Antibiotic-Resistant Bacterial Group in Field Soil Evaluated by a Newly Developed Method Based on Restriction Fragment Length Polymorphism Analysis

Spreading of antibiotic resistant bacteria into environment is becoming a major public health problem, implicating affair of the indirect transmission of antibiotic resistant bacteria to human through drinking water, or vegetables, or daily products. Until now, the risk of nosocomial infection of antibiotic resistant bacteria has mainly been evaluated using clinical isolates by phenotypic method. To evaluate a risk of community-acquired infection of antibiotic resistant bacteria, a new method has been developed based on PCR-RFLP without isolation. By comparing restriction fragment lengths of the 16S rDNA gene from bacterial mixture grown under antibiotic treatment to those simulated from the DNA sequence, bacterial taxonomies were elucidated using the method of Okuda and Watanabe [1] [2]. In this study, taxonomies of polymyxin B resistant bacteria group in field soils, paddy field with organic manure and upland field without organic manure were estimated without isolation. In the both field soils, the major bacteria grown under the antibiotic were B. cereus group, which had natural resistance to this antibiotic. In field applied with organic manure, Prevotella spp., and the other Cytophagales, which were suggested to be of feces origin and to acquire resistance to the antibiotic, were detected. When numbers of each bacterial group were roughly estimated by the most probable number method, B. cereus group was enumerated to be 3.30 × 106 MPN/g dry soil in paddy field soil and 1.32 × 106 MPN/g dry soil in upland filed. Prevotella spp. and the other Cytophagales in paddy field were enumerated to be 1.31 × 106 MPN, and 1.07 × 106 MPN∙g−1 dry soil.


Introduction
At present, various kinds of manures and compost originated from diverse biological wastes e.g., livestock feces, food waste, and sewage sludge, were introduced into field soils under a hope to be used as organic fertilizers or under a governmental policy in aim to promote recycling and re-using organic wastes.Some biological wastes were reported to include antibiotic resistant bacteria [3]- [7] which acquired antibiotic resistance during chemotherapy for human or livestock, or by the addition of antimicrobial growth promoter (AGP).At present, spreading of antibiotic resistant bacteria from animal husbandry is becoming a major public health problem, implicating affair of community-acquired infection of antibiotic resistant bacteria through underground water, or vegetables, or daily products.While the existing their surveillance method targeting specific nosocomial bacteria is not suitable for the evaluation of their environmental risk [8] [9].Because the susceptibility tests and taxonomy determinations must be broadly expanded over a large numbers of environmentally important bacterial groups for the risk assessment [10].
If the bacterial taxonomies in culture grown under antibiotics can be elucidated directly without isolation, antibiotic resistant bacterial flora in various environments will become more easily analyzed, which will contribute to establish their environmental risk assessment method.PCR-RFLP of the 16S rDNA seemed suitable for this purpose, because not only taxonomic estimation of isolates [1] [10] [11], but also that of the mixed culture without isolation [12] can be estimated by measuring number and the lengths (bp) of the restriction fragments.In this manuscript, bacterial taxonomies in culture grown under polymyxin B were determined directly without isolation by employing PCR-RFLP.Furthermore, numbers of each taxonomically different bacterial group in field soils were roughly estimated by the most probable number method after taxonomic elucidations of the mixed culture in the serially diluted culture medium.

Enumeration by Dilution Plate Method and MPN
Surface soils (depth 0 -15 cm) were obtained from upland Andosol fields (B field) at Koshi, Kumamoto, Japan (National Agricultural Research Center for Kyushu-Okinawa Region), where buckwheat have been cultivated without organic fertilizer or manure, and from the surface horizon of Gray Lowland paddy field soil (P field) at Koshi, Kumamoto, Japan, where a rice-wheat rotation under paddy-upland fields conditions with an annual application of chemical fertilizer and organic manure.Sampling was done in duplicate on October 5. Immediately after sampling, soils were sieved (<2 mm).
The number of polymyxin B resistant bacteria was estimated by the dilution plate method using the peptonepolymyxin medium (PP medium); 10 g proteose peptone (Difco, Sparks MD), 5 g NaCl, 15 g agar and 5 mg polymyxin B sulfate per liter, pH 7.0 [13] [14].After 6 days incubation at 30˚C, colony forming units (CFU) were counted.For MPN, serial 10-fold dilutions (10 −2 to 10 −7 ) prepared from soil (1 g fresh wt.) were inoculated to test vials (5 replicates) including PP medium.After 3 days incubation at 30˚C, bacterial DNA in each vial was extracted as the followings.Each bacterial group was counted by MPN after the phylogenetic estimations.

Fragment Lengths Measurement by Microchip Electrophoresis System
Fragment lengths was measured by microchip electrophoresis system (Cosmo-i SV1200; Hitachi Electronics Engineering Co., Ltd.Tokyo Japan) as described previously [10] [11].The sample was diluted by de-ionized water (10 folds for Low salt buffer) before loading on i-tip DNA (IC-1000, Hitachi Chemical Industry Co., Ltd., Tokyo Japan).DNA fragment (65 bp; 5'GCTCAGATTGAACGCTGGCGACATTTCACAACACGAGCTGGCT CAGATTGAACGTGGCGACATT 3') was used as the lower internal standard, and the PCR product amplified by 41f/1066r primers was used as the upper internal standard, which were co-applied with samples as described previously [10] [11].In the next similarity search process, the fragment smaller than 100 bp were eliminated from the both theoretical and measured MERFLs as described previously [1].

Theoretical Multiple Enzyme Restriction Fragment Lengths (MERFL)
Database Used for the Estimation The same theoretical MERFL database as that described previously [1] [10] [11] was used for this research, which was consisted from 4370 sequence files of 576 bacterial genera, and 143 uncultured and 34 unidentified bacteria.

Data Processing for Phylogenetic Estimation Using Multi-Template DNA and Phylogenetic Estimation (Figure 1)
As each MPN vial included multi-template DNA originated from heterogeneous bacteria, most of the measured MERFLP was the mixed MERFLPs digested from the heterogeneous 16S rDNA.Whereas all the theoretical MERFLs were originated from the homogeneous 16S rDNA sequence, the measured MERFLP digested from the homogeneous 16S rDNA must be used for the phylogenetic estimation.
In order to search and select the MERFLP digested from the homogeneous 16S rDNA among the mixed MERFLPs, the authors constructed the following procedures: 1) the restriction fragments (RFs) with the highest relative mole concentration (ratio of fluorescent intensity to fragment size) were selected until accumulated RF sizes did not exceed to the original 16S r DNA size (1070 bp ± measuring error), which were used as the major RFs (represented as H in Table 1).2) After subtraction of the above the major RFs from the mixed heterogeneous RFs, RFs originated from the 2nd major gene were similarly selected among the remained heterogeneous RFs, and used for similarity search (represented as M in Table 1).3) After subtraction of the above 2nd major Scheme to identify and quantify each bacterial group using MERFLP and MPN.Theoretical MERFL database was constructed from DNA sequence files by the method of Watanabe and Okuda [2] as described in the former papers [1] [10] [11].Similarity between the measured MERFLP and the theoretical MERFLP were searched based on pairwise distance according to Nei and Li [18].Relative mole ratio was a ratio of averaged relative mole concentrations of the 2nd, or 3rd major MERFL to those of the major MERFL as described in the materials and method.d Restriction enzymes used for similarity search; "Ha", "Hh", "R", and "A" stand for Hae III, Hha I, Rsa I, and Alu I. e Allowance limit for measuring error was set at 10% at first, then increased as described in materials and methods.For the measured MERFLP which had no completely identical theoretical MERFLP, the theoretical MERFLP having the highest similarity using all the RFLPs was presented with the similarity in italics as described in the materials and method.f Species name (accession number) of the theoretical MERFL having the highest similarity with the measured MERFL of the isolate.g Additional name (accession number) of the theoretical MERFL having the same MERFL for the used restriction enzymes.
RFs from the remained heterogeneous RFs, RFs originated from the 3rd major gene were similarly selected and used for similarity search (represented as L in Table 1).
The similarity between the measured RFLP (A) and the theoretical RFLP (B) was calculated as described previously [1] [10] [11] based on the pairwise distance (D AB ) by the following equation; D AB = 1 − 2N AB /(N A + N B ), where N A and N B were the numbers of fragments of each RFLPs and N AB was the number of shared fragments that indicated same sizes within an allowance limit for measuring error according to Nei and Li [18].The pairwise distance of the MERFLPs (D ABME ) was an average of all the D ABs for used restriction enzymes.Similarity (%) was (1 − D ABME ) × 100 (Table 1).
In similarity search, allowance limit for measuring error was set at 10% in the first analysis, then gradually increased to 15 % until completely identical theoretical MERFL (100% similarity) were found out (Table 1) [1] [10] [11].If the completely identical theoretical MERFL was not found out by using all of the measured MERFL data, combinations of restriction enzymes used for the analysis was changed (Table 1) [1] [10] [11].As to the measured MERFL which had no completely identical theoretical MERFL, the theoretical MERFL having the highest similarity to the measured MERFL was indicated in Table 1 [1] [10] [11].

Estimation of Numbers of Each Taxonomically Different Group by the Most Probable Number Method
Most probable numbers of each groups (A-I) were estimated for five-tube, three-decimal-dilution experiment (Table 2).Confidence limits shown in Table 2 were obtained using Woodward's method [19].

Affiliation of Bacteria by MERFLP without Isolation
Affiliations of fifty three MERFLPs in each MPN vials were summarized in Table 1.They were affiliated to be Bacillus cereus group (Group A, 15 MERFLPs), the other Bacillus spp.(Group B, 5 MERFLPs), Clostridium spp.(Group C, 4 MERFLPs), the other Firmicutes (Group D, 9 MERFLPs), Actinobacteria (Group E, 2 MERFLPs), Proteobacteria (Group F, 7 MERFLPs), Prevotella spp.(Group G, 4 MERFLPs), the other Cytophagales (Group H, 5 MERFLPs), and the other gram negative bacterial group (Group I, 2 MERFLPs).Our former results indicated that most of the bacterial 16S rDNA were amplified by using the same PCR condition, and affiliated to the corresponding bacteria.When the method was used for the mixed culture, PCR bias was found to effect on the results as the followings.The most of the major MERFL (21 vials) in the dilution vials were affiliated to be B. cereus (15 vials), while there was no vials in the 2nd major and 3rd MERFL affiliated to be B. cereus (Table 1 and Table 3).The results suggested that B. cereus had always been the major MERFL even if B. cereus existed as the minor bacteria in the vial, because 16S rDNA of B. cereus might be amplified preferentially to the other bacterial groups by the used PCR condition due to the PCR bias.Although the MERFLP from the homogeneous 16S rDNA, the major MERFL, the 2nd major MERFL, and the 3rd major MERFL, could be selected among the mixed MERFLPs, the major MERLP provided the most reliable information on bacterial taxonomy of each bacterial group by the similarity search as the followings.Ratio of the MERFLs having 100% similarity to the corresponding theoretical MERFLs was the highest in the major MERFL (90.5%), followed by the 2nd major MERFL (50.0%), and the 3rd major MERFL (25.0%) (Table 3), which indicated that the discrepancies between the measured MERFL and the corresponding theoretical MERFL was gradually increasing in the 2nd and 3rd major MERFL.Smaller peak area of 2nd major MERFL (14.5% of the major MERFL Table 3) and the 3rd major MERFL (5.73% of the major MERFL Table 3) might be one factor to increase the discrepancies caused by the undetectable fragment under the detection limit or misreading of the fragment size.The 2nd major, or the 3rd major MERFLP having the same fragment size as the major MERFLP might be another factor to increase the discrepancies.Because the fragment, which was once attributed to the major MERFLP, could not be attributed to the 2nd major, or the 3rd major MERFLP.
Table 3. Affiliated bacterial groups of the major, the 2nd major, and 3rd major MERFLs and their relation to the results of similarity search.field (Table 2 and Figure 2).Proteobacteria (Group F) were enumerated to be 0.10 × 10 6 MPN in P field and 0.16 × 10 6 MPN in B field (Table 2 and Figure 2).Prevotella spp.(GroupG), the other Cytophagales (Group H), and the other bacterial group (Group I) were enumerated to be 1.31 × 10 6 MPN, 1.07 × 10 6 MPN, and 0.63 × 10 6 MPN•g −1 in P field, while the other Cytophagales (Group H) was enumerated to 0.028 × 10 6 MPN•g −1 in B field (Table 2 and Figure 2).Most MPN scores remained low especially in lower decimal dilution except for the some groups, Group A in P field and Group F in B field (Table 2), which might be caused by the PCR bias as the followings.When specific 16S rDNA, e.g., B cereus group, was preferentially amplified, the other 16S rDNAs in the same vials could not be equally amplified, which might inhibit reliable taxonomic estimation even if they were numerically dominant genomic DNA.The PCR bias also seemed to cause the inappropriate MPN scores (3-4-0, 1-3-0 Table 2) in some bacterial groups (Group A and Group G in P field) (Table 2), where confidence limit could not be calculated.Inhibition of PCR in lower dilution vials, which was caused by incomplete removal of PCR inhibitor in soils, also caused the lower MPN score in the lower decimal dilution.Although the numbers of some bacterial groups might be underestimated due to the PCR bias and PCR inhibition described above, the method at least provided the minimum numbers of each bacterial group.

Conclusions
Polymyxins (B and E), polycationic peptide antibiotics produced by Bacillus polymyxa, were bactericidal to gram-negative bacteria (MIC 50 = 1 ppm for most gram-negative bacteria) except for Burkholderia cepacia [21] due to crossing the bacterial outer membrane by competitive divalent cation displacement by bulky polycations and little to no effect on gram-positive bacteria.The bacteria in MPN vials, which were affiliated to be gram negative bacteria (from Group F to Group I, Figure 2), were suggested to acquire antibiotic resistance [10] [22].Although polymyxin B have mainly been used in hospitals and have never been used as AGP, or applied for livestock, polymyxin E (colistin), which has a similar structure and has the same site of action, has been used as an AGP in Japan.Polymyxin B-resistant gram negative bacteria detected in this study were supposed to be colistin resistant bacteria, because large numbers of polymyxin B (over 10 6 CFU/g dry matter) and colistin-resistant bacteria (over 10 6 MPN/g dry matter) were typically found in raw livestock feces [23].
The estimated numbers of gram negative antibiotic resistant bacteria (3.11 × 10 6 MPN•g −1 in P, and 0.19 × 10 6 MPN•g −1 in B) were lower than those estimated for the other field soils (from 31.7 × 10 6 CFU•g −1 to 258 × 10 6 CFU•g −1 ) [10] [20] where large amount of liquid livestock feces had annually been applied.As considerable numbers of Prevotella spp., and the other Cytophagales (1.31 × 10 6 MPN, 1.07 × 10 6 MPN), which had been suggested to be of feces origin [10] [24] were only enumerated in P field, the gram negative antibiotic resistant bacteria were concluded to be contaminated from organic manure.
The method presented here was found to be effective and useful for the purpose of monitoring whole antibiotic resistant bacterial flora in the environment and evaluating their risk.As the method was a new method, which was different from the other known analysis methods for microbial group such as denaturing gradient gel electrophoresis (DGGE) [25] [26] or terminal restriction fragment length polymorphism (t-RFLP) [27] [28], the availability as evaluation method for the other microbial groups, such as multi-drug resistant bacteria, bacteria causing food poisoning, and the other general bacteria, a precision of the affiliation, and validation of enumeration of each microbial group will be described in the next manuscripts.

Figure 1 .
Figure1.Scheme to identify and quantify each bacterial group using MERFLP and MPN.Theoretical MERFL database was constructed from DNA sequence files by the method of Watanabe and Okuda[2] as described in the former papers[1] [10][11].Similarity between the measured MERFLP and the theoretical MERFLP were searched based on pairwise distance according to Nei and Li[18].

Figure 2 .
Figure 2. Numbers of bacterial groups estimated by MPN and MERFLP in paddy field soil (P), and upland field soil (B).Number of B. cereus group (Group A; □), the other Bacillus spp.(Group B;), Clostridium spp., (Group C; ), the other low GC content gram positive bacterial group (Group D;), Actinobacteria (Group E; ), Proteobacteria (Group F; ), Prevotella spp.(Group G;), the other Cytophagales (Group H; ), and the other bacterial group (Group I; ■) were presented.
averaged relative mole concentrations of the 2nd, or 3rd major MERFL to those of the major MERFL.

Table 1 .
Affiliation of bacteria grown in serially diluted PP medium by MERFLP a .
Grouping was based on affiliation by MERFL; Bacillus cereus group (Group A), the other Bacillus spp.(GroupB), Clostridium spp.(Group C), the other Firmicutes (Group D), Actinobacteria (Group E), Proteobacteria (Group F), Prevotella spp.(Group G), the other Cytophagales (Group H), and the other gram negative bacterial group (Group I).b The 1st letter in vial indicates soils; "P" stands for paddy filed soil, and "B" stands for upland field soil.Exponential of vial number represents the decimal dilution of the vial.The 2nd number of vial number (1-5) represents number in 5 replicates for the each decimal dilution.H of last letter represents MERFL originating from the major 16S rDNA, M represents from the 2nd major 16S rDNA, and L represents from the 3rd major 16S rDNA.c a