Close this search box.

Engineering tumor-colonizing E. coli Nissle 1917 for detection and treatment of colorectal neoplasia – Nature Communications

Strains and plasmids

All bacterial strains used were luminescent (integrated luxCDABE cassette) so they could be visualized with the In Vivo Imaging System (IVIS). The EcNΔclbA strain was engineered using the lambda-red recombineering method64. The salicylate-encoding plasmid was constructed using Gibson assembly methods or restriction enzyme-mediated cloning methods whereby isochorismate synthase genes (irp9, mbtI, menF, entC, and pchA) and pchB genes were cloned onto medium or high-copy origin plasmids and driven by the tac promoter. Pathway-engineered EcN was constructed by integrating aroGfbr, tktA, and talB genes using pSPIN plasmid. The SLIC and SLIC-3 strains were constructed as previously described33.

Bacterial preparation for oral administration

Overnight cultures of EcN-lux were diluted 1:100 into LB with 50 ng/ml erythromycin and cultured to an OD600 of 0.1–0.5 on a shaker at 37 °C. Bacteria were collected by centrifugation at 3000–5000 × g, washed three times with sterile PBS, resuspended in sterile ice-cold PBS with a total of 100–200 μL dosed orally at a concentration of ~1010–1011 CFU/ml. Salicylate-producing strains were cultured similarly with an additional 50 μg/ml kanamycin added to retain the salicylate-encoding plasmid. SLIC strains were prepared as previously described33. Briefly, growth media for SLIC and SLIC-3 strains also contained 0.2% glucose to suppress premature lysis in culture. In addition, SLIC-3 strains were grown with 50 μg/ml kanamycin.

Organoid culture

Mouse CRC BrafV600E;Tgfbr2Δ/Δ;Rnf43Δ/Δ/Znrf3Δ/Δ;p16 Ink4aΔ/Δ (BrafV600EΔTRZI, MSS CRC) and ApcΔ/Δ, KrasG12D/Δ, Trp53Δ/Δ, Mlh1Δ/Δ (AKPM, MSI CRC) organoids were generated using CRISPR/Cas9 genome engineering and expanded for injection into mice in matrigel culture as described41. Culture medium was Advanced Dulbecco’s modified Eagle medium/F12 (Life Technologies) supplemented with 1x gentamicin/antimycotic/antibiotic (Life Technologies), 10 mM HEPES, 2 mM GlutaMAX, 1xB27 (Life Technologies), 1xN2 (Life Technologies), 10 ng/ml human recombinant TGF-β1 (Peprotech). Further media supplementation with 50 ng/ml mouse recombinant EGF (Peprotech) for MSS organoids, or 100 ng/ml mouse recombinant noggin (Peprotech), 50 μM Nutlin (Sigma) and 1 mM EGFR inhibitor (Sigma-Aldrich) for MSI organoids. Immediately after each split, organoids were cultured in 10 μM Y-27632 (In Vitro Technologies), 3 μM iPSC (Calbiochem Cat #420220), 3 μM GSK-3 inhibitor (XVI, Calbiochem, # 361559) for the first 3 days.

Orthotopic mouse models of CRC

All animal experimentation involving the orthotopic CRC implant models was approved by the institutional animal ethics committee of the South Australian Health and Medical Research Institute (SAHMRI) (SAM-319, SAM-20-031, SAM-21-041). Orthotopic injections to generate distal colon tumors were undertaken as previously described41. In brief, NOD.Cg-PrkdcscidIl2rgtm1Wjl/SzJ (NSG) mice (male and female, 6–12 weeks old, recipients for MSS organoids) or C57BL/6 mice (male, 6–12 weeks old, recipients for MSI organoids) were obtained from the SAHMRI Bioresources facility and housed under SPF conditions. Digested MSS or MSI organoid clusters (equivalent to ~150 organoids) were resuspended in 20 μL 10% GFR matrigel 1:1000 India ink, 10 μM Y-27632 in PBS and injected into the mucosa of the distal colon of anaesthetized mice using colonoscopy-guided orthotopic injection (2 injection sites/mouse). Injection sites were monitored by weekly colonoscopy. EcN administration began once the tumors were clearly established at grade 3 to 4 on the Becker scale65, 4–6 weeks post organoid injection (Fig S4). Mice bearing MSS tumors were treated with broad-spectrum antibiotics to generate gut dysbiosis through administration of 0.5 g/L neomycin and 1 g/L ampicillin ad libitum in drinking water for 5 days. This was halted 6 h prior to EcN administration. All mice were regularly monitored for signs of clinical deterioration (such as body condition, absence of stool production, weight loss) and euthanized if clinical score reached 3, or timed endpoint 5 days after last EcN administration.

Min/+ mouse model of CRC

All animal experimentation related to the ApcMin/+ mouse model of CRC was approved by the Institutional Animal Care and Use Committee (Columbia University, protocols AC-AAAN8002 and AC-AAAZ4470). All mice were regularly monitored and euthanized based on veterinarian recommendation or when they reached ~20 weeks of age. In all therapeutic studies wild-type littermates of ApcMin/+ on the C57BL/6 background were used and both males and females were treated and evenly distributed among groups. For diagnostic studies, ApcMin/+ mice were purchased from Jackson Laboratories and purchased C57BL/6 mice were used as wild-type controls.

Bioluminescence imaging

To quantify the EcN-lux derived luciferase signal in our mouse models of CRC we used a Xenogen in vivo imaging system (IVIS) Spectrum Imager (Perkin Elmer Inc). Following necropsy, individual mouse tissues were collected into individual wells of a 6-well plate, weighed and background (stage alone) subtracted average radiance (photons/s/cm2/sr) measurements were used to correct for the area being measured which differed for each tissue analyzed.

Colony-forming unit assays

Excised mouse tissues were placed aseptically into 5 ml 20% glycerol in PBS and homogenized in MACS Gentle cell dissociator C tubes, one tissue per tube using program C. 100 μl of each tissue homogenate glycerol stock was serially diluted 1:100 six times. 10 μl of each dilution was spotted onto an LB agar plate with 50 μg/ml erythromycin selection with 5 technical replicates. Plates were incubated at 37 °C overnight (16 h). Colony-forming units (CFU) were calculated for each sample normalized to the weight of tissue input to generate CFU/g tissue. To generate CFU/g stool, one pellet of stool was placed into a 1.5 ml microcentrifuge tube and manually homogenized in PBS with a pipette tip and rigorous pipetting. Serial dilutions were spotted onto an LB agar place with 50 μg/ml erythromycin and incubated at 37 °C overnight. CFU was normalized to weight of the stool.

Clinical trial design

This study was an interventional, double-blind, dual-centre, prospective clinical trial (WHO Uni-versal Trial Number U1111-1225-7729, ANZCTR number ACTRN12619000210178 registered 13 Feb 2019). See supplementary information for study protocol. The study was approved by the Human Research Ethics Committee of the Central Adelaide Local Health Network (HREC/18/CALHN/751) to meet the requirements of the National Statement on Ethical Conduct in Human Research in accordance with the Declaration of Helsinki for medical research involving human subjects. The study objective was to evaluate the colonization of matched normal and neoplastic bowel tissue by the probiotic E. coli Nissle (EcN). Adult participants undergoing routine colonoscopy or surgical resection for primary colorectal cancer were recruited from St. Andrew’s Hospital and Royal Adelaide Hospital, Adelaide (N = 35). Written, informed consent was provided before participants were assigned to take either 2 tablets (109CFU) per day of non-genetically modified EcN (Mutaflor) or placebo for 14 days, prior to their procedure. Patients and treating physicians were blind to active or placebo status. Participants were provided with probiotic tablets and instructed to begin 14 days prior to resection. On the morning of resection investigators verbally confirmed with the participants that they had taken the entire probiotic course, stopping on the day before surgery. There were no alternate methods employed to validate treatment uptake, other than PCR detection of EcN DNA sequence in microbial cultures from tissue samples. Mucosal biopsies (colonoscopy) or surgical resection samples from normal and neoplastic tissue were collected from each participant at the time of their procedure. Participants were excluded if they took probiotics or antibiotics during the trial period. First participant recruited 7 March 2019, last participant 24 September 2019. Eight participants withdrew from the study primarily due to treatment plan change, i.e., no longer undergoing surgery and 2 participants did not have sufficient tumor tissue present to sample from surgery and were excluded. The trial was terminated early after accrual of 35 of the planned 110 participants due to COVID-19-related restrictions and concerns colibactin-expressing E. coli may be pro-carcinogenic. The planned primary outcome of neoplastic colonization status of the probiotic in CRC patient tissues is reported herein, the secondary outcome of microbiome analyses associated with colonization status was discontinued due to smaller than expected sample size.

Human tissue sample analysis

Initially, participants (n = 15) tissue samples were snap-frozen for subsequent DNA extraction. Interim analysis of these samples indicated that detection of EcN was hampered by the presence of host nucleic acid that far outnumbered EcN-derived nucleic acid sequences. These 15 samples were not included in further analyses. To enrich for microbial content in the samples we altered our sample collection methodology as follows. Tissue samples were weighed and collected in sterile 20% glycerol in PBS (n = 10 participants). Tissue was immediately homogenized in gentle MACS C Tubes (Miltenyi Biotec, 130-093-237), with a gentle MACS Dissociator (Miltenyi Biotec, 130-093-235), program E. Aliquoted, homogenized tissue was stored at −80 °C until further use. For culture enrichment, the equivalent of 10 mg of human tissue in homogenate was added to 1.2 ml of LB broth/sample and incubated with shaking at 37 °C for 24 h. Culture OD was monitored hourly for the first 14 h to ensure exponential growth, samples from 2 patients were excluded due to inability to attain log phase cultures from tissue homogenates. 1 ml of saturated culture at 24 h was centrifuged at 10,000 × g to collect cells and DNA extracted from cell pellet using DNeasy PowerSoil Pro kit (Qiagen, 47016) for samples from the remaining 8 patients.

Development of EcN strain-specific PCR assay for human samples

We first tested EcN pMUT2 primers ECN7/8 and 9/10 used previously to detect EcN in mouse fecal samples47, but found that they generated unacceptable false positives using gDNA isolated from human tissue samples from untreated patients. Alignment of PCR primer sets ECN7/8 or 9/10 against DNA sequences using Primer-BLAST suggested that Edwardsiella and Plesiomonas contain highly related sequences potentially also found in the human gut, that may cause false positive calls via PCR assay using these primers. To avoid this confounding amplification during EcN detection, we designed a nested PCR strategy to boost specificity and sensitivity for use with human samples using DNA sequence from pMUT2 unique to EcN in comparison with human gut microbiota sequences66 (Fig. S5). The external 283 bp amplicon spans the unique pMUT2 DNA region: ext-F 5’ CGCGAACGTTAAA-TAATCATC; ext-R 5’ TCTGTTTTAGATAAGGCCATGTCTTC, and was amplified from 50 ng DNA input using KAPA Probe qPCR Master Mix (Roche, KK4716) with PCR conditions: denaturation 95 °C for 20 s; 10 cycle s of 95 °C for 1 s, 60 °C for 20 s, and 72 °C for 25 s. Then 1 μl of this reaction was used as the template for the second 114 bp nested primer/probe-based assay. Nested primer and probe sequences were: int-F 5’ ACCCATCGATAC-CAAATGTATGT; int-R 5’ TCAATGCGTACTCGACTATTCAAA; probe 5’ /56-FAM/CCCG-CAGAT/ZEN/CACTGACCTCAATACA/3lABkFQ/ using KAPA Probe qPCR Master Mix with PCR conditions as follows: 95 °C for 20 s, 40 cycles of 95 °C for 1 s, 60 °C for 20 s, and 72 °C for 25 s. For 16S PCR, standard KAPA SYBR (non-nested) qPCR Master Mix (Roche, KK4602) with primers reported to amplify a 466 bp amplicon covering 331-797 of the E. coli 16S rRNA gene 16S-F 5’ TCCTACGGGAGGCAGCAGT and 16S-R 5’ GGACTACCAGGG-TATCTAATCCTGTT38. EcN PCR standards were generated from serially diluted DNA isolated from exponentially growing cultures from crushed Mutaflor capsule in LB at 37 °C, with CFU determined by plating of matched samples on LB agar plates. Limit of detection of the assay was based on standard curve dilution series, that is the most dilute standard for which a specific PCR amplicon was reliably generated was determined to be the limit of detection of the assay. This is indicated by a red dashed line in figure, dot points above the line have detectable PCR amplicon signal, those below are beyond the limit of reliable detection of the assay.

In vitro sample preparation for salicylate metabolite detection

Overnight cultures of  EcN-salicylate strain variants were OD600 matched to be 1. Cultures were then centrifuged at 3000 × g and 1 mL of the supernatant was collected and stored at −80 °C, the rest was decanted, and the pellet was stored at −80 °C as well until analysis. To 1 ml of supernatants, 1000 µL of extraction solvent ¬(MeOH/MeCN/H2O (2:2:1; v/v/v) containing 0.1 mg/mL of D4-salicylate) was added. Similarly, to cell pellets, 800 μL of the same extraction solvent with the internal D4-salicylate standard was added. Samples were homogenized using a Bead Ruptor 4 at speed 4 for 10 s for 5 cycles. The homogenized samples were centrifuged for 5 min at 14,000 × g. Then, the supernatant was removed and dried on the Gene Vac for 5 h and resuspended in 200 µL of MeCN/H2O before LC-MS analysis. A quality control (QC) sample was prepared by combining 20 μL of each sample to assess the reproducibility of the features through the runs.

Ultra-high performance liquid chromatography (UPLC) analysis

Chromatographic separation was carried out at 40 °C on Acquity UPLC BEH C18 column (2.1 × 50 mm, 1.8 μm) over a 7-min gradient elution. Mobile phase A consisted of water and mobile phase B was acetonitrile both containing 0.1% formic acid. After injection, the gradient was held at 99% mobile phase A for 0.5 min. For the next 4 min, the gradient was ramped in a linear fashion to 50% B and held at this composition for 1 min. The eluent composition returned to the initial condition in 0.1 min, and the column was re-equilibrated for an additional 1 min before the next injection was conducted. The flow rate was set to 450 μL/min and Injection volumes were 2 μL using the flow-through needle mode in the negative ionization mode. The QC sample was injected between the samples and at the end of the run to monitor the performance and the stability of the MS platform.

Mass spectrometry (MS) analysis

The Synapt G2-Si mass spectrometer (Waters, Manchester, UK) was operated in the negative electrospray ionization (ESI) modes. A capillary voltage of −1.5 kV and a cone voltage of 30 V was used. The source temperature was 120 °C, and desolvation gas flow was set to 850 L/h. Leucine enkephalin was introduced to the lock mass at a concentration of 2 ng/μL (50% ACN containing 0.1% formic acid), and a flow rate of 10 μL/min for mass accuracy and reproducibility. The data was collected in duplicates in the centroid data-independent (MSE) mode over the mass range m/z 50 to 650 Da with an acquisition time of 0.1 s per scan. The QC sample, D4-salicylate, and salicylate standards were also acquired in enhanced data-independent ion mobility (IMS-MSE) in negative modes for the structural assignment. The ESI source settings were the same as described above. The traveling wave velocity was set to 650 m/s and the wave height was 40 V. The helium gas flow in the helium cell region of the ion-mobility spectrometry (IMS) cell was set to 180 mL/min to reduce the internal energy of the ions and minimize fragmentation. Nitrogen as the drift gas was held at a flow rate of 90 mL/min in the IMS cell. The low collision energy was set to 4 eV, and the high collision energy was ramped from 25 to 50 eV in the transfer region of the T-Wave device to induce fragmentation of mobility-separated precursor ions.

Data pre-processing and statistical analysis for mass spectrometry

All raw data files were converted to netCDF format using DataBridge tool implemented in MassLynx software (Waters, version 4.1). Then, they were subjected to peak-picking, retention time alignment, and grouping using XCMS package (version 3.2.0) in R (version 3.5.1) environment. Technical variations such as noise were assessed and removed from extracted features’ list based on the ratios of average relative signal intensities of the blanks to QC samples (blank/QC > 1.5). Also, peaks with variations larger than 30% in QCs were eliminated. The detected signal intensity of salicylate in the samples were normalized to the signal intensity of labeled D4-salicylate. Group differences in measured salicylate levels were calculated using the Welch t-test, p-value < 0.05 in GraphPad prism. For Supplementary Fig. 9 targeted analysis was performed as previously described67. In brief, samples were separated by liquid chromatography on an Agilent 1290 Infinity LC system by injection of 3 μl of extract through an Agilent InfinityLab Poroshell 120 HILIC-Z, 2.1 × 150 mm, 2.7 μm (Agilent Technologies) column heated to 50 °C. Solvent A (100% water containing 10 mM ammonium acetate, 5 mM InfinityLab Deactivator Additive and adjusted to pH 9 using ammonium hydroxide) and Solvent B (85% acetonitrile/15% water containing 10 mM ammonium acetate, 5 mM InfinityLab Deactivator Additive and adjusted to pH 9 using ammonium hydroxide) were infused at a flow rate of 0.250 ml min−1. The 26-min normal phase gradient was as follows: 0–2 min, 96% B; 5.5–8.5 min, 88% B; 9–14 min, 86% B; 17 min, 82% B; 23–24 min, 65% B; 24.5–26 min, 96% B; followed by a 10-min post-run at 96% B. Acquisition was performed on an Agilent 6230 TOF mass spectrometer (Agilent Technologies) using an Agilent Jet Stream electrospray ionization source (Agilent Technologies) operated at 3500 V Cap and 0 V nozzle voltage in extended dynamic range, negative mode. The following settings were used for acquisition: The sample nebulizer set to 35 psi with sheath gas flow of 12 L min–1 at 350 °C. Drying gas was kept at 350 °C at 13 L min−1. Fragmentor was set to 90 V, with the skimmer set to 45 V and Octopole Vpp at 750 V. Samples were acquired in centroid mode at 1 spectra/s for m/z values from 50 to 1700.

Urine sample preparation for salicylate metabolite detection

Urine samples were collected from mice 24 h after EcN-salicylate strain variant oral dosing and frozen at −80 °C for later LC-MS analysis. For LC-MS analysis, urine samples were thawed on ice and polar metabolites were extracted with addition of 100% ice-cold LC-MS grade methanol with 0.1 M formic acid (4:1 ratio of extraction solvent volume to urine). Samples were then vortexed and centrifuged at 20,000 × g for 10 minutes at 4 °C. Supernatants were then transferred to clean LC-MS tubes and loaded onto the LC-MS autosampler, which was temperature-controlled at 4 °C.

Data acquisition and analysis for mass spectrometry

Raw data was acquired from the instrument and analyzed using previously described open-source XCMS software68. Metabolites were identified from (m/z, rt) pairs by retention time comparison with authentic standards.


For the ApcMin/+ mouse model, all intestinal tissue was excised with the cecum removed and tissue was bisected such that there were 5 total sections: duodenum, proximal jejunum, distal jejunum, ileum, and colon. Intestines were flushed with PBS, splayed open, swiss-rolled, and fixed overnight in 4% paraformaldahyde. After 24 h, the Swiss rolls were switched to 70% ethanol and sent for histology services at Histowiz, where they were paraffin-embedded, sectioned, and stained with either H&E or specific immunohistochemistry markers (HA-Tag C29F4 #3724 from Cell Signaling Technology; GranzymeB Leica Biosystems PA0291; CD3 Abcam 16669; CD8 catalog #CST98941 clone D4W2Z). Tumor sizes and IHC quantification were determined using FIJI software image analysis tools.

For the orthotopic CRC transplant model, mice were intraperitoneally injected with Pimonidazole-HCl (Hypoxyprobe, 60 mg/kg) or PBS negative control for Hypoxyprobe staining, 1 h prior to euthanasia. Tumor, nearby adjacent colonic tissue and kidneys were collected at necropsy, rinsed in PBS and fixed overnight in formalin prior to dehydration in 70% ethanol and paraffin-embedding. An additional positive control tumor sample for optimizing ISH staining was generated by intratumoral injection of EcN-lux into a dissected tumor sample from a PBS-treated mouse ex vivo, prior to fixation. Formalin-fixed and paraffin-embedded tissue sections were stained with H&E. Serial sections were also subjected to co-immunofluorescence staining against Hypoxyprobe (cat# HP12-200 Kit, 1:200 dilution) and lipopolysaccharide (LPS, cat# HM6011-100UG, 1:500 dilution) or chromogenic in situ hybridization using RNAscope technology (RNAscope 2.5 Detection Kit, Advanced Cell Diagnostics) following the manufacturer’s instructions with a custom probe to detect the lux transcript in EcN-lux or negative control probe, DapB (target region 414-862; catalog number 310043). Briefly for ISH, tissue sections were baked in a dry oven (HybEZ II Hybridization System, ACD) at 60 °C for 1 h and deparaffinized, followed by incubation with Hydrogen Peroxide (Lot# 322000, ACD) and targeted retrieval (Lot# 322330, ACD). Slides were incubated with relevant probes for 2 h at 40 °C, followed by successive incubations with Amp1 to 6 reagents. Staining was visualized with DAB. For IF studies, sections were treated with blocking buffer (X0909, Dako) for 30 min, incubated with the indicated primary antibodies overnight at 4 °C, and washed with PBS. Sections were then incubated with Alexa Fluor 488/594-conjugated secondary antibodies (1:200 dilution, Thermo Fisher Scientific) for 1 h at room temperature. The sections were then mounted with Vectashield antifade mounting medium (Cat# H-1000-10, Vector Laboratories), and fluorescence was examined using a confocal laser-scanning microscope (FV3000, Olympus).

Resected human CRC samples were fixed overnight in formalin prior to dehydration in 70% ethanol and paraffin-embedding. Formalin-fixed and paraffin-embedded tissue sections were stained with H&E. Serial sections were also subjected to immunofluorescence staining against lipopolysaccharide (LPS, cat# HM6011-100UG, 1:500 dilution). Sections were treated with blocking buffer (X0909, Dako) for 30 min, incubated with the primary antibody overnight at 4 °C, and washed with PBS. Sections were then incubated with Alexa Fluor 488-conjugated secondary antibody (1:200 dilution, Thermo Fisher Scientific) for 1 h at room temperature. The sections were mounted with Vectashield antifade mounting medium (Cat# H-1000-10, Vector Laboratories), and fluorescence was examined using a confocal laser-scanning microscope (FV3000, Olympus).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.