An increasing number of diseases have been associated with changes in the gut microbiota composition. While these changes have been characterized in detail in diseases such as inflammatory bowel disease (Hansen et al., 2012) and metabolic syndrome (Everard et al., 2013), more recent data have suggested alteration of the gut microbiota composition also as a part of pathogenesis of neurodegenerative diseases, such as Parkinson’s disease (Cosma-Grigorov et al., 2020) and Alzheimer’s disease (Vogt et al., 2017). However, it is still unclear how the changes in the gut microbiota relate to the pathological processes observed in the brain.

A few plausible mechanisms for crosstalk between the brain and the gut microbiota have been suggested, including the altered responses of hypothalamus-pituitary axis (Ait-Belgnaoui et al., 2012), immune system (Cheng et al., 2019) and the activation of the vagus nerve (Bravo et al., 2011). Recently, bacterial extracellular vesicles (EVs) secreted by the gut microbiota have been postulated to be one of the key mechanisms contributing to the gut-brain communication (Cuesta et al., 2021). EVs are nano-sized, round, double membrane encapsulated structures that are an essential part of cell biology of both eukaryotes and prokaryotes. EVs are secreted by both gram-negative (Dorward and Garon, 1990) and gram-positive bacteria (Liu et al., 2018), can harbor a variety of biomolecules as their cargo and possess ability to cross biological barriers in the body (Stentz et al., 2018; Sarshar et al., 2022). Blood–brain barrier (BBB) forms a physical and metabolic barrier between the brain and circulation to prevent the entry of harmful chemicals to the brain (Abbott et al., 2006). BBB consists of endothelial cells binding together with tight junctions, the basement membrane, pericytes within the basal lamina that regulate the capillaries and subsequently brain blood flow, and finally astrocytes (Obermeier et al., 2013). While several studies have reported that bacterial EVs can indeed cross BBB, the mechanisms of the passage have remained largely unknown.

In gram-negative bacteria EVs are formed by budding from the outer membrane, which requires detachment of the outer membrane from the peptidoglycan layer. The detachment of the layers requires local changes in the outer membrane in order to maintain its stability, such as breaking the covalent and non-covalent bonds between the constituents of the outer membrane and peptidoglycan layer (Deatherage et al., 2009), repulsion between resulting anionic charges (Beveridge, 1999), and subsequent change in the hydrostatic pressure in the periplasmic space (Schwechheimer et al., 2013) EVs secreted by gram-negative bacteria are also referred to as outer membrane vesicles (OMV) and are considered to be 20–300 nm diameter in size (Avila-Calderón et al., 2021). Another mechanism for EV formation in gram-negative bacteria is through cell death when outer inner membrane vesicles (OIMVs) and explosive outer membrane vesicles (EOMVs) are formed. OIMVs have a cytoplasmic membrane of gram-negative bacteria, a peptidoglycan layer and an outer membrane (Pérez-Cruz et al., 2013), while EOMVs have only an outer membrane (Devos et al., 2017).

The EV formation in gram-positive bacteria has been suggested to take place through membrane budding and cell lysis (Brown et al., 2015). In this process, in order to reach extracellular space, the EVs must pass through the thick peptidoglycan layer. During cell lysis this is enabled by endolysin enzyme (Andreoni et al., 2019), but the exact mechanism for EV formation in membrane budding remains largely unknown. As compared to EVs secreted by gram-negative bacteria, the EVs secreted by gram-positive bacteria have a larger size range of 20–400 nm (Díaz-Garrido et al., 2021). EVs secreted by gram-positive bacteria are also referred to as cytoplasmic membrane vesicles (CMVs), due to a lack of an outer membrane (Toyofuku et al., 2019).

While cargo embedded to bacterial EVs varies by species, the same bacteria can also produce differently loaded EVs depending on the environmental cues (Lynch et al., 2019). The EVs secreted by gram-negative bacteria often harbor molecules originating from the outer membrane, cytoplasmic membrane, peptidoglycan layer and periplasm (Beveridge, 1999), including heat shock proteins, superoxidase dismutases, adhesins, toxins, lipopolysaccharides of the outer membrane (Wispelwey et al., 1989), small RNAs (Han et al., 2019) and other pathogen associated molecular patterns (PAMPs). In addition, bacterial DNA has been observed in EVs secreted by gram-negative bacteria (Pérez-Cruz et al., 2013).

Extracellular vesicles from gram-positive bacteria lack LPS and periplasmic components but carry similar types of cargo molecules as EVs from secreted by gram-negative bacteria, including peptidoglycan, lipids, proteins, and nucleic acids (Brown et al., 2015). EVs secreted by the gram-positive bacteria have been reported to contain lipoteichoic acid, phosphatidylglycerol, cardiolipin (Resch et al., 2016), DNA (Klieve et al., 2005), toxins, enzymes and other proteins, short fatty acids (Olaya-Abril et al., 2014), microRNA, lipids and fluids (Brown et al., 2015). Gram-positive bacteria have been reported to use EVs for gene transfer (Klieve et al., 2005) and CMVs to ship bacteriophage receptors and bacteriophages, thus rendering bacteriophage resistant cells subjective to phage invasion (Toyofuku et al., 2017; Tzipilevich et al., 2017).

In general, bacterial EVs can harbor a variety of neurotransmitters as their cargo, including dopamine, serotonin, noradrenaline and enzymes contributing to synthesis of these molecules. Bacterial EVs can also harbor short chain fatty acids that are able to influence the function of neurons and microglia (Haas-neill and Forsythe, 2020).

Bacteroidetes, one of the major bacterial phyla constituting the human gut microbiota, are known to produce enzymes aiding the digestion of nutrients and send them to the gut lumen via EV secretion (Elhenawy et al., 2014). These enzymes include hydrolases that degrade complex carbohydrates (Rakoff-Nahoum et al., 2014), inositol polyphosphatases that break down dietary phytate to phosphates, inositol phosphates and inositol (Stentz et al., 2014). The commensal Bacteroides fragilis has been reported to secrete EVs containing neurotransmitter gamma-aminobutyric acid (GABA) and its precursors glutamate and α-ketoglutarate, while pathogenic Bacteroides fragilis has been reported to secrete EVs embedded with histidine decarboxylase, an enzyme catalyzing histamine synthesis (Zakharzhevskaya et al., 2017).

While bacterial EV cargo can reach the cells of biological barriers through different methods of endocytosis, intact bacterial EVs have been suggested to cross epithelial and endothelial layers using paracellular and transcellular routes. Bacterial EVs passing through the intestinal barrier and entering circulation is supported by findings of gut microbiota-associated EVs in dendritic cells of lamina propria (Shen et al., 2012), urine (Lee et al., 2017), and blood (Chang et al., 2021). Moreover, it has been reported that in mice, bacterial EVs in blood represent the gut microbiome composition of the host (Park et al., 2017). To pass through the gut epithelial layer, EVs can plausibly use transcellular transmigration through caveolae-mediated endocytosis or paracellular transmigration, where EVs pass through the intestinal barrier between the epithelial cells (Jones et al., 2020). However, the routes human gut microbiota EVs take to pass the gut epithelia likely differ between different species and utilize both transcellular and paracellular routes (Stentz et al., 2018).

While there is limited data on human microbiota EVs, EVs produced by pathogens and their translocation across biological barriers have been studied extensively. EVs of pathogenic bacterial strains have been demonstrated to increase the permeability of the intestinal barrier (Turkina et al., 2015) by weakening the extracellular matrix via embedded collagenases and hyaluronate lyases (Jeon et al., 2017) or the integrity of physiological barriers by serine protease activity (Hoy et al., 2010; Jarzab et al., 2020). Gingivitis causing bacteria Aggregatibacter actinomycetemcomitans and Porphyromonas gingivalis have been demonstrated to secrete EVs that increase the expression of vascular endothelial growth factor subsequently altering the permeability of blood vessel endothelium (Suthin et al., 2003). EVs secreted by Campylobacter jejuni have been reported to break down E-cadherin of cell junctions and occludin of tight junctions (Elmi et al., 2016). On the other hand, commensal bacteria have been reported to produce EVs that enhance the function of tight junction in the intestinal barrier by increasing zonula occludin expression and subsequently decreasing the paracellular transmigration of EVs of pathogenic strains (Alvarez et al., 2016).

Bacterial EVs have been demonstrated to enter the host cells using different endocytosis mechanisms, including micropinocytosis (Weiner et al., 2016), clathrin-mediated endocytosis (Boisvert and Duncan, 2008), clathrin-independent endocytosis (Norkin et al., 2001; Mondal et al., 2016) and cell membrane fusion (Bomberger et al., 2009). By default, bacterial EVs up taken by the host cells end up in lysosomes for degradation. However, in caveolae-mediated endocytosis EVs are protected by the plasma membrane of the host and are transported to the endoplasmic reticulum or Golgi apparatus, thus allowing bacterial EVs to evade degradation by lysosomes (Lim et al., 2014).

Extracellular vesicles secreted by the gut microbiota can undergo phagocytosis by dendritic cells of the gut lamina propria. For example, the dendritic cells detect EVs secreted by Bacteroides fragilis by recognition of capsular polysaccharide A and direct them to phagocytosis via TLR2 receptor activation (Shen et al., 2012). This process leads to an increase in the number of regulatory T-cells and transcription of IL-10 and Foxp3, protecting the body from excessive inflammatory response and autoimmunity (Izcue et al., 2009). In addition, EVs secreted by Lactobacillus rhamnosus were reported to induce IL-10 and heme oxygenase-1 expression levels in dendritic cells and subsequently increase the number of regulatory T-cells in Peyer’s patches and mesenteric lymph nodes in mice (Al-Nedawi et al., 2015).

In this systematic review we summarize the present evidence of bacterial EVs crossing the BBB and highlight the importance of future research on gut microbiota-derived EVs in the context of gut-brain communication across the BBB.

The literature search was performed using PubMed and Scopus databases (query search on 04-08-2023). Literature search parameters were set to find publications that included in their title or abstract one word of each category of the following with options for different spelling: (1) bacteria, microbiota, microbiome (2) extracellular vesicles, outer membrane vesicles (3) brain, blood–brain barrier. Search were conducted in PubMed using the following query: ((“Blood–Brain Barrier”[Mesh] OR “blood brain barrier”[Text Word] OR “Brain”[Mesh] OR Brain [Text Word]) AND (“Extracellular Vesicles”[Mesh] OR “outer membrane vesicle*”[Text Word] OR “extracellular vesicle*”[Text Word])) AND (“Bacteria”[Mesh] OR “Microbiota”[Mesh] OR “bacter*”[Text Word] OR “microbiota”[Text Word] OR “microbiome”[Text Word]). Search were conducted in Scopus using the following query: (TITLE-ABS-KEY(“*bacter*” OR “*microbiota*” OR “*microbiome*”) AND TITLE-ABS-KEY(“*extracellular vesicle*” OR “*outer membrane vesicle*”) AND TITLE-ABS-KEY(“blood–brain barrier” OR “Brain”)). The publications resulting from the literature search were imported to Covidence program for the systematic review and duplicate records were removed (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia, n.d. Available at www.covidence.org). The abstracts of the publications were screened and records irrelevant to the topic or of other publication type than original research were excluded. Full text screening was performed with the following inclusion criteria: bacterial origin of EVs and incorporation of either in vivo or in vitro model of EV biodistribution to the brain in the study setting.

The literature search yielded 218 publications from which 80 duplicate records were removed. After screening the abstract and title of 138 publications, 113 records were excluded. After full text screening, 8 publications were excluded. As a result, 17 publications were included in this systematic review. PRISMA flow diagram (Page et al., 2021) of the systematic review process is presented in the Figure 1 and the studies included in the review in the Table 1.

The studies exploring bacterial EVs passage through BBB in murines so far have been done with varying EV sources, administration locations, dosages, circulation times and most importantly with varying methods of quantification. It is important to note that the technique used in the EV administration has a wide effect on implications of the obtained results. While the passaging of bacterial EVs through BBB can be studied with different administration routes, the number of biological barriers EVs have to pass through increases if the administration takes place, e.g., via the oral cavity as compared to direct intravenous administration. Thus, the study settings and administration technique need to be planned accordingly paying attention to detail and taking into account the larger context of the study at hand. While the majority of studies tracking bacterial EV passage to the brain employ staining of the EVs and subsequent imaging analysis, the importance of the use of negative controls is highlighted. Thus, control samples representing diffusion of the dye itself need to be incorporated. It is also of importance whether a loading dye targeting the membrane or the cargo is used, in other words, distinguishing whether the staining represents intact EVs or released cargo as well.

The data on the mechanisms how bacterial EVs cross BBB remain limited. In recent studies, one of the leading hypotheses of how bacterial EVs pass through BBB is that they directly alter its permeability. Indeed, many studies on bacterial EVs and the BBB permeability report a decrease in gene expression of tight junction-related proteins in response to bacterial EV administration (Wei et al., 2020; Gong et al., 2022; Nonaka et al., 2022; Pritchard et al., 2022) while a decrease in the number of tight junctions has been demonstrated to directly increase the permeability of BBB (Hartsock and Nelson, 2008). Wei et al. (2020) reported a decrease in claudin-5 tight junction protein expression in the mouse hippocampus in response to EVs isolated from fecal samples of patients with Alzheimer’s disease (Wei et al., 2020). Gong et al. (2022) observed a reduction of zonula occludens-1 (ZO-1), occludin, claudin-5 mRNA expression and occluding protein expression in the hippocampus in response to oral administration of Porphyromonas gingivalis to mice (Gong et al., 2022). Nonaka et al. (2022) reported an induced degradation of ZO-1 and occludin in their human in vitro BBB cell model (Nonaka et al., 2022). Interestingly, in their study Xie et al. (2023) indicated that Helicobacter pylori EVs translocate from stomach to brain through transcellular pathways without disrupting the gastrointestinal or blood–brain barriers (Xie et al., 2023).

In addition to changes in BBB permeability, Pan et al. (2023) and Mi et al. (2022) demonstrated that bacterial EVs are able to” hitchhike” through BBB in neutrophils (Mi et al., 2022; Pan et al., 2023). In their study setting, Mi et al. (2022) used a mouse model in which the brain tumors of the mice were colonized by intravenous injections of bioengineered Salmonella typhimurium, a strain characterized by an enhanced tumor-homing and a capability of induction of neutrophil recruitment. They observed that EVs isolated from the culture of Salmonella typhimurium and loaded with doxorubicin (DOX) could cross BBB and reach the colonized brain tumor. While this finding demonstrated the role of EVs as carriers mediating the crossing of BBB, it implied that bacterial EVs could not cross BBB in an absence of tumor colonization and subsequent neutrophil infiltration to the brain (Mi et al., 2022). Pan et al. (2023) reported successful delivery of pioglitazone over BBB with Escherichia coli EVs in their murine stroke model (Pan et al., 2023). In their study, Pan et al. (2023) assessed pioglitazone-EV uptake to neutrophils, their stability during the hitchhike and BBB penetration capability of neutrophil-engulfed pioglitazone-EVs in vitro (Pan et al., 2023).

A number of mechanisms have been suggested for how whole pathogenic bacteria cross BBB in pathologies involving infection of the brain, such as bacterial meningitis (Kim et al., 2005, 2015). More recently, it has been speculated that EVs secreted by these pathogens could be a main factor driving the changes in the BBB permeability to enable infiltration of whole cell bacteria through the barrier (Nonaka et al., 2022). As a demonstration, a study by Kim et al. (2005) reported that Escherichia coli is able to pass through BBB in a human in vitro cell model through cytotoxic necrotizing factor 1 (CNF1) action (Kim et al., 2005). Later on, it was demonstrated that CNF1 secretion from bacteria cytoplasm is mediated via EVs (Davis et al., 2006; Kouokam et al., 2006) and that a deficiency of the protein mediating CNF1 packaging to EVs decreased the ability of the bacteria to invade the microvascular endothelial cells in the human brain (Yu and Kim, 2012). Thus, in addition to the fact that bacterial EVs are able to cross BBB themselves, their contribution to the whole bacterial cell infiltration into the brain needs further evaluation. In addition, the possible presence and role of bacterial EVs needs to be assessed in studies that report a bacterial presence in the brain based on quantification of bacterial biomolecules in brain samples. Thus, it needs to be evaluated if the reported findings are due to translocation of bacterial EVs and their cargo instead of actual presence of whole cell bacteria.

While the effect of EVs derived from oral pathogens in the brain have been extensively studied, little is known about the communication of the brain and the microbiota through EVs in health. The relevance of EV secretion from different human microbiotas, most interestingly gut microbiota, needs to be further evaluated: bacterial EVs are likely to be one of the key aspects of the communication between the vital gut microbiota and the host, so their role in health requires more research. Notably, while it is known that bacterial EVs can modify host immune responses, it is of importance to establish the role of microbiota-derived EVs in the immunomodulation processes in health and disease inside and outside of the scope of neuroinflammation.

While circulation is often considered to be the most relevant route for bacterial EVs’ to enter the brain in regard to their role in disease and their potential use in medical applications, there are few other possible routes for bacterial EVs to reach the brain. There is very little data on the bacterial EV passage through other systems and their relevance especially in a healthy state. Currently, to our best knowledge, there is no data on bacterial EV transport to the brain via the lymphatic system and only few studies explore the possibility of bacterial EV trafficking via the vagus nerve (Lee et al., 2020) or their effect on the conductivity of the afferent fibers (Al-Nedawi et al., 2015).

Bacterial EVs possess enormous potential in multiple biomedical applications, including vaccine platforms, biomarker discovery, drug delivery and discovery of novel molecules with pharmacological value. Bacterial EVs of gram-negative bacteria are already used in vaccines, although the toxicity of their LPS content is redeemed problematic (Acevedo et al., 2014). EVs of gram-positive bacteria have been suggested for vaccine development due to the absence of LPS in them (Choi et al., 2015). Bacterial EVs have potential to serve as biomarkers in different pathologies that involve invasion by pathogenic bacteria. In addition, the gut microbiota-derived EVs might be used as a marker of gut microbiota dysbiosis associated with several diseases (Hansen et al., 2012; Everard et al., 2013; Vogt et al., 2017; Cosma-Grigorov et al., 2020). Based on their studies in mice, Park et al. (2017) suggested analysis of blood sample-associated bacterial EVs as a technique for screening the dysbiosis of the gut microbiota in patients with neurodegenerative diseases (Park et al., 2017). In addition to blood sampling, changes in the gut microbiota have been screened by analyzing bacterial EVs in urine samples from individuals with autism spectrum disorder (Lee et al., 2017).

The ability to cross biological barriers, to carry variety of biomolecules, and to efficiently protect their combined with high potential for bioengineering render bacterial EVs with enormous potential in drug delivery, including targeted delivery of antibiotics and cytostatic chemotherapy. In their study suggesting bacterial EV hitchhiking across BBB in host immune cells, Mi et al. (2022) demonstrated that doxorubicin could be efficiently target-delivered to Salmonella typhimurium strain-colonized glioma with bioengineered EVs from the same strain when EVs were administrated intravenously to mice (Mi et al., 2022). In turn, Pan et al. (2023) demonstrated delivery of pioglitazone to the brain of ischemic mice via bacterial EVs (Pan et al., 2023).

In conclusion, bacterial EVs have been characterized to cross BBB in animal models and in vivo cell cultures, but the occurrence of this in the context of the human gut microbiota-brain communication remains poorly characterized. Bacterial EVs and their passage through BBB creates possibilities for their use in biomedical applications. Indeed, bacterial EVs can be considered a ready-made, co-evolution driven interkingdom transporting system.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

AK and JR: conceptualization. AK, SS, SA, and JR: methodology. AK: literature review. AK and SK: writing – original draft preparation. AK, SS, SA, JT, MT, NB, TT, and JR: writing – review and editing. SS: graphical abstract illustration. JR: supervision and project administration. TT and JR: funding acquisition. All authors contributed to the article and approved the submitted version.

JR thanks the Academy of Finland for grants 328768 and 299749. AK thanks the Finnish Cultural Foundation for grant 00220426 and Yrjö Jahnsson Foundation for grant 20217413.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.