Intestinal bacterial overgrowth includes potential pathogens in the carbohydrate overload models of equine acute laminitis

Veterinary Microbiology 159 (2012) 354–363

Contents lists available at SciVerse ScienceDirect

Veterinary Microbiology journal homepage: www.elsevier.com/locate/vetmic

Intestinal bacterial overgrowth includes potential pathogens in the carbohydrate overload models of equine acute laminitis Janet C. Onishi a,b,*, Joong-Wook Park b,c, Julio Prado d, Susan C. Eades e, Mustajab H. Mirza e, Michael N. Fugaro f, Max M. Ha¨ggblom b, Craig R. Reinemeyer d a

Rutgers Equine Science Center, Rutgers University New Brunswick, NJ, United States Department of Biochemistry & Microbiology, Rutgers University, New Brunswick, NJ 08901, United States Department of Biological and Environmental Sciences, Troy University, Troy, AL 36082, United States d East Tennessee Clinical Research, Inc., Rockwood, TN 37854, United States e Equine Health Studies Program, School of Veterinary Medicine, Louisiana State University, Baton Rouge, LA 70803, United States f Department of Equine Studies, Centenary College, Hackettstown, NJ 07840, United States b c

A R T I C L E I N F O

A B S T R A C T

Article history: Received 4 August 2011 Received in revised form 4 April 2012 Accepted 6 April 2012

Carbohydrate overload models of equine acute laminitis are used to study the development of lameness. It is hypothesized that a diet-induced shift in cecal bacterial communities contributes to the development of the pro-inflammatory state that progresses to laminar failure. It is proposed that vasoactive amines, protease activators and endotoxin, all bacterial derived bioactive metabolites, play a role in disease development. Questions regarding the oral bioavailability of many of the bacterial derived bioactive metabolites remain. This study evaluates the possibility that a carbohydrate-induced overgrowth of potentially pathogenic cecal bacteria occurs and that bacterial translocation contributes toward the development of the pro-inflammatory state. Two groups of mixed-breed horses were used, those with laminitis induced by cornstarch (n = 6) or oligofructan (n = 6) and non-laminitic controls (n = 8). Cecal fluid and tissue homogenates of extra-intestinal sites including the laminae were used to enumerate Gram-negative and -positive bacteria. Horses that developed Obel grade2 lameness, revealed a significant overgrowth of potentially pathogenic Gram-positive and Gram-negative intestinal bacteria within the cecal fluid. Although colonization of extraintestinal sites with potentially pathogenic bacteria was not detected, results of this study indicate that cecal/colonic lymphadenopathy and eosinophilia develop in horses progressing to lameness. It is hypothesized that the pro-inflammatory state in carbohydrate overload models of equine acute laminitis is driven by an immune response to the rapid overgrowth of Gram-positive and Gram-negative cecal bacterial communities in the gut. Further equine research is indicated to study the immunological response, involving the lymphatic system that develops in the model. ß 2012 Elsevier B.V. All rights reserved.

Keywords: Equine Carbohydrate overload model Acute laminitis Bacterial overgrowth Bacterial translocation Lymphadenopathy

1. Introduction The cecal microbiota changes in the equine carbohydrate overload models of acute laminitis.

* Corresponding author at: Equine Science Center, 57 US Hwy 1, Rutgers University, New Brunswick, NJ 08901, United States. Tel.: +1 732 932 9419; fax: +1 732 392 2658. E-mail address: [email protected] (J.C. Onishi). 0378-1135/$ – see front matter ß 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.vetmic.2012.04.005

Overgrowth of Lactobacillus sp. or Streptococcus sp. occurs in cecal fluid following infusions of cornstarch (CS) or oligofructan (OF) (Garner et al., 1978; Milinovich et al., 2008). The Gram-negative bacterial communities change as well but precise alterations are less defined. A seven log decrease in cultured enterobacteriaceae was reported following a bolus of CS (Garner et al., 1978). After OF infusion, a time-dependent shift in Escherichia coli strains was detected using culture-independent methods

J.C. Onishi et al. / Veterinary Microbiology 159 (2012) 354–363

(Milinovich et al., 2008). In addition to lameness, the horses develop symptoms of systemic inflammation (Belknap et al., 2009). It is not clear how changes in gut microbiota contribute to development of inflammation although bacterial derived factors such as proteases, endotoxin and vasoactive amines are postulated to play a role. The objective of this study was to explore the possibility that following carbohydrate overload in the horse an overgrowth of potentially pathogenic cecal bacteria occurs and that the bacteria translocate to extra-intestinal organs. Bacterial translocation is a process by which gut-derived bacteria cross the intestinal mucosal membrane and access the lymphatic or circulatory systems (Berg, 1999). The process is involved in the development of mucosal immunity as well as in the pathogenesis of certain foodborne pathogens (Macpherson and Uhr, 2004). If translocation of bacteria occurs and the bacteria localize in tissue, then bacterial metabolites hypothesized to contribute toward the development of acute laminitis could attain physiological active levels in target tissue, the laminae. Culture-dependent and -independent methods were used to enumerate broad groups of potentially pathogenic cecal bacteria and assess the possible contribution of bacterial translocation in carbohydrate overload models of equine acute laminitis. 2. Materials and methods 2.1. Animals Laminitis is a disease unique to hoofed animals and therefore requires the judicious use of large animals susceptible to the condition. Researchers in the field adhere to the principles of 3Rs of animal experimentation; reduce, replace and refine. In this study, the numbers of animals used was reduced by combining results generated in the course of two separate studies. The control animals in the present study served as controls in an earlier study of chronic laminitis (Onishi et al., 2012). Animals donated and enrolled in the earlier study as control horse were selected from a donor pool after the owner and treating veterinarian determined elective euthanasia was appropriate for nonhoof related disorders where the prognosis was considered poor to grave. Non-hoof related disorders included such conditions as non-healing orthopedic injuries and severe arthritis. Eight mixed breed horses, ranging in age from 8 to 25 years, and consisting of 3 geldings and 5 mares, comprised the control set. The control horses had no abnormalities associated with the hoof and were eating and drinking water normally for their health-status. Six, mixed breed horses ranging in age from 4 to 5 years and of both sexes were placed on a pelleted ration of Purina Horse Chow. After 7 days, each was given a laminitisinducing ration of cornstarch wood gruel as described by Garner (Garner et al., 1975). Six mixed breed horses ranging in age from 4 to 9 years and of both sexes were enrolled in a laminitis model induced by oligofructan (inulin) as described by Bailey et al. (2009). Each horse received 1 g/kg body weight of OF top-dressed daily on feed for 3 days prior to induction. Induction involved

355

administering an aqueous solution of 10 g/kg OF by nasogastric intubation on the fourth day. In both induction models, clinical parameters were monitored at 4-h intervals, and included heart rate, respiratory rate, mucous membrane color, capillary refill time, fecal consistency and rectal temperature. Lameness was assessed at 4 h intervals and assigned an Obel lameness score (Menzies-Gow et al., 2010). All animals were euthanized when Obel grade 2 lameness was achieved. Euthanasia protocols used were in compliance with the American Veterinary Medical Association Guidelines on Euthanasia (2007). All procedures were conducted in compliance with regulatory guidelines, and were approved by the Animal Care and Facilities Committee and Rutgers University and the Animal Care and Use Committee at Louisiana State University and at East Tennessee Clinical Research, Inc., Rockwood, TN. The control horses, as well as those used in the carbohydrate overload models of acute laminitis were de-wormed and vaccinated using accepted veterinary protocols. The control horses for the study were obtained from two facilities in New Jersey. Control horses were maintained on mixed grass pastures common to New Jersey and included Kentucky bluegrass and Orchard grass as the most prominent forages. The animals also had access to timothy hay ad libitum. Experts in the use of the carbohydrate overload models of acute laminitis were located at different facilities. In both cases, test horses were acclimated at research facilities prior to initiation of their respective laminitis induction models. Horses receiving the cornstarch-gruel induction ration were acclimated on Bermuda grass and Bermuda grass hay and Purina Horse Chow for 3 to 4 months prior to enrollment in the study. Horses receiving the oligofructose ration were acclimated for 1 to 7 months on pastures of orchard grass and were provided with timothy-fescue hay. The diets of the control and OF-infused horses were most similar. 2.2. Sample collection Blood samples were collected prior to euthanasia in sodium polyanethol sulfonate tubes (SPS) (Becton Dickinson, Franklin Lakes, NJ) and cooled on ice. Immediately after euthanasia, the hoof of one forelimb was disarticulated at the fetlock joint and processed for recovery of laminar tissue, as described below. A sterile abdominal field was established and cecal and colonic lymph nodes, liver and spleen were harvested aseptically and placed in sterile containers. A sample of cecal ingesta was collected as the final step in the harvesting process. Sterile tubes containing cecal ingesta were filled to capacity, sealed with parafilm and cooled on ice. Tissues collected from control horses were kept on ice and processed on the day of necropsy. Samples from acute laminitic horses were placed on ice and shipped on the day of necropsy by overnight express delivery to Rutgers University. Culture status was evaluated on the day after necropsy. 2.3. Enumeration of aerobic bacteria in organ homogenates The laminar tissue was collected and processed as described earlier (Onishi et al., 2012). The laminar tissue

J.C. Onishi et al. / Veterinary Microbiology 159 (2012) 354–363

356

analyzed included material connected to the P3 as well as the hoof wall. Samples of cecal/colon lymph nodes, liver and spleen were transferred to sterile whirl-pak bags with a filter insert (Nasco, Fort Atkinson, WI) and weighed. Lymphoid tissue was suspended in two volumes of PBS relative to weight, while samples of the liver and spleen were placed in an equal volume of PBS relative to weight. Tissues were homogenized by manual pressure and 100 mL aliquots were diluted in PBS as needed and 100 mL aliquots were spread on microbiological agar plates. The particulates in the cecal fluid were allowed to settle by gravity and aliquots were appropriately diluted in PBS and plated on Columbia agar supplemented with 5% fresh horse defribinated blood (Hema Resource & Supply, Aurora, Oregon). One hundred mL aliquots of the SPS tubes were plated directly on agar plates to detect bacteria. Agar plates were incubated at 37 8C for 72 h and colonies of bacteria were enumerated. The numbers of bacteria per gram of tissue were calculated to control for differences in PBS volume used to suspend the tissues. All processes were performed in a Class II biosafety hood. 2.4. Profiling facultative anaerobic organisms in the cecal fluid Facultative anaerobic organisms in the cecal fluid were enumerated using a panel of selective agar media chosen to recover groups of bacteria known to colonize the intestinal tract. Two types of media were used to recover Gram-negative enteric bacteria. MacConkey Agar was used to enumerate organisms capable of lactose fermentation such as Escherichia, Klebsiella, Enterobacter and Citrobacter as well as those that do not ferment lactose. Salmonella and Shigella (SS) agar was used to recover Salmonella and Shigella. Three media were used to recover Gram-positive bacteria. Colistin–oxolinic acid-blood (COBA) was used to recover catalase-negative, Gram-positive bacteria such as Aerococcus, Corynebacterium, Enterococcus, Lactococcus, and Streptococcus (Sawant et al., 2002). Other types of Grampositive bacteria were recovered on Mannitol Salts Agar (MSA) and Palcam Agar. The yellow colonies on MSA were counted as Staphylococcus spp., while the black colonies on Palcam agar were counted as Listeria spp. All media supplies were obtained from scientific supply companies (Becton Dickinson & Company, Franklin Lakes, NJ; Palcam

media from EMD Chemicals, Gibbstown, NJ). The bacterial colonies in the cecal fluid were enumerated and the number of colony forming units (CFU) per milliliter was determined by back-calculation from the dilution needed to recover 100–200 colonies on an agar plate. Cecal fluid was analyzed for the presence of specific target genes using proprietary primers and PCR conditions to detect pathogens associated with equine diarrheal disease (Zoologix, Inc., Chatsworth, CA). The pathogens and the target genes detected in the screen were Clostridium difficile: triosephosphate isomerase, Ehrlichia risticii: citrate synthase; Lawsonia intracellularis: superoxide dismutase C and Salmonella: invasion protein A. 2.5. DNA extraction procedure Tissue homogenates that had been stored as frozen 15% glycerol stocks at 80 8C were thawed, refrozen at 80 8C and thawed a second time and held on ice prior to DNA extraction. Aliquots equivalent to 40 mg of tissue were washed with 1 mL of ice-cold phosphate buffered saline, pH 7.2, by centrifugation at 14,000 rpm for 4 min. DNA was extracted using a commercial extraction protocol as described by the manufacturer (Extract-N-Amp; Sigma–Aldrich, St. Louis MO). Briefly, the washed pellet was suspended in 90 mL of extraction solution, and heated at 95 8C for 10 min. After allowing the extract to cool to room temperature, 90 mL of the stop solution was added and the sample was centrifuged at 14,000 rpm for 2 min to pellet the particulates. Prior to testing the DNA extract sample in PCRs, the samples were diluted one to ten in bacterial DNA-free water (Lonza water). The extraction of bacterial DNA from Gram-positive and negative bacteria was assessed by using overnight cultures grown in brain heart infusion at 37 8C of Staphylococcus aureus and E. coli diluted serially from 103 to 108 CFU without tissue as well as with 40 mg of lymph node tissue homogenate from control animal number 2. The bacterial DNA extract from the dilution series was used to determine the limit of detection in all PCR assays. 2.6. PCR amplification and agarose gel electrophoresis Primer pairs, listed in Table 1, were used to amplify the V3 region of the ribosomal 16S rRNA gene (Muyzer et al.,

Table 1 Primer sequences used for PCR. Forward and reverse primer sequences are reported 50 to 30 . Target

Forward primer

Reverse primer

V3 region of the 16S rRNA V3 region of the 16S rRNA gene for DGGE analysis 23S rRNA This study

338F-50 -ACT CCT ACG GGA GGC AGC AG-30

518R-50 -ATT ACC GCG GCT GCT GG-30

341F-GC 50 - CGC CCG CCG CGC GCG GCG GGC GGG GCG GGG GCA CGG GGG GCC TAC GGG AGG CAG CAG -30

518R-50 -ATT ACC GCG GCT GCT GG-30

256F 50 -AGTAGYGGCGAGCGAA-30 256f-GC clamp CGCCCGCCGCGCGCGGCGGGCGGGGCGGGGGCACGGGGGGAGTAGYGGCGAGCGAA 1930F 50 -GTAGCGAAATTCCTTGTCG-30 1930f-GC clamp CGCCCGCCGCGCGCGGCGGGCGGGGCGGGGGCACGGGGGGGTAGCGAAATTCCTTGTCG 1663F-CAG GCG TTT CGT CAG TAT CC

457R 50 -CCTTTCCCTCACGGTACT-30

23S rRNA This study

b-Galactosidase (lacZ)

457R 50 -CCTTTCCCTCACGGTACT-30 2241R 50 -ACCGCCCCAGTHAAACT-30 2241R 50 -ACCGCCCCAGTHAAACT-30 1870R-GGT GTT TTG CTT CCG TCA G

J.C. Onishi et al. / Veterinary Microbiology 159 (2012) 354–363

1993), the ribosomal 23S rRNA gene (this study) and the bgalactosidase (lacZ) gene (this study) of E. coli K12. A realtime PCR assay was performed by use of SYBR green nucleic acid stain detection system with melt curve analysis to assess purity of the PCR product (Step-One 7900 thermocycler, Applied Biosystems, Foster City, CA). The PCR conditions to amplify the 16S rRNA gene and the lacZ gene fragments were 95 8C for 10 min followed by 40 cycles of 94 8C for 15 s, 60 8C for 30 s, and 72 8C for 30 s (Park et al., 2011). The 23S rRNA gene fragment was amplified at 95 8C for 10 min followed by 40 cycles of 94 8C for 20 s, 52 8C for 45 s, and 72 8C for 45 s. All PCR products were evaluated using melt curve analysis which was carried out at 95 8C for 15 s then heated from 60 8C to 95 8C at a rate of 0.3 8C/s. Amplicons produced with the 16S rRNA primers were separated using electrophoresis in a 3% high resolution agarose gel in TBE (89 mM, Tris base-89 mM, borate-2 mM, EDTA pH 8.0) after staining with ethidium bromide (Sigma–Aldrich, St. Louis, MO). The mobility of the amplicons was compared to a reference ladder consisting of DNA fragments ranging from 100 base pairs to 2000 base pairs (Low DNA Mass Ladder, Invitrogen, Carlsbad, CA). 2.7. Denaturing gradient gel electrophoresis and band identification The PCR products of the 16S rRNA gene were further analyzed using the DCode TM system (Bio-Rad Laboratories, Hercules, CA) for DGGE analysis. PCR reaction conditions and cycles were as described previously. After electrophoresis, the gels were stained with ethidium bromide and photographed on a UV transilluminator. Image analyses of the DNA profiles were conducted using Quantity One1 (version 4.5.0; Bio-Rad Laboratories, Hercules, CA). DGGE bands were excised from the gel and eluted overnight in 50 mL of MilliQ H2O at 4 8C. Eluted DGGE bands were re-amplified with 341F-GC and 534R and purity was confirmed by DGGE. The eluted single DGGE bands were amplified with 341F and 534R, purified using a PCR purification kit (MoBio Laboratories Inc., Carlsbad, CA), and used as a template for DNA sequencing. DNA sequencing was performed by Genewiz, Inc. (North Brunswick, NJ). The sequences were compared to sequences deposited in GenBank and similarity searches were performed using BLAST (Altschul et al., 1997). A pairwise determination of similarity or identity between DNA sequences of the closest relative were calculated using MatGat2.01 software (Campanella et al., 2003). Phylogenetic and molecular evolutionary analyses were conducted using MEGA version 4 (Tamura et al., 2007). The PCR products of the 23S rRNA gene were analyzed using the DGGE methods described above. The primers with the GC clamp are listed in Table 1. 2.7.1. Histology Cecal/colonic lymph nodes were collected from all oligofructose-induced horses at necropsy and preserved in 10% buffered neutral formalin. Representative portions of lymph nodes were sectioned at 5–6 mm, mounted, and stained with hematoxylin and eosin. Stained sections were

357

examined at 20–400 total magnification by a boardcertified veterinary pathologist1. The pathologist had no knowledge of the study protocol prior to examining the tissues. Lymph nodes were evaluated on the basis of overall architecture and cellularity, prominence of follicles, architecture of follicles, mitotic index, edema and inflammatory infiltrates. Inflammatory infiltrates were examined for the presence of eosinophils, neutrophils, mast cells, macrophages, plasma cells, and lymphocytes. The observed architectural features and cell morphology were used to differentiate between normal, hyperplastic or neoplastic tissue. The interpretation of the hematoxylin and eosin stained tissue was subjective and based on the professional experience of the board certified veterinary pathologist. 2.7.2. Statistical analysis Statistical analyses were performed using Sigma Plot 11 (Systat Software Inc., San Jose, CA). A normality test was performed using Shapiro–Wilk and the Mann–Whitney rank sum test was used to ascertain whether the difference in median values between the control and acute laminitic horses was statistically significant; a p value of <0.05 was considered statistically significant. When tests for normality failed, the data sets were analyzed using the Kruskal– Wallis one way analysis of variance to determine the median; difference in median values was evaluated using the Dunn’s method test for multiple comparisons. A p value of < 0.05 was considered statistically significant. 3. Results 3.1. Clinical signs of laminitis Clinical signs of carbohydrate-induced laminitis were identical in the two models and no significant differences were observed between the tests in terms of the time required to develop diarrhea, fever and lameness. Consistent with published descriptions of the clinical symptoms, non-bloody, watery diarrhea developed prior to an increase in rectal temperature, which in turn preceded development of lameness. Fig. 1 demonstrates that the time to develop clinical signs was less variable among animals induced with OF compared to CS but statistically there was no difference between the models in time to symptom development. 3.2. Enumeration of bacteria in the cecal fluid Enumeration of Gram-positive and Gram-negative cecal bacteria was based on recovery of bacterial colonies on selective media. Fig. 2, panel A, demonstrates a significant increase of 2 to 3 orders of magnitude in Gram-positive bacteria recovered on COBA plates, which is a selective medium used to recover Lactobacillus spp. and Streptococcus spp., as well as other Gram-positive cocci. Mean bacterial colony numbers were significantly greater

1 Robert L. Donnell, DVM, PhD, Dip. ACVP. Department of Pathobiology, College of Veterinary Medicine, University of Tennessee, 2407 River Drive, Knoxville, TN 37996-4542, United States.

358

J.C. Onishi et al. / Veterinary Microbiology 159 (2012) 354–363

all control horses, regardless of the carbohydrate used to induce laminitis. The lymphocenters appeared as well formed structures with diverse shapes ranging from 2 to 25 mm in size. The nodes were embedded in fat tissue that adhered to the surface of the cecum and colon and were creamy-yellow to orange-yellow in color and firm in texture. The lymphocenters collected from CS- and OFinfused horses were identical in appearance. 3.4. Histolopathologic examination of lymph nodes draining cecum and colon In five of six OF-induced horses lymphoid hyperplasia was evident and graded at a level of moderate to marked. Eosinophilic infiltrates in both the intranodal or extranodal spaces were detected in six of six OF-induced horses. Fig. 1. Dot plots showing time of onset of individual animal responses to carbohydrate overload in equine acute laminitis models. All data sets were analyzed for normality and passed. The time of onset of diarrhea did not differ significantly (p = 0.45) following a bolus of cornstarch (mean (M) = 15.33 h, standard deviation (SD) = 10.09, standard error of the mean (SEM) = 4.12) compared to oligofructan (M = 12.0 h, SD = 2.53, SEM = 1.03). The time to develop a fever was not significantly different between the models (cornstarch: M = 18.33 h, SD = 9.42, SEM = 3.84; oligofructan: M = 22 h, SD = 6.93, SEM = 3.46; p = 0.53). Two animals in the oligofructan-treated group did not develop a fever and are indicated in the plot with ‘‘X’’. The time to develop Obel grade 2 lameness did not differ significantly in the two models (cornstarch: M = 29 h, SD = 6.78, SEM = 2.7l; oligofructan: M = 26 h, SD = 2.19, SEM = 0.89; p = 0.33).

(p < 0.05) in both models compared to control animals. Fig. 2, panels B, C and D, indicates that other groups of Gram-positive and Gram-negative cecal bacteria had increased by the time Obel grade 2 lameness was apparent. Increases of 3 and 5 orders of magnitude were detected on the Mannitol Salts agar media used to select Staphylococcus spp., but induced and control animals were not significantly different, due to large individual animal variation. Gram-negative bacteria recovered on MacConkey and Salmonella/Shigella agar were 5 and 3 orders of magnitude greater in OF-infused horses compared to controls (p < 0.05). Numbers of Gram-negative colonies increased in CS-infused compared to control horses but were not significantly different, due to large individual animal variation. The recovery of bacteria on MacConkey and Salmonella/Shigella agar indicates that cecal fluid from both CS- and OF-infused horses had increased levels of both Enterobacteriaceae and Salmonella spp. 3.3. Gross anatomy of lymph nodes draining cecum and colon Cecal and colonic lymph nodes collected from the control horses appeared physically different than those from horses with CS- or OF-induced laminitis. The lymphatic tissues collected from all control horses were identical to those described by Saar and Getty (1975). The cecal/colon lymphocenters were embedded in fatty tissue attached to the surface of the cecum and colon and were present as numerous small, opaque white nodules, 1–2 mm in size. The cecal/colonic lymphocenters collected from horses with acute laminitis were distinctly different from those of

3.5. Profiling bacteria in cecal/colon lymph nodes Cecal/colonic lymph nodes collected from control and all acute laminitic horses were culture-negative on Columbia agar with 5% horse blood. On selective agar media, lymph nodes in all but two horses were culture negative. One control horse had 3  103 CFU/g of a Grampositive bacterium tentatively identified as Listeria spp. based on the appearance of black colonies on Palcam agar while one OF-induced horse had bacteria at 3  103 CFU/g tissue on COBA plates. Culture-independent methods to detect bacterial DNA in lymph nodes homogenates were based on the detection of PCR products using universal bacterial primers to the ribosomal 16S rRNA gene. Analysis of the PCR product by agarose gel electrophoresis revealed the presence of two bands in five out of eight tissue samples from control horses (Fig. 3). One product had a molecular weight of 200 bp while a second amplicon had a smaller molecular weight. Based on comparison to control samples prepared from bacteria alone or lymph node tissue alone, the band at 200 bp was concluded to originate from bacterial DNA. In order to identify the smaller of the two PCR products generated with the primers for the 16S rRNA gene, the PCR product was re-amplified with primers containing GC clamps and the amplicons were separated by DGGE analysis. Only one DGGE band was observed in tissue samples and the DNA sequencing data was 100% identical to the Equus caballus (horse) ribosomal 18S rRNA gene. PCR products obtained from pure bacterial cultures of E. coli and S. aureus either alone or spiked at 1  108 CFUs into lymph node tissue samples showed bands in the DGGE gels that were larger and well separated from the horse 18S rRNA amplicon. Since the primers to the universal bacterial 16S rRNA gene also detected the equine 18S rRNA gene, the sensitivity of the assay for detecting bacteria was reduced due to primer bias. In an attempt to develop the most sensitive method to detect bacterial DNA in the presence of mammalian DNA, tissue extracts were tested for the presence of bacterial ribosomal 23S rRNA gene and for bgalactosidase (lacZ) using a quantitative PCR assay. The limit of detection of the qPCR assay was determined by preparing a dilution series of pure E. coli B, extracting the

J.C. Onishi et al. / Veterinary Microbiology 159 (2012) 354–363

359

Fig. 2. Dot plots showing the recovery of cecal bacterial colonies on selective agar from cecal fluid collected at necropsy from control, cornstarch- and oligofructan-infused horses. The normality test (Shapiro–Wilk) failed with all data sets; the data were therefore analyzed using the Kruskal–Wallis one way analysis of variance test. The median values (M) of CFU/ml cecal fluid for each data set on each selective media are as follows. On COBA, the control M: 1.7  105; the CS M: 2.7  107 and the OF: M: 5.0  108; On MSA the control M: 1.4  103; the CS M: 2.3  105 and the OF M: 1.5  106; on MacConkey the control M: 2.2  103; the CS M: 9.5  106 and the OF M: 2.0  108; on Salmonella/Shigella agar the control M: 3.5  102; the CS M: 4.5  105 and for OF M: 2.7  106. To determine differences between groups on a single media with significant differences, data sets were analyzed using Dunn’s method for pairwise comparisons. Sets in which significance (p  0.05) in median values of treated and control animals differed are indicated by ‘‘*’’.

DNA and testing the dilution series in the qPCR assay. The limit of detection for this present/absent (+/) assay corresponded to 1  106 CFUs and 7.5  104 CFUs, for the ribosomal 23S rRNA and for b-galactosidase genes, respectively. Amplicons formed in the presence of the ribosomal 23S rRNA primer were present in extracts prepared from five of eight control horses while none of the samples were positive for the b-galactosidase gene, lacZ. Samples from 11 of 12 acute laminitic horses were negative for both genes. A single horse had DNA that could be accounted for by at least 7.5  107 CFUs of bacteria, based on PCR assay with ribosomal 23S rRNA primers or 3.0  105 CFUs of enterobacteria based on the use of lacZ primers.

rRNA gene. Distinct, single bands labeled A, B and C in Fig. 4, panel A, were present in three of five samples. Phylogenetic analysis of the sequenced bands (Fig. 4B) revealed that bands B and C corresponded to two Gammaproteobacteria, Pseudomonas fluorescens and E. coli, respectively. The percent identity with reference strains was 100%. Band A, strongly present in one sample, showed 75.9% identity with a Betaproteobacteria, Comamonas testosteroni LMG1800. This level of homology does not allow for identification of the phylotype with confidence.

3.6. Identification of bacterial 23S rRNA gene amplicons by DGGE

The total number of aerobic bacteria in laminae homogenates from control samples did not differ significantly from those with acute laminitis. Five of eight control samples were culture negative for aerobic bacteria, but bacteria were recovered from the laminae of three horses. Five times more total bacteria were recovered from

DGGE revealed the presence of multiple amplicons in DNA lymph node extracts prepared from five of five control horses generated with primers specific to the bacterial 23S

3.7. Culture-dependent evaluation of bacteria in extraintestinal tissue

360

J.C. Onishi et al. / Veterinary Microbiology 159 (2012) 354–363

Fig. 3. High resolution agarose gel electrophoresis of the PCR amplified 16S rRNA genes from lymph node extracts prepared from control horses (figure A) or horses with acute laminitis induced with cornstarch (lanes 1–6) or oligofructan (lanes 7–12) (figure B). Numbers at the top of the lanes indicate the horse identification. The mobility of the amplified products is compared to standard molecular weight oligonucleotide ladder (L) with the weights indicated in the figure.

horses induced with CS (M = 1264, SD = 1792) compared to control horses (M = 229, SD = 344). The difference, however, was not statistically significant (p = 0.14). Bacterial populations in laminae collected from horses induced with OF (M = 141, SD = 219, p = 0.54) did not differ significantly compared to control. Tissue homogenates prepared from liver and spleen from control and acute laminitic horses were culture negative. All blood samples were culture negative. 4. Discussion It is well established that changes in cecal microbiota are associated with the development of lameness in the equine carbohydrate overload models of acute laminitis. The results of this study show that the changes may be

broader than currently appreciated and include an overgrowth of potentially pathogenic bacteria. Culturedependent and -independent methods have shown that following oral infusion of either CS or OF, cecal Grampositive populations of Lactobacillus sp. and Streptococcus sp. increase (Garner et al., 1978; Milinovich et al., 2008). The increase is time-dependent with overgrowth of lactobacilli preceding the increase in streptococci. We confirmed that the numbers of cecal Gram-positive bacteria increased by more than 2 orders of magnitude following CS infusion and an increase of 3 orders of magnitude occurred following OF infusion. The increase was statistically significant in both sets of animals when bacteria were enumerated on COBA agar, a selective media used to recover lactobacilli and streptococci. Although the horses in the present study originated from multiple

J.C. Onishi et al. / Veterinary Microbiology 159 (2012) 354–363

361

Fig. 4. Phylogenetic analysis targeting bacterial 23S rRNA gene in the lymph node tissues of non-laminitic horses (A) PCR-DGGE analysis and (B) neighborjoining tree based on DNA sequences of three DGGE bands and related 16S rRNA genes. Phylogenetic tree with their relatives derived from the 205 bp regions of the 23S rRNA gene. The neighbor-joining method was used to construct phylogenetic trees, and bootstrap values (1000 replicates) higher than 50% were indicated at the branch points. [GenBank accession numbers: Band A, JN244993; Band B, JN244994; Band C, JN244995].

facilities and diet prior to the start of the test were not identical, there is agreement between the present and earlier studies that following infusion of either OF or CS, Gram-positive bacteria increase in the cecal fluid within 26 to 29 h of carbohydrate overload. The results also show that the increase is not restricted to lactobacilli and streptococci but is extended to include at least staphylococci. This study also showed that compared to control horses, numbers of Gram-negative cecal bacteria increased by the time lameness developed. A significant increase of 5 and 3 orders of magnitude was detected for Enterobacteriaceae and Salmonella, respectively, in OF-infused horses. The same groups of bacteria increased in the CS model but there was greater variation between individuals. The significance of the results is that in a functional sense, there is an overgrowth of Gram-negative cecal bacteria rather than a die-off of Gram-negative bacteria by the time that mild lameness had developed. Longitudinal studies in which culture-independent methods were used, a timedependent shift in E. coli strains was detected (Milinovich et al., 2008). The concept that a die-off of Enterobacteriaceae occurs is based on a comparison of bacterial cell counts in which the cecal fluid at an initial time point had more than 5  1010 bacteria (Garner et al., 1978). Both the present study and the study by Garner et al. (1978) incubated the MacConkey agar plates aerobically to enumerate this group of intestinal bacteria. The tests differ however in that horses were fitted with cannulaes placed in cecal fistulas in the early study but not in the present study. There is a difference of 7-logs in the enumeration of Enterobacteriaceae in the control animals between the two studies; further study is required to establish numbers of viable gut bacteria in control horses. We propose that the discrepancy can be explained if acid-sensitive organisms die but acid-resistant bacteria survive and overgrow in the bowel. By the time that mild lameness has developed in the model, bacteria have

fermented the carbohydrate used to induce lameness and have produced lactic acid. An overgrowth of diverse bacterial communities following carbohydrate overload is not unexpected since it is known that acid-resistant Gram-negative bacteria exist in nature (Rowbury, 2001). Numerous potentially pathogenic bacteria in the intestinal tract have lactate dehydrogenase and therefore the potential for growth using lactic acid as a carbon source (Garvie, 1980). Many of these organisms are Gramnegative bacteria, such as E. coli, Pseudomonas sp. and Salmonella sp., but Gram-positive bacteria, such as Staphylococcus sp are also known to metabolize lactic acid. Despite decades of research, it remains unclear how sudden changes in the gut microbiota and development of both localized and systemic inflammation in the carbohydrate overload models of acute laminitis are connected. Bacterial derived bioactive metabolites such as endotoxin and vasoactive amines and possibly activated proteases are postulated to play a role in the disease (Bailey et al., 2003; Garner et al., 1975; Mungall et al., 2001; Sprouse et al., 1987; van Eps, 2006). To this date, it remains unclear how these metabolites are involved and whether plasma levels, particularly in the case of the vasoactive amines, reach physiologically relevant concentrations in the model. In this study we explored the possibility that bacterial translocation provides a mechanism to deliver bioactive metabolites produced by gut bacteria to the circulatory system and to laminar tissue. Bacterial translocation is defined as the movement of intestinal bacteria to extra-intestinal sites (Berg, 1999) and is detected in rodent models using culture dependent methods (Deitch et al., 1990; Pang et al., 2007). Using these methods, evidence of bacterial translocation to the lymph was only detected in two horses, one control and one OF-infused horse. The strongest evidence that bacterial translocation occurs in horses is the detection of bacterial DNA in cecal/colon lymph nodes in five of eight control animals. Using the universal bacterial 16S rRNA

362

J.C. Onishi et al. / Veterinary Microbiology 159 (2012) 354–363

gene primers, and based on the limit of detection established using a dilution series of pure E. coli, it was estimated that at a minimum 1.0  106 CFU equivalents of E. coli could be detected in the culture-independent approach. Further analysis using primers to the 23S rRNA gene showed that DNA from multiple Gram-negative bacteria was present in cecal/colonic lymph nodes from three out of eight control horses. Bacteria identified with the highest degree of certainty included E. coli and P. fluorescens, both known to inhabit the mammalian gastrointestinal tract (Madi et al., 2010). It is concluded that the bacterial DNA in the lymph nodes from control animals was not due to contamination but rather represents gut-derived Gram-negative bacteria that were translocated and processed within the lymphatic system presumably to generate mucosal immunity (Macpherson and Uhr, 2004). Precautions to prevent contamination of these samples by laboratory contaminants included preparation of a sterile field at necropsy, placement of the sample at necropsy in sterile containers, processing of the samples in biological safety hoods, homogenization of the samples manually using sterile bags with filter inserts and using separate laboratory areas for DNA extraction and PCR set-up. The failure to detect bacterial DNA in all samples is likely to be related to the limit of detection of the existing assays and the need for assay optimization including tissue stabilization protocols and DNA extraction protocols. In the acute laminitic horses, there was no evidence that an overgrowth of potentially pathogenic bacteria in the cecum was associated with seeding of extra-intestinal tissues with bacteria. All tissues, including liver, spleen and the majority of cecal/colon lymph nodes as well as blood were culture negative; the numbers of bacteria in the laminar tissue did not differ from control animals. Using culture-independent methods, bacterial translocation to the lymph nodes was not detected. We suggest that the failure to detect bacterial DNA in lymph nodes collected from acute laminitic horses is a false negative result related to the fact that the tissue is immunologically activated and that experimental protocols require optimization to accurately detect microbes in the tissue. The detection of amplicons due to equine 18S rRNA in all lymph node extracts from the acute laminitic horses shows that the inability to detect amplicons due to bacterial 16S rRNA genes is not due to the presence of a PCR inhibitor in the extracts. The differences in bacterial cecal bacterial communities and the physical differences observed between the cecal/ colon lymph nodes collected from control compared to all horses receiving carbohydrate rations is striking and does not appear to be simply due to chance. Based on the study design, it is not clear whether the alterations are due to differences in horse management practices prior to carbohydrate infusion or to carbohydrate associated pathology. Further, the histological findings were observational and restricted by the fact that objective criteria for measuring inflammatory cell infiltrates in equine lymph nodes have not been established. The sets of animals with the most similar horse management routines prior to the carbohydrate infusion were the control and the OF-infused horses. The cecal/colon lymph nodes from all horses receiving the carbohydrate ratios were larger than those collected from all control horses and formed the basis for

the conclusion that lymphadenopathy was present. This conclusion is based on gross anatomical examination of the tissue collected at necropsy with a comparison of control equine cecal/colon lymph tissue as described (Saar and Getty, 1975). The conclusion that hyperplasia and eosinophilia was present in all OF-infused horses was based on the microscopic examination of H & E stained tissue by a veterinary pathologist, who was blinded, with extensive experience in reviewing lymph nodes from equine samples submitted for diagnostic purposes. Clarifying the exact response in cecal bacterial communities and in the cecal/ colon lymph nodes in the carbohydrate overload model of acute laminitis will require a research study involving the use of appropriate control animals. Although lymphadenopathy and eosinophilia are commonly associated with parasite infection, other conditions may lead to these conditions (Bochner and Gleich, 2010). Based on the rapid development of clinical signs in this model, it is unlikely that eosinophilia detected in the lymph nodes could be accounted for by an overgrowth of intestinal parasites. Changes in gut microbiota have been associated with intestinal eosinophilia in infants with bloody stools (Nevoral et al., 2009). If hyperplasia and eosinophilia in the lymph nodes collected from OF-infused horses is correct, we suggest that the tissue is immunologically activated. The tissue could be activated by translocation of intact gut-derived bacteria or their components. Activated eosinophils produce reactive oxygen species, cytokines, chemokines, lipid mediators, cytotoxic cationic proteins, elastase as well as other factors (Yan and Shaffer, 2009). When viable gut-derived bacteria translocate to the cecal/colon lymph nodes, it is thought that microbes will be rapidly killed and presumably degraded to a wide range of bacterial components that are capable of further activating an immunological response. Factors that could activate an immune response, called PAMPS, include endotoxin, peptidoglycan fragments, flagellin proteins and DNA (Doyle and O’Neill, 2006). We propose that when the lymphatic tissue is activated, bacterial killing leads to DNA degradation which accounts for our inability to detect bacterial DNA in the acute laminitic horses. The factors in the activated lymph tissue could disseminate hematogenously to account for the systemic and localized inflammatory response known to occur in the model. This proposal, however, fails to explain how vasoactive amines contribute toward the clinical symptoms found in the model. It is possible that these bacterial derived bioactive metabolites remain in the lumen of the gut and directly affect the function of the intestinal epithelial cells (IEC). Several lines of evidence show that intestinal epithelial cells are altered in the model. Within 24 h of CS infusion, the cells surrounding the crypt openings on the surface of the intestinal epithelium are swollen and the ultrastructure of epithelial cells reflected the initiation of primary degeneration process (Krueger et al., 1986). Within the same time frame, increased recovery of technetium Tc99m diethylenetriaminopentaacetate (99m Tc-DTPA) in urine in ponies following a CS ration indicates that the intestinal tract was unable to retain a biomarker with a molecular weight of 500 g/mol (Weiss et al., 1998). With molecular weights

J.C. Onishi et al. / Veterinary Microbiology 159 (2012) 354–363

well below that of 99m Tc-DTPA, it is not clear why vasoactive amines are not detected in the plasma. Research to understand how IECs respond to rapid shifts in gut microbiota and their metabolites is an unexplored in the carbohydrate overload model of acute laminitis. An IEC response could contribute to the overall immunological response as IEC are now known to be lined with a range of Toll-like receptors that could respond to rapid shifts in the gut microbiota (Abreu, 2010). 5. Conclusion Carbohydrate overload models of equine acute laminitis involve changes in cecal bacterial communities and production of bioactive metabolites such as lactic acid, vasoactive amines and endotoxin. It is not clear how these metabolites contribute toward the development of both systemic and localized inflammatory responses that in this model lead to damaged laminar tissue. The results of this study confirm the changes in Gram-positive cecal bacteria described by others but also show that significant overgrowth of potentially pathogenic bacteria Gram-positive and Gram-negative bacterial communities occur in the equine cecum within the time frame of early stage lameness. Translocation of viable bacteria to extraintestinal tissues was not detected in the model but preliminary evidence that the cecal/colon lymphatic system is activated is presented. The process leading to laminitis is postulated to begin with rapid and massive changes in the gut bacterial communities which serve as a trigger to drive an immune response. Future research to study how the intestinal epithelial cells and the underlying lymphatic system respond to sudden changes in gut bacterial communities as well as their metabolites will advance understanding how laminitis develops following rapid ingestion of non-digestible carbohydrates. Acknowledgements Financial support for this project was provided by the New Jersey Agricultural Experiment Station through the State Equine Initiative with the endorsement of the Rutgers Equine Science Center and the director, Karyn Malinowski, PhD. The authors have no competing financial interests. References Abreu, M., 2010. Toll-like receptor signalling in the intestinal epithelium: how bacterial recognition shapes intestinal function. Nat. Rev. Immunol. 10, 131–143. Altschul, S., Madden, T., Schaffer, A., Zhang, A., Miller, W., Lipman, D., 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, 3389.s–3402.s. Bailey, S., Adair, H., Reinemeyer, C., Morgan, S., Brooks, A., Longhofer, S., Elliott, J., 2009. Plasma concentrations of endotoxin and platelet activation in the developmental stage of oligofructose-induced laminitis. Vet. Immunol. Immunopathol. 129, 167–173. Bailey, S., Baillon, M., Rycroft, A., Harris, P., Elliott, J., 2003. Identification of equine cecal bacteria producing amines in an in vitro model of carbohydrate overload. Appl. Environ. Microbiol. 69, 2087–2093. Belknap, J., Moore, J., Crouser, E., 2009. Sepsis-from human organ failure to laminar tissue. Vet. Immunol. Immunopathol. 129, 155–157. Berg, R.D., 1999. Bacterial translocation from the gastrointestinal tract. Adv. Exp. Med. Biol. 473, 11–30.

363

Bochner, B.S., Gleich, G.J., 2010. What targeting the eosinophil has taught us about their role in diseases. J. Allergy Clin. Immunol. 126, 16–25. Campanella, J.J., Bitincka, L., Smalley, J., 2003. MatGAT: an application that generates similarity/identity matrices using protein or DNA sequences. BMC Bioinformatics 4, 29. Deitch, E., Morrison, J., Berg, R., Specian, R., 1990. Effect of hemorrhagic shock on bacterial translocation, intestinal morphology, and intestinal permeability in conventional and antibiotic-decontaminated rats. Crit. Care Med. 18, 529–536. Doyle, S.L., O’Neill, L.A.J., 2006. Toll-like receptors: from the discovery of NFkB to new insights into transcriptional regulations in innate immunity. Biochem. Pharmacol. 72, 1102–1113. Garner, H.E., Coffman, J.R., Hahn, A.W., Hutcheson, D.P., Tumbleson, M.E., 1975. Equine laminitis of alimentary origin: an experimental model. Am. J. Vet. Res. 36, 441–444. Garner, H.E., Moore, J.N., Johnson, J.H., Clark, L., Amend, J.F., Tritschler, L.G., Coffman, J.R., Sprouse, R.F., Hutcheson, D.P., Salem, C.A., 1978. Changes in the caecal flora associated with the onset of laminitis. Equine Vet. J. 17, 249–252. Garvie, E., 1980. Bacterial lactate dehydrogenase. Microbiol. Rev. 44, 106–139. Krueger, A.S., Kinden, D.A., Garner, H.E., Sprouse, R.F., 1986. Ultrastructural study of the equine cecum during the onset of laminitis. Am. J. Vet. Res. 47, 1804–1812. Macpherson, A., Uhr, T., 2004. Induction of protective IgA by intestinal dendritic cells carrying commensal bacteria. Science 303, 1662–1665. Madi, A., Svinareff, P., Orange, N., Feuilloley, M.J., Connil, N., 2010. Pseudomonas fluorescens alters epithelial permeability and translocates across Caco-2/TC7 intestinal cells. Gut Pathog. 2, 16. Menzies-Gow, N.J., Stevens, K.B., Sepulveda, M.F., Jarvis, N., Marr, C.M., 2010. Repeatability and reproducibility of the Obel grading system for equine laminitis. Vet. Rec. 167, 52–55. Milinovich, G.J., Burrell, P.C., Pollitt, C.C., Klieve, A.V., Blackall, L.L., Ouwerkerk, D., Woodland, E., Trott, D.J., 2008. Microbial ecology of the equine hindgut during oligofructose-induced laminitis. ISME J. 2, 1089–1100. Mungall, B., Kyaw-Tanner, M., Pollitt, C.C., 2001. In vitro evidence for a bacterial pathogenesis of equine laminitis. Vet. Microbiol. 79, 209–223. Muyzer, G., de Waal, E., Uitteerlinden, A., 1993. Profiling of complex microbial populations by denaturing gradient gel electrophoresis analysis of polymerase chain reaction-amplifed genes coding of 16S rRNA. Appl. Environ. Microbiol. 59, 695–700. Nevoral, J., Rada, V., Vlkova´, E., Bla´hova´, K., Bronsky´, J., Buba´kova´, D., Killer, J., 2009. Intestinal microbiota in exclusively breast-fed infants with blood-streaked stools. Folia Microbiol. (Praha) 54, 167–171. Onishi, J.C., Park, J.W., Ha¨ggblom, H.M., Fennell, M.J., Fugaro, M.N., 2012. Chronic laminitis is associated with potential bacterial pathogens in the laminae. Vet. Microbiol., http://dx.doi.org/10.1016/j.vetmic.2012.02.024. Pang, H.J., Lo, C.Y., Matthews, K.R., 2007. Influence of immunomodulation on the development of Listeria monocytogenes infection in aged guinea pigs. FEMS Immunol. Med. Microbiol. 49, 415–424. Park, J.W., Krumins, V., Kjellerup, B.V., Fennell, D.E., Rodenburg, L.A., Sowers, K.R., Kerkhof, L.J., Ha¨ggblom, M., 2011. The effect of cosubstrate activation on indigenous and bioaugmented PCB declorinating bacterial communities in sediment microcosms. Appl. Microbiol. Biotechnol. 89, 2005–2017. Rowbury, R., 2001. Extracellular sensing components and extracellular induction component alarmones give early warning against stress in Escherichia coli. Adv. Microb. Physiol. 44, 215–257. Saar, L.I., Getty, R., 1975. Equine lymphatic system. In: Getty, R. (Ed.), The Anatomy of the Domestic Animal. W. B. Saunders Company, Philadelphia, pp. 619–630. Sawant, A.A., Shreekuma, R., Pillai, R., Jayarao, B.M., 2002. Evaluation of five selective media for isolation of catalase-negative gram-positive cocci from bulk tank milk. J. Dairy Sci. 85, 1127–1132. Sprouse, R., Garner, H., Green, E., 1987. Plasma endotoxin levels in horses subjected to carbohydrate induced laminitis. Equine Vet. J. 19, 25–28. Tamura, K., Dudley, J., Nei, M., Kumar, S., 2007. MEGA4: molecular evolutionary genetics analysis (MEGA) software version 4.0. Mol. Biol. 24, 1596–1599. van Eps, A.W.a.C.C.P., 2006. Equine laminitis induced with oligofructose. Equine Vet. J. 38, 203–208. Weiss, D.J., Evanson, O.A., MacLeay, J., Brown, D.R., 1998. Transient alteration in intestinal permeability to technetium Tc99m diethylenetriaminopentaacetate during the prodromal stages of alimentary laminitis in ponies. Am. J. Vet. Res. 59, 1431–1434. Yan, B.M., Shaffer, E.A., 2009. Primary eosinophilic disorders of the gastrointestinal tract. Gut 58, 721–732.