The role of small extracellular vesicles in ovarian cancer drug resistance: from bench clues to clinical implications

Abstract

Ovarian cancer (OC) represents the most lethal gynecologic malignancy, characterized by complex molecular heterogeneity, immune evasion, and therapy resistance. Chemotherapy resistance constitutes a formidable challenge in the global management of ovarian cancer, severely compromising therapeutic efficacy and precipitating adverse clinical outcomes. The chemoresistance phenotype originates from intrinsic resistance traits or genetic alterations, encompassing gene mutations and chromosomal variations. Emerging evidence has elucidated the capacity of small extracellular vesicles (sEVs) within the tumor microenvironment (TME) to horizontally transfer functional proteins and non-coding RNAs, thereby conferring chemoresistance phenotypes to drug-sensitive cells. This review delineates the packaging, sorting, and phenotypic transfer mechanisms of sEVs in the TME, elucidating how these processes facilitate the dissemination of resistance phenotypes across ovarian cancer progression. We also reviewed the established sEV-based diagnostic and prognostic utilities, as well as sEV-mediated drug delivery platforms. The study provides novel insights into related mechanisms and technologies in drug resistance of ovarian cancer.

Introduction

Ovarian cancer exhibits complex etiology and pathological characteristics, posing significant threats to women health while imposing substantial socio-economic burdens. OC ranks as the second deadliest gynecologic malignancy and the sixth leading cause of cancer-related mortality among women worldwide [1]. This grim clinical reality primarily stems from the absence of specific early symptoms, resulting in approximately 70% of patients being diagnosed at advanced stages (III/IV). Despite standard treatment comprising cytoreductive surgery followed by platinum-based (e.g., cisplatin, carboplatin) and paclitaxel-containing adjuvant chemotherapy that yield acceptable response, up to 70% of cases ultimately relapse due to the development of chemoresistance, with the 5-year survival rate remaining stagnant around 30–50%. This therapeutic bottleneck underscores the critical urgency of chemoresistance as a central challenge in ovarian cancer clinical management [2], highlighting the imperative need to comprehensively elucidate its underlying molecular mechanisms and develop effective intervention strategies. It represents the critical barrier underlying treatment failure and disease recurrence, enabling cancer cells to evade drug-induced apoptosis under selective pressure [3]. The mechanisms underlying drug resistance include tumor microenvironmental alteration, drug-metabolizing genes mutation, drug efflux transporter expression, enhanced drug efflux activity, etc. [2, 4]. Based on emergence and cellular origin, drug resistance is broadly classified into intrinsic or acquired resistance [5–7]. Intrinsic resistance denotes a pre-existing state driven by genomic instability [8]. In contrast, acquired resistance poses substantially greater clinical challenges, as tumors initially demonstrate robust therapeutic responses but subsequently relapse or progress following weeks to months of treatment, manifesting as a dynamic transition from chemo-sensitive to chemo-resistant phenotypes. The mechanisms lying behind encompass both genetic mutations and microenvironment-mediated non-genetic adaptations, which become progressively entrenched through clonal selection or epigenetic reprogramming under sustained therapeutic pressure [2, 9]. While genetic mutations confer permanent, heritable drug resistance, emerging evidence demonstrates that non-genetic pathways are equally critical in establishing chemoresistance [9]. Non-genetic resistance originates from reversible gene expression regulation, including epigenetic reprogramming that silences pro-apoptotic genes such as DAPK and BIM [10–13], or histone deacetylase upregulation in drug resistance-associated chromatin regions [14]. Additional mechanisms involve NF-κB and HIF-1α-mediated activation of anti-apoptotic programs, and post-transcriptional regulation via miR-21 [15] and miR-155 [16] that downregulates tumor suppressor genes. The dynamic interplay between Lamarckian induction and Darwinian selection provides a rapid yet reversible mechanism for establishing tumor resistance phenotypes, thereby increasing the probability of subsequent resistance-conferring genetic mutations. Notably, intrinsic and acquired resistance are not mutually exclusive. Intrinsically resistant clones may become enriched during therapy, while acquired-resistant cells may perpetuate resistance phenotypes through gene amplification, forming an interactive transfer network [17, 18].

Researches on sEVs originated from observations of 30–150 nm extracellular vesicles released from sheep reticulocytes, with early investigations focusing on their metabolic clearance functions. Seminal studies by Ratajczak and Skog established that sEVs transport mRNA and miRNA which modulate translation and gene expression in recipient cells, thereby validating their role in intercellular communication [19–21]. The 2013 Nobel Prize in Physiology or Medicine marked a monumental recognition for sEV research. Rothman’s SNARE zipper model elucidated sEV-target membrane fusion specificity [22], Südhof’s calcium-triggered fusion mechanism provided the molecular switch linking microenvironmental stress to accelerated secretion [23], and Schekman’s yeast genetic screens identified over 30 genes governing sEV biogenesis, subsequently recognized as direct regulators of sEV release [24–26]. sEVs are now widely acknowledged as critical mediators in a broad range of physiological and pathological processes. They participate in immune regulation [27], tissue repair [28], and cellular homeostasis [29, 30], while also playing pivotal roles in disease progression, including cancer development, neurodegenerative disorders, and cardiovascular diseases [31]. Through their capacity to transfer bioactive molecules such as proteins, lipids, and nucleic acids, sEVs facilitate complex intercellular signaling networks and contribute to the dynamic regulation of the cellular microenvironment [32]. These discoveries not only integrated sEVs into the canonical framework of cellular secretion but also catalyzed the paradigm of non-genetic horizontal phenotype transfer.

Traditional chemoresistance research focused predominantly on genetic or epigenetic alterations within tumor cells. However, accumulating evidence reveals that sEV-mediated, non-genetic phenotype transmission has emerged as a potent accelerator of resistance dissemination [9]. Unlike genetic resistance, which requires multiple cell divisions for establishment, exosomes can disseminate resistance phenotypes throughout the tumor microenvironment within hours, enabling rapid acquisition of resistance by drug-sensitive cells and synchronizing drug tolerance across the entire tumor population. This review delineates sEV packaging and sorting mechanisms, elucidates the phenotype transfer processes underlying resistance trait dissemination in OC, and explores the clinical potential of sEV-targeted therapeutic strategies in the management of ovarian cancer.

Biogenesis and cargo sorting of sEVs

ESCRT-dependent and independent sEV biogenesis

The generation of sEVs primarily relies on a set of protein complexes known as the Endosomal Sorting Complex Required for Transport (ESCRT). The ESCRT complex comprises five core subunits, including ESCRT-0 to III and VPS4 ATPase. ESCRT mediates intraluminal vesicles (ILVs) formation within multivesicular bodies (MVBs) for subsequent secretion upon plasma membrane fusion [33–35] (Fig. 1). VPS4-driven ATP hydrolysis induces ESCRT-III-mediated membrane scission, a prerequisite for packaging molecules into sEVs [36]. The Bro1 protein allosterically activates Vps4 ATPase activity via its V domain, accelerating ESCRT-III disassembly to upregulate sEV secretion [37]. Concordantly, disruption of ESCRT-associated proteins, including syndecan, syntenin, and Alix, diminished sEV release in MCF-7 breast cancer cells [38, 39]. Extending these findings, phosphorylated HRS overexpression drives immunosuppressive sEV secretion and restricts CD8⁺ T-cell infiltration in a murine model [40]. As a key ESCRT-0 component mediating initial cargo recognition and MVB sorting, HRS represents a critical regulatory node in sEV biogenesis [41].

Fig. 1.

The biogenesis, structure, and functions sEVs. SEVs are bilayered phospholipid vesicles ranging from 30 to 100 nm in diameter. ILVs are formed by the fusion of primary endocytic vesicles. MVBs fuse with the plasma membrane through the SNARE complex and release sEVs into the extracellular space. While a proportion of these vesicles undergo exocytosis into the extracellular space, others are shunted to lysosomes for different fates. After uptake by recipient cells through endocytosis, exosomal cargo is discharged into the cytosol to drive phenotypic transfer. Exosomal ncRNAs further modulate DNA transcription in target cells, governing phenotype regulation indirectly. The figure was originally created by the authors using Adobe Illustrator

On the other side, ESCRT-independent pathways encompass ceramide- and tetraspanin-mediated routes. Ceramide’s conical lipid geometry spontaneously induces negative membrane curvature to drive ILV formation. Trajkovic et al. demonstrated that sphingolipid ceramide induced the transfer of sEV-associated domains into the lumen of the endosome through an ESCRT-independent machinery [42]. In parallel, tetraspanins establish cargo-enriched microdomains that facilitate ESCRT-independent sorting. Tetraspanin CD63 directly participates in ESCRT-independent sorting of the PMEL luminal domain to ILVs, but not of traditional ESCRT-dependent cargoes. Inactivating CD63 impaired amyloid genesis and downstream melanosome morphogenesis both in vivo and in vitro [43].

At the genetic level, genome-wide CRISPR screening identified RAB31-GTP as a functional switch for the ESCRT-independent pathway that recruits PI4KIIβ to drive ILV formation while suppressing MVE degradation [44]. These findings suggest that cancer cells may dynamically toggle between pathway modes, thereby identifying a novel strategy targeting drug-resistant sEV biogenesis.

Cargo sorting of sEVs

Tumor cells employ sophisticated sorting machineries to determine sEV cargo composition, with core mechanisms encompassing post-translational modification and recognition [45]. Various proteins that contribute to sEV biogenesis are enriched on the inner leaflet of exosomal membranes [46]. Additional post-translational modifications, such as ubiquitination, SUMOylation, and phosphorylation, have also been implicated in cargo sorting [46].

Protein sorting predominantly occurs via the ESCRT-dependent machinery, wherein ubiquitinated membrane proteins are recognized by ESCRT-0 to III and invaginated into ILVs [39, 47, 48]. ESCRT-independent mechanisms rely on tetraspanin-enriched microdomains, including CD9, CD63, CD81 [49, 50] that recruit proteins like PMEL, or N-terminal myristoylation that mediates membrane anchoring and loading of cytosolic proteins [51]. RNA sorting mechanisms center on an active loading system wherein RNA-binding proteins (RBPs) recognize specific sequence motifs, including hnRNPA2B1 (GGAG, AGG, or UAG), hnRNPK (AsQGnA), SYNCRPI (GGCU), and FMRP (AAUGC) [46, 52]. For instance, hnRNPA2B1 was reported to facilitate colorectal cancer liver metastasis and bladder cancer lymphatic metastasis by regulating the loading of miRNAs into sEVs [53, 54]. The RNA sorting process is further modulated by SUMOylation and phosphorylation, and requires ER–MVB contact sites to target RBP–RNA complexes for transport [55]. By contrast, passive RNA sorting machinery merely depends on intracellular RNA abundance. DNA sorting remains relatively underexplored. However, recent findings reveal that non-apoptotic cell death-associated sEVs actively encapsulate cytoplasmic dsDNA, ssDNA and mtDNA from genomically unstable tumor cells, maintaining cellular homeostasis through expulsion of harmful DNA [46]. High abundance of DNA was detected in sEVs isolated from prostate cancer patient plasma [56]. Whole genome sequencing demonstrated that these sEV-derived DNA is enriched for genes frequently altered in metastatic prostate cancer (e.g., MYC, AKT1, PTK2, KLF10, and PTEN). These sorting mechanisms not only confer cell type-specific molecular fingerprints upon sEVs but also represent key nodes for tumor diagnostic biomarker and targeted therapeutic intervention.

Biogenesis regulation of sEVs in tumor microenvironment

The physicochemical condition of the tumor microenvironment contributes to sEV synthesis and release, thereby establishing intercellular communication networks [27, 57, 58]. Hypoxia induces HIF-1α-mediated upregulation of sEV biogenesis, promoting drug resistance. Jiang et al. reported that HIF1α bound to the lincROR promoter, driving the transcription and exosomal packaging in an hnRNPA1-dependent axis to disseminate chemoresistance to recipient cells [59]. Besides, a hypoxic exosomal circRNA-mediated mechanism in pancreatic cancer cells has been reported. CircZNF91 competitively bound miR-23b-3p, derepressing SIRT1 inhibition. Consequently, upregulated SIRT1 enhanced deacetylation-dependent HIF-1α stabilization, promoting gemcitabine resistance in recipient cells [60]. The acidic tumor microenvironment is considered another factor driving sEV release and accelerating drug efflux [45, 61, 62]. Inhibition of sEV biogenesis via proton pump inhibitors (omeprazole, lansoprazole), the ceramide synthesis inhibitor GW4869, or the ABCA3 inhibitor indomethacin reduced chemotherapeutic packaging into sEVs and restored chemosensitivity [63, 64]. A combined treatment of 0.5µM PEGylated liposomal doxorubicin and GW4869 in U937 acute myeloblastic leukemia cells produced cytotoxicity comparable to 1 µM of either agent alone [65]. Notably, chemotherapeutic agents can be directly sequestered and expelled by sEVs, while DNA damage concomitantly activates c-Myc, which enhances sEV-mediated horizontal transfer of miR-425-3p by binding its promoter in lung cancer [66]. Exosomal miR-425-3p targets AKT1 to promote autophagic activation in recipient cells, culminating in chemoresistance [67].

Molecular mechanisms of sEV-mediated chemoresistance transfer

TME-derived sEVs participate in numerous pathological processes such as cancer proliferation, metastasis, immune suppression, and chemotherapy resistance. Chemoresistance is categorized into intrinsic resistance, driven by genomic alterations, and acquired resistance, which evolves through interactions within the tumor microenvironment [68]. SEV-mediated transfer of chemoresistant phenotypes to drug-sensitive cells through drug efflux, ATP-binding cassette (ABC) transporters, anti-apoptotic signaling upregulation, and epithelial-to-mesenchymal transition (EMT) [69]. Beyond protein cargo, sEVs harbor non-coding RNAs such as lncRNAs and miRNAs, which modulate protein expression in recipient cells to activate chemoresistant pathways (Fig. 2).

Fig. 2.

SEV-mediated chemoresistance in ovarian cancer. SEVs induced chemoresistance in ovarian cancer in multiple ways. 1) SEVs accelerate drug efflux. 2) SEVs enhanced chemoresistance via drug transporters (ABC transporters). 3) Exosomal ncRNAs participate in intercellular epigenetic regulation. 4) Exosomal protein targets various resistance-associated signaling pathways. 5) SEVs act as messengers in EMT-associated chemoresistance in ovarian cancer. The figure was originally created by the authors using Adobe Illustrator

SEV-mediated drug efflux

A central hallmark of tumor chemoresistance is reduced intracellular therapeutics concentration, wherein sEV-mediated drug efflux plays a pivotal role. Hydrophobic chemotherapeutic agents tend to be encapsulated by exosomal membrane lipids, subsequently being secreted to diminish intracellular drug concentrations [45]. Safaei et al. reported that cisplatin-resistant ovarian cancer cells released sEVs containing more than 2-fold higher cisplatin levels compared to their sensitive counterparts in vivo [70]. Meanwhile, cisplatin resistance is closely linked to lysosomal dysfunction, characterized by significantly reduced lysosomal expression and downregulated LAMP1/2 proteins, while paradoxically exhibiting elevated LAMP1 in secreted sEVs. These results indicate compromised lysosomal storage capacity in ovarian cancer [70]. The tumor microenvironmental hypoxia exacerbates this process. Hypoxic ovarian cancer cells significantly increase their sEV release by upregulating Rab27a while downregulating Rab7, LAMP1/2, and NEU-1, promoting a secretory lysosomal phenotype against cisplatin treatment [71].

Protein post-translational modifications further contribute to sEV-mediated drug resistance. Aberrant O-GlcNAcylation and endoplasmic reticulum (ER) stress disequilibrium drive CSN5-mediated platinum resistance in epithelial ovarian cancer (EOC), with CSN5 inhibition achieving tumor regression and overcoming intrinsic platinum resistance in preclinical models [72]. Similarly, Saini et al. identified TMEM205-mediated platinum resistance in ovarian clear cell carcinoma via exosomal pathways, wherein TMEM205 undergoes ligand-independent constitutive endocytosis and Rab11 co-localization to modulate late recycling endosomes [73]. Notably, CLPTM1L (CRR9) confers chemoresistance both through intrinsic plasma membrane and exosomal transfer to bystander cells in a domain-dependent manner, with anti-CLPTM1L monoclonal antibodies successfully reversing drug resistance in animal models [74]. Collectively, these studies reveal that sEV-mediated drug efflux operates through both cell-autonomous and non-autonomous mechanisms in ovarian cancer, regulated at multiple levels by hypoxia, metabolic reprogramming, and protein modifications.

Exosomal drug efflux transporters

Apart from direct chemotherapeutics efflux, exosomal drug efflux transporters also facilitate horizontal transfer of drug resistance phenotypes in ovarian cancer [75], representing the most extensively investigated mechanism of sEV-mediated resistance propagation [76]. The ATP-binding cassette (ABC) transporter family comprises over 20 members that efflux chemotherapeutic agents. Elevated expression of these transporters positively correlates with chemoresistance in ovarian cancer [75]. These transporters enable the expulsion of a wide range of structurally and functionally diverse drug molecules, forming the classic molecular basis for multidrug resistance (MDR) [45, 77]. For instance, ABCB1, 4, 6 was demonstrated to induce the resistance against docetaxel, paclitaxel and doxorubicin in ovarian cancer [78–80]. Among these, P-glycoprotein (P-gp) plays a pivotal role in chemoresistance phenotypes transfer across multiple malignancies, including ovarian carcinoma. SEVs enable direct transfer of functional transporters from resistant to sensitive cells, thereby conferring acquired resistance independent of genetic mutations. Intercellular P-gp transfer between tumor cells occurs without direct contact, a phenomenon subsequently confirmed to be sEV-mediated both in vitro and in vivo [81]. Zhang et al. reported that paclitaxel-resistant human ovarian cancer cells (A2780/PTX) released P-gp-enriched sEVs at markedly higher rates than parental sensitive cells (A2780/WT), with exosomal transfer redistributing chemotherapeutics in recipient cells and conferring more than 5-fold elevated resistance to both doxorubicin and paclitaxel. Similarly, elevated miR-1246 and Cav1 overexpression from paclitaxel-resistant ovarian cancer sEVs resensitize OC cells via the Cav1/P-gp/M2 macrophage axis [82]. These findings underscore that sEV-mediated efflux transporter transfer, regulated by ESCRT-dependent MVBs, constitutes a critical mechanism for rapid chemoresistance dissemination throughout the tumor microenvironment.

Exosomal transfer of ncRNAs

Non-coding RNAs (ncRNAs) constitute a diverse class of regulatory transcripts that play critical roles in cancer pathogenesis [83]. Exosomal ncRNAs, predominantly miRNAs together with lncRNAs and circRNAs, serve as efficient messengers for horizontal transfer of chemoresistance phenotypes between cancer cells, modulating resistance-related gene expression to establish drug tolerance [68]. SEV-encapsulated miRNAs are protected from intracellular RNase degradation, enabling stable intercellular phenotypic transfer in ovarian cancer (Table 1). MiR-21 was reported to be highly enriched in sEVs from cancer-associated adipocytes (CAAs) and fibroblasts (CAFs), and its transfer to ovarian cancer cells suppressed apoptosis and conferred paclitaxel resistance by targeting APAF1 [83]. Similarly, miR-21-3p promoted cisplatin resistance through NAV3 downregulation [84]. MiRNAs also confer chemoresistance via cell cycle modulation. MiR-433 induced senescence and paclitaxel resistance by targeting CDK6, MAPK14, E2F3, and CDKN2A [85]. In addition, exosomal miR-221-3p derived from CD163⁺ tumor-associated macrophage (TAM) activated the AKT pathway and induced EMT by downregulating ADAMTS6, exhibiting cisplatin resistance within ovarian cancer ascites [86]. Furthermore, miR-223 was enriched in TAM-derived sEVs under hypoxia, which enhanced drug resistance in co-cultured ovarian cancer cells (A2780) [87]. Exosomal metabolic reprogramming represents another potential mechanism. MiR-21-5p derived from cisplatin-resistant SKOV3 promoted glycolysis and inhibited chemosensitivity by targeting PDHA1 in its progenitor [88], while plasma exosomal miR-21-5p contributed to OC progression by regulating CDK6 [89]. Interestingly, exosomal miR-484 derived from ovarian cancer cells promoted vascular normalization, thereby enhancing tumor chemosensitivity and prolonging survival in tumor-bearing mice receiving chemotherapy [90]. Decreased serum exosomal miR-484 expression correlated with aggressive clinicopathological features and diminished survival, particularly in patients with elevated CA-125 [91].

Table 1.

Exosomal ncRNAs associated with chemoresistance in ovarian cancer

Type	Exosomal cargo	Mechanism	Effect	References
miRNA	miR21-3p	NAV3	Inducing cisplatin resistance in A2780, potentially by targeting the NAV3 gene.	[84]
	miR433	CDK6/MAPK14/E2F3/CDKN2A	Enhancing survival rate of PEO1 and PEO4 OC cells against paclitaxel treatment.	[85]
	miR221-3p	ADAMTS6/EGFR/AKT, SNAIL1/ZEB1	Overexpression of ADAMTS6 reducing the IC50 of cisplatin in ovarian cancer cells.	[86]
	miR223	PTEN-PI3K/AKT	Hypoxic macrophages-derived sEVs enhanced the resistance phenotype in A2780 and SKOV3	[87]
	miR21-5p	PDHA1	Suppressed SKOV3 cisplatin sensitivity and promoted cell viability and glycolysis via PDHA1 inhibition	[88]
	miR21-5p	CDK6	Inhibited the expression of CDK6 by targeting 3’-UTR, increasing tumor volume, size, and weight in vivo.	[89]
	miR484	VEGF-A	Improved the vascular normalization and sensitized the cancer to chemotherapy	[90]
	miR484	-	Combination of miR484/CA-125 showed an elevated AUC in identifying OC patients	[91]
	miR6836	DLG2/Yap1/TEAD1	Exosomal miR-6836 reversed cisplatin response in EOC cells.	[92]
	miRNA296-3p	PTEN/AKT, SOCS6/STAT3	Promoted the proliferation, migration, invasion, and drug resistance of OC in vitro.	[93]
lncRNA	UCA1	miR143/FOSL2	Knockdown of UCA1 inhibited cell proliferation and promoted cisplatin-induced cell apoptosis.	[94]
	PANDAR	SRSF9/SIRT4/SIRT6	PANDAR led to a rapid adaptation of OC cell survival under cisplatin stress exposure.	[95]
circRNA	circ_WHSC1	miR-145, miR-1182/MUC1, hTERT	Induced tumor metastasis through peritoneal mesothelium.	[96]
	circ_C20orf11	miR-527/YWHAZ	Induced cisplatin resistance and promoted macrophage M2 polarization.	[97]
	circ_PIP5K1A	miR942-5p/NFIB	Promoted proliferation and migration, as well as increased cisplatin-resistance in OC	[98]

Research on exosomal lncRNA mechanisms in ovarian cancer chemoresistance remains considerably less comprehensive, leaving substantial scope for future investigation. lncRNA urothelial carcinoma-associated 1 (UCA1) was upregulated in serum sEVs from cisplatin-resistant patients and modulated cisplatin resistance through the miR-143/FOSL2 axis [94]. OC-derived sEVs contained high level of lncRNA promoter of CDKN1A antisense DNA damage activated RNA (PANDAR), which significantly enhanced cell survival and chemoresistance in vitro while promoting tumor progression and metastasis in vivo by recruiting SRSF9 to nuclear bodies under cisplatin stress [95].

Exosomal circRNAs predominantly mediate chemoresistance transfer by modulating miRNA signals through a “sponge effect”. Zong et al. demonstrated that circWHSC1 was upregulated in ovarian cancer tissues, with expression inversely correlating with tumor differentiation, and promoted oncogenesis by sponging miR-145 and miR-1182 to upregulate MUC1 and hTERT [96]. CircC20orf11 enhanced cisplatin resistance and attenuated apoptosis by sponging miR-527 to promote YWHAZ expression [97]. Exosomal circ-PIP5K1A facilitated intercellular transmission of cisplatin resistance through the miR-942-5p/NFIB axis [98]. These ncRNAs function as not only molecular vehicles for chemoresistance phenotype transfer but also as predictive biomarkers for therapy response and precision therapeutic targets for overcoming drug resistance.

Exosomal proteins and chemoresistance

Plasma gelsolin (pGSN), a secreted actin-binding protein encoded by GSN and transported via sEVs, ranks as the fourth most abundant blood protein and demonstrates clearance efficacy for multiple substances. Though conventional OC biomarker CA125 has long been effective in monitoring chemotherapy responses, there is currently no biomarker capable of evaluating treatment response prior to chemotherapy. Exosomal pGSN (Ex-pGSN) holds considerable potential for predicting chemotherapy efficacy in ovarian cancer. The ratio of total pGSN and exosomal pGSN was a promising predictive marker for chemoresistance [99]. Asare-Werehene et al. revealed that Ex-pGSN upregulated HIF1α-mediated pGSN expression in chemoresistant OC cells through an autocrine feedback loop, as well as conferred cisplatin resistance in chemosensitive OC cells [100]. Besides direct phenotypic induction, pGSN induces chemoresistance through immunomodulation within the tumor microenvironment. Cancer cell-derived exosomal pGSN was reported to induce CD8⁺ T cell apoptosis, while promoting glutathione synthesis and cisplatin resistance [101]. The team also reported that elevated epithelial pGSN skewed macrophage polarization toward a pro-tumorigenic phenotype, reducing M1 macrophage viability through caspase-3 activation while suppressing iNOS and TNFα production [101]. Additionally, OC-derived Ex-pGSN induced NK cell death through direct contact [102]. Mechanistically, Ex-pGSN increased T-cell immunoglobulin and mucin-domain-containing-3 expression (TIM-3) on activated NK cells which decreased interferon-γ production [102].

A disintegrin and metalloprotease 17 (ADAM17), another key mediator of chemoresistance, promotes tumor cell proliferation and survival through EGFR pathway activation via ligand cleavage (e.g., amphiregulin) [69]. ADAM17 serves as a critical regulatory hub linking cell surface events to downstream signaling, exerting multifaceted functions in inflammation, cell proliferation, and TME modulation, constituting a promising therapeutic target. Hedemann et al. identified ADAM17 as an essential upstream regulator of AREG release under chemotherapeutic treatment in ovarian cancer cell. Inhibition of ADAM17 with either GW280264X or anti-ADAM17 antibody D1 as well as silencing of ADAM17 by siRNA sensitized cancer cells to cisplatin-induced apoptosis and significantly reduced cell viability [103]. In addition, chemotherapeutic treatment stimulated OC cells to release bioactive ADAM17 via sEVs, while patient-derived sEVs induced AREG shedding and restored chemoresistance in ADAM17-deficient cells [104].

DNA methyltransferase 1 (DNMT1) is the core maintenance enzyme for DNA methylation, preferentially targets hemi-methylated DNA to preserve epigenetic information, significantly contributes to chemoresistance [105–107]. DNMT1 was upregulated in ovarian cancer tissues and cell lines, with DNMT1 transcripts markedly enriched in sEVs derived from OC-conditioned medium. Exposure to these sEVs stimulated endogenous DNMT1 expression in recipient cells, conferring resistance to cisplatin cytotoxicity. In vivo administration of exosomal DNMT1 accelerated xenograft progression and significantly reduced overall survival in mouse models. Furthermore, treatment with the sEV inhibitor GW4869 greatly restored chemosensitivity in resistant cells [108].

Annexin A3, a calcium-dependent phospholipid-binding protein, plays a critical role in various malignancies, influencing tumor cell proliferation, invasion, metastatic potential, and mediate resistance to chemotherapeutics. As a member of the phospholipid-binding protein family, annexin A3 mediates membrane fusion, exocytosis, ion channel modulation, and signal transduction [109]. Its aberrant expression serves as a biomarker for early diagnosis, as well as a potential therapeutic target for reversing tumor drug resistance. Yan et al. reported that significantly elevated annexin A3 was detected in patients with platinum-resistant ovarian cancer than those platinum-sensitive patients. The intracellular platinum concentration and platinum-DNA adduct was significantly lower in annexin A3-overexpressing cells than those in parental cells. The decreased cisplatin concentration was also accompanied by reduced induction of p53, which could be restored by downregulation of annexin A3 [110]. The team further documented that significantly increased secretion of sEVs in high annexin A3-expressing ovarian cancer cells has been observed, indicating that sEV-mediated release as the principal mechanism for annexin A3 secretion in ovarian cancer [111].

SEV-associated EMT in ovarian cancer

Epithelial-mesenchymal transition denotes a fundamental biological process whereby epithelial cells undergo phenotypic conversion to acquire mesenchymal characteristics, including enhanced motility and invasiveness. This transformation facilitates basement membrane breach and distant metastasis while activating cancer stem cells (CSCs) that exhibit self-renewal capacity and chemotherapeutic resistance to reinitiate cancer population [5]. In ovarian cancer, sEV-mediated EMT transmission critically contributes to chemoresistance. Pretreatment of A2780 cells with sEVs derived from platinum-resistant counterparts conferred a 2-fold increase in carboplatin viability compared to controls, accompanied by enhanced EMT marker expression [112]. The following DNA sequencing identified SMAD4 mutations within platinum-resistant cells. Notably, SMAD4-mutant A2780 cells exhibited both upregulated EMT signatures and 1.7-fold increased sEV secretion compared to parental cells, providing seminal evidence that acquired SMAD4 mutations augment chemoresistance while promoting sEV-mediated EMT phenotype propagation. Additionally, among primary tumor-derived, metastasis-derived, CAFs-derived, and normal fibroblasts-derived sEVs from stage IIIC ovarian cancer patients, CAFs-derived sEVs exhibited highest TGFβ1 level compared to other groups. Upon uptake by SKOV-3 and CAOV-3 ovarian cancer cell lines, CAF-derived sEVs induced malignant phenotypes, including enhanced migration, invasion, and SMAD pathway-dependent EMT [113].

Therapeutic potentials of sEVs in chemoresistance

In recent years, accumulating evidence substantiates sEVs as promising therapeutic targets in ovarian cancer [27, 58, 114]. Their role as key mediators of intercellular communication provides unique endogenous platforms for therapeutic intervention, with emerging strategies targeting sEV-mediated chemoresistance, offering novel avenues for cancer therapy [115]. The related researches have been summarized in Table 2 in detail.

Table 2.

Therapeutic implications of sEVs in OC chemoresistance

Therapeutic target	Resistance	Mechanisms	Reference
Amiloride	Cisplatin	STAT3/Rab	[71]
Heparin	Cisplatin	p38/JNK	[116]
Exosomal miR21	Paclitaxel	APAF1	[117]
TP/miR497	Cisplatin	PI3K/AKT/mTOR	[118]
M1exo-CISs	Cisplatin	PTEN/miR-130a/P-gp	[119]
Exo-PTX	Paclitaxel	P-gp	[120]
Exo-B2C	MDR	NADH oxidation	[121]
LncRNA PLADE	Cisplatin	HNRNPD/DNA Hybrids	[122]
Exo-TMP	Paclitaxel	MDR1/MRP1/GSTP1	[123]
Exo-miR484	Cisplatin	Vascular normalization	[90]
Exo-miRNA296-3p	-	TEN/AKT and SOCS6/STAT3	[93]

Exosomal biogenesis as therapeutic targets

Given the multifaceted roles of sEVs in OC pathogenesis and chemoresistance development, strategies targeting sEVs release, uptake, and biogenesis processes represent a promising therapy. Hypoxic ovarian cancer cells exhibited enhanced cisplatin efflux via sEVs. Co-culture with resistant OC-derived sEVs led to significantly decreased dsDNA damage and increased cell survival in response to cisplatin treatment. Pharmacological blockade of sEV release using amiloride or STAT3 inhibitors significantly increased cisplatin-induced apoptosis and decreased clonogenic survival [71]. Similarly, cisplatin treatment led to the release of sEVs which induced invasion and increased resistance when taken up by bystander cells. The application of sEV uptake inhibitors (heparin) reversed this adaptive response, reducing the IC50 of A2780 cells from 31.3 to 21.2 µM (p < 0.0001) [116, 124–126]. Targeting specific exosomal cargo also holds therapeutic promise. Exosomal miR-21 transfer from CAAs or CAFs to OC cells conferred paclitaxel resistance, establishing miR-21 as a targetable mediator of chemoresistance in metastatic and recurrent disease [117].

Engineered sEVs as therapeutic platforms in ovarian cancer

While sEVs offer inherent advantages as endogenous nanocarriers, clinical translation is hindered by limitations in biodistribution specificity, drug-loading capacity, and targeting efficiency [127–129]. Through precise engineering of surface proteins, cargo sorting, and membrane architecture, engineered sEVs can be developed as therapeutic platforms with enhanced loading, active targeting, and stimuli-responsive properties, thereby improving chemotherapy efficacy [127]. One promising strategy involves engineering sEVs as therapeutic vehicles for tumor-targeted drug delivery. EV-liposome hybrid nanoparticles were developed by fusing CD47-expressing tumor sEVs with RGD-modified liposomes for co-delivery of miR-497 and TP in ovarian cancer [118]. These hybrids demonstrated effective tumor accumulation and significant anticancer activity in vivo, while promoting dephosphorylation of hyperactivated PI3K/AKT/mTOR signaling and enhancing ROS generation to overcome cisplatin resistance. Similarly, umbilical cord blood-derived M1 macrophage sEVs loaded with cisplatin (M1exoCISs) were evaluated for treating platinum-resistant ovarian cancer [119]. In multiple OC cell lines (A2780, SKOV3, A2780/DDP, and SKOV3/DDP), M1exoCISs increased cytotoxicity 3.3-fold against resistant A2780/DDP cells and 1.4-fold against sensitive A2780 cells, effectively reversing cisplatin resistance. Additionally, paclitaxel-loaded sEVs exhibited more than 50-fold increased solubility and cytotoxicity against MDR cell lines (MDCKMDR1), overcoming P-gp-mediated drug efflux [120].

Another potential therapeutic approach involves engineering sEVs to modulate tumor chemoresistance phenotypes. Kang et al. demonstrated that small sEVs reversed drug resistance by circumventing efflux mechanisms and delivering small molecules directly to MDR cancer cells [121]. The team developed GSH-depleting benzoyloxy dibenzyl carbonate (B2C)-encapsulated sEVs, which overcame the efflux system to exert highly potent anticancer activity against human MDR ovarian cancer cells (OVCAR-8/MDR) by depleting GSH, thereby inducing oxidative stress and apoptotic cell death in both resistant and parental cell line [121]. Platinum sensitivity-related LncRNA from Ascites-Derived Exosomes (PLADEs) transmitted by ascites sEVs enhanced platinum sensitivity in high-grade serous ovarian cancer. Mechanistically, exosomal PLADE downregulated heterogeneous nuclear ribonucleoprotein D (HNRNPD) by VHL-mediated ubiquitination, thus inducing an increased DNA hybrids (R-loop) formation which promoted cisplatin sensitivity in HGSOC [122]. RGD-modified sEVs loaded with miR-484 induced vessel normalization and in turn sensitized the cancer cells to chemotherapy-induced apoptosis [90]. Nevertheless, tetramethylpyrazine (TMP) encapsulated via an sEV loading strategy (EXO-TMP) leveraged the innate tumor-targeting properties and biocompatibility of sEVs to enhance delivery efficiency and specificity for OC chemotherapy [123]. The study demonstrated that EXO-TMP reversed the MDR phenotype in OC by downregulating the expression of multidrug resistance protein 1 (MDR1), multidrug resistance-associated protein 1 (MRP1), and glutathione S-transferase Pi 1 (GSTP1), thereby inhibiting drug efflux and reducing metabolic inactivation. Furthermore, EXO-TMP exhibited specific targeting capacity toward A2780T cells and demonstrated synergistic antitumor efficacy when combined with paclitaxel.

Clinical transitions of sEV-based strategies

Beyond mechanistic insights, sEV-based strategies have rapidly transitioned into clinical applications with numerous trials currently evaluating their utility in cancer diagnosis, prognosis prediction, and therapeutic intervention. The ExoDx Prostate test represents the first FDA-recognized EV-based liquid biopsy. By analyzing three exosomal RNA transcripts (PCA3, ERG, and SPDEF) in urine, the process may identify high-grade prostate cancer with a negative predictive value of 91%, thereby reducing unnecessary biopsies [130]. Similarly, the EXONERATE study demonstrated that an exosome-based liquid biopsy incorporating cell-free and exosomal miRNA signatures could robustly predict progression-free survival and overall survival in metastatic colorectal cancer patients receiving anti-EGFR therapy, outperforming traditional tumor sidedness approaches [131]. Engineered sEVs have also entered therapeutic clinical trials. MSC-derived exosomes loaded with siRNA targeting oncogenic KRAS^G12D, have shown feasibility and safety in a Phase I trial for metastatic pancreatic cancer [132], with preliminary evidence of reduced level of circulating mutant DNA [133]. Additionally, commercially developed platforms such as exoIL-12 (surface-displayed IL-12 for cutaneous T-cell lymphoma), exoSTING (STING agonist-loaded exosomes for solid tumors), and exoASO-STAT6 (antisense oligonucleotide delivery for hepatocellular carcinoma) have completed or are undergoing initial clinical evaluation [133]. These trials collectively underscore the translational momentum of sEV-based strategies, while regulatory approval and large-scale efficacy validation remain pending. A summary of other relevant clinical studies is provided in the table below (Table 3).

Table 3.

Clinical trials of sEV-based strategies in cancer management

Cancer Type	Application	Study Design	Source/Feature	Reference
Ovarian cancer	Prognosis	Prospective observation	Peritoneal fluid	[134]
Lymphomas	Prognosis	Prospective observation	CAR + EVs	[135]
omCSPC	Prognostic/ Diagnostic	Biomarker validation	Plasma	[136]
ESCC	Prognostic/ Diagnostic	Phase I clinical trial	Salivary extracellular vesicles	[137]
Pancreatic cancer	Therapeutic	Phase I, open-label	MSCs	[138]
Lung cancer	Therapeutic	Phase II, open-label	Dendritic cells	[139]
Hepatocellular carcinoma	Therapeutic	Phase I, dose escalation	MSC	-
Colorectal cancer	Therapeutic	Phase I, open-label	Plasma	-
Metastatic colorectal cancer	Prognostic	Biomarker validation	Plasma	[131]

Limitations and future perspectives

Despite the promising advances in understanding sEV-mediated chemoresistance in ovarian cancer, several critical limitations must be acknowledged. A standardized isolation protocol is imperative to ensure reproducibility and comparability across different studies. Current sEV isolation techniques, including ultracentrifugation, size-exclusion chromatography, and precipitation-based methods, yield heterogeneous populations with varying degrees of contamination from lipoproteins, protein aggregates, and other non-vesicular components [140]. This heterogeneity compromises the reproducibility of functional studies and biomarker validation across different laboratories [141]. The updated MISEV2023 guidelines provide improved reporting standards and advocate for the use of the EV-TRACK database to enhance transparency [142, 143]. Yet universal adoption across the field remains incomplete, and no single gold standard method currently exists.

Functional specificity and cargo heterogeneity pose significant hurdles. The dynamic nature of sEV cargo loading, influenced by cellular stress, metabolic status, and microenvironmental cues, results in context-dependent functional outputs which can be difficult to predict or standardize [141]. Distinguishing sEV subpopulations based solely on biophysical properties remains technically challenging, as ultracentrifugation-based methods may cause particle aggregation and loss of integrity, while precipitation methods introduce higher levels of contaminants [144]. The structural and biomolecular heterogeneity of sEV isolates obtained through different methods further complicates mechanistic interpretations.

Clinical translation barriers are substantial. While preclinical studies demonstrate therapeutic potential, scalable GMP-compliant manufacturing of clinical-grade sEVs, quality control metrics, and regulatory frameworks remain underdeveloped [145]. Producing sEVs at scale while preserving their functional integrity is resource-intensive, and downstream purification processes are often time-consuming and costly. Additionally, systemic inhibition of sEV secretion or uptake may disrupt physiological intercellular communication, underscoring the need for tumor-specific targeting strategies to minimize off-target effects. Currently, a major barrier in sEV research is the lack of standardized isolation methods that are readily translatable to clinical applications [146].

Future studies should prioritize the development of standardized isolation protocols, single-vesicle analysis technologies, and prospective clinical trials to validate sEV-based diagnostics and therapeutics in cancer management. Emerging technologies, such as nano-flow cytometry combined with high-resolution microscopy, offer promising avenues to address sEV heterogeneity at the single-particle level [147].

Conclusion

SEV-mediated horizontal transfer has emerged as a critical non-genetic accelerator of chemoresistance in ovarian cancer, enabling tumor cells to rapidly disseminate resistance-conferring cargo to drug-sensitive bystanders. This rapid phenotype synchronization, regulated through multiple biogenesis pathways, provides revolutionary insights into therapeutic failure by demonstrating that chemoresistance may propagate independently of genetic mutations or clonal selection. Clinically, sEV-based liquid biopsy demonstrates transformative theragnostic potential, offering exceptional diagnostic accuracy for early cancer detection and surpassing static genomic profiling through real-time monitoring of tumor evolution and therapeutic response [148]. Despite these advances, clinical translation faces substantial hurdles. Standardized isolation protocols and comprehensive biomarker validation remain under debate, while systemic sEV inhibition poses risks of disrupting normal intercellular communication. Engineered sEVs, though promising as therapeutic delivery vehicles, require scalable manufacturing processes and rigorous safety assessment. Future research should prioritize tumor-specific targeting mechanisms, integrated detection platforms, and patient-oriented therapeutic strategies. Combining sEV-targeted interventions with conventional therapies offers the potential to disrupt horizontal resistance networks while eradicating resistant OC clones, ultimately transforming these molecular mediators from drivers of therapeutic failure into precision tools for cancer treatment. Collectively, this study provides novel insights into the role of sEVs in ovarian cancer resistance and highlights their potential for therapeutic exploitation.

Authors’ contributions

JC wrote the manuscript. ZG finished the figures. YZ, MZ. collected the data. FS, KH, GZ and JD provided supervision. All authors reviewed the manuscript.

Funding

The study was funded by Clinical Research Project of Shanghai Municipal Health Commission (Project number: 20254Y53).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.