A programmable encapsulation system improves delivery of therapeutic bacteria in mice

Abstract

Living bacteria therapies have been proposed as an alternative approach to treating a broad array of cancers. In this study, we developed a genetically encoded microbial encapsulation system with tunable and dynamic expression of surface capsular polysaccharides that enhances systemic delivery. Based on a small RNA screen of capsular biosynthesis pathways, we constructed inducible synthetic gene circuits that regulate bacterial encapsulation in Escherichia coli Nissle 1917. These bacteria are capable of temporarily evading immune attack, whereas subsequent loss of encapsulation results in effective clearance in vivo. This dynamic delivery strategy enabled a ten-fold increase in maximum tolerated dose of bacteria and improved anti-tumor efficacy in murine models of cancer. Furthermore, in situ encapsulation increased the fraction of microbial translocation among mouse tumors, leading to efficacy in distal tumors. The programmable encapsulation system promises to enhance the therapeutic utility of living engineered bacteria for cancer.

Fig. 1. Programmable CAP system for control over bacterial encapsulation and in vivo delivery profiles.

a, We engineered the biosynthetic pathway of bacterial CAP for tunable and dynamic surface modulation of the probiotic E. coli Nissle 1917 with synthetic gene circuits. This approach enables increased CAP levels upon induction to control immune evasion and clearance. b, The programmable CAP system enhances systemic delivery of bacteria by transiently expressing CAP. Non-CAP bacteria (thin gray cells) elicit toxicity by exposing the immunogenic bacterial surface, and permanently CAP-expressing bacteria (thick black cells) lead to overgrowth. The iCAP (blue cells) system enables transient encapsulation of bacteria, thus reducing initial inflammation while effectively clearing bacteria over time. c, The CAP system controls bacterial translocation among tumors. The iCAP system allows in situ activation of CAP in one tumor, which results in inducible bacteria translocation to distal, uncolonized tumors.

Fig. 2. sRNA knockdown screen identifies key genes in CAP biosynthesis.

a, Schematic of K5 CAP biosynthesis in EcN. CAP is composed of an alternating polymer chain of GlcA and GlcNAc connected to a poly-Kdo linker. Subsequently, CAP is transported from the inner bacteria membrane to the outer membrane. b, Quantification of microbial growth parameters of EcN KD strains in nutrient-rich media (LB), blood or phage-containing media. Growth rate denotes maximum specific growth rate (h⁻¹) obtained by fitting growth curve to measured OD₆₀₀ over time. Blood viability is defined as bacterial CFU ml⁻¹ after 6-hour incubation in human blood. Phage sensitivity is calculated by area under the curve of bacterial turbidity over 6 hours of incubation with LB media containing ΦK1-5. c, Phage sensitivity of EcN and EcN ΔkfiC. Plaque-forming assay demonstrates complete absence of infection and lysis in EcN ΔkfiC. The representative images show difference between serially diluted PFU of bacteria with and without CAP. Error bars represent s.e.m. over three independent samples. d, TEM images showing CAP encapsulation of the cellular outer surface. kfiC KO results in the absence of CAP nanostructure on the cell surface of EcN ΔkfiC. White arrows indicate cell surface. WT, wild-type.

Fig. 3. Design and characterization of the iCAP system.

a, Inducible gene circuit diagram whereby the kfiC gene was cloned under the control of a lac promoter to allow inducible CAP expression via the small molecule IPTG. b, SDS–PAGE gel stained with alcian blue showed elevating levels of CAP production corresponding to the IPTG concentration (top). The densitometric analysis of CAP bands demonstrated that CAP production reaches maximum at approximately 1 µM IPTG (bottom). c, SDS–PAGE gels and densitometric analysis show CAP kinetics upon induction (left) and decay (right). For b and c, source data are provided as a Source Data file. d, Ruthenium-red-stained TEM images showing change in CAP in titrating IPTG concentration. Histograms reveal shift in cellular outer layer thickness as IPTG concentration increases. Insets show representative images of bacteria and zoomed outer surface structure. Dotted lines indicate inner and outer (white) perimeters of CAP. Scale bars, 40 nm (left) or 200 nm (right) in each inset. Five cells from each group were analyzed to generate the histograms. All error bars represent s.e.m. over two independent samples. OP, outer polysaccharide. Source data

Fig. 4. Tunable interaction of the CAP system with host immune factors.

a, Bacteria were encapsulated using the iCAP system and exposed to human whole blood to test CAP-mediated protection. Elevating levels of CAP activation with IPTG enabled a corresponding increase in bacterial survival in human whole blood. Bacteria were pre-induced with IPTG before blood exposure. b, Representative images of bacteria spotted on LB agar plate after 1-hour incubation in human whole blood (right). c, Survival kinetics using varying levels of IPTG induction before incubation with human whole blood. 10², 10³ or 10⁴ nM IPTG were added to the bacteria overnight culture to pre-induce the iCAP system. d, Induced or un-induced iCAP bacteria were co-cultured with BMDMs to test CAP-mediated protection from phagocytosis. e, BMDMs were lysed after incubating with bacteria for 30 minutes to enumerate phagocytosed bacteria (**P = 0.007, two-sided unpaired t-test). f, Representative fluorescence microscopy images showing bacteria (GFP, top) in phagocytes (bright-field overlayed with GFP, bottom). Scale bars, 10 µm. g, Human THP-1 cells were co-incubated with EcN, EcN ∆kfiC or EcN iCAP (pre-induced with 10 µM IPTG) to test for immunogenicity. h–j, Cytokine levels in the culture media were measured using Luminex multiplex assay including TNFα (**P = 0.001 and ***P = 0.0002, respectively, at 24 hours after incubation), IL1β (*P = 0.017 at 4 hours after incubation) and IL6 (*P = 0.029 at 4 hours after incubation). All error bars represent s.e.m. over three independent samples, and statistical analyses were performed using one-way ANOVA with Tukey’s multiple comparison test. LOD at 2 × 10² CFU ml⁻¹ (b and c) and 1 × 10² CFU ml⁻¹ (e).

Fig. 5. Transient CAP activation improves systemic bacterial delivery and efficacy in vivo.

a, Host response was evaluated after bacterial administration in mice. EcN iCAP was pre-induced with 10 µM IPTG and allowed to gradually attenuate CAP over time. b–f, Serum cytokine levels after 5 × 10⁶ CFU bacterial administration. IL1β (b, *P = 0.039 at 4 hours p.i.), IL6 (c, **P = 0.0037 at 4 hours p.i.), IL10 (d, **P = 0.0089 and *P = 0.014, respectively, at 24 hours p.i.), G-CSF (e, ****P < 0.0001 and **P < 0.01, respectively, at 24 hours p.i.) and GM-CSF (f, *P = 0.019 at 4 hours p.i.) were measured. g, Toxicity was evaluated by bacterial administration with elevating doses. h, Change in animal body weight after i.v. bacterial administration at 5 × 10⁶ CFU (**P = 0.004; n = 10, 5 and 5 mice for EcN iCAP, EcN and EcN ΔkfiC groups, respectively). i, Survival curve (>15% body weight reduction; n ≥ 5 mice per group) after bacterial administration at 1~7 × 10⁷ CFU. j, Dose–toxicity curve with MTD = 4.4 × 10⁷ CFU, 5.8 × 10⁶ CFU and 9.6 × 10⁶ CFU for EcN iCAP, EcN and EcN ΔkfiC, respectively. MTD was calculated based on TD₅₀ (>10% body weight drops p.i.; non-linear regression with least squares fit; n ≥ 5 per group). k, Mice bearing tumors were intravenously injected with EcN MTD, EcN ΔkfiC MTD and EcN iCAP MTD (pre-induced with 10 µM IPTG) or EcN iCAP low (pre-induced with 10 µM IPTG) at 5 × 10⁶, 1 × 10⁷, 5 × 10⁷ or 5 × 10⁶ CFU, respectively. Bacteria were engineered to produce TT. l, Bacterial growth trajectories in subcutaneous CT26 tumors measured by bacterial luminescence (****P < 0.0001; n = 14, 13, 9 and 13 tumors for EcN MTD, EcN ΔkfiC MTD, EcN iCAP MTD and EcN iCAP low groups, respectively). Luminescence values are normalized to basal luminescence of individual strains. m, n, Therapeutic efficacy measured by relative tumor size over time in a syngeneic CT26 model (m; ****P < 0.0001, ***P = 0.0008, **P = 0.003; n = 14, 13, 9, 13 and 11 tumors for EcN MTD, EcN ΔkfiC MTD, EcN iCAP MTD, EcN iCAP low and PBS groups, respectively) and in a genetically engineered spontaneous breast cancer MMTV-PyMT mouse model (n; *P = 0.0197; n = 15, 15 and 9 tumors for EcN MTD, EcN iCAP MTD and PBS groups, respectively). MMTV-PyMT tumors were measured by calipering three orthotopic regions in mammary glands (upper left, upper right and bottom). Mice in PBS groups reached study endpoint 10 days p.i. Statistical analyses were performed using one-way ANOVA (b–f) and two-way ANOVA (h, i–n) with Tukey’s multiple comparison test. Bacteria were engineered to produce TT (l–n). All error bars represent s.e.m. over three independent samples unless otherwise noted. All ‘n’ denotes number of biological replicates. a.u., arbitrary units; GM-CSF, granulocyte–macrophage colony-stimulating factor; p.i., post injection.

Fig. 6. In situ activation of the CAP enables bacterial translocation and drug delivery to distal tumors.

a, Schematic of iCAP-mediated bacterial translocation. EcN iCAP is injected into one tumor. iCAP activation enables bacteria translocation to distal tumors. b, Mice harboring multiple tumors are injected with EcN iCAP into a single tumor (treated tumor). Mice are fed 10 mM IPTG water to activate iCAP in situ. To quantify bacterial biodistribution, mice were imaged daily for bacterial bioluminescence, and organs were harvested and bacterial colonies were counted after 3 days. c–e, Inducible translocation of EcN iCAP to distal tumors in CT26 syngeneic (c), 4T1 orthotopic (d) and MMTV-PyMT genetically engineered (e) mouse tumor models. Representative IVIS images showing bacterial translocation in vivo. White arrows indicate location of bacterial injection. Black arrows indicate location of bacterial translocation. Translocation is quantified by fraction of bacteria found in distal tumor compared to treated tumor. Bacteria number is measured by performing biodistribution for CFU g⁻¹ enumeration for CT26 (*P = 0.032; n = 4 and 5 tumors for −IPTG and +IPTG groups, respectively), 4T1 (*P = 0.029; n = 4 tumors for both −IPTG and +IPTG groups) and MMTV-PyMT (**P = 0.003; n = 14 and 11 tumors for −IPTG and +IPTG groups, respectively) models. All error bars represent median, and statistical analyses were performed using a two-sided Mann–Whitney test. f, Representative images of ex vivo organ images taken with IVIS showing bacterial tumor translocation in 4T1 orthotopic mouse model. g, Schematics of engineered EcN capable of inducible translocation and therapeutic expression. Translocation and therapeutic production are externally controlled by IPTG and AHL, respectively. h–i, Therapeutic efficacy in treated and distal CT26 tumors measured by relative tumor growth over time. All bacteria were engineered to produce TT. Bacteria were injected into a single treated tumor. Three days p.i., AHL was administered. Mice were fed with (h) or without (i) IPTG water, and tumor size was measured (NS P = 0.83 and **P = 0.004, respectively; two-way ANOVA, n = 6 and 5 for treated and distal tumors, respectively). All error bars represent s.e.m. All ‘n’ denotes number of biological replicates. NS, not significant; p.i., post injection.