Application of artificial intelligence-driven protein design in nano-vesicle analysis, preparation and drug delivery

Abstract

The present disclosure provides an engineered delivery system comprising bacterial membrane vesicles derived from bacteria and one or more non-bacterial proteins for anchoring to the bacterial membrane vesicles.

Description

[0001] Cross-reference related applications This application claims priority to U.S. Provisional Application No. 63 / 447,384, filed February 22, 2023, and U.S. Provisional Application No. 63 / 517,149, filed August 2, 2023, the entire contents of which are incorporated herein by reference.

[0002] sequence list reference The sequence list conforming to WIPO standard ST.26 is incorporated herein by reference. This sequence list is an electronic document created on February 22, 2024, named “Accure-2024-02-22-SEQL.xml”, and is 73,446 bytes in size. Background Technology

[0003] The success of mRNA-based COVID-19 vaccines has sparked interest in using mRNA to deliver therapeutic proteins. In many cases, it is necessary to deliver mRNA to specific targets, cells, or tissues and maintain persistent, high levels of expression in the target cells. Developing safe and more efficient delivery technologies is crucial to realizing the potential of targeted mRNA therapies for treating both rare and common diseases. Summary of the Invention

[0004] Embodiments of this disclosure provide an engineered delivery system. The engineered delivery system includes bacterial membrane vesicles derived from bacteria and one or more non-bacterial proteins for anchoring to the bacterial membrane vesicles.

[0005] In some embodiments, the one or more nonbacterial proteins include mammalian membrane-associated proteins or fragments thereof.

[0006] In some embodiments, the one or more nonbacterial proteins are further linked by direct linkage or by linking peptides to polypeptide conjugates.

[0007] In some embodiments, the polypeptide conjugate is displayed on the outer side of the vesicle membrane.

[0008] In some embodiments, the linker peptide is a glycine-serine linker peptide.

[0009] In some embodiments, the glycine-serine linker peptide is GGGGS.

[0010] In some embodiments, the polypeptide conjugate is a synthetic polypeptide.

[0011] In some embodiments, the polypeptide conjugate includes a nucleic acid binding domain.

[0012] In some embodiments, the polypeptide binder comprises an amino acid sequence as set forth in SEQ ID NO: 4, 5, 22, 23, 24, or 61.

[0013] In some embodiments, the bacteria is a gram-negative bacteria.

[0014] In some embodiments, the gram-negative bacteria is Escherichia coli.

[0015] In some embodiments, the one or more non-bacterial proteins comprise a mammalian protein voltage-dependent anion-selective channel 1 (VDAC1), mitochondrial carrier homolog 2 (MTCH2), or acyl-CoA synthetase long-chain family member 1 (ACSL1), or a fragment thereof.

[0016] In some embodiments, the mammalian protein comprises an amino acid sequence as set forth in SEQ ID NO: 1, 2, or 3.

[0017] In some embodiments, the full amino acid sequence of the non-bacterial protein is as set forth in SEQ ID NO: 6, 7, 8, 9, 10, 11, 12, 13, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43, 45, 47, 49, 51, 53, 55, 57, or 59.

[0018] In some embodiments, the engineered delivery system further comprises a nucleic acid, a protein, a complex thereof, or a combination thereof, located within the bacterial membrane vesicle.

[0019] In some embodiments, the nucleic acid is an mRNA, a circular RNA, or an antisense oligonucleotide.

[0020] Embodiments of the present disclosure also provide a synthetic polypeptide that binds to Programmed Death-Ligand 1 (PD-L1). The synthetic polypeptide comprises an amino acid sequence as set forth in SEQ ID NO: 4, 5, 23, or 24, including conservative mutants thereof.

[0021] In some embodiments, the synthetic polypeptide is linked to a small molecule tag located at the N-terminus or C-terminus of SEQ ID NO: 4, 5, 23, or 24.

[0022] In some embodiments, the synthetic polypeptide further comprises a second polypeptide located at the N-terminus or C-terminus of SEQ ID NO: 4, 5, 23, or 24.

[0023] In some embodiments, the synthetic polypeptide further comprises a bacterial signal peptide.

[0024] In some embodiments, the complete amino acid sequence is as shown in SEQ ID NO: 14, 15, 16, 17, 18, 19, 20, 21, 28, 30, 34, 36, 40, 42, 46, 48, 52, 54, 56, 58, 60, 62, 63 or 64.

[0025] Embodiments of this disclosure also provide a synthetic polypeptide that specifically binds to Claudin 18.2. The synthetic polypeptide comprises the amino acid sequence shown in SEQ ID NO: 22, including its conserved mutants.

[0026] In some embodiments, the synthetic polypeptide is further linked to a small molecule tag located at the N-terminus or C-terminus of SEQ ID NO: 22.

[0027] In some embodiments, the synthetic polypeptide further includes a second polypeptide located at the N-terminus or C-terminus of SEQ ID NO: 22.

[0028] In some embodiments, the synthetic polypeptide further includes a bacterial signaling peptide.

[0029] In some embodiments, the complete amino acid sequence is as shown in SEQ ID NO: 26, 32, 38, 44 or 50.

[0030] Embodiments of this disclosure also provide a method for preparing microbial vesicles. The method includes: computationally designing a membrane vesicle (EV) scaffold protein comprising a human membrane anchor peptide, a synthetic polypeptide conjugate, and a bacterial signal peptide domain; cloning the EV scaffold protein into a bacterial expression vector; expressing the EV scaffold protein in *Escherichia coli* or other Gram-negative bacteria; purifying the EV scaffold protein from the bacterial expression culture; and verifying the targeted binding and cellular uptake capabilities of the EV scaffold protein.

[0031] Brief description of the attached figures Figure 1 The size and quantification of Spirulina membrane vesicles (EVs) determined using nanoparticle tracking analysis (NTA) are shown. The average size of Spirulina EVs is approximately 177 nm.

[0032] Figure 2This is a schematic diagram of in vitro targeted binding TiMES detection. This detection is used to verify whether engineered bacterial membrane vesicles (EVs) bind to their target proteins (such as PD-L1 and Claudin 18.2) in vitro. Engineered bacterial EVs expressing the designed scaffold bind to the target proteins coated on magnetic particles (MPs), followed by electrochemical detection using universal EV markers. TiMES detection and equipment are employed, such as those disclosed in U.S. Patent No. 11,125,745 and U.S. Patent Application Publication No. 2021 / 0208169, the entire contents of which are incorporated herein by reference.

[0033] Figure 3 This is a schematic diagram illustrating the targeted delivery of RNA to mammalian cells using engineered microbial vesicles. Cellular uptake of the vesicles is achieved through the interaction of the engineered vesicles with target biomarker proteins on the cell surface.

[0034] Figure 4 This is a schematic diagram illustrating the targeted RNA delivery to mammalian gastric cells using engineered microbial vesicles expressing a human gastric tissue-specific target biomarker protein (ATP4A). The uptake of the vesicles by the gastric cells is achieved through the interaction of the engineered vesicles with ATP4B expressed on the cell surface.

[0035] Figure 5 It is a flowchart demonstrating the experimental workflow for the design, preparation, and validation of engineered microbial membrane vesicles.

[0036] Figure 6 It is a schematic diagram demonstrating the design, preparation, isolation, RNA loading and cellular uptake of engineered microbial membrane vesicles, as well as RNA expression in target cells.

[0037] Figure 7 This is a schematic diagram of TiMES detection, which uses artificial intelligence-designed protein conjugates to capture or detect EV biomarkers.

[0038] Figure 8 It is a schematic diagram and a fluorescence microscope image demonstrating the detection of PD-L1 expression in cells using PD-L1 conjugates designed with artificial intelligence.

[0039] Figure 9 It is a schematic diagram and a fluorescence microscope image demonstrating how artificial intelligence-designed PD-L1 conjugates can block the interaction between cellular PD-L1 and its antibody.

[0040] Figure 10 It is a graph showing the standardized PD-L1 binding signal from various engineered E. coli EVs and control samples.

[0041] Figure 11It is a graph showing the standardized Claudin 18.2 binding signal from various engineered E. coli EVs and control samples.

[0042] Figure 12 This is a schematic diagram and fluorescence microscope image illustrating the uptake of engineered microbial membrane vesicles and the expression of eGFP mRNA in target cells expressing PD-L1.

[0043] Figure 13 It is a flowchart demonstrating the design and preparation of protein conjugates and their application in analyte capture, detection and blocking.

[0044] Detailed description Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. The following definitions supplement those in the prior art and apply to this application, and should not be attributed to any related or unrelated case, such as any jointly owned patent or application. While this disclosure may be practiced or tested using similar or equivalent methods and materials as described herein, preferred materials and methods are described herein. Therefore, the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting.

[0045] In this specification and appendix claims, unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “the” include plural references. For example, reference to “an analyte” includes a plurality of such analytes, and so on.

[0046] The term “about” as used in this article means that the value of a given quantity varies within ±10% of that value.

[0047] Unless otherwise specified, the scope listed in this article includes endpoints.

[0048] This article defines or otherwise characterizes many additional terms.

[0049] The success of mRNA COVID-19 vaccines has sparked interest in using mRNA to deliver therapeutic proteins. However, the development of mRNA therapies faces additional challenges. Compared to mRNA vaccines, mRNA therapies require up to 1000 times higher protein levels to reach the therapeutic threshold. In many cases, it is necessary to deliver mRNA to specific targets, cells, or tissues and maintain persistent and high levels of expression in the target cells. Most mRNA therapies currently under development rely on lipid nanoparticles (LNPs). However, LNPs administered systemically accumulate primarily in the liver; efficient delivery to other solid organs remains challenging. Another major obstacle to LNP circulation is immunogenicity. Even with advanced LNPs, repeated administration in chronic diseases can activate innate immunity, leading to cytotoxicity and rapid clearance of LNPs. On the other hand, oral administration is the most widespread form of drug delivery, but the gastrointestinal tract (GI) presents numerous barriers to LNP-based RNA delivery. In particular, acidic pH and the formation of coronal masses in gastrointestinal fluids significantly affect LNP stability and cellular uptake. Therefore, developing safe and more efficient delivery technologies is crucial to realizing the potential of targeted mRNA therapies for treating both rare and common diseases.

[0050] Extracellular vesicles (EVs) are naturally derived, lipid-encapsulated nanoparticles involved in intercellular communication across all living organisms. They carry various biological payloads (e.g., DNA, RNA, proteins) and transport these payloads between cells via endocytosis and efflux. Therefore, EVs have been explored as drug carriers to leverage their inherent tissue-directed delivery capabilities to specific cells or tissues. Compared to conventional synthetic carriers (e.g., LNPs), EVs offer several advantages, such as biocompatibility, relatively high stability in biological fluids, and low immunogenicity. Based on their favorable safety profile, several EV-based therapeutic interventions have entered Phase I or II clinical trials for various diseases, including tissue regeneration, stroke, and cancer (clinicaltrials.gov / ). Despite extensive research, broader clinical translation of EVs as drug carriers remains hampered by a lack of robust EV engineering strategies, low-cost preparation systems, and effective analytical methods.

[0051] Microbial systems, including nonpathogenic bacteria, probiotics, and microalgae, are cost-effective systems for producing EVs. For example, *Escherichia coli* (E. coli) is a common Gram-negative bacterium found in the lower intestine of mammals. Many *E. coli* strains are part of the normal gut microbiota and are harmless or even beneficial to humans. Another example is *Spirulina* (Arthrospira platensis), a globally consumed edible Gram-negative cyanobacterium known for its high protein content and other nutritional benefits. Bioencapsulation within *Spirulina* biomass protects therapeutic payloads from the low pH and high concentrations of pepsin in the gastric environment. Orally delivered therapeutic antibodies expressed by *Spirulina* have shown safety for human administration in a phase I clinical trial. Nonpathogenic *E. coli* and *Spirulina* are potentially advantageous systems for producing EVs as drug carriers: i) they combine the safety of a nonpathogenic host with the high yield and low cost of a microbial platform; ii) *E. coli* and *Spirulina* cells release large quantities of EVs with a size distribution similar to human EVs. Figure 1 Furthermore, the outer membrane of Spirulina differs chemically from other Gram-negative bacteria, containing virtually no pro-inflammatory endotoxins. Spirulina-derived EVs are not only a safe carrier for oral delivery, but also, due to their low immunogenicity, may be easier to purify for systemic delivery.

[0052] Most human EV membrane-anchored proteins cannot be readily expressed in microbial hosts due to diverse membrane targeting pathways and a lack of specific post-translational modification mechanisms within microbial systems. To address unmet needs in EV engineering and analysis, this disclosure presents two recently developed techniques to advance EV-based drug delivery. i) Sequence- and structure-based artificial intelligence algorithms for designing target-specific EV protein scaffolds. Significant progress has been made in protein folding and design using deep learning methods. Diffusion models have shown unprecedented success in image and language generation modeling and have also demonstrated excellent performance in topologically constrained protein design, protein conjugate design, and motif scaffold design. Design speed and success rates have significantly improved in recent years, and de novo designed proteins can now achieve high binding affinity and specificity for structured and flexible targets. EV biomarker databases have been established, including EV proteins that are substantially stable in human biological fluids, as well as EV proteins enriched for specific tissues or diseases. These resources allow for the optimization of artificial intelligence parameters to design compact and target-specific EV protein scaffolds. ii) Advanced EV analysis systems based on integrated magnetoelectric sensing technology, such as TiMES, have been developed, as disclosed in, for example, U.S. Patent No. 11,125,745 and U.S. Patent Application Publication No. 2021 / 0208169, the entire contents of which are incorporated herein by reference. The TiMES system automates EV separation and detection on a single platform and offers significant advantages: a) it allows direct analysis of EVs from complex media without filtration or centrifugation; b) it provides superior analytical performance with a limit of detection (LOD) of approximately 10⁴ EVs (approximately 1000 times more sensitive than ELISA) and a dynamic range spanning four orders of magnitude; c) it enables parallel analysis of multiple EV biomarkers with a total detection time of less than one hour. In this disclosure, the invention combines AI-driven EV design, TiMES EV analysis, and a spirulina-based preparation method to accelerate the development of targeted or tissue-specific drug delivery. Figure 2 A schematic diagram of in vitro targeted binding TiMES assay is shown. This assay is used to verify whether engineered bacterial membrane vesicles (EVs) bind to their target proteins (e.g., programmed death-ligand 1 (PD-L1), Claudin 18.2) in vitro. Engineered bacterial EVs expressing the designed scaffold bind to the PD-L1 protein coated on magnetic particles (MPs), followed by electrochemical detection using a universal EV marker, employing TiMES detection and apparatus.

[0053] As an example, EV scaffolds have been designed to facilitate the delivery of targeted RNA to PD-L1 positive cells. Figure 3The study considered technical and commercial factors: i) PD-L1 is a well-established therapeutic target for various solid tumors, including gastrointestinal cancers that can be treated via systemic and oral routes; ii) a large number of cell lines and other reagents are available for evaluating and benchmarking the efficacy and specificity of PD-L1 targeting; and iii) there is clinical and commercial interest in developing combination therapies with programmed cell death protein 1 (PD-1) inhibitors (anti-PD-1 / PD-L1). PD-L1-targeted EVs could provide a strategy for combining anti-PD-1 / PD-L1 antibodies with novel mRNA tumor suppressor drugs.

[0054] As another example, EV scaffolds have been designed to facilitate the delivery of targeted RNA to Claudin 18.2 positive cells. Figure 3 (i) Claudin 18.2 is an emerging therapeutic target for various solid tumors, including gastric and pancreatic cancer, which can be treated via systemic or oral administration; (ii) there is clinical and commercial interest in developing combination therapies with anti-Claudin 18.2 drugs. Targeting EVs with Claudin 18.2 could provide a strategy for combining anti-Claudin 18.2 antibodies with novel mRNA tumor suppressor drugs.

[0055] As another example, EV scaffolds can be designed to facilitate the delivery of targeted RNA to ATP4B-positive cells. Figure 4 ATP4B is a gastric tissue-specific protein biomarker. Engineered microorganisms (EVs) expressing the extracellular domain of human ATB4A or expressing an AI-redesigned version of ATB4A (e.g., hypothetical ATB4A) can be used to deliver therapeutic nucleic acids (e.g., gene editors, tumor suppressor RNA, protein degraders) to normal gastric tissue or gastric cancer cells. Figure 4 ).

[0056] PD-L1, Claudin 18.2, or ATP4B-targeted EVs are merely examples. Microbial EV platforms can be used directly to target other cell types (e.g., EGFR-mutant tumor cells, T cells, astrocytes) and deliver a variety of RNA payloads (e.g., tumor suppressor RNA, protein degraders, chimeric antigen receptors, gene editors).

[0057] In this disclosure, the present invention provides a method for targeted or tissue-specific drug delivery by combining artificial intelligence-driven vesicle design and analysis with a low-cost microbial preparation and delivery system. Key features and competitive advantages are discussed below and in Table 1.

[0058] Table 1. Competitive advantages of engineered microorganisms (EVs) for RNA delivery

[0059] Figure 5 This is a flowchart illustrating the design, preparation, and validation methods for engineered microbial membrane vesicles. It is understood that the operations shown in this method are not exhaustive; other operations may be performed before, after, or between the operations illustrated. Furthermore, some operations may be performed simultaneously or in conjunction with... Figure 5 The different execution sequences are shown.

[0060] refer to Figure 5 Method 500 begins with operation 502 and involves computationally designing EV scaffold proteins comprising human membrane anchoring peptides, synthetic peptide conjugates, and bacterial signaling peptide domains. In some embodiments, deep learning-based structure prediction, diffusion generation models, and EV biomarker databases can be used to design novel EV scaffold proteins with high accuracy and speed. This approach allows for: a) rapid design of candidate EV scaffolds to facilitate membrane anchoring and vesicle release from the microbial host; b) efficient loading of RNA and protein payloads; and c) in vitro screening of candidate designs to improve targeted uptake and minimize non-targeted interactions.

[0061] In step 504, the EV scaffold protein is cloned into an expression vector. In some embodiments, natural and engineered EVs of human or bacterial origin can be used as drug carriers for RNA and protein therapies. Preliminary studies have shown that EVs derived from human cells (e.g., hematopoietic cells, HEK293T) are generally safe with no significant adverse reactions. However, preparation challenges (including drug loading efficiency, batch-to-batch variability, and cost) limit their scalability. The engineering and preparation of microbial EVs may be more easily scalable. However, most microbial engineering strategies utilize viral proteins for membrane targeting and anchoring, which can lead to toxicity and immunogenicity. The method disclosed herein relies on unique mammalian or humanized recombinant proteins to minimize toxicity and immunogenicity.

[0062] In Operation 506, EV scaffold protein is expressed in Escherichia coli or other Gram-negative bacteria.

[0063] In operation 508, EVs expressing engineered scaffold proteins were purified from bacterial expression cultures.

[0064] In operation 510, the targeted binding and cellular uptake capabilities of the EVs are verified. In some embodiments, an automated analysis system (TiMES) can be used to rapidly monitor the quantity and stability of engineered microbial EVs after preparation and administration. For EV quality assessment, the cell targeting and RNA loading properties of engineered EVs can be monitored in near real-time (approximately 1 hour from sample to response). Furthermore, human cell-derived EVs can be used non-invasively to analyze treatment response after intracellular release of the therapeutic payload (e.g., in tumor, immune, or intestinal cells).

[0065] As an example, modular design of EV scaffold proteins can be achieved using state-of-the-art protein design algorithms and parameters (e.g., RFdiffusion, ProteinMPNN). Membrane anchoring, payload (RNA) binding, and target binding structures can be designed independently first to improve computational throughput, and then combined to evaluate structural stability and target binding. The design can be optimized through successive noise addition and denoising (partial diffusion). A design can be considered successful if the Alphafold2 (AF2) prediction alignment error (pAE) between the designed protein and the target (e.g., PD-L1) is <10, the root mean square deviation (RMSD) between the designed protein and the AF2 prediction is <2 Å, and the AF2 local structure prediction reliability score (pLDDT) is >80. For example, EV protein scaffolds targeting PD-L1 or Claudin 18.2 have been designed.

[0066] In some embodiments, computational protein-protein interaction analysis may be performed to assess targeted binding and nonspecific interactions. Successful targeted binding designs from the above steps can be evaluated using protein complex prediction algorithms. The top 50 designs based on pDockQ, IPTM, and pAE scores can be further evaluated for nonspecific interactions with other targets in a proprietary database, including EV and receptor proteins highly expressed in normal tissues or immune cells, as well as proteins associated with drug toxicity.

[0067] In some embodiments, computational tools (e.g., patcHwork) can be used to evaluate the designed protein and assess how pH affects its stability and binding interface.

[0068] Genes encoding codon-optimized proteins can be synthesized and cloned into E. coli protein expression vectors. Following plasmid transformation and protein expression induction in E. coli cells (e.g., Lemo21), EVs can be harvested and purified from the bacterial supernatant using the ExoBacteria OMV isolation kit or ultracentrifugation. Lemo21 is only an example. In practice, other non-pathogenic bacterial strains can also be used, including but not limited to BL21(DE3), E. coli Nissle1917, Lactobacillus, Bifidobacterium, and Bacillus licheniformis.

[0069] In some embodiments, codon-optimized genes encoding designed proteins may be cloned into an integration vector for Spirulina transformation. When the engineered Spirulina strain is nearly completely isolated, the total cell biomass and cell culture supernatant can be harvested using the ExoBacteria OMV isolation kit or ultracentrifugation. Spirulina is only an example. In practice, other microalgae strains may also be used, including but not limited to Synechocystis sp. PCC 6803, Prochlorococcus marinus subsp. pastoris str. CCMP1986, Cyanophora paradoxa, and Tetraselmis chuii.

[0070] In some embodiments, codon-optimized genes encoding one or more nonbacterial proteins may be synthesized and cloned into an *E. coli* protein expression vector. Following plasmid transformation and protein expression induction in *E. coli* cells (e.g., Lemo21), EVs can be harvested and purified from the bacterial supernatant using an exogenous bacterial outer membrane vesicle (ExoBacteria OMV) isolation kit or ultracentrifugation. Lemo21 is merely an example. In practice, other non-pathogenic bacterial strains may also be used, including but not limited to BL21(DE3), *E. coli* Nissle 1917, *Lactobacillus*, *Bifidobacterium*, and *Bacillus licheniformis*. In some embodiments, the engineered delivery system comprises bacterial membrane vesicles derived from bacteria; and one or more nonbacterial proteins for anchoring to the bacterial membrane vesicles. In some embodiments, the one or more nonbacterial proteins comprise mammalian membrane-associated proteins or fragments thereof.

[0071] In some embodiments, the mammalian membrane-associated protein is further linked to a polypeptide conjugate, either directly or via a linker peptide. The polypeptide conjugate may be a synthetic polypeptide, or it may include a nucleic acid binding domain. In some embodiments, the polypeptide conjugate comprises an amino acid sequence as shown in SEQ ID NO: 4, 5, 22, 23, 24, or 61.

[0072] In some embodiments, the polypeptide conjugate may be displayed on the outer side of the vesicle membrane. The membrane orientation of the conjugate is determined by the protein sequence and structure of the membrane-associated protein and the conjugate. The linker peptide may be a glycine-serine linker peptide, and the glycine-serine linker peptide may be GGGGS. A codon-optimized gene encoding the fusion protein may be synthesized and cloned into an *E. coli* protein expression vector to generate the fusion protein and membrane vesicles displaying the fusion protein.

[0073] In some embodiments, Gram-negative bacteria engineered EVs can be used as a delivery system. In particular, the Gram-negative bacteria can be Escherichia coli.

[0074] In some embodiments, the one or more nonbacterial proteins may include mammalian proteins such as voltage-dependent anion-selective channel 1 (VDAC1), mitochondrial carrier homolog 2 (MTCH2), or acyl-CoA synthase long chain family member 1 (ACSL1) or fragments thereof. The mammalian protein may include the amino acid sequence shown in SEQ ID NO: 1, 2, or 3.

[0075] In some embodiments, the complete amino acid sequence of the nonbacterial protein is as shown in SEQ ID NO: 6, 7, 8, 9, 10, 11, 12, 13, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43, 45, 47, 49, 51, 53, 55, 57 or 59.

[0076] In some embodiments, the engineered EV may further comprise nucleic acids (such as DNA or RNA), proteins (such as enzymes or antibodies), complexes thereof, or any combination thereof. The nucleic acid may be mRNA, circular RNA, or antisense oligonucleotides.

[0077] Purified EVs can be analyzed using an automated TiMES instrument. This assay uses magnetic particles (MPs) coated with target proteins (e.g., PD-L1 or Claudin 18.2) to capture bacterial EVs expressing a designed targeted scaffold. Antibodies against universal bacterial EV markers (e.g., Enolase, OmpA, GroEL) can then be introduced and conjugated with an oxidase (horseradish peroxidase, HRP). The MP-EV complex is mixed with a chromogenic electron medium (3,3′,5,5′-tetramethylbenzidine, TMB) and magnetically concentrated on the electrode; HRP catalyzes the oxidation of TMB, which is then reduced by accepting electrons from the electrode, generating an electrical signal as the analytical readout. Figure 2 The net signal difference (ΔI) between engineered EV samples and control samples (EV purified from control E. coli) can be obtained. ΔI can be further normalized to the total bacterial EV concentration in the cell culture.

[0078] In some embodiments, the binding of engineered microbial vesicles to target proteins can also be verified using conventional immunoassays (such as enzyme-linked immunosorbent assay (ELISA)) or flow cytometry through methods similar to those described above.

[0079] In some embodiments, to assess the stability of engineered microbial EVs in biofluids, purified EVs can be immersed in human plasma or gastric juice, the EV concentration can be remeasured, and the binding of the EVs to the target protein can be tested to see if these biofluids affect their performance.

[0080] In some embodiments, RNA (e.g., eGFP mRNA, eGFP circular RNA) can be loaded into engineered microbial EVs via electroporation. Modified eGFP mRNA or circular RNA can be synthesized via in vitro transcription and encapsulated into microbial EVs using an optimized electroporation protocol to maintain EV membrane integrity and mRNA stability. After electroporation, exogenous mRNA can be removed using RNase. In some embodiments, RNA can be loaded onto the surface of microbial EVs via RNA-protein binding (i.e., “RBD backpack”). RNA carrying a specific RNA-binding domain (RBD) recognition sequence can be synthesized via in vitro transcription and loaded into microbial EVs by binding to the corresponding RBD on a designed EV scaffold protein. In some embodiments, modified eGFP mRNA or circular RNA can be synthesized via in vitro transcription and encapsulated into microbial EVs using acoustic perforation. eGFP RNA is only an example. In practice, a variety of RNA payloads can be loaded into engineered microbial EVs, including but not limited to tumor suppressor RNA, protein degraders, chimeric antigen receptors, and gene editors.

[0081] In some embodiments, engineered, RNA-loaded microbial EVs can be administered to human cell lines expressing high or low levels of target proteins (proteinatlas.org). Examples shown in Table 2 can be used to load eGFP mRNA or circular RNA into microbial EVs as described above, or directly added to cell culture media without any carrier. eGFP expression can be measured every 24 hours by imaging and / or flow cytometry to assess the expression level and duration of RNA in target cells. Delivery efficiency e is defined as: e = percentage of eGFP-positive cells (P) × mean fluorescence intensity (MFI). To further optimize cellular uptake efficiency, different RNA-EV ratios and EV-cell ratios can be tested to estimate the minimum EV and RNA dose required to achieve sustained RNA expression in target cells. Figure 6 ).

[0082] Table 2. Experimental design for microbial EV cell uptake

[0083] In some embodiments, the functionality and safety of engineered microbial EVs can be tested in wild-type and cell line-derived xenograft (CDX) mouse models. Cell lines used to generate CDX mouse models include, but are not limited to, RKO colon cancer cells, HCC4006 lung cancer cells, Hs 746T gastric cancer cells, SNU-423 hepatocellular carcinoma cells, Panc 08.13 pancreatic cancer cells, and HuP-T4 pancreatic cancer cells. Natural or engineered microbial EVs carrying GFP mRNA or circular RNA can be mixed with excipients (e.g., spirulina powder) and administered orally to mice. GFP expression in cancer cells and intestinal cells (targeted or untargeted) can be measured by stereofluorescence microscopy. Alternatively, natural or engineered bacterial EVs carrying GFP RNA can be purified and administered to mice via systemic or intranasal administration. GFP expression in targeted (or untargeted) cells can be measured by stereofluorescence microscopy. Serum cytokines and liver and kidney histopathology can be assessed to determine toxicity.

[0084] In some embodiments, naturally occurring or engineered bacterial EVs carrying therapeutic RNA (e.g., tumor suppressor RNA, protein degraders, chimeric antigen receptors, gene editors) can be administered orally, systemically, or intranasally to mice (or clinical study participants). Toxicity and therapeutic efficacy can then be assessed.

[0085] In some embodiments, computationally designed protein scaffolds or conjugates can be used for in vitro analytical detection. In some embodiments, the protein scaffold or conjugate further comprises a second polypeptide located at the N-terminus or C-terminus of SEQ ID NO: 4, 5, 22, 23, or 24. The protein scaffold or conjugate may further comprise a bacterial signal peptide. The complete amino acid sequence of the PD-L1 binding polypeptide may be as shown in SEQ ID NO: 14, 15, 16, 17, 18, 19, 20, 21, 28, 30, 34, 36, 40, 42, 46, 48, 52, 54, 56, 58, 60, 62, 63, or 64. The complete amino acid sequence of the Claudin 18.2 binding polypeptide may be as shown in SEQ ID NO: 26, 32, 38, 44, or 50. The codon-optimized gene encoding the polypeptide sequence may be synthesized and cloned into the pET-29b(+) E. coli plasmid expression vector. The plasmid can be transformed into chemically competent *E. coli* Lemo21 cells [or BL21 Star™ (DE3) cells]. Bacteria can be cultured at 37°C (or 30°C) in LB medium containing 50 μg / mL kanamycin and 30 μg / mL chloramphenicol. Cells can then be cultured at 37°C (or 30°C, 15°C) in TB medium containing 50 μg / mL kanamycin, 30 μg / mL chloramphenicol, 2 mM MgSO4, and 0.5 mM L-rhamnose until an OD600 of 0.4–0.6 is reached, followed by IPTG induction for 15–20 hours. Cells can be harvested and lysed by centrifugation at 4°C at 4300 g, followed by treatment with DNase I and a protease inhibitor. The clarified lysate supernatant can be batch-bound with equilibrated Ni-NTA resin and subsequently washed three times with washing buffer. Proteins can be further purified by size exclusion chromatography and characterized by SDS-PAGE.

[0086] In some embodiments, computationally designed protein conjugates can be used as capture or detection agents in immunoassays. Purified conjugate proteins (e.g., PD-L1 conjugates) can be immobilized onto plastic surfaces, magnetic particles, or biosensors to capture free target proteins (e.g., PD-L1), cells expressing the target protein, or extracellular vesicles. These captured cells, vesicles, or proteins can then be labeled with another protein conjugate or antibody for signal detection. In some embodiments, cells, vesicles, or free proteins expressing the target can be first captured using a capture antibody and then labeled with a designed protein conjugate (including biotinylated or tag-linked conjugates) for signal detection. In some embodiments, PD-L1 or Claudin18.2 binding peptides are linked to small molecule tags. In particular, PD-L1 binding peptides are linked to small molecule tags located at the N-terminus or C-terminus of SEQ ID NO: 4, 5, 22, 23, or 24. In some embodiments, the small molecule tag can be a biotin molecule linked to the peptide via chemical or enzymatic biotinylation. In some embodiments, the PD-L1 or Claudin 18.2 binding peptide is further fused to a second peptide (e.g., his-tag, MBP-tag, Avi-tag, eGFP-tag) located at the N-terminus or C-terminus of SEQ ID NO: 4, 5, 22, 23, or 24. These computationally designed protein conjugates are particularly valuable when i) no functional antibody is available, or ii) only one functional antibody is available. Computationally designed protein conjugates can be advantageous when nonspecific antibody interactions are a concern.

[0087] In some implementations, computationally designed protein conjugates can be used as capture or detection agents in magnetoelectric sensing. Purified conjugate proteins (e.g., PD-L1 conjugates) can be immobilized to magnetic particles (MPs) to capture free target proteins (e.g., PD-L1) or extracellular vesicles expressing the target protein. These captured vesicles or proteins can then be labeled with another protein conjugate or antibody for signal detection. Figure 7 In some embodiments, vesicles or free proteins may first be captured using a capture antibody and then labeled with a designed protein conjugate, including a biotinylated or tag-linked conjugate. The MP-EV complex may then be magnetically concentrated on a sensing electrode; redox reactions and electron transfer from the electrode generate an electrical signal as an analytical readout.

[0088] In some embodiments, computationally designed protein conjugates can be used for microscopy or flow cytometry. These computationally designed protein conjugates can be used to detect the expression of their target binding proteins in cells or cell-derived extracellular vesicles by microscopy or flow cytometry. In some embodiments, a PD-L1 or Claudin 18.2 binding peptide is linked to a small molecule tag. In particular, the PD-L1 binding peptide is linked to a small molecule tag located at the N-terminus or C-terminus of SEQ ID NO: 4, 5, 22, 23, or 24. In some embodiments, the small molecule tag can be a fluorescent probe (e.g., an organic dye, quantum dot) that can be crosslinked to the peptide. In some embodiments, the small molecule tag can be a biotin molecule linked to the peptide via chemical or enzymatic biotinylation. In some embodiments, the PD-L1 or Claudin 18.2 binding peptide is further fused to a second peptide (e.g., his-tag, MBP-tag, Avi-tag, eGFP-tag) located at the N-terminus or C-terminus of SEQ ID NO: 4, 5, 22, 23, or 24. In some embodiments, the labeled PD-L1 binding peptide can be used to directly stain PD-L1 protein on the cell surface using fluorescence microscopy or flow cytometry. In some embodiments, a his-tag-linked peptide conjugate can be applied first to bind to the target on the cell surface, followed by the application of a fluorescently labeled anti-his-tag antibody to indirectly detect target expression. Figure 8 ).

[0089] In some implementations, computationally designed protein conjugates can serve as blocking agents in cell assays. Purified protein conjugates can be applied to bind target proteins on the cell surface and block the interaction of the target proteins with other molecules (e.g., antibodies, antibody-drug conjugates, natural ligands), thereby modulating the cellular function of the target cells. Figure 9 ).

[0090] Example 1: Preparation of E. coli vesicles capable of binding PD-L1 In some embodiments, codon-optimized genes encoding polypeptide sequences (SEQ ID NO: 4, 5, 14, 15, 16, 17, 18, 19, 20, 21, 23, 24, 28, 30, 34, 36, 40, 42, 46, 48, 52, 54, 56, 58, 60, 62, 63, 64 and their conserved mutants) can be synthesized and cloned into pET-29b (+) or pET-21b (+) *E. coli* plasmid expression vectors. The plasmids can be transformed into chemically competent *E. coli* Lemo21 cells. The bacteria can be cultured at 37°C (or 30°C) in LB medium containing 50 μg / mL kanamycin and 30 μg / mL chloramphenicol. Cells can then be cultured at 37°C (or 30°C) in TB medium containing 50 μg / mL kanamycin, 30 μg / mL chloramphenicol, 2 mM MgSO4 and 0.5 mM L-rhamnose until OD600 reaches 0.4-0.6, and then induced with IPTG for 12-20 hours.

[0091] In some embodiments, bacteria can be precipitated by centrifugation at 4300 xg for 20 minutes at 4°C. The culture supernatant can be collected and filtered through a 0.45 μm vacuum filter. The filtered supernatant containing membrane vesicles can be used directly for downstream analysis or further purified using a membrane vesicle isolation kit (e.g., SBI ExoBacteria OMV Isolation Kit). For certain downstream analyses (e.g., nanoparticle tracking analysis, RNA loading), an additional buffer exchange step can be performed to use an appropriate buffer.

[0092] In some embodiments, automated magnetoelectric sensing can be used to analyze clarified bacterial supernatant or purified membrane vesicles to test for binding to PD-L1. This assay can begin by capturing bacterial membrane vesicles (EVs) directly from clarified culture supernatant or purified bacterial membrane vesicle solution using magnetic particles (MPs), based on the binding of the PD-L1 protein on the MP to the PD-L1 conjugate displayed on the surface of the bacterial EV. The captured EVs can then be labeled with enzyme-linked detection antibodies (e.g., anti-GroEL, anti-Enolase, anti-OmpA, anti-his) to detect the presence of a universal bacterial EV marker or a designed protein scaffold marker. The MP-EV complex can then be magnetically concentrated on a sensing electrode; redox reactions and electron transfer from the electrode generate an electrical signal as the analytical readout. Figure 2 and Figure 10 The readings can be further normalized to the bacterial concentration in the cell culture. Bacterial EVs exhibiting functional PD-L1 binding on their surface will bind to PD-L1 and generate a signal above the reference threshold; bacterial EVs without functional PD-L1 binding on their surface cannot bind to PD-L1 and only generate a background signal below the reference value. Figure 10The binding of engineered bacterial membrane vesicles to PD-L1 can also be verified using routine immunoassays (such as enzyme-linked immunosorbent assay (ELISA)) or flow cytometry through methods similar to those described above.

[0093] In some embodiments, the size determination and quantification of engineered vesicles can be performed using nanoparticle tracking and analysis software on NanoSight. Samples can be diluted, tested, and equilibrated at room temperature (RT) before analysis.

[0094] Example 2: Preparation of E. coli vesicles capable of binding Claudin 18.2 In some embodiments, codon-optimized genes encoding polypeptide sequences (SEQ ID NO: 22, 26, 32, 38, 44, and 50 and their conserved mutants) can be synthesized and cloned into the pET-29b (+) *E. coli* plasmid expression vector. The plasmid can be transformed into chemically competent *E. coli* Lemo21 cells. The bacteria can be cultured at 37°C (or 30°C) in LB medium containing 50 μg / mL kanamycin and 30 μg / mL chloramphenicol. The cells can then be cultured at 37°C (or 30°C) in TB medium containing 50 μg / mL kanamycin, 30 μg / mL chloramphenicol, 2 mM MgSO4, and 0.5 mM L-rhamnose until the OD600 reaches 0.4–0.6, followed by IPTG induction for 12–20 hours.

[0095] In some embodiments, bacteria can be precipitated by centrifugation at 4300 xg for 20 minutes at 4°C. The culture supernatant can be collected and filtered through a 0.45 μm vacuum filter. The filtered supernatant containing membrane vesicles can be used directly for downstream analysis or further purified using a membrane vesicle isolation kit (e.g., SBI ExoBacteria OMV Isolation Kit). For certain downstream analyses (e.g., nanoparticle tracking analysis, RNA loading), an additional buffer exchange step can be performed to use an appropriate buffer.

[0096] In some embodiments, automated magnetoelectric sensing can be used to analyze clarified bacterial supernatant or purified membrane vesicles to test for binding to Claudin 18.2. This assay can begin by capturing bacterial membrane vesicles (EVs) directly from clarified culture supernatant or purified bacterial membrane vesicle solution using magnetic particles (MPs), based on the binding of the Claudin 18.2 protein on the MP to the Claudin 18.2 conjugate displayed on the surface of the bacterial EVs. The captured EVs can then be labeled with enzyme-linked detection antibodies (e.g., anti-GroEL, anti-Enolase, anti-OmpA, anti-his) to detect the presence of a universal bacterial EV marker or a designed protein scaffold marker. The MP-EV complex can then be magnetically concentrated on a sensing electrode; redox reactions and electron transfer from the electrode generate an electrical signal as the analytical readout. Figure 2 and Figure 11 The readings can be further normalized to the bacterial concentration in the cell culture. Bacterial EVs exhibiting functional Claudin 18.2 conjugates on their surfaces will bind to Claudin 18.2 and generate a signal above the reference threshold; bacterial EVs without functional Claudin 18.2 conjugates on their surfaces will not bind to Claudin 18.2 and will only generate a background signal below the reference value. Figure 11 The binding of engineered bacterial membrane vesicles to Claudin 18.2 can also be verified using routine immunoassays (such as enzyme-linked immunosorbent assay (ELISA)) or flow cytometry via methods similar to those described above.

[0097] In some embodiments, the size determination and quantification of engineered vesicles can be performed using nanoparticle tracking and analysis software on NanoSight. Samples can be diluted, tested, and equilibrated at room temperature (RT) before analysis.

[0098] Example 3: Cellular uptake of engineered E. coli membrane vesicles (EVs) In some embodiments, modified eGFP mRNA or eGFP circular RNA can be synthesized via in vitro transcription and encapsulated into engineered *E. coli* EVs via electroporation. For example, electroporation can be performed under different settings to achieve optimal encapsulation efficiency. Examples include, but are not limited to, 30 ms at 0.4 kV / cm, 30 ms at 0.44 kV / cm, 30 ms at 0.53 kV / cm, 40 ms at 0.67 kV / cm, 10 ms at 0.8 kV / cm, 25 ms at 0.8 kV / cm, 20 ms at 1.2 kV / cm, 12 ms at 1.7 kV / cm, 10 ms at 2.3 kV / cm, 2 ms at 10 kV / cm, 2 ms at 21 kV / cm, and other suitable combinations. Each set of electroporation conditions may result in different loading efficiencies. Electroporation can be performed using commercially available Invitrogen™ Neon™ transfection systems, BTX ECM630, Bio-Rad GenePulser, or custom electroporation equipment.

[0099] In some embodiments, human cells (e.g., RKO colon cancer cells, PD-L1 positive) can be cultured in 96-well plates at 37°C in growth medium. eGFP RNA-encapsulated *E. coli* EVs can be added to the culture medium of the human cancer cell line. GFP expression in target cells can be measured by fluorescence imaging (40x, EVOS M7000) to assess EV uptake and RNA translation. GFP expression in target RKO cells can be detected 24 hours after the addition of eGFP mRNA-encapsulated *E. coli* EVs. Figure 12 As a control, directly adding eGFP mRNA to the cell culture medium did not lead to GFP expression. As another control, mixing eGFP mRNA with *E. coli* EVs without electroporation also did not lead to GFP expression.

[0100] Example 4: Cellular uptake of microalgal membrane vesicles (EVs) In some embodiments, two wild-type microalgal strains (Spirulina and Tetraselmis Chuii) can be purchased from Algae Research Supply and cultured under their standard culture conditions. The supernatant of the growth medium can be harvested from the algal culture and centrifuged first at 500g for 5 minutes at 4°C, then at 3000g for 20 minutes at 4°C. The supernatant can then be filtered through a 0.45μm vacuum filter. The filtered supernatant containing membrane vesicles can be used directly for downstream analysis or further purified using a membrane vesicle isolation kit (e.g., SBI ExoBacteria OMV Isolation Kit). For certain downstream analyses (e.g., nanoparticle tracking analysis, RNA loading), an additional buffer exchange step can be performed to use appropriate buffers.

[0101] In some embodiments, the size and quantification of microalgal vesicles can be performed using nanoparticle tracking and analysis software on NanoSight. Samples can be diluted, tested, and equilibrated at room temperature (RT) before analysis. Figure 1 ).

[0102] In some embodiments, modified eGFP mRNA or eGFP circular RNA can be synthesized via in vitro transcription and encapsulated into microalgae EVs via electroporation. For example, electroporation can be performed under different settings to achieve optimal encapsulation efficiency. Examples include, but are not limited to, 30 ms at 0.4 kV / cm, 30 ms at 0.44 kV / cm, 30 ms at 0.53 kV / cm, 40 ms at 0.67 kV / cm, 10 ms at 0.8 kV / cm, 25 ms at 0.8 kV / cm, 20 ms at 1.2 kV / cm, 12 ms at 1.7 kV / cm, 10 ms at 2.3 kV / cm, 2 ms at 10 kV / cm, 2 ms at 21 kV / cm, and other suitable combinations. Each set of electroporation conditions may result in different loading efficiencies. Electroporation can be performed using commercially available Invitrogen™ Neon™ transfection systems, BTX ECM 630, Bio-RadGenePulser, or custom electroporation equipment.

[0103] In some embodiments, human cells (e.g., RKO colon cancer cells, PD-L1 positive) can be cultured in 96-well plates in a growth medium at 37°C. Microalgal EVs encapsulated with eGFP RNA can be added to the culture medium of the human cancer cell line. GFP expression in the target cells can be measured by fluorescence imaging (40x, EVOS M7000) to assess EV uptake and RNA translation.

[0104] Figure 13This is a flowchart illustrating the design and preparation of protein conjugates and their application in analyte capture, detection, and blocking. Method 1300 may include: 1) step 1302, calculating and designing a protein conjugate that binds to a portion of a target analyte; 2) step 1304, expressing and purifying the protein conjugate from *E. coli* or other microbial protein expression systems; 3) performing one or more of the following steps: step 1306, immobilizing the purified protein conjugate onto a plastic surface, magnetic particles, or a biosensor to capture cells or extracellular vesicles expressing the target protein, or to capture the free target protein; step 1308, labeling the purified protein conjugate with a biotinylated or fluorescent tag and using the labeled protein conjugate as a target detection agent in fluorescence microscopy, flow cytometry, or immunoassay; or step 1310, treating cells with the purified protein conjugate to block the interaction between the surface-expressed target protein and molecules including antibodies, antibody-drug conjugates, or natural ligands.

[0105] It should be understood that the detailed description section, rather than the abstract and summary sections, is used to interpret the claims. The abstract and summary sections may list one or more, but not all, exemplary embodiments conceived by the inventors, and are therefore not intended to limit the claims of this disclosure or the appendix in any way.

[0106] While this disclosure has been described herein with reference to exemplary embodiments in exemplary fields and applications, it should be understood that this disclosure is not limited thereto. Other embodiments and modifications thereof are possible and are within the scope and spirit of this disclosure. For example, without limiting the generality of this paragraph, embodiments are not limited to the software, hardware, firmware, and / or entities illustrated and / or described herein. Furthermore, embodiments (whether or not explicitly described herein) have significant utility in fields and applications beyond those described herein.

[0107] This document describes embodiments using functional building blocks, illustrating the implementation of specified functions and their relationships. For ease of description, the boundaries of these functional building blocks are arbitrarily defined. Alternative boundaries can be defined as long as the specified functions and their relationships (or their equivalents) are appropriately executed. Furthermore, alternative embodiments may execute functional blocks, steps, operations, methods, etc., in a different order than that described herein.

[0108] The breadth and scope of this disclosure should not be limited by the exemplary embodiments described above, but should be defined solely by the following claims and their equivalents.

[0109] All publications, patents and patent applications mentioned in this specification reflect the ordinary knowledge of a person skilled in the art and are formally incorporated herein by reference, with the same legal effect as if each individual publication, patent or patent application were specifically and clearly indicated to be incorporated by reference.

Claims

1. An engineered delivery system, comprising: Bacterial membrane vesicles originating from bacteria; as well as One or more nonbacterial proteins used for anchoring to the bacterial membrane vesicles.

2. The engineered delivery system according to claim 1, wherein, The one or more nonbacterial proteins include mammalian membrane-associated proteins or fragments thereof.

3. The engineered delivery system according to claim 1, wherein, The one or more nonbacterial proteins are further linked directly or via linker peptides to polypeptide conjugates.

4. The engineered delivery system according to claim 3, wherein, The polypeptide conjugate is displayed on the outer side of the vesicle membrane.

5. The engineered delivery system according to claim 3, wherein, The linker peptide is a glycine-serine linker peptide.

6. The engineered delivery system according to claim 5, wherein, The glycine-serine linker peptide is GGGGS.

7. The engineered delivery system according to claim 3, wherein, The polypeptide conjugate is a synthetic polypeptide.

8. The engineered delivery system according to claim 3, wherein, The polypeptide conjugate includes a nucleic acid binding domain.

9. The engineered delivery system according to claim 3, wherein, The polypeptide conjugate comprises an amino acid sequence as shown in SEQ ID NO:4, 5, 22, 23, 24 or 61.

10. The engineered delivery system according to claim 1, wherein, The bacteria in question are Gram-negative bacteria.

11. The engineered delivery system according to claim 10, wherein, The Gram-negative bacteria mentioned are Escherichia coli.

12. The engineered delivery system according to claim 1, wherein, The one or more nonbacterial proteins include mammalian protein voltage-dependent anion-selective channel 1 (VDAC1), mitochondrial carrier homolog 2 (MTCH2), or acyl-CoA synthase long chain family member 1 (ACSL1) or a fragment thereof.

13. The engineered delivery system according to claim 12, wherein, The mammalian protein comprises an amino acid sequence as shown in SEQ ID NO: 1, 2 or 3.

14. The engineered delivery system according to claim 7, wherein, The complete amino acid sequence of the nonbacterial protein is shown in SEQ ID NO: 6, 7, 8, 9, 10, 11, 12, 13, 25, 27, 29, 31, 33, 35, 37, 39, 41, 43, 45, 47, 49, 51, 53, 55, 57 or 59.

15. The engineered delivery system according to any one of claims 1 to 14, further comprising nucleic acids, proteins, complexes thereof, or combinations thereof located within the bacterial membrane vesicles.

16. The engineered delivery system according to claim 15, wherein, The nucleic acid is mRNA, circular RNA, or antisense oligonucleotide.

17. A synthetic polypeptide that binds programmed death-ligand 1 (PD-L1), comprising an amino acid sequence as shown in SEQ ID NO: 4, 5, 23 or 24, including its conserved mutants.

18. The synthetic polypeptide according to claim 17, wherein, The synthetic polypeptide is linked to a small molecule tag located at the N-terminus or C-terminus of SEQ ID NO: 4, 5, 23 or 24.

19. The synthetic polypeptide according to claim 17 further comprises a second polypeptide located at the N-terminus or C-terminus of SEQ ID NO: 4, 5, 23 or 24.

20. The synthetic polypeptide according to claim 17 further comprises a bacterial signal peptide.

21. The synthetic polypeptide according to claim 17, wherein, The complete amino acid sequence is shown in SEQ ID NO: 14, 15, 16, 17, 18, 19, 20, 21, 28, 30, 34, 36, 40, 42, 46, 48, 52, 54, 56, 58, 60, 62, 63 or 64.

22. A synthetic polypeptide that specifically binds to Claudin 18.2, comprising the amino acid sequence shown in SEQ ID NO: 22, including its conserved mutants.

23. The synthetic polypeptide according to claim 22, wherein, The synthetic polypeptide is further linked to a small molecule tag located at the N-terminus or C-terminus of SEQ ID NO:

22.

24. The synthetic polypeptide according to claim 22 further comprises a second polypeptide located at the N-terminus or C-terminus of SEQ ID NO:

22.

25. The synthetic polypeptide according to claim 22 further comprises a bacterial signal peptide.

26. The synthetic polypeptide according to claim 22, wherein, The complete amino acid sequence is shown in SEQ ID NO: 26, 32, 38, 44 or 50.

27. A method for engineering microbial vesicles, comprising: Computational design includes human membrane anchor peptides, synthetic peptide conjugates, and membrane vesicle (EV) scaffold proteins with bacterial signaling peptide domains; The membrane vesicle scaffold protein was cloned into a bacterial expression vector; The membrane vesicle scaffold protein is expressed in Escherichia coli or other Gram-negative bacteria; Purification of the membrane vesicle scaffold protein from bacterial expression cultures; and Verify the targeted binding and cellular uptake capabilities of the described membrane vesicle scaffold protein.

Images