Targeted inhibition of tumor-derived exosomes as a novel therapeutic option for cancer

Mounting evidence indicates that tumor-derived exosomes (TDEs) play critical roles in tumor development and progression by regulating components in the tumor microenvironment (TME) in an autocrine or paracrine manner. Moreover, due to their delivery of critical molecules that react to chemotherapy and immunotherapy, TDEs also contribute to tumor drug resistance and impede the effective response of antitumor immunotherapy, thereby leading to poor clinical outcomes. There is a pressing need for the inhibition or removal of TDEs to facilitate the treatment and prognosis of cancer patients. Here, in the present review, we systematically overviewed the current strategies for TDE inhibition and clearance, providing novel insights for future tumor interventions in translational medicine. Moreover, existing challenges and potential prospects for TDE-targeted cancer therapy are also discussed to bridge the gaps between progress and promising applications.


INTRODUCTION
Since the discovery of exosomes in 1983, a new method of cell-tocell communication was introduced to extend our perspectives of numerous physiological and pathological processes 1,2 . Exosomes are specific extracellular vesicles generated from the endosomal system instead of outward budding, typically ranging in size from 30 to 150 nm. Exosomes contain a large number of active constituents (e.g., proteins, lipids, and nuclear acids) and are considered key mediators of intercellular transportation 3,4 . Exosomes can be secreted by multiple kinds of donor cells, among which tumor cell-derived exosomes (TDEs) attract the most interest since they are involved in a series of critical functions, such as tumor growth and metastasis 5,6 .
Accelerating evidence suggests that ubiquitous TDEs in the tumor microenvironment (TME) play critical roles in tumor progression (Fig. 1). Transferring biological information locally and distantly, TDEs regulate the fate of their target cells through autocrine and paracrine pathways 7,8 . TDEs communicate with tumor cells, immune cells, cancer-associated fibroblasts (CAFs), and host vasculature in the TME and from a distance 9,10 . It has been indicated that TDEs can be taken up by tumor cells and inhibit the further release of TDEs, forming a negative feedback loop regulation. Moreover, TDEs also modulate the function of recipient tumor cells in proliferation and metastasis 11,12 . Immune cells are critical antitumor effectors in the TME 13,14 . Suppressing T-cell proliferation and inhibiting CD8 T-cell activation, TDEs contribute to immune escape 15,16 . In addition, TDEs may also induce T-cell differentiation into a suppressive regulatory T-cell (Treg) phenotype, favoring immunosuppression [17][18][19] . Producing immunoglobulins and presenting antigens, B cells play critical roles in cancer immunity. Regulatory B cells (Bregs) can be induced by TDEs, facilitating immune tolerance 20,21 . In addition, TDEs also present suppressive effects on natural killer cells 22,23 , dendritic cells [24][25][26] , and macrophages [27][28][29] in the TME, promoting tumor progression and benefiting immune escape 30 .
CAFs constitute the major cellular component in solid tumors. Distinct from normal fibroblasts, CAFs are able to secrete multiple proinflammatory factors that contribute to tumor growth and metastasis 31,32 . Tumor cells communicate with CAFs through TDEs. TDEs may reprogram normal fibroblasts into CAFs, mainly through the delivery of critical proteins and microRNAs (miRNAs) to activate the transforming growth factor-β (TGF-β) signaling pathway [33][34][35] . In addition, TDEs could also induce CAFs to transform epithelial cells to a mesenchymal phenotype, promoting tumor metastasis 36 . TDEs accelerate angiogenesis in the TME to construct a new blood vessel network for tumor progression 37,38 . Containing vascular endothelial growth factor (VEGF) and other critical modulators in angiogenesis and targeting endothelial cells (ECs), TDEs reprogram ECs and activate angiogenic signaling pathways, inducing neovascularization in the TME [39][40][41] . In addition, ligand/receptor-mediated interactions are also involved in TDE-induced angiogenesis of ECs 42 .
In addition to the critical roles of TDEs in tumor progression, TDEs also contribute to drug resistance and impede the effective response to antitumor immunotherapy 43,44 . Accelerating evidence indicates that TDEs contain a large number of nucleic acids that may transition drug-sensitive cancer cells to a resistant phenotype 45,46 . In addition, TDEs deliver P-glycoprotein (P-gp), an ATPdependent multidrug transporter, in an autocrine way to induce the extrusion of cytotoxic drugs 47 . Acquired drug resistance could also be achieved by the transmission of proto-oncogenes such as PTPRZ1-MET 48 . Checkpoint blockade immunotherapies targeting the programmed cell death-1 receptor (PD-1)/programmed cell death 1 ligand 1 (PD-L1) axis have emerged as promising treatments for cancer patients 49 . Nevertheless, the response rate is not satisfactory and partially due to the adaptive resistance mechanism mediated by TDEs 50,51 . Carrying PD-L1 inherited from their donor cells, TDEs bind to immune cells through the PD-1/PD-L1 axis, leading to the dysfunction of antitumor effectors [52][53][54] . Studies also suggested that PD-L1 on TDEs could directly bind to anti-PD-L1 antibody and may lead to immunotherapy resistance 55 .
Due to their involvement in tumor progression and therapy resistance, TDEs need to be inhibited to improve the prognosis of cancer patients. In the present review, we systematically overviewed the current strategies for TDE inhibition and clearance and suggested opportunities for tumor interventions in future translational medicine. Moreover, existing challenges, as well as potential prospects for TDE-targeting cancer therapy, are also discussed to bridge the gaps between TDE inhibition and the promising future of cancer therapy.

TDE BIOGENESIS AND SECRETION
As an important subset of exosomes, TDEs share common mechanisms of biogenesis with non-TDEs (Fig. 2) that begin with the inward membrane budding of early endosome (EE) to form multivesicular bodies (MVBs) 56 . It is well known that the endosomal sorting complex required for transport (ESCRT) machinery pathways are critically involved in the mechanism of TDEs. The ESCRT system consists of four complexes (named ESCRT-0, ESCRT-I, ESCRT-II, ESCRT-III) and accessory components 57,58 . The early ESCRTs (ESCRT-0 and ESCRT-I) are more responsible for cargo sorting. Initially, the ESCRT-0 complex binds phosphatidinositol-3phosphate (PI3P) on the EE. Then, ESCRT-I and ESCRT-II are recruited and aggregated on the endosomal membrane. With high affinities, the rigid polyvalent membrane binding structures of ESCRT-I and ESCRT-II promote early endosomal membrane deformation and inward budding to form a narrow membrane neck. ESCRT-III is subsequently recruited to cleave the neck, forming the intraluminal vesicles (ILVs) of MVB 58 . TDEs are consequently released by the fusion of MVBs with the tumor cell plasma membrane 59 . Studies have reported that most of the key regulators, for example, ESCRT-0 proteins (HRS, STAM1, STAM2) 60-62 , ESCRT-I protein TSG101 63 , ESCRT-III proteins (CHMP4A, CHMP4B, CHMP4C) 64 , and accessory protein Alix 65 , are overexpressed and/or hyperactivated in various tumor cells, contributing to the aberrant secretion of TDEs (Fig. 3). In this context, although the mechanism is shared, the biogenesis and secretion activity of TDEs are different from those of non-TDEs, owing to aberrantly expressed regulators.
Colombo et al. revealed that ESCRT-0/I (HRS, STAM1, and TSG101) knockdown critically inhibited TDE secretion by HeLa-CIITA cells 66 . In addition to abundant numbers, the function of TDEs strongly depends on the encapsulated critical protein cargos, such as PD-L1 55 . In this context, fully revealing the mechanism of cargo sorting would provide new insights for TDE inhibition. ESCRT-0 is generally considered the driver responsible for sorting ubiquitinated protein cargo into ILVs. Bache et al. revealed that HRS, STAM proteins and EPS15 formed a multivalent complex, which had multiple ubiquitin-binding domains to cluster ubiquitin cargos on the endosomal membrane 67 . Deletion of certain ESCRT subunits leads to changes in the protein composition of TDEs. In our previous study, when HRS was downregulated in malignant melanoma cells, a decrease in the level of exosomal PD-L1, as well as an increase in cellular PD-L1, was achieved, suggesting HRS as a potential target for functional cargo sorting of TDEs 52 . Further elucidation of critical TDE cargo sorting regulated by ESCRT-0 would be beneficial for functional TDE inhibition.
Baietti et al. indicated that the Syndecan-Syntenin-Alix axis was crucial for TDE biogenesis and cargo sorting manipulation in breast cancer cells 68 . Syndecan is a ubiquitinated transmembrane protein that is supplied with heparan sulfate on the membrane. Alix binds to syntenin, the cytoplasmic adaptor of syndecan, regulating the biogenesis of ILVs. Recent evidence demonstrated that Alix also recruited ESCRT-III proteins to regulate TDE biogenesis and specific cargo sorting, such as tetraspanins, independent of other ESCRT subunits 69,70 . Being able to bind both ubiquitinated and nonubiquitinated proteins, Alix may be considered a potential target for TDE inhibition. However, Monypenny et al. indicated that Alix depletion resulted in defective PD-L1 trafficking through MVBs. Loss of Alix promoted PD-L1 redistribution to the cell surface and conferred an EGFR-dependent immunosuppressive phenotype 71 . Taken together, future studies should take combinations between TDE inhibition and the related effects into consideration to achieve effective TDE-targeted cancer therapy.
It has also been found that the lipid raft microdomains distinctly segregated protein cargos on the endosome membrane 72,73 . Neutral sphingomyelinase 2 (nSMase2) is able to hydrolyze the sphingomyelin of the endosome membrane into conical ceramide, resulting in negative membrane curvature by its cone-shaped structure. Subsequently, the endosome membrane bends toward the inner cavity and sorts the lipids and protein cargos into the ILVs 73 . When nSMase2 was knocked out in the PC3 prostate cancer cell line, the secretion of TDEs was critically inhibited 53 .
Before the release of TDEs, secretory MVB is transported to the cell periphery and docked to the plasma membrane. The Rab GTPase family contributes to the underlying mechanism of this trandport 74,75 . It has been noted that the 2 isoforms of Rab27, i.e., Rab27a and Rab27b, are involved in membrane transport with distinguished roles. Specifically, Rab27a regulates the docking and fusion of MVBs with the plasma membrane, while Rab27b participates in membrane transfer to MVEs from the Golgi network (TGN) 76 . Rab27a or Rab27b knockdown reduces TDE secretion of various types of cancer cells, such as HeLa cervical cancer cells 76 and T24 bladder cancer cells 77 . Rab7 has also been found to play a critical role in TDE secretion by MCF-7 human breast cancer cells 68 . Other Rab GTPases, such as Rab11 and Rab35, participate in exosome formation of human retinal pigmented epithelial 1 (RPE1) cells 78 , also indicating a potential role in TDE biogenesis. After MVB docking, the soluble Nethylmaleimide-sensitive component attachment protein receptor (SNARE) complexes start to drive membrane fusion and subsequent TDE secretion. The SNARE complexes are formed by v-SNAREs on the secretory MVB and t-SNAREs on the plasma membrane [79][80][81] . It has been indicated that syntaxin 6 81 , YKT6 82 , and VAMP7 83 , all SNARE proteins, regulate the TDE secretion of C4-2B and CWR-R1 prostate cancer cells, A549 lung cancer cells and K562 leukemic cells, respectively.
In addition to universal regulation, the exclusive TME may also benefit the biogenesis and secretion of TDEs in multiple manners (Fig. 3). Hypoxia and low pH are the key features of the TME 84 . Under hypoxic and acidic conditions, the biogenesis and secretion of TDEs are crucially regulated, making them different from non-TDEs 85,86 . Mechanistically, hypoxia-inducible factor-1alpha (HIF-1α) promoted the expression of pyruvate kinase 2 (PKM2) 87 . PKM2 was responsible for the phosphorylation of Ser95 of synaptosomeassociated protein 23 (SNAP-23), a critical component of the SNARE complex, thereby promoting the secretion of TDEs 88 . Low pH is also considered a key microenvironmental factor that regulates the biogenesis and secretion of TDEs. It has been elucidated that acidic conditions could promote the secretion of TDEs by enhancing the function of caveolin-1 86 , which is an important initiator of exosome biogenesis through the regulation of cholesterol contents.
Due to the specific genomic mutations of tumor cells, there are distinguishing mechanisms underlying the biogenesis and secretion of TDEs compared with non-TDEs (Fig. 3). RAS (H-RAS, N-RAS, and K-RAS) is the most frequently mutated oncogene in cancers 89 . Several key differences have been revealed between mutant RAS and wild-type RAS cell line-derived exosomes. Increased secretion levels and more oncogenic proteins were found in mutant K-RAS cell-derived exosomes 90 . Mechanistically, activation of the RAS/ERK pathway, as a result of RAS mutation, was associated with the ESCRT-dependent biogenesis and secretion of exosomes. In addition, mutation of epidermal growth factor receptor (EGFR), which is considered one of the characteristics of lung cancer, also leads to the increased biogenesis and secretion of TDEs 91 . Collectively, the aberrant levels of shared regulators, the exclusive TME and the specific genomic mutations may combine to contribute to the promoted secretion and altered molecular contents of TDEs.

INHIBITION STRATEGIES FOR TDES
Based on the mechanisms of TDE biogenesis and release, emerging strategies are employed for targeted inhibition of TDEs ( Fig. 4 and Table 1). Genetic manipulation and pharmacological inhibitors are the most studied approaches.

Genetic manipulation
Effective biotechnological tools, such as RNA interference (RNAi) and the CRISPR-Cas9 system, have been widely used to reduce or completely disrupt the expression of critical genes involved in TDE biogenesis and secretion 92   cells and found that the secretion of exosomes was significantly reduced, as well as the protein contents, such as TSG101 and MT1-MMP, the critical matrix-degrading proteinase 93 . In our previous study, exosomal PD-L1 from malignant melanoma cells was found to contribute to immunosuppression. Encouragingly, when the expression of HRS was disrupted by RNAi, the level of exosomal PD-L1 decreased significantly 52 . Taken together, the above evidence suggests that HRS may be targeted for both TDE and effective cargo inhibition.
In MCF-7 human breast cancer cells, a portion of exosomes are formed through the Syndecan-Syntenin-Alix pathway. The disruption of Syndecan, Syntenin, or Alix by RNAi led to the reduced secretion of exosomes 68 . In another study, the biogenesis of syntenin-Alix exosomes was found to be regulated by the small GTPase ARF6 and its effector PLD2. Downregulation of ARF6 or PLD2 affected ILV formation and caused defects in MVB formation and subsequent exosome secretion 94 . In that case, ARF6 depletion did not implicate the specific cargo sorting of exosomes, which led to an aimless inhibition of both TDEs and non-TDEs.
Rab GTPases, which are responsible for intracellular vesicle transport, have also been used as targets for TDE inhibition.
Evidence indicated that knockdown of Rab27a or Rab27b by RNAi inhibited the secretion of exosomes from a variety of tumor cells, such as cervical cancer cells 76 , bladder cancer cells 77 , HNSCCs 93,95 , etc. The downregulation of Rab27a or Rab27b blocked the transport of MVB to the cell periphery as well as the docking to the cell membrane, resulting in TDE inhibition 76 . Notably, Poggio et al. achieved the inhibition of exosomal secretion of PC3 prostate cancer cells by knockout of nSMase2 or Rab27a based on CRISPR-Cas9, indicating a potential method of TDE inhibition 53 . However, since the deletion of nSMase2 resulted in only a partial loss of TDEs, more effective strategies or targets should be revealed in future studies. In addition, RNAi targeting Rab7 in MCF-7 breast cancer cells also inhibited the transport of MVB and led to reduced TDE secretion 68 . A recent study presented that Rab31 regulated exosome biogenesis in HeLa cells through an ESCRT-independent pathway and drove the sorting of protein cargos such as EGFR, providing a novel target for TDEs as well as cargo inhibition in future therapeutics 96 . For SNARE proteins that mediate the membrane fusion of MVB and cell membrane, Peak et al. found that downregulation of t-SNARE protein syntaxin 6 resulted in significantly reduced TDE secretion Fig. 4 Emerging strategies employed for TDE targeted inhibition. Genetic manipulation has been proven to inhibit TDEs. With RNAi and the CRISPR-Cas9 system to disrupt the genes that regulate TDE biogenesis and secretion, TDE inhibition was achieved. Pharmacological inhibitors have also been adopted to block TDEs by targeting critical regulators in TDE secretion.

Pharmacological inhibition
Over recent decades, pharmacological inhibitors have been extensively studied for suppressing TDEs, providing broad prospects in therapeutic applications. GW4869 is considered the most commonly used inhibitor of TDEs 97 . In both in vitro and in vivo cases, GW4869 inhibited the secretion of exosomes from a variety of tumor cells, including breast cancer cells 98 , bladder cancer cells 77 , epidermal cancer cells 99 , head and neck squamous cell carcinoma cells 93 , and malignant melanoma cells 100 , facilitating antitumor immunity. Yang et al. indicated that GW4869 inhibited TDEs from MDA-MB-231 breast cancer cells, resulting in a decrease in total protein content 98 . In a mouse model of breast cancer, GW4869 arrested tumor growth by inhibiting TDEs to promote antitumor effects, significantly enhancing the therapeutic effect of an anti-PD-L1 antibody. Given the hydrophobic nature of the GW4869 molecule, Wang et al. constructed a hyaluronic acid (HA)-based nanoplatform (HGF NPs) to deliver GW4869 in vivo 101 . The secretion of TDEs was dramatically decreased by HGFs assembled from GW4869 with a ferroptosis inducer (Fe 3+ ), leading to the antitumor effect of cytotoxic T lymphocytes as well as longlasting immunological memory. Although inhibiting roles on TDEs have been reported in numerous cases, the practical application of GW4869 is limited due to certain shortcomings. GW4869 is a blocker of nSMase2, which mediates the biogenesis and secretion of both TDEs and non-TDEs via an ESCRT-independent way 102 . Thus, direct application of GW4896 without tumor-targeted delivery may lead to nonspecific inhibition of both TDEs and non-TDEs. Additionally, considering that both ESCRT-dependent and ESCRT-independent machineries are involved in the regulation of TDE biogenesis and secretion 103 , application of GW4869 to inhibit the nSMase-mediated ESCRT-independent pathway may achieve only limited effects. In addition, GW4869 abrogates nSMase in a noncompetitive way 102 , which may further lead to limited efficiency of TDE inhibition. Finally, biosafety assessment of GW4869 is required before its clinical application because nSMase2 also contributes to multiple central biological processes 104 , thereby resulting in unpleasant side effects in vivo.
Based on the mechanisms that account for the biogenesis and secretion of TDEs, targeted inhibition strategies of TDEs were revealed. The genomic mutations that contribute to the aberrant biogenesis and secretion of TDEs could provide specific therapeutic targets. In this regard, Datta et al. conducted highthroughput drug screening and found a natural bacterial metabolite, manumycin-A (MA), which selectively affected RAS/ RAF/ERK1/2 by targeting farnesyltransferases (FTases), resulting in the inhibited secretion of TDEs by castration-resistant prostate cancer (CRPC) cells 105 . In addition, tipifarnib, another FTase inhibitor, was also found to inhibit TDEs of prostate cancer cells, suggesting that FTase inhibitors can function as a class of potential effectors to block TDEs 106 . In a recent study, ketoconazole was adopted as an adjunctive therapy to enhance the efficacy of sunitinib in renal cell carcinoma treatment by inhibiting TDEs through downstream ERK signaling, providing updated evidence for the use TDE inhibitors as a novel option for tumor therapy 107 . Sasabe et al. employed erlotinib, an EGFR inhibitor, to suppress the negative effects of TDEs in oral squamous cell carcinoma (OSCC) 108 . This finding suggested that anti-EGFR agents may be effective for the treatment of cancer patients with EGFR mutations by not only blocking the EGFR signaling pathway but also attenuating the unpleasant roles of TDEs.
In addition to the specific inhibition of genomic mutations or exclusive TME, a shared regulatory mechanism could also be employed. Since most of the key regulators were found to be overexpressed and/or hyperactivated in tumor cells, they may serve as potential candidates for TDE inhibition. Wu et al. indicated that heparan sulfate (HS) analogs (heparin, low molecular weight heparin, and 6-O-desulfated heparin) specifically and efficiently inhibited TDE secretion by targeting Syndecan-Syntenin-Alix, leading to weakened tumor proliferation and invasion 109 . When B16F10 melanoma cells were treated with different HS analogs, both TDE secretion and protein cargo were inhibited. By screening 1163 drugs from FDA-approved libraries, Im et al. revealed that sulfisoxazole (SFX) selectively inhibited the secretion of TDEs from breast cancer cells 110 . SFX is generally employed as an oral antibiotic that is noncytotoxic at effective doses. By suppressing the transcription of Rab GTPases (Rab5, Rab7, and Rab27a) and ESCRT components (Alix, VPS4B), SFX inhibited the formation and secretion of MVB and induced their degradation within lysosomes, ultimately leading to TDE inhibition. In breast cancer xenograft mouse models, SFX presented significant antitumor and antimetastatic effects by inhibiting TDEs. Endothelin receptor A (ETA) was identified as the downstream effector of SFX, providing a potential target for TDE inhibition in breast cancer cells. In a recent study, SFX was found to be effective in reducing the level of circulating exosomes carrying PD-L1 in CT26 tumor-bearing mice, which reinvigorated the function of CD8 cytotoxic T cells and enhanced the efficacy of anti-PD-1 immunotherapy 111 . However, the specificity is limited since non-TDEs would also be inhibited, leading us to focus on tumor-targeted delivery of the drugs for improvement. Based on the key features of the TME (e.g., hypoxia and low pH), hypoxiaand/or pH-responsive drug delivery systems may be developed for targeted drug delivery. Taking lessons from well-designed TME-responsive systems, pharmacological inhibitors may be encapsulated and specifically delivered to tumor tissues, thereby achieving selective inhibition of TDEs.

CLEARANCE STRATEGIES FOR TDES
In addition to the strategy based on genetic manipulation or pharmaceutical inhibition of TDEs in vitro and in vivo, Orme et al. pioneered the removal of TDEs from the circulation through therapeutic plasma exchange (TPE) in patients with malignant melanoma 112 . To discard the circulating exosomes, the plasma from patients was extracted by apheresis equipment and replaced by colloid solutions. With the diminished level of circulating TDEs, especially the critical cargos that modulate immunosuppression, the efficacy of immunotherapy may be improved. Dialysis is also a commonly employed treatment to remove harmful substances from the circulatory system. As widely adopted therapies for kidney diseases, hemofiltration (HF) and hemoperfusion (HP) have also been developed to treat cancer. Taking advantage of a semipermeable membrane (diameter < 1 nm) to remove poison, an appropriately sized (diameter > 200 nm) microporous membrane would be promising for the clearance of TDEs.
Although not yet applied in the clinic, affinity adsorbents have been adopted to selectively remove immunosuppressive cytokines, which may be a promising strategy for cancer treatment. It has been revealed that the cytokine network of the TME is involved in tumor progression and metastasis, leading Wang et al. further to utilize polyvinyl alcohol (PVA) microspheres coupled with heparin to remove tumor-induced cytokines 113 . The developed approach efficiently adsorbed immunosuppressive cytokines, such as VEGF and TGF-β, in the blood of tumor patients, facilitating cancer therapy. It is worth noting that PVA also presented promising biosafety, making it suitable for hemoperfusion in future translational medicine. Wu et al. also applied silica microspheres with a hemofiltration device to achieve the selective capture and removal of abundant circulating tumor cells as well as TDEs, providing potential choices for tumor therapy 114 . Recently, the rapid development of microfluid chips has paved the way for TDE filtration. Benefitting from the powerful compatibility and tiny size, microfluid chips assemble a large number of units with antibody coating for TDE elimination, which enables rapid and straightforward TDE clearance 115 .

CURRENT CHALLENGES AND FUTURE PERSPECTIVES
As outlined, approaches have been developed for the effective inhibition and clearance of TDEs. However, compared with the growing demand for scientific research and clinical application, more steps are needed (Fig. 5).
Despite its promising application in cancer research, TDE inhibition based on genetic manipulation still faces challenges in Off-target effects and biosafety risks are the leading factors that limit the application of gene manipulation for TDE inhibition. Efforts should be devoted to developing an efficient and safe delivery system as well as targeting TDE-specific effectors. For promising pharmaceutical TDE inhibition, considerations on specific TDE blocking as well as controllable side effects should be taken ahead of translational medicine. The combination of high-throughput drug screening strategies and targeted drug delivery systems would promote the development of clinically available pharmacological inhibitor-based tumor therapies. For direct TDE clearance, the risk of trauma and infection as well as nonspecific elimination hamper the clinical translation to a large extent. In this regard, more friendly and specific strategies should be developed to pave the way for future clinics. certain aspects 116,117 . First, off-target effects would result in unintended gene deletions, insertions, or mutations, leading to reduced block effect and inducing safety concerns 118 . Second, the specificity of the current inhibition strategy is limited. By blocking both TDEs and non-TDEs, potential adverse effects in tumor treatment were induced. Third, genetic modification may cause biosafety problems with virus-based systems, such as adeno-associated viral vectors or lentiviral vectors, resulting in uncertain virus-host interactions such as severe immune reactions and cancer progression [119][120][121] . To obtain eligible TDE inhibition with genetic manipulation, efforts should be devoted to developing an efficient and safe delivery system as well as targeting TDE-specific effectors [122][123][124] . To this end, the mechanism of TDE biogenesis and secretion should be well recognized, providing more potential targets for future translation. With the understanding of the detailed mechanism of TDE biogenesis and secretion, multitargeting strategies may be developed to inhibit every critical step in TDE generation, leading to a whole pathway of TDE inhibition 53,125 . In addition, although the safety concerns of viral vectors have been dispelled in most in vitro and in vivo studies, considerations should be addressed on more alternatives in tumor therapy 126 .
For TDE inhibitors, serious considerations should also be taken ahead of translational medicine. First, it is difficult to target a single molecule with an inhibitor to effectively block TDEs since complicated pathways are involved in TDE biogenesis and secretion. On the other hand, with a heterogeneous population of exosomes in circulation, including but not limited to TDEs, obstacles also exist in specific inhibition of TDEs with the current strategies 2,127 . In these cases, multitarget pharmacological inhibitors should be developed to block as many pathways as possible that are critical for TDE biogenesis and secretion. Moreover, to avoid potential side effects to non-TDEs, precise release of TDE inhibitors should also be achieved. Considering that the exclusive TME (e.g., hypoxia and low pH) benefits TDE biogenesis and secretion, reshaping the TME or targeting the related downstream signaling pathways would provide new opportunities for precise TDE inhibition. It should also be noted that the development of clinically available pharmacological inhibitors is a time-consuming work with substantial costs 128,129 . In this regard, it would be more effective to screen TDE inhibitors with a high-throughput system. The quantitative analytical methodology should also be constructed for simultaneous screening with a wide range of candidate inhibitors 105,106,130 .
For direct TDE clearance, current strategies based on extracorporeal devices present limitations, such as invasion-induced trauma, bleeding risks and potential infections, hampering clinical translation. Moreover, the removal of total exosomes by blood purification would not only eliminate TDEs but also clear up the potential positive exosomes, leading to unknown effects that may further burden the tumor patients. In this case, more friendly and specific clearance strategies should be developed in future translation 112,131 . To this end, activating the in vivo phagocytosis system mediated by macrophages instead of developing a clearance system in vitro would be a beneficial approach 132 . Evidence has indicated that phagocytotic clearance by macrophages might be altered in the TME. The elevated PD-1 expression on macrophages in the TME was negatively correlated with phagocytic potency 133 . In this case, inhibiting or blocking PD-1 in macrophages in the TME might be an effective approach for activating phagocytosis-mediated clearance of TDEs. Furthermore, Lu et al. recently demonstrated that head and neck squamous cell carcinoma (HNSCC)-derived TDEs inhibited phagocytosis of macrophages through CD73, thereby triggering immune suppression and aggressive tumor growth 134 . Therefore, future studies may also attempt to block CD73 on TDEs to enhance the clearance of TDEs by phagocytes, thus facilitating tumor therapy.

CONCLUSIONS
To conclude, TDEs play critical roles in tumor progression and mediate therapy resistance, leading to poor clinical outcomes. Taking lessons from the mechanism of TDE biogenesis and secretion, we discussed the emerging strategies for TDE inhibition and clearance, providing opportunities for future cancer therapy. To address the issues occurring in current inhibition approaches, considerations should be taken into account to achieve more specific and effective methods for genetic manipulation as well as pharmacological inhibition. Alternatively, friendly TDE clearance strategies should also be introduced. Taken together, targeted inhibition or clearance of TDEs may provide novel therapeutic options for future cancer treatment.