Cell line and culture conditions
The murine mammary cell line 4T1 (American Type Culture Collection, Rockville, MD, USA) was transfected with red fluorescent protein using a dsRed-expressing lentiviral vector. This cell line was originally isolated from a spontaneously arising mammary tumor in BALB/cfC3H mice . Successful transfection with dsRed was confirmed by fluorescence microscopy (Axiolmager 2, Carl Zeiss MicroImaging, GmbH, Jena, Germany). 4T1 cells were cultured in RPMI-1640 medium (Bio-Whittaker, Verviers, Belgium) supplemented with 10% Foetal Calf Serum (Sigma-Aldrich, Steinheim, Germany), 100 units/ml penicillin, 100 μg/ml streptomycin, 2% L-glutamine (All from Bio-Whittaker, Verviers, Belgium). The cells were maintained at 37°C in 5% CO2 and 95% air, and were seeded and used at ~ 80% confluence in all experiments.
All the animal experiments were performed in accordance with the regulations of the Norwegian State Committee for Animal Research ("Forsøkdyrutvalget", approval number 1279) and approved by the Local Institution Board at the University of Bergen (approval number 2008076BB). The investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health.
In vivo experiments
Female NOD/SCID mice (18-24 g) were used in this study. Generally, the NOD/SCID mice used expressed eGFP in all nucleated cells, the exception being histological and immunohistochemical experiments, where plain NOD/SCID mice were used. The emission of green fluorescence has previously been observed in muscle, pancreas, kidney, heart and other organs of the mice, confirming the fluorescent phenotype . The breeding was performed at the animal facility at the University of Bergen, as described by Niclou et al. . All animal experiments were performed under Isoflurane (Rhone-Puolenc Chemicals, France) and N2O gas-anaesthesia. A 17β-estradiol pellet (0.18 mg/pellet--60 day release, Innovative Research of America, Sarasota, FL, USA) was inserted into the interscapular area of all mice prior to tumor cell injection. The pellets provide a continuous release of estradiol to give serum concentrations of 150-250 pM. 3 × 106 4T1 cells dissolved in 0.15 ml PBS were injected in the mouse mammary fat pad, in the groin area. 4T1 tumors had a 100% take rate, and a 6 days latency period to a palpable tumor estimated to be ~5 mm in diameter.
The tumors were measured externally with a calliper at day 1 (pre HBO exposure), day 4 and 8 (post HBO exposure). The location of the tumor excluded external measurement in more than two dimensions. The tumor volume was therefore calculated assuming a cylindrical shape of the tumor, according to the formula: π/6 ·a2·b, where a is the shortest and b is the longest transversal diameter.
A 27 l Hyperbaric Animal Research Chamber (OXYCOM 250 ARC, HYPCOMOY, Tampere, Finland) with an inner diameter of 25 cm, and an inner length of 55 cm was used. The chamber was supplied with pure O2. After reaching 100% O2 (15 min), the pressure was raised over a period of approximately 5 min to 2.5 bar (equivalent to 15 meters sea water). The 2.5 bar pure oxygen atmosphere was maintained for a period of 90 min. To maintain > 97% O2 atmosphere, the chamber was flushed with pure oxygen for 3-5 min every 10-30 min depending on the number of mice in the chamber. After treatment, the mice were slowly decompressed over a 10 min period.
Three separate groups of mice were studied. The first group of tumor bearing mice was exposed to intermittent HBO treatment (2.5 bar and 100% O2, 3 exposures à 90 min on day 1, 4 and 7). The second group was exposed to daily HBO treatments (2.5 bar and 100% O2, 7 daily exposures à 90 min), whereas the control group was housed under normal atmosphere for the experimental period of 8 days.
In situ and ex vivo imaging
We used a fluorescence dissection microscope (Model C-DSD230, Nikon, Japan) with UV-filter optics for dsRed and eGFP, to observe the tumor in situ. After the mice were sacrificed with saturated KCl during anaesthesia, the tumors were excised. The tumors were processed in three different ways: 1) formalin (4%), and later embedded in paraffin. 2) frozen in liquid nitrogen and stored at -80°C until further use. 3) paraformaldehyde (PFA) prior to freezing, and then embedded in Prolong Gold (Invitrogen, CA, USA) after sectioning. PFA fix was performed to conserve the fluorescent traits, when visualized under the microscope (Leica TCS SP5, Wetzlar, Germany).
Histology and immunohistochemistry
Paraffin embedded tumor sections from all three groups were H&E stained, and examined by an experienced pathologist.
Frozen tumor sections (20 μm) were used for immunostaining of blood vessels using rat anti-mouse CD31 (AbD Serotec, Morphosys UK Ltd, Oxford, UK) as primary antibody and biotinylated rabbit anti-rat as secondary antibody (Vectastatin ABC kit, peroxidase IgG PK 4004, Bioteam AS, Trondheim, Norway). An ABC kit (Vectastatin ABC kit, peroxidase IgG PK 4004, Bioteam AS, Trondheim, Norway) and Diaminobenzidine tetrahydrochloride (3.3 DAB, Sigma Aldrich, Germany) were used as a chromogen to visualize blood vessels. Richardson stain was used as a nuclear counterstain. The cross-sectional density of CD31 positive structures was quantified per mm2 using a counter grid, covering the viable tumor area. Tumor cell proliferation was assessed by staining with an anti-rabbit Ki67 antibody diluted 1:100 (Millipore, Billerica, MA), and biotinylated goat anti-rabbit secondary antibody (DACO Patts, Glostrup, Denmark) on frozen tumor sections. Four Ki67 labelled "hot spots" within selected high power fields of view (HPFs) were quantified using the NIS-Elements™ BR 3.1 software (Nikon Corporation, Tokyo, Japan), under 400× magnification. The immunopositive cells were counted and expressed as a fraction (%) of the total cells. Four fields of vision per section from each animal were included in the quantification. Cell death was examined by the terminal transferase-mediated dUTP nick-end-labeling (TUNEL) method (Boehringer Mannheim, Mannheim, Germany), performed according to the manufacturers recommendation on frozen tumor sections. For quantifying TUNEL labelled cells, threshold levels of pixel intensity were determined and expressed as % positive cells/area fraction at x200 magnification. All sections were examined using a Nikon light microscope (THP Eclipse E600, Nikon Corporation) and the images were captured with a Nikon Digital Camera (DXM 1,200F, Nikon Corporation). The quantification was performed blindly.
Fluorescence-activated cell sorting (FACS)
Freshly isolated 4T1 tumors were dissociated by mincing the tissue with scalpels, followed by incubation with 1 mg/ml collagenase/dispatase (Roche Diagnostics GmbH, Germany) and 0.125% DNase I (Sigma-Aldrich) dissolved in DMEM medium for 60-90 min at 37°C. Incompletely dissociated tissue was digested a second time using the same procedure. The dissociated tumor tissue was then washed with ice-cold FACS buffer (PBS with 2% FBS) and filtered twice through a 70-μm cell strainer. The cell suspension was then centrifuged at a speed of 500 G for 10 min (4°C). The cell pellets were resuspended in FACS buffer for further analysis and sorting. The single cell suspension was filtered through a 40-μm cell strainer in order to remove any clumping cells before sorting. The cells were sorted using a cell sorter (FACS Aria SORP, BD Biosciences, Erembodegem, Belgium) based on the single-cell viability and the fluorescence intensity of eGFP and dsRed, and separation was confirmed by fluorescence microscopy (Nikon ellipse 2000, Nikon, Japan).
RNA isolation and quantification
Sorted cells were lysed using 350 μl Buffer RLT Plus with β-mercaptoethanol (β-ME). The lysate was homogenized by passing it 5 times through a 23 G needle. Further, the protocol for purification of total RNA from animal cells (RNeasy Plus Mini Handbook, Qiagen AB, Sweden) was used according to the manufacturer's recommendations. RNA concentration (ng/μl) and purity (260/280 ratio) were determined using NanoDrop 1,000 Spectrophotometer (Thermo Scientific, Sweden).
Gene expression analysis
Global gene expression analysis was performed in order to identify key molecular mechanisms and differences in gene programs in stromal and tumor cells, as well as changes following HBO treatment. A total of 18 tumor bearing mice were used. Seven mice served as control animals (non-HBO exposure), 5 mice were HBO treated intermittently and 6 mice treated daily. Stromal cells and tumor cells from the different tumors were isolated as described above. Total RNA purification, cRNA labeling, microarray hybridization and features extraction were performed as previously described . The Agilent G4122F Whole Mouse Genome (4 × 44 k) Oligo Microarray Kit with SurePrint Technology (Agilent Technologies, Inc., Palo Alto, CA) was used to analyze samples in the present study. The normalized channel values were log (2) transformed and combined into a gene expression data matrix. Data were formatted in a J-Express-file suitable for additional data mining (http://www.molmine.com/) . Following normalization, the treated tumor cells and stromal cells were analysed against the respective non-treated controls of stromal and tumor cells. We also analysed untreated stromal cells against untreated tumor cells to highlight and characterize the stromal cell signatures. A similar approach was performed to analyse how the treated tumor cells responded to HBO treatment versus treated stromal cells (results not shown).
For the gene expression data, we used analysis of variance (ANOVA), SAM (Significant Analysis of Microarray), GSEA (Gene Sets Enrichment Analysis)  and Gaussian kernels of the J-Express program package for identification of differentially expressed genes. Following SAM we selected genes with False Discovery Rate (FDR) < 5% as basis for GSEA and Hierarchical clustering. Gene sets consisting of more than 18 genes and less than 200 genes were selected for further analysis of cellular processes, pathways and molecular function. Annotated microarray data were uploaded in the BASE database and formatted and exported to ArrayExpress at the European Bioinformatics Institute (http://www.ebi.ac.uk/arrayexpress (Accession number: E-TABM-1103)) in agreement with the MIAME guidelines. Two-tailed unpaired t-test (normalized data) or the non-parametric Mann-Whitney test (non-normalized data) was used for testing the statistical differences between groups. P < 0.05 was considered statistically significant. GraphPad InStat 3 (GraphPad Software, Inc., La Jolla, USA) and SPSS for Windows (IBM Corporation, NY, USA) were used for the statistical analysis.