The therapeutic potential of bacterial EVs depends on their isolation, administration, and distribution within host systems. However, at present, inconsistencies in methodologies pose challenges for standardizing outcomes and interpreting results across studies.

Isolation of bacterial EVs typically involves sequential steps of bacterial cell cultivation and removal of intact cells followed by vesicle concentration and purification. Common techniques include ultracentrifugation, ultrafiltration, and size-exclusion chromatography but these methods continue to be marked by a lack of standardization, leading to variations in the size and subpopulations of isolated EVs (87, 88). While influenced by bacterial strain and analytical technique, Gram negative-derived EVs typically exhibit dimensions in the range of 20 to 200 nm, while Gram positive-derived EVs show a broader range of 50 to 300 nm diameter (29). Additional sources of variability include exosome contamination from components of the culture media, differences in media composition, and the bacterial growth phase at which EVs are harvested (87, 88). These inconsistencies may impact their function and biodistribution, making it difficult to compare results across studies. Addressing these challenges requires consistent protocols that ensure reproducibility and enable meaningful comparisons, alongside transparent and accurate reporting of methods.

Techniques for the quantification of EVs also vary, with most studies using relative protein abundance, while others count particle numbers (62, 89, 90). Standardizing EVs by protein concentration can be misleading due to variations in cargo and surface protein load caused by different growth conditions. Normalizing by particle number may reduce this variability, ensuring more consistent comparisons between samples, while standardization of EVs per colony forming units directly links vesicle production to bacterial cell number, providing a complementary perspective on EV yield relative to bacterial population. However, relating EV dose to bacterial cell number remains challenging because few studies establish a direct equivalence between EVs and live bacteria. For example, in a murine norovirus infection model, a dose of approximately 1 × 10¹⁰ EVs yielded similar beneficial effects as 1 × 10⁸ CFUs of Lactobacillus johnsonii N6.2 (72). Noteworthy, while EVs can mimic some of the beneficial effects of live bacteria, their transient nature would require repeated administration, whereas live bacteria may colonize and continuously release EVs.

But this is not the only source of deviations in final applied EV-concentration. For example, in in vivo cancer-related experiments, two separate studies, one on prostate cancer (A. muciniphila) and one on colorectal cancer (Lc. paracasei) administered each 40 μg of EVs per mouse. However, A. muciniphila EVs were injected intravenously after tumor formation while Lc. paracasei EVs were injected subcutaneously together with the tumor-inducing HCT116 colorectal cancer cells. Both experiments showed a decrease in tumor burden (61, 80). In vitro studies on hepatic and chemo-resistant colorectal cancer cells, in contrast, demonstrated substantial differences in administered concentration, with 50–200 μg/mL Lc. rhamnosus GG EVs or 0.1–10 μg/mL Lp. plantarum EVs, respectively (70, 74). Importantly, Lc. rhamnosus GG EVs only showed a cytotoxic effect with 100 μg/mL for 24 h, while other concentrations did not yield the same results, underlining the importance of including a relatively broad range of concentrations especially if there is no prior work to base estimates on.

Another example demonstrating that the effective EV formulation may be different between species is a study by Choi et al., where standardized protein concentrations were used to assess the impact of different pEVs’ anti-depressive effects (66). Here, the fold change in expression of neurotropic factor-regulators in vitro in hippocampal cells was more pronounced with B. subtilis EVs compared to those from Lp. plantarum or A. muciniphila. Contrastingly, in vivo, the impact of B. subtilis EVs on the same genes was similar to other bacterial families tested, pointing out the different effective concentrations in vivo and in vitro (66). Another study explicitly studying the effects of Lp. plantarum EVs in the same context was very similar in outcomes, however the administered concentration of EVs was approximately 0.0020–0.0025 μg (0.1 μg/kg) as compared to 6 μg of protein (approximately 240–300 μg/kg) in the study by Choi et al.; highlighting a magnitude difference of ca. 10³ times (66, 71). Despite the significant difference in dosages, both studies reported antidepressant-like effects following EV treatment, though the specific mechanisms and potential dose-dependent effects would require further investigation to be fully understood.

The impact and range of concentration effects can also be seen in the context of intestinal inflammation, where the use of L. kefirgranum EVs on DSS-induced colitis led to more pronounced effects on intestinal integrity at higher (3.0 mg/kg) doses as compared to lower doses (0.03 mg/kg), while for example Lp. plantarum Q7 EVs did not show significant differences in outcome between 0.5 and 1.0 mg/kg (51, 57). In vitro experiments on intestinal epithelial cells also showed dose-dependent effects, for example on cytokine profiles, toll-like receptors, and tight junctions in concentration ranges as small as 0.1 to 5.0 μg A. municiphila EVs (91) and between 100 and 150 μg/mL Lc. casei EVs (56).

Overall, while the rationale behind the dosage form is often not clearly stated in the reports, the evaluation of those has a significant impact on the results, as dose-dependent effects are often seen and contribute to the rigor and confidence on the results observed.

Experiments evaluating the positive impact of bacterial EVs on wound healing investigated wound treatment via subcutaneous injections, direct topical application, or intraperitoneal injections, post-wounding, across studies. Results consistently demonstrated a beneficial effect of EVs on wound healing, whether the treatment occurred only twice over 12 days, six times over 15 days, or four times over 7 days when administered through injections. Notably, topical treatment every other day did not lead to improved healing compared to untreated controls (63, 78, 85). Despite researchers universally reporting accelerated healing via systemic administration, the mechanisms described exhibited slight variations, suggesting the presence of strain-specific effects that partially overlap. Taken together, these examples highlight that the overall functions of a diverse group of bacterial EVs are very similar in a narrow context and that different bacterial sources can lead to the same overall results despite different mechanisms, whereas the route of administration had a clear impact on effectiveness of the treatment (63, 78, 85). It is worth to mention the importance of simplicity in application in terms of standard clinical practice, which would make direct application options an interesting area to explore further. This was addressed very recently by Kuhn et al. (32), who embedded Lp. plantarum and Lc. casei EVs in a hydrogel matrix for cutaneous application to accelerate wound healing.

The importance of timing of EV-treatment is showcased in a study testing the effects of Lp. plantarum-derived EVs in a model of dermatitis, an inflammation of the skin (68). In vitro tests included assessing the immunogenicity of EVs on untreated cells and their protective effect when pre-treated—before stimulation with S. aureus induced EVs—versus co-treated. Results demonstrated a dose-dependent preventive effect of Lp. plantarum EVs on skin inflammation while co-treatment groups showed no significant impact. In 2022, Chen and team (77) compared preventative (co-treatment) with therapeutic (post-treatment) treatment options in osteonecrosis, revealing that probiotic EVs could only alleviate symptoms in early stages of the disease. Unlike many other studies focusing on a single mode of administration, this article underscores the substantial influence of administration timing on outcomes. The choice of treatment option must be well-reasoned and clinically relevant. Especially with bone diseases like osteoporosis and osteoarthritis increasing in risk with age, a preventative treatment is a realistic and promising option (92, 93). The study also included controls for EV-treated healthy cells and mouse models, demonstrating that some effects are consistent in both healthy and diseased conditions, while others are specific to either. This holds particular significance in the context of preventative treatments, where the condition may not have fully manifested at the point of administration, potentially leading to unintended side effects if controls are not established during development.

The timing of effect-measurement is equally critical, as effects may be noticeable shortly after the treatment but not several days post-intervention as demonstrated by the change in fibrosis markers and an attenuation of hypertonic scar formation by administration of L. druckerii EVs on a scleroderma mouse model. Changes in mRNA and protein levels of fibrosis markers were evident 24 to 48 h post-treatment. In contrast, a study applying Synechococcus elongatus PCC 7942 EVs on wound healing and associated differentiation of dermal fibroblasts did not assess early changes, but after 12 days, no significant differences in those markers were detected (63, 78, 85).

Lastly, the location of measurement may impact the interpretation of results and should be mechanistically reasoned. For example, oral administration of A. muciniphila-derived EVs increased serotonin levels in the hippocampus and colon of mice, but decreased them in the serum, reflecting the importance of monitoring tissue-specific as well as systemic markers to capture localized effects in target tissues (76). In general, understanding the biodistribution of EVs is critical for assessing their therapeutic mechanisms and off-target effects. Studies on bone diseases have employed oral, rectal, or intravenous routes, demonstrating systemic vesicle distribution to the liver, spleen, brain, muscle, lungs, gastrointestinal tract, kidneys, and bones within an hour of administration (62, 67, 77). However, these studies relied on labor-intensive methods such as fluorescent labeling or polyclonal antibody production to trace EVs, highlighting the urgent need for reliable biomarkers.

In summary, the studies reviewed emphasize the critical need for standardized and well-reasoned administration protocols. This is particularly crucial given the variations in concentrations and measurement methodologies. Moreover, the significance of treatment timing has been clearly demonstrated, as seen in the comparison between preventive and post-treatment options for bone diseases. Additionally, the location of measurement plays a pivotal role, with distinct results obtained when measuring cytokine levels in serum compared to tissue samples. Although the information about EV distribution obtained using fluorescent labels from the experiments mentioned above is exceptionally valuable, existing studies lack well-established biomarkers for observing the systemic distribution of specific vesicles, with only one publication directly addressing this concern to the best of our knowledge, which has identified Sdp as biomarkers for the systemic distribution of EVs derived from L. johnsonii N6.2 (33).

Research into bacterial EVs has utilized both in vivo and in vitro models to assess their impact on immune responses and intestinal health. In vivo studies, often using murine models of gastrointestinal disease, such as colitis in mice, have demonstrated that EVs from probiotics and commensal bacteria can alleviate symptoms, restore gut homeostasis, and regulate inflammation (51, 57, 58, 63, 94). For example, L. johnsonii-derived vesicles have been shown to inhibit macrophage migration and induce a tolerogenic immune profile by manipulating the M2/M1 macrophage ratio (62, 73). On the other hand, EVs from E. coli Nissle 1917 promote stronger bacterial killing activities in macrophages and modulate downstream cytokine production, highlighting their role in enhancing host defense mechanisms (49). Furthermore, in models of inflammatory bowel disease (IBD), EVs derived from gut commensals have demonstrated significant protective effects by regulating inflammation, restoring immune balance, and enhancing intestinal barrier integrity (51, 57, 58, 63, 94). Mouse models are historically useful for studying the immune system in health and disease because of their genetic similarity to humans, their ease of manipulation, and their well-characterized immune responses (95). However, pEV-studies are even extended to other animals, such as chickens, reinforcing the broad applicability of EVs in diverse biological systems (96).

In contrast, in vitro experiments offer controlled environments to study EVs’ effects on specific cell lines. Studies use different intestinal epithelial cell lines or monolayers to investigate the direct effects of EVs on inflammation. HT-29 and Caco-2 are both human colorectal adenocarcinoma cell lines commonly used to model the intestinal epithelium in vitro. While Caco-2 cells produce no mucin but often stronger tight junctions, HT-29 cells can build a mucus layer that influences pathogen adhesion and invasion (97). To trigger inflammation in the model, proinflammatory inducers such as LPS, tumor necrosis factor-alpha (TNF-α), dextran sodium sulfate (DSS), or IL-1β are applied (57–59, 89). One study, specifically comparing the effects of EVs on inflammation in HT-29 cells, revealed a specificity to stimulation with LPS compared to IL-1β or TNF-α, prompting a critical examination of studies that assume a single inducer is adequately representative of inflammatory conditions (89).

Several studies have explored the protective impact of EVs on intestinal barrier integrity, particularly in IBD. Both in vitro and in vivo gene expression analysis and protein level assessments consistently show that probiotic EVs enhance the integrity of tight junctions (TJs) crucial for maintaining intestinal barrier function (37, 40). For example, E. coli Nissle 1917 OMVs, when tested in DSS-induced colitis models, observed protein levels of TJ proteins like claudin-14 and ZO-1, which correlated well with expression levels, indicating a transcriptional regulation of TJ proteins (94, 98, 99). Beyond IBD, barrier protective functions of EVs have been studied in various conditions such as obesity (52), alcohol-induced liver injury (75), Staphylococcus aureus challenged IECs (47), and infections with enteropathogenic E. coli (EPEC) (100), consistently showing alleviation of symptoms. Notably, during EPEC infection, barrier dysfunction was reversed through the reorganization of proteins associated with TJ function (such as ZO-1 and occludin) and the cytoskeleton (e.g., F-actin), without corresponding changes in mRNA expression profiles. This finding highlights the importance of comprehensive assessments of EV-induced effects on different levels of expression and organization.

In general, the measurement of parameters such as cytokine levels during inflammation, should be standardized or expanded to include mRNA as well as protein levels in culture supernatants, tissues, or serum. Cytokine levels across studies were mainly measured with RT-qPCR or ELISA in cell culture supernatants (56, 57, 62, 89, 90) or in colon tissues and serum (51) while intestinal barrier markers were measured either on mRNA or protein level (57, 91) in cell cultures or tissue sections. Most information can be drawn from experiments combining several of these methods, as has been exemplified earlier. This comprehensive approach will contribute to a more nuanced understanding and facilitate the translation of promising findings into effective clinical applications.

A notable gap in current research lies in the limited direct comparison between EV treatments on healthy and diseased tissues or cells, highlighting the need for future investigations to aid in contextualizing these outcomes. Research on healthy intestinal epithelial cells has shown both anti-inflammatory and pro-inflammatory effects depending on the EV dose and bacterial strain (43–45, 48, 56, 91, 101). Some studies partially replicated the effects observed during inflammation, such as the augmentation of TJ proteins or the creation of a tolerogenic anti-inflammatory environment, while others revealed opposing effects, often in a dose-dependent manner, as summarized in Table 1 (102–104). Several studies have undertaken comparative analyzes between commensal bacteria and probiotic strains, showcasing strain-specific or general responses (43, 48, 55, 103). For instance, comparing commensal E. coli strains with probiotic E. coli Nissle 1917 revealed that the probiotic strain induced a more intricate T cell response, while the commensal strains seemed to suppress Th1 responses to LPS (48). A frequently observed response is the general priming of the immune system, a phenomenon achievable with both commensals and probiotics (43–45).

Also, the comparison of effects between parent bacterium and treatment with purified EVs can give valuable information on the mechanistic effects, such as for example Lc. casei was shown to increase TLR9 gene expression as well as IL-10 and IFNγ levels in Caco-2 cells whereas EVs significantly decreased TLR9 gene expression and decreased the level of IFNγ, while raising the level of anti-inflammatory cytokines (56). More rarely seen is an explicit comparison of EV-treatment to stimulation with cell-free supernatant, cell-EV-free supernatant, or with the media and purified protein, to evaluate for putative active factors secreted in a soluble form and their specific active components (99). A good example is Lc. rhamnosus JB-1 EVs, which are enriched for heat shock protein components like chaperonins, where administration of a purified protein could reproduce regulatory and neuronal effects in vitro and in vivo (102). In a model of norovirus infection, similar efficacy was observed between L. johnsonii N6.2 whole cells, derived EVs and SH3b2 proteins in liposomes, both in macrophage cell lines and a mouse model (72). Finding these specific bioactive molecules associated with EV-function is an essential step towards their safe and efficient therapeutic application.

Naturally occurring EVs from several bacteria were investigated for their potential in treatment of cancer (61, 70, 74, 80). In the context of hepatic and colorectal cancer an increase in apoptosis markers after incubation with EVs was observed, indicating a cytotoxic effect of Lactobacillus-EVs on cancer cells. Lp. plantarum EVs were specifically tested for their anti-proliferative effect on chemo-resistant colorectal cancer cells, revealing no such impact on non-resistant HCT116 cancer cells. In vivo experiments including A. muciniphila and Lc. paracasei EVs also integrated controls for cytotoxicity on healthy cell lines and tissues, revealing no significant damage to non-tumor tissues. While both studies administered EVs systemically, the timing and duration of EV-administration varied. In the prostate cancer model, EV-administration commenced after tumor development and persisted for 13 days, while in the colorectal cancer model, EVs were co-administered with tumor-inducing HCT116 cells only once (61, 70, 74, 80).

Despite variations in probiotics, administration methods, and mechanisms these experiments demonstrate comparable end-results. However, recognizing the diversity of functions of the same probiotic EVs in different contexts underscores the importance of incorporating controls for systemic effects. Addressing this need, biomarkers to track EV-distribution in the body would be highly beneficial. For instance, a study by Bulut et al. (63) showcased worsened tumor formation due to commensal EVs suppressing cell-mediated immunity against the vaccinated tumor antigen, further emphasizing the importance of understanding off-target effects. Especially in the treatment of cancer, co-administration with other tumor-growth inhibiting medications may be relevant, and the impact of EVs on drug-induced weakened immune systems warrants consideration, given their earlier demonstrated influence on immune responses. In contrast, co-treatments with EVs of different origins could potentially activate complementary pathways, leading to increased therapeutic effects. One very recent study investigated the combination of Lc. rhamnosus GG EVs with anti-PD-1 immunotherapy against colorectal cancer and reported an increase in the therapies’ efficiency (105). This promise of exploring combination therapies has moreover been shown by a study comparing three kefir-derived Lactobacillus EVs—L. kefir, L. kefiranofaciens, and L. kefirgranum—that found a synergistic effect improving IBD when used in combination (59). This suggests the potential merit of investigating synergies between different strains or bacterial sources in future therapeutic strategies. Lastly, we want to mention the emergence of engineered EVs as, for example, cancer drug delivery systems increasing their therapeutic potential (106).

In brief, the study of EVs in interkingdom interactions requires careful consideration of their strain- and dose-dependent effects as well as their context-specific role. In vitro studies show that the elicited responses may depend on experimental conditions, such as the choice of cell lines, proinflammatory inducers and measurement methods. Explicit comparison between EVs and their parent bacteria as well as specific effector molecules enriched in EVs will hopefully reveal their mechanism of action and mechanistic targets. Key challenges include identifying biomarkers to track EV distribution and understanding the synergy of EVs with other treatments to further elucidate the complexity of their interactions.

EV isolation and purification methods need to be standardized to ensure reliability and allow for comparison between bacterial strains. Different isolation techniques can yield vesicles with varying composition and properties, therefore it is important to assess the chosen method’s efficiency, purity, and preservation of vesicle integrity to ensure accurate interpretation of results (107). A possible improvement is including size-exclusion based separation for removal of co-isolated protein impurities. This approach has previously significantly improved the purity and anti-inflammatory activity of EVs produced by Gram positive and Gram negative probiotic bacteria (108). Establishing common criteria and methodologies will ensure the reproducibility of findings, advancing the field collectively as summarized in Figure 2.

Some open questions and limitations revolve around the surrounding and culture conditions in which EVs are formed or processed. These variations in conditions could lead to differences in EV production and function. Monitoring specific beneficial functions of probiotics as well as their EVs is difficult due to a lack of information on their in vivo metabolic activities. Additionally, distinguishing effects produced by delivered probiotic EVs from those of indigenous microbes with similar effects complicates the selection of a specific strain for targeted applications (109, 110). This complexity further hampers reproducibility, given the interactions with other indigenous strains. Similarly, the properties of EVs produced by bacteria under in vitro conditions may differ from those produced in vivo, emphasizing the importance of considering intended application and administration. If administration of live bacteria is anticipated, studies should include EVs produced under conditions that closely mimic gastrointestinal settings. This consideration is crucial, especially given previous evidence that growth medium and culture conditions significantly influence EV composition and quantity (31, 32). Purification and isolation methods may not be entirely effective in separating EVs from protein aggregates or membrane fragments, therefore additional steps and measures must be implemented to ensure the quality of the end product (111–113). Failure to do so may introduce biases in the results.

Another source for bias, related to EV purification, could be the process of vesicle uptake. Different internalization routes result in the sorting of vesicle cargo to specific subcellular locations and determining the factors influencing route selection remains an unresolved question. For instance, the uptake and distribution of EVs may hinge on their size, a characteristic that could be influenced by the purification and concentration methods used, potentially leading to the enrichment of specific EV subpopulations with distinct functions (37, 113).

The importance of standardizing criteria and methods to guarantee the traceability and integrity of the results cannot be overemphasized. The heterogeneity of methods also concerns the measurement of results and parameters such as cytokine levels, which are often difficult to compare within a specific context at the time being. In this case, it is important to reason at what level—gene, protein, or three-dimensional organization—to measure, at which time-point and what location in the organism. Similar guidelines will have to be established for the induction of the condition and the start and interval of EV treatment. In both cases, the concentration administered, the duration and timing of treatment (pre- and co-treatment or post-treatment) must be carefully considered and systematically compared. Since results are often dose-dependent and the doses of EVs used vary in a range of several orders of magnitude (circa 0.1 to >1,000 μg/kg in vivo and 0.01 to 200 μg/mL in vitro), a range of dosages should be included and compared in all studies where there is no prior work available to reasonably base dose-estimations on (57, 71, 74).

In general, direct comparisons between disease-models, between bacterial strains and exploration of combination therapies as well as addressing the nuances of simultaneous secondary conditions remain limited. In addition, meaningful comparisons with drugs already on the market will provide a reference for the potential impact of EVs in clinical practice and will help to guide interpretations. For example, treatment with Lactobacillus-EVs and imipramine, a tricyclic antidepressant, showed similar antidepressant-like effects in mice (71). The selection of administration routes, with a preference for simple and safe methods like oral or topical administration warrants a deeper understanding of the distribution and systemic effects in the host. To ensure the safe and efficient application of EVs in clinical settings, there is an urgent need for reliable biomarkers to monitor their distribution. To date, studies have predominantly focused on understanding the functions and effects of EVs at specific sites of interest, few studies included controls for side effects.

Looking ahead, it is crucial to invest in identifying biologically active components associated with vesicles and understanding the targets on human cells they interact with. Information is available on the mechanistic properties of how selected probiotics and postbiotics interact with the human host, and it seems reasonable that these mechanisms at least partially overlap with those employed by bacterial vesicles (18, 114, 115). Although numerous reports suggest molecules and pathways involved in these EV-transferred functions, detailed information is still lacking, necessitating an effort in that direction in coming work (29). This knowledge not only aids in understanding how these vesicles contribute to maintaining homeostasis, enhancing immune function, or promoting tissue repair but also helps discern the differences between specific probiotic effects and common traits of non-pathogenic bacterial vesicles when applied strategically. In some cases, opposing effects have been observed in different contexts, emphasizing the importance of comparisons between control groups and experimental conditions. Including appropriate controls, such as vesicle-depleted supernatants or vesicles from non-beneficial bacteria helps delineate the specific effects of EVs. Lastly, understanding the experimental context, such as the microbial environment, host cell type, and physiological conditions contributes to the contextual relevance of the findings. Hence, we want to advise combining in vitro experiments with in vivo counterparts whenever possible.

A future avenue touched upon earlier is the use of engineered EVs as targeted delivery systems. Engineered bacterial EVs have shown promise as vaccine delivery vectors and as carriers in biomedical research (116, 117). For instance, Lc. rhamnosus GG vesicles were modified to deliver intrinsic miRNA specifically to bone tissues as treatment of osteoporosis, and E. coli-derived OMVs have been used as transport system for chemotherapeutic drugs (106, 118). But not only in the therapeutic space, also in the food industry could EVs be applied as a convenient delivery system. Their biocompatibility and intrinsic ability to transport metabolic components and influence intestinal barrier permeability could be exploited to increase the bioavailability of food additives with limited stability or solubility, thus contributing to improving the nutritional value of foods (119). Lc. paracasei EVs for instance were observed to aggregate or fuse with milk EVs, potentially changing the availability of associated beneficial effector molecules like the proteins p40 and p75 (120). The relationships between diet, gut microbiome and host are complicated, and both diet- and host-derived bioactive components like microRNAs could modulate the gut microbiota and its EV composition and jointly influence the availability and uptake efficiency of EVs, dietary metabolites and components (121, 122). This application, though promising, warrants an effort to know effector molecules and bioactive components present in or on the EVs and precisely understand their interaction with each other and the host.

The fact that microbial, especially probiotic EVs, interact closely with other microorganisms, food components and host systems has led to a keen interest in their use to enhance the therapeutic potential of certain foods, marking a step towards establishing functional foods and targeted probiotic therapies. As mentioned previously, pEVs can have substantial implications in the treatment of metabolic diseases like obesity and diabetes (52, 73). Since dietary habits were shown to impact amount, size, and content of gut microbiota–derived EVs, this as well concludes implications for the prevention and treatment of such metabolic diseases (110). For example, succinate production triggered by dietary protein intake, or β-mannan administration as an example for carbohydrate-rich food, were shown to influence the production and composition of EVs produced by bacteria (123, 124). Therefore, research should aim at understanding how these processes can alter the bioactive properties of vesicles. In this regard, comprehensive metabolomic studies could help link the presence of specific EVs to the health effects of diet.

In addition to dietary habits, the production of these foods will also affect their nutritional value and beneficial effects on health and wellbeing. EVs together with other postbiotics are naturally found in fermented foods and are used as functional components in the food industry (125). While much remains to be discovered about the role of EVs produced by microbial communities in fermented foods, these vesicles likely contribute to microbial community dynamics by modulating growth, metabolism, and resilience of the consortia that drive fermentation. Gaining insights into microbial communication within food ecosystems may ultimately enhance the development of fermented products with improved safety, nutritional quality, and health-promoting benefits for consumers (126). Effects of food processing techniques such as drying however might impact EVs’ structure, content and biodistribution requiring future mechanistic investigation and establishment of biomarkers. Beyond their given use in some fermented foods, pEVs could also be used deliberately in food production, for example as natural antibacterial agents. For example, Lp. plantarum-derived EVs have been utilized to enhance the quality of tuna fish, showcasing their efficacy in food storage technology. These vesicles did inhibit the growth of Shewanella putrefaciens, a strain which can cause food poisoning and food spoilage (127). Similarly, Lc. casei BL23 EVs have demonstrated inhibition of the formation of Salmonella enteritidis biofilms on polystyrene surfaces, hinting at potential applications in both healthcare and the food industry (128). Further, pEVs may be used as natural treatment and protection against bacterial infections during meat production, avoiding overuse of antibiotics (96). Finally, in addition to extending EV applications to non-human organisms, we would like to emphasize that it could be lucrative to extend research in humans to new tissues and organs such as the cardiovascular or respiratory system and muscles, as presented studies have shown that pEVs also accumulate at these sites after systemic administration (67, 77).

However, as in all research areas discussed here, the same main limitations considering mechanistic understanding, composition variability, host-interactions and clinical relevance must be overcome, as summarized in Figure 2. Little guidance is currently available, and long-term studies are needed to assess possible adverse effects in the human body, be it in the area of precision nutrition, viral infection or cancer therapeutics. Addressing these gaps through rigorous research and standardized methods is key to unlocking the full potential of probiotic extracellular vesicles and enabling their well-characterized, safe and effective use in clinical, biotechnological and dietary applications.