All animal experiments were approved by the local animal care and use committee and by the French Ministry of Research (registration number: APAFIS 3735-2016012017118094). Experiments were conducted on 17 male Wistar rats (Janvier Labs, Le Genest Saint Isle, France) weighing between 55 and 85 g and aged 3 weeks at the time of their arrival in our animal facility. The RF-EMF-exposed and control (non-exposed) groups comprised 9 and 8 animals, respectively. Each group was housed in a separate anechoic chamber with a 12 hr:12 hr dark/light cycle and controlled thermoneutral air temperature (24 ± 1°C), relative air humidity (mean ± standard deviation (SD): 39±12%) and air velocity (<0.2 m/s). Rats were housed individually in plastic cages (425 mm x 266 mm x 185 mm) within the anechoic chamber. Standard chow (3436EXF12, Serlab, Montataire, France) and drinking water were available ad libitum.
The RF-EMF exposure protocol was the same as that used by Bosquillon de Jenlis et al. 36. After 4 days of acclimatization, the RF-EMF exposure (23 hours per day, over a 5-week period) was initiated. The climatic chambers were equipped with RF-EMF antennae powered by a generator (model RFS 900–64, RFPA, Artigues-près-Bordeaux, France) emitting a continuous-wave 900 MHz EMF. Antennae (model 800–10465, KATHREIN-Werke KG, Rosenheim, Germany) were placed horizontally in the climatic chamber, 80 cm above the exposed rats’ boxes. The generator’s power was set to obtain a field intensity of 1.8 V/m. The animals’ estimated specific absorption rate was 30 mW/kg. The dose of RF-EMF exposure was checked with a radiofrequency dosimeter (PMM EP600, Narda Safety Test Solutions, Hauppauge, NY) and monitor on computer software (Win EP 600, Narda Safety Test Solutions). To allow animal care, the generator was turned off for one hour a day.
Exposure, sacrifice, and sample collection
After 5 weeks of RF-EMF exposure, the rats were sacrificed by heart puncture under isofluorane 2.5% anesthesia (Iso-Vet 1000 mg/g, Piramal Healthcare UK Ltd, Morpeth, United Kingdom; airflow: 1 L/min). The blood from the heart puncture was collected. Parts of the spleen and thymus were collected postmortem and placed in phosphate buffer saline (PBS, Corning, New York, NY, USA). Splenic or thymic cells were dissociated with the gentleMACSTM Octo Dissociatior (Miltenyi Biotec, Bergisch Gladbach, Germany), using a spleen program. The samples were passed through a 100 µm sieve and then frozen at -150°C in a solution composed of 9:1 v/v fetal bovine serum (FBS): dimethylsulfoxide (DMSO).
All sample preparation steps (collection, centrifugation, dilution, and inoculation into culture media) were performed on the same day for anaerobe and aerobe assays. Anaerobic conditions were applied first; to protect the anaerobic bacteria, all steps were performed as quickly as possible.
After weighing, thawed sample of intestine and other organs were mashed in an equal weight of cysteine broth and homogenized in a sterile blender in order to maintain adherent bacteria. Next, 1 mL samples of organ homogenate (intestine and organs) were diluted and plated on various selective and non-selective media for qualitative and quantitative cultures of aerobic and anaerobic microbiota (100 µL of diluted medium per plate). Standard microbiological techniques and media were used, as described elsewhere 52 53. All aerobic culture plates were incubated at 37°C for 48 h. All anaerobic culture plates were incubated in an anaerobic chamber (Bactron Anaerobic, Sheldon Manufacturing, Cornelius, OR, USA) and examined after 4 days of incubation at 37°C. The culture plates were evaluated by a researcher who was blinded to the samples’ group assignment. Isolated colonies grown on Petri dishes were counted using an automatic colony counter (Scan® 500, INTERSCIENCE, Saint-Nom-la-Bretèche, France) and expressed as log10 colony-forming units per gram of tissue (log(CFU)/g).
After macroscopic and microscopic observations (Gram staining procedures), biochemical assays were performed. A number of specific chromogenic media were also used. Again, the results were read by a researcher who was blinded to the samples’ group assignment.
Bacterial translocation to the spleen
The spleen was removed for the culture of translocated organisms. Bacteria were counted and identified according to their characteristic colonies and microscopic features and by using matrix-assisted laser desorption ionization time-of-flight mass spectroscopy (as described by Joly et al. 54) under aseptic conditions. Thawed tissue samples were placed in sterile, preweighed glass containers. On the day of inoculation, organs were weighed, placed in sterile lab blender bags, and then homogenized in Ringer’s saline solution (Bio-Rad, Marnes-la-Coquette, France). Lastly, plates were inoculated with the resulting homogenate. The conventional media have been described elsewhere 53. After incubation, bacteria were identified and counted using standard microbiological techniques by a researcher who was blinded to the samples’ group assignment 52. The lactobacilli and bifidobacterial were considered to be beneficial microbes, whereas enterococci, enterobacteria, clostridia, Bacteroides and staphylococci were considered to be pathogenic.
Data acquisition and analysis
On the day after sacrifice, a complete blood count was determined on a KT-6400 analyzer (Genrui Biotech Inc., Shenzhen, China). Next, peripheral blood mononuclear cells (PBMCs) from non-exposed or exposed animals were separated using gradient separation in Ficoll Pacques Plus© (GE-Healthcare, Little Chalfont, United Kingdom). The top Ficoll phase (i.e. the plasma sample) was frozen at -150°C. The PBMC band was collected, washed in PBS, pelleted at 300 x g, and then frozen at – 150°C in a solution composed of 9:1 v/v FBS:DMSO.
Flow cytometry analyses
Single-cell suspensions prepared from the spleen, thymus or PBMCs were thawed in RPMI (Roswell Park Memorial Institute) medium (PAN-Biotech, Aidenbach, Germany) supplemented with 10% FBS (Eurobio, Courtaboeuf, France), 2 mM L-glutamine (Sigma Aldrich, Saint-Louis, Missouri, USA) and 10 U/mL penicillin-streptomycin (Sigma Aldrich) for three hours. The cells were then stained with fluorochrome-conjugated antibodies (Abs). The following Abs were used for flow cytometry experiments: PE-anti-rat-CD3 (clone REA223), APC-anti-rat-CD4 (clone REA482), FITC-anti-CD8a (clone REA437), FITC-anti-rat-immunoglobulin (Ig)M (clone ES26-13D3.4), APC-anti-rat-CD45R (clone REA450), APCVio770-anti-rat-CD45 (clone REA504), FITC-anti-rat-CD161 (clone REA227), APC, anti-rat-IgG1 (clone RG11/39.2) (Miltenyi Biotec). Lymphoid cell populations were analyzed after gating (1 x 105 viable cells), according to their forward scatter and side scatter. B lymphocytes (CD45+/CD3-/CD45R+), active B lymphocytes (CD45+/CD3-/IgG1+), immature B lymphocytes (CD45+/CD3-/IgM+), switched B lymphocytes (CD45+/CD3-/IgG1+), memory B lymphocytes (CD45+/CD3-/CD45R+/CD27+), T lymphocytes (CD45+/CD3+), helper T lymphocytes (CD45+/CD3+/CD4+CD8a-) and cytotoxic T lymphocytes (CD45+/CD3+/CD4-CD8a+) subset were detected according to a previously described gating strategy 55–57. All sample were acquired using a MACSQuantify (Miltenyi Biotec). The flow cytometry data were analyzed using FlowJo software (version 10.2).
Immunoglobulin concentrations in plasma samples were determined using an Antibody Isotyping 6-Plex Rat ProcartaPlex© assay (Invitrogen, Carlsbad, CA). Plasma samples were first diluted 4000-fold, as recommended by the manufacturer. Bead immunoassay results were read on a MagPix Milliplex Map system (Merck, Fontenay-sous-Bois, France) and analyzed with ProcartaPlex© Analysis software (Invitrogen).
The normality of the data distribution was checked with the Agostino-Pearson test. In order to compare control and exposed groups and probe correlations between the two, we applied Mann Whitney, Wilcoxon, chi-squared and Pearson’s r tests, as appropriate (see the figure legends). Statistical analyses were performed with GraphPad Prism software (version 8.4.0, GraphPad Software, San Diego, CA, USA). Quantitative variables were expressed as the median (range) in the figures and as the mean ± SEM in the table. The threshold for statistical significance was set to p<0.05.