|Home | About | Journals | Submit | Contact Us | Français|
Three repetitive-element PCR techniques were evaluated for the ability to type strains of Lactobacillus species commonly identified in the chicken gastrointestinal tract. Enterobacterial repetitive intergenic consensus PCR (ERIC-PCR) produced species- and strain-specific profiles for Lactobacillus crispatus, Lactobacillus gallinarum, Lactobacillus johnsonii, and Lactobacillus reuteri isolates. The technique typed strains within these species equally as well as pulsed-field gel electrophoresis. DNA concentration and quality did not affect the ERIC-PCR profiles, indicating that this method, unlike other high-resolution methods, can be adapted to high-throughput analysis of isolates. Subsequently, ERIC-PCR was used to type Lactobacillus species diversity of a large collection of isolates derived from chickens grown under commercial and necrotic enteritis disease induction conditions. This study has illustrated, for the first time, that there is great strain diversity within each Lactobacillus species present and has revealed that chickens raised under commercial conditions harbor greater species and strain diversity than chickens raised under necrotic enteritis disease induction conditions.
Lactobacilli are normal inhabitants within the microflora of the chicken gastrointestinal tract (GIT) (27, 39). Species frequently identified within the chicken GIT include Lactobacillus crispatus, Lactobacillus gallinarum, Lactobacillus johnsonii, and Lactobacillus reuteri (1, 7, 27). The first three of these species are members of the Lactobacillus acidophilus complex (LAC) (22, 32), a closely related group of species which are difficult to differentiate using traditional techniques, such as physiological and biochemical tests (34). While molecular methods, such as DNA-DNA hybridization (22), ribotyping (71), sodium dodecyl sulfate-polyacrylamide gel electrophoresis (19), randomly amplified polymorphic DNA PCR (19), and 16S rRNA gene sequencing (41), have been used with some success, many of these techniques are not readily adaptable to high-throughput applications required for large-scale ecological studies.
While numerous studies have reported Lactobacillus species distribution within the chicken GIT, the strain diversity within species has not been explored. Previously, Hagen et al. (28) investigated L. gallinarum isolates present within the crops of commercial chickens, revealing a high level of strain diversity among the isolates examined (17 strains represented among 38 isolates). These results indicate that there could be great diversity within and among the lactobacilli present within the chicken GIT. Examining and typing large numbers of lactobacilli from the chicken GIT to the strain level may facilitate a better understanding of microflora dynamics and niche competition. These studies may also result in the identification of strains which could be used in competitive exclusion applications and potentially in the development of probiotics or live vectors for the delivery of therapeutic recombinant proteins to specific sites within the chicken GIT.
Lactobacilli have been proposed as possible competitive exclusion agents or probiotics (30, 36, 42) against Clostridium perfringens, the causative agent of necrotic enteritis (NE) in broiler chickens. The withdrawal of antimicrobial growth promoters in Europe has led to an increase in the incidence of NE (10, 64), prompting the investigation of alternative methods for controlling C. perfringens in the broiler chicken GIT. Various models have been developed to study NE under research conditions, as recently reviewed by Dahiya et al. (15). The conditions used in these NE models and the effects they have on GIT microflora may adversely impact the identification of antimicrobial alternatives, such as probiotic strains, for use within commercial broiler chickens. Application of a high-throughput typing method capable of distinguishing species and strains is needed to determine if differences exist between the Lactobacillus populations of chickens raised under NE and commercial conditions.
While pulsed-field gel electrophoresis (PFGE) has been used widely for genotyping Lactobacillus strains (49, 50, 65), it is time-consuming, labor-intensive, expensive, and suitable only for low-throughput analysis of isolates (54, 65). In contrast, repetitive-element PCR (Rep-PCR) has been developed for genotypically fingerprinting bacteria and is a fast and reliable high-throughput genotyping system (66, 67). Rep-PCR has been used successfully to identify strains of a variety of genera (33, 44, 48, 55). Several Rep-PCR primers, including the repetitive extragenic palindromic (REP) primers (6, 9, 16, 65), the enterobacterial repetitive intergenic consensus (ERIC) primers (6, 65), and the (GTG)5 primer (24, 40, 62), have been used in the typing of lactobacilli in studies which have generally focused on species important to the dairy industry and food fermentations. Very few studies have examined the application of Rep-PCR to species commonly identified in the chicken GIT (62, 65). To our knowledge, only a single study has actually applied Rep-PCR to type Lactobacillus isolates from chickens (62) for the identification of potential probiotics to control Salmonella enterica serovar Enteritidis in egg-laying hens.
The aim of this study was to investigate whether Rep-PCR could be used to analyze Lactobacillus species and strains in the chicken GIT. Several Rep-PCR techniques [REP-, ERIC-, and (GTG)5-PCR] were compared for the ability to type strains of several Lactobacillus species commonly isolated from the chicken GIT (L. crispatus, L. gallinarum, L. johnsonii, and L. reuteri). ERIC-PCR was able to simultaneously type isolates to the species and strain levels, and its strain differentiation ability was comparable with that of PFGE. ERIC-PCR was further applied to high-throughput analysis of a large number of isolates collected from chickens raised under NE and commercial conditions.
The bacterial strains used in this study are outlined in Table Table1.1. All cultures of lactic acid bacteria were grown in MRS broth (Difco, Detroit, MI) or agar at 37°C for 24 or 48 h, respectively, under anaerobic conditions by use of a BD GasPAK EZ container (Becton Dickinson, North Ryde, Australia) with an AnaeroGen sachet (Oxoid, Thebarton, Australia).
Lactobacillus isolates were collected from the GITs of Ross 308 chickens raised under two different environmental and dietary regimens, referred to as “NE chickens” (n = 8, sacrificed on day 24) and “commercial chickens” (n = 10, sacrificed on day 42). The “NE chickens” were raised under the NE disease model conditions previously described (35). Briefly, the chickens were hatched in a commercial hatchery, raised in a research facility under conditions similar to those of commercial broiler chicken producers, and fed an antibiotic-free chicken starter diet containing 20% protein for 13 days followed by a wheat-based diet containing 50% fish meal (see Table S1 in the supplemental material). The chickens were challenged with C. perfringens on days 20 and 21, as outlined previously (35), and sacrificed by CO2 asphyxiation on day 24. All animal experiments were assessed, approved, and monitored by the Australian Animal Health Laboratories Animal Ethics Committee.
The GITs of “commercial chickens” were sourced from a commercial abattoir. The chickens were raised under conditions and diet, including antimicrobial growth promoters, typical of that used in the Australian Poultry Industry.
The crop, duodenum, jejunum, ileum, cecum, and colon were removed from the birds and opened, and swabs were collected from the lining of the tissue. The swabs were dilution streaked onto MRS agar plates. Approximately 10 isolated colonies per gut section per bird were randomly picked into 2-ml, 96-well (round deep well) plates (Axygen Scientific, Inc., Union City, CA) containing 1 ml MRS broth. The plates were sealed using breathable sealing film (Axygen) and incubated as outlined above. Following propagation, the cells were resuspended by pipetting, and 80 μl was transferred to half-skirt, 96-well PCR plates (Axygen) and mixed with 80 μl of MRS broth containing 50% glycerol (vol/vol). The plates were sealed using 96-well PCR AxyMats (Axygen) and stored at −80°C.
DNA was extracted from cultures by using a method adapted from Walter et al. (68). Briefly, “lysates” were prepared using overnight cultures pelleted in a 1.5-ml microfuge tube at 5,000 × g for 5 min and washed twice with 1 ml of TN150 buffer (10 mM Tris-HCl, 150 mM NaCl, pH 8). The cultures were resuspended in 500 μl of TN150 buffer, transferred to sterile 1.5-ml microfuge tubes containing 0.3 g of sterile, 0.1-mm zirconium-silica beads (Biospec Products, Bartlesville, OK), and then lysed in a tissue lyser (Qiagen, Doncaster, Australia) for 3 min at 30 Hz with a 2 by 24 adapter set (for up to 24 2-ml microcentrifuge tubes) (Qiagen), centrifuged as described above, and stored at −20°C. This method was adapted for high-throughput analysis as follows. Overnight cultures grown in 96-well plates as outlined above were centrifuged at 5,000 × g for 5 min. The pellets were resuspended twice in 1 ml TN150 buffer and centrifuged as described above. The pellet was resuspended in 500 μl TN150 buffer and transferred to a new 96-well plate containing approximately 0.3 g of sterile, 0.1-mm zirconium-silica beads per well. The 96-well plate was placed within a 96-well plate adapter (Qiagen) and shaken for 3 min at 30 Hz in a tissue lyser (Qiagen). The plates were centrifuged as described above and stored at −20°C.
Where indicated, lysates were further “purified” using phenol-chloroform extraction followed by ethanol precipitation as described previously (68). The DNA pellet was resuspended in 20 μl of 2 mg/ml DNase-free RNase A (Sigma-Aldrich, Castle Hill, Australia). DNA concentrations were determined spectrophotometrically with a Nanodrop ND-1000 instrument (Thermo Scientific, Noble Park, Australia).
The oligonucleotide primers used for amplified rRNA gene restriction analysis (ARDRA) and 16S rRNA gene sequencing are listed in Table Table2.2. The strains were identified to the species level using ARDRA as described by Guan et al. (27). Briefly, the Lb16a and 23-1B primers were used to amplify the entire 16S rRNA gene and the 16S-23S rRNA intergenic region. The PCR was performed as described previously (27), and PCR products were digested with HaeIII and/or MseI (New England Biolabs, Ipswich, MA) according to the manufacturer's directions. The digested products were analyzed on 2% (wt/vol) agarose gels. Sequencing of the 16S rRNA genes was performed by PCR amplification of the full 16S rRNA gene using the Ec16SrRNA1538Fwd and Ec16SrRNA1538Rev primers (Table (Table2)2) and PCR parameters reported previously (25). The PCR products were purified using a Wizard purification kit (Promega, Alexandria, Australia) according to the manufacturer's directions. Sequencing of the purified products was performed by the Australian Genome Research Facility (AGRF, St. Lucia, Australia).
Preparation of the cells and PFGE plugs was performed using a protocol described previously (28). Samples and the midrange PFG marker I (New England Biolabs) were run on 1% (wt/vol) agarose gels in 0.5× TBE buffer (0.045 M Tris-borate, 0.001 M EDTA, pH 8.3; Amresco, Solon, OH) at 14°C using a CHEF-DR III system (Bio-Rad, Gladesville, Australia). The conditions used for the L. crispatus, L. gallinarum, and L. johnsonii isolates were 6.0 V/cm, a 3-s to 17-s switching time, a linear ramping factor, a 120° angle, and a run time of 27 h. The conditions used for the L. reuteri isolates were 6.0 V/cm, a 1-s to 6-s switching time, a linear ramping factor, a 120° angle, and a run time of 17 h. The PFGE gels were visualized by staining with ethidium bromide (0.5 μg/ml) for 20 min and destaining in distilled H2O for 15 min and viewed under UV light with a Bio-Vision 1000 gel documentation system (Peqlab, Fareham, United Kingdom) and Vision-Capt 14.1a software (Vilber-Lourmat, Torey, France).
The primers used in this study are outlined in Table Table2.2. The 20-μl PCR mixture for each PCR type contained 2 μl of 10× PCR buffer [670 mM Tris-HCl, pH 8.8, 166 mM (NH4)2SO4, 4.5% Triton X-100 (vol/vol), 2 mg/ml gelatin; Astral Scientific, Caringbah, Australia], 3 mM MgCl2 (Astral Scientific), 200 μM of each deoxynucleoside triphosphate (Promega), 0.5 U Taq DNA polymerase (Astral Scientific), 1 μl of template DNA, and 1 μM of each primer (Sigma-Aldrich) for each PCR type. The ERIC-PCR (65), REP-PCR (65) and (GTG)5-PCR (67) programs were performed as previously described. PCR amplifications were conducted using a Palm-Cycler thermal cycler (Corbett Life Science, Mortlake, Australia). The Rep-PCR amplicons were analyzed on 1.5% (wt/vol) agarose gels. GeneRuler DNA ladder mix (Fermentas, Burlington, Ontario, Canada) was used as a molecular size marker according to the manufacturer's directions. Gels were visualized by ethidium bromide staining as outlined above.
All gels were analyzed using the BioNumerics version 5.10 software package (Applied Maths, Sint-Martens-Latem, Belgium). Dendrograms for PFGE gels were generated using the Dice similarity coefficient and the unweighted-pair group method using arithmetic averages (56), with 1% optimization and 1% position tolerance. The dendrograms for REP-PCR, ERIC-PCR, and (GTG)5-PCR comparisons were generated using the Pearson correlation similarity coefficient and the unweighted-pair group method using arithmetic averages, with 1% optimization.
Three different Rep-PCR primer sets were compared for the ability to type different strains of lactobacilli from the closely related LAC species L. crispatus, L. gallinarum, and L. johnsonii, which are commonly found in the chicken GIT (Fig. (Fig.11 and Table Table3).3). The strains used in this comparison had previously been typed to the species level using ARDRA (data not shown). The type strains of L. acidophilus and Lactobacillus amylovorus were also included in this comparison. Each of the primer sets generated profiles that varied within and among the species and that differed in both the number and the size of the bands produced. Where multiple strains were analyzed, species-specific bands were generated in each PCR (Table (Table3),3), with the exception of L. gallinarum in the REP-PCR. Strains of the same species clustered together and were similar to various degrees.
The REP-PCR generated profiles for the L. crispatus strains with an average of 19 bands (Table (Table3),3), which allowed easy interpretation and visual differentiation (Fig. (Fig.1A).1A). The profiles generated for the L. acidophilus and L. amylovorus strains contained more bands (≥23), while those for the L. gallinarum and L. johnsonii strains contained fewer bands (≤17). The similarities within species were ≥60%, ≥30%, and ≥30% for L. crispatus, L. gallinarum, and L. johnsonii, respectively. The L. acidophilus and L. amylovorus strains were grouped within the L. gallinarum strains, indicating that the technique cannot differentiate among these species.
The (GTG)5-PCR generated complex profiles comprised of a large number of bands (≥22 average bands) (Table (Table3)3) and similarly sized bands among strains and species (Fig. (Fig.1B).1B). The complexity of these profiles makes strain differentiation difficult without the use of gel analysis software, like BioNumerics. Strains from the same species clustered together, and the technique separated the L. acidophilus and L. amylovorus strains from the L. gallinarum strains. The similarities within species were ≥54%, ≥69%, and ≥45% for L. crispatus, L. gallinarum, and L. johnsonii, respectively, while the L. acidophilus and L. amylovorus strains were 52% similar to each other.
The ERIC-PCR generated profiles with a moderate number of bands (12 to 28 bands) (Fig. (Fig.1C1C and Table Table3).3). The species-specific bands (Table (Table3)3) are obvious in the ERIC-PCR profiles of each strain. ERIC-PCR clustered each of the strains from each species together with high similarity (L. crispatus at ≥80%, L. gallinarum at ≥82%, and L. johnsonii at ≥62%), while the L. acidophilus and L. amylovorus strains were less than 55% similar to each other and the other species. The ability of the ERIC-PCR to discern species is complemented by the ability to differentiate strains within a species. The profiles contained bands that were sufficient to separate and identify each strain while being easy to visually interpret and compare. Therefore, the ERIC-PCR technique was selected for further analysis of Lactobacillus isolates.
The abilities of PFGE and ERIC-PCR to type strains of common chicken Lactobacillus species were compared (Fig. (Fig.2).2). A collection of chicken Lactobacillus isolates were typed to the species level using ARDRA (data not shown), and 10 L. crispatus, 12 L. gallinarum, 9 L. johnsonii, and 10 L. reuteri isolates (including each type strain) were analyzed.
The species-specific banding profiles mentioned above are clearly visible in the ERIC-PCR profiles. In contrast, the PFGE profiles do not contain any species-specific patterns. This observation was supported by the BioNumerics analysis. When the ERIC-PCR profiles of the individual strains of the four species were compared, they clustered by species, whereas the PFGE profiles did not (data not shown). Therefore, ERIC-PCR provides easy typing of isolates to the species level, which is not possible with PFGE.
The similarities of the PFGE profiles varied among the different species, from ≥43% (L. gallinarum isolates) to ≥70% (L. reuteri isolates). PFGE profiles which were determined to indicate the same strain by use of the criteria of Tenover et al. (59) were highlighted in both the PFGE and the ERIC-PCR panel (Fig. (Fig.2).2). The ERIC-PCR profiles of these isolates were ≥97% similar. There were a few examples where the isolates generated very similar ERIC-PCR profiles and less-related PFGE profiles (for example, L. crispatus ANU 22-31 and ANU 29-31 were 97% and <50% similar by ERIC-PCR and PFGE, respectively).
Overall, ERIC-PCR simultaneously types isolates to the strain and species levels, while PFGE can type only to the strain level.
The ability of ERIC-PCR to produce similar profiles with poor-quality DNA and low concentrations of DNA was assessed to determine whether the technique is sufficiently robust to be readily adapted to high-throughput analysis of Lactobacillus isolates.
The use of 10-fold and 100-fold dilutions of stock DNA preparations (209 to 347 ng) as the PCR template does not adversely affect the ERIC-PCR profiles (Fig. (Fig.3A).3A). A 10-fold dilution of the DNA resulted in profiles which were 95 to 99% similar, while a 100-fold dilution resulted in 92 to 96% similarity and less-intense bands. The degrees of similarity of the profiles of the stock DNA and the 10-fold dilution of L. johnsonii ATCC 33200 were comparable to those observed for the other strains. The 100-fold dilution was a sole exception, having fewer and less-intense bands, and while the profiles were visually similar to those for the higher concentrations, the resulting similarity coefficient was 71%.
The effect of template DNA quality on ERIC-PCR profiles was tested to determine if crude lysates could be used instead of purified DNA for strain typing. Although some of the bands in the lysate profiles were fainter (Fig. (Fig.3B),3B), the lysate DNA generated profiles which were visually the same as those from the purified DNA (profiles generated from lysate and purified DNA were 97 to 99% similar). Again, L. johnsonii ATCC 33200 was the exception, as the lysate and purified ERIC-PCR profiles were 77% similar, while they appeared to be virtually identical visually. Therefore, the purity of the template DNA does not affect the ability of the ERIC-PCR to type Lactobacillus isolates to the strain level.
Collectively, the results indicate that the ERIC-PCR profiles are consistent regardless of variation in template DNA quantity and quality which may occur in high-throughput applications.
The ERIC-PCR technique was successfully employed in a high-throughput investigation of the diversity of species and strains of lactobacilli isolated from the GITs of NE and commercial chickens (Fig. (Fig.44 and and5,5, respectively).
In total, 500 isolates were characterized using ERIC-PCR (Table (Table4).4). The species-specific bands revealed that the majority of isolates belonged to the L. crispatus, L. gallinarum, L. johnsonii, and L. reuteri species. These four species comprised 56% and 57% of the total isolates from the NE and commercial birds, respectively (ERIC-PCR profiles of these species are presented in Fig. Fig.4A4A and and5A,5A, respectively). The representations of L. crispatus, L. gallinarum, L. johnsonii, and L. reuteri varied between the two bird types. Isolates from the NE birds were largely L. reuteri (45%) (there was one L. gallinarum isolate [0.5%], and there were no L. crispatus isolates), while the isolates from the commercial birds represented all four species (L. crispatus, 14%; L. gallinarum, 9%; L. johnsonii, 20%; and L. reuteri, 13%). The remaining isolates from the NE and commercial birds produced ERIC-PCR profiles that were not consistent with the four species examined above (Fig. (Fig.4B4B and and5B5B).
In analyzing the data with the BioNumerics software, it was noted that visually identical profiles vary in similarity values between 84 and 100%, while those less similar than 84% contain visible banding differences. Variations in similarity among identical profiles run on different gels have been reported previously (45) and are due to slight gel-to-gel shifts in the profile and the normalization process used to compare samples run on separate gels. Therefore, a cutoff of 84% similarity was used to define an “ERIC type.” The number of ERIC types represented within each species was determined (Table (Table4).4). To compare the diversity of strains, or ERIC types, between the two groups of birds, the number of ERIC types was divided by the number of isolates to determine an ERIC diversity index (EDI) (Table (Table4).4). The higher the EDI, the more ERIC types are represented among the isolates of a given species. The EDI values for the L. crispatus and L. gallinarum species could not be determined for the NE birds, while the commercial birds had EDI values of 0.28 and 0.67, respectively. The NE birds had less diversity within L. johnsonii (0.28 EDI) and L. reuteri (0.31 EDI) than did the commercial birds (0.48 and 0.79 EDI, respectively). Collectively, these results indicate that the representation and diversity of these four species are greater in the commercial birds than in the NE birds.
Both the NE and the commercial birds contained a large number of isolates (96 and 124 isolates, respectively) with species-specific bands which did not match the species-specific bands of the four species examined above. Typing of these isolates to the species level was performed by ARDRA and 16S rRNA gene sequencing (data not shown). The NE birds contained multiple Lactobacillus vaginalis, Enterococcus faecalis, and Enterococcus faecium isolates (Table (Table4),4), while these species were not present within the commercial birds. Isolates belonging to the species Lactobacillus agilis, Pediococcus acidilactici, Pediococcus pentosaceus, and Veillonella sp. were identified in both groups of birds. Over 25% of the commercial chicken isolates belonged to the Lactobacillus salivarius species, which is the largest species group present within the commercial birds, while no L. salivarius isolates were identified in the NE birds. A small number of isolates representing several other Lactobacillus and Enterococcus species were identified in the commercial birds. The larger number of species and the higher EDI results indicate that the commercial birds have a greater diversity of lactic acid bacteria than do the NE birds.
The number of isolates representing the same ERIC types within and between the two chicken groups was investigated. Several ERIC types which were isolated from multiple GIT sites from multiple birds were identified (Fig. (Fig.44 and and5).5). Interestingly, a combined comparison of the ERIC-PCR profiles from the NE and commercial isolates revealed that only one ERIC type was present within both groups (data not shown). This P. acidilactici ERIC type (Fig. (Fig.4B4B and and5B)5B) was identified only once within the commercial isolates and five times within the NE isolates.
Comparison of the Rep-PCR protocols for representative Lactobacillus strains revealed that each of the three techniques [REP-, ERIC- and (GTG)5-PCR] had the ability to separate the strains based upon the profiles generated. The ERIC-PCR profiles were easy to interpret, contained species- and strain-specific bands, and were able to sufficiently differentiate the isolates. The comparison of several L. crispatus, L. gallinarum, L. johnsonii, and L. reuteri isolates using both PFGE and ERIC-PCR revealed that the ERIC-PCR has the ability to discriminate isolates similarly to PFGE. ERIC-PCR was able to be applied to high-throughput analysis of lysates of cultures grown within 96-well plates. This technique was applied to the high-throughput typing of a large number of isolates collected from MRS agar plates from NE and commercial chickens. Various lactobacilli, along with a variety of enterococci, pediococci, and Veillonella isolates, were differentiated using the criteria defined for ERIC types. The results suggest that the lactic acid bacterium microflora of the two groups of birds varied with respect to species represented and diversity of ERIC types present.
The results indicate that ERIC-PCR is applicable to high-throughput analysis of Lactobacillus isolates propagated from the chicken GIT. The amount and purity of DNA template have little effect on the profile generated for a given strain. Gel-to-gel differences generate greater variation (data not shown), which is consistent with intraexperimental variation reported previously (45). Consequently, the number of gels should be minimized within an experiment. In this study, the gel-to-gel variations were taken into account by determining a similarity cutoff for profiles which were visually the same. As a result, the ERIC type classification of isolates in this study is likely to be a conservative estimation of the strain diversity present within the NE and commercial birds.
ERIC-PCR has several advantages over PFGE for the identification of large collections of isolates. Only one PCR is required to simultaneously type isolates to the species and ERIC type level, while a speciation step is required prior to PFGE. Crude DNA lysates can be used, thus minimizing the time required for DNA preparation, while the DNA preparation for PFGE is labor-intensive and time-consuming. Specialized PFGE equipment is not required. The PCR can also be scaled up for use in high-throughput analysis of a large number of isolates, while PFGE is low-throughput only. Finally, ERIC-PCR is relatively cheap to perform (hardware and reagents required) in comparison with PFGE and is considerably quicker.
Several other studies have applied Rep-PCR techniques to characterize lactobacilli from a variety of sources. Ventura and Zinc (65) examined L. johnsonii isolates from several sources and determined that ERIC-PCR was a highly reliable typing technique compared with PFGE. The results presented here support those of Ventura and Zinc and also demonstrate that ERIC-PCR is comparable with PFGE for typing isolates of the species L. crispatus, L. gallinarum, and L. reuteri. This is significant as it is the first time multiple strains of these species have been typed using Rep-PCR. These three species are consistently isolated from commercial broiler chickens around the world. Investigating the diversity of strains within these species may identify particular strains that have beneficial properties which could be used in commercial products.
(GTG)5-PCR has previously been used for the identification of a variety of food fermentation Lactobacillus isolates and several other lactic acid bacteria (24) and for the identification of Lactobacillus from the cloacae and vaginas of laying hens (62). Here we have applied ERIC-PCR to type lactobacilli and a variety of other lactic acid bacteria from the broiler chicken GIT and demonstrated the ability of the technique to define and differentiate isolates of closely related species. This technique could be of significant use to researchers investigating Lactobacillus populations in animals and humans. The current research extends the application of ERIC-PCR, which has previously been used for typing a wide variety of gram-negative bacteria in chickens, including Arcobacter spp. (5), Escherichia coli (17), Salmonella enterica (13), Campylobacter jejuni (70), Pasteurella multocida (52), and Ornithobacterium rhinotracheale (60). Collectively, the current research and these past studies highlight the utility of ERIC-PCR in the typing of a variety of gram-negative and gram-positive bacteria and its application to the study of microbial ecology.
REP- and ERIC-PCR have previously been used to differentiate strains of closely related cheese-making Lactobacillus species (Lactobacillus paracasei, Lactobacillus rhamnosus, and Lactobacillus zeae) (6). Here the application of ERIC-PCR to type other closely related lactobacilli was demonstrated. ERIC-PCR was used to type isolates of the L. amylovorus, L. crispatus, L. gallinarum, and L. johnsonii species, all of which belong to the LAC. The technique was also able to type isolates of other closely related species (20), including isolates from the L. reuteri group (Lactobacillus panis, L. reuteri, and L. vaginalis); the L. salivarius group (L. agilis, L. salivarius, and Lactobacillus saerimneri); and the closely related Pediococcus species P. acidilactici and P. pentosaceus. Isolates of enterococci were also able to be typed to the species and ERIC type levels, supporting a previous assessment of the ERIC-PCR for the differentiation of E. faecium isolates (18). The results presented here demonstrate the versatility of ERIC-PCR to type a wide variety of lactic acid bacteria to the genus, species, and strain levels.
The species identified within this study from broiler chickens raised in Australia are similar to those identified from broiler chickens around the world, supporting the notion that these species are autochthonous inhabitants within the chicken GIT. For example, L. agilis (7, 57, 62), L. amylovorus (11), L. crispatus (27, 62), L. gallinarum (27, 28, 62), L. johnsonii (27, 57, 62), L. panis (43), L. reuteri (27, 43, 62), L. salivarius (27, 62), L. saerimneri (62), L. vaginalis (7, 57, 62), P. acidilactici (1, 38), P. pentosaceus (53), Enterococcus cecorum (7), E. faecium (26, 31), E. faecalis (46), Enterococcus villorum (61), and Veillonella spp. (69) have previously been described as inhabitants of the chicken GIT. The identification of the Veillonella species isolates was interesting, as they are not lactic acid bacteria but rather belong to the order Clostridiales. Here we have demonstrated that these gram-negative bacteria were isolated from the chicken GIT, grown on MRS media, and typed using ERIC-PCR. Growth on MRS media may be possible due to the fermentative ability of Veillonella spp. (51).
There are a variety of factors that could contribute to the differences in the species and ERIC types represented between the two bird groups examined in this study. The Lactobacillus and lactic acid bacteria component of the chicken intestinal microflora is generally established by 14 days of age (3, 27), suggesting that the difference in age between the two bird groups (i.e., 24 and 42 days for the NE and commercial birds, respectively) is likely to be a minor contributor. Furthermore, the birds were the same breed, and efforts were made to minimize the effect of housing conditions by raising the NE chickens as close to commercial conditions as possible within the research facility. Therefore, the composition and presence/absence of antimicrobials in the diets (23, 29, 39, 69) and the C. perfringens challenge involved in the NE disease model (21) are the most likely contributors. With respect to the NE challenge aspect, our results are consistent with a recent study by Feng et al. (21) that reported on the representation of Lactobacillus species in their model. Interestingly they also found that while the L. reuteri population was not affected, the L. salivarius population decreased as a result of the C. perfringens challenge. Furthermore, they reported a decrease in the abundance of lactobacilli within the ileum, in particular, of the Lactobacillus aviarius species. Harrow et al. (29) also reported smaller L. salivarius populations within the GITs of chickens fed diets containing protein sources comprised of meat and bone meal, which are poorly digested in the proximal chicken GIT. This may suggest that in birds fed a high-protein diet, such as the NE birds in this study, the L. salivarius component of the “normal microflora” is replaced by other bacteria, such as C. perfringens, that can readily utilize the protein and amino acids in the ileum.
Several species were isolated from both groups of birds (L. agilis, L. gallinarum, L. johnsonii, L. reuteri, P. acidilactici, P. pentosaceus, and Veillonella spp.). A comparison of the ERIC types between the two groups revealed that only one P. acidilactici ERIC type was identified in both groups. This particular ERIC type may have a competitive advantage that enables it to colonize the GITs of birds under both dietary conditions. Comparison within the two groups revealed that several ERIC types were isolated from multiple GIT sections of multiple birds, suggesting that these types also may have particular attributes, such as adhesion factors, which allow them to be competitive and colonize multiple birds. It would be interesting to determine whether these strains could persist under the NE and/or commercial diet when reintroduced into birds in an attempt to determine whether they are truly persistent and whether diet would affect colonization. REP-PCR has previously been used to identify and monitor probiotic L. crispatus isolates within humans (4). The ERIC-PCR technique developed in this study could be used to track and monitor Lactobacillus strains in chickens.
Over recent years, several Lactobacillus genomes have been sequenced (2, 8, 12, 14, 37, 47, 63). While these genome sequences provide large amounts of information about these species, no two strains of the same species have yet been sequenced, nor have chicken lactobacilli been included to date. It would be extremely interesting to compare the genomes of multiple strains of the same species isolated from the same and/or different sources in an attempt to identify specific factors which allow them to colonize a particular niche. The work presented here highlights the great strain diversity within isolates from the same source and the need for more strains of the same species to be sequenced.
In summary, ERIC-PCR was selected from several Rep-PCR techniques for the typing of chicken lactobacilli. The technique was comparable to PFGE for typing strains of L. crispatus, L. gallinarum, L. johnsonii, and L. reuteri and generated both species- and strain-specific banding profiles for a variety of lactobacilli and other lactic acid bacteria. The method was adapted for, and applied successfully to, the high-throughput analysis of chicken Lactobacillus isolates. The data suggest that commercial chickens harbor greater lactic acid bacterial diversity than NE chickens.
This research was supported by the Rural Industries Research and Development Corporation (RIRDC) Chicken Meat Program (project ANU-72A), the Australian National University, and the Commonwealth Scientific and Industrial Research Organization (CSIRO).
We thank Todd Klaenhammer (North Carolina State University) for providing many of the reference cultures.
Published ahead of print on 11 September 2009.
†Supplemental material for this article may be found at http://aem.asm.org/.