Laboratory animals
C57BL/6J and C56BL/6 IL-10−/− mice were obtained from The Jackson Laboratories (Bar Harbor, ME). A breeding colony was established in a Campylobacter/Helicobacter-free facility, and the MouSeek database (Caleb Davis, Baylor College of Medicine, Houston, TX) was used to track all mice bred and used throughout the study. Germ-free C57BL/6J mice were also obtained from the same containment building at The Jackson Laboratories. All mouse experiments were performed according to recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. Protocols were reviewed and approved by the Michigan State University Institutional Animal Use and Care Committee (approval numbers 06/12-107-00 and 06/15-101-00). Age-matched male and female mice were used for all experiments. A portion of the mice in each experiment possessed humanized microbiota generated as described previously [45] and are indicated with the prefix Hu (Humicrobiota). Briefly, germ-free mice were inoculated by gavage with a human fecal slurry, bred, and the microbiota allowed to pass from mother to offspring without intervention within the germ-free incubator. After initial characterization in founder mice described in Collins et al. 2015 [45], the Humicrobiota mice were separated into two groups, and a new colony was established (Hu-C57BL/6) by LS Mansfield by transferring mice in sterile filter top cages within sterile dog crates to Michigan State University. Humicrobiota mice were housed under specific pathogen-free conditions (SPF), and bred for six generations in closed cages on an Innovive (San Diego, CA, USA) mouse rack with filtered air flow and sterile food and water. All cage changes and other manipulations were performed in a laminar flow hood with gowned and gloved personnel using sterile technique to avoid introduction of microorganisms from the environment.
In the pilot experiment, age-matched 10–12-week-old Humicrobiota C57BL/6 genetically wild-type (Hu), conventional microbiota genetically wild-type (Convmicrobiota), and Convmicrobiota congenic IL-10-deficient mice (Conv-IL-10−/−) were used. Mice were inoculated with either tryptone soy broth (TSB; vehicle control), C. jejuni 260.94, or C. jejuni 11168 and handled with sterile or specific pathogen-free (SPF) technique resulting in six groups (Table 1). For SPF technique, all personnel that were handling animals wore Tyvek coveralls, impermeable plastic booties, face mask, hair bonnet, and gloves. All cage changes were performed on a laminar flow cage changing station. For sterile technique, all personnel that were handling animals wore impermeable plastic booties, face mask, hair bonnet, sterile surgical gown, and sterile surgical gloves. All breeding mice were and continue to be handled using sterile technique to avoid introducing extraneous organisms to the microbiota. All cage changes for breeding mice were performed in a sterile laminar flow hood that was disinfected after each use. To determine if handing would alter outcomes in TSB sham-inoculated Humicrobiota mice, we inoculated 20 Humicrobiota mice with TSB and handled them with either sterile technique (Hu-TSB (Ster.)) or SPF technique (Hu-TSB (SPF)). Two other Humicrobiota groups were generated by inoculating Humicrobiota mice with either C. jejuni 260.94 (Hu-260.94 (Ster.)) or C. jejuni 11168 (Hu-11168 (Ster.)) and handling them with sterile technique. As a positive control for gastroenteritis, we inoculated and compared outcomes in Convmicrobiota wild-type C57BL/6 and C57BL/6 IL-10−/− mice inoculated with C. jejuni 11168 and handled with SPF technique. Mice were sacrificed at 5 weeks post-inoculation.
In Experiment 1, age-matched C57BL/6 Humicrobiota and Convmicrobiota mice were inoculated with TSB, C. jejuni 260.94, or C. jejuni 11168 (Table 2). In Experiment 1, all mice were handled with SPF technique, observed for 7 weeks post-inoculation, and then sacrificed. In all, six experimental groups were generated in Experiment 1; Convmicrobiota TSB inoculated (Conv-TSB), Convmicrobiota C. jejuni 260.94 infected (Conv-260.94), Convmicrobiota C. jejuni 11168 infected (Conv-11168), Humicrobiota TSB inoculated (Hu-TSB), Humicrobiota 260.94 infected (Hu-260.94), and Humicrobiota 11168 infected (Hu-11168).
Enteric pathogen screening
DNA was extracted from feces collected from all mice before experimental inoculation and at necropsy for enteric pathogen screening as described [30]. In all cases, no control mice were positive for C. jejuni PCR using gyrA-specific primers [46]. Also, we screened all samples for Campylobacter spp. (16S rRNA gene), Helicobacter spp. (16S rRNA gene), Citrobacter rodentium (espB gene), and Enterococcus faecalis (ddl gene). Dedicated sentinel mice were used to assess extraneous infection with bacteria, protozoa, and viral agents (Charles River Laboratories, Wilmington, MA) and were monitored by the MSU Campus Animal Resources (CAR).
C. jejuni strains and inoculum preparation
C. jejuni strains 260.94 (ATCC BAA-1234) and NCTC 11168 (ATCC 700819) were obtained from the American Type Culture Collection (Manassas, VA). C. jejuni 260.94 is a Guillain-Barré syndrome patient strain that elicits GM1 and GD1a anti-ganglioside antibody responses in C57BL/6 IL-10−/− mice [31]. C. jejuni 11168 is an enteric disease patient strain isolated from a patient with severe gastroenteritis. C. jejuni 11168 has a GM1 ganglioside mimic on its surface [47] but is not associated with GBS and has not been shown to elicit significant anti-ganglioside antibody responses in C57BL/6 IL-10−/− mice [31]. Inocula were prepared in the same manner for both experiments. Inocula of both C. jejuni strains were prepared by streaking frozen stocks onto tryptone soy agar (TSA) (Accumedia) supplemented with 5% defibrinated sheep blood (Cleveland Scientific, Bath Ohio) (TSAB). Plates were incubated at 37 °C in anaerobic jars equilibrated to 10% CO2, 10% H2, and 80% N2 for 48 h and a portion of the growth re-suspended in tryptone soya broth (TSB) to give an A600 of 0.2 to 0.3. One-hundred microliters of this suspension was spread on two plates per mouse and the plates incubated for 16 h in the 10% CO2, 10% H2, and 80% N2 gas mixture. The resulting cells were collected and suspended in TSB; the suspension was adjusted to give an A600 of approximately 1.0 when diluted 1:10 (approximately 1 × 1010 CFU/mL final concentration). Purity, morphology, and motility were verified by microscopy and Gram straining. Finally, 0.2 mL per mouse of the resulting inoculum or the vehicle (i.e., TSB) was carried to the containment facility on ice and delivered to infected and control mice, respectively, by oral gavage, resulting in six groups (Tables 1 and 2). Limiting dilution analysis was used to determine the actual titer of the inoculum delivered to the mice.
Experimental design
Following infection, all mice were observed at least once daily (twice daily after clinical signs were noted) by trained individuals for a period of 5 (Pilot) or 7 (Experiment 1) weeks to ensure mice were euthanized at a humane endpoint. In Experiment 1, 1 week before infection (i.e., baseline) and once each week for 7 weeks post-inoculation, mice underwent behavioral phenotyping in an open-field test in a sterile rat cage (18′′ × 8′′) divided into four quadrants located in a laminar flow hood. At 5- (Pilot) or 7-weeks (Experiment 1), the mice were sacrificed, and tissues were collected and stored for further analysis. Prior to humane euthanasia by CO2 overdose, fecal samples were collected, placed in TSB, frozen on dry ice, and quickly moved to a −80 °C freezer until thawing for DNA extraction. After euthanasia, mice were weighed and blood was collected by cardiac puncture, immediately mixed with 0.1 mL of 3.8% citrate, spun down, and plasma stored at −80 °C for analysis of plasma antibodies. During necropsy, two veterinarians (a pathologist and a gastroenterologist) observed and recorded any gross pathology prior to the removal of the GI tract. For the Pilot and Experiment 1, the cecum and colon were harvested, cut in half, and the halves flash frozen or streaked on TSAB-CVA plates for cytokine analysis and quantification of C. jejuni in these compartments, respectively. In Experiment 1, the ileocecocolic junction was harvested, infiltrated with 10% neutral-buffered formalin (NBF), placed in a cassette, and further fixed in NBF for 20–24 h and stored in 60% ethanol until processed for histological analysis.
Bacterial DNA isolation from feces and 16S ribosomal RNA gene analysis
In the pilot experiment, DNA was extracted from fecal samples using the QIAamp DNA stool kit (QIAGEN) according to manufacturer’s instructions. DNA concentrations were determined using a NanoDrop ND-1000 spectrophotometer and concentrations normalized. The quantity of Clostridium group 1, Clostridium group 1, Bacteroidetes, and Enterobacteriaceae were measured using an IQTM5 Multicolor Real-Time PCR Detection System. In Experiment 1, DNA was extracted from fecal samples using bead beating and the FastDNA SPIN Kit for Soil (MP Biomedicals, LLC) according to manufacturer’s instructions. The resulting DNA samples were delivered to the Michigan State University Research Technology Support Facility for library preparation and 16S rRNA gene amplicon analysis. In all, 62 samples were submitted for sequencing, including 60 mouse samples, the original fecal slurry used for inoculation of founder mice, and a mock community (HM-782D, BEI) for estimation of sequencing error. The V4 region of the 16S rRNA gene was amplified using dual-indexed primers [48]. PCR products were normalized using an Invitrogen SequalPrep DNA Normalization plate and the normalized products pooled. After quality control and quantitation, the pool was loaded on a standard MiSeq v2 flow cell and sequenced with a 500 cycle MiSeq v2 reagent kit (paired-end 250 base pair reads). Base calling was performed by Illumina Real-Time Analysis (RTA) v1.18.54 and output of RTA was de-multiplexed and converted to FastQ format files with Illumina Bcl2fastq v1.8.4.
16S rRNA gene amplicon analysis was performed using mothur (v. 1.35) and protocols available at http://www.mothur.org/wiki/MiSeq_SOP accessed December, 2015. Alignment was achieved using the Silva 16S ribosomal gene database [49]. Chimeric sequences and any sequences classified as chloroplast, mitochondria, Archaea, or Eukaryota were removed from the dataset using uchime and the mothur formatted version of the Ribosomal Database Project (RDP) training set version 9, respectively, per the mothur protocol. Sequences were clustered in operational taxonomic units (OTUs) of 97% sequence identity yielding 128 OTUs. Analyses were performed in mothur and PAST 3.07 [50]. Sequence read data has been made available in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) as documented in “Availability of data and materials.” A full record of the code used to develop the heat map that appears in Fig. 2, is based on the mothur protocol cited above. An annotated markdown file with the code for the heat map appears in Additional files 4 and 5.
Clinical signs assessments
We used a clinical sign score sheet developed to discern humane endpoints for gastrointestinal and neurological disease in mice; these have been approved by the MSU institutional animal care and use committee (IACUC) and published [30, 33, 35]. Briefly, mice were observed once a day by trained animal handlers and when clinical signs were discerned, they were documented and the mice were thereafter observed twice a day. A score sheet was filled out each time a mouse showed a clinical sign. Each sign has a point value and the scores for all signs observed were totaled for that observation period. If the score equaled or exceeded 9 then the mouse was humanely euthanized. Mice were assigned scores for a battery of clinical signs according to this scoring system: (1) Eating/Drinking (0 = yes, 1 = No); (2) Respiration (0 = normal, 1 = abnormal (increased), 10 = labored); (3) Rough hair coat (0 = no, 2 = yes), Hunched posture (0 = no, 9 = yes), Tremors (0 = no, 10 = yes), Movement (0 = normal, 1 = subdued (moves with stimulation), 2 = unresponsive to handling), Crusty eyes (0 = no, 1 = one eye, 2 = 2 eyes), Diarrhea on fur (0 = no, 1 = yes), Cool to the touch (0 = no, 10 = yes), and Body weight (0 = 0–1% weight loss, 1 = 1–5% weight loss). Endpoints resulting in a score greater than 9 include loss of body weight greater than 5%, cool to touch, blue extremities, or points adding up to greater than 9 in other criteria.
Quantification of C. jejuni in the cecum and colon
C. jejuni in the colon and cecum were quantified using a standardized semi-quantitative scoring system [30]. Briefly, colon and cecum tissue segments of the same size were collected at necropsy and were streaked on TSAB containing cefoperazone (2 μg/mL), vancomycin (10 μg/mL), and amphotericin B (2 μg/mL) (all antibiotics were obtained from Sigma-Aldrich, St. Louis MO) agar plates and grown in anaerobic jars equilibrated with CampyGen sachets (Oxoid) at 37 °C for 48–72 h. The resulting growth was assigned a score on a scale of 0–4 based on the density of growth; 0 (no growth), 1 (1–20 CFU), 2 (20–200 CFU), 3 (200–400 CFU), and 4 (confluent growth) as described [30].
Neurological phenotyping
Starting 1 week before experimental infections and then daily after inoculation with a C. jejuni strain, mice were observed daily for evidence of enteric and neurological disease. Daily monitoring was based on previously published clinical exam score sheets designed to score feature of gastrointestinal and neurological signs [30, 33]. Additionally, open-field testing was performed to detect neurological signs and changes in behavior due to inoculation with either GBS-associated or enteric-associated strains of C. jejuni. All TSB sham-inoculated control mice served as controls for phenotyping. The activity of all experimental mice was video-recorded once per week for 1 week before inoculation and once per week for 7 weeks post-inoculation. Briefly, mice were placed in the center of an 18′′ × 8′′ sterile rat cage divided into four marked quadrants and allowed to move freely for 90 s. At the completion of the experiment, a single investigator (PTB), who was blinded to mouse group identity, recorded the number of quadrants crossed and the number of rears for each mouse. Quadrants crossed were counted starting with the first line crossed after establishing all four limbs in a single quadrant. Rears were counted as the extension of hind limbs and placement of both front limbs on the side of the cage.
Scoring of ileocecocolic junction histopathology
Tissue samples were collected at necropsy, placed in cassettes, fixed in 10% NBF (Fisher Scientific) for 20–24 h, and then transferred to 60% ethanol until final processing. Samples were submitted to the Michigan State University Investigative Histopathology Laboratory where they were processed in the following manner: fixed samples were vacuum-infiltrated with paraffin on the Sakura VIP 2000 tissue processor; followed by embedding with a ThermoFisher HistoCentre III embedding station. Paraffin-embedded blocks were sectioned at 4–5 μm with a rotary microtome, dried at 56 °C in a slide incubator for 2–24 h, and stained with Hematoxylin and Eosin (H&E). Scoring of the distal ileum, cecum, and proximal colon was performed as described [30]. Briefly, the lumen, epithelium, lamina propria, and submucosa of the ileocecocolic junction (ICCJ) of each mouse were observed for histopathological changes by a single investigator (LSM) blinded to sample identity, and a score from 1 to 41 was assigned based on lesions using a standardized scoring system. Specific features evaluated among others were as follows: (1) excess mucus and inflammatory exudates in the lumen; (2) surface integrity, intraepithelial lymphocyte number, goblet cell hypertrophy, goblet cell depletion, crypt hyperplasia, crypt atrophy, crypt adenomatous changes, and crypt inflammation in the epithelium; (3) increased immune cells in the lamina propria; (4) and fibrosis in the submucosa.
Cytokine analysis
RNA was extracted from proximal colon samples that were flash frozen at the time of necropsy. Equal sized 5-mm-cubed tissue snips were homogenized using micropestles, and RNA was extracted following the RNeasy Plus Mini Kit protocol (QIAGEN). RNA concentrations were measured using the Nanodrop ND-1000 spectrophotometer and standardized to a concentration of 50 ng/μL. cDNA was obtained by PCR with random primers. A master mix was assembled using reagents from Promega GoTaq qPCR kit and added to the samples. This reaction was run using the following thermal cycler conditions: step 1, 5 min 25 °C; step 2, 20 min 42 °C; step 3, 70 °C; and step 4, 4 °C min—Hold. Interleukin 4 (IL-4) and Interferon gamma (IFNγ) cytokine levels were measured using qPCR on an iQ5 thermocycler (Bio-Rad) with standardization. ANOVAs were performed on 2-ΔΔct data to find the linear fold change in gene expression and are presented as mean fold change of three replicates over levels of the housekeeping gene hypoxanthine-guanine phosphoribosyltransferase (HPRT).
Enzyme-linked immunosorbent assays
Indirect enzyme-linked immunosorbent assays (ELISAs) were performed to test for the presence of antibodies reactive with bulk C. jejuni antigen and/or gangliosides GM1, GD1a, and GQb1 in the plasma of experimental mice, referred to as anti-Campylobacter and anti-ganglioside antibodies, respectively. Preparation of the bulk C. jejuni antigen was performed as previously described [30, 35]. Positive controls (highly reactive plasma samples that tested strongly for the presence of the antigen in previous experiments) and negative controls (monoclonal mouse anti-Toxoplasma gondii, ViroStat) were used in all cases. All samples were run in triplicate and the mean values used for statistical analysis. We tested for antibodies to gangliosides GM1 (Sigma), GD1a (USBio), and mixed GM1-GQ1b (Sigma, Calbiochem, respectively) [33]. Immunoglobulin (IgG) subtypes were determined using biotinylated goat anti-mouse-IgG1, IgG2b, IgG2c, and IgG3 (Jackson ImmunoResearch, West Grove, PA) secondary antibodies. Methods for C. jejuni-specific antibody ELISAs were described previously [30] and ganglioside ELISAs were conducted similarly [33].
Quantification of F4/80 positive cells in sciatic nerves and dorsal root ganglia
Sciatic nerves and 2–3 lumbar dorsal root ganglia (DRG) from L3, L4, and L5 were dissected, isolated, and fixed in 10% formalin pH 7.0. After that, tissues were embedded en bloc in order to assess the segmental nature of any GBS lesions [33]. Slides were prepared by the Michigan State University Investigative Histopathology Laboratory. Briefly, 3–5 μm sections were placed on charged slides, dried at 56 °C for approximately 12 h, and subsequently deparaffinized in xylene and hydrated through descending grades of ethyl alcohol to distilled water. Slides were placed in Tris-buffered saline (TBS) pH 7.4 (Scytek Labs—Logan, UT) for 5 min for pH adjustment. Following TBS, epitope retrieval was performed using Citrate Plus Retrieval Solution pH 6.0 (Scytek) in a vegetable steamer for 30 min followed by a 10-min countertop incubation and several changes of distilled water. Following pretreatment standard, avidin-biotin complex staining steps were performed at room temperature on the DAKO Autostainer. All staining steps are followed by 2-min rinses in Tris-buffered saline and Tween 20 (Scytek). After blocking with Normal Rabbit Serum (Vector Labs—Burlingame, CA) for 30 min, sections were incubated with avidin-biotin blocking system for 15 min each (Avidin D—Vector Labs/d-Biotin—Sigma). Primary antibody slides were incubated for 60 min with the Monoclonal Rat anti-Mouse F4/80 diluted at 1:100 (AbD Serotec—Raleigh, NC) in normal antibody diluent (NAD) (Scytek). Reaction development utilized Vector Nova Red Kit peroxidase chromogen incubation of 15 min followed by counterstaining in Gill’s Hematoxylin (Cancer Diagnostics—Durham, NC) for 30 s, differentiated with 1% acetic acid, dehydrated, and mounted with Permount (Sigma). F4/80 stained cells were counted and normalized for tissue area using ImageJ version 2.0.0-rc43/1.50e [51].
Statistical analysis
Statistical analyses were performed using GraphPad Prism 6.0 h for Mac OS X (GraphPad Software, La Jolla, California USA) with the exception of 16S rRNA gene amplicon analysis. Data were entered and then checked for normality and equal variance. If they passed both tests, one-way ANOVA was performed. If they failed either test, a Kruskal-Wallis test was performed instead, followed by Dunn’s post-test, with P < 0.05 constituting significance. Statistical analysis of histopathological scoring of ICCJ was performed using a Kruskal-Wallis test followed by Dunn’s post-test. Statistically significant comparisons in histopathology were further analyzed using Fisher’s exact test (http://vassarstats.net/fisher2x3.html) and corrected for multiple comparisons with the Holm-Sidak step-down procedure [30]. Two-way repeated measures ANOVA and Tukey’s post-test were used for analysis of open-field and rearing behavior in Experiment 1. Analysis of 16S rRNA gene amplicon data was performed using PAST 3 [50]; statistical procedures are indicated in figure legends.
For comparison of anti-ganglioside antibody levels between experimental groups, all datasets had unequal variances by one-way ANOVA so Kruskal-Wallis test was used in PAST. If the full table had a significant P in the Kruskal-Wallis test, pairwise tests between groups were conducted using the Mann-Whitney test; P values were adjusted for multiple comparisons using the Bonferroni procedure.