Pancreatic cancer diagnosis by free and exosomal miRNA

Abstract

Patients with pancreatic adenocarcinoma (PaCa) have a dismal prognosis. This is in part due to late diagnosis prohibiting surgical intervention, which provides the only curative option as PaCa are mostly chemo- and radiation resistance. Hope is raised on a reliable non-invasive/minimally invasive diagnosis that is still missing. Recently two diagnostic options are discussed, serum MicroRNA (miRNA) and serum exosomes. Serum miRNA can be free or vesicle-, particularly, exosomes-enclosed. This review will provide an overview on the current state of the diagnostic trials on free serum miRNA and proceed with an introduction of exosomes that use as a diagnostic tool in serum and other body fluids has not received sufficient attention, although serum exosome miRNA in combination with protein marker expression likely will increase the diagnostic and prognostic power. By their crosstalk with host cells, which includes binding-initiated signal transduction, as well as reprogramming target cells via the transfer of proteins, mRNA and miRNA exosomes are suggested to become a most powerful therapeutics. I will discuss which hurdles have still to be taken as well as the different modalities, which can be envisaged to make therapeutic use of exosomes. PaCa are known to most intensely crosstalk with the host as apparent by desmoplasia and frequent paraneoplastic syndromes. Thus, there is hope that the therapeutic application of exosomes brings about a major breakthrough.

Key Words: Pancreatic cancer; Exosomes; MicroRNA; Diagnosis; Serum

Core tip: Patients with pancreatic adenocarcinoma have a dismal prognosis due to late diagnosis prohibiting surgical intervention, which is further burdened by chemo- and radiation resistance. Hope is raised on a minimally invasive diagnosis that is still missing. Recently two diagnostic options are discussed, serum microRNA (miRNA) and serum exosomes. Serum miRNA can be free or vesicle-, particularly, exosomes-enclosed. This review presents an overview on the current state on miRNA as a cancer diagnostics and discusses arguments in favor of tumor exosomes as a diagnostic tool that additionally could provide a powerful therapeutic option in the near future.

INTRODUCTION

Pancreatic adenocarcinoma (PaCa) ranks fourth in mortality among cancer-related deaths. With an overall 5-year survival rate of below 1% and a mean survival time of 4-6 mo it is the deadliest cancer[1,2]. There has been considerable progress in the treatment of patients with early stage PaCa. But late initial diagnosis that prohibits resection, chemotherapy and radiation resistance and the early metastatic spread of PaCa account for the non-satisfactory progress in therapy[3,4]. Thus, research has focused on defining a reliable non-invasive or minimally invasive diagnosis. So far, serum markers allowing for a non-invasive diagnosis and follow up studies are rare. CA19-9 is still the most reliable diagnostic serum marker, but should be used in conjunction with other diagnostic tools. Additional markers are carcino-embryonic antigen (CEA), CA125 and CA242, their specificity and particularly sensitivity being below that of CA19-9[5-7]. However, recently, two non-invasive diagnostic tools have come into focus. First, serum microRNA (miRNA) was repeatedly described to allow for differential diagnosis of cancer, where PaCa patients’ serum miRNA might allow differentiating between benign and malignant tumors as well as inflammation[8,9]. Second, tumor-derived exosomes are readily detected in body fluids. Their protein, mRNA and miRNA profiles might well serve as diagnostic tools[10]. In addition, exosomes are hotly debated as potent therapeutics[11-13].

TUMOR DIAGNOSIS AND miRNA

Recovery of non-coding RNA in body fluids

A new class of small noncoding RNA known as miRNA endogenously regulates gene expression at the posttranscriptional level[14]. miRNA range in size from 19 to 25 nucleotides. They regulate translation and degradation of mRNA through base pairing to complementary sites mostly in the untranslated region[15]. MiR constitute only 1%-3% of the human genome, but control about 30% of the coding genes[16], most miR controlling multiple mRNA[17]. miR biogenesis is a multistep process, where a long primary transcript (pri-miR) is processed into a 70-100 nt hairpin precursor pre-miR. The pre-miR is translocated to the cytoplasm, where it is cleaved by the ribonuclease Dicer into a mature miR duplex, which is incorporated into the RNA-induced silencing complex (RISC) resulting in degradation of the duplex and binding to target mRNA by complementary base pairing at the 3’-untranslated region[14]. Seed sequence complementarity of about 7 base pairs enables miRNA to bind the target mRNA, which results in inhibition of translation or a reduction in mRNA stability[18]. miRNA in the serum may derive from necrosis, apoptosis[19] or be actively released in microvesicles[20]. Free extracellular miRNA is associated with argonaute proteins (Ago) The Ago2-miRNA complex accounts for the stability of the free miRNA[21,22].

In advance of discussing serum miRNA as a potential diagnostic tool, it should be stated that data normalization is an important factor and that due to any fluctuation, epigenetic factors or others, like age, gender, diurnal changes and many more, cohort sizes should be large. Also due to these variabilities, it is very unlikely that a set of reference housekeeping miRNA with universal applicability can be identified[23,24]. Furthermore, it has to be kept in mind that most miRNA regulate more than one mRNA. Thus, in turn, a given miRNA may be deregulated in multiple diseases, including different types of cancer[25,26].

miRNA and cancer

The increased knowledge on miRNA greatly fostered progress in oncology, where miRNA could be linked to prognosis, disease progression, local recurrence and metastasis[24,27-29]. As summarized in a recent review[30] miRNA plays an important role in epithelial-mesenchymal transition (EMT), maintenance of cancer stem cells as well as tumor invasion and migration. EMT is regulated by the mir-200 family, miR-141, miR-429 and miR-205. The expression level of miR-200 negatively correlates with znc finger E-box-binding homeobox (ZEB)1 and 2, which inhibit E-cadherin expression[31]. In PaCa, downregulation of miR-30 correlates with EMT, targets being vimentin and snail-1[32]. Examples for the involvement of miR in cancer stem cell (CSC) control, including pancreatic cancer, are the tumor suppressor miR-34 that regulates Notch and Bcl2[33,34] and miR-21 that correlates with chemoresistance[35]. Instead, miR-9, regulating E-cadherin expression, is suggested to be of major importance for metastasis-associated mobility and invasiveness[36,37]. miR-34a overexpression can inhibit metastasis by regulating CD44[38] and miR-340 suppresses invasion and metastasis by regulating c-Met and via c-Met MMP2 and MMP9[39,40].

For PaCa Jamieson et al[41] performed microarray analysis on resected PaCa tissue on a cohort of 48 and 24 patients. They describe associations with lymph node involvement, tumor grading and overall survival, where high expression of miR-21 and low expression of miR-34a significantly correlated with poor survival. Additional studies on PaCa tissue, non-transformed pancreatic ductal cells, CP samples and on PaCa culture lines by array or RT-PCR[42-46] have been summarized by Li et al[47], which also provides an overview on their function as tumor suppressors (miR-15a, miR-34a, miR-96, miR-375) or oncogenes (miR-27a, miR-132, miR-155, miR-194, miR-200b, miR-220c, miR-429, miR-212, miR-214, miR-301a, miR-421, miR-483-3p) and potential molecular targets, which include besides others WNT3A, p53, K-Ras, Akt, 14-3-3zeta and Smad4[43,48-56].

Taken together, there is increasing evidence that miRNA plays a central role in carcinogenesis and tumor progression, where the recovery of miRNA in body fluids may, additionally, provide a minimally invasive diagnostic tool. This has created hope particularly for most deadly PaCa, late diagnosis considerably contributing to the poor prognosis.

Serum miRNA as a diagnostic tool in pancreatic cancer

The stability of free miRNA in serum and other body fluids has fostered the hope for a minimally invasive diagnostic tool that may also be of prognostic value[57-59], which meanwhile has been experimentally supported for different types of cancer[60-62] including PaCa, where it will be particularly important as late diagnosis prohibits a curative intervention.

In an earlier study 4 miRNA, miR-21, miR-210, miR-155 and miR-196a have been found to differentiate PaCa patients’ serum from that of healthy controls, where miR-155 is a biomarker of early PaCa and miR-196a correlates with progression[63]. Evaluating a combination of CA19-9 with plasma miRNA in PaCa revealed 4 miRNA, miR-155, miR-181a, miR-181b and miR-196a, to differ significantly from healthy donors’ miRNA, where only miR-16 and miR-196a allowed for discrimination from chronic pancreatitis (CP). Including CA19-9 increased sensitivity and specificity of the analysis, 85.2% of PaCa samples being positive even at stage 1[64]. An elegant recent study on serum miRNA in PaCa based on sequencing of pooled samples, a selection phase based on quantitative reverse transcriptase PCR (qRT-PCR) followed by a testing phase revealed upregulation of miR-20a, miR-21, miR-24, miR-25, miR-99a, miR-185 and miR-191 in the serum of PaCa patients as compared to healthy controls. The authors also confirmed that these 7 miRNA allowed for differentiation towards CP, where expression in CP did not differ significantly from that of healthy donors[9]. Additional studies mostly confirmed abundance of miR-21, miR-155, miR-196a, miR-210 and miR-16 in PaCa patients’ sera[65-69]. A statistical meta-analysis, which includes 9 studies, from which 5 were performed with tissue and 4 with serum or plasma[9,63,64] suggests a potential role for miRNA assays in screening for and confirming PaCa diagnosis[70]. However, the authors also point out that none of these miRNA is selective for PaCa. An additional concern should be mentioned. A differential analysis of free versus vesicular, particularly, exosomal miRNA in the serum of PaCa patients appears to be missing. An exosomal miRNA analysis may well be advantageous as exosomal miRNA derives from living cells, whereas free miRNA may mostly derive from dead cells and thus could significantly change particularly during therapy or in late stage PaCa[19,71,72]. Serum exosome screening could have an additional advantage. Membrane integrated PaCa markers will be recovered on exosomes, thus allowing for a concomitant screening of miRNA and proteins.

EXOSOMES AS A DIAGNOSTIC TOOL

Exosomes are small 40-100 nm vesicles, which derive from the fusion of the intraluminal vesicles of multivesicular bodies (MVB) with the plasma membrane[10,73]. Their homogeneous size is one of the major criteria to differentiate exosomes from apoptotic blebs, microparticles and microvesicles, which vary in size[74]. Exosomes are composed of a lipid bilayer; they contain selected proteins, mRNA and miRNA[75]. Exosomes are secreted by many cells and abundantly by tumor cells[76] and are found in all body fluids[77]. Due to their presence in all body fluids and the expression of selected markers, exosomes are suggested to be optimal candidates for non-invasive diagnosis[78,79]. Exosomal proteins, mRNA and miRNA being functionally active[80,81] and transferred into target cells[13,81-85], exosomes are the most important intercellular communicators[75] and are suggested to become a very powerful therapeutic tool[12,86,87]. To reach the goals of exosomes as diagnostics and therapeutics great efforts are taken to elaborate the prerequisites, such as exosome assembly and exosomal message transfer.

Exosome assembly and secretion

It is well known that the relative abundance of proteins, mRNA and miRNAs differs between exosomes and donor cells, which implies active sorting into MVB. Indeed, the sorting of proteins into exosomes is a highly regulated process, where monoubiquitinylation as well as the endosomal sorting complex required for transport (ESCRT) play a role, some components of ESCRT, like Tsg101 and Alix being recovered in exosomes. The ESCRT machinery consists of 3 complexes, ESCRT I, II and III, where Tsg1 in the ESCRT complex I binds ubiquitinated protein and recruits ESCRT II. ESCRT III becomes recruited via ESCRT II or Alix. ESCRT III recruits a deubiquitinating enzyme that removes the ubiquitin tag from the cargo proteins prior to sorting into MVB[88,89]. However, not all proteins require the ESCRT complex for incorporation into exosomes. Alternatively, proteins in detergent resistant membrane complexes can become incorporated into MVB like MHC II molecules in dendritic cells[90]. Lipid affinity also can account for MVB incorporation[91]. Tetraspanins and other proteins with high affinity for cholesterol and sphingolipids are partitioned into membrane domains which according to their physical properties are prone for internalization[92-95]. Proteins also may become recruited by associated proteins such as integrins associated with tetraspanins or the transferin receptor (TfR), which associates with heat shock proteins (HSP)[94]. In particular for tetraspanin-associated molecules it has been described that protein complexes rather than singular molecules are recruited into exosomes. This complex binding severely influences exosome targeting and the crosstalk with target structures[96-98]. Besides members of the tetraspanin family (CD9, CD63, CD81, CD82, CD151, Tspan8), where tetraspanins are constitutive components of exosomes[91,99] and are frequently used to differentiate exosomes from other extracellular vesicles[75,91], additional molecules most abundantly recovered in exosomes are HSP[100,101], proteases[102,103], MHC molecules, cytoskeletal proteins and signal transduction molecules[104], where engulfment of cytosolic proteins involves proteins located close to the outer membrane of MVB by automicrophagy[105].

Interest in exosomes has steeply increased, when it was reported that exosomes contain mRNA and microRNA that will be transferred into target cells[106]. Exosomal mRNA and miRNA also differs from that in the donor cell. mRNA recruitment can be guided by a zip code in the 3’-UTR[107]. Exosomal mRNA is less abundant than exosomal miRNA. Exosomal mRNA are mostly involved in cell cycle progression, angiogenesis, migration, or histone modification[98,108,109]. Exosomes also contain selected miRNA. miRNA recruitment is facilitated by coupling of RISCs (RNA-induced silencing complexes) to components of the sorting complex[110,111], the release of miRNA being controlled through ceramide-dependent machinery associated with exosome secretion[112]. Exosomes contain > 120 miRNA from a selected number of genes. Network based analysis of exosomal miRNA points towards an involvement in stem cell differentiation (let-7), organogenesis (miR-1), hematopoiesis (miR-181) tumorigenesis (miR-17, miR-18, miR-19a, miR-20, miR-19b-1, miR-93-1)[13,14] and metastasis[105].

As exosomes are found in all body fluids[77], the selective enrichment of “marker” proteins as well as of miRNA makes exosomes a very attractive means for non-invasive diagnosis[104,113].

Tumor diagnosis by serum exosomes

Exosomes are separated by sequential centrifugation steps followed or preceded by 0.2 μm filtering. For pre-evaluation exosomes should be further purified by sucrose density gradient centrifugation[114,115]. This, however may not be possible for large sample number evaluation and also may not be feasible with the amount of available serum. According to our experience and in line with literature reports, 1 mL of serum will be sufficient for screening of a limited number of proteins and miRNA. Particularly for miRNA screening, recently a thorough comparative evaluation of mRNA preparation has been published[116], which should be taken into account as in dependence of the exosome source minor differences may lead to a pronounced loss of miRNA. Besides these “home made” exosomes, several commercially available exosome purification kits are available that were described to reveal comparable results. In addition, there are special diagnostic kits on the market, which will be helpful, if a clearly defined question is to be answered, e.g., searching for one marker or a few selected miRNA. As far as one is interested to find out the protein marker or miRNA profile of exosomes of a tumor entity that has not yet already been analyzed, it may be preferable to start open minded without any preselection. In concern of the readout system, I strongly recommend for miRNA the protocol of Liu et al[9] described above for free serum miRNA, starting with a microarray of pooled serum exosomes from patients and control donors. According to our unpublished experience the ten most abundant miRNA are with high likeliness recovered in exosome pools of different patients. As the serum contains much more exosomes that are not tumor-derived, taking into account that only platelets account for roughly 50% of serum exosomes[117] , the comparison to healthy donors’ exosomes provides already a good means to select out non-tumor exosomal miRNA. As an additional control, I would recommend exosomes from culture supernatant of tumor lines from the same cancer type.

It also should be remembered that exosome collect a limited number of mRNA and miRNA that does not correlate to the mRNA or miRNA profile of the cell, which we confirmed for a rat pancreatic cancer line and exosomes derived thereof[109]. Our unpublished study on human PaCa serum exosomes confirms this inasmuch as the miRNA profile of serum exosomes and of culture supernatant exosomes show abundance of the same miRNA. In addition, the absence of a miRNA that is recovered in serum exosomes from healthy donors and PaCa patients provides a strong hint towards this miRNA being not derived from tumor exosomes. Having selected for miRNA abundant in pools of PaCa patients serum exosomes, one can proceed with verification by qRT-PCR.

In concern of serum exosome marker profiles one should also start with pooled healthy donors serum exosomes and select for markers that are undetectable on healthy donors’ exosomes. Antibodies against constitutive exosome markers may serve as controls. After this screening one can either proceed with enzyme-linked immunosorbent assay (ELISA)[118] or flow cytometry, where latex beads can be coated with antibody in advance or latex beads are loaded with exosomes and marker expression is evaluated by incubation with antibodies after blocking free binding sites of the latex beads[114,119,120]. Both procedures have advantages and disadvantages and it depends on the individual question, which to prefer. For diagnostic purposes several kits are commercially available.

So far, at least to my knowledge studies being concerned about serum diagnosis or diagnosis in other body fluids, like the urine, by miRNA have rarely taken into account the particular profile of exosomes. To give a few examples. In glioblastoma serum exosomes miR-21 was 40-fold increased[108]. In serum exosomes from ovarian cancer patients, 8 miRNA were significantly increased[121] and in prostate cancer urine exosomal miR-107 and miR-574-3p are upregulated[122].

In concern of the comparably rare trials on serum or other body fluids exosomes as diagnostic tool, I want to stress again that only exosomal miRNA is delivered by live cells[19,71,72]. Thus, this miRNA can be expected to be particularly selected for functional relevance. In addition, CSC/migrating tumor cells are suggested to be enriched in the serum[123,124] and could well contribute to the serum exosome pool and to its diagnostic validity, cancer progression relying on the small population of CSC, which account for drug resistance, metastasis and late recurrence[125-127]. Finally, exosomes being delivered by live tumor cells, the amount of exosomal miRNA may change with the size of the tumor, but the miRNA profile most likely will be stable.

Serum exosomes as a diagnostic tool have an additional advantage, as besides tumor miRNA, membrane bound tumor markers can be searched for. Thus, in ovarian cancer, CD24⁺ and EpCAM⁺ exosomes were recovered in ascites of tumor patients and in serum CD24⁺ exosomes were detected, the absence of EpCAM⁺ exosomes in serum being due to cleavage by exosomal ADAM10[128,129]. Also in ovarian cancer claudin4 was upregulated in 32 of 63 patients’ serum exosomes, but only in 1 of 50 control serum exosomes[130]. In plasma exosomes of prostate cancer patients’ survivin is upregulated compared to controls and benign prostate hyperplasia[131]. In urine exosomes of prostate cancer also PCA3 and TMPRSS2: ERG, deriving from a chromosomal rearrangement were detected, verifying body fluid exosomes as diagnostic marker[132], though in another study on prostate cancer urinary exosomes PSA and PSMA were detected, but exosomes in urine showed great variability[133]. Also in plasma exosomes from melanoma patients caveolin-1 and CD63 were consistently elevated[134] and tumor exosomes could be efficiently isolated with anti-HER2/neu from ascites of cancer patients[135]. Last, not least, the tumor-specific epidermal growth factor receptor VIII (EGFRVIII) was detected in 7 out of 25 glioblastoma patients serum exosomes[108] and our ongoing study on pancreatic cancer serum exosomes confirms recovery of exosomes carrying PaCa stem cell markers[124].

Taken together, comparably few studies on cancer patients serum/plasma or urinary exosomes confirmed the suggestion that exosomes in body fluids can serve as a diagnostic tool. Unfortunately, at least according to my stage of knowledge, PaCa serum exosomes have not yet been evaluated, where I strongly recommend to take into account that exosomes offer the possibility to evaluate both miRNA and protein markers. Our ongoing studies strongly suggest that combining the analysis of these two parameters most likely will bring about a considerable improvement in early PaCa diagnosis.

EXOSOMES AS A THERAPEUTIC TOOL

Exosomes are hotly debated as the most potent gen therapeutic option of the futur4[12]. In advance of discussing this option, I should briefly introduce what is known so far about the interaction between exosomes and target cells. I will first discuss exosome binding and uptake and proceed giving a brief overview on exosome binding and uptake-induced target modulation.

Exosome binding and uptake

In advance of considering options for the therapeutic use of exosomes, it is a conditio sine qua non to be aware, which cells in the body are potentially targeted by exosomes. Though it is well appreciated that exosomes only interact with selected targets[97,98,136], the mode of selection requires further clarification. Several options, which are mutually not exclusive are discussed, receptor-ligand interactions, attachment, fusion with the target cell membrane, or internalization[136-138].

Due to inward budding of endosomes into MVB, the outer membrane of exosomes is characterized by phosphatidylserine (PS), which can trigger exosome uptake by binding to scavenger receptors, integrins, complement receptors and PS receptors (TIM), particularly TIM-4[139,140]. In line with this, macrophages (M) very rapidly bind exosomes, binding being efficiently blocked by anti-CD11b[141]. However, in vivo studies did not provide evidence that exosome uptake is dictated by scavenger receptors. Furthermore, the selectivity of exosome uptake argues for PS facilitating binding, but not for being involved in exosome uptake[97,141,142].

Instead, already in 2004 evidence was presented that exosome uptake by dendritic cells (DC), Kupffer cells and some macrophages (M) involves, besides PS, milk fat globulin-E8, CD11a, CD54, CD9 and CD81 on exosomes and requires αvβ3, CD11a and CD54 as ligands on DC[143] suggesting exosome binding and uptake to involve receptor-ligand interactions that may vary depending on the protein pattern on exosomes and target cells[144]. Notably, this early study also pointed towards a later on confirmed contribution of tetraspanins[97,145,146]. We additionally unraveled that target cell ligands are also located in internalization prone protein clusters, which include annexins, chaperons, molecules involved in vesicular transport, tetraspanins and tetraspanin-associated molecules[97]. Thus, internalization by donor cells and the exosome uptake by target cells use similar fusion/fission machineries, maintenance of internalization complexes and re-use of these complexes for exosome uptake apparently being a common theme[146-148]. Furthermore, antibody blocking of CD91, a common receptor for several HSP interferes with exosome activity[149]. Of note, exosomes also bind with high avidity several matrix proteins[102], where matrix protein binding is selective and requires defined tetraspanin-adhesion molecule complexes[103]. Less is known about the discussed mechanism allowing for fusion of exosomes with their target cell. However, it has been shown that exosome fusion is facilitated or requires an acid pH[150].

Thus, exosomes display target cell selectivity, which at least partly builds on the engagement of protein complexes in internalization prone membrane domains.

Target modulation by exosomes

First to note, exosomal proteins, mRNA and miRNA are function competent[112,145]. Accordingly, there are several modes, whereby exosomes can modulate their targets. Binding-induced target modulation mostly relies on activation of exosome ligands and protein cleavage by exosomal proteases. Exosome uptake-initiated changes can be brought about by transferred proteins, mRNA and miRNA. These distinct activities of exosomes are far from being comprehensively understood, but all have exemplary been confirmed. I will mention some examples, as I feel it is important to be aware of this ongoing research to understand the potential power of an exosome based therapy.

Exosome-binding induced target modulation

Exosomes are rich in proteases[102], which modulate the exosomes protein profile as well the ECM and target cells.

A tumor creates its own matrix, but also influences the host matrix to generate surroundings promoting tumor cell migration and survival. The phenomenon is poorly understood and the impact of tumor exosomes is largely unexplored. First to note, exosome proteases modulate the exosome protein profile, described for L1 and CD44 shedding by ADAM10 and for EpCAM, CD46, TNFR1 by unknown metalloproteinases[151-153]. Exosomal proteases also modulate the ECM, where exosomal tetraspanins due to their association with proteases and integrins become important[152,154-156]. The collagenolytic and laminin-degrading activity of exosomes facilitates angiogenesis and metastasis[142,157-162], degradation of aggrecan increases invasiveness[163,164] and exosomal MMP2, MMP9, MMP14 and cathepsinB correlate with invasiveness[160,165]. Focalizing exosomal matrix degrading enzymes allows for paving the path of metastasizing CSC towards the premetastatic niche, which we confirmed for a rat metastasizing pancreatic adenocarcinoma[103,166]. As the ECM also is a storage of bioactive compounds[167], modulation of the ECM by exosomal proteases[168] can account for cytokine/chemokine and protease liberation and generation of cleavage products that promote motility, angiogenesis and stroma cell activation[102]. Thus, the modulation of the ECM by exosomal proteases creates a path for migrating cells, favors a tumor growth promoting microenvironment, angiogenesis and premetastatic niche establishment.

Exosome-initiated signal transduction: Exosome-initiated signal transduction can be promoted by exosome binding and exosome uptake, which in most instances is experimentally difficult to decipher. Nonetheless, the impact of tumor exosome binding-initiated signal transduction on tumor immunity, angiogenesis, tumor growth/metastasis has been convincingly demonstrated.

DC-exosomes are one of the best explored examples for exosome binding-initiated signal transduction. DC-exosomes can replace DC in immune response induction and exosome-based therapy was first explored using DC-exosomes as a cancer vaccine. DC also take up exosomes secreted by other cells, including tumor cells, which they internalize and process for presentation. Thus, DC use exosomes as a source of antigen and produce exosomes that suffice for T cell activation, both features expanding the operational range of DC[143,169-171].

Tumor exosomes also affect the immune system[172]. Tumor exosomes inhibit CD4⁺ T cell proliferation, which is accompanied by up-regulation and stronger suppressive activity of regulatory T cells (Treg) due to exosome-associated transforming growth factor beta 1 (TGF-β1)[168]. NK activity also becomes impaired via tumor exosome inhibiting activation of Stat5, Jak3, cyclinD3 expression and perforin release[173] or due to blocking NK cells via NKG2D binding[174]. Induction of myeloid-derived suppressor cell (MDSC) is promoted by exosomal TGFβ and PGE2[175]. Via stimulating TGFβ1 secretion by M, tumor exosomes suppress anti-tumor immune responses allowing for tumor growth and metastasis formation in allogeneic mice[176] and by high ICAM1 expression, tumor exosomes block the interaction between T cells and endothelial cells, thereby decreasing T cell recruitment[177]. On the other hand. high level HSP expression on tumor exosomes-HSP functioning as an endogenous danger signal-promotes NK activation and tumor cell lysis[178,179] and supports T cell activation and effector functions[180] as well as induction of costimulatory molecule expression in DC[181,182]. Tumor exosomal chemokines attract and activate DC and T cells, such that intratumoral injection efficiently inhibits tumor growth[183]. Tumor exosomes also can be an efficient antigen source, which induce a potent Th, CTL and B cell response, even where lysates of the same tumor are non-immunogenic[141,184].

Taken together, there is an intense crosstalk between tumor exosomes and the immune system that may be due predominantly to exosome binding-initiated signal transduction. Depending on the individual tumor’s exosome composition, immune responses are suppressed, but also can be strengthened and in combination with DC tumor exosomes could well contribute to cancer immunotherapy.

Angiogenesis induction being one of the hallmarks of cancer, intense efforts have been taken to elaborate the contribution of tumor exosomes. Tumor exosomes containing tumor necrosis factor alpha (TNF-α), IL1β, TGFβ and TNFR1 recruit endothelial cell (EC) progenitors, promote angiogenesis[107] and stimulate EC by paracrine signaling[185]. Delta-like4 bearing tumor exosomes confer a tip cell phenotype to EC with filipodia formation, enhanced vessel density and branching[186], which involves activation of PPARα and NFκB activation[187]. In a feedback, prostate cancer exosomes lead to activation of fibroblasts, which then shed exosomes that increase tumor cell migration via CX3C-CX3CR1[188].

Another elegant examples of tumor exosome-mediated signal transduction describes overexpression of CD9 or CD82 promoting formation and secretion of exosomes that contain β-catenin, thereby reducing its cellular content and impairing Wnt signaling, which proceeds via tetraspanin-associated E-cadherin[189]. Besides indicating that the cargo of exosomes differs depending on ESCRT- or tetraspanin-initiated internalization, this study demonstrates that by depletion of inhibitors or stimulators tumor exosomes can opposingly affect signal transduction[190]. Also, tumor exosome-promoted tumor growth may vary for individual tumors. Thus, a deficit in Rab27a leading to reduced exosome production affected growth of a tumor line that required recruitment of neutrophils, but not of another neutrophil-independent line[191].

Briefly, binding of tumor exosomes to hematopoietic cells, EC and stroma cells can severely affect the target cell, which may become activated or suppressed. Additionally, the export of proteins into tumor exosomes affects the tumor cell itself. It also has to be kept in mind that tumor exosome-initiated signaling varies with the origin and composition of tumor exosomes. Last and importantly, the strength of tumor exosome initiated signaling relies on their accessibility throughout the body.

Exosome uptake promoted target cell modulation: Early reports on the information transfer via exosomes showed that embryonic stem cell exosomes transfer messages into hematopoietic progenitor cells that promoted survival and expression of early pluripotency markers[20]. Adult tissue exosomes, too, had the capacity to alter the phenotype of their target such that upon coculture bone marrow cells (BMC) express markers found on the exosome donor cell[192], where uptake of exosome proteins, mRNA and miRNA are contributing. These findings also account for tumor exosomes, which transfer receptor and oncoproteins or miRNA[20,193].

One of the first evidences to support tumor exosome-uptake plays a critical role in autocrine stimulation of tumor growth revealed that the intercellular transfer of the oncogenic receptor EGFRVIII via tumor exosomes to glioma cells, lacking this receptor, causes transformation of indolent glioma cells[194] and reprograms growth factor pathways in EC[126]. Other oncogenes, like Ras, Myc, SV40T also induce signaling and gene expression[195-197], where e.g., exosomal amphiregulin, an EGFR ligand, increased tumor invasiveness 5-fold compared to the recombinant protein[198].

Tumor exosome uptake-induced changes in recipient non-tumor cells can be transient, but also suffice to drive tumor growth as described for tissue transglutaminase and fibronectin[199] or high level c-Met uptake by BMC, which leads to their re-education to support premetastatic niche formation for melanoma cells, where in melanoma patients, too, circulating BM-derived cells express Met[200]. Tumor exosomes also transport apoptosis inhibitory proteins[201] and present TGFβ. This drives differentiation of fibroblasts towards myofibroblasts, which support tumor growth[202]. Adipose-tissue derived mesenchymal stem cells (MSC) also can be driven into myofibroblasts by tumor exosomes[203]. Lung cancer tumor exosome uptake stimulates IL8, VEGF, LIF, oncostatin and MMP secretion, which promotes tumor growth[204]. Instead, uptake of tumor suppressor genes from non-transformed cells can mitigate cancer cell aggressiveness[12,205].

An involvement of exosomes in metastasis was first described for platelet-derived exosomes, which transferred the αIIb integrin chain to lung cancer cells, stimulated the MAPK pathway and increased expression of MT1-MMP, cyclin D2 and angiogenic factors and enhanced adhesion to fibrinogen and human umbilical vein EC[206]. We explored that exosomes from a PaCa together with a soluble tumor matrix facilitated recruitment of hematopoietic progenitors from the BM as well as activation of stroma cells and leukocytes in premetastatic lymph nodes such that a non-metastatic tumor line settled and formed metastases[207]. The recruitment of tumor cells also becomes facilitated by exosomal HSP90, a complex of exosomal HSP90 with MMP2 and tissue plasminogen activator promoting together with exosomal annexin II plasmin activation tumor cell motility[208]. As already mentioned, the transfer of c-Met contributes to premetastatic niche formation mostly via bone marrow cell modulation[200]. Thus, tumor exosomes enhance migration and homing of tumor cells in sentinel lymph nodes due to stroma and hematopoietic cell as well as matrix modulation[77,108,200,207]. Finally, uptake of exosomes from non-transformed cells in the tumor surrounding can affect tumor cells such that fibroblast-exosomes promote breast cancer motility via Wnt planar polarity signaling[209].

Tumor exosome uptake also accounts for EC modulation. Colorectal cancer exosomes, enriched in cell cycle-related mRNA, promote EC proliferation[210]. Glioblastoma-exosome-induced angiogenesis relies on the transfer of exosomal proteins and mRNA[108]. Uptake of EGFR-positive tumor exosomes by EC elicit EGFR-dependent responses including activation of the MAPK and Akt pathway and VEGFR2 expression[211]. Transfer of exosomal Notch-ligand-delta-like-4 increases angiogenesis[183] and tumor exosomes expressing a complex of Tspan8 with CD49d preferentially are taken up by EC and EC progenitors, which initiates progenitor maturation and EC activation including VEGFR transcription[60]. Chronic myeloid leukemia (CML)-exosomes induce angiogenic activity in EC, where a Src inhibitor affects exosome production as well as vascular differentiation[212].

As mentioned tumor exosome uptake-induced target cell modulation frequently represent the net result of protein transfer-initiated signal transduction, transferred mRNA translation and mRNA silencing by miRNA. Though a separation between these activities appears somewhat artificial, a few reports describing preferential activities of mRNA and miRNA should be mentioned.

By the transfer of miR-150 in AML-exosomes to hematopoietic progenitors CXCR4 expression becomes reduced and HSC migration is impaired[213]. CD105⁺ renal cell CSC exosomes carry proangiogenic mRNA and miRNA, which trigger the angiogenic switch[158]. mRNA and miRNA of exosomes from a metastasizing PaCa are recovered in lymph node stroma and lung fibroblasts, and transferred miRNA significantly affects mRNA translation, which was exemplified for abundant exosomal miR-494 and miR-542-3p, which target cadherin17. Concomitantly, MMP transcription, accompanying cadherin17 downregulation, was up-regulated in lymph node stroma cells transfected with miR-494 or miR-542-3p or co-cultured with tumor exosomes. Thus, tumor exosome miRNA uptake affected premetastatic organ stroma cells towards supporting tumor cell hosting[109]. Exosomes from virus transfected cells transfer viral miRNA[214,215]. Leukemia cell exosomes contain miR-92a that is transferred into EC, downregulates CD49e and increases migration and tube formation[216]. In lung cancer exosomes miR-21 and miR-29a act as a ligand of mouse TLR7 or human TLR8, functioning as agonist and leading to NFκB activation and IL6 and TNFα secretion, which promotes metastasis[217]. Hepatocellular carcinoma exosomes abundantly contain miR-584, miR-517c, one of the potential targets, TGFβ activated kinase 1, activates JNK and MAPK pathway and NFκB, where transfer of exosomal miRNA in coculture promoted anchorage-independent growth and apoptosis resistance[218].

Stroma cells also release exosomes, whose miRNA can influence tumor cells. BM stroma cell exosomes inhibit the growth of multiple myeloma, but those derived from patients with multiple myeloma force multiple myeloma progression, the latter exosomes showing a lower content of tumor suppressor miR-15a, but high levels of oncogenic proteins, cytokines and adhesion molecules[219]. Tumor-associated M secret exosomes with high miR-223, that binds Mef2c, causing nuclear accumulation of β-catenin[220]. Monocyte exosomal miR-150, when transferred to EC, promotes migration[221].

Taken together, transferred exosomal miRNA can reprogram target cells, the linkage between exosomal miRNA and the targeted mRNA remaining to be elaborated in detail in many instances. In concern of the described impact of transferred proteins and mRNA, the question on long-lasting in vivo efficacy awaits clarification. Exosomes being a most powerful means of intercellular communication that function across long distance, it is utmost important to answer these open questions. Nonetheless, therapeutic exploitation of exosomes appears promising.

EXOSOMES AS THERAPEUTICS

Exosomes are discussed as most potent gene delivery system, as they are easy to manipulate and efficiently transfer proteins and genes. This could offer a means to interfere with tumor exosome promoted angiogenesis and metastasis, two major targets in cancer therapy[91,222]. In addition, exosomes are discussed as cancer vaccine[172]. Nonetheless, in advance of discussing the possibilities to interfere with tumor growth and progression via exosomes, I want to stress three points. First, uptake by selective target cells needs to be most thoroughly controlled. Second, the pathway whereby exosomes affect a selected target cells has to be well defined. Besides the still open question, whether transferred proteins, mRNA and miRNA or a combination account for observed effects, the multiple targets of individual miRNA could create problems such that side effects at the present state of knowledge can not be excluded. Third, it should be mentioned that the indispensability of exosome transfer in human cancer remains questionable. In A431 PS blocking inhibits uptake of exosomes by EC, but the antiangiogenic effect was only transient[194]. Also a blockade of cellular vesiculation (TSAP6, acidic sphingomyelinase) did not prevent tumorigenesis[223,224]. Furthermore, blocking of Rab27a involved in exosome biogenesis exerts distinct effects on primary versus metastatic tumor growth and also differs between tumors[200,225]. These findings should not be taken to discourage attempts to translate experimental studies on the power of exosomes into therapeutic settings, but should foster the point that clinical translation in many instances essentially awaits progress in elaborating the mode of exosome activities. These clauses account particularly for active interference with tumor exosomes. Instead, DC exosomes are already used as a vaccine[226,227].

Exosomes to substitute or support dendritic cells

Exosome research became highly stimulated, when it was noted that antigen presenting cells release exosomes derived from MVB of the MHC class II compartment, which can stimulate T cells in vitro and in vivo[228]. Several studies report that DC-exosomes were well tolerated, induced an antigen-specific response and or NK recovery and that the disease-free survival time was mostly prolonged. For the therapeutic translation it is also beneficial that exosomes can be stored at -80 °C and that recovery is high. Limitation were mostly restricted to the requirement of large amounts of DC-exosomes[229-231].

Though tumor exosomes can be immunosuppressive, this does not affect their use for loading DC. Several groups report that exosomes delivered from DC after coculture with tumor exosomes might be superior to exosomes derived from peptide-pulsed DC. DC pulsed with exosomes of an AML line provoked a strong anti-leukemia response[232]. In line with this, directing tumor-associated, non-mutated antigens like CEA and HER2 to exosomes by coupling to lactadherin increased their immunogenicity[233]. Targeting prostate-specific antigen or prostatic acid phosphatase via lactadherin to exosomes also induced a superior immune response[234]. Furthermore, anticancer drug force the release of HSP-bearing exosomes, which efficiently activate NK cells[235]. Taking this into account, tumor exosomes should be particularly helpful as antigen source, when immunogenic entities of a tumor are unknown.

Competing with tumor exosomes

Even taking into account that an individual tumor may not essentially depend on exosomes for survival and progression, tumor exosomes doubtless support the tumor by modulating the host. Thus, competing with tumor exosomes might be a means to retard metastasis formation.

Blocking of exosome uptake could be performed at the exosome or the target cell level[229], where PS blocking of tumor exosomes only transiently inhibited angiogenesis[194]. Instead, in a rat PaCa, where exosomes expressing the tetraspanin Tspan8 induced a lethal systemic consumption coagulopathy, blocking exosomes by a Tspan8-specific antibody completely prevented undue angiogenesis, although primary tumor growth was not impaired[236,237]. Based on this finding and our ongoing studies that exosomes bind via tetraspanin-complexes to ligands also located in internalization prone membrane domains[59], we speculate that a scrutinized analysis of an individual tumors’ exosome-binding complex should provide the information for hampering undue tumor exosome-initiated angiogenesis and premetastatic niche formation, where exosomes from non-transformed cells modulated to express the tumor exosome-binding complex will be most promising[59]. As an alternative approach, tumor exosomes can be removed by affinity plasmapheresis known as Aethlon ADAPT™[238]. Blocking of tumor exosomes also can affect drug and radiation resistance due to enhanced release of export transporter MRP2, ATP7A and ATP7B or Annexin A3[239,240].

Tailored exosomes for drug delivery

Greatest hope in exosome therapy is based on the discovery of horizontal transfer of mRNA and miRNA[106,241], which can be translated or mediate RNA silencing[20,73,242].

As exosomes are natural products, are small and flexible, which allows them to cross biological membranes and to protect their cargo from degradation by a lipid bilayer[138], they are discussed as ideal and possibly the most potent gene delivery system[73,86,138,241,243]. Notably, exosome electroporation efficiently transfers siRNA into exosomes[112]. Furthermore, special devices can be developed, e.g., to cross the blood-brain barrier, which was explored for the delivery of BACE1 siRNA, where mast cell exosomes were equipped with a brain penetrating peptide fused to the vesicular membrane protein Lamp2[244,245]. Also, curcumin or Stat3 inhibitor delivery confirmed exosomes to be well suited for drug delivery[246,247], where chemotherapeutic drug efficacy was increased by lowering the pH of exosomes[150,248]. Adenoviral vectors associated with exosomes displayed higher transduction efficacy than purified AAV vectors[249]. As exosomes from non-tumor cells contain tumor-suppressive miRNA, it was suggested to use exosomes loaded with those miRNA, which was exemplified for miR-143 as a therapeutic strategy in cancer[213]. In a mouse hepatoma, systemic administration of miR-26a, inducing cell cycle arrest, exerted a dramatic protective effect without toxicity[250]. Additional approaches like miRNA inhibitors (miRNA sponges), antagomirs, locked-nucleic-acid-modified oligonucleotides are reviewed in[23].

At the present state of knowledge miRNA based therapies have to be considered as double-edged sword as most miRNA have a multitude of targets. However, as soon as the above mentioned hurdles are solved, rapid progress in clinical translation can be expected[251,252].

CONCLUSION

The recovery of tumor-associated miRNA and of tumor exosomes in serum and other body fluids has created hope for non/minimally invasive diagnostics, where our own, unpublished data indicate that an exosome-based screening may be advantageous as it offers the possibility to search concomitantly for tumor-related protein markers as well as tumor-associated miRNA. Taking into account that the poor prognosis of PaCa patients despite considerable progress in surgical treatment is mostly due to late diagnosis, a reliable serum-based diagnosis at early stages could already significantly contribute improving the rate of curative treatment.

Beyond diagnosis, the discovery of exosomes as intercellular communicators throughout the body fostered reconsideration of many aspects of tumor biology and is hoped to bring a major breakthrough in therapy. The power of exosomes is due to their ubiquitous presence, their particular protein profile and their equipment with mRNA and miRNA as well as their most efficient transfer in target cells. Together with the ease of transfecting exosomes, there should be hardly any limits in the use of exosomes as therapeutics. The therapeutic use of exosomes from non-transformed cells to compete, to induce an immune response or to silence immunosuppression should not become a danger for the patient’s organism. Instead, therapeutic approaches based on tailored tumor exosomes still awaits answers to the targeting receptors and their ligands, which most likely will offer modalities to further restrict the panel of potential targets of natural tumor exosomes and a precise knowledge on miRNA targets and consequences on release from repression. Answering these questions will take time, but is not an insurmountable hurdle.

PaCa are burdened by desmoplasia and early metastatic spread. Both features essentially depend on the crosstalk with the host, which has been convincingly demonstrated to be to a considerably degree mediated by tumor exosomes. Thus, it is my personal opinion that PaCa treatment/rognosis will particularly profit from unraveling the option of exosome-based therapy.