Engineering bacterial membrane nanovesicles for improved therapies in infectious diseases and cancer

Abstract

Research on bacterial membrane vesicles (BMVs) is an emerging topic, and the goal is to address whether BMVs can bring translational tools to improve current therapies. In this review, we provided the updated studies on BMVs including their production, their types, and therapeutic regimens for treating infectious diseases and cancers. We described several platforms of BMVs, such as outer membrane vesicles (OMVs), inner membrane vesicles (IMVs) and double membrane vesicles (DMVs), and those structures were produced from Gram-negative or Gram-positive bacteria. We also discussed how to engineer and formulate new and novel BMVs using chemical, physical, and genetic methods. For therapies, we analyzed current methods for loading drugs in BMVs and discussed their limitations. Finally, we reviewed several therapeutic platforms of BMVs that have been exploited in improving the treatments of infectious diseases and cancers. Although BMVs offer the promising biomedical applications, it is needed to develop rigorous approaches and methods to generate reproducible and scalable drug delivery systems for translation.

1. Introduction

Living cells secrete membrane-formed vesicles (MVs) that contain biologically active molecules, and these vesicles play a central role in mediating intercellular communications and cellular functions [1]. Both eukaryotic and procaryotic cells can actively or passively produce MVs. Membrane-formed vesicles from eukaryotic cells have been investigated [2–4]. Based on their biogenesis and sizes, there are three types of membrane vesicles: exosomes, microvesicles and apoptotic cell bodies. It is challenging to distinguish these three types of vesicles because they have similar sizes and lipid membrane properties. Therefore, they are usually called extracellular vehicles (EVs). EVs can be produced from a wide range of eukaryotic cells, such as immune cells (neutrophils [5], monocytes [6] or macrophages [7]), red blood cells [8,9] and stem cells [10,11]. EVs also exist in a variety of tissues [12] and circulation systems [13]. EVs can be engineered and loaded with cargos for treating acute lung inflammation [14–17], stroke [18], sepsis [19], and cancer [20]. Several reviews have covered this topic [4,21,22]. MVs derived from procaryotic cells become an emerging research topic and they show the unique features different from EVs derived from mammal cells [23]. This review is focused on discussing bacterial membrane-derived vesicles (BMVs).

The first discovery of BMVs produced by Gram-negative bacteria was traced back more than 50 years ago [24] (Fig. 1). Ten years later, this observation was further confirmed and the interest in BMVs dramatically increased [25,26]. The studies found that Gram-negative and Gram-positive bacteria spontaneously release BMVs in culture media, but interestingly, the production of BMVs is increased when bacteria are stimulated or activated [27].

Fig. 1.

Milestones on BMVs derived from Gram-negative and Gram-positive bacteria. BMVs, bacterial membrane vesicles; OMVs, outer membrane vesicles; IMVs, inner membrane vesicles; DMVs, double membrane vesicles.

BMVs were initially considered as cellular debris or artifacts because they were made of bacterial membrane, unlikely possessing biological functions [28]. Recent studies and advances in microscopy suggest that BMVs are important mediators in regulating interactions between the host and bacteria. For example, BMVs remove pathogen-stimulating molecules released from bacteria in antimicrobial effects. In other hand, BMVs may support bacterial survival through transfer of genes and sequestering of nutrients to assist the biofilm formation. Interestingly, BMVs also support the development of antimicrobial resistance [29–32]. Fig. 1 shows the milestones of BMVs made from Gram-negative or Gram-positive bacteria. BMVs derived Gram-negative bacteria have been intensively studied, such as the discovery and biogenesis of OMVs, and creating inner membrane vesicles (IMVs) and double membrane vesicles (DMVs). The proteomics shows the unique properties of BMVs, and these properties enable to develop a wide range of biomedical applications, such as targeted drug delivery. In contrast, BMVs made from Gram-positive bacteria are less studied, but it is an emerging area to expand the biomedical applications of BMVs.

Advances in microbiology, immunology and microscopy have revealed many roles of BMVs in bacterial ecology, physiology, and host–microbe interactions. BMVs are also strongly correlated to disease development, so BMVs have demonstrated the potential applications in treating infectious diseases and inflammatory disorders. Furthermore, advances in nanotechnology, bioengineering and genetic modifications allow us to design and fabricate varied types of bacterial membrane structures for targeted drug delivery and vaccine development. There are several reviews on outer membrane vesicles (OMVs) and their biomedical applications [33–35]. The growth on studying BMVs is very rapid and there are many emerging technologies and approaches used to engineer BMVs.

In this review, we expand a scope from the discovery of BMVs to their biogenesis and engineering. We also discuss their applications in improved therapies to infectious diseases and cancers. Furthermore, we describe several studies to illustrate the concept on designing and engineering BMVs for vaccine development and targeting tumor microenvironments for improved cancer treatment. Finally, we discuss the barriers to translate the technologies of BMVs in targeted drug delivery and we also offer our views on opportunities for BMVs.

2. Generation of BMVs

Bacteria bud their membrane and release them into surrounding environments and this process is a spontaneous event. After these components are isolated and purified, they show lipid bilayer-enclosed structures that are morphologically similar to eukaryotic EVs [14,15,36]. Bacteria release BMVs with sizes with 20 to 400 nm in diameter [37,38]. Both Gram- negative and Gram-positive bacteria can spontaneously secrete BMVs. Their secretion is also dependent on surrounding environments, affecting diverse biological processes, such as virulence, horizontal gene transfer, export of cellular metabolites, phage infection and cell-to-cell communications [39] (Fig. 2).

Fig. 2.

The components and functions of BMVs. BMVs are composed of phosphatidylglycerol, phosphatidylethanolamine, and membrane of proteins and lipopolysaccharides. Nucleic acids, metabolites and metal ions are incorporated in BMVs during their formation. The functions of secretion of BMVs are for bacterial growth (such as bacterial survival and nutrition acquisition) and for immune responses (signal communication and material transfer).

There are two types of bacteria based on gram staining, such as, Gram-negative, and Gram-positive. Their structural difference is that Gram-negative bacteria have a thin peptidoglycan layer and an outer lipid membrane while Gram-positive bacteria have a thick peptidoglycan layer and no outer lipid membrane. Therefore, the formation and structures of BMVs depend on bacterial types. Fig. 3 shows the different types of BMVs made from Gram-negative or Gram-positive bacteria.

Fig. 3.

Schematic illustration of varied types of BMVs made from Gram-negative or Gram-positive bacteria and their applications in infections and cancers. The outer membrane of Gram-negative bacteria forms OMVs. IMVs (inner membrane vesicles) and DMVs (double membrane vesicles) are created using biological or physical methods. BMVs are also produced from genetically engineered Gram-negative and Gram-positive bacteria.

2.1. BMVs formed from Gram-negative bacteria

2.1.1. OMVs

BMVs derived from Gram-negative bacteria have been extensively investigated [21,40,41]. Gram-negative bacteria possess a distinctive structure made of lipopolysaccharide (LPS) and an outer lipid membrane. Gram-negative bacteria spontaneously release the outer lipid membrane to form membrane vesicles. This structure is called OMVs (outer membrane vesicles) [38,39]. The production process of OMVs is comprised of the following steps: 1) The membrane proteins connect an outer lipid membrane to peptidoglycan to form a stable Gram-negative envelope. The proteins begin to destabilize during the formation of OMVs. 2) The linking proteins between the outer lipid membrane and peptidoglycans are disassociated by either relocation of the linking proteins or by breaking of the connection between outer lipid membrane and peptidoglycans. 3) OMVs are released after the breaking of linking proteins. The formation of OMVs strongly depends on local environments, so OMVs may contain different cargos, such as proteins or genetic molecules. Proteins linking the outer lipid membrane and peptidoglycans may be also present in OMVs.

The mechanisms of formation of OMVs are complex so that OMVs are heterogeneous in their sizes and compositions [31,38]. OMVs are comprised of phospholipids, proteins, nucleic acids, proteins, and virulence factors (such as LPS) [42]. Sometimes, metal ions, signaling molecules and metabolites are also present in OMVs [43,44] (Fig. 2). Since proteins play an important role in OMVs for tissue targeting and signal transduction, the proteomics have been performed to analyze proteins in OMVs. There are several approaches and methods, for example, sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS-PAGE) [45,46], 2D-electrophoresis [47] followed by in situ mass spectrometry (MS) [45] or direct trypsin digestion of electrophoresis gels followed by liquid chromatography (LC)-MS/MS [46,48]. The proteomics show that OMVs are mainly composed of outer membrane-associated proteins rather than from either the inner membrane or cytoplasm. The findings strongly support the hypothesis of OMVs biogenesis.

Formation of OMVs from Gram-negative bacteria strongly depend on local cues, such as stress [49], and stimulators [50] (Table 1). Therefore, the optimization of culture media and conditions may increase the production of OMVs and control their compositions.

Table 1.

Current technologies used to enhance the production of BMVs.

Types	Methods	Genes/Agents	Strains	Key Results	Ref.
OMVs	Biological	Mutation of tolR and galU	Shigella sonnei	Increased vesiculation	[94]
		Mutation of tolR gene	Helicobacter pylori	Increased vesiculation	[93]
		T4 phage infection	Escherichia coli	Enhanced generation of OMVs	[79]
	Chemical	IPTG	Escherichia coli	Enhanced generation of OMVs	[49]
		Gentamicin	Pseudomonas aeruginosa	Production of OMVs increased by 3 times	[78]
		Sodium bicarbonate	Vibrio cholerae	Increased production of OMVs	[82]
		Arabinose	Escherichia coli	Increased production of OMVs and concentration dependent	[49]
		Cysteine	Neisseria meningitidis	Depletion of cysteine increased OMVs	[77]
		Oxygen	Neisseria meningitidis	Oxygenation increased release of OMVs	[83,84]
	Temperature Stress	Heat shock	Yersinia pestis;	Increased production of OMVs	[50]
			Escherichia coli	Increased production of OMVs	[49]
			Escherichia coli	Increased membrane blebs	[68]
		Cold shock	Yersinia pestis	Increased production of OMVs in low temperature	[50]
	Growth Conditions	Bacterial density	Shigella sonnei	Bacterial growth density mediated production of OMVs	[94]
		Increased volumes	Neisseria meningitidis	Continuous culture achieved 4.0 × 1014 OMVs per liter per day	[88]
IMVs	Physical	Extrusion	Escherichia coli	Production of IMVs	[51]
DMVs	Physical	Cavitation	Pseudomonas aeruginosa	High yield of BMVs	[19,37]

Note: Bacterial membrane vesicles (BMVs); Outer membrane vesicles (OMVs); Inner membrane vesicles (IMVs); Double membrane vesicles (DMVs); Isopropyl ß-D-1-thiogalactopyranoside (IPTG).

2.1.2. Inner membrane vesicles (IMVs)

The inner membrane of Gram-negative bacteria is covered by the outer membrane and the peptidoglycan layer, so it is not easy to form membrane vesicles naturally. Because OMVs contain LPS, they may cause the immune toxicity in their medical applications. Kim et al. [51,52] proposed to generate inner membrane vesicles from Gram-negative bacteria, so called bacterial protoplast-derived nanovesicles (PDNVs). Protoplast is a state of bacteria without a toxic outer membrane and a peptidoglycan layer (cell wall) that were removed by lysozyme. After the outer membrane was removed, the protoplast generated IMVs. The IMVs were successfully produced by a serial extrusion approach and were used to develop a universal adjuvant-free vaccine. The IMVs showed the higher efficacy and a safety profile than the vaccine developed using OMVs. In addition, IMVs induced the strong antigen-specific humoral and cellular immune responses.

2.1.3. DMVs

OMVs or IMVs are made from a single lipid membrane, thus their stability and cargo loading may be problematic as drug delivery systems. Gram-negative bacteria have a distinct structure with a double-layer lipid membrane linked by peptidoglycans, and integrated membrane contains the functions of parent bacteria. Developing vesicles made from integrated bacterial membrane may be critical for medical applications of BMVs [53,54]. Wang et al. [37] proposed an innovative approach using nitrogen cavitation to generate DMVs. Using cryogenic transmission electron microscopy (cryo-TEM), biochemistry and proteomics, the authors have shown that DMVs possessed the whole bacterial membrane including the inner and outer lipid membranes and two membrane linkers of peptidoglycans. The unique feature is that DMVs contained many antigens required for development of vaccines. In a mouse model of sepsis induced by Pseudomonas aeruginosa, it was found that DMVs dramatically increased mouse survival compared to OMVs made from Pseudomonas aeruginosa. The molecular mechanism of improved survival is associated with the enhanced adaptive immunity and distinct biodistribution of DMVs compared to OMVs, possibly due to more pathogen-associated molecular patterns (PAMPs) on DMVs. This study shows that the nitrogen cavitation approach would be a promising technology to make DMVs from a wide range of bacteria.

The same research group extended DMVs to target multiple cells (immune cells and endothelium) in tumor microenvironments [20]. In the study, the authors designed bacterial (Escherichia coli) DMVs expressing arginine-glycine-aspartate (RGD) peptides and with endogenous targeting ligands from bacteria. The DMVs can spontaneously bind to neutrophils, monocytes, and endothelial cells in tumor microenvironments. Importantly, the authors can load doxorubicin (Dox) inside DMVs via a pH gradient, and the loading efficiency of Dox was very high at 12% (w/w). In a mouse model of melanoma, DMVs loaded with Dox dramatically shrank the tumor sizes compared to DMVs without expression of targeting ligands. This study demonstrates that DMVs are novel targeted drug delivery carriers to efficiently deliver therapeutics for cancer therapies.

2.2. BMVs made from Gram-positive bacteria

Compared to Gram-negative bacteria, Gram-positive bacterial membrane vesicles are recently attractive and become an emerging topic [55]. There are more than 30 species of Gram-positive bacteria have shown that they are capable of secreting BMVs [56].

Gram-positive BMVs contain a wide range of cargo molecules, such as nucleic acids, proteins, lipids, enzymes, and toxins [55,57] (Fig. 2). Compared to OMVs, Gram-positive BMVs lack of LPS and periplasmic components (Fig. 3). Studies have shown that the formation of BMVs is regulated by an active metabolic process.

The mechanism of biogenesis of Gram-positive BMVs remains unclear. Unlikely the biogenesis of OMVs, OMVs are pinched off the outer membrane of Gram-negative bacteria because Gram-positive bacteria possess a thick cell wall made of peptidoglycans that may inhibit the budding of lipid membrane from Gram-positive bacteria. One hypothesis is that enzymes may destroy the peptidoglycan layer, and then exposed lipid membrane may be easy to bud to form BMVs [55,58].

The functions of Gram-positive BMVs include transport of molecules and increased bacterial survival [59] (Fig. 2). For example, it was observed that bacterial chromosomal DNA was transported in Ruminococcin spp [60] and BMVs transported surface receptors [61]. Interestingly, Gram-positive bacteria also secrete BMVs containing nutrient uptake enhancer (Iron-binding factor) [62], antibiotic degrading enzymes (β-lactamase) [63] and even toxic factor to enhance bacterial survival [64,65]. The sizes of Gram-positive BMVs range in 10–400 nm [56].

3. Engineering BMVs

Advances in bioengineering and nanotechnology enable to engineer BMVs to solve the limitations on BMVs, for example, low production yields, toxicity and tissue targeting. Here we discuss several approaches and methods used in engineering BMVs for solving these barriers.

3.1. Enhanced production of BMVs

BMVs are secreted during bacterial growth, but this type of production of BMVs results in the challenges for biomedical applications. Various strategies have been developed to solve these problems. Several approaches and methods were used by selecting bacterial strains and changing culture environments.

OMVs constitutively release during the bacterial growth [66], and interestingly, bacterial strains strongly affect the production of OMVs. For example, Escherichia coli [67,68] Salmonella typhimurium [69,70], Borrelia burgdorferi [71], and Gallibacterium anatis [72] were observed to produce more OMVs. In addition, it was reported that the late culture phase in Neisseria meningitides [73] and Francisella novicida [74] produced more OMVs. Another bacterium, Legionella pneumophila formed OMVs and their growth was related to all phases of the cell growth cycle [75], while for Helicobacter pylori, the release of OMVs was inversely dependent on the cell growth. Interestingly, the production of OMVs was also correlated with bacterial shape transitions [76].

The generation of OMVs is very sensitive to culture environments, such as temperature [49,50,68], lacking of amino acids [77], antibiotics [78], or phage infection [79]. It was shown that the cysteine depletion caused the oxidative stress that mediated the secretion of OMVs in Neisseria meningitidis [77]. Interestingly, adding chemicals in culture media can also induce more production of OMVs, such as supplement of hexadecane in Acinetobacter calcoaceticus [80], glucose in Yersinia pestis [81] and sodium carbonate in Vibrio cholera [82]. Oxidative stress can also induce the release of OMVs in Neisseria meningitidis [83].

To increase the yield of OMVs, detergents or ethylenedi-aminetetraacetic acid (EDTA) [40,84] was used to extract more OMVs from bacterial membrane, but the production is still limited by the culture process. Thus, the fermentation with bioreactors [85–87], were employed. A continuous generation of Neisseria meningitidis OMVs is a way to increase culture volumes. This process can be performed for more than 600 h and the production yield was at 4.0 × 10¹⁴ OMVs per liter a day. The results showed that features of OMVs were similar to those produced by traditional culture procedures, suggesting that the bioreactors may be feasible for the scalable production of OMVs for medical applications [88].

The generation of OMVs is dependent on peptidoglycans between inner and outer lipid membrane, and some proteins regulate the peptidoglycans. The Tol-Pal system regulated the membrane peptidoglycans [89]. The studies show that the Tol-pal operon regulated seven downstream genes, and five genes encoded proteins for a Tol-Pal system, called tolQ, tolR, tolA, tolB, and pal. These 5 proteins regulated the assembly a complex of peptidoglycan, and controlled the outer membrane tightness [90]. Mutation of these genes may cause a malfunction of the protein complex, resulting in increased vesiculation of bacteria [91,92]. For example, deletion of tolB increased vesiculation of Helicobacter pylori [93]. Similarly, simultaneous mutation of the tolR gene and the galU gene was also found to increase the production of OMVs in Shigella sonnei [94]. Local protein aggregations may also impair the Tof-Pal system, leading to increased OMV production. It was also observed that a high level of TolR induced vesiculation in Escherichia coli [95].

Recently, the protoplast membrane vesicles [51] and double membrane vesicles derived from a whole bacterial membrane [20,37] were also created. These new types of BMVs demonstrated the high production and scalability for potential medical applications.

We summarized the approaches and methods used in increasing the production of BMVs in Table 1. Current technologies including chemical and biological methods show a promise on translating BMVs in clinic.

3.2. Reduction of virulence in BMVs

BMVs can cause immune responses when they are used as drug carriers because they contain antigens expressing on bacterial membrane. The research is focused on how to reduce this virulence when BMVs are produced. There are two major resources of toxins: LPS (lipopolysaccharide) and toxic proteins in BMVs, either on the membrane or inside BMVs. It is needed to develop novel approaches and methods to attenuate immunogenetic agents in BMVs when they are used as drug delivery carriers.

LPS is a major component of the outer membrane on Gram-negative bacteria, comprising lipid A, core oligosaccharides, and O-antigen [96,97]. When LPS binds to a receptor complex consisting of TLR4, CD14 and MD2, myeloid differentiation factor (MyD88) activates nuclear factor-κB (NF-κB) and mitogen-activated protein kinases, leading to inflammatory responses. The inflammatory responses may cause a severe disease, such as sepsis [98,99]. Thus, removing LPS from BMVs is critical to develop BMVs as drug delivery vehicles.

To reduce LPS levels on BMVs, controlling the expression of LPS on bacteria may be a strategy. Genetic modification of acylation of the lipid A moiety showed the reduced inflammatory responses in BMVs [100]. Lipid A needs enzymes for its biosynthesis [96], therefore, downregulating these enzymes may be a novel approach to control the synthesis of lipid A. Knockout of lpxL and lpxM have demonstrated the low toxicity [101–103]. The modified phosphorylation of a lipid A component may be also a novel approach to lower the toxicity of LPS. The lipid A was phosphorylated, and it was observed that monophosphorylated lipid A is less immunogenic [104]. Monophosphorylated lipid A (MPL) has been approved by FDA for the use as an adjuvant [105].

Endotoxin removal by detergent-extraction was used in the first vaccine made from OMVs, and this vaccine has successfully prevented the outbreak of Neisseria meningitidis serogroup B [106]. Based on the structural features of LPS, non-ionic detergents and chelating agents were used for sequential reduction of LPS during the production of OMVs.

There are other virulence factors including collagenase and hyaluronate lyase that disrupt extracellular matrix (ECM) [107], and serine proteases [108]. The exfoliative toxins [109] can disrupt blood barriers. It is shown that BMVs made from Bacillus anthracis, Staphylococcus aureus, Streptococcus pneumoniae, Streptococcus pyogenes, and Streptococcus agalactiae, carried hemolysins and/or pore forming toxins [64,109–111]. Other membrane proteins, VacA, OstA, FilC, pilA, PagL, EstA and aaaA were also reported to be toxic [38]. Genetic engineering, as well as natural strain selectivity, may be a good strategy for the reduction of toxins.

3.3. Engineering BMVs for antigen presentation

Peptides and proteins can be engineered on the bacterial surface by fusing them with an anchoring motif. For example, the Mycobacterium tuberculosis antigens ESAT6, Ag85B, and Rv2660c were linked to the surface of Escherichia coli OMVs via the fusion to hemoglobin protease (Hbp). The OMVs decorated with multiple heterologous antigens caused antigen-specific immune responses, thus they may be novel vaccines to prevent the infection from Mycobacterium tuberculosis [112].

It is noticed that configuration of antigens on the surface of OMVs is important in immune responses. For example, OspA (outer surface protein A) was fused to different regions of factor H binding protein (fHbp) (a meningococcal surface-exposed lipoprotein) and was tested whether OspA was displayed on the surface of OMVs. When mice were immunized with the OMVs, the OspA-specific antibodies were observed. This study indicated that the correct expression of antigen on the surface of OMVs was critical to develop a vaccine system [113].

To extend the Hbp platform, heterologous proteins were coupled to internal sites of the Hbp passenger. For example, SpyTags were coupled to the Salmonella OMV surface and efficiently interacted with the SpyCatcher fusion proteins. These studies show that a modular Hbp system was established for developing recombinant OMVs-based vaccines and therapies [114,115].

3.4. Engineering BMVs for targeting

Living bacteria naturally exhibit tumor tropism because bacteria favor the hypoxia where tumor cells exist [116], thus bacterium-based anticancer therapies demonstrate the benefits overcoming the limitations of current cancer therapies. Clinical trials showed that it was safe to administrate attenuated strain of bacteria, but with the limited antitumor effect [117]. These studies suggest that it is needed to increase bacterial targeting. Genetical modifications using bioengineering allows to develop new and novel bacterium-derived therapeutic systems.

Proteins or peptides can be genetically engineered on the surface of BMVs. Bacterial OMVs fused with heterologous proteins were studied using a system of vesicle-associated toxin ClyA of Escherichia coli [118]. In the study, it is found that chimeric ClyA fusion proteins were localized on the surface of bacterial OMVs and retained its activity, demonstrating that ClyA is a useful tool to modify the surface of bacteria and their OMVs. Using this tool, they fused a fragment of anti-digoxin single-chain Fv to the C terminus of ClyA to make a designer, so called “immuno-MVs”.

Based on this concept, Kim et al. [51] fused epithelial growth factor (EGF) with PrsA, an inner membrane protein, and prepared EGF-decorated IMVs. The A549 lung cancer expresses an EGF receptor (EGFR), therefore EGF-expressed IMVs would target the tumors. The in vitro and in vivo studies confirmed this hypothesis. Similarly, Gao et al. [20] fused a RGD peptide to the ClyA protein in bacteria, and subsequently they prepared RGD-expressed DMVs using the nitrogen cavitation method. The in vitro and in vivo studies indicated that RGD-expressed DMVs dramatically increased their tumor accumulation via multiple targeting of immune cells, tumor vasculature and tumor cells.

The motif of RGD peptide is a novel targeting molecule because they are small and specific to certain tissues [119,120]. For example, Chen et al. [121] conjugated RGD to the surface of OMVs for increased targeting of OMVs. The OMVs also carried polymeric micelles and they were tested in the synergistic therapeutic effect of OMVs. The nanocomplexes were able to accumulate in tumor tissues in the mouse breast xenograft model.

3.5. Drug loading in BMVs

The current approaches for loading drugs in BMVs include incubation, extrusion and sonication [36,38], but the loading efficiency is very low because the methods are the passive loading processes. Bacteria have a function to eliminate antibiotics via secretion of OMVs as a mechanism for antimicrobial resistance, so it is possible that antibiotics may be loaded in OMVs when bacteria secrete OMVs in the presence of antibiotics. To verify this hypothesis, the authors added gentamicin during culturing Pseudomonas aeruginosa strain PAO1, and then they harvested OMVs. They found that gentamicin was incorporated in OMVs at 4 ng of drug per μg of proteins [122,123]. Tashiro et al. [124] changed the timing of adding antibiotics during the culture of bacteria. Buttiauxella agrestis - was grown until the late phase, and then gentamicin was added at a concentration of 4 folds higher than the minimum inhibition concentration (MIC). After the OMVs were collected, they observed the loading of gentamicin in OMVs, but they did not quantitatively analyze the actual loading yield of gentamicin in OMVs. Similarly, Huang et al. [125] used Acinetobacter baumanii and added levofloxacin in the culture media at sub-MIC concentrations. They found that OMVs incorporated levofloxacin. These studies suggest that the strains, drugs, and culture conditions may affect the drug loading in OMVs.

BMVs are composed of phospholipid layers, thus hydrophobic drugs may be incorporated in the lipophilic leaflets. Using this approach, Kuerban et al. [53] loaded Dox in OMVs after Dox was incubated with OMVs in PBS at 37 °C for 4 h. The loading efficiency of Dox was estimated to be 2% after Dox was determined by LC-MS. Similarly, after the prodrug of taxol, DSEP-PEG-CA-PTX was dissolved in a small volume of OMVs, the drug was successfully loaded in OMVs [126].

BMVs are made of lipid bilayer vesicles. The lipid layer is a barrier to load therapeutics inside BMVs, so the passive loading of drugs in BMVs is not an efficient method to load the drugs. It is critical to establish novel concepts and methods to load therapeutics. Inspired by the remote loading of Dox in liposomes (Doxil^® and Doxebo^®) [127,128], Wang lab proposed to load Dox in BMVs using pH gradient [20]. To achieve the pH gradient between inner and outer BMVs, they incorporated cholesterol in lipid layers of BMVs. They found that establishing pH gradient required the incorporation of cholesterol. Using this approach, they found that Dox loading in DMVs was 12% (w/w). This drug loading level in BMVs offers the potential translation of BMVs in clinic. In addition, genetical methods were used to express the interest of proteins or genes in BMVs. We summarize the current approaches and methods used in loading drugs in BMVs as shown in Table 2.

Table 2.

Drug loading strategies of BMVs and the results.

Types	Methods	Cargo	Key results	Ref.
OMVs	Incubation	Gentamicin	4 ng/μg OMVs (protein weight)	[122]
	Incubation	Gentamicin	146 μg/ml (unknown in OMVs)	[124]
	Incubation	Levofloxacin	23.2 mg per 1.4 × 1014 OMVs	[125]
	Incubation	Dox	2% (w/w)	[53]
	Incubation/genetical	TRAIL; ICG	632 pg/μg (TRAIL)	[129]
	Incubation/electroporation	siRNA; PTX	Unknown	[126]
	Cloaking	Tegafur/micelles	7.1% (w/w)	[121]
	Hybrid/cloaking	ICG/NPs	Unknown	[130]
IMVs	Genetically engineering	GFP/OmpA SAcoagulase	Unknown	[51]
DMVs	Remote loading	Dox	12% (w/w)	[20]

Note: BMVs, bacterial membrane vesicles; OMVs, outer membrane vesicles; IMVs, inner membrane vesicles; DMVs, double membrane vesicles; Dox, doxorubicin; GFP, Green Fluorescence Protein; SAcoagulase, Staphylococcus aureus specific coagulase; ICG, indocyanine green; TRAIL, tumor necrosis factor related apoptosis-inducing ligand; PTX, paclitaxel.

4. Infectious diseases

4.1. BMVs as antigens for vaccination

Bacterial membrane vesicles, especially OMVs are derived from various pathogens, so they have been developed for immunogenic vaccines against the respective or closely related organisms [29]. Upon administration of BMVs, they induce the innate and adaptive immune responses. Antigen presentation cells (such as dendritic cells (DCs)) encounter BMVs and subsequently DCs activate T cells and B cells to produce cytokines and antibodies, thus equipping the immune system to combat pathogen infections [37].

There are several advantages of OMVs as vaccines: 1) Easily decorating OMVs with heterologous proteins for enhanced vaccine features. 2) Incorporating heterologous glycans to OMVs for development of glycoconjugate vaccines. 3) Engineering OMVs for desired immune responses. 4) Incorporating multiple antigens in BMVs to make effective vaccines and 5) Possible scalability and translation of OMVs for clinic [40]. There are several reviews [33,40,131,132] have summarized recent progress on engineering OMVs in vaccine development. Although the virulence derived from bacteria membrane is necessary to initiate immune responses, management of the toxicity levels is critical in the development of OMVs-based vaccines. Removal of other antigens is a method before OMVs are produced, for example, detergents are used to remove lipoxins of Gram-negative bacteria for the vaccines of OMVs [133].

IMVs and DMVs were designed to increase vaccine performance. Kim et al. [51] generated an adjuvant-free vaccine delivery platform by depleting the toxic outer membrane of bacteria, called PDNVs. The PDNVs exhibited the higher production and better safety profile than OMVs and induced strong immune responses. After bacterial antigens were loaded in PDNVs and PDNVs were administered to mice, immunized mice showed the effective protection to sepsis. Wang et al [37] fabricated a new novel vaccine platform, so-called DMVs (double layer membrane vesicles) derived from Pseudomonas aeruginosa, and mice were treated with DMVs against the infection by Pseudomonas aeruginosa. Compared with OMVs, the platform of DMVs increased the mouse survival and the immune response to the infection.

4.2. BMVs as adjuvants in vaccination

Protein-based vaccines often require adjuvants for increased immune responses [134,135]. The vaccine adjuvants include cholera toxin, alum, and diphtheria toxin, but they exhibit several adverse side effects such as, inflammation, toxicity, and less effective mucosal immunity [136,137]. BMVs may be the potential adjuvants due to their ability to stimulate the innate immune system. LPS and peptidoglycan present in BMVs can activate sensors of the innate immune system [138–140]. BMVs were capable of effectively inducing immune responses [141]. BMVs can also act as antigen carriers [142].

Antigens, AnAPN1 and Pfs48/45, or ovalbumin were used to evaluate OMVs derived from Escherichia coli as an adjuvant in intranasal vaccination. In the study, they compared with the adjuvant cholera toxin (CT) and parenteral adjuvant MF59C.1. The vaccine was administered intranasally or subcutaneously. When OMVs were used as an adjuvant in intranasal immunization, the antibody production and cellular responses were observed. This study provides the rationale for the future development of OMVs-based adjuvants in other infectious diseases [143].

Influenza is a major of cause for acute respiratory diseases, resulting in global health issues. The immunization is considered a promising route for vaccination [141]. OMVs can be manipulated to produce low endotoxic vaccines. This can be done through modifying the structure of LPS. In this study, the modified OMVs showed the decreased toxicity compared to native OMVs. In the immunization studies, intranasal administration of OMVs dramatically enhanced antibody production and T cell functions. Interestingly, vaccinated mice increased the survival when they were challenged at a lethal dose of viruses. These studies indicate that OMVs may be used in intranasal vaccines.

4.3. Anti-microbial effects

During the OMV formation in many gram-negative bacteria, OMVs entrap several periplasmic components, such as alkaline phosphatase, phospholipase C and peptidoglycan hydrolase [78,122,144]. For example, a study demonstrated that BMVs made from Pseudomonas aeruginosa PAO1 were able to lyse Staphylococcus aureus, Escherichia coli, and another Pseudomonas strain [122]. Pseudomonas aeruginosa OMVs possessed the most lytic activity [145].

Goes et al. [146] generated OMVs from Cystobacter velatus Cbv34 and Cystobacter ferrugineus Cbfe23 and analyzed their effects in intracellular killing of Staphylococcus aureus. The results showed that both Cbv34 and Cbfe23 OMVs killed Staphylococcus aureus via the intracellular mechanism after OMVs were taken up by macrophages and epithelial cells. Interestingly, the uptake of OMVs did not kill macrophages. The molecular mechanism of the antimicrobial feature of myxobacterial OMVs may be related to cystobactamid [147].

4.4. Probiotic-related BMVs for host defense

Lactobacilli are probiotics, and they can prevent cellular apoptosis from cytokines, thus decreasing the pathogenicity, such as Escherichia coli and vancomycin-resistant enterococcus (VRE) [148,149]. For example, Lactobacillus acidophilus NCFM prolonged the survival of nematodes when exposed to VRE. Lactiplantibacillus plantarum-derived BMVs prevented the worms from VRE invasion. The molecular mechanism for this protection may be related to upregulation of host defense genes by BMVs made from Lactiplantibacillus plantarum. The development of probiotic derived BMVs may be a new approach to treat antimicrobial resistant pathogens [150].

4.5. BMVs as antibiotic carriers

When Pseudomonas aeruginosa PAO1 bacteria were incubated with gentamicin, the bacteria secreted membrane vesicles containing gentamicin (g-MVs) and peptidoglycan hydrolase. The studies showed that g-MVs delivered gentamicin to IIIa Burkholderia cepacian. The further studies showed that strain CEP0248 was killed by g-MVs, but the strain C5424 was not because the second strain was resistant to gentamicin [123]. This passive loading of antibiotics to BMVs has the low drug loading, thus the therapeutic outcome was low [89]. To resolve this issue, Wu et al.[151] developed strategy in which they first loaded rifampicin in mesoporous silica nanoparticles and then coated the nanoparticles with OMVs. In this approach, mesoporous silica nanoparticles increased the drug loading. The authors coated the rifampicin nanoparticles using - Escherichia coli OMVs and the nanoparticle complexes had a longer drug release. The OMV-coated MSNs were taken up by Escherichia coli cells and this uptake was more efficient than uncoated nanoparticles or free drugs. However, it is hard to eliminate bioactive molecules during the production of BMVs, therefore these molecules may degrade antibiotics resulting in low efficacy in treating infectious diseases [63,152].

4.6. Phage therapy reinforcement

Phages can interact with bacterial cells. To limit Vibrio cholerae infection and the spread of cholera, new preventative methods are still required. Due to the quick production of phages, a bacteriophage (phage)-based therapy is an alternative to treat bacterial infections. However, a recent study showed that bacteria secreted OMVs to neutralize phage binding for prevention of phage predation to Vibrio cholerae. This finding supported that OMVs secreted by bacteria are decoys to prevent phage therapy [153].

We summarized the current technologies and methods used in treating infectious diseases using BMVs (as shown in Fig. 4 and Table 3). The studies are mainly focused on developing effective vaccines and delivering antibiotics using BMVs.

Fig. 4.

Molecular mechanisms in which BMVs treat infectious diseases through vaccination and direct killing bacteria. For vaccination, BMVs are recognized by DCs and subsequently DCs activate T cells and B cells to recognize and neutralize bacteria, and finally macrophages phagocytose bacteria to resolve infections. In addition, engineering BMVs may directly kill bacteria. BMVs, bacteria membrane vesicles. DCs, dendritic cells.

Table 3.

The applications of BMVs in infectious diseases.

Types	Sources	Regimens	Antigens/Drugs	Targets	Ref.
Gram-negative
OMVs	Vibrio cholerae	Vaccination	Natural PAMPs	Vibrio cholerae	[154]
	Escherichia coli	Vaccination	Genetically engineering of exogeneous FhuD2/ Csa1A/LukE/ SpAkkaa/ Hla	Staphylococcus aureus	[155]
	Campylobacter jejuni	Vaccination	Genetically overexpression of cjaA	Campylobacter jejuni	[156]
	Escherichia coli	Vaccination	Genetically engineering of SpyCEP/StrepO/ Spy0269/SAM1372/R-TEM b-lac	Streptoccus pyogenes	[157]
	Yersinia pestis	Vaccination	Genetically overexpression of LcrV	Yersinia pestis	[158]
	Neisseria meningitidis	Vaccination	Natural PAMPs with LPS detoxification	Neisseria meningitidis	[159]
	Burkholderia pseuomallei	Vaccination	Natural PAMPs	Burkholderia pseuomallei	[160]
	Edwardsiella tarda	Vaccination	Natural PAMPs	Edwardsiella tarda	[161]
	Salmonella typhimurium	Vaccination	Natural PAMPs	Falmonella enterritidis; Escheerichia coli	[162]
	Burkholderia pseudomallei	Vaccination	Natural PAMPs	Burkholderia pseudomallei	[163]
	Escherichia coli	Vaccination	Natural PAMPs	Escherichia coli	[164]
	Burkholderia pseuomallei	Vaccination	Condition tuned OMVs; Intracellular-stage proteins	Burkholderia pseuomallei	[165]
	Acinetobacter baumanni	Vaccination	Natural PAMPs with LPS detoxification	Acinetobacter baumanni	[166]
	Gallibacterium anatis	Vaccination	Natural PAMPs Hydrostatic filtration	Gallibacterium anatis	[167]
	Helicobater pylori	Vaccination	Natural PAMPs	Helicobater pylori	[168]
	Escherichia coli	Vaccination	ESTA6, Ag85BN, Ag85BC Rv2660c	Escherichia coli	[112]
	Cystobacter ferrugineus	Therapy	Anti-microbial potency	Cystobacter ferrugineus	[146]
	Cystobacter velatus	Therapy	Anti-microbial potency	Escherichia coli	[147]
	Pseudomonas aeruginosa	Therapy	In vivo gentamicin loading	Burkholderia cepacia	[123]
	Pseudomonas aeruginosa	Therapy	Loading ceftriaxone, amikacin, azithromycin, ampicillin, levofloxacin, ciprofloxacin and norfloxacin by incubation	Lactobacillus acidophilus	[125]
	Klebsiella pneumoniae	Therapy	BSA nanoparticle cloaking	Klebsiella pneumoniae	[169]
	Aggregatibacter actinomycetemcomitans	Therapy	Nano decoy	Aggregatibacter actinomycetemcomitans	[170]
	Pseudomonas aeruginosa	Therapy	Loading gentamicin via incubation	Staphylococcus aureus, Escherichia coli, and Pseudomonas aeruginosa	[122]
	Pseudomonas aeruginosa	Therapy	Loading gentamicin via incubation	Bacillus subtilis; Staphylococcus aureus	[171]
	Buttiauxella. agrestis	Therapy	Loading gentamicin via incubation	Buttiauxella agrestis, Escherichia coli	[124]
	Helicobater pylori	As adjuvant	Not appliable	Helicobater pylori	[172]
	Escherichia coli	Therapy	Rifampicin-loaded nanoparticle coating	Escherichia coli	[151]
IMVs	Escherichia coli	Vaccination	Genetical engineering of OmpA	Escherichia coli	[51]
DMVs	Pseudomonas aeruginosa	Vaccination	Natural PAMPs; Prepared by nitrogen cavitation	Pseudomonas aeruginosa	[37]
Gram-positive
BMVs	Lactobacillus acidophilus	Vaccination	Natural PAMPs	Enterococcus faecium	[150]
	Staphylococcus aureus	Therapy	Vancomycin-loaded nanoparticle cloaking	Staphylococcus aureus	[173]

Note: BMVs, bacterial membrane vesicles; OMVs, outer membrane vesicles; IMVs, inner membrane vesicles; DMVs, double membrane vesicles; LPS, lipopolysaccharide; PAMPs, pathogen-associated molecular patterns; BSA, bovine serum albumin; Hbp, hemoglobin protease.

5. Cancer therapies and diagnosis

5.1. Cancer immunotherapy using BMVs

The tremendous progress on cancer immunotherapy over the past few decades has shown that cancer can evade the immune system and metastasize. This understanding offers new ways to treat cancer by targeting tumor microenvironments rather than targeting cancer cells. The present cancer immunotherapeutic strategies include cancer vaccines, immune checkpoint blockade therapies, targeting immunosuppressive myeloid cells and chimeric antigen receptor (CAR)-T cell adoptive immunotherapy [174–176].

During cancer immunotherapy, BMVs are able to induce both innate and adaptive immune responses, leading to direct and immunoglobulin-mediated killing of cancer cells. However, a significant barrier in cancer immunotherapy is how to specifically target and deliver immune modulators to tumor microenvironments.

Bacteria and BMVs possess the natural tumor tropism because they can interact with tumor tissues. To verify this idea in which OMVs can reprogram tumor microenvironments (TME), Kim et al. [177]designed a regimen with multiple doses of OMVs derived from Escherichia coli in a mouse CT26 tumor model and found that OMVs effectively induced antitumor immune responses to completely eliminate established tumors. In addition, the authors addressed the molecular mechanism of this antitumor effect. They found that administered OMVs accumulated in tumor locations, and subsequently induced the production of antitumor cytokines, CXCL10 and interferon-γ. The further study in interferon-γ-deficient mice showed that this antitumor effect was interferon-γ dependent. The results showed that OMVs may be anticancer agents to regulate tumor immunity.

Aiming to enhance the tumor checkpoint therapy efficiency and tumor targeting of OMVs, Li et al. [178] reported the genetic engineering of OMVs in which the surface of OMVs was modified by adding the ectodomain of programmed death 1 (PD1). The engineered OMV-PD1 can bind the programmed death ligand 1 (PD-L1) on the tumor cell surface and facilitate its reduction of PD-L1 on the surface, thereby inhibiting the interactions of PD1-PD-L1 between T cells and cancer cells. This study suggests that bioengineered BMVs would be effective immunotherapeutic agents to improve cancer immunotherapies.

BMVs may be as antigen carriers to develop cancer vaccines. A eukaryotic-prokaryotic vesicle (EPV) nanoplatform was constructed by fusing melanoma cytomembrane vesicles (CMVs) to attenuated Salmonella OMVs. This design allowed EPV to integrate melanoma antigens with natural adjuvants to form a single therapeutic agent. In vivo prophylactic studies revealed that the EPV nano formulation stimulated the immune system and triggered the antitumor immune response, combating tumorigenesis. The eukaryotic-prokaryotic fusion strategy provides new methods to design hybrid vaccine platforms [130].

5.2. BMVs as drug carriers

The challenges to translate BMVs are the low drug loading because BMVs have the membrane barriers to prevent drug loading inside BMVs. Nanoparticles (NPs) possess the high drug loading efficiency and nanoparticle coated with eukaryotic cell membrane vesicles show the improved drug loading for cancer therapies [179–181], thus the similar concepts have been applied to BMVs for combination therapies.

Chen et al. [121] prepared OMVs made from attenuated Salmonella and fused DSPE-PEG-RGD to OMVs through the coextrusion of OMVs and DSPE-PEG-RGD, named OMV-DSPE-PEG-RGD (OR). Tegafur-loaded micelles (FT) were encapsulated with OR for the formation of OR-coated F127 micelles (ORFT). It was hypothesized that ORFT nanoparticles could possess the antitumor effect in a mouse melanoma model because OMVs stimulated the host immune response and the drugs sensitized tumor cells to CTLs for tumor killing. The experiments were designed to test this hypothesis and the results supported it. The studies show that bioinspired immunomodulatory nanomedicine may be a novel strategy to the improve current cancer immunotherapies.

Another study is to harness the benefit of immunotherapy from OMVs through the design of hollow polydopamine (HPDA)-mediated photothermal therapy (PTT) in the mouse model of melanoma. An OMV-CC hybrid membrane comprised of a bacterial OMV and B16-F10 cancer cell (CC) membrane was coated onto HPDA nanoparticles. When injected intravenously, nanoparticles of HPDA@[OMV-CC] targeted melanoma tissues and mediated the immune response via activation of dendritic cells (DC) in lymph nodes. The results showed that combinatory therapies of immunotherapy and PTT are needed to eliminate tumor in this mouse model of melanoma [182].

BMVs are liposome-like structures, therefore their enclosed compartments provide various space for loading of hydrophilic and hydrophobic chemotherapeutics. Guo et al. [126] designed a pH-sensitive drug delivery system made of OMVs derived from Gram-negative bacteria loaded with paclitaxel (PTX) and DNA damage response 1 (Redd1)-siRNA to simultaneously regulate tumor metabolism microenvironment and suppressing tumor growth. In a series of in vitro and in vivo experiments, the co-loaded OMVs repolarized tumor-associated macrophages (TAM), suppressed tumor growth, and remodeled TME in the triple-negative breast cancer model. This study shows that co-delivery of multiple drugs using OMVs may be a novel platform to enhance cancer treatment. Furthermore, Gao et al. [20] expressed RGD peptides on bacterial surface, and generated a drug delivery system made of bacterial (Escherichia coli) DMVs to tumor vasculature. Furthermore, the authors loaded DOX in DMVs using a pH gradient driven drug loading method. In a melanoma xenograft mouse model, the authors found that DMVs can target neutrophils and monocytes in vivo and they mediated the transport of DMVs across blood vessel barriers. Interestingly, DMVs can also bind to the tumor vasculature and tumor cells, leading to enhanced deposition of DMVs in tumor microenvironments (TMEs). Notably, DMVs were genetically conferred a RGD targeting group by fusing the RGD-4C peptide to ClyA, a common membrane anchoring protein on bacterial surface, and this strategy significantly enhanced the delivery efficiency of Dox to TME. The study reveals that DMVs are a powerful tool to develop drug carriers for enhanced tumor targeting.

5.3. Probiotic BMVs for tumor inhibition

Probiotic BMVs derived from Gram-positive Lactobacillus rhamnosus were isolated from conditioned medium and HepG2 cells were treated with BMVs. The studies showed that BMVs regulated the expression of Bcl-2 and Bax genes in HepG2. Interestingly, at only 100 μg/ml BMVs showed a significant killing to cancer cells. They studied the apoptotic factors (Bax/Bcl-2 expression ratio) after treating cancer cells with BMVs at 50 and 100 μg/ml, and the increased ratio of Bax/Bcl-2 was related to cancer cell killing, suggesting that the cellular death may be mediated by apoptotic signaling [183]. However, the findings are limited by the in vitro studies, but it is not clear whether this observation can be applied to in vivo cancer treatment.

5.4. Tumor diagnosis

BMVs can be used for cancer diagnosis like EVs [184]. Gujrati et al. [185] investigated bioengineered OMVs derived from Escherichia coli for enhanced optoacoustic (photoacoustic) imaging. The authors prepared OMVs encapsulating biopolymer-melanin (OMV_Mel) comprised of a tyrosinase transgene. The in vivo results in a 4 T1 breast cancer harboring model showed that OMV_Mel generated strong optoacoustic signals for tumor imaging.

Chen et al. [186] reported a one-pot synthetic strategy to engineer OMVs-based sensors for both antigen binding and signal generation. INP-Scaf3 surface scaffold was fused to the Z-domain for antibody recruiting. To assess the capability of the OMV sensor for cancer cell detection, the cancer-specific surface marker, MUC1, was selected as a target. Meanwhile, a GFP was assembled onto the displayed INP-Scaf3-Z scaffold. The results showed that HeLa cells were detected only when the anti-MUC1 antibody was added, with low background interaction between OMVs and HeLa cells.

We have summarized the current technologies and approaches using BMVs to improve the cancer treatment in Fig. 5 and Table 4. These approaches demonstrate the impact of BMVs in biomedical applications via harnessing tumor immunity.

Fig. 5.

Several platforms of BMVs used in cancer therapies. (A) BMVs as cancer antigen carriers for cancer immunotherapy. (B) BMVs as drug carriers to target cancer cells. (C) BMVs are naturally tumoricidal agents to induce cancer cell apoptosis. (D) BMVs as imaging carriers for cancer imaging and diagnosis. (E) BMVs combine with drug nanoparticles to enhance cancer therapy. BMVs, bacteria membrane vesicles. DCs, dendritic cells.

Table 4.

The applications of BMVs in cancer therapy.

Types	Bacterial strain	Engineering and regimens	Therapeutical cargo	Diseases	Ref.
OMVs	Escherichia coli; Vibro cholera; Shigella flexneri	CaP decorating/Immunotherapy	NA	4 T1 breast cancer; CT26 colon cancer	[187]
	Escherichia coli	Co-loading of OMVs	PTX; siRNA	4 T1 breast cancer	[126]
	Escherichia coli	Modified OMVs	PD1	B16 melanoma	[178]
	Escherichia coli	Drug carrier	5-FU	CT26 colon cancer	[188]
	Escherichia coli	Decoration/conjugation drug loading	ICG; TRAIL	B16 melanoma	[189]
	Escherichia coli	Hybrid OMVs	ICG	B16 melanoma	[130]
	Escherichia coli	Transdermal delivery	ICG; PEP	B16 melanoma	[129]
	Escherichia coli	Photothermal therapy	Melanin	4 T1 breast cancer	[185]
	Escherichia coli	OMVs/B16 melanoma cell membrane vesicles.	HPDA	B16 melanoma	[182]
	Klebsiella pneumonia	Attenuated strain/drug loading	Dox	A549 non-small-cell lung cancer	[53]
	Salmonella Typhimurium	Therapy in combination with PTX; Immunotherapy	NA	HTC116 colorectal carcinoma; MCF-7 breast cancer; HepG2 hepatocellular carcinoma	[190]
	Salmonella typhimurium; VNP20009; Staphylococcus aureus	Photothermal/Immunotherapy	NA	CT26 colon cancer	[191]
IMVs	Escherichia coli	Targeting ligand decoration Drug loading	Dox	A549 lung carcinoma	[52]
DMVs	Escherichia coli	RGD ligand decoration Drug loading	Dox	B16 melanoma	[20]
BMVs	Lactobacillus rhamnosus	Immunotherapy	NA	HepG2 liver cancer	[183]

Note: BMVs, bacterial membrane vesicles; OMVs, outer membrane vesicles; IMVs, inner membrane vesicles; DMVs, double membrane vesicles; Dox, doxorubicin; TRAIL, tumor necrosis factor related apoptosis-inducing ligand; ICG, indocyanine green; PD1, programmed cell death protein 1; 5-FU, fluorouracil; RGD, arginine-glycine-aspartate peptide; NA, not appliable.

6. Conclusion and perspective

In this review, we have discussed the unique features of BMVs and their functions. There are two types of bacteria, Gram-negative and Gram-positive. They have the different membrane structures, so the formation of BMVs is different. For example, Gram-negative bacteria can form OMVs. Advances in nanotechnology enable to design and generate many types of BMVs. We can make OMVs, IMVs and DMVs from Gram-negative bacteria. To generate the translational products, a novel approach of nitrogen cavitation was developed to fabricate DMVs, and this approach may be applied to any bacteria. We also discussed the current approaches and methods to generate BMVs using chemical, physical, and biological tools. Biological engineering is a novel and appealing tool to design BMVs for improved their tissue targeting and therapeutic outcomes. Furthermore, we discussed the status on loading of therapeutics in BMVs. There is a challenge to effectively load therapeutics inside BMVs because BMVs form a lipid bilayer barrier. The loading of drug inside BMVs using passive methods was less than 2% (w/w). This is far away to translate them in clinic. To resolve this problem, Gao et al. [20] proposed the remote loading of drugs via pH-gradient between outer and insider membrane of BMVs. The approach improved the drug loading to 12% (w/w). Finally, we discussed the biomedical applications of BMVs in infectious diseases and cancers. The current data demonstrate that the technologies of BMVs could be applied to treat a wide range of diseases even though we only focused on discussing infectious diseases and cancers. Bacteria and BMVs have the natural tumor tropism, therefore, they would be novel drug delivery systems to deliver a wide range of cargo including chemo drugs and genetic materials to tumor microenvironments.

Although BMVs-based therapeutics seem promising, many questions remain studied. 1) Drug loading is too low to make the impact of BMVs in clinic. Current research is mainly focused on developing new structures of BMVs. There is lacking effort to develop novel approaches and methods to increase the drug loading to BMVs. Remote loading of therapeutics in BMVs may be a new direction for translational research. 2) Reproducibility of BMVs. There are so many methods to generate BMVs, and there are no standard protocols. Most importantly, the current methods generate the heterogeneous BMVs (such as sizes and compositions), thus it is difficult to warrant the batch-to-batch consistency of products required in clinic [192,193]. It is needed to address what factors impact on generation of BMVs. Can we develop physical methods to produce BMVs? Can we develop proteomics and lipidomics to analyze the compositions of BMVs? We need to develop rigorous methods to address those questions. 3) Immunotoxicity of BMVs. BMVs are derived from pathogens, so BMVs may contain toxins that may cause the side effects. We need to develop methods to control immunotoxicity of BMVs when they are used in clinic. 4) Spontaneous tumor targeting of BMVs. The studies show that BMVs have the tumor tropism and preclinical studies show that BMVs may be a novel drug carrier to improve cancer treatments. We need to understand the molecular mechanism of tumor targeting of BMVs.

In summary, we have discussed the current studies on BMVs including their generation, varied types and therapies for infectious diseases and cancers. The unique structures of bacteria (Gram-negative and Gram-positive bacteria) offer us to design and produce a wide range of types of BMVs compared to mammal cells. This feature may allow us to develop the translational BMVs-based products. There is the lacking information of how BMVs interact with the host. This knowledge would provide the opportunities to develop tissue targeting drug delivery systems. Finally, we expect that the research of BMVs will lead to the translational therapeutics for treating a wide range of diseases, such as inflammatory disorders, autoimmune diseases, and cardiovascular diseases.