A synthetic probiotic engineered for colorectal cancer therapy modulates gut microbiota

Abstract

Background: Successful chemoprevention or chemotherapy is achieved through targeted delivery of prophylactic agents during initial phases of carcinogenesis or therapeutic agents to malignant tumors. Bacteria can be used as anticancer agents, but efforts to utilize attenuated pathogenic bacteria suffer from the risk of toxicity or infection. Lactic acid bacteria are safe to eat and often confer health benefits, making them ideal candidates for live vehicles engineered to deliver anticancer drugs.

Results: In this study, we developed an effective bacterial drug delivery system for colorectal cancer (CRC) therapy using the lactic acid bacterium Pediococcus pentosaceus. It is equipped with dual gene cassettes driven by a strong inducible promoter that encode the therapeutic protein P8 fused to a secretion signal peptide and a complementation system. In an inducible CRC cell-derived xenograft mouse model, our synthetic probiotic significantly reduced tumor volume and inhibited tumor growth relative to the control. Mice with colitis-associated CRC induced by azoxymethane and dextran sodium sulfate exhibited polyp regression and recovered taxonomic diversity when the engineered bacterium was orally administered. Further, the synthetic probiotic modulated gut microbiota and alleviated the chemically induced dysbiosis. Correlation analysis demonstrated that specific bacterial taxa potentially associated with eubiosis or dysbiosis, such as Akkermansia or Turicibacter, have positive or negative relationships with other microbial members.

Conclusions: Taken together, our work illustrates that an effective and stable synthetic probiotic composed of P. pentosaceus and the P8 therapeutic protein can reduce CRC and contribute to rebiosis, and the validity and feasibility of cell-based designer biopharmaceuticals for both treating CRC and ameliorating impaired microbiota. Video abstract.

Keywords: AOM/DSS model of colitis-associated cancer; Akkermansia; Alanine racemase; DLD-1 xenograft; Lactobacillus rhamnosus CBT LR5 (KCTC 12202BP); Microbiome; Turicibacter.

Fig. 1

Module design of a lactic acid bacterium-based drug delivery system for optimal P8 productivity. a A schematic outline depicting the expected mode of action of the synthetic probiotic PP*-P8 with the alr complementation system. alr, the alanine racemase gene. b Constructs with various promoters for dual expression of the P8 therapeutic protein fused to the 27-residue Usp45 leader peptide. GK, glucose kinase; LDH, l-lactate dehydrogenase; PK, pyruvate kinase; ChoS, choline ABC transporter permease and substrate binding protein. c Concentrations of P8 secreted from PP*-P8 that were quantified using ELISA, indicating that the PK-PK promotor system had the highest amount of secreted P8

Fig. 2

Anti-tumor efficacy of the PP*-P8 probiotic in the DLD-1 xenograft mouse model. a Increased sizes of DLD-1-derived tumors recorded each week. Mice (n = 10 in each group) were subcutaneously inoculated with 2 × 10⁶ DLD-1 cells in the rear right flank and then received 0.9% saline (control), 60 mg/kg body weight gemcitabine (dFdC; intraperitoneal injection, twice a week), 1 × 10¹⁰ CFU/head P. pentosaceus alr (pCBT24-2-alr) (PP*; oral administration, five times a week), or 1 × 10¹⁰ CFU/head P. pentosaceus alr (pCBT24-2-PK-p8-PK-p8-alr) (PP*-P8; oral administration, five times a week). ***P < 0.001 for control vs. dFdC, control vs. PP*-P8, dFdC vs. PP*, PP* vs. PP*-P8. b Extracted tumor tissues from each treatment group 6 weeks after the DLD-1 xenograft. c Inhibition ratios for tumor growth calculated from the mean tumor weights of the control group and the test groups. ***P < 0.001. d Relative fold changes in the expression of cell cycle regulatory factors between PP* with control and PP*-P8 with control. Each vertical bar represents the arithmetic mean of three replicates. *P < 0.05, **P < 0.01, ***P < 0.001

Fig. 3

AOM/DSS-induced mouse model of colitis-associated colon carcinogenesis. a The experimental scheme for tumor induction by azoxymethane (AOM) and dextran sodium sulfate (DSS). Mice (n = 10 in each group) were intraperitoneally injected with 12.5 mg/kg body weight AOM on day 1 and on day 5; they were given water containing 2% w/v DSS for 5 days, followed by regular water for 16 days, which was repeated three times during the 68-day treatment. Treatment groups: untreated control (0.9% saline, oral), fluorouracil (5-FU; 40 mg/kg body weight, intraperitoneal, twice a week), wild-type P. pentosaceus (PP WT; 1 × 10¹⁰ CFU/head, oral, five times a week), P. pentosaceus alr (pCBT24-2-alr) (PP*; 1 × 10¹⁰ CFU/head, oral, five times a week), and P. pentosaceus alr (pCBT24-2-PK-p8-PK-p8-alr) (PP*-P8; 1 × 10¹⁰ CFU/head, oral, five times a week). Schedules for fecal sampling are indicated with arrows. b Temporal dynamics of the PP*-P8 population in relative abundance during the experimental period. Dashed lines represent the DSS treatment episodes. c Bleeding scores were assessed every 5 days by hemoccult testing and visible signs. Dashed lines represent the DSS treatment episodes. d Comparison of bleeding scores in days 25, 45, and 68, the last days of each stage. *P < 0.05, **P < 0.01, ***P < 0.001

Fig. 4

Effects of PP*-P8 on general health and tumorigenesis in the AOM/DSS mouse model. a Changes in the bodyweight of mice were recorded each week. **P < 0.01. b Kaplan-Meier survival curves for mouse with five different treatment groups. Log-rank test was performed to measure the statistical significance. *P < 0.05. c Macroscopic and histopathological appearance of polyps. d Colon length and e number of polyps were measured after 68 days. *P < 0.05, **P < 0.01, ***P < 0.001

Fig. 5

Longitudinal analyses of the gut microbiota of AOM/DSS mice treated with PP*-P8. a Changes in alpha diversity indices of microbial communities in the fecal samples. Species richness and evenness are plotted as the number of observed ASVs and inverse Simpson and Shannon indices. b Principal coordinate analysis based on Bray-Curtis dissimilarity. Each dot indicates a single sample and each group is shown in a different color. P values correspond to the permutational multivariate analysis of variance results. c Microbial composition at the family level is shown as relative abundance. Except for stage 0.5, which shows a single sample, proportions are the averages of five samples. D0, day 0

Fig. 6

Specific microbial taxa likely associated with differences between the treatment groups. a Linear discriminant analysis effect size of samples after the final DSS administration. b Positive and negative correlation matrix between the top 20 most abundant bacterial taxa. Results of a pairwise Spearman’s rank correlation after the final DSS administration are shown. Correlations with P values less than 0.05 are marked with asterisk symbols and adjusted P values less than 0.05 by the Benjamini-Hochberg FDR method are colored black. Related genera based on Euclidean distance were clustered together. Red, positive correlation; blue, negative correlation. Uncultured (unc) or unassigned (una) genera were labeled with the initials of their family names, [M], [L], and [R] represent Muribaculaceae, Lachnospiraceae, and Ruminococcaceae, respectively. c Scatterplot with Spearman’s rank correlations between the relative abundances of six genera and Shannon indices. Those with significant P values in b are shown in c. d Quantification of the number of cells for Akkermansia and Turicibacter as measured by quantitative PCR in fecal samples on day 68 after the final DSS treatment. *P < 0.05, **P < 0.01, ***P < 0.001. e Discriminative functional pathway abundant between control versus PP*-P8, PP* versus PP*-P8, and control versus 5-FU