Colorectal cancer (CRC) is one of the main cancers that threaten life and health, causing a serious social burden. Long term use of NSAIDs can significantly reduce its incidence rate, but it can not be completely prevented and treated. Resection is an early intervention measure that can be combined with chemotherapy and radiation therapy, but currently there are not many effective methods for detecting and treating colorectal cancer. And engineering probiotics provide new opportunities for improving the screening, prevention, and treatment of colorectal cancer.
This week, we will explain a research paper published by Tal Danino et al. from Columbia University in Nature Communications titled “Engineering tuber colizing E. coli Nissle 1917 for detection and treatment of colorectal neoplasia”. This study delivered Escherichia coli Nissle 1917 (EcN) to genetically engineered mouse models susceptible to CRC and ortho models of CRC, and found selective colonization of colorectal adenomas. Afterwards, the author conducted an interventional, double-blind, double-center, prospective clinical trial in which CRC patients received placebo or EcN for two weeks before tumor and adjacent normal colorectal tissue resection. The author detected EcN enrichment in tumor samples from normal tissues of patients treated with probiotics (the main result of the experiment). Next, the author develops early CRC intervention strategies. In order to detect lesions, the author modified EcN to produce a small molecule – salicylate. Compared with the healthy control group, oral administration of this strain resulted in an increase in the content of salicylates in the urine of mice carrying adenomas. To evaluate the therapeutic potential, the authors designed EcN to locally release cytokines GM-CSF and block nanoantibodies against PD-L1 and CTLA-4 at the tumor site, and demonstrated that oral administration of this strain can reduce the burden of adenomas by approximately 50%. These results support the use of EcN as an oral delivery platform for disease detection and treatment of CRC through the production of screening and therapeutic molecules.
Adenoma colonization of Escherichia coli Nissle 1917 (EcN) in a genetically engineered mouse model susceptible to CRC
The author explored the colonization of adenomas by orally administering the integrated luxCDABE box (EcN lux) EcN encoding the genome to ApcMin/+mice with intact immune systems and microbiome. In vivo imaging of mice taking EcN lux showed that the bioluminescent bacteria in the intestines of healthy wild-type (WT) mice were eliminated, but after oral administration, the bioluminescent bacteria remained in the intestines of ApcMin/+mice for up to 7 weeks. The enrichment of EcN lux in tumors was further demonstrated through in vitro imaging of intestinal tissue, where bioluminescence was co localized with visible large adenomas, and more bioluminescence was typically observed in the distal small intestine with the highest burden of adenomas. Subsequently, the homogenized intestinal tissue was spread on an antibiotic selective Luria broth (LB) plate specific to EcN lux, indicating that detectable EcN lux could not be recovered from wild-type mouse tissue. This indicates that unless there is tumor tissue present, long-term ecological niches will not form in the intestine. In order to further investigate bacterial localization, EcN lux was designed to release proteins labeled with human influenza hemagglutinin (HA) under the control of a lysis circuit and orally delivered to ApcMin/+mice. After 4 weeks of oral administration, the mice were euthanized and tested for positive HA signal in their intestinal tissue. Dark positive staining with EcN labeled payloads was identified in adenomas of different sizes, indicating functional delivery of payloads released from the cleavage loop after oral administration of engineered strains.
Due to increasing concerns about the carcinogenicity of bacteria like EcN that produce colistin, the author knocked out the clbA gene (EcN Δ ClbA was used to prevent the production of colistin, and compared with the parent EcN strain, there was no change in bacterial growth kinetics. The author orally administered EcN to ApcMin/+mice Δ ClbA lux was used to explore tumor colonization, and co localization of bioluminescent bacteria with large adenomas was observed, indicating that colonization does not depend on the presence of clbA genes or complete E. coli encoded operons. When using EcN lux or EcN Δ No significant differences in physical condition and weight were observed in ApcMin/+mice treated with clbA lux strain. 48 hours after administration, EcN was collected from the liver, spleen, cecum, colon, and small intestine of each mouse and spread on antibiotic selective LB plates for recovery Δ ClbA lux. Compared with the healthy control group, fewer or even no bacteria were recovered from the liver and spleen of ApcMin/+mice or wild-type control mice, indicating minimal off target localization.
picture
Figure 1 Adenoma colonization of Escherichia coli Nissle 1917 (EcN) in CRC genetically engineered mouse model
Related Services
E.coli Metabolic Engineering Service
E. coli CRISPR-Cas9 genome editing
Tumor colonization of EcN in CRC in situ mouse model
Next, the author tested the selective colonization of isolated lesions by evaluating the EcN lux of two CRC in situ models (representing MSS and MSI disease subtypes), in which mouse CRC like organs were injected into the distal colon of recipient mice and tumor grade was tracked through weekly colonoscopy. Firstly, MSS CRC mice carrying tumors were pretreated with broad-spectrum antibiotics to disrupt the normal microbial community composition, which is a common phenomenon in gastrointestinal diseases including CRC. Five days after oral administration of EcN lux, in vivo imaging showed co localization of bioluminescent EcN lux with colon tumors. Subsequently, the excised organs were homogenized and seeded on antibiotic selective LB plates, confirming significant enrichment of EcN lux in the tumor compared to adjacent healthy tissues and peripheral organs. The median diameter of EcN lux implanted tumors is 2 millimeters (+/-1.2 millimeters), indicating that the size of tumor lesions detected using this EcN lux platform is similar to that of human polyps (0-5 millimeters). The specific localization of EcN Lux in these tumors was analyzed through RNA in situ hybridization, indicating that bacteria mainly exist in nests on the surface of the tumor cavity and can coexist with hypoxic tumor areas, which may further promote bacterial growth in the tumor space. Similarly, the oral dose of EcN lux in the MSI CRC model, without antibiotic pretreatment, resulted in significant selective enrichment and colonization of tumors compared to adjacent tissues and organs measured through in vitro luminescence imaging and CFU electroplating.
Figure 2 Tumor colonization of Escherichia coli Nissle 1917 (EcN) in an in situ mouse model and human CRC patients
Tumor Colonization of EcN in Human CRC
Next, the author will determine whether Gram negative bacteria can be observed in tumor samples of CRC patients using pan Gram negative lipopolysaccharide (LPS) bacterial staining. Consistent with previous reports on the microbiome of tumor associated CRC, the author observed LPS+bacteria associated with human CRC. Subsequently, the author conducted a clinical trial to examine EcN colonization in CRC patients. Before tissue resection, CRC patients take commercially available non GMO forms of EcN, Mutaflor, or placebo orally for two weeks. Cultivate homogenates of matched normal and tumor tissues (n=8 patients) to enrich microbial content, then isolate DNA and perform qPCR detection. Although the sample size was less than expected, EcN specific PCR amplicons indicated significant enrichment of this bacterium in tumor tissue cultures of patients taking Mutaflor, but not in the placebo control group.
Design an EcN platform based on feces and urine for non-invasive adenoma tracking
The phenomenon of EcN tumor selective colonization indicates that it can serve as a platform for monitoring the presence of adenomas. As fecal testing is a common non-invasive screening tool for CRC, the author first determined whether the recoverable EcN lux in feces can be used for non-invasive monitoring of the presence of adenomas. To this end, the author orally administered EcN lux to healthy wild-type (WT) and ApcMin/+mice, collected fecal particles at predetermined time points, homogenized, and finally spread the feces on antibiotic selective LB agar plates. In the first 7 hours after administration, similar EcN lux CFU shedding was observed in the feces of both WT and ApcMin/+mice, corresponding to the time of substance transport through the intestine. However, by 24 hours, EcN lux levels were not detected in some healthy mice. By 48 hours, the author was unable to recover EcN lux from healthy mouse feces, but EcN lux could still be recovered from ApcMin/+mouse feces. Compared with normal control mice, using MSI CRC model or EcN Δ The ApcMin/+model of clbA lux mutant strain dose showed significant retention of similar EcN lux in fecal samples carried by tumors.
Although this fecal test may be a useful method for tracking adenomas, the author aims to investigate a more accessible diagnostic reading. For this reason, the author designed EcN to generate a molecule that can be easily recovered from body fluids. They chose to encode the production of salicylate because of its role in CRC chemoprevention and because it can be detected in urine. In order to optimize the production of salicylates, the author designed a strain variant library to express genes crucial to the shikimate pathway, which is responsible for converting endogenous bacterial cholismite into salicylates. Specifically, genes mbti, irp9, menF, entC, or pchA were cloned onto plasmids, and pchB with low (sc101 *) or high (ColE1) copy number origins were also encoded. Then, these plasmids were cloned into two different EcN strains – EcN Δ ClbA lux (EcN) mutant or EcNATT Δ ClbA lux (EcNATT), which includes genome integrated aroG *, tktA, and talB genes. Due to their involvement in the oxalate and pentose phosphatase pathways, it is assumed that their integration will lead metabolic flux towards increased salicylate production. The author observed that some variants containing high copy plasmids were unable to grow or contained gene mutations of interest, which may be due to the toxic effects of higher levels of salicylates on bacterial cell viability. Continuing to use viable EcNATT strains, they used liquid chromatography-mass spectrometry (LC-MS) to detect salicylates in the supernatant collected from overnight cultures. Generally speaking, higher copy number variants produce more salicylic acid, while EcNATT EntC ColE1 produces about 15 microns per 109 bacterial species. In addition, comparing the highest producing strains of two copy number variants sc101 * – Irp9 and EntC ColE1 showed that compared to the EcN mutant, the salicylate production was higher when encoded by EcNATT. In summary, the author determined that the variant with the highest salicylic acid production is the EcNATT EntC ColE1 strain.
To evaluate the diagnostic potential of this optimized strain, the authors injected ApcMin/+and WT mice with EcNATT EntC ColE1 strain and collected feces and urine at predetermined time points. Use EcN recovered from feces collected 48 hours after forced feeding to confirm colonization or deficiency of strains in ApcMin/+and WT mice. In addition, feces were spread on kanamycin plates with selectivity for salicylate encoding plasmids to confirm plasmid retention. Collect urine before administration to establish baseline, and then collect urine again 24 and 48 hours after administration. Using LC-MS to detect the presence of salicylates, the authors observed that the salicylate content in ApcMin/+mice was five times higher than baseline levels after 48 hours of administration, while the salicylate levels in WT mice did not change over time. In separate studies, elevated levels of salicylic acid (the main metabolite of salicylates) were also detected in the urine of ApcMin/+mice. In summary, these data suggest that engineered bacteria can be transported into mice with tumors, maintain their genetic circuits, and serve as agents for non-invasive adenoma tracking, possibly for early detection in fecal and urine testing.
Figure 3 Engineering of EcN platform based on feces and urine for non-invasive adenoma tracking
Immunotherapy using EcN can reduce the burden of adenomas and alter the tumor immune microenvironment
Based on the non-invasive ability to determine the presence of adenomas, the author next attempts to address whether the screening system can adapt to treatment objectives and reduce the burden of polyps. Due to the fact that the ApcMin/+model is considered a microsatellite stable (MSS) model and traditionally has poor response to immunotherapy methods, the authors hypothesize that bacteria in the probiotic platform will act as immune adjuvants and provide a chassis for multiple immunotherapy payloads. They combined therapeutic delivery with genome encoded cleavage loop optimization (SLIC) of EcN lux strains to maximize the release of immunotherapy and assist in biological containment by controlling the EcN population. According to previous work, the use of this lysis based release mechanism is necessary for effective release therapy and is crucial for therapeutic efficacy. In addition, bacterial dissolution leads to immune adjuvants, further enhancing the therapeutic effect of immunotherapy. SLIC is used to deliver nano antibodies that block PD-L1 and CTLA-4 targets, as well as cytokine GM-CSF (SLIC-3). It has been previously demonstrated that it can enhance the efficacy of checkpoint blockade therapy in a subcutaneous mouse colorectal model when delivered within the tumor. Here, ApcMin/+mice were orally administered PBS or SLIC-3 twice within 3-4 days, and then euthanized approximately 1 month later. The histological analysis of tumors stained with hematoxylin and eosin showed that the overall area and number of adenomas decreased by about 47% after SLIC-3 treatment. It is worth noting that an increase in the percentage of smaller adenomas was observed in mice treated with SLIC-3, while mice treated with PBS often had larger adenomas. Furthermore, this reduction is not specific to a specific location, but can be observed throughout the entire small intestine. Interrogation of the immune phenotype of tissue slices indicates that the reduction in adenoma burden is associated with increased infiltration of CD3+and CD8+cells and production of granzyme B within the adenoma, indicating immune mediated tumor cell death in SLIC-3 treated mice. In addition, compared to untreated mice, the use of SLIC alone increased the staining of granzyme B, which may be due to the inherent immunogenicity of lytic bacteria, which is also the basis for the additional benefits brought by utilizing bacterial based platforms.
Figure 4 shows the use of EcN for treatment to generate immunotherapy, reduce the burden on adenomas, and alter the tumor immune microenvironment.
In summary, this study demonstrated the selectivity and robust colonization of adenomas and tumors in two different in situ mouse models and human CRC patients treated with oral probiotic EcN. By utilizing this colonization ability, the possibility of engineered EcN in the diagnosis of adenomas has been demonstrated through non-invasive fecal and urine testing. In addition, by encoding EcN to generate checkpoint inhibitor therapy and cytokine GM-CSF, the therapeutic potential has been demonstrated, which can significantly reduce the burden of adenomas in the MSS CRC model through oral administration.
