Sonogenetics-controlled synthetic designer cells for cancer therapy in tumor mouse models

Summary

Bacteria-based therapies are powerful strategies for cancer therapy, yet their clinical application is limited by a lack of tunable genetic switches to safely regulate the local expression and release of therapeutic cargoes. Rapid advances in remote-control technologies have enabled precise control of biological processes in time and space. We developed therapeutically active engineered bacteria mediated by a sono-activatable integrated gene circuit based on the thermosensitive transcriptional repressor TlpA₃₉. Through promoter engineering and ribosome binding site screening, we achieved ultrasound (US)-induced protein expression and secretion in engineered bacteria with minimal noise and high induction efficiency. Specifically, delivered either intratumorally or intravenously, engineered bacteria colonizing tumors suppressed tumor growth through US-irradiation-induced release of the apoptotic protein azurin and an immune checkpoint inhibitor, a nanobody targeting programmed death-ligand 1, in different tumor mouse models. Beyond developing safe and high-performance designer bacteria for tumor therapy, our study illustrates a sonogenetics-controlled therapeutic platform that can be harnessed for bacteria-based precision medicine.

Graphical abstract

Highlights

• •Developing a high-sensitivity, precisely controllable SINGER for thermosensitivity
• •Engineering VNP20009 cells with SINGER for ultrasound-controlled transgene expression
• •Intravenously administered VNP20009 with SINGER suppresses tumors in mouse models
• •VNP20009 with SINGER delivers multiple therapeutics for enhanced tumor suppression

Gao et al. develop engineered bacteria equipped with a sono-activatable integrated gene circuit (SINGER) in which a short, ultrasound-triggered thermal stimulus activates the local expression and release of therapeutic payloads. Subsequently, they demonstrate the therapeutic utility of the designer bacteria against multiple tumor mouse models using multiple therapeutic proteins.

Introduction

Genetically engineered bacteria have been attracting increasing attention as living and intelligent devices for disease diagnosis and therapy.^1,2,3,4 Engineered bacteria have been deployed as sentinels to monitor gastrointestinal disorders,⁵ liver metastasis,⁶ and liver dysfunction,⁷ promote amelioration of metabolic disorders,^8,9,10 and protect against pathogen infections.^{11,12,13,14,15} Further, certain obligate or facultative anaerobic bacteria have been shown to selectively colonize tumors, as they can infiltrate tumors, particularly immune-deficient hypoxic cores.^{16,17,18,19,20,21,22} Therefore, these bacteria have been engineered as tumor-specific drug delivery carriers to locally release therapeutic payloads for suppressing tumor growth. For example, Escherichia coli Nissle 1917 has been engineered to specifically target and suppress B16F10 melanoma and 4T1 breast cancer through the secretion of azurin.²³ The therapeutic application of these engineered bacteria has required the design of remote-control systems that are functionalized to perform user-defined transgene expression regulation and thus precisely control the dosage of therapeutic outputs in vivo.

Initial work in this area used chemical cues such as salicylate or L-arabinose as remote-control triggers for gene expression.^24,25 Still, they are inadequate for most therapeutic applications, as chemicals are systemically administered and incapable of spatiotemporal controllability.²⁶ Optogenetics has recently rapidly developed as a non-invasive technique with high spatiotemporal specificity and reversibility. Photosensitive proteins such as the light-oxygen-voltage domain²⁷ and bacteriophytochrome P1²⁸ have been used to engineer optogenetic switches to regulate gene expression in bacteria; however, their in vivo applications are hindered by poor penetration into deep tissues and potential phototoxicity.²⁹ Alternatively, ultrasound (US), characterized by non-invasiveness, spatiotemporal specificity, safety, and tissue penetrability,^30,31,32 has attracted significant attention in biomedical engineering for diagnostic and therapeutic applications.^{33,34,35,36,37}

Recently, two US-inducible transgene expression systems have been developed in bacteria for cancer therapy based on the thermosensitive bacteriophage λ repressor TcI.^38,39 After bacterial colonization, the tumors were irradiated by focused US to induce an elevated temperature of 42°C–45°C to activate the pR-pL tandem promoter to drive the expression of therapeutic outputs for immune activation. However, the integrase-based state switch cannot be controlled once activated. As a result, sustained production of immune checkpoint inhibitors may lead to immune-related adverse effects.³⁸ Another US-responsive bacterium was constructed by assembling a US-triggered therapeutic circuit into a laboratory strain of E. coli MG1655.^38,39 Further, 42°C–45°C may lead to thermal damage to healthy tissues. Consequently, developing US-inducible transgene expression systems with higher controllability and enhanced sensitivity is necessary. Ideally, these systems should be activated at lower temperatures ranging from 39°C to 40°C and constructed within safer non-pathogenic or attenuated bacterial strains for therapeutic purposes.

Hence, we here developed a sono-activatable integrated gene circuit (SINGER) in an attenuated strain of Salmonella typhimurium, VNP20009, the tolerance and safety of which have been proven in patients in several phase 1 clinical trials.⁴⁰ The circuit was carried on a plasmid and designed based on the thermosensitive transcriptional repressor TlpA₃₉, a mutant of TlpA derived from S. typhimurium that undergoes temperature-dependent allosterism and specifically binds to synthetic cognate promoters containing TlpA-specific binding sites in its low-temperature dimeric state and blocks transcription.⁴¹ By engineering different synthetic promoters and ribosome binding site (RBS) screening, we obtained an optimized US-triggered transgene expression system with a low background and high induction efficiency. We demonstrated US-induced gene expression and protein secretion mediated by the SINGER system in VNP20009. Moreover, therapeutic cargos produced by tumor-colonizing VNP20009 under US irradiation were successfully released to the tumor microenvironment, suppressed tumor growth, and significantly increased the survival rate of mice bearing multiple model tumors (Figure 1). The characteristics of the SINGER system, including non-invasiveness, robustness, strong tissue penetrability, and precise controllability, show great promise for developing innovative bacteria-based anticancer therapies.

Figure 1.

Schematic representation for bacteria-based tumor therapy in xenograft mouse tumor models triggered by ultrasound (US)

Engineered designer bacteria harboring synthetic genetic circuits are administered into circulation, after which they can accumulate, colonize, and proliferate inside tumors. The mild temperature increase caused by US stimulation leads to dissociation of the thermosensitive transcriptional repressor TlpA₃₉ from its synthetic promoter P_TlpA. This triggers the localized expression and release of therapeutic agents, such as azurin and PD-L1 nb, from the engineered bacteria. These agents can directly induce apoptosis in tumor cells or stimulate innate and adaptive antitumor immune responses, including activating T cells and NK cells. This comprehensive approach aims to eradicate tumors and prevent tumor relapse.

Results

Design and optimization of a US-triggered transgene expression system

Seeking to develop improved sonogenetic systems in bacteria for therapeutic delivery, we constructed a SINGER that can be controlled non-invasively by US based on the thermosensitive transcriptional repressor TlpA₃₉ and the reporter LuxCDABE driven by a synthetic promoter, P_TlpA, containing a TlpA₃₉ specific DNA-binding sequence (O_TlpA). As a thermosensitive coiled-coil protein, TlpA₃₉ can rapidly and reversibly dissociate from a homodimer form to a monomer as the temperature increases to 39°C. When the temperature is below 39°C, TlpA₃₉ homodimers bind to the synthetic chimeric promoter P_TlpA and block transcription of the LuxCDABE cassette. When the temperature increases to 39°C, the TlpA₃₉ protein dissociates from P_TlpA, thus enabling LuxCDABE transcription (Figure 2A). Cells of a VNP20009 strain harboring the US-inducible gene switch were cultured under a range of temperatures: the bioluminescence signal intensity increased by about 20-fold following a thermal transition from 37°C to 39°C. Additionally, we noted that the intensity of this transcriptional activation was induction-time dependent (Figures 2B and 2C).

Figure 2.

Construction and optimization of the SINGER system in Salmonella VNP20009

(A) Schematic representation of the US-inducible transgene expression switch. In the absence of US, the transcriptional repressor TlpA₃₉ forms a homodimer, binds to its specific promoter P_TlpA, and blocks LuxCDABE expression. Upon US, the mild temperature elevation leads to the dissociation of TlpA₃₉ from a homodimer to a monomer and dissociation from its promoter P_TlpA, which drives LuxCDABE expression.

(B) Quantification of transgene expression by engineered bacteria under various induction temperatures.

(C) Induction-time-dependent transgene expression kinetics at 39°C.

(D–F) Screening various ribosome binding sites (RBSs) (D), the number of tandem repeats of TlpA₃₉ binding element (O_TlpA) (E), and constitutive promoters of differing strengths (P_const) to drive TlpA₃₉ expression (F).

(G) Heatmap visualization of the correlation between the temperature of bacteria solution and the examined irradiating acoustic parameters.

(H) Schematic experimental procedure for SINGER-mediated transgene expression.

(I) Irradiation-time-dependent SINGER-mediated transgene expression kinetics.

(J) SINGER-mediated transgene expression in different bacteria strains.

All samples in (B)–(F) were maintained at 37°C after the indicated period of temperature induction (for a total experimental duration of 12 h from the start of induction to the sample collection) before the bioluminescence signals were quantified using a microplate reader. All samples in (I) and (J) were maintained at 37°C after the indicated period of US irradiation (for a total experimental duration of 12 h from the start of the US stimulation to the sample collection) before the bioluminescence signals were quantified using a microplate reader. Data in (B)–(F), (I), and (J) are presented as means ± SD; n = 3 independent experiments. Each data point represents the mean of three technical replicates.

See also Figure S1 and Tables S1–S3.

Seeking reduced basal activation and higher induction efficiency, we constructed and assessed a variety of configurations to optimize the induction profiles in VNP20009 cells using LuxCDABE as a reporter (SINGER-LuxCDABE). Specifically, we tested six RBSs with various translational efficiencies, three synthetic promoters (P_TlpA) with different tandem repeat variants of O_TlpA, and four different constitutive promoters for driving TlpA₃₉ expression (Tables S1–S3).^41,42 The combination of RBS3, P_TlpA1 (1×O_TlpA), and P_J23100 showed the highest fold induction (∼37-fold) of the reporter cultured at 39°C (Figures 2D–2F). This optimized circuit design achieved the best induction performance in VNP20009. To achieve SINGER-mediated transgene expression that US can control non-invasively, we adjusted acoustic parameters including US intensity and irradiation duration time. Exposing bacterial cultures to 0.5 W/cm² US irradiation with a 1:1 pulse pattern (1 s on, 1 s off) raised their temperature to a steady range of 39°C–40°C (Figure 2G). This temperature range is favorable to transcriptional activation. Additionally, our tests on different ultrasonic intensities showed that continuous US exposure at 0.4 and 0.5 W/cm² led to lower gene expression levels than pulsed US at 0.5 W/cm² with a 1:1 pulse pattern (Figure S1). The bioluminescence signal displayed US irradiation-time-dependent increases under 0.5 W/cm² US irradiation with a 1:1 pulse (0.5 W/cm², pulse of 1 s on, 1 s off) (Figures 2H and 2I). Moreover, we observed that the bioluminescence signal continued to rise for up to 6 h following the cessation of US irradiation (Figure S2).

To assess possible side effects of US on bacteria, VNP20009 cells carrying SINGER-LuxCDABE were exposed to US irradiation (0.5 W/cm², pulse of 1 s on, 1 s off) over a time course. The growth curve of bacteria indicated that US has no adverse effect on cell viability (Figures S3A and S3B). To further confirm that US does not cause bacterial lysis and death, we employed live/dead bacterial stains. After US stimulation, we observed good bacterial viability through fluorescence microscopy (Figure S3C). We subsequently evaluated the impact of US on both tissue cells and tumor cells by exposing hMSC-TERT, ARPE-19, NIH-3T3, HEK-293T, B16F10, and CT26 cells to US irradiation (0.5 W/cm², pulse of 1 s on, 1 s off) for a duration of 1 h. The cell viability was assessed using the CCK-8 assay, which indicated that US irradiation did not have adverse effects on the viability of either the tissue cells or the tumor cells (Figures S3D and S3E). We also introduced the system to five other bacteria. We observed that the gene expression was US-irradiation dependent in all tested strains (Figure 2J), supporting the fact that the SINGER system works in multiple bacteria. These results support that our SINGER system can be used in various bacteria for US-induced, adjustable transgene activation.

Construction and characterization of US-controlled engineered bacteria for antitumor applications

For tumor therapy, the functional proteins produced by engineered bacteria must be efficiently released into the tumor.⁴³ We evaluated several secretion signal peptides, including commonly used ones such as pelB,⁴⁴ ompT,⁴⁵ and lamB,⁴⁶ for protein secretion in our engineered bacterial system. However, immunoblotting analyses failed to detect any proteins in the bacterial supernatant. Consequently, we turned to explore the Yersinia YopE N-terminal signal peptide as an alternative. YopE functions as a type III secreted effector, and its N-terminal 15 amino acids encode a class type III secretion sequence.⁴⁷ We fused the YopE_1–15 type III secretion sequence to the Gaussia luciferase (Gluc) reporter protein to guide its extracellular secretion (Figure 3A). Under US irradiation, the bioluminescence signal of the Gluc reporter in the culture supernatants was increased in a time-dependent manner (Figure 3B). Subsequent comparative analysis on the secretion of Gluc, with and without the YopE_1–15 sequence, revealed that the secretion is facilitated by YopE_1–15 (Figure S4A). Moreover, the SINGER system exhibited excellent reversibility in response to exposure and US withdrawal cycles at 12 h intervals (Figures 3C and 3D).

Figure 3.

Cytotoxic effects of azurin secreted by US-mediated VNP20009 cells against melanoma and colorectal cancer cells

(A) Schematic representation of SINGER-mediated protein expression and secretion in VNP20009 cells.

(B) The expression and secretion of a Gaussia luciferase (Gluc) reporter in VNP20009 cells after US stimulation.

(C) Schematic for the time schedule and experimental procedure for assessing the reversibility of the SINGER system.

(D) Reversibility of SINGER-mediated transgene expression. VNP20009 cells (OD₆₀₀ = 0.25) transformed with pGT240 were cultivated for 36 h while alternating irradiation with US (0.5 W/cm², pulse of 1 s on, 1 s off) for 1 h or without US. Gluc expression in the culture supernatants was quantified every 2 h. Cell density was adjusted to OD₆₀₀ = 0.25 when the pattern changed.

(E) Schematic of the experimental procedure for analysis of secretion proteins.

(F and G) The expression and secretion of azurin in VNP20009 cells after US stimulation. Azurin in the culture supernatants was checked by immunoblotting analysis (F) and ELISA (G).

(H) Measurement of cell viability by CCK-8 assays. B16F10 and CT26 cells were treated with the culture supernatants of VNP20009 harboring SINGER-azurin with (right) or without (left) US stimulation.

(I) Fluorescence microscopy images for the detection of apoptosis by TUNEL assay. Scale bar, 50 μm.

(J–L) Representative flow cytometry plots using annexin V-fluorescein isothiocyanate (FITC)/propidium iodide (PI) staining for apoptosis (J and K) and quantification of apoptosis (L).

Data in (B), (D), (G), (H), and (L) are presented as means ± SD; n = 3 independent experiments. Each data point represents the mean of three technical replicates. p values were calculated by one-way analysis of variance (ANOVA) with Tukey’s post-test. ∗∗∗p < 0.001.

See also Figures S2–S4.

Azurin is a protein initially identified in the pathogenic bacterium Pseudomonas aeruginosa capable of entering cancer cells and inducing apoptosis.⁴⁸ We applied the SINGER system to control the expression and secretion of a YopE_1–15-azurin fusion protein (termed “SINGER-azurin”) to induce the apoptosis of tumor cells. Initial immunoblotting tests and ELISAs confirmed that the culture supernatants of VNP20009 cells carrying SINGER-azurin, when exposed to US (0.5 W/cm² with a 1 s on, 1 s off pulse) for 1 h, contained approximately 321.7 ng/mL azurin protein (14 kDa). In contrast, no azurin was detected in the supernatants when the cells were not exposed to US (Figures 3E–3G). We further confirmed that the secretion of azurin is guided by YopE_1–15 (Figure S4B). Subsequent exposure of B16F10 melanoma and CT26 colon carcinoma tumor cell cultures to supernatants showed that the azurin-containing supernatants from cultures of VNP20009 cells carrying SINGER-azurin led to an increased extent of cell death compared to the supernatants of the control VNP20009 cells lacking SINGER-azurin (Figure 3H). TUNEL and annexin V-propidium iodide analyses showed that the azurin-containing supernatants induced apoptosis in tumor cells (Figures 3I–3L). Thus, US can induce the expression and secretion of the YopE_1–15-azurin fusion protein by engineered VNP20009 cells, and this protein can induce apoptosis in tumor cells.

Bacterial proliferation, US-triggered transgene expression, and tumor-suppressive effects in mice

To evaluate the proliferation ability of VNP20009 in tumor tissues, we stably integrated the LuxCDABE bioluminescent reporter cassette into the VNP20009 genome (VNP20009_LuxCDABE) as driven by a constitutive promoter (P_frr).⁴⁹ C57BL/6 mice bearing melanoma tumors (B16F10) were intratumorally injected with VNP20009_LuxCDABE at a dose of 5 × 10⁵ colony-forming units when the tumor volume was about 100 mm³ (Figure S5A). Bioluminescence imaging and colony counting showed that VNP20009_LuxCDABE cells proliferated in tumors. In fact, on day 17 after bacteria injection, the bacterial population in tumors increased to approximately 3 × 10⁹, which was an order of magnitude higher than that in tumors on day 1 (Figures S5B and S5C).

We then examined the functionality of the SINGER system for gene activation in mice. C57BL/6 mice bearing B16F10 melanoma tumors were injected intratumorally with VNP20009 carrying SINGER-LuxCDABE, and the tumors were exposed to US (0.5 W/cm², pulse of 1 s on, 1 s off) for different time periods (Figure 4A). During US irradiation, an infrared thermal imager was used to monitor the temperature of the tumor site, which was stabilized at about 39°C (Figure S6). The bioluminescence signal revealed that US-inducible transgene expression mediated by SINGER in mice was irradiation-time dependent (Figures 4B and 4C). Notably, the expression level in vivo continued to rise for approximately 10 h following the cessation of US irradiation (Figure S7). Moreover, we also intratumorally injected SINGER-azurin cells into mice bearing B16F10 melanoma tumors and examined the in vivo expression and release of azurin induced by US. The expression and secretion of azurin in tumors post-US irradiation were confirmed by immunoblotting and immunofluorescence staining; however, azurin was not detected in tumors that were not exposed to US (Figures 4D and 4E).

Figure 4.

SINGER-mediated transgene expression in a tumor xenograft mouse model

(A) Schematic illustration of the experimental design for evaluating SINGER-mediated transgene expression in tumors.

(B and C) Time-course-dependent SINGER-mediated transgene expression kinetics in tumors revealed by bioluminescence images in mice. C57BL/6 mice bearing B16F10 melanoma tumors received an intratumoral injection of VNP20009 cells (5 × 10⁵ colony-forming units [CFUs]) transformed with pGT240, after which the mice were exposed to US (0.5 W/cm², pulse with 1 s on, 1 s off) for the indicated durations. The bioluminescence signal intensity was quantified with an in vivo imaging system 12 h after US irradiation. The data in (C) are presented as means ± SEM; n = 4 mice. p values were calculated by one-way ANOVA with Tukey’s post-test. ∗∗p < 0.01 and ∗∗∗p < 0.001.

(D) Immunoblotting for azurin levels in tumor tissues.

(E) Immunofluorescence staining of tumor sections. The azurin expressed and secreted by US-mediated VNP20009 cells was stained with an anti-FLAG antibody (red). Bacterial cells were stained with an anti-DnaK antibody (green). Nuclei were stained with DAPI (blue). Azurin-FLAG and DnaK were detected by fluorescence confocal microscopy. Scale bar, 10 μm.

See also Figures S5–S7.

Having shown that US can induce SINGER-mediated azurin expression in tumor tissues, we next assessed the potential antitumor impacts of tumor-resident azurin. Mice bearing B16F10_luci melanoma tumors were injected intratumorally with PBS, untransformed VNP20009 cells, or VNP20009 cells carrying SINGER-azurin. Twenty-four hours after bacteria injection, the tumors received US irradiation (0.5 W/cm², pulse of 1 s on, 1 s off) for 1 h every 2 days (Figure 5A). Significant therapeutic benefits were observed for the mice treated with VNP20009 cells harboring SINGER-azurin with US induction (SINGER-azurin+US); indeed, the mice that received SINGER-azurin and US irradiation achieved complete eradication of tumors (Figures 5B–5F), with all mice surviving to the pre-defined endpoint of 50 days (Figure 5E). Although treatment with untransformed VNP20009 cells (VNP20009), untransformed VNP20009 cells with US (VNP20009+US), and VNP20009 cells harboring SINGER-azurin without US (SINGER-azurin) resulted in delayed tumor growth compared to PBS controls, presumably due to the activation of innate immune by bacteria, the final tumor volumes and survival rates were not statistically different from PBS-treated mice (Figures 5C–5E). These data suggested that SINGER-mediated azurin production by VNP20009 within the tumor elicits antitumor activity.

Figure 5.

Antitumor performance of VNP20009 cells harboring SINGER-azurin in mice bearing B16F10_luci melanoma tumors

(A) Schematic illustrating the experimental design for evaluating the antitumor performance of VNP20009 cells harboring SINGER-azurin against B16F10_luci melanoma tumors. C57BL/6 mice bearing B16F10_luci melanoma tumors received intratumoral injection of PBS alone, US alone (US), untransformed VNP20009 cells (VNP20009), untransformed VNP20009 cells with US (VNP20009+US), VNP20009 cells harboring SINGER-azurin with US induction (SINGER-azurin+US), or VNP20009 cells harboring SINGER-azurin without US induction (SINGER-azurin).

(B) Serial in vivo bioluminescence imaging of B16F10_luci-tumor-bearing mice in the groups depicted in (A). Five representative mice per treatment group are shown.

(C) Bioluminescence quantification of B16F10_luci-tumor-bearing mice treated with different treatments.

(D) Tumor volume measurement in the indicated mouse groups.

(E) Kaplan-Meier curves for mouse survival.

(F) Individual growth curves of B16F10_luci melanoma tumors in different mouse groups.

Data in (C)–(E) are presented as means ± SEM; n = 5 mice per group. p values were calculated by one-way ANOVA with Tukey’s post-test (∗∗∗∗p < 0.0001).

US-mediated tumor suppression by SINGER-azurin VNP20009 cells delivered intravenously

We then assessed the therapeutic utility of SINGER-azurin in tumor xenograft mouse models through systemic intravenous administration of the therapeutic cells. First, the specific targeting and selective colonization of intravenously injected VNP20009 cells on tumors was verified. VNP20009_LuxCDABE cells were delivered into C57BL/6 mice bearing melanoma tumors (B16F10) through tail vein injection, and bioluminescence imaging of mice was performed at various time points after bacterial injection. Moreover, tumors, hearts, livers, spleens, lungs, and kidneys were harvested, homogenized, and plated on Luria-Bertani (LB) medium for colony counting (Figure S8A). No bacterial colonization was detected in the hearts, livers, spleens, lungs, or kidneys 48 h after injection (Figures S8B–S8F). Through in vivo bioluminescence imaging and colony counting in tumor tissues and five main organs, VNP20009_LuxCDABE was found to preferentially colonize and proliferate in tumors within 3 days (Figures S8G–S8I). Thus, the engineered bacteria can specifically colonize tumors upon systemic intravenous delivery.

Intravenous delivery of the bacteria would be desirable as it would enhance their applicability to tumors of unknown location. C57BL/6 mice bearing B16F10_luci tumors were intravenously injected with PBS, VNP20009 cells, or VNP20009 cells harboring SINGER-azurin; 3 days after bacteria injection, the tumors received US irradiation (0.5 W/cm², pulse of 1 s, on 1 s off) for 1 h every 2 days (Figure 6A). Significant therapeutic effects were observed for the SINGER-azurin+US group. Compared to each of the control groups (mice treated with PBS, US, VNP20009, VNP20009+US, or SINGER-azurin), the SINGER-azurin+US group mice achieved superior tumor clearance and prolonged survival (Figures 6B–6E and S9), indicating that engineered VNP20009 cells colonized tumors and secreted azurin in situ. We evaluated the levels of azurin in the tumors and surrounding normal tissues (SNTs) through immunoblotting analysis, which demonstrated negligible leakage of the therapeutic protein azurin into the SNTs (Figure S10). These findings suggest that SINGER-mediated therapy specifically targets tumor cells without causing significant off-target effects on the surrounding healthy tissues. Given the reported mechanisms of azurin-mediated cytotoxicity and our results from the assays mentioned above with B16F10 and CT26 cells, we collected tumor specimens 7 days after treatment to examine apoptosis and proliferation in tumor cells. Compared to the mice treated with PBS, the SINGER-azurin+US group tumors showed increasing apoptosis and significant antiproliferative activity (Figure S11).

Figure 6.

SINGER-mediated antitumor performance of engineered VNP20009 delivered via intravenous injection in different tumor xenograft mouse models

(A) Schematic illustration of SINGER-mediated antitumor performance against B16F10_luci melanoma tumors. C57BL/6 mice bearing B16F10_luci melanoma tumors received intravenous injections of PBS, VNP20009, US, VNP20009+US, SINGER-azurin, or SINGER-azurin+US.

(B) Serial in vivo bioluminescence imaging of B16F10_luci tumors expressing luciferase in the groups described in (A).

(C) Bioluminescence quantification of B16F10_luci-tumor-bearing mice treated with different treatments.

(D) Tumor volume measurement in the indicated mouse groups.

(E) Kaplan-Meier curves for mouse survival in (B).

(F–H) Tumor volume measurement in mice bearing various xenograft tumors, including CT26 (F), A20 (G), and H22 tumors (H).

(I–K) Kaplan-Meier curves for mouse survival in (F)–(H).

Data in (C)–(K) are presented as means ± SEM; n = 4–6 mice. p values were calculated by one-way ANOVA with Tukey’s post-test (∗∗∗∗p < 0.0001).

See also Figures S8–S11.

To assess the broader applicability of US-mediated tumor suppression by engineered VNP20009 cells, we also examined the antitumor efficacy of SINGER-azurin by intravenous administration in additional mouse tumor models. Mice bearing other syngeneic tumors, including murine CT26 colorectal carcinoma, murine A20 B cell lymphoma, and murine H22 hepatocellular carcinoma, received a single treatment with different formulations. All mice treated with SINGER-azurin+US displayed significant therapeutic efficacy, with tumors partially or completely regressing, and survived significantly longer than mice in the three control groups (Figures 6F–6K). These results indicate that the SINGER-azurin system can confer antitumor effects against various tumors in a US-dependent manner.

US-mediated immunological synergism for tumor clearance

Although promising, complete ablation of tumors is exceedingly difficult through US-activated apoptosis owing to the presence of any residual tumor mass at the untreated margins.⁵⁰ We developed an improved cancer therapy combining SINGER-mediated tumor apoptosis and immunotherapy for enhanced antitumor efficacy. Specifically, we introduced a nanobody targeting programmed death-ligand 1 (PD-L1 nb) into the SINGER system to create SINGER-PD-L1. This PD-L1 nb can reactivate T cell-mediated antitumor immunity by inhibiting the interaction between PD-L1 on tumor cells and PD-1. Immunoblotting analysis revealed that the PD-L1 nb, when guided by YopE_1–15, is secreted into the culture supernatants of VNP20009 cells carrying SINGER-PD-L1 after 1 h of exposure to US (0.5 W/cm² with a 1 s on, 1 s off pulse pattern). In contrast, the PD-L1 nb was not detectable when US or YopE_1–15 was absent (Figures S12A–S12C). To validate the binding of the PD-L1 nb to PD-L1, we conducted in vitro experiments. H22 cells were coincubated with the culture supernatants of VNP20009 cells carrying SINGER-PD-L1, either irradiated or non-irradiated with US, and a fluorescently conjugated anti-PD-L1 monoclonal antibody (mAb) that specifically targets epitopes recognized by the 10F.9G2 antibody on H22 cells. Flow cytometry analysis demonstrated a reduction in fluorescence intensity of the 10F.9G2 mAb upon adding supernatants containing the PD-L1 nb (Figure S12D).

Subsequently, we evaluated the therapeutic efficacy of combining SINGER in tumor xenograft mouse models through systemic intravenous administration of the therapeutic cells. C57BL/6 mice bearing H22 liver tumors, which responded poorly to the SINGER-azurin+US treatment, were intravenously injected with PBS, VNP20009, SINGER-azurin, SINGER-PD-L1, or an equal mixture of SINGER-azurin and SINGER-PD-L1 (SINGER-2). Three days after bacterial injection, the tumor received US irradiation (0.5 W/cm², pulse of 1 s on, 1 s off) for 1 h every 2 days (Figure 7A). Immunoblotting analysis verified the successful production of the PD-L1 nb in mouse tumors treated with VNP20009 cells equipped with SINGER-PD-L1 following US exposure (Figure S12E). We found that the tumor growth was completely inhibited in mice treated with SINGER-2 and US induction (SINGER-2+US), which achieved 100% survival within 50 days (Figures 7B, 7C, and S13). Moreover, significantly less lymph node metastasis was observed in mice treated with SINGER-2+US compared to other groups (Figure 7D). To assess the effects of azurin and the PD-L1 nb on tumor suppression, we administered purified azurin or PD-L1 nb protein to H22 tumor model mice via intraperitoneal injection. The outcomes revealed that the individual administration of either azurin or the PD-L1 nb did not significantly inhibit tumor growth or improve the survival rates of the mice (Figure S14). This demonstrates that the therapeutic efficacy of azurin or the PD-L1 nb, when induced by US in our engineered bacteria, substantially exceeds the effects of administering these proteins directly.

Figure 7.

SINGER-mediated immune activation for clearance of tumors

(A) Schematic illustration of SINGER-mediated antitumor therapy. C57BL/6 mice bearing H22 tumors received an intravenous injection of PBS, VNP20009, US, VNP20009+US, SINGER-azurin+US, SINGER-PD-L1+US, an equal mixture of SINGER-azurin and SINGER-PD-L1 (SINGER-2), or SINGER-2+US.

(B) Tumor volume measurement in the indicated mouse groups.

(C) Kaplan-Meier curves for mouse survival in (B).

(D) The number of visible metastatic nodules in each group.

(E) Schematic diagram illustrating the experimental procedure for the analysis of immune cells as well as cytokines using flow cytometry and ELISAs. C57BL/6 mice bearing H22 tumors were treated with PBS, VNP20009+US, SINGER-azurin+US, SINGER-PD-L1+US, SINGER-2, and SINGER-2+US.

(F–L) Splenocytes isolated from tumor-bearing mice on day 11 after the indicated treatments were analyzed by flow cytometry for the presence of CD8⁺ T cells (F), IFNγ⁺CD8⁺ T cells (G), IFNγ⁺Foxp3⁻CD4⁺ T cells (H), Foxp3⁺CD4⁺ T cells (I), LAG3⁺CD8⁺ T cells (J), TIM3⁺CD8⁺ T cells (K), and NK cells (L).

(M and N) Tumor tissues isolated from tumor-bearing mice on day 11 after the indicated treatments were analyzed by ELISA for the levels of TNF-α (M) and IL-10 (N).

Data in (B)–(D) and (F)–(N) are represented as means ± SEM; n = 4–6 mice. p values were calculated by one-way ANOVA with Tukey’s post-test (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001, n.s., no significance).

See also Figures S12–S23.

To further understand the immune mechanisms underlying tumor inhibition, we analyzed immune responses in spleens and tumor-draining lymph nodes (TDLNs) (Figure 7E). Compared to the other four groups, the mice treated with SINGER-PD-L1+US and SINGER-2+US had a significantly increased population of CD3⁺CD8⁺ T cells (Figures 7F and S15A) and an increased population of cytotoxic CD8⁺ T cells (interferon [IFN]γ⁺CD8⁺ and IFNγ⁺Foxp3⁻CD4⁺) (Figures 7G, 7H, S15B, and S15C) but a significant decrease of regulatory T cells (Tregs; Foxp3⁺CD4⁺) and exhausted T cells (LAG3⁺CD8⁺ and TIM3⁺CD8⁺) in spleens (Figures 7I–7K and S15D–S15F). Further analysis found a significant increase of natural killer (NK) cells in spleens (Figures 7L and S15G). We also observed an increased population of CD3⁺CD8⁺ T cells (Figures S16A and S17A) and cytotoxic CD8⁺ T cells (IFNγ⁺CD8⁺) (Figures S16B and S17B), a significant decrease of Tregs (Foxp3⁺CD4⁺) and exhausted T cells (LAG3⁺CD8⁺) (Figures S16C, S16D, S17C, and S17D), and a significant increase of NK cells in TDLNs (Figures S16E and S17E). It is worth noting that the SINGER-azurin+US treatment also elicited certain immune responses in comparison with the PBS group, which are assumed to be induced by the inherent immune activity of the VNP20009 and the release of tumor cell contents following tumor apoptosis. We also showed the aforementioned gating strategy that we used to compare T cells and NK cells in different treatments (Figure S18). Moreover, we detected the cytokine levels in tumors, including tumor necrosis factor α (TNF-α), IFN-γ, interleukin (IL)-10, and transforming growth factor β (TGF-β). Our findings demonstrate a significant increase in the levels of the proinflammatory cytokines TNF-α and IFN-γ in tumors treated with SINGER-2+US compared to those receiving other treatments, suggesting an enhanced inflammatory response conducive to antitumor activity (Figures 7M and S19A). Conversely, the anti-inflammatory cytokine IL-10 and TGF-β levels were lower in the SINGER-2+US group, indicating a potential shift toward a more proinflammatory tumor microenvironment (Figures 7N and S19B). Thus, US-mediated azurin and PD-L1 nb production by VNP20009 effectively induced a durable therapeutic response by activating adaptive immunity synergism apoptosis.

Activating a robust immune response is anticipated to generate immune memory, potentially preventing tumor metastasis. Consequently, we explored the possibility of SINGER-2+US inducing such an immune memory. Our study utilized a distant H22 tumor model in mice with tumors on both flanks. The mice were intravenously injected with PBS, VNP20009, SINGER-azurin, SINGER-PD-L1, or SINGER-2. Three days post-injection, only one tumor per mouse was exposed to US irradiation at 0.5 W/cm² in a pulsed mode (1 s on, 1 s off) for an hour every two days. We monitored the growth of both treated and untreated tumors (Figure S20A). Consistent with earlier findings, the SINGER-2+US treatment significantly decreased the tumor volume of the directly treated tumor (Figure S20B). Intriguingly, the untreated tumor on the opposite flank also showed a significant reduction in tumor volume following the treatment (Figure S20C), indicating the potential presence of an abscopal effect.

We additionally evaluated the safety of the engineered bacteria during tumor therapy. Firstly, we assessed the acute inflammatory response in mice by analyzing the levels of inflammation-related cytokines, IL-6 and TNF-α, in the serum following intravenous injection of SINGER-2. Compared to the PBS group, serum IL-6 and TNF-α levels increased 3 h after the injection of SINGER-2, which subsequently returned to normal levels within 24 h (Figure S21). This indicates that the acute inflammation induced by the engineered bacteria was moderate and well tolerated by the mice with no signs of chronic toxicity, as the inflammation resolved within a day. Further, C57BL/6 mice bearing H22 tumors were intravenously injected with PBS, VNP20009, or SINGER-2; 3 days after bacterial injection, the tumors received US irradiation (0.5 W/cm², pulse of 1 s on, 1 s off) for 1 h every 2 days. After 14 days of treatment, blood samples were collected for routine examination and serum biochemical analyses, and the mice were euthanized for harvest of organs for histological analysis. Staining of histological sections revealed that the mice given the SINGER-2+US treatment had no pathological abnormalities in the hearts, livers, spleens, lungs, or kidneys (Figure S22). Further, a comprehensive assessment of serum biochemistry indexes and complete blood panels was performed (Figure S23). The results indicated the absence of inflammation and renal or liver dysfunction, supporting the fact that the engineered bacteria and US irradiation were well tolerated by mice.

Discussion

We established a US-based sonogenetic platform allowing remote and non-invasive control of gene expression in engineered bacteria for therapeutic applications. The US-controlled engineered bacteria could colonize the hypoxic and immune-privileged core of the tumor, where they locally expressed and secreted therapeutic proteins without causing non-specific off-tumor toxicity. Specifically, our findings indicate that the fusion of the YopE N-terminal secretion sequence to heterologous proteins facilitates their release, potentially through the Salmonella flagellar type III secretion system. The SINGER system, with high sensitivity, modularity, and compatibility, allows rapid engineering of multiple payloads in response to US irradiation. We have shown that a single dose of engineered bacteria harboring SINGER-azurin delivered intratumorally or intravenously resulted in substantial tumor suppression in diverse mouse tumor models under moderate intensity of US stimulation with the short-pulsed patterns (0.5 W/cm², pulse of 1 s on, 1 s off) for 1 h every other day. Moreover, we have also explored the therapeutic combination of bacterial-delivered azurin and PD-L1 nb. Although our study may not reveal unique aspects of the immune mechanism beyond previous research, it distinctively integrates immune activation with apoptosis induction. This approach has been proven to yield a lasting therapeutic effect by significantly reducing tumor cell growth. The dual strategy not only enhances the suppression of tumor growth but also provides a more effective and comprehensive treatment approach. Consequently, future versions of the SINGER system are anticipated to facilitate the exploration of various cancer treatments through testing multiple therapeutic strategies.

In recent years, US-based cell therapies have been moving into the spotlight owing to the non-invasiveness, spatial-temporal specificity, and strong tissue-penetrability properties of US. Previous studies have applied US to control the expression of the CAR protein in mammalian cells based on either the mechanically activated cation channels (Piezo1)³⁶ or the heat shock protein (HSP) promoters.³⁷ However, Piezo1 requires microbubbles for activation, limiting in vivo applications, while HSP systems can be non-specifically activated by various stress factors. The SINGER system based on transcription factor TlpA₃₉ responding to US-triggered thermal stimulus shows precise regulatable control of transgene expression with high flexibility and orthogonality. It is worth mentioning that in vivo biosafety is still a concern for the clinical implementation of bacteria-based therapies. In our study, long-term treatment with SINGER-mediated engineered bacteria caused no significant tissue damage to or adverse effects in mice, benefiting from a multifaceted design: compared to existing sonogenetic tools, the SINGER system based on TlpA₃₉ is of higher sensitivity, which can be activated by conventional US-induced hyperthermia (39°C), minimizing thermal damage to healthy tissue.^39,51 By promoter engineering and RBS screening, the SINGER system was optimized with a wide dynamic regulation range, reducing the hypothetical adverse effects caused by background leakage. Moreover, VNP20009, which was selected as a carrier of the SINGER system for cancer therapy, has been widely used in pre-clinical and clinical studies^52,53 and could selectively colonize and proliferate at the hypoxic region of tumors upon systemic administration,⁴⁰ making it possible to deliver therapeutic cargos upon US irradiation locally.

While US offers several advantages for therapeutic applications, its application has inherent challenges, particularly with deep-seated or metastatic tumors. Its effectiveness in gene expression is hindered by tumor depth and spread requiring more precise heat transfer methods. High-intensity focused US and external beam radiations are potential solutions for more targeted therapy. Ongoing research and integration with other treatments are vital to maximizing the potential of SINGER in complex cancer treatments, underscoring its promise as a therapeutic modality but also highlighting the need for technological advancements and strategic treatment combinations.

Although we demonstrated the significant antitumor efficacy of the SINGER-mediated engineered bacteria, there are limitations in the current bacteria-based therapy for future clinical use. While we used an equal mixture of SINGER-azurin and SINGER-PD-L1 for tumor therapy, we did not investigate varying ratios of these two therapeutic agents. The SINGER platform’s flexibility to test different combinations and ratios of agents is likely an advantage allowing for the exploration and optimization of treatment efficacy by adjusting the proportions of these agents. To enable future optimization, we intentionally avoided combining both therapeutic outputs (azurin and the PD-L1 nb) in a single engineered bacterial cell. Further, we used strains transiently transformed with plasmids instead of strains with therapeutic genes integrated into the genome. This approach might limit the treatment’s effective duration due to the potential loss of gene circuit components.

Our engineered bacteria, equipped with the SINGER system in which US-triggered thermal stimulus activates the local expression and release of therapeutic protein into tumor regions, have shown excellent anticancer effects in diverse mouse tumor models. Moreover, the SINGER system is highly modular and applicable to the rapid engineering of multiple outputs in various bacteria, potentially treating numerous diseases beyond tumors. For instance, the US-mediated SINGER system could be deployed to locally produce hormones, enzymes, or anti-inflammatory agents by engineered gut microbes for alleviating metabolic disorders and intestinal inflammation. Further, a promising application of this system is the development of targeted therapies for infections or autoimmune diseases where fever is a primary symptom. Engineered bacteria could be programmed to produce and release therapeutic compounds, such as anti-inflammatory cytokines or antimicrobial peptides, in response to fever-associated elevated body temperatures, often within the 39°C–40°C range. Nevertheless, adapting the SINGER system for diseases beyond tumors presents several challenges, particularly regarding the targeted delivery of bacteria in the body and minimizing in vivo toxicity. Optimization of bacterial platforms and the development of portable US devices are vital areas for advancing this technology for personalized, non-invasive treatments. We anticipate that our US-based sonogenetic platform will provide broad application scenarios and usher in an era of genetically engineered bacteria-based therapies controlled by US precisely and non-invasively.

Limitations of the study

Our research has not yet established the efficacy of our engineered bacteria in treating deep-seated in situ tumors in mouse models. The performance of our US-based system is constrained by the depth and extent of tumors, with deep-seated or metastatic tumors presenting significant challenges for activating gene expression via US. This may require more refined methods of heat transfer or combinations with other therapeutic modalities, such as high-intensity focused US or external beam radiation, to achieve targeted treatment. Moreover, our study utilized a fixed ratio of SINGER-azurin and SINGER-PD-L1 in tumor therapy without investigating the potential benefits of varying these ratios. Additionally, our methodology involved the use of strains transiently transformed with plasmids as opposed to integrating gene circuits directly into the bacterial genome. This approach may shorten the effective treatment period due to the possible loss of gene circuit elements over time, highlighting a key area for future improvement.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-His primary antibody	YEASEN	Cat#30401ES10
anti-Flag primary antibody	Proteintech	Cat#80010-1-RR; 80010-1-RR
HRP-conjugated anti-rabbit secondary antibody	Proteintech	Cat#SA00001-2; SA00001-2
rabbit polyclonal anti-Ki67	Proteintech	Cat#27309-1-AP; RRID:AB_2756525
Alexa Fluor 546-conjugated secondary antibodies	Proteintech	Cat#SA00013-4; RRID:AB_2810984
CY3-labelled anti-rabbit IgG (H + L) antibody	Servicebio	Cat#GB21303;RRID:AB_2861435
AF488-labelled anti-mouse IgG(H + L)	Servicebio	Cat#GB25301; RRID:AB_2904018
PE/Cyanine7 anti-mouse CD45	Biolenged	Cat#103114; RRID:AB_312979
FITC anti-mouse CD3ε	Biolenged	Cat#100306; RRID:AB_312671
APC anti-mouse CD8a	Biolenged	Cat#100712; RRID:AB_312750
Brilliant Violet 510™ anti-mouse CD8a	Biolenged	Cat#100751; RRID:AB_2561389
PerCP/Cyanine5.5 anti-mouse CD4	Biolenged	Cat#100434; RRID:AB_893330
PE anti-mouse NK-1.1	Biolenged	Cat#108708; RRID:AB_313395
Brilliant Violet 605™ anti-mouse CD366	Biolenged	Cat#119721; RRID:AB_2616907
Brilliant Violet 421™ anti-mouse CD223	Biolenged	Cat#125221; RRID:AB_2572080
Brilliant Violet 421™ anti-mouse Foxp3	Biolenged	Cat#126419; RRID:AB_2565933
PE anti-mouse IFN-γ	Biolenged	Cat#505808; RRID:AB_312750
Bacterial and virus strains
E. coli DH5α	Sangon Biotech	Cat#B528413
E. coli TOP10	Sangon Biotech	Cat#B528412
E. coli BL21 (DE3)	Sangon Biotech	Cat#B528414
E. coli Nissle 1917	Shenzhen Institutes of Advanced Technology	N/A
S. typhimurium VNP20009	Nanchang University	N/A
S. enteritidis 3934	Universidad P $\acute{u}$ blica de Navarra-CSIC-Gobierno de Navarra	N/A
Chemicals, peptides, and recombinant proteins
RIPA lysis buffer	YEASEN	Cat#20101ES60
Paraformaldehyde	Servicebio	Cat#G1101
LB Broth Powder, Miller	Sangon Biotech	Cat#A507002
LB Agar Powder, Miller	Sangon Biotech	Cat#A507003
Ampicillin sodium	Sangon Biotech	Cat#A610029
Tween 20	Sangon Biotech	Cat#9005-64-5
Sterile PBS	Sangon Biotech	Cat#A610100
D-Luciferin solution	Shanghai Science light Biology Science & Technology	Cat#uc001
Isoflurane	RWD Life Science	Cat#R510-22-10
DAPI	Vector Laboratories	Cat#H-1200-10
Dulbecco’s modified Eagle’s medium (DMEM)	Gibco	Cat#12100061
Roswell Park Memorial Institute-1640 medium (RPMI-1640)	Gibco	Cat#11875093
Fetal bovine serum (FBS)	Gibco	Cat#16000-044
0.25% Trypsin-EDTA (1X), Phenol Red	Gibco	Cat#25200072
Penicillin/streptomycin	Beyotime	Cat#ST488-1/ST488-2
RBC lysis buffer	Biolenged	Cat#420301
Critical commercial assays
SYTO9/PI Live/Dead Bacterial Double Stain Kit	MKbio	Cat#MX4234-40T
Mouse IL-6 ELISA Kit	Multi Sciences	Cat#EK206
Mouse TNF-α ELISA Kit	Multi Sciences	Cat#EK282
Mouse TGF-β ELISA Kit	Multi Sciences	Cat#EK981
Mouse IL-10 ELISA Kit	Multi Sciences	Cat#EK310HS
Mouse IFN-γ ELISA Kit	Multi Sciences	Cat#EK280HS
MultiS One Step Cloning Kit	Vazyme	Cat#C113-01
Cell counting Kit-8	Beyotime	Cat#C0037
Annexin V-FITC/PI double-staining apoptosis detection kit	Beyotime	Cat#C1062M
BCA Protein Assay Kit	Beyotime	Cat#P0010
One Step TUNEL Apoptosis Assay Kit	Beyotime	Cat#C1086
Zombie NIR™ Fixable Viability Kit	Biolenged	Cat#423106
Foxp3 Transcription Factor Staining Buffer kit	eBioscience	Cat#00-5523-00
Experimental models: Cell lines
Mouse: B16F10	Meisen CTCC	Cat#CTCC-400-0304
Mouse: B16F10luci	Meisen CTCC	Cat#CTCC-0488-Luc1
Mouse: H22	Bluefbio Inc.	Cat#BFN608007276
Mouse: CT26	ATCC	Cat#CRL-2638
Mouse: A20	ATCC	Cat#TIB-208
Mouse: ARPE-19	ATCC	Cat#CRL-2302
Human: HEK-293T	ATCC	Cat#CRL-11268
Human: hMSC-TERT	ATCC	Cat#SCRC-4000
Human: NIH-3T3	ATCC	Cat#CRL-1658
Oligonucleotides
DNA primers sequences, see Table S4	This paper	N/A
Other DNA sequences, see Tables S1–S3	This paper	N/A
Recombinant DNA
Plasmids generated in this study, see Table S4	This paper	N/A
Experimental models: Organisms/strains
Mouse: C57/BL6	East China Normal University Laboratory Animal Center	N/A
Mouse: BALB/c	East China Normal University Laboratory Animal Center	N/A
Software and algorithms
GraphPad Prism 8	GraphPad	http://www.graphpad.com/
Snap Gene	Dotmatics	https://www.snapgene.com/
Gen5 v2.04.	Biotek	https://www.biotekmilano.com/
Living Image v4.3.1	Perkin Elmer	https://www.perkinelmer.com/
Fluorescent western blot imaging system	LI-COR Odyssey Biosciences	https://www.licor.com/
FlowJo v.10.6.1	FlowJo	https://www.biorender.com/
LAS X	Leica	https://www.leica-microsystems.com/products/microscopesoftware/p/leica-las-x-ls/downloads/

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Haifeng Ye (hfye@bio.ecnu.edu.cn).

Materials availability

All engineered VNP20009 cells and the SINGER system components generated in this study will be made available on request with a completed Materials Transfer Agreement, but a fee may be applicable to cover the costs of material preparation and shipping. All requests for resources and reagents should be directed to the lead contact author.

Data and code availability

• •The data that support the findings of this study are available from the corresponding author upon reasonable request.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Experimental model and study participant details

Bacterial strains

The Escherichia coli K-12 strains TOP10 and DH5α, the E. coli BL21 (DE3), the attenuated Salmonella typhimurium strain VNP20009, the Salmonella enteritidis strain 3934, and the probiotic E. coli Nissle 1917 were used in this study. S. typhimurium VNP20009 was kindly provided as a gift from Tingtao Chen’s lab (Nanchang University). S. enteritidis 3934 was kindly provided as a gift from I $\tilde{n}$ igo Lasa’s lab (Universidad P $\acute{u}$ blica de Navarra-CSIC-Gobierno de Navarra). The other bacteria strains were preserved by our lab. These strains were cultured in lysogeny broth (LB) medium (10 g/L sodium chloride, 10 g/L tryptone, and 5 g/L yeast extract).

Mammalian cells

B16F10_luci (Luciferase tagged mouse B16F10 melanoma cell line), B16F10 melanoma cell line (Meisen CTCC) and mouse colon carcinoma cell line (CRL-2638, ATCC) were cultured in RPMI-1640 medium (11875093, Gibco) with 10% (v/v) fetal bovine serum (FBS; 16000-044, Gibco) and 1% (v/v) penicillin/streptomycin (ST488-1/ST488-2, Beyotime). The murine A20 B-cell lymphoma cell line (TIB-208, ATCC) was cultured in GI 1640 medium (11875093, Gibco) with 10% (v/v) fetal bovine serum (FBS; 16000-044, Gibco) and 1% (v/v) penicillin/streptomycin (ST488-1/ST488-2, Beyotime) plus HEPES (25 mM) and β-mercaptoethanol (50 μM). The murine H22 hepatocellular carcinoma cell line (BFN608007276, Bluefbio Inc.), human embryonic kidney cells (HEK-293T, CRL-11268, ATCC), telomerase-immortalized human mesenchymal stem cells (hMSC-TERT, SCRC-4000, ATCC), mouse fibroblast NIH3T3cell line (CRL-1658, ATCC), and adult retinal pigment epithelial cells (ARPE-19, CRL-2302, ATCC) were cultured in Dulbecco’s modified Eagle’s medium (DMEM; C11995500BT, Gibco) supplemented with 10% (v/v) fetal bovine serum and 1% (v/v) penicillin/streptomycin solution. All mammalian cells were cultured at 37°C in a humidified atmosphere containing 5% CO_2. The concentration and viability of the cell lines were evaluated using a Countess II Automated cell counter (AMEP4746, Life Technologies).

Animals

6-week-old C57BL/6 male mice and 6-week-old BALB/c male mice were obtained from the ECNU Laboratory Animal Center and maintained in a climate controlled sterile environment. Animals were housed under pathogen-free conditions with food and water ad libitum. All experiments and procedures were performed following the protocol m20211008 and m20220404 approved by the East China Normal University (ECNU) Animal Care and Use Committee and were in direct accordance with the Ministry of Science and Technology of the People’s Republic of China on Animal Care guidelines. All researchers involved in animal experiments complied with relevant animal-use guidelines and ethical regulations during this study. All mice were euthanized after termination of the experiments.

Method details

Plasmids construction

All plasmids constructed for this study (Table S4) were made using Gibson Assembly or restriction enzyme-mediated cloning methods. The constructed plasmid sequences were verified by Sanger sequencing (Genewiz). Constructed plasmids were transformed into E. coli and Salmonella through electroporation. Amino acid sequences of proteins used in this study were listed in Table S5.

SYTO-9/PI staining assay

The viability of bacteria was assessed using the SYTO9/PI live/dead bacterial double stain kit (MX4234-40T, Maokangbio) according to the manufacturer’s instructions. In brief, VNP20009 cells were grown to OD₆₀₀ = 0.25 and irradiated by US for 1 h. SYTO-9 and PI were combined in an equal volume ratio to create the staining solution, which was then added to bacterial suspension (3 μL/mL). The mixtures were incubated for 15 min in the dark at room temperature. Post-incubation, the bacteria were examined under a confocal laser scanning microscopy (SP8 LIGHTNING, Leica, Germany) with 1000× magnification. Bacteria subjected to a thermal treatment at 65°C for 1 h were used as a positive control.

US regulation in vitro

For US stimulation of bacterial cultures in vitro, VNP20009 harboring SINGER system was propagated in LB medium with 100 μg/mL ampicillin overnight at 37°C, 210 rpm. After dilution to OD₆₀₀ = 0.1, the cells were propagated at 37°C, 210 rpm, until reaching an OD₆₀₀ = 0.25. The cultures were transferred to sterile flasks, and the pre-sterilized ultrasonic probe (delta 07 FND, Delta) installed in the flask lid was positioned above it and in direct contact with the culture. A fiber optic thermometer (HT9815, XINTEST) was inserted into the culture and monitored the temperature of the cultures at the site of insonation.

Protein secretion assay

Engineered bacterial strains were cultured in LB medium with 100 μg/mL ampicillin at 37°C, 210 rpm for 12 h. After dilution to OD₆₀₀ = 0.1, the cells were propagated at 37°C, 210 rpm, until reaching an OD₆₀₀ = 0.25. The cultures received US stimulation (0.5 W/cm², pulse of 1 s on, 1 s off), and the temperature maintained at 39°C was verified by a fiber optic thermometer.

To verify the secretion of Gluc by engineered VNP20009, the samples of VNP20009 cells transformed with pGT198 (P_TlpA1-RBS3-YopE_1-15-Gluc-P_j23100-TlpA₃₉) or pGT309 (P_TlpA1-RBS3-Gluc-P_j23100-TlpA₃₉) with US irradiation were centrifuged at 12,000 × g for 5 min, and 10 μL supernatants was transferred to a 384-well assay plate (Costar black/clear bottom) after filtration through a 0.2-mm pore size filter. Gluc luminescence was then triggered by adding 10 μL Gluc buffer and measured immediately using a Synergy H1 hybrid multimode microplate reader (BioTek Instruments Inc.) with Gen5 software (version 2.04).

To verify the secretion of Azurin-Flag or PD-L1 nb-His by engineered VNP20009, the samples of VNP20009 cells transformed with pGT252 (P_TlpA1-RBS3-YopE_1-15-Azurin-Flag-P_j23100-TlpA₃₉), pGT307 (P_TlpA1-RBS3-Azurin-Flag-P_j23100-TlpA₃₉), pGT305 (P_TlpA1-RBS3-YopE_1-15-PD-L1 nb-His-P_j23100-TlpA₃₉), or pGT308 (P_TlpA1-RBS3-PD-L1 nb-His-P_j23100-TlpA₃₉) after prescribed US stimulus was centrifuged and filtered. Then, the proteins released into the culture supernatants were extracted by methanol/chloroform protein precipitation. One volume of supernatant samples was mixed with four volumes of methanol, two volumes of chloroform, and three volumes of water, successively. After vortexed, the mixtures were centrifuged at 10,000 × g for 10 min at room temperature. The top aqueous layer was pipetted off while keeping the white precipitate carefully. Then, another three volumes of methanol were added and vortexed gently. Finally, the proteins were precipitated as pellets through centrifugation at 10,000 × g for 10 min. The supernatants were pipetted off as much as possible without disturbing the pellets, which were then resuspended in PBS and prepared for immunoblotting analysis.

For the quantification of Azurin-Flag secreted by VNP20009 cells transformed with pGT252 (P_TlpA1-RBS3-YopE_1-15-Azurin-Flag-P_j23100-TlpA₃₉), the samples after US irradiation were centrifuged and filtered to remove cells and debris. The concentration of Azurin-Flag in the supernatant was then determined using an ELISA kit designed for Flag-tagged proteins, following the manufacturer’s protocol (Flag ELISA kit, JN7162, Jining Shiye Inc.). The absorbance was measured using a Synergy H1 hybrid multi-mode microplate reader, equipped with Gen5 software (version: 2.04). To quantify Azurin, we utilized the molar relationship between Azurin and the FLAG tag. The formula used was:

$$C_{\text{Azurin}}=C_{\text{Flag}}/M_{\text{Flag}}\times M_{\text{Azurin}}$$

Where C_Azurin is the concentration of Azurin, C_Flag is the concentration of Flag (determined by ELISA), M_Flag is the molar amount of Flag, and M_Azurin is the molar amount of Azurin. For creating the standard curve, we prepared Flag standards at various concentrations (100, 50, 25, 12.5, 6.25, and 0 ng/mL), which were used to calibrate and determine the concentrations of our experimental samples.

Immunoblotting analysis

The protein samples prepared as above were resolved on a 10% SDS-PAGE and electroblotted onto a Nitrocellulose membrane (1620115, Bio-Rad). After blocked in 5% skim milk (w/v of TBST) at room temperature for 2 h, the membrane was probed with anti-Flag primary antibody (80010-1-RR, Proteintech) and anti-His primary antibody (30401ES10, YEASEN) overnight at 4°C and HRP-conjugated anti-rabbit secondary antibody (SA00001-2, Proteintech) at room temperature for 2 h, successively. The membrane was visualized with a fluorescent western blot imaging system (LI-COR Odyssey Biosciences).

Functional validation of PD-L1 nb in vitro

To investigate the function of PD-L1 nb, VNP20009 cells transformed with pGT305 (P_TlpA1-RBS3-YopE_1-15-PD-L1 nb-His-P_j23100-TlpA₃₉) were grown to OD₆₀₀ = 0.25 and irradiated by US (0.5 W/cm², pulse of 1 s on 1 s off) for 1 h (+) or not (−). The samples were maintained at 37°C for another 11 h, and the culture supernatants were collected and concentrated by ultrafiltration. H22 cells seeded in 6-well plates (1×10⁶ cells per well) were co-incubated with the concentrated culture supernatants for 2 h, co-incubated with a PE/Cy7-conjugated anti-PD-L1 monoclonal antibody (10F.9G2, BioLegend) specific to epitopes recognized by the 10F.9G2 mAb on H22 cells for 0.5 h, and then analyzed using a LSRFortessaTM Flow Cytometer (BD Biosciences).

Cell viability assay

Cell viability was assessed using a cell counting Kit-8 (CCK8, C0037, Beyotime Biotechnology) according to the manufacturer’s instructions. To evaluate the effects of US irradiation on the viability of tissue cells or tumor cells, different mammalian cells, including hMSCs, ARPE-19, NIH-3T3, HEK-293T, B16F10, and CT26 cells, were cultured in 6 cm cell-culture dishes. The pre-sterilized ultrasonic probe (delta 07 FND, Delta) was installed in the lid of the dish, allowing direct contact with the cell culture. The cells were irradiated with US (0.5 W/cm², pulse of 1 s on, 1 s off) for 1 h. Cells were then incubated at 37°C for 2 h after the CCK8 reagent was added. The absorbance (optical density) at 450 nm was measured using a Synergy H1 hybrid multimode microplate reader.

To assess the activity of Azurin secreted by VNP20009 cells carrying SINGER-Azurin in vitro, B16F10 and CT26 cells were seeded into 96-well culture plates at a density of 2×10⁴ cells per well in 150 μL of complete medium. After cultured for 12 h, 20 μL of the supernatants of engineered bacteria were added into the cell cultures and treated for 24 h before the cell viability was assessed using CCK8.

Detection of cell apoptosis in vitro

Terminal deoxynucleotidyl transferase (TdT) dUTP nick end labeling (TUNEL) analysis was performed according to the manufacturer’s instructions (Beyotime #C1086). In brief, cells (4×10⁴ cells per well, 24-well plate) were seeded on the glass slides and treated with 100 μL of the supernatants of engineered bacteria for 12 h. The cells were washed three times with PBS and fixed in a freshly prepared fixation solution (4% paraformaldehyde). After 30 min of fixation, the cells were washed twice with PBS and permeabilized in a freshly prepared permeabilization solution (0.3% Triton X-100 in PBS) for 5 min. Afterward, the cells were washed twice with PBS and incubated with a TUNEL reaction mixture at 37°C for 60 min in a humidified dark atmosphere. After washing twice with PBS, the samples were imaged using a BX53 upright microscope (Olympus, Japan).

Cell apoptosis was also assessed using the Annexin V-FITC/Propidium Iodide (PI) double-staining apoptosis detection kit (C1062M, Beyotime Biotechnology) to detect the presence of phosphatidylserine on the cell membrane surface during apoptosis. Cells were seeded into 6-well plates (2×10⁵ cells/well) and treated with 150 μL of the supernatants containing Azurin protein produced by SINGER-Azurin for 12 h. The collected cells were washed 3 times with PBS, gently resuspended in 195 μL Annexin V-FITC binding solution and 5 μL Annexin V-FITC, followed by the addition of 10 μL propidium iodide staining solution and gentle mixing. The mixture was then incubated at room temperature in the dark for 15 min. Subsequently, the apoptosis data acquisition and analysis were performed by BD LSR Fortessa flow cytometry.

Tumor models

B16F10_luci (5×10⁵) and H22 (1×10⁶) tumor cells were subcutaneously implanted into the hind flank of C57BL/6 male mice, while CT26 (2×10⁶) and A20 (3×10⁶) tumor cells were subcutaneously implanted into the hind flank of BALB/c male mice. Tumors were grown to an average volume of approximately 100 mm³ before treatment with different bacterial strains. The tumor growth was externally measured with a digital caliper, and tumor volumes were calculated using formulations V = (L×W×W)/2, where V is tumor volume, W is tumor width, and L is tumor length. Animals were euthanized with carbon dioxide asphyxiation when a tumor size greater than 1000 mm³.

Tumor treatment

Bacteria were grown in LB medium with 100 μg/mL ampicillin overnight at 37°C, 210 rpm. After dilution to OD₆₀₀ = 0.1, the cells were propagated at 37°C, 210 rpm, until an OD₆₀₀ = 0.25. Bacteria were then centrifuged at 3,000 × g for 5 min and washed three times with sterile ice-cold PBS. Bacteria were delivered intratumorally at a concentration of 10⁷ CFU/mL in PBS with a total of 50 μL injected per mouse or intravenously at a concentration of 5 × 10⁶ CFU/mL in PBS with a total of 100 μL injected per mouse when the tumor volume was about 100 mm³. The tumor location of mice received US irradiation (0.5 W/cm², pulse of 1 s on, 1 s off) for 1 h every two days. During the irradiation, an infrared thermal imager (UTi260B, UNI-T) was used to monitor the temperature of the tumor site.

In vivo imaging

Mice bearing B16F10 melanoma tumors were intraperitoneally injected with 50 μL of 15 mg/mL D-luciferin (Shanghai Sciencelight Biology Science & Technology Inc.) and anesthetized with 2% isoflurane. Ten minutes after injection, bioluminescence images of the mice were obtained using an IVIS Lumina II in vivo imaging system (PerkinElmer, USA), and the total radiant efficiency (in photons/sec/cm²/Sr) of the tumor was analyzed with the Living Image 4.3 software.

Biodistribution of bacteria in tumor-bearing mice

The bacterial strains harboring luxCDABE cassette could be visualized with the in vivo imaging system. Images were taken using the IVIS Lumina II in vivo imaging system (PerkinElmer, USA), and regions of interest were quantified as average radiance (p/sec/cm²/sr) using Living Image 4.3 software. Moreover, the major organs of mice, including heart, liver, spleen, lung, kidney, and tumor, were collected, weighed, and homogenized at 4°C in sterile PBS. The homogenates were serially diluted, plated on LB medium, and incubated at 37°C for 12 h. Bacterial colonies were counted, and the bacterial titer was calculated as colony-forming units (CFU) per gram of tissue.

Immunofluorescence analysis

Tumor tissues were collected from mice and fixed in 4% neutral paraformaldehyde for 24 h at 4°C. The tissues were washed in PBS and soaked in 30% sucrose solution overnight at 4°C. Fixed tissues were embedded in optimal cutting temperature (O.C.T) compound (4583, Tissue Tek) and kept at ˗80°C overnight. The tissues were then sectioned (10 μm thickness) using a microtome (Thermo Scientific) and mounted on glass slides, and the tumor tissue slides were heated to 100°C in antigen retrieval buffer for 30 min, blocked with 5% BSA for 0.5 h. For the assessment of tumor cell proliferation, slides bearing tumor tissue sections were incubated with rabbit anti-flag primary antibody (1:500; 80010-1-RR, Proteintech) or rabbit polyclonal anti-Ki67 (1:500; 27309-1-AP, Proteintech) overnight at 4°C. The sections were washed three times with PBST (PBS containing 0.1% Triton X-10) and then incubated with Alexa Fluor 546-conjugated secondary antibodies (1:500; SA00013-4, Proteintech) overnight at 4°C. After being washed three times with PBST, the sections were mounted in antifade mounting medium with DAPI (H-1200-10, Vector Laboratories) and examined with a BX53 upright fluorescence microscope (Olympus, Japan).

To confirm the in vivo secretion of Azurin, the slides bearing tumor tissue sections were incubated with rabbit anti-flag primary antibody (1:500; 80010-1-RR, Proteintech) and mouse anti-DnaK primary antibody (1:2000; ab69617, Abcam) overnight at 4°C, and then incubated with AF594-labelled anti-rabbit IgG (H + L) (1:500; SA00013-4, Proteintech) and AF488-labelled anti-mouse IgG(H + L) (1:500; SA00013-1, Proteintech) overnight at 4°C. After being washed three times with PBST, the sections were mounted in antifade mounting medium with DAPI (H-1200-10, Vector Laboratories) and examined with fluorescence confocal microscopy (Leica, Germany) using a 40× dry objective.

TUNEL assay

An in situ cell apoptosis detection kit (C1086, Beyotime Biotechnology) was used to evaluate the apoptosis of tumor tissue following the manufacturer’s instructions. Briefly, frozen tumor tissue sections were treated with a freshly prepared permeabilization solution (0.3% Triton X-100 in PBS) for 5 min and then stained with fluorescein deoxyuridine triphosphate (green) for 60 min. Fluorescence images were obtained from a BX53 upright fluorescence microscope (Olympus, Japan). The nuclei of apoptotic cells were stained green, and all other nuclei were stained blue.

Flow cytometry

After the indicated treatments, spleens and tumor-draining lymph nodes (TDLNs) of mice were isolated on day 11, ground up, and passed through a 70 μm cell strainer (93070, SPL life sciences). The collected spleen tissues were treated with RBC lysis buffer (420301, Biolenged) for 5 min. Subsequently, the samples were centrifuged at 400 × g for 10 min to remove the lysed RBCs in the supernatants. The pellets containing lymphocytes were washed using PBS and resuspended in 1 mL PBS. The collected TDLN samples were centrifuged at 400 × g for 10 min, and the pellets containing lymphocytes were resuspended in 1 mL PBS. For immunophenotyping of spleens and TDLNs, specific cell subsets (CD8⁺ T cells, Foxp3⁺CD4⁺ T cells, NK1.1⁺ cells) were stained with fluorescence-labeled antibodies and then measured using flow cytometry. A Zombie NIR Fixable Viability Kit (423106, Biolenged) was used as a live/dead marker. Cell surface markers were identified using a panel of fluorochrome-conjugated antibodies: PE/Cyanine7 anti-mouse CD45 (103114, clone 30-F11, Biolenged), PE/Cyanine5 anti-mouse CD45 (103110, clone 30-F11, Biolenged), FITC anti-mouse CD3 (100306, clone 145-2C11, Biolenged), APC anti-mouse CD8 (100712, clone 53–6.7, Biolenged), Brilliant Violet 510 anti-mouse CD8a (100751, clone 53–6.7, Biolenged), PerCP/Cyanine5.5 anti-mouse CD4 (100434, clone GK1.5, Biolenged), PE anti-mouse NK-1.1 (108708, clone PK136, Biolenged), Brilliant Violet 605 anti-mouse CD366 (Tim-3) (119721, clone RMT3-23, Biolenged), and Brilliant Violet 421 anti-mouse CD223 (LAG-3) (125221, clone C9B7W, Biolenged). For the intracellular cytokine staining and transcription factor staining, surface Ab-stained cells were first fixed and permeabilized using the Foxp3 Transcription Factor Staining Buffer kit (eBioscience, 00-5523-00) following the manufacturer’s instructions. Cells were further stained with Abs against intracellular proteins for 30 min at 4°C. Intracellular Abs were used as follows: Brilliant Violet 421 anti-mouse Foxp3 (126419, cloneMF-14, Biolenged), PE anti-mouse IFNγ (505808, clone XMG1.2, Biolenged). Samples were run on a BD LSR Fortessa flow cytometer and analyzed with FlowJo 10.

Cytokine detection

C57BL/6 mice bearing H22 tumors were subject to different treatments. Tumor tissues were isolated on day 11 and cytokines (TGF-β, TNF-α, IFN-γ, and IL-10) in tumors were measured using corresponding enzyme-linked immunosorbent assay (ELISA) kits (TGF-β, EK981, Multi Sciences; TNF-α, EK282, Multi Sciences; IFN-γ, EK280HS, Multi Sciences; IL-10, EK310HS, Multi Sciences), according to the manufacturer’s instructions.

C57BL/6 mice bearing H22 tumors were intravenously injected with PBS or SINGER-2. Blood was collected at 3 h and 24 h post-injection, and cytokines (IL-6 and TNF-α) in plasma were quantified using corresponding ELISA kits (IL-6, EK206, Multi Science; TNF-α, EK282, Multi Sciences) according to the manufacturer’s instructions.

Tumor treatment with recombinant proteins

Recombinant Azurin and PD-L1 nb were produced by E. coli BL21 (DE3) and purified using a process developed by Shanghai Saiheng Biotechnology Co., Ltd., resulting in final preparations in PBS at concentrations of 4 μg/μL for Azurin and 2 μg/μL for PD-L1 nb. C57BL/6 mice were implanted subcutaneously with H22 cells (1×10⁶). When the tumor volume reached approximately 100 mm³, mice were randomly divided into three groups and received intraperitoneal injections of PBS, Azurin (50 μg), or PD-L1 nb (100 μg) at days 0, 3, and 6.

Histology analysis

Tissue sections of the hearts, livers, spleens, lungs, and kidneys were stained with hematoxylin-eosin staining (H&E) and analyzed by light microscopy for postmortem histopathological analysis. Briefly, mice were killed at 14 days post-treatment, and the tissues were collected and fixed in 4% neutral paraformaldehyde for 24 h, dehydrated by passage through a graded series of ethyl alcohol, cleaned in xylene, embedded in paraffin, and the tissues were then sectioned (5 μm thickness) with a rotary microtome (Leica RM2235, Manual Rotary Microtome) and mounted on glass slides. Paraffin-embedded hearts, livers, spleens, lungs, and kidneys sections were stained with the H&E Staining Kit (G1120, Solarbio) following the manufacturer’s instruction and examined under a BX53 upright fluorescence microscope (Olympus, Japan).

Complete blood count and blood biochemistry analysis

Mice were euthanized, and whole blood was collected for a complete blood count using the Sysmex XT-2000iV hematology analyzer (Sysmex). Blood biochemistry indicators, including alanine aminotransferase, aspartate aminotransferase, blood urea nitrogen, and creatinine, were determined using an automatic biochemical analyzer BX-3010 (Sysmex). All biochemical serum evaluations were performed at the same time to minimize analytical variability.

Quantification and statistical analysis

Unless otherwise mentioned, all in vitro data represent means ± SD of three independent biological replicates. For the animal experiments, each treatment group consisted of randomly selected mice (n = 4–6). Comparisons between the two groups were performed using two-tailed unpaired t-tests. One-way ANOVA was used to evaluate differences between multiple groups with a single intervention, followed by Tukey’s post-test, and the results are expressed as means ± SEM. Statistical significance was determined using GraphPad (Prism v.8), and significance was assigned at ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, and ∗∗∗∗p < 0.0001. n and p values are described in the figures or figure legends.

Published: April 11, 2024