Development of a Limosilactobacillus reuteri therapeutic delivery platform with reduced colonization potential

Abstract

Bacterial biotherapeutic delivery vehicles have the potential to treat a variety of diseases. This approach obviates the need to purify the recombinant effector molecule, allows delivery of therapeutics in situ via oral or intranasal administration, and protects the effector molecule during gastrointestinal transit. Lactic acid bacteria have been broadly developed as therapeutic delivery vehicles though risks associated with the colonization of a genetically modified microorganism have so-far not been addressed. Here, we present an engineered Limosilactobacillus reuteri strain with reduced colonization potential. We applied a dual-recombineering scheme for efficient barcoding and generated mutants in genes encoding five previously characterized and four uncharacterized putative adhesins. Compared with the wild type, none of the mutants were reduced in their ability to survive gastrointestinal transit in mice. CmbA was identified as a key protein in L. reuteri adhesion to HT-29 and enteroid cells. The nonuple mutant, a single strain with all nine genes encoding adhesins inactivated, had reduced capacity to adhere to enteroid monolayers. The nonuple mutant producing murine IFN-β was equally effective as its wild-type counterpart in mitigating radiation toxicity in mice. Thus, this work established a novel therapeutic delivery platform that lays a foundation for its application in other microbial therapeutic delivery candidates and furthers the progress of the L. reuteri delivery system towards human use.IMPORTANCEOne major advantage to leverage gut microbes that have co-evolved with the vertebrate host is that evolution already has taken care of the difficult task to optimize survival within a complex ecosystem. The availability of the ecological niche will support colonization. However, long-term colonization of a recombinant microbe may not be desirable. Therefore, strategies need to be developed to overcome this potential safety concern. In this work, we developed a single strain in which we inactivated the encoding sortase, and eight genes encoding characterized/putative adhesins. Each individual mutant was characterized for growth and adhesion to epithelial cells. On enteroid cells, the nonuple mutant has a reduced adhesion potential compared with the wild-type strain. In a model of total-body irradiation, the nonuple strain engineered to release murine interferon-β performed comparable to a derivative of the wild-type strain that releases interferon-β. This work is an important step toward the application of recombinant L. reuteri in humans.

Keywords: adhesins; biotherapeutic; colonization; engineered probiotic; probiotic.

Fig 1

Conceptual overview of a strain with reduced colonization potential. (a) The locations of 10 putative adhesion proteins in the L. reuteri VPL1014 genome are indicated. (b) In this study, we mutated each putative adhesion mutant individually and sequentially to yield a nonuple mutant. Depicted here are graphical examples of single mutants and the nonuple mutant. (c) We hypothesize that the nonuple mutant, lacking multiple adhesion proteins, will have reduced adherence to human intestinal cells. In this graphic, wild-type L. reuteri can adhere to mucus or epithelial cells, while the nonuple mutant is unable to adhere.

Fig 2

Mutant barcoding scheme and dual-recombineering efficiencies. (a) L. reuteri VPL4011 contains an insertion of an inactivated chloramphenicol acetyltransferase gene (cat*) coded on the negative strand. Incorporation of oVPL3848, which contains a series of degenerate bases (“N’s”) at the stop codon position (underlined) and at adjacent wobble bases, restores the cat gene while barcoding the strain with random bases. (b) A sequence logo generated from the sequencing results of 96 barcodes in VPL4011 transformed with oVPL3848. Positions displayed correspond to the barcoded bases (“N’s”) in oVPL3848. Positions 1, 5, and 6 are the wobble base positions, and positions 2–4 are at the location of the replaced stop codon. (c) Distribution of amino acids that restored cat* in VPL4011 following transformation with oVPL3848. (d) Recombination efficiencies of VPL4011 dual-transformed with oVPL236 (236) and oVPL3848 (3848). Rifampicin-resistant colonies out of 100 patched (patch) colonies from agar plates supplemented with chloramphenicol to agar plates supplemented with rifampicin. The results shown are average values derived from three independent experiments ± standard error of the mean.

Fig 3

Construction, recovery, and growth analysis of adhesin mutants. (a) The gene targets and recombineering oligonucleotides used in this study are listed on the left. DNA sequences of the targeted adhesion proteins of L. reuteri are shown aligned with each recombineering oligonucleotide, with the encoded amino acid listed above. Numbers above the first and last amino acids encoded by each sequence indicate the amino acid positions. The directions of the lagging strands are indicated, and underneath are the mismatches in the recombineering oligonucleotides, which result in internal stop codons in each gene. Resulting amino acid mutations are indicated underneath the mismatched nucleotides. The coding strand is indicated in the third column, and barcodes derived from oVPL3848 are listed in the fourth column. (b) Scheme to optimize adhesin mutant recovery. Barcoding oligonucleotide oVPL3848 (3848, black) was dual-transformed into VPL4011 with an oligonucleotide targeting different adhesins (####; red, yellow, and blue) via electroporation. From the three transformations, colonies were selected on agar supplemented with chloramphenicol (Cm). Thirty CFUs from each transformation were screened for mutant genotypes by MAMA PCR. Wild-type genotypes have an expected size of 1 kb, while recombinant genotypes are 0.5 kb. If no recombinants were recovered from one of the transformations (yellow, in this example), the corresponding recombineering oligonucleotide was re-transformed alongside oligonucleotides targeting two additional adhesins. This process was repeated until each adhesin mutant was recovered. (c) Growth curves of all single mutants. Wild-type control (WT) is VPL4052, which contains the cat* gene insertion restored to cat with oVPL283. Data for WT are the same across the three growth curves, while the mutants are split across the three. The results shown are averages from three independent experiments ± standard error of the mean. right.

Fig 4

Mouse gastrointestinal survival of adhesin mutants and adhesion to human colon cancer cells. (a) Mice (n = 5–8) were administered with 10⁸ CFU of each mutant for two consecutive days. At 15, 27, and 39 h after the second gavage, fecal material was collected, resuspended to 100 mg/mL in PBS, and plated for quantification. (b) At 15 h, we measured the survival of each adhesion mutant following transit through the murine GI tract (n = 5). A mix of VPL4011 (n = 8) transformed with an oligonucleotide conferring each mutant barcode served as a control (WT mix). (C) Persistence of each mutant is depicted as the CFU recovered over the course of the in vivo experiment. Data for WT are the same across both graphs. (d) Relative percent adhesion of each mutant and its complemented strain to HT-29 cells was compared with chloramphenicol-resistant wild-type control (VPL4052) and empty vector control of L. reuteri (VPL31134), respectively. VPL1014 served as the WT control. The results shown are averages from six independent experiments with three technical replicates each, ±standard error of the mean, *; P < 0.05, **; P < 0.01, no statistical label; P > 0.05 (two-tailed unpaired t-test between adhesion mutant and complemented strain) (e-g) Change in relative ratio of each strain within each sample recovered from enteroid monolayers (T_F) compared with the respective ratio of each strain in the starting mixture (T₀). Data are presented as the change in relative percent (Δ%(T_F- T₀) for the barcode control mix (e), mutant mix (f), and the complemented mix (g) based on sequencing reads targeting the cat barcode. Positive numbers indicate an increase in relative ratio while negative numbers indicate a decrease. The results shown are averages from three independent experiments with three technical replicates each, ±standard error of the mean; *, P < 0.05.

Fig 5

In vivo characterization and therapeutic efficacy of the nonuple mutant. (a) Growth curve and mitomycin C induction of L. reuteri VPL1014 (WT) and nonuple mutant. (b) At the endpoint of the growth experiment (T₈), samples derived from uninduced and induced cultures of L. reuteri VPL1014 (WT) and the nonuple variant were processed to quantify phage production (PFU) (P > 0.3). (c) Adhesion of L. reuteri VPL1014 (WT) and nonuple mutant to monolayers of HT-29 cells (P > 0.4). (d) Adhesion competition experiment on human enteroid monolayers between wild-type control (WT) and the nonuple mutant. Multiplicities of infection (MOI) ratios of 5:1 and 30:1 were tested. Results are expressed as a ratio of wild-type CFU recovered compared with nonuple CFU recovered from the adhesion assay. T0 and TF represent the ratio of WT and nonuple cells before and after the adhesion assay, respectively. (a-b) The results shown are averages from three independent experiments ± standard error of the mean. For panels (c) and (d), the results shown are averages from three independent experiments (with three technical replicates each) ±standard error of the mean. *, P < 0.05; ***, P < 0.005. (e) Persistence of WT and the nonuple mutant in the mouse GI tract. Male B6 mice (n = 6/group) were orally gavaged with 100 µL (10⁹ CFU/mouse) WT or nonuple mutant, and persistence of each bacteria was monitored daily for7 days post-oral gavage. The clearance rate was determined from absolute slope value of linear curve between bacterial concentration [Log (CFU/100 mg)] and post-oral gavage days, constructed using data from 0 to 5 days post-oral gavage. ***, P < 0.001 (two-tailed unpaired t-test). (f) Survival of mice exposed to partial body irradiation (PBI) was comparable between the treatment groups LR-IFN-β and nonuple-IFN-β. Twenty-four hours following PBI (13.35 Gy), female C57BL/6 mice (n = 15/group) were gavaged with 200 µL saline (sham), LR-IFN-β, or nonuple-IFN-β, each 10⁹ CFU. The Survival of animals was monitored for 30 days. Significance in survival was determined by a two-sided log-rank test with P < 0.05 regarded as statistically significant.