Skip to main content
NCBI home page
Journal of Extracellular Biology logoLink to Journal of Extracellular Biology
. 2025 Nov 23;4(11):e70099. doi: 10.1002/jex2.70099

Oral Delivery of Semaglutide and Tirzepatide Using Milk‐Derived Small Extracellular Vesicles

Yuefei Zhang 1,2,3,✉, Jianyi Han 1,2,3, Wei Wu 1,2,3, Bobo Dang 1,2,3,✉
PMCID: PMC12640765  PMID: 41293773

ABSTRACT

Therapeutic proteins and peptides have revolutionized modern biomedicine, but their oral delivery is limited by gastrointestinal degradation and barriers. Small extracellular vesicles (sEVs), which are resistant to biochemical degradation and capable of traversing mucus and cellular barriers, hold great promise as next‐generation oral delivery vehicles. Oral semaglutide, the first approved oral GLP‐1 receptor agonist (GLP‐1RA), employs vesicle‐mediated transcellular transport, highlighting the potential of sEVs as an effective delivery vehicle. In this study, we demonstrate the successful oral delivery of two GLP‐1RAs, semaglutide and previously unexplored tirzepatide, using milk‐derived sEVs. Both peptides were efficiently loaded onto sEVs in vitro, and their oral administration effectively reduced blood glucose levels in diabetic db/db mice. Compared with the current SNAC technology, which is limited exclusively to semaglutide, our sEV platform provides broader applicability and versatility for oral peptide drug delivery.

1. Introduction

Since insulin was extracted to treat diabetes, more than 80 peptide drugs have been marketed in the therapeutic areas to treat diabetes, cancer, osteoporosis, and other diseases (Muttenthaler et al. 2021). However, peptide drugs suffer from low oral bioavailability, limiting them to invasive injection as the administration route. To address this issue, extensive research has been devoted to developing effective oral delivery systems for peptides. This involves utilizing a diverse range of innovative materials and technologies to enhance the oral bioavailability of peptides, focusing on overcoming degradation caused by digestive fluids and enhancing peptide absorption through the gastrointestinal barrier (Drucker 2020). Despite significant efforts, oral administration of peptides remains a major hurdle, with only a few approved oral peptide drugs exhibiting low oral bioavailability (typically less than 1%) (Brown et al. 2020). The limited oral bioavailability not only compromises the efficacy of these drugs but also necessitates higher and more frequent dosages for achieving desired therapeutic outcomes (Davies et al. 2017).

The first oral GLP‐1RA, semaglutide, was approved in 2019 for type 2 diabetes treatment using the excipient SNAC, which prevents pepsin degradation and enhances gastric absorption (Brown et al. 2020; Drucker 2020; Buckley et al. 2018). Despite these advancements, oral semaglutide's bioavailability remains low (0.4%–1%), which necessitates higher and more frequent dosages, leading to an increased economic burden on patients (Granhall et al. 2019; Brayden et al. 2020). Moreover, the excipient (SNAC) used in oral semaglutide inadequately enhances oral absorption of structurally similar peptides, restricting its applicability and limiting its broader potential beyond semaglutide (Buckley et al. 2018). In glucose‐lowering and weight‐loss peptide drugs, there is growing interest in multi‐targeted therapies. For example, tirzepatide, a dual GLP‐1R/GIPR agonist, showed superior glucose‐lowering effects over semaglutide in a phase III trial (Frías et al. 2021). The success of such drugs will likely increase demand for oral formulations; however, the semaglutide formulation may not be suitable for other peptides. This underscores the need for an efficient, convenient, and broadly adaptable oral delivery system (Brown et al. 2020). Oral semaglutide is absorbed via vesicle‐mediated transcellular transport in the mucous epithelial cells, without disrupting junctional complexes (Buckley et al. 2018). This non‐paracellular‐directed absorption shows the possibility of improving bioavailability by using natural vesicles in intercellular communication, such as small extracellular vesicles (sEVs).

sEVs are naturally occurring nanovesicles secreted by cells, acting as carriers of signalling molecules to facilitate intercellular communication in various pathophysiological processes (van Niel et al. 2018). Being of natural origin, sEVs offer several advantages, including excellent biocompatibility and low immunogenicity when compared to synthetic or viral vehicles. These inherent characteristics position sEVs as promising candidates for the next‐generation drug delivery vehicles (Cheng and Hill 2022; Herrmann et al. 2021). Serving as versatile carriers, sEVs are released by almost all cell types, leading to diverse and abundant sEV populations (Kalluri and LeBleu 2020). Additionally, some sEVs have shown remarkable stability in the gastrointestinal tract and the ability to cross biological barriers, rendering them promising candidates for oral drug delivery vehicles (Song et al. 2022). Currently, oral sEVs are primarily sourced from milk, although some studies have explored sEVs derived from cell lines or plants (Carobolante et al. 2020; Mao et al. 2021). Milk‐derived sEVs exhibit low immunogenicity and high stability in the stomach and intestine (Zhong et al. 2021). The widespread availability and cost‐effectiveness of milk provide a strong foundation for the industrial‐scale production of milk‐derived sEV‐based drugs, paving the way for significant advancements in therapeutic applications (Herrmann et al. 2021).

In 2016, a few studies started to use milk sEVs to orally deliver small‐molecule chemotherapeutic drugs (including paclitaxel and docetaxel) (Munagala et al. 2016). Curcumin, anthocyanin (Aqil et al. 2017), α‐mangostin (Qu et al. 2022), and resveratrol (Esfahani et al. 2024) were also shown to be orally delivered using milk sEVs. Furthermore, experiments have demonstrated that milk sEVs possess the potential to orally deliver nucleic acids, peptides and proteins, due to their resistance to degradation and ability to cross the intestinal barrier (Shandilya et al. 2017; Zhang et al. 2022, 2023; Yan et al. 2022; Oshchepkova et al. 2023; Grossen et al. 2021). Oral milk sEVs loaded with insulin were shown to lower blood glucose in type I diabetic rats (Wu et al. 2022). Septic mice were relieved by oral FGF21 encapsulated within milk‐derived sEVs (Li et al. 2024). Liraglutide was also loaded into milk sEVs for oral delivery. While sublingual delivery has been reported to lower blood glucose, oral gavage failed to show efficacy (Xu et al. 2022). Among current studies, there is a lack of comprehensive comparison between different sEV sources or loading methods for the oral delivery of peptide or protein drugs.

In this study, we investigated the potential for oral delivery of semaglutide and tirzepatide, utilizing sEVs derived from natural sources. We isolated, purified, and extensively characterized various sEV types obtained from milk (whole or defatted fresh milk), plants (immature or mature coconut water), and the 293F cell line. Among these sources, sEVs from defatted milk exhibited notably high yields and demonstrated favourable characteristics in terms of particle size, morphology, marker expression, and purity. As a result, we selected defatted milk‐derived sEVs as the optimal oral delivery vehicles. Through a thorough comparison of various loading methods, we discovered that room temperature (RT) incubation produced an sEV‐semaglutide complex with the highest encapsulation rate (56.4%) and a high positive rate (94.9%). In vitro assays confirmed that sEV–semaglutide and sEV–tirzepatide maintained potency comparable to their free peptide counterparts. In vivo distribution analysis showed that the capsule sEV formulation remained in the gastrointestinal tract significantly longer than the liquid form after oral administration. Moreover, efficacy studies in a diabetic mouse model demonstrated that oral capsule formulations of sEV–semaglutide and sEV–tirzepatide produced significant and sustained blood glucose–lowering effects. Together, these findings indicate that milk‐derived sEVs can markedly enhance the oral absorption of semaglutide and tirzepatide, presenting a promising strategy for advanced oral peptide drug delivery.

2. Results

2.1. Defatted Milk sEVs Are Utilized as Oral Delivery Vehicles

Currently, the sources of oral sEVs are mainly focused on milk, with a few derived from cell lines or plants (Carobolante et al. 2020; Mao et al. 2021). However, these sEVs have not been studied comparatively. HEK293‐F (293F) cells are widely utilized for producing macromolecules, such as recombinant proteins and viral vectors. For EV production, 293F‐derived EVs exhibit consistent physicochemical properties and batch‐to‐batch compositional stability, along with an excellent safety profile, making them a promising, safe, and reliable platform for drug delivery applications (Chen et al. 2025). Coconut water, as a direct source of plant‐derived liquid, requires no juicing process and offers cost‐effective, high‐yield production, analogous to milk, which is advantageous for industrial‐scale applications. Furthermore, coconut water EVs have been thoroughly characterized (Zhao et al. 2018; Yu et al. 2019), providing a solid foundation for their use in drug delivery. Therefore, we focused on representative sEVs from milk (whole or defatted), coconut water (immature or mature), and 293F cells in this study (Zhao et al. 2018; Si et al. 2022; Agrawal et al. 2017). A comprehensive analysis was conducted through nanoparticle tracking analysis (NTA), transmission electron microscopy (TEM), and western blot. We analysed the particle sizes of the sEVs, which were shown to be around 100–200 nm (Figure 1A). TEM data showed that the sEVs we isolated have a ‘cup‐shaped’ structure (Figure 1B). The TEM morphology of milk EVs contains ‘non‐vesicle’ structures identified as lipoproteins, which are the primary non‐EV protein components in this fraction (Somiya et al. 2018). In negative‐staining TEM, lipoproteins exhibit a characteristic spherical and bright morphology (Yu et al. 2016), resembling the structures in Figure 1B. Here we utilized the yield normalized to input volume to facilitate comparisons between sEV isolated from different sources. We found that milk sEVs have the highest yields: the yields of defatted milk sEV reach 1011 sEVs/mL input, which is slightly higher than whole milk sEVs (Figure 1A). The final sEV product of defatted milk has the concentration of 1013–1014 particles/mL, consistent with high‐yield standards (Marsh et al. 2021). We therefore selected milk sEVs for further characterizations. The phospholipid bilayers of milk sEVs were clearly observed by cryo‐EM (Figure 1C). TEM and cryo‐EM show that there are more lipoprotein particles (indicated by black arrowheads) in whole milk sEV samples compared with sEVs from defatted milk, suggesting low purity of whole milk‐derived sEVs (Figure 1B,C). We then analysed the expression of generally recommended sEV‐positive markers (CD9, CD81 and TSG101) and negative marker (GM130). The results showed that the expression of positive marker proteins in defatted milk sEVs was higher than that in whole milk‐derived sEVs (Figure 1D). The data indicated that defatted milk sEVs exhibited higher purity compared to whole milk sEVs. Considering factors such as yield, morphology, and purity, we selected defatted milk‐derived sEVs as the preferred oral delivery vehicles. Henceforth, sEVs will refer specifically to defatted milk sEVs.

FIGURE 1.

FIGURE 1

Open in a new tab

sEV isolation and characterization. (A) sEV size distribution and yield determined by nanoparticle tracking analysis (n = 3 to 7 per group). Morphological characterization of sEVs through TEM (B) or cryo‐EM (C). White arrows indicate sEVs with a cup‐shape in TEM or a lipid bilayer in cryo‐EM. Black arrows indicate lipoprotein particles. (D) Western blot analysis of sEV positive markers CD81, CD9, and TSG101 and negative marker GM130 (35 µg total protein loaded per sample).

2.2. In Vitro Loading of Semaglutide or Tirzepatide Into sEVs

The drug loading efficiency of sEVs is significantly influenced by the loading methods employed, which include incubation, sonication, extrusion, freeze‐thaw cycles, and detergent treatment (Vader et al. 2016). We used encapsulation efficiency (EE) and positive rate to compare different loading methods. The EE is defined as the concentration of sEV‐incorporated drugs over the initial concentration used to prepare the formulation (Piacentini 2016). EE (%) = (Amount of peptides loaded in EVs/Initial amount of peptides) × 100. The positive rate represents the ratio of drug‐loaded sEVs to the total sEVs. To track and quantify semaglutide or tirzepatide loading, we synthesized semaglutide‐fluorescein amidite (semaglutide‐FAM) (Figure S1) or tirzepatide‐fluorescein amidite (tirzepatide‐FAM) (Figure S2) to monitor the loading process (Hintzen et al. 2024). sEVs are significantly larger in size compared to semaglutide or tirzepatide. Consequently, during fast protein liquid chromatography (FPLC), sEVs elute at volumes earlier than 10 mL, whereas semaglutide or tirzepatide elutes after 20 mL (Figure S3). Therefore, EE can be calculated by measuring the relative proportions of FAM fluorescence intensity at these two distinct elution volumes. We loaded semaglutide‐FAM into sEVs using incubation, sonication, extrusion, freeze‐thaw, and saponin treatment methods, and calculated EE. The results showed that incubation at room temperature has the highest EE of 56.4% (Figure 2A). The sEV‐semaglutide‐FAM has a positive rate of 94.9% determined by nanoFCM (Figure 2B). We also observed colocalization of semaglutide‐FAM and sEVs (Figure 2C) using super‐resolution imaging. In addition, sEV‐tirzepatide‐FAM prepared by incubation has 23% EE and an 83.4% positive rate, respectively (Figure S4). Both semaglutide and tirzepatide contain a long chemically modified lipid tail in their structures. We believe their loading into sEVs most likely occurs through insertion of this lipid tail into the sEV membrane. Consequently, the peptide portion would remain exposed on the vesicle surface rather than residing inside the lumen.

FIGURE 2.

FIGURE 2

Open in a new tab

Production of sEV‐semaglutide. (A) Efficiency of loading semaglutide‐FAM into sEVs. (B) Positive rate of sEVs loaded with semaglutide‐FAM. (C) Super‐resolution microscopy of sEVs loaded with semaglutide‐FAM. The upper panel shows three EV aggregations; there are about 3‐4 EVs in each aggregation. The lower panel shows enlarged pictures of EV aggregation.

2.3. In Vitro and In Vivo Activity of sEV‐Semaglutide and sEV‐Tirzepatide

To evaluate the in vitro and in vivo activities of sEV‐semaglutide and sEV‐tirzepatide, we employed a cell‐based assay using 293 cells expressing human GLP‐1 receptor (hGLP‐1R) coupled with CRE‐driven firefly luciferase, along with an in vivo diabetic mouse model (db/db mice) (Lau et al. 2015). We mixed semaglutide with semaglutide‐FAM at a 10:1 ratio and incubated the mixture with sEVs at room temperature. Using the fluorescence standard curve generated from semaglutide‐FAM, we determined the amount of semaglutide loaded into the sEV‐semaglutide samples. Tirzepatide was prepared using an identical protocol. Cell assays revealed that sEV‐semaglutide and sEV‐tirzepatide exhibited EC50 values comparable to those of the peptide‐only groups (Figures 3A,B). Furthermore, subcutaneous injection studies demonstrated that sEV‐semaglutide exhibited glucose‐lowering efficacy comparable to free semaglutide (Figure 3C). Although semaglutide has a prolonged half‐life exceeding 100 h in humans, its half‐life in mice is substantially shorter, typically under 10 h, resulting in a correspondingly shorter duration of efficacy (FDA 2017). Consistent with our findings, a previously published efficacy study by Novo Nordisk demonstrated that subcutaneous administration of semaglutide (30 nmol/kg) maintained blood glucose reduction for up to 48 h (Lau et al. 2015), but not beyond this time frame (Figure 3C).

FIGURE 3.

FIGURE 3

Open in a new tab

Efficacy of sEV‐peptides. (A) GLP‐1R activation (EC50) of semaglutide and sEV‐semaglutide. (B) GLP‐1R activation (EC50) of tirzepatide and sEV‐tirzepatide. (C) Blood glucose lowering efficacy study of semaglutide and sEV‐semaglutide in diabetic db/db mice after subcutaneous (sc) dosing (30 nmol/kg, n = 5 to 6 per group). (D) Blood glucose lowering efficacy study of tirzepatide and sEV‐tirzepatide in diabetic db/db mice after sc dosing (30 nmol/kg, n = 5 to 6 per group). (E) Blood glucose level of diabetic db/db mice after p.o. dosing with semaglutide or sEV‐semaglutide in liquid formulation (n = 4 to 6 per group). All data are represented as mean ± SEM. Statistical significance was determined by an unpaired two‐tailed t‐test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Similarly, in the tirzepatide treatment group, sEV‐tirzepatide effectively reduced blood glucose levels to an extent comparable to free tirzepatide (Figure 3D). Notably, tirzepatide demonstrated superior blood glucose‐lowering efficacy compared to semaglutide, aligning with results from a recent clinical study (Figure 3D) (Frías et al. 2021).

The approved oral semaglutide (Rybelsus) is administered at a dose of 5 mg per individual, equivalent to approximately 17 nmol/kg for a 70 kg body weight (Buckley et al. 2018). In a diabetic mouse model, we initially attempted oral delivery of a 10‐fold dose (170 nmol/kg) of semaglutide or sEV‐semaglutide in a liquid formulation. However, neither group exhibited a significant reduction in blood glucose levels (Figure 3E). SNAC is the critical enhancer in promoting oral absorption of semaglutide (Buckley et al. 2018). We next tested an increased dose of oral semaglutide (2800 nmol/kg) in a solution supplemented with SNAC. However, this formulation still failed to lower glucose levels compared to the control group. In the commercial oral formulation, semaglutide is co‐formulated with SNAC in a tablet form. Mechanistic studies have shown that semaglutide absorption occurs in the stomach, where the oral semaglutide tablets remain and sustain a high local drug concentration (Buckley et al. 2018). We hypothesize that the solid formulation prolongs the residence time of semaglutide in the digestive tract, facilitating its absorption. Consequently, the failure to reduce blood glucose levels in our study was likely attributable to the use of the liquid formulation.

2.4. sEV Capsules Have a Longer Half‐Life Than the Liquid Formulation In Vivo

We then lyophilized semaglutide or tirzepatide‐loaded sEVs into a dry powder for oral capsule preparation. Direct lyophilization led to a reduction in sEV concentration and altered size distribution. This phenomenon has been reported in the past, when no protectant was added during EV lyophilization (Golan and Stice 2024). Milk EVs also suffer from destruction after lyophilization (Dogan et al. 2025). However, these adverse effects can be mitigated by the addition of the cryoprotectant trehalose, as we also observed (Figure 4A, Figure S5) (Popowski et al. 2022).

FIGURE 4.

FIGURE 4

Open in a new tab

Biodistribution of orally administered bovine milk‐derived sEVs. (A) Analysis of sEV size distribution and concentration, both with and without lyophilization, was performed to assess the impact of the lyophilization process. (B)–(C) db/db mice were administered a single dose of VivoTrack680‐labelled sEVs (≈3 × 1011 particles) by oral gavage (p.o.) in liquid or capsule formulation. In vivo imaging and statistical analysis of the mice after drug administration (n = 3 to 5 per group). (D) Representative ex vivo imaging of the tissues and organs. All data are represented as mean ± SEM.

Additionally, we prepared VivoTrack680‐labelled sEVs to investigate the oral absorption process of both sEV liquid and capsule formulations (Zhang et al. 2020). The signal from the liquid formulation rapidly diminished following oral administration, with most of the signal disappearing within 8 h, indicating that a significant amount of sEVs were excreted from the body. In contrast, the sEV signal from the capsule formulation showed an initial increase followed by a decrease, with the peak signal observed around 4 h (Figure 4B). The signal from the capsule formulation remained stronger than that of the liquid formulation from 2 to 48 h (Figure 4C). Organ distribution analysis following oral administration revealed an enhanced signal in the gastrointestinal tract for the capsule group (Figure 4D, Figure S6). These results suggest that capsule‐formulated sEVs persist longer in vivo compared to the liquid formulation.

The stomach is the primary absorption site for oral semaglutide, where SNAC transiently enhances absorption, with permeability reverting to baseline within 30 min and no prolongation of tablet residence time (Buckley et al. 2018; Solis‐Herrera et al. 2024). In contrast, sEVs can remain in the stomach for up to 18 h (Figure 4D, Figure S6), providing a basis for extended residence time and potentially enhanced bioavailability.

2.5. sEV‐Semaglutide and sEV‐Tirzepatide Oral Capsules Lower Blood Glucose Level

We then performed the efficacy study using encapsulated sEV‐semaglutide and sEV‐tirzepatide powders, comparing them with semaglutide and SNAC‐semaglutide in db/db mouse models. The oral enhancer SNAC was critical for the successful oral delivery of semaglutide, as evidenced by the expected blood glucose‐lowering effect in the SNAC‐semaglutide group, while the control group without SNAC showed no effect. Both sEV‐semaglutide and sEV‐tirzepatide significantly reduced blood glucose levels compared to the control group (Figure 5), suggesting that sEVs may play a similar role as SNAC in enhancing the oral availability of semaglutide, thereby reducing blood glucose levels. Moreover, sEV‐tirzepatide exhibited a more profound and long‐lasting effect in this diabetic mouse model, as evidenced by the lower blood glucose levels at later time points (Figure 5).

FIGURE 5.

FIGURE 5

Open in a new tab

Oral capsules containing sEV‐semaglutide and sEV‐tirzepatide effectively lowered blood glucose levels in db/db mouse models. Different formulations of semaglutide and tirzepatide were administered orally to db/db mice (n = 4–8). Blood glucose levels were monitored at various time points and compared between different groups. All data are represented as mean ± SEM. Statistical significance was determined by an unpaired two‐tailed t‐test (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Currently, semaglutide is approved for daily oral administration (Rybelsus) for treating type 2 diabetes. However, the bioavailability of orally administered semaglutide is approximately 1% (US010933120B2), necessitating daily dosing compared to weekly subcutaneous injections. Due to inherently shorter pharmacokinetic profiles in rodents (< 10 h) compared to humans (> 100 h), oral administration naturally results in shorter and less sustained efficacy in mice. While previous studies have not reported the efficacy of oral semaglutide in rodents, our findings demonstrate that both semaglutide/SNAC and our newly developed semaglutide/sEV formulation achieve similar blood glucose‐lowering effects at 18 h post‐oral dosing (Figure 5). Importantly, tirzepatide has not yet been developed for oral administration; our data indicates that the sEV‐based oral delivery approach can be successfully extended to tirzepatide with better efficacy, representing a clear advantage over SNAC technology (Figure 5).

In conclusion, milk‐derived sEVs enhanced the oral delivery of semaglutide and tirzepatide, demonstrating significant potential for the development of oral peptide therapeutics.

3. Discussion

Peptide drugs have long been difficult to be utilized orally, and only a few oral formulations have been approved (Brown et al. 2020). In this study, we developed sEV capsules capable of successfully delivering semaglutide orally. We further demonstrated that this delivery strategy could be readily extended to tirzepatide, highlighting its broad applicability and clear advantages. Liraglutide has previously been tested for oral delivery using milk‐derived sEVs in a liquid formulation; however, it failed to reduce blood glucose levels in mouse models, aligning with our findings that liquid formulations are ineffective for oral peptide delivery (Xu et al. 2022). Although SNAC has been demonstrated to enhance oral delivery of semaglutide, it does not facilitate the oral administration of other peptides, underscoring its limited applicability beyond semaglutide (Buckley et al. 2018). Recently, hybrid vesicles combining milk exosomes and liposomes have been investigated for semaglutide delivery, but their broader application remains uncertain due to the complexity of the formulation, posing additional challenges for future development (Xiao et al. 2024). In contrast, our developed sEV capsules successfully facilitated oral delivery of both semaglutide and tirzepatide, indicating that sEV capsule formulations possess broader applicability and more significant potential for delivering various related peptide drugs orally (Salim et al. 2022).

Both semaglutide and tirzepatide contain a long chemically modified lipid tail in their structures. We believe their loading into sEVs most likely occurs through insertion of this lipid tail into the sEV membrane. Consequently, the peptide portion would remain exposed on the vesicle surface rather than residing inside the lumen. This lipid‐mediated insertion is widely used in EV drug loading (Chen et al. 2022). For example, DSPE‐PEG polymers serve as linkers between sEVs and proteins, with thorough characterization (Wang et al. 2022). Super‐resolution images show aggregated sEVs, with approximately three to four DiI‐labelled sEVs per 600 nm cluster, rather than an increased size of individual loaded sEVs. To be exact, incubation holds potential for broader application in loading lipidated macromolecules into sEVs.

Extraction methods influence the yield or oral performance of milk sEVs (Xia et al. 2024). Therefore, optimizing these methods could enhance the effectiveness of sEVs in oral delivery applications. Given the heterogeneity and existence of distinct subpopulations within milk sEVs, it might be important to identify and characterize subtypes most suitable for oral delivery (Cocozza et al. 2020; Mathieu et al. 2019). Additionally, exploring advanced formulations, such as layer‐by‐layer self‐assembly or hydrogels, may further improve the efficiency of sEV‐mediated peptide oral delivery (Deng et al. 2023; Gan et al. 2022).

Regarding clinical translation challenges, our data demonstrate several advantages of milk‐derived sEVs for scalable production. First, we achieved high yields from defatted milk (Figure 1A), which is significantly more abundant and cost‐effective than cell culture‐derived sources. Second, the cryoprotectant trehalose effectively maintained sEV stability during lyophilization (Figure 4A, Figure S5), addressing the storage and distribution hurdle. However, we acknowledge manufacturing consistency requires further optimization through GMP‐compliant studies to establish standardized milk sourcing/processing protocols, rigorous quality controls for drug loading efficiency, and scalable purification methods to eliminate impurities—all critical aspects for regulatory approval and commercial viability.

Room temperature incubation resulted in the differential EE between tirzepatide (23%) and semaglutide (56.4%). This observation likely reflects intrinsic physicochemical differences between these peptides—tirzepatide's larger molecular size and higher hydrophilicity may reduce loading efficiency compared to semaglutide, yet its superior pharmacological profile enabled more profound and sustained glucose‐lowering effects in our db/db model despite lower encapsulation (Figure 5). From a scalability perspective, three key factors mitigate this limitation: First, our milk source is abundant, and the milk‐derived sEV platform achieves high yields that can readily support large‐scale production and dose escalation. Second, although the loading efficiency is lower for tirzepatide, the capsule formulation of tirzepatide‐loaded sEVs demonstrated superior functional efficacy compared with semaglutide‐loaded sEVs, suggesting that loading efficiency may not represent a critical bottleneck for downstream scale‐up or translation; however, higher peptide drug loading could further improve functional outcomes. Third, the loading process itself is technically simple—relying on room‐temperature incubation—and is inherently amenable to large‐scale implementation. Future efforts can focus on improving tirzepatide loading efficiency through optimization of buffer conditions or loading strategies, while preserving these scalability advantages.

EV‐based therapies represent an emerging field that requires a case‐specific regulatory approach due to their complexity, with key challenges including route of administration, dosing, and safety optimization. Currently, no specific guidelines exist for EV therapies, though cell‐based product standards provide valuable frameworks for characterization. Though promising, the lack of clear regulatory pathways presents unique challenges for EV therapy development, necessitating careful consideration of these factors to advance towards clinical application.

Although we observed significant glucose‐lowering effects with oral sEV‐peptides, direct detection of systemic intact peptides remains technically challenging due to their low bioavailability and matrix interference from milk EVs. Future studies in larger animal models may help address these technical constraints while maintaining focus on therapeutic outcomes.

Regarding the precise cellular and molecular mechanisms of sEV‐enhanced oral peptide absorption, including target cells and endocytic pathways, existing evidence and our observations provide some insights. As reported, milk‐derived sEVs bind surface natural IgG, enabling FcRn‐mediated transcytosis (Xia et al. 2025; Samuel et al. 2021). Our study shows a similar characteristic of retention in specific regions (e.g., the stomach and intestine) to other studies (Xia et al. 2025; Samuel et al. 2021), which mainly depends on the passive physical stability of milk sEVs. This retention trait is unique to milk‐derived sEVs, because non‐milk sEVs degrade rapidly in the gut. The small intestine is the core absorption site, with intestinal epithelial cells as key targets rather than gastric parietal cells or M cells. Our study mainly focused on verifying the feasibility of sEV‐based oral delivery of semaglutide and tirzepatide, which was successfully achieved. Future studies can explore the full complexity of these processes, as a more comprehensive mechanism would further strengthen the rationale for sEV‐based oral peptide delivery systems.

In summary, we have explored various sEV sources and peptide drug‐loading methods to develop a capsule formulation that effectively enabled oral delivery of semaglutide, successfully reducing blood glucose levels in a mouse model. Currently, oral delivery of tirzepatide has not been established; however, our method was successfully extended to tirzepatide, which demonstrated even greater and more sustained efficacy. These findings highlight the potential for oral delivery of tirzepatide and underscore the broad compatibility and distinct advantages of our approach. By demonstrating compatibility with multiple peptide drugs, we believe this method holds significant potential for broader application to other therapeutics with similar characteristics.

4. Materials and Methods

4.1. General Information on Peptide Synthesis

4.1.1. Materials for Peptide Synthesis

Fmoc/Boc‐protected amino acids, as well as Fmoc‐Gly‐Wang Resin, were purchased from GL Biochem (Shanghai). Ethyl ether and dichloromethane (DCM) were purchased from Sinopharm Chemical Reagent Co. Ltd. N,N‐dimethylformamide (DMF), trifluoroacetic acid (TFA, HPLC grade), and acetonitrile (HPLC grade) were purchased from J&K Scientific. TIPS (triisopropylsilane), DIEA (N,N‐diisopropylethylamine), diisopropylcarbodiimide (DIC), ethyl 2‐(hydroxyimino)cyanoacetate (Oxyma), HCTU (5‐chloro‐1‐[bis(dimethylamino)methylene]‐1H‐benzotriazolium 3‐oxide hexafluorophosphate), 4‐methylpiperidine, [2‐[2‐(Fmoc‐amino)ethoxy]ethoxy]acetic acid, tetrakis(triphenylphosphine)palladium, 18‐(tert‐butoxy)‐18‐oxooctadecanoic acid, 20‐(tert‐butoxy)‐20‐oxoicosanoic acid, and 5(6)‐carboxyfluorescein (5(6)‐FAM) were purchased from Bidepharm (Bidepharm).

4.1.2. General Procedures for SPPS

The main chain of Semaglutide/Tirzepatide was synthesized by standard Fmoc solid‐phase peptide synthesis (Fmoc‐SPPS) using a single‐channel microwave peptide synthesizer (Liberty Blue, CEM, North Carolina, USA). The Wang resin, preloaded with glycine (0.53 mmol/g), or Rink amide resin (0.47 mmol/g), was solubilized in DMF for 5 min prior to use. The Fmoc protecting groups were removed by treatment with 20% piperidine in DMF at 40°C. Amino acid coupling was carried out at 80°C with the reaction of Fmoc amino acids (5 equivalents), Oxyma (5 equivalents), and DIC (5 equivalents) for 3 min. To avoid side reactions at high temperatures, BOC‐L‐His(Trt)‐OH and Fmoc‐L‐Arg(Pbf)‐OH were reacted at 50°C for 10 min.

For the synthesis of fluorescently labelled semaglutide, 5(6)‐carboxyfluorescein was introduced at the end of the main chain. Specifically, 5(6)‐carboxyfluorescein (2 equivalents), HCTU (2.1 equivalents), and DIEA (4.2 equivalents) were reacted overnight.

For side‐chain synthesis, Fmoc‐Lys (Alloc)‐OH was introduced, and the Alloc protecting group was removed using tetrakis(triphenylphosphine)palladium. Reaction conditions: the fully protected peptide was washed with DMF and DCM, followed by the addition of 246 µL of phenylsilane, which was reacted with 6 mL of DCM solution containing 11 mg of tetrakis(triphenylphosphine)palladium for 20 min, repeated three times. When the reaction was finished, the resin was washed and drained, and the side‐chain synthesis was carried out sequentially according to the sequence. The coupling of amino acids was carried out at room temperature with the reaction of Fmoc amino acids (5 equivalents), HCTU (4.75 equivalents), and DIEA (10 equivalents) for 2 h. The reaction was repeated three times.

After peptide chain synthesis, the resin was thoroughly washed with DCM and dried under vacuum. The resin was then treated with 2% (v/v) water, 2.5% (v/v) triisopropylsilane, and 95.5% (v/v) trifluoroacetic acid (TFA) for 2.5 h at room temperature. The excised resulting peptide solution was precipitated with ice‐cold ether and washed three times. Crude peptides were dissolved using a mixture of water and acetonitrile (H2O:CH3CN = 1:1) containing 0.1% trifluoroacetic acid and lyophilized for processing.

4.1.3. HPLC and Mass Spectrometry

The crude peptides were purified by semi‐preparative reversed‐phase high‐pressure liquid chromatography (RP‐HPLC), and the column used for peptide purification was an Agilent ZORBAX X300SB‐C18 (9.4 × 250 mm, 5 µm) column, with a flow rate of 5 mL/min for the mobile phase. The ultraviolet detection wavelengths of the HPLC were 214 and 280 nm. Solution A was water (containing 0.1% trifluoroacetic acid), and solution B was acetonitrile (containing 0.1% trifluoroacetic acid). The peptide purity was above 95% as determined by RP‐HPLC and characterized by electrospray ionization mass spectrometry (Agilent 1260‐6230 single TOF LC/MS).

4.1.4. Preparation of Peptides or Peptide‐FAM

Semaglutide or Semaglutide‐FAM was prepared using preloaded Glycine Wang Resin (0.53 mmol/g, 0.1 mmol scale) and synthesized according to the standard Fmoc‐SPPS method, following the general procedure (Merrifield 1963). For Tirzepatide or Tirzepatide‐FAM, we also used preloaded Glycine Wang Resin (0.47 mmol/g, 0.1 mmol scale).

4.1.5. sEV Isolation

Defatted bovine milk was purchased from a local supermarket. Defatted bovine milk‐derived sEVs were isolated by the acetic acid/ultracentrifugation (AA/UC) method (Somiya et al. 2018). After incubation with acetic acid [milk/acetic acid = 100 (vol.)] for 5 min at room temperature, samples were centrifuged at 16,000 × g for 20 min at 4°C. The supernatant was filtered with a 0.22 µm membrane and named whey. The whey was ultracentrifuged at 130,000 × g for 120 min at 4°C using an SW32Ti rotor and an Optima XPN‐100 Ultracentrifuge (Beckman Coulter). A pellet of sEVs was resuspended by phosphate‐buffered saline (PBS). This process was repeated. After centrifugation at 10,000 × g for 5 min at 4°C, the residual precipitates were removed. The sEV samples were then stored in aliquots at −80°C until use.

Whole bovine milk was purchased from a local supermarket. A conventional isolation method was performed according to the previous article (Munagala et al. 2016; Somiya et al. 2018). Briefly, whole milk was centrifuged at 13,000 × g at 4°C for 30 min. Then the whey was collected by passing through a cheesecloth. Whey was collected in the middle layer and ultracentrifuged at 100,000 × g for 60 min at 4°C to remove larger particles. The supernatant was further ultracentrifuged at 130,000 × g for 60 min at 4°C. At 130,000 × g for 60 min at 4°C, the pellet was washed with PBS twice. The sEV samples were resuspended in PBS and stored in aliquots at −80°C until use.

Coconuts were purchased from a local supermarket. The sEV isolation method was performed according to the previous article (Zhao et al. 2018). Briefly, samples were centrifuged at 3000 × g for 60 min at 4°C. Then the supernatant was ultracentrifuged at 18,000 × g for 60 min at 4°C. Next, supernatant was ultracentrifuged at 120,000 × g for 60 min at 4°C using an SW32Ti rotor. The pellet was washed by PBS. Then ultracentrifugation at 120,000 × g for 60 min at 4°C. The pellet of sEVs was resuspended in PBS, and the residual precipitates were removed via centrifugation at 10,000 × g for 5 min at 4°C. The samples were filtered (0.22 µm). The remaining samples were stored at −80°C for further analysis. HEK 293F cells were cultured with a serum‐free medium. For sEV purification experiments, the cell supernatant was isolated by four steps at 4°C as previously described (Zhang et al. 2021).

Above sEVs also can be resuspended by 25 mM cryoprotectant trehalose (TRE) in PBS and stored in aliquots at −80°C until use.

4.1.6. sEV Characterization

For particle size distribution detection, we used NTA. sEVs were diluted 10‐to 10000‐fold to achieve between 20 and 100 objects per frame in the Nanosight NS 500 system (Malvern). We carried out the experiments and analysed data as previously mentioned (Zhang et al. 2021).

For sEV morphological observation, we utilized TEM and cryogenic electron microscopy (Cryo‐EM) as before (Zhang et al. 2021). For TEM, 3 µL of sEV solution was incubated for 1 min on the carbon film and then discarded. Next, 5 µL of 1% uranium acetate (filtered by 0.2 µm membrane) was pipetted onto the carbon film and incubated for 2 min. Images were acquired using a 120 kV TEM instrument (Talos L120C G2, Thermo Scientific). For cryo‐EM, sEV samples were applied to the grids (Quantifoil 1.2/1.3) and then prepared with a freeze unit (Vitrobot, Thermo Fisher). Samples were transferred under liquid N2 temperatures into a 200 kV TEM (Glacios, Thermo Fisher).

CD81 (A22986PM, Abclonal), CD9 (ab92726, Abcam), TSG101 (ab125011, Abcam) and GM130 (ab52649, Abcam) primary antibodies were purchased for expression analysis of sEV markers. sEV samples were lysed in RIPA lysis buffer containing PMSF (Beyotime), and protein concentration was quantified using the Enhanced BCA Protein Assay Kit (Beyotime). Samples were then lysed in 1× loading buffer and boiled. Proteins were separated by SurePAGE (Genscript), and PVDF membranes were incubated with indicated primary antibodies, followed by HRP‐labelled Goat Anti‐Rabbit IgG (H+L) (Beyotime). Signals were detected with Amersham Imager 600 (GE Healthcare).

4.1.7. EE

We used five ways for loading semaglutide‐FAM (5 µM) into sEVs (220 µg/mL) and evaluated their EE (Haney et al. 2015). In this experiment, we used Superose 6 increase columns in AKTA Go (Cytiva) equipped with fluorescence detector RF‐20A (Shimadzu). The following methods were done in the dark. For the incubation at room temperature (RT), sEVs were incubated with semaglutide‐FAM for 12 h. For saponin treatment, a mixture of sEVs and semaglutide‐FAM was supplemented with 0.1 mg/mL saponin (S875797, Macklin) for 10 min at RT. For freeze‐thaw, sEVs and semaglutide‐FAM were incubated for 1 h, then rapidly frozen at −80°C, and thawed at RT. The freeze‐thaw cycle was repeated three times. For extrusion, the mixture solution was extruded (×11 times) through an extruder (Morgec) with a 100 nm pore diameter. For sonication, the solution was sonicated (low intensity, 10 cycles by 30 s on/30 s off) by a sonication device (Diagenode). The EE of tirzepatide‐FAM was evaluated similarly as described above.

4.1.8. Positive Rate

To evaluate the rate of sEVs successfully loaded with semaglutide‐FAM, we used Flow NanoAnalyzer U30 (nanoFCM), a nanoflow cytometer (nFCM), following the manufacturer's instructions and as described previously (Silva et al. 2021). We prepared the solution by incubating sEVs (10 mg/mL) with semaglutide‐FAM (200 µM) for 12 h at RT, in the dark. Then the samples were purified by ultracentrifugation as mentioned above. Briefly, the samples were diluted by PBS (0.22 µm filtrated) and processed by the machine in a conventional routine. Light scattering and fluorescence of individual sEVs were collected using side scatter and 488/24 bandpass filters. Positive rates were calculated in comparison to the sEV‐only group. The nFCM can also be applied in the measurement of the concentration and size of sEVs. The positive rate of sEVs loaded with tirzepatide‐FAM was evaluated similarly as described above.

4.1.9. Super‐Resolution Microscopy

We prepared the solution by incubating sEVs (1 mg/mL) with semaglutide‐FAM (20 µM) overnight at RT, in the dark. Then the samples were purified by ultracentrifugation as above mentioned. sEV membrane was labelled by DiI kit (C1036, Beyotime) according to instructions. Free DiI was discarded by the PD‐10 desalting columns (17085101, Cytiva). Coverslips (WHB‐12‐CS, WHB scientific) were immersed in this sample (1:10 dilution by PBS) overnight at 4°C, in the dark. The next day, these coverslips remained with a thin layer of solution and were then mounted with anti‐fade mounting medium (H‐1000‐10, VECTASHIELD). Images were acquired with a STELLARIS 8 STED system (Leica) in gSTED + Lightning model. The imaging experiments were performed with a 100× objective and two channels: Ex 488 nm, Em 495‐560 nm, Depletion 592 nm; Ex 550 nm, Em 562‐640 nm, Depletion 660 nm.

4.1.10. In Vitro Functional Assay

We established 293 cell lines that express both the hGLP‐1R and CRE firefly luciferase. This in vitro functional assay was done as previously reported (Lau et al. 2015). We prepared the sEV‐peptide solution by incubating sEVs (10 mg/mL) with peptide (200 µM) and peptide‐FAM (20 µM) overnight at RT, in the dark. Then samples were purified by ultracentrifugation as mentioned above. The fluorescent standard curve of peptide‐FAM allowed us to calculate the amount of peptide in the sEV‐peptide samples. Cells were plated out into 96‐well plates at 20,000 cells/well in a volume of 50 µL assay buffer (DMEM, 10 mM HEPES, 1× GlutaMAX (Gibco 35050061)). Compounds to be tested were diluted in assay buffer with final assay concentrations of 1 × 10−14−1 × 10−7 M. The plate was incubated for 3 h at 5% CO2 at 37°C. The luciferase assay system (E1501, Promega) was used to detect firefly luciferase activity. Cells were incubated in 1× lysis buffer and shaken at room temperature for 30 min, in the dark. After adding luciferase assay reagent, the plate was read in a microplate reader (Thermo). With imported data, EC50 values were determined by GraphPad Prism version 6.

4.1.11. sEV Encapsulation

For sEV encapsulation, we utilized lyophilization to make sEV powders. Briefly, the sEV pellet was successively resuspended by 25 mM trehalose in saline, frozen at ‐80°C and lyophilized for 24 h as previously described (Popowski et al. 2022; Bosch et al. 2016). These sEV powders were filled into capsules (size M, Torpac) according to instructions.

4.1.12. In Vivo Biodistribution of sEVs

We studied the biodistribution of milk sEVs administered by oral delivery in male db/db mice. sEVs were labelled with VivoTrack 680 (fluorescence) according to the manufacturer's instructions (Zhang et al. 2020). 10 mg/mL sEVs in 1 mL PBS were incubated at the concentration of 20 µM VivoTrack 680 at RT for 1 h. Then the solution was diluted by PBS and ultracentrifugation at 130,000 × g for 2 h, as mentioned previously, to remove the unbound dye. Finally, the pellets were resuspended in 1 mL saline (25 mM trehalose). The same amount of labelled sEVs were administered via liquid or capsule formulation. The only dye group was processed simultaneously as a control. Mice were further imaged using a small animal imaging system (PerkinElmer). For ex vivo imaging, organs were rapidly excised from euthanized animals. Images were analysed by the IVIS imaging system (PerkinElmer). The relative intensities were measured and compared with control groups.

4.1.13. Animal Studies

Tests for effect on blood glucose and test procedures were conducted as previously reported (Lau et al. 2015). Briefly, male db/db mice (GemPharmatech) were subcutaneously injected with drugs formulated in a phosphate buffer (50 mM phosphate, 0.05% Tween 80, pH 8). sEV‐peptide was prepared as mentioned in the section on in vitro functional assay and sEV encapsulation. For oral gavage, mice were fasted in advance with free access to water (Buckley et al. 2018). Three hours after administration, the access to food was regained. Capsules (size M, Torpac) were filled with drug powder and delivered according to instructions (Zhu et al. 2023). The ratio between SNAC and semaglutide is 60:1. For the SNAC‐semaglutide group, the rest part is described as patent US10933120B2. For the control‐semaglutide group, there is no SNAC. Blood glucose levels were measured by the tail tip blood using the glucose oxidase method (glucose test strips and glucose analyser, 580, Yuwell). All animal maintenance and experimental procedures were conducted in accordance with the Institutional Animal Care and Use Committee (IACUC) guidelines of Westlake University (Hangzhou, China).

4.1.14. Data Analysis and Statistics

Data analysis was performed with a one‐way ANOVA or a two‐tailed Student's t‐test using GraphPad version 6.0 (GraphPad, La Jolla, CA, USA). The number of experimental repeats is indicated in the relevant figure legends. A p value < 0.05 was considered significant and is indicated in the figures as *(p < 0.05), **(p < 0.01), ***(p < 0.001), or ****(p < 0.0001).

Author Contributions

Yuefei Zhang: conceptualization, methodology, software, data curation, investigation, validation, formal analysis, supervision, funding acquisition, visualization, project administration, resources, writing – original draft, writing – review and editing. Jianyi Han: methodology, conceptualization, visualization, validation, investigation, software, data curation, formal analysis. Wei Wu: methodology, data curation, investigation, validation, formal analysis, software. Bobo Dang: conceptualization, software, data curation, investigation, validation, supervision, funding acquisition, visualization, project administration, resources, writing – original draft, writing – review and editing, formal analysis, methodology.

Funding

This work was supported by China Postdoctoral Science Foundation (2021TQ0284), Research Center for industries of the Future (RCIF) at Westlake University, and Westlake Laboratory of Life Sciences and Biomedicine. This study was also partly supported by the National Key Research and Development Program of China (2024YFC3407400), the Natural Science Foundation of China (32120103013), and the “Pioneer” and “Leading Goose” R&D Program of Zhejiang (No. 2024SDXHDX0002, 2025SDXHDX0002).

Conflicts of Interest

No, there is no conflict of interest

Supporting information

Figure S1. Characterization of semaglutide and semaglutide‐FAM. (A) Structure and LCMS analysis of semaglutide. (B) Structure and LCMS analysis of semaglutide‐FAM.

Figure S2. Characterization of tirzepatide and tirzepatide‐FAM. (A) Structure and LCMS analysis of tirzepatide. (B) Structure and LCMS analysis of tirzepatide‐FAM.

Figure S3. UV280 or FAM analysis in FPLC for semaglutide‐FAM (A), tirzepatide‐FAM (B) and sEVs (A‐B).

Figure S4. (A) The efficiency of loading tirzepatide‐FAM into sEVs. (B) Positive rate of sEVs loaded with tirzepatide‐FAM.

Figure S5. sEV size distribution and concentration analysis after lyophilization or not.

Figure S6. Representative ex vivo imaging of tissues and organs following oral administration of bovine milk‐derived sEVs.

JEX2-4-e70099-s001.docx (5.5MB, docx)

Acknowledgements

We thank the Mass Spectrometry & Metabolomics Core Facility, Laboratory Animal Resources Center and Center for Biomedical Research Core Facilities of Westlake University for sample analysis.

Zhang, Y. , Han J., Wu W., and Dang B.. 2025. “Oral Delivery of Semaglutide and Tirzepatide Using Milk‐Derived Small Extracellular Vesicles.” Journal of Extracellular Biology 4, no. 11: e70099. 10.1002/jex2.70099

Bobo Dang is the lead contact.

Contributor Information

Yuefei Zhang, Email: zhangyuefei@westlake.edu.cn.

Bobo Dang, Email: dangbobo@westlake.edu.cn.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

  1. Agrawal, A. K. , Aqil F., Jeyabalan J., et al. 2017. “Milk‐Derived Exosomes for Oral Delivery of Paclitaxel.” Nanomedicine 13: 1627–1636. 10.1016/j.nano.2017.03.001. [DOI] [PubMed] [Google Scholar]
  2. Aqil, F. , Jeyabalan J., Agrawal A. K., et al. 2017. “Exosomal Delivery of Berry Anthocyanidins for the Management of Ovarian Cancer.” Food & Function 8: 4100–4107. 10.1039/c7fo00882a. [DOI] [PubMed] [Google Scholar]
  3. Bosch, S. , De Beaurepaire L., Allard M., et al. 2016. “Trehalose Prevents Aggregation of Exosomes and Cryodamage.” Scientific Reports 6: 36162. 10.1038/srep36162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Brayden, D. J. , Hill T. A., Fairlie D. P., Maher S., and Mrsny R. J.. 2020. “Systemic Delivery of Peptides by the Oral Route: Formulation and Medicinal Chemistry Approaches.” Advanced Drug Delivery Reviews 157: 2–36. 10.1016/j.addr.2020.05.007. [DOI] [PubMed] [Google Scholar]
  5. Brown, T. D. , Whitehead K. A., and Mitragotri S.. 2020. “Materials for Oral Delivery of Proteins and Peptides.” Nature Reviews Materials 5: 127–148. 10.1038/s41578-019-0156-6. [DOI] [Google Scholar]
  6. Buckley, S. T. , Bækdal T. A., Vegge A., et al. 2018. “Transcellular Stomach Absorption of a Derivatized Glucagon‐Like Peptide‐1 Receptor Agonist.” Science Translational Medicine 10: eaar7047. 10.1126/scitranslmed.aar7047. [DOI] [PubMed] [Google Scholar]
  7. Carobolante, G. , Mantaj J., Ferrari E., and Vllasaliu D.. 2020. “Cow Milk and Intestinal Epithelial Cell‐Derived Extracellular Vesicles as Systems for Enhancing Oral Drug Delivery.” Pharmaceutics 12: 226. 10.3390/pharmaceutics12030226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Chen, Q. , Wu D., Wang Y., and Chen Z.. 2022. “Exosomes as Novel Delivery Systems for Application in Traditional Chinese Medicine.” Molecules (Basel, Switzerland) 27: 7789. 10.3390/molecules27227789 . [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Chen, Z.‐Q. , Tang T.‐T., Tang R.‐N., et al. 2025. “A Comprehensive Evaluation of Stability and Safety for HEK293F‐Derived Extracellular Vesicles as Promising Drug Delivery Vehicles.” Journal of Controlled Release 382: 113673. 10.1016/j.jconrel.2025.113673. [DOI] [PubMed] [Google Scholar]
  10. Cheng, L. , and Hill A. F.. 2022. “Therapeutically Harnessing Extracellular Vesicles.” Nature Reviews Drug Discovery 21: 379–399. 10.1038/s41573-022-00410-w. [DOI] [PubMed] [Google Scholar]
  11. Cocozza, F. , Grisard E., Martin‐Jaular L., Mathieu M., and Théry C.. 2020. “SnapShot: Extracellular Vesicles.” Cell 182: 262–262.e1. 10.1016/j.cell.2020.04.054. [DOI] [PubMed] [Google Scholar]
  12. Davies, M. , Pieber T. R., Hartoft‐Nielsen M.‐L., Hansen O. K. H., Jabbour S., and Rosenstock J.. 2017. “Effect of Oral Semaglutide Compared With Placebo and Subcutaneous Semaglutide on Glycemic Control in Patients With Type 2 Diabetes: A Randomized Clinical Trial.” JAMA 318: 1460. 10.1001/jama.2017.14752. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Deng, C. , Hu Y., Conceição M., et al. 2023. “Oral Delivery of Layer‐by‐Layer Coated Exosomes for Colitis Therapy.” Journal of Controlled Release 354: 635–650. 10.1016/j.jconrel.2023.01.017. [DOI] [PubMed] [Google Scholar]
  14. Dogan, A. B. , Marsh S. R., Tschetter R. J., et al. 2025. “Stabilizing Milk‐Derived Extracellular Vesicles (mEVs) Through Lyophilization: A Novel Trehalose and Tryptophan Formulation for Maintaining Structure and Bioactivity During Long‐Term Storage.” Journal of Biological Engineering 19: 4. 10.1186/s13036-024-00470-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Drucker, D. J. 2020. “Advances in Oral Peptide Therapeutics.” Nature Reviews Drug Discovery 19: 277–289. 10.1038/s41573-019-0053-0. [DOI] [PubMed] [Google Scholar]
  16. Esfahani, S. K. , Dehghani S., Hosseinzadeh H., et al. 2024. “An Exosomal Approach for Oral Delivery of Resveratrol: Implications for Inflammatory Bowel Disease Treatment in Rat Model.” Life Sciences 346: 122638. 10.1016/j.lfs.2024.122638. [DOI] [PubMed] [Google Scholar]
  17. FDA . 2017. “Non‐Clinical Review(s) of Ozempic (Semaglutide).” FDA. https://www.accessdata.fda.gov/drugsatfda_docs/nda/2017/209637Orig1s000PharmR.pdf. [Google Scholar]
  18. Frías, J. P. , Davies M. J., Rosenstock J., et al. 2021. “Tirzepatide Versus Semaglutide Once Weekly in Patients With Type 2 Diabetes.” New England Journal of Medicine 385: 503–515. 10.1056/NEJMoa2107519. [DOI] [PubMed] [Google Scholar]
  19. Gan, J. , Sun L., Chen G., Ma W., Zhao Y., and Sun L.. 2022. “Mesenchymal Stem Cell Exosomes Encapsulated Oral Microcapsules for Acute Colitis Treatment.” Advanced Healthcare Materials 11: e2201105. 10.1002/adhm.202201105. [DOI] [PubMed] [Google Scholar]
  20. Golan, M. E. , and Stice S. L.. 2024. “Extracellular Vesicle Lyophilization for Enhanced Distribution to the Point of Care.” Extracellular Vesicle 3: 100041. 10.1016/j.vesic.2024.100041. [DOI] [Google Scholar]
  21. Granhall, C. , Donsmark M., Blicher T. M., et al. 2019. “Safety and Pharmacokinetics of Single and Multiple Ascending Doses of the Novel Oral Human GLP‐1 Analogue, Oral Semaglutide, in Healthy Subjects and Subjects With Type 2 Diabetes.” Clinical Pharmacokinetics 58: 781–791. 10.1007/s40262-018-0728-4. [DOI] [PubMed] [Google Scholar]
  22. Grossen, P. , Portmann M., Koller E., et al. 2021. “Evaluation of Bovine Milk Extracellular Vesicles for the Delivery of Locked Nucleic Acid Antisense Oligonucleotides.” European Journal of Pharmaceutics and Biopharmaceutics 158: 198–210. 10.1016/j.ejpb.2020.11.012. [DOI] [PubMed] [Google Scholar]
  23. Haney, M. J. , Klyachko N. L., Zhao Y., et al. 2015. “Exosomes as Drug Delivery Vehicles for Parkinson's Disease Therapy.” Journal of Controlled Release 207: 18–30. 10.1016/j.jconrel.2015.03.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Herrmann, I. K. , Wood M. J. A., and Fuhrmann G.. 2021. “Extracellular Vesicles as a Next‐Generation Drug Delivery Platform.” Nature Nanotechnology 16: 748–759. 10.1038/s41565-021-00931-2. [DOI] [PubMed] [Google Scholar]
  25. Hintzen, J. C. J. , Devrani S., Carrod A. J., Bayik M. B., Tietze D., and Tietze A. A.. 2024. “Fluorescence Labeling of Peptides: Finding the Optimal Protocol for Coupling Various Dyes to ATCUN‐Like Structures.” ACS Organic & Inorganic Au Journal 4: 517–525. 10.1021/acsorginorgau.4c00030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kalluri, R. , and LeBleu V. S.. 2020. “The Biology, Function, and Biomedical Applications of Exosomes.” Science 367: eaau6977. 10.1126/science.aau6977. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Lau, J. , Bloch P., Schäffer L., et al. 2015. “Discovery of the Once‐Weekly Glucagon‐Like Peptide‐1 (GLP‐1) Analogue Semaglutide.” Journal of Medicinal Chemistry 58: 7370–7380. 10.1021/acs.jmedchem.5b00726. [DOI] [PubMed] [Google Scholar]
  28. Li, X. , Yu D., Chen X., Huang Z., and Zhao Y.. 2024. “A Strategy for Oral Delivery of FGF21 for Mitigating Inflammation and Multi‐Organ Damage in Sepsis.” International Journal of Pharmaceutics 656: 124115. 10.1016/j.ijpharm.2024.124115. [DOI] [PubMed] [Google Scholar]
  29. Mao, Y. , Han M., Chen C., et al. 2021. “A Biomimetic Nanocomposite Made of a Ginger‐Derived Exosome and an Inorganic Framework for High‐Performance Delivery of Oral Antibodies.” Nanoscale 13: 20157–20169. 10.1039/d1nr06015e. [DOI] [PubMed] [Google Scholar]
  30. Marsh, S. R. , Pridham K. J., Jourdan J., and Gourdie R. G.. 2021. “Novel Protocols for Scalable Production of High Quality Purified Small Extracellular Vesicles From Bovine Milk.” Nanotheranostics 5: 488–498. 10.7150/ntno.62213. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Mathieu, M. , Martin‐Jaular L., Lavieu G., and Théry C.. 2019. “Specificities of Secretion and Uptake of Exosomes and Other Extracellular Vesicles for Cell‐to‐Cell Communication.” Nature Cell Biology 21: 9–17. 10.1038/s41556-018-0250-9. [DOI] [PubMed] [Google Scholar]
  32. Merrifield, R. B. 1963. “Solid Phase Peptide Synthesis. I. The Synthesis of a Tetrapeptide.” Journal of the American Chemical Society 85: 2149–2154. 10.1021/ja00897a025. [DOI] [Google Scholar]
  33. Munagala, R. , Aqil F., Jeyabalan J., and Gupta R. C.. 2016. “Bovine Milk‐Derived Exosomes for Drug Delivery.” Cancer Letters 371: 48–61. 10.1016/j.canlet.2015.10.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Muttenthaler, M. , King G. F., Adams D. J., and Alewood P. F.. 2021. “Trends in Peptide Drug Discovery.” Nature Reviews Drug Discovery 20: 309–325. 10.1038/s41573-020-00135-8. [DOI] [PubMed] [Google Scholar]
  35. Oshchepkova, A. , Zenkova M., and Vlassov V.. 2023. “Extracellular Vesicles for Therapeutic Nucleic Acid Delivery: Loading Strategies and Challenges.” International Journal of Molecular Sciences 24: 7287. 10.3390/ijms24087287. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Piacentini, E. 2016. "Encapsulation Efficiency." In Encyclopedia of Membranes, edited by Drioli E. and Giorno L., 706–707. Springer. 10.1007/978-3-662-44324-8. [DOI] [Google Scholar]
  37. Popowski, K. D. , Moatti A., Scull G., et al. 2022. “Inhalable Dry Powder mRNA Vaccines Based on Extracellular Vesicles.” Matter 5: 2960–2974. 10.1016/j.matt.2022.06.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Qu, S. , Han Y., Liu Y., et al. 2022. “Milk Exosomes Facilitate Oral Delivery of Drugs Against Intestinal Bacterial Infections.” Journal of Agricultural and Food Chemistry 70: 16069–16079. 10.1021/acs.jafc.2c04971. [DOI] [PubMed] [Google Scholar]
  39. Salim, M. , Eason T., and Boyd B. J.. 2022. “Opportunities for Milk and Milk‐Related Systems as 'New' Low‐Cost Excipient Drug Delivery Materials.” Advanced Drug Delivery Reviews 183: 114139. 10.1016/j.addr.2022.114139. [DOI] [PubMed] [Google Scholar]
  40. Samuel, M. , Fonseka P., Sanwlani R., et al. 2021. “Oral Administration of Bovine Milk‐Derived Extracellular Vesicles Induces Senescence in the Primary Tumor but Accelerates Cancer Metastasis.” Nature Communications 12: 3950. 10.1038/s41467-021-24273-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Shandilya, S. , Rani P., Onteru S. K., and Singh D.. 2017. “Small Interfering RNA in Milk Exosomes Is Resistant to Digestion and Crosses the Intestinal Barrier In Vitro.” Journal of Agricultural and Food Chemistry 65: 9506–9513. 10.1021/acs.jafc.7b03123. [DOI] [PubMed] [Google Scholar]
  42. Si, Y. , Chen K., Ngo H. G., et al. 2022. “Targeted EV to Deliver Chemotherapy to Treat Triple‐Negative Breast Cancers.” Pharmaceutics 14: 146. 10.3390/pharmaceutics14010146. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Silva, A. M. , Lázaro‐Ibáñez E., Gunnarsson A., et al. 2021. “Quantification of Protein Cargo Loading Into Engineered Extracellular Vesicles at Single‐Vesicle and Single‐Molecule Resolution.” Journal of Extracellular Vesicles 10: e12130. 10.1002/jev2.12130. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Solis‐Herrera, C. , Kane M. P., and Triplitt C.. 2024. “Current Understanding of Sodium N‐(8‐[2‐Hydroxylbenzoyl] Amino) Caprylate (SNAC) as an Absorption Enhancer: The Oral Semaglutide Experience.” Clinical Diabetes 42: 74–86. 10.2337/cd22-0118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Somiya, M. , Yoshioka Y., and Ochiya T.. 2018. “Biocompatibility of Highly Purified Bovine Milk‐Derived Extracellular Vesicles.” Journal of Extracellular Vesicles 7: 1440132. 10.1080/20013078.2018.1440132. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Song, M. , Cui M., Fang Z., and Liu K.. 2022. “Advanced Research on Extracellular Vesicles Based Oral Drug Delivery Systems.” Journal of Controlled Release 351: 560–572. 10.1016/j.jconrel.2022.09.043. [DOI] [PubMed] [Google Scholar]
  47. Vader, P. , Mol E. A., Pasterkamp G., and Schiffelers R. M.. 2016. “Extracellular Vesicles for Drug Delivery.” Advanced Drug Delivery Reviews 106: 148–156. 10.1016/j.addr.2016.02.006. [DOI] [PubMed] [Google Scholar]
  48. van Niel, G. , D'Angelo G., and Raposo G.. 2018. “Shedding Light on the Cell Biology of Extracellular Vesicles.” Nature Reviews Molecular Cell Biology 19: 213–228. 10.1038/nrm.2017.125. [DOI] [PubMed] [Google Scholar]
  49. Wang, Z. , Popowski K. D., Zhu D., et al. 2022. “Exosomes Decorated With a Recombinant SARS‐CoV‐2 Receptor‐Binding Domain as an Inhalable COVID‐19 Vaccine.” Nature Biomedical Engineering 6: 791–805. 10.1038/s41551-022-00902-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Wu, L. , Wang L., Liu X. I., et al. 2022. “Milk‐Derived Exosomes Exhibit Versatile Effects for Improved Oral Drug Delivery.” Acta Pharmaceutica Sinica B 12: 2029–2042. 10.1016/j.apsb.2021.12.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Xia, B. , Hu R., Chen J., et al. 2024. “Oral Administration Properties Evaluation of Three Milk‐Derived Extracellular Vesicles Based on Ultracentrifugation Extraction Methods.” Advanced Healthcare Materials 13: e2401370. 10.1002/adhm.202401370. [DOI] [PubMed] [Google Scholar]
  52. Xia, B. , Xu F., Chen J., et al. 2025. “Site‐Specific Adaptive Nanovesicles for Oral Insulin Delivery.” Science Advances 11: eady6386. 10.1126/sciadv.ady6386. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Xiao, P. , Wang H., Liu H., et al. 2024. “Milk Exosome‐Liposome Hybrid Vesicles With Self‐Adapting Surface Properties Overcome the Sequential Absorption Barriers for Oral Delivery of Peptides.” ACS Nano 18: 21091–21111. 10.1021/acsnano.4c02560. [DOI] [PubMed] [Google Scholar]
  54. Xu, M. , Chen G., Dong Y., et al. 2022. “Liraglutide‐Loaded Milk Exosomes Lower Blood Glucose When Given by Sublingual Route.” ChemMedChem 17: e202100758. 10.1002/cmdc.202100758. [DOI] [PubMed] [Google Scholar]
  55. Yan, C. , Chen J., Wang C., et al. 2022. “Milk Exosomes‐Mediated miR‐31‐5p Delivery Accelerates Diabetic Wound Healing Through Promoting Angiogenesis.” Drug Delivery 29: 214–228. 10.1080/10717544.2021.2023699. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Yu, S. , Zhao Z., Xu X., Li M., and Li P.. 2019. “Characterization of Three Different Types of Extracellular Vesicles and Their Impact on Bacterial Growth.” Food Chemistry 272: 372–378. 10.1016/j.foodchem.2018.08.059. [DOI] [PubMed] [Google Scholar]
  57. Yu, Y. , Kuang Y.‐L., Lei D., et al. 2016. “Polyhedral 3D Structure of Human Plasma Very Low Density Lipoproteins by Individual Particle Cryo‐Electron Tomography1.” Journal of Lipid Research 57: 1879–1888. 10.1194/jlr.M070375. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Zhang, G. , Huang X., Xiu H., et al. 2020. “Extracellular Vesicles: Natural Liver‐Accumulating Drug Delivery Vehicles for the Treatment of Liver Diseases.” Journal of Extracellular Vesicles 10: e12030. 10.1002/jev2.12030. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Zhang, J. , Chen H., and Chen J.. 2022. “Colostrum‐Derived Exosome Based System: The Next‐Generation Nucleic Acid Delivery System With Enhanced Biosafety?” Molecular Therapy Nucleic Acids 30: 449–450. 10.1016/j.omtn.2022.10.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  60. Zhang, Y. , Belaid M., Luo X., et al. 2023. “Probing Milk Extracellular Vesicles for Intestinal Delivery of RNA Therapies.” Journal of Nanobiotechnology 21: 406. 10.1186/s12951-023-02173-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Zhang, Y. , Li Y., Liu P., et al. 2021. “Phosphatase Shp2 Regulates Biogenesis of Small Extracellular Vesicles by Dephosphorylating Syntenin.” Journal of Extracellular Vesicles 10: e12078. 10.1002/jev2.12078. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Zhao, Z. , Yu S., Li M., Gui X., and Li P.. 2018. “Isolation of Exosome‐Like Nanoparticles and Analysis of MicroRNAs Derived From Coconut Water Based on Small RNA High‐Throughput Sequencing.” Journal of Agricultural and Food Chemistry 66: 2749–2757. 10.1021/acs.jafc.7b05614. [DOI] [PubMed] [Google Scholar]
  63. Zhong, J. , Xia B., Shan S., et al. 2021. “High‐Quality Milk Exosomes as Oral Drug Delivery System.” Biomaterials 277: 121126. 10.1016/j.biomaterials.2021.121126. [DOI] [PubMed] [Google Scholar]
  64. Zhu, W. , Chao Y., Jin Q., et al. 2023. “Oral Delivery of Therapeutic Antibodies With a Transmucosal Polymeric Carrier.” ACS Nano 17: 4373–4386. 10.1021/acsnano.2c09266. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Figure S1. Characterization of semaglutide and semaglutide‐FAM. (A) Structure and LCMS analysis of semaglutide. (B) Structure and LCMS analysis of semaglutide‐FAM.

Figure S2. Characterization of tirzepatide and tirzepatide‐FAM. (A) Structure and LCMS analysis of tirzepatide. (B) Structure and LCMS analysis of tirzepatide‐FAM.

Figure S3. UV280 or FAM analysis in FPLC for semaglutide‐FAM (A), tirzepatide‐FAM (B) and sEVs (A‐B).

Figure S4. (A) The efficiency of loading tirzepatide‐FAM into sEVs. (B) Positive rate of sEVs loaded with tirzepatide‐FAM.

Figure S5. sEV size distribution and concentration analysis after lyophilization or not.

Figure S6. Representative ex vivo imaging of tissues and organs following oral administration of bovine milk‐derived sEVs.

JEX2-4-e70099-s001.docx (5.5MB, docx)

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Articles from Journal of Extracellular Biology are provided here courtesy of Wiley on behalf of the International Society for Extracellular Vesicles (ISEV)

RESOURCES