . With time, they may reduce in fitness because of growth delay in the fertilized soil with decreasing TCs; thus, discarding some MGEs could be a possible strategy to achieve recovery of the ecological niche, which can be a source of donor cells for TRGs. Sequencing of nearly full length tetL and tetX in radomly selected TRB (Tables 3 and 4) revealed that about 63.6% (7/11) and 66.7% (4/6) of hosts harboring tetL and tetX probably acquired them from other TRB, indicating a diversity of hosts as well as common spread events for TRGs among TRB. All tetL or tetX sequences derived from different hosts were highly homologous; in addition, the hosts were mostly found in pig manure or fertilized soil samples, indicating severe TRG diffusion and spread from pig manure to the soil via manuring. Overall, TRGs had diverse and distinct hosts in pig manure, untreated soil, and fertilized soil in this study, suggesting that the spread of TRGs from pig manure to the soil remains a public concern.