Suppression of Lymphocyte Functions by Plasma Exosomes Correlates with Disease Activity in Patients with Head and Neck Cancer

Head and neck cancers (HNCs) are among the most immunosuppressive human malignancies. These tumors produce a variety of inhibitory factors, which interfere with antitumor immune responses (1–3). Immune abnormalities correlate with disease activity and are more pronounced in patients with active disease (AD) than those with no evident disease (NED) following therapy. More recently, immune dysfunction in patients with HNC has also been linked to disease progression and outcome (4–6). Thus, the integrity of antitumor immune responses is critical for the patients' ability to deal with the disease and to respond to therapies.

Among numerous mechanisms shown to mediate immune suppression in patients with HNC (7–9), tumor-derived exosomes or TEX, which carry immunosuppressive factors and/or ligands, are emerging as an important negative regulator of antitumor immune responses (10, 11). TEX carrying inhibitory ligands that bind to cognate receptors on immune cells deliver tolerogenic signals to immune cells and blunt antitumor immunity (reviewed in ref. 10). Importantly, we have reported in a series of ex vivo studies with human T cells, natural killer (NK) cells and dendritic cells (DCs) that these immune cells can be protected, at least in part, from suppressive signals delivered by TEX by pretreatment with a mix of cytokines produced by PHA-activated peripheral blood lymphocytes and called IRX-2 (12–15). In aggregate, our earlier data indicated that exosomes represent a ubiquitous as well as very effective mechanism of tumor escape from the host immune system, and that this mechanism could be therapeutically controlled.

To date, most studies of vesicle-mediated immune suppression were performed with extracellular vesicles (EVs) isolated from supernatants of tumor cell lines and, less frequently, from plasma of patients with cancer (16, 17). Exosomes, the smallest of EVs (30–150 nm), are derived from the endocytic compartment of the parent cells (18), carry unique cargos, and are functionally distinct from the larger EVs (microvesicles and apoptotic bodies) (19). Their isolation from supernatants or body fluids and subsequent molecular/functional characterization require methods allowing for separation of exosomes from a heterogeneous mix of EVs. We have recently described a method combining differential centrifugation, ultrafiltration, and size exclusion chromatography that allows for an efficient, high-throughput isolation of morphologically intact, functionally active exosomes from plasma of patients with cancer (20). This method has been used to obtain exosome fractions from plasma of patients with HNC and to evaluate their effects on normal human immune cell subsets. Furthermore, our data indicate that exosomes present in the peripheral circulation of patients with HNC play a key role in immune regulation during cancer progression and response to therapy. Our data suggest that monitoring the protein content, molecular profiles, and suppressive functions of exosomes isolated from patients' plasma offers an opportunity for defining the extent and levels of immune suppression prior to and during therapy. In aggregate, we demonstrate that exosome-mediated immune suppression can be reliably measured ex vivo and may represent a clinically useful biomarker for the integrity of the immune system in patients at diagnosis and during oncologic therapies.

Peripheral venous blood specimens were collected from patients with HNC (n = 38) or healthy volunteers (n = 14) after informed consent was obtained from all individuals. The study was approved by the Institutional Review Board of the University of Pittsburgh (IRB #960279, IRB#0403105, and IRB #0506140) and was conducted in accordance with the International Ethical Guidelines for Biomedical Research Involving Human Subjects (CIOMS). The patients with HNC were seen at the UPMC Otolaryngology Clinic between years 2014 and 2016. Samples were obtained from 19 patients with active disease (AD) prior to any therapy, 15 patients with NED following oncological therapies (selected at random in respect to time since last therapy) and four patients with recurrent disease (REC). The blood samples were delivered to the laboratory and immediately centrifuged at 1,000 × g for 10 minutes to separate the plasma from blood components. Plasma was stored in 2 mL aliquots at −80°C. Peripheral blood obtained from healthy volunteers was used for the peripheral blood mononuclear cells (PBMCs) isolation on Ficoll-Paque Plus gradients (GE Healthcare Bioscience). Cells were washed in RPMI medium (Lonza) supplemented with 10% (v/v) exosome-depleted fetal bovine serum (FBS, Gibco, Thermo Scientific) and immediately used for experiments. CD8⁺ T cells, CD4⁺ T cells, and NK cells were isolated by negative selection on AutoMACS (Miltenyi) using an isolation kit for CD8⁺, CD4⁺ T cells, and NK cells (Miltenyi) as previously described (21). CD4⁺CD39⁺ Tregs were isolated as described by Schuler and colleagues (22).

Thawed plasma samples were differentially centrifuged 2,000 × g for 10 minutes at room temperature (RT) and then at 14,000 × g for 30 minutes at 4°C. Next, plasma was ultrafiltrated using a 0.22-μm filter (EMD Millipore). Self-made mini-SEC columns were prepared as previously described (20). The void volume fractions (1 mL) were collected, with exosomes eluting in fractions #3–5. The fraction #4 contained the bulk of eluted exosomes as previously reported (20). The eluted exosomes were characterized for the protein content, size, particle numbers, morphology, molecular profiles, and suppressive functions as described in the information to follow.

Protein content and concentration of the isolated exosomal fractions were determined using the Pierce BCA protein assay kit (Pierce Biotechnology) following the manufacturer's instructions. The protein concentrations are calculated as μg protein/1 mL plasma loaded onto the mini-SEC column.

Size ranges and concentrations of isolated exosome fractions were measured using TRPS as recommended by the system manufacturer Izon. Nanopores NP150 were coated with different buffers from the reagent kit supplied by Izon. Immediately before and after each experiment, calibration beads provided in the kit (200 EV, at the 1:1 ratio) were tested under the same conditions used for the samples. A small volume (10 μL) of the exosome fraction #4 was diluted 1:10 in 0.03% Tween-20 in PBS and loaded on the Nanopore. The measurement conditions for the sample were as follows: NP#37266, stretch 45.6 mm, voltage 0.68 V, current 144–150 nA, and two pressure steps 5–12 mbar. Each particle was measured by a short drop of the current (blockade). At least 500 particles and two pressure levels were recorded for both, samples and calibration beads. The Izon software version 3.2 was used for data recording and for calculating nanoparticle size ranges and concentrations.

TEM was performed as previously described (23) at the Center for Biologic Imaging at the University of Pittsburgh. Freshly isolated exosomes were layered on copper grids with 0.125% Formvar in chloroform and stained with 1% uranyl acetate in ddH₂O. Immediately, exosomes on the grids were visualized by the transmission electron microscope JEOL JEM-1011.

For analyses by Western blots, exosome fraction #4 was concentrated on Vivaspin 500 (VS0152, 300,000 MWCO, Sartorius, Göttingen, Germany). Exosomes were tested for the presence of exosomal markers, including TSG101 and of other protein markers of interest as previously described (20). Briefly, exosomes were lysed in Lane Marker Reducing Sample Buffer (Pierce, Thermo Scientific) separated on 7–15% SDS/PAGE gels (Bio-Rad) always applying 10 μg protein/lane and transferred onto an Immobilon-P PVDF membrane (EMD Millipore) for Western blotting. Membranes were incubated overnight at 4°C with antibodies purchased from various vendors as indicated and used at the following dilutions: from Abcam: TSG101 (1:500, ab30871), Fas (1:1000, ab133619), FasL (1:500, ab68338), CTLA-4 (1:5000, ab134090), TRAIL (1:500, ab2056), PD-L1 (1:400, ab58810), CD80 (1:1000, ab134120), OX40L/CD252 (1:1000, ab108083), CD70 (1:1000, ab96323); from Santa Cruz Biotechnology: CD39 (1:400, sc-33558), CD73 (1:400, sc-25603), COX-2 (1:500, sc-1745), Bcl-2 (1:500, sc-509); from R&D: PD-1 (1:500, MAB1086); from Cell Signaling: TGF-β1 (1:1000, #3711) or from Sigma Aldrich: PRAME (1:1000, SAB1401590). HRP-conjugated secondary antibody (1:3000–1:5000, Pierce, Thermo Fisher) was added for 1 hour at room temperature (RT), and blots were developed with ECL detection reagents (GE Healthcare Biosciences). The blots were semiquantitated by densitometry using software Image J. The integrated pixel value was determined for each protein band by multiplying image intensity and the band area after subtracting the mean background value.

Freshly isolated CD8⁺ T cells (10⁶/mL) were activated with CD3/CD28 T-cell activator (25 μL/mL, Stemcell,) and IL2 (150 IU/mL, PeproTech, Bionity, Rocky Hill) in freshly prepared RPMI for 48 hours. In some experiments, CD8⁺ Jurkat cells (10⁶/mL) were plated in a 96-well plate (10⁵ cells/well) in MV-depleted RPMI for 24 hours. Next, exosomes (2–3 μg in 50 μL PBS) isolated from plasma of patients with HNC with AD, REC, or NED and from plasma of NDs were added and cocultures were incubated for 24 hours. Cocultures containing no exosomes (=50 μL PBS) served as controls. Apoptosis of CD8⁺ T cells was measured by flow cytometry using an Annexin V assay (Beckman Coulter) and Accuri flow cytometer (BD Bioscience). Additionally, Caspase-3/7 activity was measured in some experiments using Caspase-3/7 Glo and Cell TiterGlo Viability assays following the manufacturer's instructions (Promega) and assessed in GloMax 96-Microplate Luminometer (Promega).

CD4⁺ T cells isolated from PBMCs of normal donors were labeled with 1.5 μmol/L CFSE (Cell Trace, Thermo Scientific) in 0.1% BSA in PBS (w/v) for 10 minutes at 37°C. The staining was quenched with an equal volume of exosome-depleted FBS (21). CFSE-labeled T cells (10⁵ cells/well) were activated with CD3/28 beads (cell to bead ratio 1:1, T-cell activation/expansion kit, Miltenyi) for 24 hours and coincubated with exosomes (2–3 μg of fraction #4 in 50 μL PBS) for 3 days at 37°C. Proliferation of T cells was measured on day 4 by flow cytometry, and the data were analyzed by Modfit (Verity Software House). The percentage suppression of proliferation in cocultures with exosomes was calculated as described by us (24) and compared with control T cells incubated alone or with exosomes isolated from plasma of NC.

Freshly isolated, normal human resting CD4⁺ T cells (10⁵/100 uL) were coincubated with exosomes (2–3 μg in 50 μL PBS) isolated from patients with AD or NED and from NDs' plasma in the presence of 20 μmol/L of exogenous ATP. Samples incubated with no exosomes (PBS only), and no ATP were used as controls. CD4-PE (Beckman Coulter) and CD39-FITC (eBioscience) expression and appropriate isotype controls were used to phenotype the T cells by flow cytometry after 20 hours of coincubation.

25,000 freshly isolated, normal human resting CD4⁺CD39⁺ in 50 μL PBS were coincubated with exosomes isolated from plasma of patients with AD, NED, and NDs under conditions described above in the presence of 20 μmol/L ATP for 1 hour (25). Concentrations of 5′-AMP and adenosine (ADO) and their degradation products (Inosine, Hypoxanthine and Xanthine) were measured by mass spectrometry as previously described (25). As controls, Treg incubated without ATP and ATP in PBS only were used.

Resting PBMCs obtained from healthy donors were coincubated with or without exosomes as described above for 24 hours. Exosome-induced downregulation of NKG2D expression levels (MFI) and percentages of NKG2D+ cells were determined by gating on CD3⁻CD56⁺ NK cells as previously described (20).

NKG2D-PE antibody and matching isotype control were purchased from Beckman Coulter.

Flow cytometry-based cytototoxicity assays (26) with K562 serving as target cells and activated NK cells treated or not with ± exosomes as effector cells were performed as described by us with minor modifications. NK cells (10⁶ cells/mL) isolated from PBMCs of NDs were activated with IL2 (1000 IU/mL, Peprotech) for 48 hours. On day 2, exosomes (2–3 μg of fraction #4 in 50 μL PBS) isolated from plasma of patients with HNC or ND were added to 150 × 10³ activated NK cells placed in wells of a 96-well plate. Aliquots of PBS (50 μL; No exosomes) were added to control wells. Following 24-hour coincubation, 30 × 10³ CFSE-labeled K562 target cells were added to the wells to obtain the NK: K562 ratio of 5:1. In preparation for the assay, K562 cells were labeled with CFSE for 10 minutes, washed twice with complete RPMI medium and reconstituted to the concentration of 10⁶ cells/mL. Plated K562 targets and effector NK cells ± exosomes were coincubated for 4 hours at 37°C in the atmosphere of 5% CO₂ in air. Next, cells were stained with 7-AAD (eBioscience). Cytotoxicity was determined by gating on CFSE-positive K562 cells and estimating the percentage of these cells that were also stained with 7-AAD. In parallel, NKG2D downregulation was determined in activated NK cells only ± exosomes.

Data were summarized by descriptive statistics (IBM SPSS, version 23) such as means and standard errors (SE). Statistical analysis was performed using GraphPad Prism (version 6). As parametric tests unpaired t test or one-way ANOVA and for nonparametric data Mann–Whitney U or Kruskal–Wallis tests were used. T-cell proliferation in CFSE-based assays was assessed with Modfit (Verity software house, version 4). Flow analyses were performed with VenturiOne version 5.0 or Kaluza v1.5 (Beckman Coulter). A P value of <0.05 was considered to be statistically significant.

The clinicopathological characteristics of the patients who contributed specimens for the study are listed in Supplementary Table S1. The patient age and gender are representative for patients with HNC, with an average age of 60 years (range, 24–79 years) and predominantly male (76% vs. 24%). Primary tumor sites were oral cavity (61%) and oropharynx (34%). Only two larynx patients (5%) were included. At blood draw, disease status was AD in 19 patients (50%), REC in four patients (11%), and NED in 15 patients (39%). The numbers of patients with the early-stage (UICC stage I or II) versus late-stage (UICC stage III or IV) disease were at the ratio of 2:3. Most patients (39%) had small tumors (T₁ or T₂) or a negative nodal status (N₀, 55%). No patient had distant metastases at the time of diagnosis. The majority of 23 patients had intermediately differentiated tumors (G₂). HPV status, routinely determined by p16 immunohistochemistry, was positive in 10 patients, negative in eight patients, and undetermined in 20 patients with nonoropharyngeal cancer. The majority (50% patients) consumed alcohol and/or tobacco at the time of diagnosis. As treatment, 19 patients (50%) underwent surgery alone, 18 patients (47%) had surgery with an adjuvant therapy (radiotherapy and/or chemoradiotherapy) and one patient was treated with primary chemoradiotherapy.

Exosome recovery from plasma (μg protein/mL of plasma in the fraction #4) was significantly greater in patients with AD than in patients with NED or NDs (Fig. 1A). Importantly, in patients with NED after therapy, protein levels in total exosomal fractions were the same as those in ND (Fig. 1A). Exosomes isolated from plasma of patients with AD with early disease (stage I/II) had significantly lower exosome protein levels than those isolated from patients with AD with advanced (stage III/IV) disease (Fig. 1B). Interestingly, somewhat higher exosomal protein levels (NSD) were observed in patients who underwent surgery with adjuvant therapy (chemotherapy alone or chemoradiotherapy) compared with patients who had surgery only (Fig. 1C).

Tunable Resistive Pulse Sensing (TRPS or qNano) measurements were performed with exosomes recovered in fractions #4 from plasma of three patients with AD and three patients with NED as well as three NDs. Size ranges and mean sizes were similar in all three groups (size range ∼40–150 nm, mean diameter ∼71–84 nm). The average particle count was highest in patients with AD (363 × 10⁹ particles/mL), intermediate in patients with NED (207 × 10⁹ particles/mL), and lowest in ND specimens (7 × 10⁹ particles/mL). These differences in particle concentration levels are significant for all cohorts and are in agreement with protein levels for the exosomal fractions in AD, NED, and NDs, as shown in Fig. 2. Thus, both the protein content and exosome numbers were highest in plasma of patients with AD.

Although the recovery of exosomes was greater in patients with HNC than in NDs, it was necessary to prioritize the assays selected for exosome characterization. Thus, not all functional assays could be performed in all patients. In most cases, the results obtained were representative of eight to 10 subjects in each patient cohort (AD vs. NED). In NDs, low levels of plasma-derived exosomes limited exosome analysis to few parameters.

The representative TEM images of exosomes isolated by mini-SEC from plasma of patients with HNC with AD or NED and from plasma of NDs are shown in Fig. 3. Freshly isolated exosomes eluting in #4 fractions appeared as intact vesicles ranging in size from 30 to 100 nm and were morphologically identical.

Western blots were performed following exosome concentration by Vivaspin 500 (MWCO 300,000), which was necessary to be able to consistently load 10 μg exosomal protein/lane on SDS gels. The composite Western blot shown in Fig. 4 presents exosome protein profiles for four patients with AD, four patients with NED, and one ND. The objective was to look for differences in the cargo of exosomes obtained from plasma of patients with HNC with AD versus NED. Among the patients, the first two vertical blot profiles are for exosomes of patients with UICC stage I/II disease, the next two are for patients with UICC stage III/IV disease. Because exosomes are known to carry immunosuppressive as well as immuno-activating proteins (27), we selected and respectively blotted for several of these proteins, as shown in the blot horizontal sections A and B. Section C shows expression of TSG101 used as an exosome marker and of PRAME used here as a marker of exosomes potentially derived from the tumor (28). Figure 4A shows that the protein profiles are different for every subject. The NED protein profiles tend to resemble those seen with the exosomes of ND. The presence of multiple immunosuppressive molecules at various levels in exosomes of patients as well as of NDs is evident. Densitometry measurements of individual protein bands for nine immunosuppressive proteins (PD-1, PD-L1, Fas, FasL, CTLA-4, TRAIL, CD73, COX₂, and TGFβ-LAP) visualized by Western blots (Fig. 4A) allowed us to compare relative densities of the bands in exosomes isolated from plasma of patients with AD versus NED (Table 1 and Supplementary Fig. S1). Densities for 6/9 of immunosuppressive proteins were higher in exosomes of patients with AD than patients with NED even though exosomes of only four patients in each cohort could be blotted for all nine proteins. Statistical analysis was not attempted, but representative densitometry data for TGF-β/LAP in Supplementary Fig. S1 indicate higher expression levels of this protein in exosomes of patients with AD than patients with NED. In Fig. 1B, we showed that exosome protein levels were higher on patients with AD with stage III/IV than stage I/II disease. Densitometry of the respective bands in Fig. 4A also suggested that this might be the case for some of the inhibitory proteins (data not shown). Figure 4B also indicates that levels of activating proteins were somewhat lower in exosomes of patients with AD with stage III/IV than stage I/II disease and tended to increase, albeit variably, in exosomes of patients with NED. Exosomal levels of PRAME did not change from patient to patient, and its presence in exosomes of ND indicates it cannot be considered as a potential tumor marker in HNC or a marker of exosome origin from tumor cells. Overall, the cargos of exosomes from plasma of patients with AD with advanced disease were enriched in proteins such as COX-2, TGFβ-LAP, PD-1, CTLA-4, and TRAIL. Their levels were variably decreased in exosomes obtained from plasma of patients with NED (Table 1).

Because Fas and FasL were often detected on Western blots of exosomes from patients with HNC, biological activity of these molecules was tested by coincubating exosomes with CD8⁺ Jurkat cells previously shown by us to carry CD95 (29). As shown in Fig. 5A, upon coincubation with exosomes from plasma of patients with HNC, CD8+Jurkat cells bind Annexin V and undergo apoptosis. Furthermore, Fig. 5B shows that levels of FasL expression on exosomes isolated from plasma of patients with AD appears to be somewhat higher than those on exosomes of with NED (NSD). Also, Annexin V binding is significantly greater (P < 0.05) in CD8+ T cells coincubated with exosomes of patients with AD than that in CD8+ T cells coincubated with exosomes of patients with NED (Fig. 5C). However, as illustrated in Supplementary Fig. S2, there was no correlation between FasL expression on exosomes and Annexin V binding to CD8+ Jurkat cells. Thus, FasL+ exosomes from patients' plasma induced apoptosis of CD8+ Jurkat cells and primary activated CD8⁺ T cells, as measured in assays based on Annexin V binding or caspase-3/7 activity (Fig. 5C and D). Further, AD and REC exosomes induced significantly more apoptosis in CD8⁺ T cells than exosomes from plasma of patients with NED or ND (Fig. 5C and D). Caspase-3/7 activity in CD8⁺ T cells was significantly elevated upon coincubation with exosomes from patients with AD compared with the control with no exosomes (P = 0.04). Importantly, caspase-3/7 activity decreased upon incubation with exosomes from plasma of the different patient cohorts and NDs: AD &REC>NED>ND (Fig. 5D), albeit given the small numbers of subjects in each group this trend was not statistically significant.

As previously shown by us, HNC plasma-derived exosomes suppress the proliferation of activated CD4⁺ T cells (20). We suspected that the ability of exosomes to mediate immune suppression was related to disease activity in plasma donors. Therefore, exosomes isolated from plasma of patients with AD or NED or ND were coincubated with normal pre-activated CD4⁺ T cells, and proliferation of T cells ± exosomes was measured in CFSE-based assays. Representative data shown in Fig. 6A indicate that exosomes of patients with AD induced the strongest suppression of proliferation (48%). Exosomes of patients with NED induced less suppression (35%) and ND exosomes were the least inhibitory (26%). In Fig. 6B, suppression levels of all experiments performed are shown, indicating that the suppression levels mediated by exosomes decrease in concert with reduced disease activity (AD vs. ND P < 0.05, AD vs. NED P < 0.05).

Previous evidence showed that exosomes derived from HNC cell line supernatants upregulated expression of CD39 ectonucleotidase on the surface of CD4+CD39+ Treg (30) and increased adenosine production by these Treg (25). We coincubated exosomes derived from HNC patients' plasma with isolated resting CD4⁺ T cells and found that only exosomes from plasma of patients with AD significantly upregulated CD39 expression levels on T cells, whereas exosomes isolated from plasma of patients with NED or NDs did not (Fig. 7A and B). As exosomes derived from plasma of patients with HNC invariably carry CD39 and CD73 (Fig. 4; ref. 25), we wished to determine whether the ectonucleotidases tethered to these exosomes retained enzymatic activity. CD4+CD39+ Treg coincubated with exosomes derived from plasma of patients with AD and NED significantly upregulated conversion of exogenous ATP to adenosine (Fig. 7C). Exosomes of patients with AD had higher CD73 activity and induced significantly more adenosine production than did exosomes from patients with NED. Exosomes from plasma of NDs induced significantly less adenosine production than exosomes from HNC patients' plasma. These data are consistent with previously observed higher levels of CD39 and CD73 carried by patients' exosomes relative to ND's exosomes (25, 30). No ATP conversion to adenosine occurred in the absence of exosomes. Interestingly, levels of 5′AMP production by CD4+CD39+ Treg alone were high and did not change upon coincubation with exosomes, suggesting that the rate of exosome-mediated conversion of ATP to 5′AMP exceeded that of 5′AMP conversion to adenosine. High levels of adenosine produced by CD4+ T cells in the presence of patients' exosomes could be responsible for inhibition of CD4 T cell proliferation as reported above.

Exosomes from plasma of patients with HNC were previously observed to induce suppression of cytotoxicity in normal human NK cells (20). To show that this exosome-mediated suppression was also related to disease activity in HNC patients, we monitored downregulation of NKG2D expression levels on the surface of normal NK cells coincubated with AD or NED exosomes. The data shown in Fig. 8A and B show the strongest and significant NKG2D downregulation in NK cells coincubated with exosomes of patients with AD and no downregulation in NK cells coincubated with exosomes of patients with NED or NDs. Also, cytotoxicity of CD3⁻CD56⁺ NK cells against K562 targets measured by flow cytometry was significantly inhibited by AD exosomes relative to no suppression seen with NED or ND exosomes (Fig. 8C).

Suppression of antitumor immune responses by tumor-derived soluble factors, including inhibitory cytokines, has long been recognized as a mechanism contributing to tumor progression (31). Although the existence of cross-talk between tumor and recipient immune cells is widely accepted, the mechanisms responsible for sustaining this communication network have remained obscure. With the recent emergence of exosomes as vehicles engaged in conveying information between tissue and hematopoietic cells (32), attention has focused on the role exosomes present in plasma of all patients with cancer might play in promoting tumor progression or modulating response to therapy. Accumulating evidence suggests that exosome fractions in plasma of patients with cancer are enlarged relative to those in plasma of healthy donors (33) and that the protein content, numbers of exosomes present as well as the molecular cargo of exosomes could serve as potential biomarkers of disease progression and even disease outcome following therapy (34). Indeed, our study of exosomes isolated from plasma of patients with HNC shows that plasma levels of exosomal proteins, exosome numbers, their molecular cargo, and their ability to alter functions of normal immune cells reflect the presence and activity of the disease as well as response to oncologic therapies. It is important to note that our studies reported over a decade ago showed that CD8⁺ T cells freshly isolated from the circulation of patients with HNC with AD were highly sensitive to spontaneous apoptosis and that the fraction of apoptosis-prone CD8⁺ effector T cells was highest in patients with AD (35–37). At the time, we interpreted these results as evidence that immune effector cells in the circulation of HNC patients with AD were induced in vivo into early apoptosis by tumor-derived “factors.” A decade later, plasma exosomes recapitulate the effects we have described (35–37) and emerge as a distinct mechanism of disease-related immune suppression.

The presence, quantity as well as molecular and functional characteristics of plasma-derived exosomes appear to correctly differentiate patients with HNC with AD from those with NED after oncologic therapy. Furthermore, among patients with AD studied prior to any therapy measurements of the protein levels in isolated exosome fractions were sufficient to discriminate between patients with stage I/II and III/IV disease. Patients with recurrent disease tended to have exosome protein levels similar to those with AD. We expected that the molecular content of plasma-derived exosomes would be a better discriminator between AD and patients with NED than exosome protein levels. To this end, we began our studies using mass spectrometry analysis of plasma-derived exosomes. Unfortunately, none of the inhibitory proteins responsible for suppression of immune cells by these exosomes were detected among hundreds of the identified proteins (data not shown). These proteins, present on exosomes at femtomolar levels, require antibody amplification for signal detection, and they are readily identifiable in Western blots. This, of course, limits the number of exosome components that can be examined and can only provide a semiquantitative analysis by densitometry measurements of individual protein bands. Nevertheless, even with these limitations, Western blot analyses showed enrichment of several immunoregulatory proteins in exosomes from plasma of patients with AD that discriminated them from patients with NED.

By far the most consistent and convincing results were obtained when we measured effects of the plasma-derived exosomes on various functions of human immune cells. In these assays, patients' exosomes were titrated into fully functional immune cells isolated from the blood of normal donors. Invariably, exosomes in patients with AD induced significantly stronger apoptosis of CD8+ T cells, greater inhibition of T-cell proliferation or of NKG2D expression on NK cells and better upregulation of suppressor functions in CD4+CD39+ Treg than exosomes of patients with NED. In aggregate, these data provide evidence that exosomes in plasma of patients with cancer could serve as reliable indicators of exosome-mediated immune dysfunction in cancer that correlates with disease activity (38). We expected that exosome-induced apoptosis, as measured by Annexin V binding in Jurkat cells, would correlate with FasL expression levels in exosomes (39). However, no correlation was observed, perhaps because FasL expression in exosomes was measured by Western blots, which detect membrane-bound and intra-vesicular proteins. In our previously reported experiments, suppressive effects of exosome-associated Fas,L TGF-β1, or CD73 on immune cells were partially but never completely blocked by anti-FasL Ab, anti-TGFβ1 Ab, or pharmacologic inhibitors of CD73, respectively (25, 29). These results suggested that not one but several different inhibitory pathways are activated in immune cells interacting with exosomes that carry multiple inhibitory ligands. Indeed, the Western blot profiles of exosomes of exosomes from HNC plasma indicate these vesicles carry at least nine different immunosuppressive molecules and deliver them to immune recipient cells as ”bundles” capable of inducing apoptosis or inhibiting other functions. For example, the presence of FasL and TRAIL in exosome cargo, as shown in Fig. 4A, suggests that efficient exosome-mediated apoptosis of T cells we measured ex vivo was likely induced by simultaneous ligation of both of these death ligands to receptors expressed on T cells. In aggregate, our data emphasize the potential of plasma-derived exosomes to serve as noninvasive indicators of disease activity before or after oncological therapies as well as indicators of immune dysfunction that these exosomes are capable of inducing in patients with cancer.

The potential of exosomes to serve as noninvasive biomarkers of cancer progression on the one hand, and as metrics of immune dysfunction in cancer on the other, is of great current interest. This study provides preliminary evidence supporting the further development of methodologies and assays for the exploration of plasma exosomes as future cancer biomarkers. This development is only slowly being realized because of limited understanding of the exosome biology and by the presence in cancer patients' plasma of numerous exosome subsets derived from various tissue cells, including tumor cells and immune cells (10, 11). Although total exosome fractions, such as those used in our experiments and those reported by others (40), may provide clinically valuable information, the future use of exosomes as a “liquid biopsy” is likely to depend on isolation from body fluids of tumor-derived exosomes as suggested by Melo and colleagues (41). Studies of supernatants of cultured human tumor cells, which contain only tumor-derived exosomes that we dubbed as TEX and no other exosomes, have clearly shown that TEX carry immunosuppressive loads and suppress functions of immune cells (16, 17, 42). However, it is possible that non-TEX could also mediate immune suppression.

The optimal use of exosomes as biomarkers of tumor progression and/or immune suppression in cancer depends on the ability to isolate exosomes from patients' plasma and to then separate TEX from non-TEX in order to obtain a tumor-specific exosome signature. This strategy presents a considerable technical barrier requiring reagents specific for tumor antigens such as glypican-1 in the Melo and colleagues report (41). Our study indicates that an alternative strategy to developing plasma exosomes as future cancer biomarkers could be based on functional evaluations of total exosome fractions ex vivo, using assays with normal immune cell subsets to discern and measure immunological abnormalities induced by exosomes. Immune competence of a patient whose tumor produces highly immunosuppressive exosomes is likely to be compromised. One such assay that could be readily standardized for potential clinical development is the flow-based apoptosis assay, where CD8+ Jurkat cells are coincubated with plasma-derived exosomes, providing a relative measure of Fas/FasL-mediated activity these exosomes can deliver. Exosome isolation from plasma using miniSEC is a high-throughput procedure (20). The development and subsequent validation of this platform for linking exosome-induced immune suppression with clinical endpoints is likely to provide useful prognostic information. The results we have presented are a “proof of principle” that could be extended, quantitated and validated in the future to harness plasma exosomes as biomarkers of cancer-related immune deficiencies during the cancer progression or as biomarkers of therapy-related immune recovery after immune therapies.

No potential conflicts of interest were disclosed.

Conception and design:S. Ludwig, E.K. Jackson, T.L. Whiteside

Development of methodology:S. Ludwig, T. Floros, E.K. Jackson, T.L. Whiteside

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.):S. Ludwig, T. Floros, M.-N. Theodoraki, C.-S. Hong, E.K. Jackson, T.L. Whiteside

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis):S. Ludwig, T. Floros, E.K. Jackson, S. Lang, T.L. Whiteside

Writing, review, and/or revision of the manuscript:S. Ludwig, T. Floros, E.K. Jackson, S. Lang, T.L. Whiteside

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases):S. Ludwig, E.K. Jackson

Study supervision:T.L. Whiteside

This work was supported in part by NIH grants R01 CA168628 and R21 CA205644 to T.L. Whiteside and also used UPCI shared resources that are supported in part by NIH award P30CA047904.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.