Bacterial cancer therapy in autochthonous colorectal cancer affects tumor growth and metabolic landscape

Bacterial cancer therapy (BCT) shows great promise for treatment of solid tumors, yet basic mechanisms of bacterial-induced tumor suppression remain undefined. Attenuated strains of Salmonella enterica serovar Typhimurium (STm) have commonly been used in mouse models of BCT in xenograft and orthotopic transplant cancer models. We aimed to better understand the tumor epithelium–targeted mechanisms of BCT by using autochthonous mouse models of intestinal cancer and tumor organoid cultures to assess the effectiveness and consequences of oral treatment with aromatase A–deficient STm (STmΔaroA). STmΔaroA delivered by oral gavage significantly reduced tumor burden and tumor load in both a colitis-associated colorectal cancer (CAC) model and in a spontaneous Apcmin/+ intestinal cancer model. STmΔaroA colonization of tumors caused alterations in transcription of mRNAs associated with tumor stemness, epithelial-mesenchymal transition, and cell cycle. Metabolomic analysis of tumors demonstrated alteration in the metabolic environment of STmΔaroA-treated tumors, suggesting that STmΔaroA imposes metabolic competition on the tumor. Use of tumor organoid cultures in vitro recapitulated effects seen on tumor stemness, mesenchymal markers, and altered metabolome. Furthermore, live STmΔaroA was required, demonstrating active mechanisms including metabolite usage. We have demonstrated that oral BCT is efficacious in autochthonous intestinal cancer models, that BCT imposes metabolic competition, and that BCT has direct effects on the tumor epithelium affecting tumor stem cells.


Tissue isolation and analysis
At the end of the STm ∆aroA treatment, mice were killed and colons (CAC) or SI (Apc min/+ ) were dissected out, cut longitudinally and washed thoroughly in cold PBS. Tumors were counted under a stereo microscope using an eyepiece graticule with 10x10 mm grid. Tumors and normal tissue were then micro-dissected and either snap frozen in liquid N2 for later analysis or used for assessing bacterial colonization (CFU).

Colony forming unit analysis
Tissue was homogenized in PBS containing 0.1% triton-X using a hand-held homogenizer. After serial dilution, 50 uL was plated onto LB agar plates containing ampicillin (200 ug/mL) and incubated at 37°C overnight.
Colonies were counted and CFU calculated per gram of tissue.

Scanning electron microscopy
Colons were dissected out, cut longitudinally and thoroughly washed in PBS. Sections around 1 cm were cut and fixed in 2.5% glutaraldehyde in 0.1% phosphate buffer for 24 hours at 4°C. Tissue was then dehydrated in increasing concentration of ethanol (50% -> 70% -> 80% -> 90% -> 95%-> 100% -> 100%) for 15 minutes each, then tissues substituted with t-butyl alcohol followed by freeze drying. Before mounting samples, tumors were cut on the sagittal plane to reveal the tumor core then mounted onto aluminum stubs and were metal-coated using a magnetron sputter (MSP-1S; Vacuum Device), and examined by scanning electron microscopy (VE-7800; Keyence).
PCR amplification using Ex Taq HS (Takara) was performed for 25 cycles with the condition as previously described (4). Amplified PCR products were purified using AMPure XP (Beckman Coulter). DNA concentration of each sample was measured using Quant-iT PicoGreen ds DNA Assay Kit (Life Technologies).
Mixed samples were prepared by pooling approximately equal amounts of PCR amplicons from each sample.
The DNA library with 20% denatured PhiX spike-in was sequenced by Miseq sequencer using 500 cycles kit (Illumina).
Taxonomic assignments and estimation of relative abundance of sequencing data were performed using the analysis pipeline of the QIIME software package (Qiime version 1.9.1) (6). Chimera checking was performed using UCHIME (7). OTUs were clustered at 97% similarity. We performed taxonomy assignment against Greengenes database version 13.8 using RDP classifier (8,9). Alpha diversity and beta diversity were assessed by calculating Chao1 index, Shannon diversity index, Simpson's diversity index and weighted Unifrac distance for each sample at equal sequencing depths (10,000 reads/ sample). We have visualized the beta diversity using principal coordinates analysis (PCoA) of weighted UniFrac distance between samples (10).

RNA isolation
RNA was isolated using Qiagen RNeasy Mini kits, as per manufacturer instructions. Briefly, tissue stored in -80°C was placed in buffer RLT + b-mercaptoethanol and homogenized by bead beating. Lysates were then processed as recommended, with the optional DNase digest. RNA was quantified using a DeNovix DS-11. cDNA library prep and RNA sequencing RNA integrity was assessed by Agilent Bioanalyzer before proceeding RIN values of 8 or greater were used. 2 ug of RNA was used to prepare a cDNA library using TruSeq RNA Library Prep Kit v2 (Illumina). Sequencing was performed on an Illumina HiSeq 1500 System in a 1 x 50 bp single read mode. Sequenced reads were mapped against the mouse reference genome (mm10) using TopHat (11), and gene expression was quantified by Cufflinks (12). Gene ontology enrichment analysis was performed using DAVID (13,14). The original RNAseq data is uploaded and available online (Gene Expression Omnibus: GSE136029).
cDNA prep and qPCR cDNA was prepared using standard oligoDT and M-MLV reverse transcriptase. Quantitative real-time PCR was performed with the LightCycler® 480 Real-Time PCR System (Roche) and SYBR Premix Ex Taq (Takara).
Gene-specific primers (Eurofins Genomics, Japan) are listed in Table S1.

Immunofluorescence
Freshly dissected colon tissue was swiss-rolled and placed in 4% PFA overnight. Samples were paraffin embedded and 5 uM sections cut. Paraffin sections were rehydrated and washed with PBS. Samples were boiled in citrate buffer for 15 minutes for antigen retrieval. Then incubated with 1% BSA/PBS supplemented with 5% normal serum donkey serum to quench the nonspecific binding of antibodies. Goat anti-E-cadherin (R&D Systems, 1:200) and Ki67 antibody (1:200; CST) was incubated overnight, washed, secondary stain with antigoat IgG or anti-rabbit IgG (LifeTechnologies) and counterstained with DAPI. Samples were imaged using an Axiovert Slidescanner (Zeiss) and images processed using Zen software (Zeiss).

Flow cytometry
Ex vivo: Tumors were excised from Lgr5-GFP CAC-induced mice that had been treated for 24hours with mCherry-STm ∆aroA . Tumors were minced into small pieces and incubated with 5 mL digestion buffer (DMEM with 2.5% FBS, ampicillin, 200U/mL collagenase D and DNAseI (2ml/mL) for 30 mins at 37°C with shaking.
In vitro: Tumor organoids derived from Lgr5-GFP CAC-induced mice were grown and infected with mCherryexpresssing STm ∆aroA as described below. Following 24 hrs infection organoids were recovered from Matrigel using Cell Recovery Solution (BD), spun at 1200 rpm for 5 mins then resuspended in 500 μL TrypLE with ROCK inhibitor (Y27632, 10mM used at 1:500) and mixed by pipetting and incubated in a 37°C water bath for 3 mins with regular agitation. Another 500 μL of basic medium (see Table 2) with ROCK inhibitor (1:500) was added and organoids thoroughly pipetted to dissociate using a 10 μL tip on the end of a 1 mL tip. These were then spun before resuspending in FACS buffer (PBS/2%FBS) and stained in a 96-well plate (as above but excluding EpCAM).

GC-MS analysis of metabolites
Extraction and measurement of metabolites were previously described (15) with some modifications. Tumor and normal intestinal tissues and tumor organoids (approximately 10 mg) were added to 125 μl methanol, 150 μl Milli-Q water containing internal standard (100 μmol/l 2-isopropylmalic acid) and 60 μl CHCl3 and disrupted with zirconia beads using Micro Smash MS-100 (Tomy Seiko). All samples were shaken at 1,200 rpm for 30 min at 37°C. After centrifugation at 16,000 × g for 5 min at room temperature, 250 μl of the supernatant were transferred to a new tube and 200 μl of Milli-Q water added. After being mixed, the solution was centrifuged at 16,000 × g for 5 min at room temperature, and 250 μl of the supernatant were transferred to a new tube. Samples were evaporated for 20 min at 40°C, and then lyophilized using a freeze dryer. Dried extracts were firstly methoxymated with 40 μl of 20 mg/ml methoxyamine hydrochloride (Sigma-Aldrich) dissolved in pyridine.
After adding the derivatization agent, samples were shaken at 1,200 rpm for 90 min at 30°C. Samples were then silylated with 20 μl of N-methyl-N-trimethylsilyl-trifluoroacetamide (GL Science) for 30 min at 37°C with shaking at 1,200 rpm. After derivatization, samples were centrifuged at 16,000 × g for 5 min at room temperature, and the supernatant transferred to glass vial for gas chromatography-tandem mass spectrometry measurement using a GCMS-TQ8030 triple quadrupole mass spectrometer (Shimadzu) with a capillary column (BPX5, SGE Analytical Science). The GC program was previously described (15). Data processing was performed using LabSolutions Insight (Shimadzu).
The quantified metabolome data was statistically analyzed using Orthogonal Partial Least Square-Discriminant Analysis (OPLS-DA) with SIMCAP+ software (Version 12.0.1.0, Umetrics, Umeå, Sweden) using UV scaling method. Potential metabolites were selected based on the Variable Importance in Projection (VIP) score greater than 1.0 and uploaded to MetaboAnalyst 3.0 (http://www.metaboanalyst.ca) for pathway analysis. The functional pathway analysis of potential biomarkers was based on the database source of the Kyoto Encyclopedia of Genes and Genomes (http://www.genome.jp/kegg/).

Tumor organoid establishment and culture
Tumor organoids were established as previously described (16), with some alterations. Buffers and culture medium components are listed in Table S2. Tumors were dissected from the colons of mice that had CACinduced tumors and from the SI and colon of Apc min/+ mice. Tumors were washed in cold PBS then incubated in 10 mL chelation buffer with 2mM EDTA for 60 mins on ice. Tubes were shaken vigorously by hand (removes most normal epithelium) and EDTA buffer was removed and tumor tissue then washed 3x with chelation buffer.
Tissue was then incubated in digestion buffer for 30 minutes at 37°C with shaking. Supernatant was filtered through a 70 µM filter, remaining tissue wash once with basic medium and also filtered. Cells were then pelleted (300 g for 3 minutes) washed once then resuspended in 100-200 uL Matrigel (Corning 356231) (50 uL per well) and cultured in 24-well plate with 500 µL complete medium. Once established for 1 to 2 weeks, cultures were switched to EGF-only medium.
For maintenance, organoids were split every week. In brief, all tubes and tips were pre-coated with FBS, organoids were manually disrupted from the Matrigel using P500 pipette and transferred to a 50 mL Falcon tube. Tubes were centrifuged at 300 g for 5 mins at 4 deg. Excess medium was removed carefully to not removed Matrigel containing the cell pellet. Organoids were resuspended in 4 mL Basic medium, transferred to a 15 mL tube and pipetted thoroughly using a fire-polished glass pipette to break up the organoids. These were centrifuged and resuspended in Matrigel and plated out. Split ratios of around 1:4 to 1:6 depending on density.
STm ∆aroA infection of tumor organoids Tumor organoids were infected at day 5 post-split to ensure good organoid size and integrity. STm ∆aroA was grown overnight and sub-cultured as described above. 5 µL containing 1x10 8 CFU was dropped into the culture medium (or PBS control) and left for 2 hours to allow for bacterial invasion of the Matrigel and organoids. After 2 hours medium was removed and Matrigel washed 2x with PBS and medium replaced and gentamycin added.
These were cultured overnight and organoids collected 24 hours after initial infection for analysis. For qPCR analysis, buffer RLT was added directly to the culture plate (after removing medium and washing with PBS) which completely dissolved the Matrigel, these were then processed for RNA isolation as described above. 1 well was used for each technical replicate. For metabolome and succinate assays, culture medium was removed and the Matrigel washed with PBS. BD Cell Recovery Solution was added and plate kept on ice for 1 hr to dissolve the Matrigel. Organoids were collected into Eppendorf tubes spun and washed twice with cold PBS.
Organoids were then snap frozen and stored at -20 until analysis.
Heat killed STm ∆aroA was prepared by incubating at 95°C for 5 minutes. Effective killing was tested by plating out on LB agar. SN was prepared by growing STm ∆aroA in tumor culture medium until an OD600 of approximately 0.7. SN was filtered with a 10 kDa cut-off columns. 1x10 5 heat-killed STm ∆aroA were used as this is the average CFU count obtained from the live infections after 24 hours, and the amount of SN added was also calculated based on this CFU.

Stem-forming Assay
Organoids were grown and infected as described above. After 24 hours, Matrigel was digested with CellRecovery Solution (BD) for 1 hour at 4°C. Organoids were transferred to a 15 mL tube, spun at 300g for 5 mins, supernatant removed and 1 mL of TrypLE added. Tubes were placed in a water bath at 37°C and shaken every minutes for approximately 4 minutes. Cells were then pipetted with a fine 10uL tip to aid dissociation, washed and resuspended for counting. Equal numbers of organoids (20-50 000, depending on experiment) were then resuspended in Matrigel and plated out. Following passages were done as per maintenance (with thorough pipetting of organoids without TrypLE) and were reseeded at equal split ratio (eg 1:12).

Succinate Assay
Succinate assay (Sigma) was performed on cell lysates as per the manufacturers protocol and 96-well plates were measured on a spectrophotometer (Spectrostar Nano -BMG Labtech) at the respective wavelengths.

MTT Assay
Following dissociation of organoids for the stem-forming assay (above), 4, 000 cells were re-seeded per well into a flat-bottom 96-well plate in 10 μL Matrigel with150 μL tumor medium, with each sample seeded in triplicate (results averaged). At the indicated timepoints, medium was aspirated and replaced with 50 μL tumor medium and 50 μL MTT reagent (AbCAM) and then incubate at 37 °C for 1 -3 hrs. The reaction was stopped when visible purple formazan crystals were formed by addition of MTT solvent (4 mM HCl, 0.1% NP40 in isopropanol). The plate was measured on a plate-reader at OD560 (BioRAD).

Caspase3 activity Assay
A colorimetric caspase-3 activity assay kit (Sigma) was used. Organoids were infected with STm ∆aroA or control as outlined above. After 24 hrs, organoids were recovered from Matrigel using CellRecovery Solution (BD) and

Statistical analysis
All data is presented as mean +/-SD. One-way ANOVA, Students T-test or non-parametric statistical tests were used, as indicated for each figure, and were conducted using GraphPad Prism 8. P<0.05 was considered statistically significant.    Figure S1: CFU in peripheral organs and SI 24 hours after STm ∆aroA treatment Apc min/+ and Apc +/+ mice were given 5x10 9 CFU of STm ∆aroA by oral gavage and bacterial colonization was assessed 4 hours (not all tissues), 24 hours, 1 week and 2 weeks later. An additional timepoint of 72 hours for feces analysis was also conducted (taken from mice for the 1 and 2 week timepoints) . Spleen, mesenteric lymph nodes, Peyer's patches, normal small intestinal (SI) tissue and SI tumors were dissected and Ileal contents and feces collected, homogenized in PBS -0.1% triton-X and plated onto LB agar plates containing ampicillin. Each dot represents one mouse.

Figure S2. Mouse weight change during AOM/DSS CAC induction
Weight change in mice from Figure 1B. Mice were treated with AOM and DSS (see Figure  1A and methods) to induce colitis-associated cancer. Mouse weight was monitored over the induction period and mice allocated to treatment or non-treatment groups after the last dose of DSS, assuring similar DSS-induced weight change between groups.        Top altered metabolites detected as described in Figure 6. Each point represents an independent well of organoids. Figure S14: Long term passaging recovers STm ∆aroA treated organoid growth. SI (A) or Colon (B) tumor organoids (Apc min/+ -derived) were infected as described. After 24 hours they were dissociated into single cells, counted and seeded at equal density, then passaged at equal rations until the end pf 4 weeks (weekly passage). Representative images shown. (C) shows the cell number at the end of passage 1 and passage 4. Each dot represents an independent well. Two-way T-test. Data is representative of 3 experiments.