Novel approaches towards cancer-directed immune checkpoint inhibition

Tumor-derived extracellular vesicles (EVs) carry potent immunosuppressive factors that affect the antitumor activities of immune cells. A significant part of the immunoinhibitory activity of EVs is attributable to CD73, a GPI-anchored ecto-5 ′ - nucleotidase involved in the conversion of tumor-derived proinflammatory extracellular ATP (eATP) to immunosuppressive adenosine (ADO). The CD73-antagonist antibody oleclumab inhibits cell surface-exposed CD73 and is currently undergoing clinical testing for cancer immunotherapy. However, a strategy to selectively inhibit CD73 exposed on EVs is not available. Here, we present a novel bispecific antibody (bsAb) CD73xEpCAM designed to bind with high affinity the common EVs surface marker EpCAM and concurrently inhibit CD73. Unlike oleclumab, bsAb CD73xEpCAM potently inhibited the immunosuppressive activity of EVs from CD73 pos /EpCAM pos carcinoma cell lines and patient-derived colorectal cancer cells. Taken together, selective blockade of EVs-exposed CD73 by bsAb CD73xEpCAM may be useful as an alternate or complementary targeted approach in cancer immunotherapy.


INTRODUCTION
Exosomes are endosome-derived nanoscale (30-150 nm) extracellular vesicles (EVs) that are actively secreted by numerous cell types, including malignant tumor cells. High levels of tumor-derived EVs can be detected in bodily fluids of cancer patients 1,2 . Typically, tumor-derived EVs carry a unique repertoire of cancer-associated (cell surface) proteins, lipids, RNA (miRNAs, mRNAs, lncRNAs) and DNA that reflects the cancer cells of origin. This repertoire may include molecules that are able to mediate pro-tumoral effects including angiogenesis, proliferation, metastasis 3 . Importantly, emerging evidence indicates that EVs also expose immunoinhibitory molecules like PD-L1 which protect cancer cells from elimination by immune surveillance 4 . Recently, it was shown that a significant part of the immunoinhibitory activity of EVs is attributable to ectonucleotidase CD73 [5][6][7] .
Normally, ectonucleotidases CD39 and CD73 are involved in down-regulation of pro-inflammatory immune responses by sequentially dephosphorylating extracellular ATP (eATP) into adenosine (ADO), which is one of the most powerful immunosuppressive molecules in the human body. In this process, CD73 is rate limiting in the conversion of adenosine monophosphate (AMP) to ADO. This catalysis step results in a rapid local increase of ADO levels, both in and outside of the tumor microenvironment (TME). Subsequently, elevated ADO levels engage the immunosuppressive actions of adenosine A2A and A2B receptors on locally present immune cells, thereby providing a self-limiting and counterbalancing mechanism to timely and locally resolve the immune response 8 .
Due to high metabolic stress, cancer cells typically excrete high levels of proinflammatory eATP. However, this eATP is rapidly hydrolyzed to anti-inflammatory ADO by cancer cell surface-overexpressed CD73. These ADO molecules may diffuse from the TME, thereby forming an extratumoral immunosuppressive 'halo' that is able to chronically suppress the anticancer immune activities of a broad variety of immune effector cells 9,10 . Therapeutic blockade of the CD73/ADO immune checkpoint on cancer cells using antagonistic antibody oleclumab appears to be a promising approach to overcome immunosuppression by restoring ADO-suppressed anticancer activities of various immune effector cells 11 .
Besides cell-bound CD73, cancer cells secrete EVs that may expose high levels of CD73 in the TME [5][6][7] . Nanoscale vesicles like EVs can easily propagate to tumordraining lymph nodes and locally inhibit anticancer immune responses 12,13 . Moreover, EVs can reach the systemic circulation and subsequently contribute to a cancerfavorable microenvironment in the pre-metastatic niches. Additionally, EVs may fuse with the plasma membrane of (cancer) cells, providing these cells with CD73-based immunoinhibitory and pro-oncogenic activities 14 . Consequently, CD73-exposing EVs may prime an initially immunocompetent organ microenvironment and render it amenable for subsequent metastatic cell colonization and/or tumor progression. Therefore, blockade of CD73 on EVs may be useful as an approach to overcome ADOmediated immune suppression in various malignancies. In this respect, it is noteworthy that the capacity of oleclumab to inhibit CD73 activity exposed on EVs has not been evaluated in any detail. Moreover, thus far therapeutic approaches to selectively inhibit CD73 exposed on EVs have not been developed. In this respect, bispecific antibodies (bsAb) appear better suited to improve efficacy and safety of cancer immunotherapy as they can be engineered to more selectively direct immune checkpoint blockade to cancer cells and derived EVs [15][16][17][18] .
For the current study, we developed bsAb CD73xEpCAM with engineered tetravalent bispecific capacity to concurrently bind to CD73 and Epithelial Cell Adhesion Molecule (EpCAM), in order to promote selective blockade of CD73 exposed on CD73 pos /EpCAM pos carcinoma cells and carcinoma-derived EVs. We selected the clinically relevant pan-carcinoma target antigen EpCAM as the tumor-directing second specificity of our CD73-blocking bsAb CD73xEpCAM. EpCAM is selectively overexpressed on broad variety of human carcinomas 19 , and abundantly exposed on carcinoma-derived EVs 20,21 . Importantly, EpCAM expression in nonmalignant tissue is low and mostly limited to the basolateral surface of epithelia, which would further reduce on-target/off tumor binding upon intravenously administration of bsAb CD73xEpCAM.
Here we demonstrate that due to the concurrent binding of bsAb CD73xEpCAM to CD73 and EpCAM, it binds with higher affinity to EpCAM-expressing cells and EVs. Moreover, unlike oleclumab, bsAb CD73xEpCAM has potent capacity to selectively inhibit the enzyme activity of EVs derived from EpCAM-expressing cancer cells. Importantly, bsAb CD73xEpCAM potently inhibited CD73 exposed on EVs isolated from a CD73 pos /EpCAM pos colon carcinoma patient with advanced disease.
Additionally, bsAb CD73xEpCAM showed remarkable capacity to restore the anticancer activity of EVs-suppressed T cells. This unique mode-of-action is not available in oleclumab or any other CD73-blocking antibody. To the best of our knowledge, this is the first report on a bsAb-based approach that selectively and potently inhibits immune suppression mediated by both cancer cell-and EVs-exposed CD73. BsAb CD73xEpCAM may be useful as an alternative or complementary approach for cancer immunotherapy to overcome CD73-induced immunosuppression.

MATERIALS AND METHODS Antibodies and reagents
Fluorescently labeled secondary antibody used for flowcytometry: goat anti-human IgG APC (SouthernBiotech).

Production of recombinant bsAbs
The Expi293 expression system (ThermoFisher) was used to produce bsAbs CD73xEpCAM-IgG2s, CD73xMock-IgG2s and MockxEpCAM-IgG2s. Briefly, Expi293 cells were transfected with plasmid encoding the bsAb of choice and cultured on a shaker platform (125 rpm) at 37°C, 8% CO 2 for 7 d. Next, conditioned culture supernatant was harvested and cleared by centrifugation (4000 x g for 30 min), after which bsAbs were purified using an HiTrap Mabselect column connected to an ӒKTA Start chromatography system (GE Healthcare Life Sciences).

SDS-PAGE analysis of bsAb CD73xEpCAM
Purified bsAb CD73xEpCAM and oleclumab (5 μg) were separated by SDS-PAGE (10% acrylamide) under reducing or non-reducing conditions, followed by staining of the gel with Coomassie brilliant blue. The calculated molecular weight of bsAb CD73xEpCAM adds up to 166 kDa.

Tumor-derived EVs and HIPEC patient-derived EVs isolation
H292, OvCAR3 and DLD-1 cells were cultured as indicated above with 10% FCS until confluency of 80-90% was reached. Subsequently, cancer cells were cultured using serum free medium for 48 h after which EVs were isolated from the spent culture medium by stepwise (ultra)centrifugation.
Cancer patient-derived EVs were isolated from rinsing fluid obtained during Hyperthermia Intraperitoneal Chemotherapy (HIPEC) surgery after written informed consent (institutional approval by University Medical Center Groningen, nr. METc2012/330). After surgical tumor resection, the abdomen was rinsed with Mitomycin C for 90 min. Subsequently, the surgeon rinsed the abdomen with 1.5 L saline and 240 ml of rinsing fluid was collected directly from the patient's abdomen and promptly used for EVs isolation.
Briefly, spent culture medium or HIPEC rinsing fluid were centrifuged at 300 x g at 4°C for 15 min to remove dead cells and debris, followed by 3,000 x g centrifugation at 4°C for 25 min to remove apoptotic bodies. The supernatant was filtrated using a 0.2 μm pore filter to remove large vesicles. Next, supernatant was subjected to ultracentrifugation at 110,000 x g at 4°C for 2 h in a SW 32 Ti Rotor Swinging Bucket rotor (k factor of 204, Beckman Coulter) to sediment EVs. EVs pellets were washed with PBS and recovered by another round of ultracentrifugation.

Nanoparticle Tracking Analysis of EVs
Size distribution and concentration were analyzed by the Nanosight LM14 apparatus, equipped with a blue laser (405 nm) and a high sensitivity digital CMOS camera. Briefly, EVs were diluted 500 times in particle-free PBS and injected into the LM14 unit with a 1-mL sterile syringe. EVs were visualized by laser light scattering and Brownian motion was captured in a video of 60 s. In total 5 videos were captured for each individual sample to provide a representative concentration measurement. Videos were analyzed using NTA software 3.0 to provide nanoparticle concentrations and size distribution profiles.

Immunoblot analysis
Lysed cell samples and isolated EVs were determined for protein content using the Bradford method after which samples with equal protein content (20 μg) were subjected to SDS/PAGE. In short, samples were mixed with 4x Laemmli sample buffer (containing 355 mM 2-mercaptoethanol) and heated at 95°C for 5 min. Subsequently, the separated proteins were transferred to a 0.45 µm nitrocellulose membrane. Separate lanes of the membrane were incubated with indicated primary antibodies at 4°C for 16 h and subsequently incubated with an appropriate horseradish peroxidaseconjugated secondary antibody at room temperature for 2 h. Immunoreactive bands were visualized with SuperSignal West Pico Chemiluminescent Substrate (Thermofisher).

Assessment dual binding activity of bsAb CD73xEpCAM to cells
Dual binding activity of bsAb CD73xEpCAM for CD73 and EpCAM on cells was assessed by flow cytometry using CHO.CD73 cells and CHO.EpCAM cells and comparison to parental CHO cells. In short, cells were incubated in the presence of increasing concentrations of bsAb CD73xEpCAM, bsAb CD73xMock and bsAb MockxEpCAM (0.01-10 µg/ml) at 4°C for 45 min after after which the bsAbs were detected using an APClabeled anti-human-Ig-antibody after incubation at 4°C for 45 min. Similar, the overall binding strength of bsAb CD73xEpCAM (1 μg/ml) was accessed using EpCAM pos /CD73 pos H292 cells in the presence of sCD73, sEpCAM or a combination thereof (10 μg) at 4°C for 20 min. Subsequently, the binding was detected using an APC-labeled anti-human-Ig-antibody after incubation at 4°C for 45 min.
Residual binding of bsAb CD73xEpCAM during the various experimental conditions was assessed using the Guava easycyte flow cytometer (Merck Millipore) and analyzed with the Guava software (Guava Soft 3.2). In short, the respective CHO cells were gated based on their forward (X-axis) versus sideward (Y-axis) scatter profile, which was used to evaluate only viable cells in the absence of cell clusters or cell debris. Subsequently, binding of bsAb CD73xEpCAM or controls to the gated CHO cells (counts on Y-axis) was evaluated using standard Red-R channel settings of the flow cytometer for Mean Fluorescent Intensity (MFI on the X-axis).
At the indicated timepoints remaining capacities to bind to CD73 and inhibit CD73 enzyme activity were evaluated using H292 cells essentially as described sections 2.9 and 2.11, respectively.

Assessment capacity of bsAb CD73xEpCAM to inhibit cancer cell-expressed CD73
The enzyme activity of CD73 was assessed using a colorimetric malachite green-based Pi assay kit (ab65622, Abcam), which detects the amount of inorganic phosphate (Pi) formed during the hydrolysis of AMP to adenosine (ADO). In short, H292 cells were treated with bsAb CD73xEpCAM or indicated controls (1 µg/ml) at 37°C for 40 min. Subsequently, cells were washed (20 mM HEPES, 120 mM NaCl, 5 mM KCl, 2 mM MgCl2, 10 mM Glucose, pH 7.4) to remove phosphates and incubated with AMP (100 µM, Sigma-Aldrich) at 37°C for 40 min. The supernatant was mixed with phosphate reagent and color development was evaluated by measuring the absorbance at 650 nm using a microplate reader (VERSA max, Molecular Devices) and corrected by subtracting background levels. The % of enzyme inhibition was calculated by the following formula: ) * 100) X = is the OD value measured in a given experiment minus the background (OD650exp-OD650background) OD650max = the amount of Pi present in the conditioned supernatant in the absence of bsAb Assessment capacity of bsAb CD73xEpCAM to inhibit EVs-exposed CD73 H292 derived-EVs (10 µg) or HIPEC patient-derived EVs (10 µg

CD73 and EpCAM are abundantly present on EVs
EVs derived from cancer cell lines H292 (non-small cell lung cancer), OvCAR3 (ovarian cancer) and DLD1 (colon carcinoma) were purified by stepwise (ultra)centrifugation. Nanoparticle Tracking Analysis (NTA) indicated that the size of these EVs ranged between 118 nm and 136 nm (supplementary Figure 1). Immunoblot analysis indicated that EV markers CD9 and TSG101 were present in EVs, whereas endoplasmic reticulum protein calnexin was only present in cellular extracts, demonstrating that the isolated vesicles are indeed EVs and not contaminated with cellular protein ( Figure 1A). Compared to the originating cancer cell extracts, EpCAM and immune checkpoint CD73, but not PD-L1, were selectively enriched in EVs ( Figure 1A). As expected, CRISPR/Cas9-mediated CD73-KO and EpCAM-KO H292 cells produced EVs devoid of CD73 and EpCAM exposure, respectively ( Figure 1B and supplementary Figure 1B and 2A). Taken together, EVs are enriched in CD73 and EpCAM. Therefore, we reasoned that EpCAM is a suitable target for tumor/EVs-selective blockade of the CD73 immune checkpoint.

BsAb CD73xEpCAM has dual binding specificity for CD73 and EpCAM
BsAb CD73xEpCAM was constructed in a so-called bispecific taFv-Fc format 27 ( Figure 2B). Results demonstrate that only bsAb CD73xEpCAM bound dose-dependently to CHO.CD73 and CHO.EpCAM cells, and not to parental CHO cells ( Figure 2D, supplementary Figure 2C and 2D). Importantly, binding of bsAb CD73xEpCAM towards H292 cells was only partially reduced in the presence of excess amounts of soluble CD73 (sCD73), whereas excess amounts of soluble EpCAM (sEpCAM) strongly inhibited the binding. Of note, binding of bsAb CD73xEpCAM was only fully abrogated in the combined competing presence of sCD73 and sEpCAM ( Figure 2E), indicating bsAb CD73xEpCAM selectively and simultaneously binds to CD73 and EpCAM. Taken together, bsAb CD73xEpCAM has dual binding specificity for CD73 and EpCAM.  Figure 1A and 1B is due to a reduction of exposure time during bioluminescent-based detection to prevent overexposure of the very high EpCAM signal from H292-derived EVs in Figure 1B.

BsAb CD73xEpCAM potently inhibits the enzyme activity of CD73 on cancer cells and EVs in an EpCAM-directed manner
The dynamics of AMP hydrolysis by cancer cell-and EVs-exposed CD73 were analyzed using H292 cells and corresponding EVs, respectively. To this end, increasing amounts of H292, H292 CD73-KO cancer cells and corresponding EVs were incubated in medium supplemented with AMP. Inorganic phosphate (Pi) produced by CD73-mediated hydrolysis of the added AMP was evaluated using a colorimetric malachite green-based Pi assay. We observed that H292 cells and derived EVs displayed prominent CD73 enzyme activity, whereas H292 CD73-KO cells and corresponding EVs were devoid of this activity ( Figure 3A and 3B). Next, we compared bsAb CD73xEpCAM and oleclumab for capacity to inhibit the enzyme activity of CD73 exposed on H292 cells and corresponding EVs. The results indicated that bsAb CD73xEpCAM outperformed oleclumab both for inhibiting CD73 on cells (~72% vs. ~53%, respectively) and EVs (~75% vs. ~35%, respectively) ( Figure 3C and 3D). Additionally, the CD73-inhibitory activity of bsAb CD73xEpCAM towards H292-derived EVs was strongly reduced in the presence of excess amounts of sEpCAM ( Figure 3E) and was limited towards EpCAM-KO EVs ( Figure 3F). Taken together, bsAb CD73xEpCAM has the unique capacity to potently inhibit the enzyme activity of CD73 exposed on both cancer cells and EVs and does so in an EpCAM-directed manner.

BsAb CD73xEpCAM overcomes EVs-induced suppression of T cell proliferation
We investigated the ability of CD73 pos EVs to suppress T cell proliferation and the capacity of bsAb CD73xEpCAM and oleclumab to overcome this suppression. In short, CFSE-labeled PBMCs were incubated with H292 parental or CD73-KO EVs in medium supplemented with AMP. Next, T cells were stimulated to expand using activation beads and live cell imaging was used to quantify the size of activated T cell clusters over time ( Figure 4A). Activation beads promoted proliferation of T cells, whereas incubation with H292-derived EVs potently inhibited this proliferation. Importantly, CD73-KO EVs showed negligible capacity to inhibit the proliferative capacity of T cells. Next, we investigated whether bsAb CD73xEpCAM or oleclumab could overcome EVs-suppressed T cell proliferation. Representative light-microscopy pictures shown in Figure 4B indicated that bsAb CD73xEpCAM, but not oleclumab, fully abrogated the EVsmediated suppression of T cell proliferation as can be appreciated from a dramatic increase in size and number of T cell clusters. These results corroborated the quantification of the activated T cell clusters size in time ( Figure 4C). Incubation of PBMCs with H292-derived EVs and AMP resulted in a ~2-fold decrease in proliferation compared to activation beads only, and bsAb CD73xEpCAM fully restored the proliferative capacity of T cells ( Figure 4D). Taken together, bsAb CD73xEpCAM has potent capacity to overcome EVs-induced suppression of T cell proliferation.

BsAb CD73xEpCAM restores the anticancer activity of EVs-suppressed T cells
We investigated the ability of CD73 pos EVs to suppress the anticancer activity of T cells and the capacity of bsAb CD73xEpCAM vs. oleclumab to restore this activity. To this end, PBMCs were cultured in medium supplemented with AMP in the presence (or absence) of H292 parental or CD73-KO EVs. Subsequently, cytotoxic T cells were stimulated and re-directed to kill EpCAM-expressing PC3M cancer cells using an EpCAM-directed/CD3-agonistic bispecific antibody (BIS-1) 26 , in the presence of a conditionally fluorescent caspase 3/8 probe. Live cell imaging technology was used to assess the induction of apoptotic cancer cell death over time. Induction of cancer cell death, evident from high caspase-3/8 activation levels in target cells, dropped dramatically when cytotoxic T cells were incapacitated by subjecting them to AMP plus H292-derived EVs. Incubation with CD73-KO EVs only marginally reduced the anticancer activity of cytotoxic T cells ( Figure 5A). Importantly, bsAb CD73xEpCAM, but not oleclumab, restored the capacity of EVs-suppressed T cells to kill cancer cells ( Figure 5B). These results corroborated ELISA data quantifying the restored capacity of these cytotoxic T cells to secrete IFN-γ ( Figure 5C). Taken together, bsAb CD73xEpCAM restores the capacity of EVs-suppressed cytotoxic T cells to kill cancer cells.

BsAb CD73xEpCAM inhibits the enzyme activity of CD73 on cancer patientderived EVs
Nanoscale vesicles were purified from rinsing fluid obtained during Hyperthermia Intraperitoneal Chemotherapy (HIPEC) surgery. Vesicle size distribution ranged from 117 nm to 176 nm ( Figure 6A and supplementary Figure 4). Immunoblot analysis revealed that TSG101 was present in vesicles from 6 out of 6 cancer patients ( Figure 6B).  Additionally, 5 out of 6 cancer patient-derived EVs exposed CD73, which is in agreement with the respective OD 650nm values representing the enzyme activity of CD73 ( Figure 6B and 6C). Of note, only EVs derived from patient # 6 exposed both CD73 and EpCAM. As expected, bsAb CD73xEpCAM outperformed oleclumab for inhibiting the enzyme activity of EVs-exposed CD73 from patient # 6 (~75% vs. ~55%, respectively) ( Figure 6D). Taken together, bsAb CD73xEpCAM potently inhibits the enzyme activity of CD73 pos /EpCAM pos cancer patient-derived EVs-exposed CD73.  (20 ug

DISCUSSION
Cancer cells exploit cell surface overexpression of the inhibitory immune checkpoint molecule CD73 to evade elimination by the immune system. Moreover, many cancer types excrete large amounts of CD73-exposing EVs, which further contributes to immune escape and tumor progression [5][6][7] . Therefore, strategies that inhibit both cancer cell-exposed and EVs-exposed CD73 may be of significant importance for cancer immunotherapy. Our data demonstrate that oleclumab, a clinically-relevant CD73blocking antibody, has only limited capacity to block the enzyme activity of EVsexposed CD73 and essentially fails to reactivate the proliferative and tumor cell-killing capacity of EVs-suppressed (anticancer) T cells. Apparently, conventional monospecific CD73-blocking antibodies lack sufficient binding avidity and tumor-selectivity.
Recently, we demonstrated that tumor-directed blockade of immune checkpoints PD-L1 and CD47 can be achieved using bsAb-based approaches [15][16][17][18] . In this study, we found that bsAbs appear to be better equipped to selectively block CD73 on cancer cells and corresponding EVs. BsAbs are an emerging class of recombinant therapeutic protein drugs that can be engineered to selectively target, modulate and interconnect biologic activities of otherwise separately acting surface receptors and ligands in a pre-designed manner 28 . Moreover, tetravalent bsAbs are known to have significantly enhanced avidity towards cells that simultaneously express both targets antigens of interest, as they have up to four binding sites available for enhancement of functional interactions 29 . Thus, use of the tetravalent molecular format of bsAb CD73xEpCAM may result in an enhanced avidity towards carcinoma cells and carcinoma-derived EVs due to its capacity for multivalent interactions with the coexposed CD73 and EpCAM molecules. Although not formally studied here, it is likely that the avidity characteristic of tetravalent bsAb CD73xEpCAM is the main mechanism-of-action for its unique capacity to efficiently and selectively antagonize the immunoinhibitory activity of CD73 exposed on carcinoma-derived EVs.
In the current study, we demonstrate that bsAb CD73xEpCAM blocks CD73 in an EpCAM-directed manner, both on carcinoma cells and EVs. The tumor-selective and inhibitory activity of bsAb CD73xEpCAM towards cancer cells and derived EVs strongly outperformed that of oleclumab. This remarkable activity of bsAb CD73xEpCAM is most likely attributable to its unique tetravalent, bispecific binding capacity for surface coexposed CD73 and EpCAM.
Intriguingly, while selected carcinoma types showed relatively low levels of cell surface-exposed CD73, we detected that the corresponding EVs were strongly enriched in CD73 exposure. Moreover, based on protein content, we calculated that the AMPto-ADO converting capacity of 0.75 μg EVs may equal that of up to 400,000 of the originating cancer cells. Possibly, the locally increased density of CD73 exposure on EVs dramatically enhances its capacity to catalytically convert AMP to ADO. Alternatively, a variety of local factors, including (charged) lipid composition, pH and substrate concentration may also affect the enzyme activity of EVs-exposed CD73. Nevertheless, bsAb CD73xEpCAM showed potent dose-dependent capacity to inhibit both cancer cell-exposed and EVs-exposed CD73 up to a maximum of ~75% ( Figure  3), whereas oleclumab CD73-inhibitory activity dropped to 35% on EVs. Moreover, bsAb CD73xEpCAM showed potent capacity to restore the anticancer activity of EVssuppressed T cells, whereas oleclumab essentially failed to do so.
Obviously, other immunosuppressive factors exposed on (e.g. PD-L1) and/or carried as cargo by EVs may not be inhibitable by bsAb CD73xEpCAM 30,31 . Indeed, it was previously demonstrated that EVs may expose TGF-β with an immunoinhibitory potency that exceeds that of its soluble form 32 . Luo et al. showed that breast cancerderived EVs suppressed the proliferative capacity of T cells through EVs-exposed TGFβ, which could be partly restored by neutralizing anti-TGF-β antibodies 33 . Additionally, it is noteworthy that it was recently uncovered that EVs not only expose CD73 to autonomously produce ADO, but may also directly deliver encapsulated ADO as part of their cargo. This EVs-stored ADO represents a mechanism for delivery of cancer cellproduced ADO to distal sites that is fully independent of EVs-exposed CD73 and thus cannot be inhibited by bsAb CD73xEpCAM 34 . The above advocates that the clinical efficacy of bsAb CD73xEpCAM may be enhanced when used in a combinatorial setting.
In the current study, we detected that EVs isolated from rinsing fluid obtained during HIPEC surgery of colon carcinoma patients showed significant CD73 enzyme activity. It is tentative to speculate that ADO-mediated immunosuppression may play a role in the highly malignant features of peritoneal carcinomatosis. In this respect, it is encouraging that bsAb CD73xEpCAM outperformed oleclumab by potently inhibiting CD73 exposed on EVs derived from a CD73 pos /EpCAM pos cancer patient.
In conclusion, bsAb CD73xEpCAM has unique abilities to: (1) simultaneously bind to both CD73 and EpCAM, resulting in enhanced avidity towards CD73 pos /EpCAM pos cancer cells and corresponding EVs; (2) inhibit the enzyme activity of CD73 on cancer cells and (EpCAM pos patient-derived) EVs in an EpCAM-directed manner; (3) overcome EVs-mediated suppression of T cell proliferation; and (4) restore the anticancer activity of EVs-suppressed cytotoxic T cells.
Taken together, the use of bsAb CD73xEpCAM may represent a promising strategy for cancer-directed blockade of the CD73-ADO immune checkpoint and may be of value to cancer immunotherapy of various EVs-producing carcinomas.