The lack of a specific targeting strategy against cancer stem cells in current cancer treatment regimens is at least partly responsible for life-threatening cytotoxicity for patients undergoing traditional chemotherapy. An effective cancer stem cell targeting system is urgently required for the next generation of cancer medicine. Epithelial cell adhesion molecule (EpCAM) is overexpressed in most solid cancers and it has recently been identified as a cancer stem cell marker. In this study, we isolated a 40-base RNA aptamer that binds to EpCAM from a random oligonucleotide library using systematic evolution of ligands by exponential enrichment. The aptamer was further truncated to 19 bases. This 19-nt RNA aptamer interacts specifically with a number of live human cancer cells derived from breast, colorectal, and gastric cancers that express EpCAM, but not with those not expressing EpCAM, as analyzed using flow cytometry and confocal microscopy. The binding affinity of the EpCAM RNA aptamer to human cancer cells is approximately 55 nM. Importantly, this EpCAM RNA aptamer is efficiently internalized after binding to cell surface EpCAM. To our knowledge, this is the first RNA aptamer against a cancer stem cell surface marker being developed. Such cancer stem cell aptamers will greatly facilitate the development of novel targeted nanomedicine and molecular imaging agents for cancer theranostics. (Cancer Sci 2011; 102: 991–998)
The epithelial cell adhesion molecule EpCAM (also known as CD326 or ESA) is a pleiotropic molecule, capable of both promoting and preventing epithelial cell–cell adhesion.(1) It is a 30–40 kDa type I glycosylated membrane protein expressed at a low level in a variety of human epithelial tissues. EpCAM is overexpressed in most solid cancers.(2–4) For example, intense expression of EpCAM is found in more than 98% patients with colorectal cancer.(5) Two decades of studies have shed light on the roles that EpCAM plays in tumorigenesis. Rather than antagonising apoptosis, EpCAM acts by inducing proliferation with a direct impact on cell cycle control, upregulating the proto-oncogene c-myc and cyclins A and E, and signal transduction into the cell nucleus by way of the wnt pathway.(2,3,6–8)
Recently, it has been recognized that a small proportion of cancer cells possess unlimited proliferation potential and are able to self-renew and to generate differentiated cancer cell progeny. These so-called cancer stem cells (also known as cancer initiating cells) are resistant to chemotherapy and radiotherapy.(9) It is thought that cytotoxic drugs and radiation kill mainly the bulk tumor cells but spare the cancer stem cells and thus a cure or even long-term control of macroscopic solid cancers by chemotherapy is still an exception rather than the rule.(10) Therefore, in order to eradicate cancer, one must target and eliminate cancer stem cells.
Epithelial cell adhesion molecule has been identified to be a cancer stem cell marker in a number of solid cancers, including breast cancer,(11) colorectal cancer,(4) pancreatic cancer,(12) and liver cancer.(13) It has been shown that the expression of EpCAM in various cancers is inversely related to the prognosis of the patients.(14) Anti-EpCAM antibodies can be used to identify circulating tumor cells (metastatic cells) in the blood for cancers of the breast, prostate, pancreas, stomach, and lung to provide prognostic information enabling individualized treatment of cancer.(15) However, initial clinical trials with anti-EpCAM antibodies failed to provide objective clinical response.(16–18) It is thought that the large size of the antibody is a limitation to the distribution and delivery of mAbs.(16) In addition, the antibody-dependent cytotoxicity relies on the carbohydrate composition in the CH2 domain of the antibody, which can vary significantly during antibody production. Thus, a smaller and more effective EpCAM targeting molecule is needed for targeted cancer therapy.
Aptamers, also known as chemical antibodies, are short single-stranded DNA or RNA isolated by a method known as SELEX (systematic evolution of ligands by exponential enrichment).(19,20) They fold into a 3-D structure and are capable of binding to target molecules with high affinity and selectivity.(21,22) Aptamers can be chemically synthesized with minimal batch variation and the cost of manufacturing large quantities of aptamers is lower. They are also more stable, non-immunogenic, and non-toxic. Because aptamers are 10–20 times smaller than an antibody, they have superb tissue penetration properties and are, therefore, advantageous over antibodies in cancer targeting. One aptamer, pegaptanib sodium, has already been approved by the US Food and Drug Administration(23) for clinical use, and several aptamers are in clinical trials.(24,25)
Given the varying success of current anti-EpCAM antibody therapy, we sought to develop a nuclease-resistant aptamer targeting the cancer stem cell marker that could have diagnostic and therapeutic potential. Herein, we describe the selection and characterisation of an RNA aptamer against EpCAM.
Materials and Methods
Flow cytometry assays. Cells were harvested at 80% confluence with trypsin digestion and resuspended in Dulbecco’s PBS (DPBS) supplemented with 5 mM MgCl2 and enumerated. Following centrifugation (1000g for 5 min), the pellet was resuspended in assay buffer (DPBS supplemented with 5 mM MgCl2, 0.1 mg/mL tRNA, 0.1 mg/mL salmon sperm DNA, 0.2% [w/v] sodium azide, and 5% FCS) and diluted to 1 × 106/mL. The blocking step was carried out at 4°C. The binding of the aptamers was carried out at 37°C for 30 min or 4°C for 1 h, with the final concentration of magnesium chloride being 2.5 mM in all binding assays.
To confirm aptamer binding to the target protein, RNA from iterative rounds were labeled at the 3′-ends with FITC according to a previously described method.(26) Amber tubes were used throughout to minimize photo-bleaching. Briefly, RNA was oxidized with sodium periodate. The oxidation was terminated with the addition of 10 mM ethylene glycol, followed by ethanol precipitation. Then FITC was added at a 30-molar excess, and the reaction was completed overnight at 4°C. Fluorescein isothiocyanate-labeled RNA (1 μM) was incubated with trypsinized 5 × 105 Kato III or U118-MG cells in 100 μL binding buffer (DPBS containing 5 mM MgCl2, 0.1 mg/mL tRNA, and 0.1 mg/mL salmon sperm DNA) for 1 h on ice, followed by washing three times and resuspension in 300 μL assay buffer. Fluorescent intensity was determined with a FACS Canto II flow cytometer (Becton Dickinson, NSW, Australia) by counting 50 000 events each sample. The FITC-labeled RNA from the unselected library and an EpCAM-negative cell line were used to determine non-specific binding.
The binding for each round was calculated after subtracting the mean fluorescence intensity of the binding of round zero RNA to target cells as well as that for binding to negative control cells according to a method described by Ellington and colleagues.(27) The dead cells were gated and excluded from the analyses by staining with 2.5 μg/mL propidium iodide and 0.5 mg/mL RNase A in PBS.
Confocal microscopy. Twenty-four hours prior to labeling, cells were seeded at a density of 75 000 cells per cm2 in an 8-chamber slide (Lab-Tek I; Nunc, Victoria, Australia). DY647-EpDT3 and the control aptamer were prepared in the same manner as for flow cytometry. Following removal of media, cells were incubated in assay buffer at 37°C for 15 min, and washed twice prior to incubation with 100 nM aptamer with 2.5 MgCl2 for 30 min at 37°C. Bisbenzimide Hoechst 33342 (3 μg/mL; Sigma, NSW, Australia) was added to the cells during the final 15 min of incubation. The aptamer solution was removed and the cells washed for 5 min each in binding buffer three times prior to visualization using a FluoView FV10i laser scanning confocal microscope (Olympus, NSW, Australia).
Inhibition of endocytosis. This was carried out essentially as described for confocal microscopy with minor modifications. Briefly, cells were pre-treated with either a potassium-depleted (50 mM HEPES, 140 mM NaCl, 2.5 mM MgCl2, and 1 mM CaCl2) or a hypertonic buffer (potassium-depleted buffer containing 3 mM KCl and 450 mM sucrose) for 1 h at 37°C. These buffers were also used in the incubation step with aptamers and all rinsing steps. The effectiveness of these treatments in inhibiting endocytosis was evaluated by qualitatively characterizing the internalization of human transferrin conjugated to Alexa Fluor 488 (Invitrogen, Victoria, Australia). Transferrin (5 μg/mL) was incubated with the cells for 30 min at 37°C following pretreatment. The cells were washed three times in their respective buffers and visualized using the FluoView FV10i confocal microscope.
Robust SELEX facilitates selection of EpCAM aptamers. To select RNA aptamers that recognize cell surface EpCAM of native conformation, we devised a modified SELEX procedure (see Data S1). In our SELEX procedure, the cytoplasmic domain (the C-terminal of EpCAM) was attached to the solid support by the His-tag epitope leaving the extracellular domain of the transmembrane protein (the N-terminal of EpCAM) freely exposed to the selection solution to facilitate the selection of aptamers that bind to the extracellular domain of the target protein. A random RNA library of approximately 1 × 1014 species containing 2′-fluoro-modified ribose on all pyrimidines was used to incubate with immobilized recombinant EpCAM. After washing to remove unbound RNA, the bound RNA was amplified by RT-PCR in situ. In total, 12 iterative rounds of SELEX were carried out and enrichment was monitored through the use of both non-radioactive RFLP (data not shown) and flow cytometry of live EpCAM-expressing cancer cells. The RFLP facilitates the confirmation of the progression of the DNA in SELEX rounds. The ability of selected RNA aptamers to interact specifically to EpCAM-positive human cancer cells but not to other sites on a human cell surface non-specifically was assessed using live cells by flow cytometry. As shown in Figure 1, RNA aptamers from SELEX round 7 displayed 1.2-fold higher binding to EpCAM-positive human gastric cancer cells (Kato III) than to EpCAM-negative human glioblastoma cells (U118MG) and that of the unselected library.
Generation of a small EpCAM aptamer by post-SELEX engineering. Following successful enrichment of aptamers to EpCAM, individual aptamers were cloned and sequenced. A total of 60 clones were sequenced from rounds 7 and 11, and sequence homology was determined using ClustalX2.(28) The binding affinity (apparent dissociation constant, K’d) of six distinct families was analyzed using FITC-labeled RNA using EpCAM-positive human cancer cells as well as EpCAM-negative human cell lines (data not shown). The EpCAM aptamer with the highest affinity, Clone D (5′-GGG ACA CAA UGG ACG UCC GUA GUU CUG GCU GAC UGG UUA CCC GGU CGU ACA GCU CUA ACG GCC GAC AUG AGA G-3′), is 73 nt long and was found to have a K’d of 211 ± 36.2 nM for human gastric carcinoma cells Kato III. For in vivo cancer molecular imaging and targeted therapy, an aptamer of small size is desirable. Thus, we carried out serial truncation of the original EpCAM aptamer Clone D, based on the assumption that the end loop on the left of the 2-D structure of Clone D (Fig. 2A) was responsible for the binding of the aptamer to EpCAM. Two rounds of truncation shortened Clone D first to 43 nt, then further down to 19 nt (Fig. 2B,C). The K’d for the shortest 3′-FITC-labeled EpCAM aptamer, EpDT3 (5′-GCGACUGGUUACCCGGUCG-3′), was 12 ± 6.5 nM, when analyzed using Kato III cells (Fig. 2D–F).
All of the RNA aptamers described here were produced by in vitro transcription. In order to study the utility of our aptamer in targeted cancer nanomedicine in vivo, we switched to the approach of total chemical synthesis of RNA aptamers. We synthesized RNA aptamer EpDT3 containing 2-fluoropyrimidines and a 3′-inverted deoxythymidine cap, leading to a 3′–3′ linkage that inhibits degradation by 3′ exonucleases. This aptamer was also labeled with a different fluorescent tag on the 5′-end of the RNA. The DY647 was chosen for future downstream applications, as it has a far-red emission spectra, allowing us to study its biodistribution profile in animal studies. Interestingly, the apparent K’d was increased, from approximately 12.0 nM to 54.5 nM, with this Dy647-labeled EpDT3. This could be due to several factors, including the different size of the fluorophore (389 Da for FITC versus 1008 Da for Dy647) and the presence of an inverted dT rather than FITC at the 3′-end.
Ep-DT3-DY647 binds to EpCAM-positive human cancer cells highly specifically. Kato III cells, previously shown by other researchers to possess a high number of EpCAM molecules on their cellular surface,(30) was initially used to determine the affinity of our aptamer to EpCAM. To test whether the EpCAM aptamer is able to bind EpCAM on the surface of human cancer cells of different histopathological origins, we studied the interaction of DY647-EpDT3 with a panel of EpCAM-positive human cancer cell lines using flow cytometry. Apart from Kato III, we used a breast adenocarcinoma-derived cell line MCF-7, a colon adenocarcinoma-derived cell line SW480, a ductal breast epithelial tumor-derived cell line T47D, a colon adenocarcinoma grade II-derived cell line HT-29, and a breast adenocarcinoma-derived cell line MDA-MB-231. We confirmed the presence or absence of EpCAM protein in all cell lines studied using anti-EpCAM antibodies and flow cytometry (Fig. S1). As shown in Figure 3(A–F), DY647-EpDT3 was able to bind to six different types of human cancer cells well, with a K’d of approximately 60 nM. To confirm that the interaction, observed using flow cytometry, is specific between EpDT3 and the human cancer cells tested, we carried out the binding assay using a control DY647-labeled aptamer that has the same nucleotide sequence as EpDT3 but with a 2′-O-methyl (2′-OMe) modification instead of a 2′-fluoro (2′-F) in all the pyrimidines. This control aptamer has the same secondary structure as EpDt3 as modeled using RNA secondary structure prediction software.(29) As indicated in Figure 3(G–L), the DY647-control aptamer did not bind the six EpCAM-positive cancer cell lines examined. Thus, the binding of DY647-EpDT3 to cell surface EpCAM is attributed to the unique 3-D structure of the aptamer. In order for this novel EpCAM aptamer to become an effective cancer targeting agent, it must have negligible interaction with cells that do not express EpCAM. To this end, we used a number of different types of human cells, both non-transformed and cancerous, to study the specificity and selectivity of the EpCAM aptamer. As shown in Figure 3(M–O), DY647-EpDT3 did not bind to EpCAM-negative human cells, such as human embryonic kidney 293T cell line HEK293T, neuroblastoma-derived cell line SK-N-DZ, or erythroleukemia-derived cell line K562. The specific interaction between EpDT3 RNA aptamer and EpCAM-positive cells, but not with EpCAM-negative cell lines, was further verified using flow cytometry analysis (Fig. 3P,Q) and confocal microscopy (Fig. 4A–C).
EpDT3-DY647 internalized through receptor-mediated endocytosis. For cancer therapy and molecular imaging, the targeting moiety and/or its conjugated nanoparticle should ideally be internalized after binding to the cancer cells. We investigated the fate of our aptamer after binding with cancer cells using confocal microscopy. We asked whether EpDT3-DY647 stays on the cell surface, or is internalized after binding. After incubating the EpDT3 aptamer with cancer cells, we observed a particular intracellular pattern of red fluorescence (for EpDT3-DY647) (Fig. 4A, upper panels, C), indicative of an entry through endocytosis. This cell internalization was specific for EpDT3-DY647 as the control 2′-OMe aptamer did not bind to the other six cancer cell lines (Fig. 4A, lower panels). To confirm this, we pre-treated cells with agents known to block endocytosis, that is, potassium depletion and hypertonic treatment.(31–33) The effectiveness of such treatments in blocking endocytosis was first confirmed by their ability to block the internalization of transferrin in the relevant cell lines (Fig. S2). Indeed, after treatment with endocytosis blockers, a ring pattern of red fluorescence for EpDT3-DY647 was observed in the four different human cancer cell lines tested, including human gastric, colorectal, and breast cancers (Fig. 4D). Thus, after binding with cell surface EpCAM, our aptamer is internalized by human cancer cells, possibly through clathrin-dependent endocytosis.
Novel approaches for early detection and effective treatment of cancer will significantly improve clinical outcomes of cancer treatment. Epithelial cell adhesion molecule is an attractive target for novel cancer-targeting therapy because it is overexpressed in most human adenocarcinomas and it is a marker for cancer stem cells in several solid cancers.(2) However, the results from clinical trials with at least five different mAbs to EpCAM as tumor immunotherapy have been disappointing.(24,34–36) The lack of success of EpCAM antibody monotherapy has been attributed, at least in part, to the size of the immunoglobulin, the affinity of the antibody, and its high immunogenicity.(16,18,37) Therefore, we attempted to isolate small RNA aptamers against EpCAM.
The robustness of our aptamer selection regime was shown through the use of flow cytometric binding assays assisted by RFLP analysis, with successful evolution of our randomized library that allowed us to clone the aptamers at the seventh cycle of SELEX. The K’d of all three versions of our aptamer (55–211 nM) are in good agreement with previously reported aptamers against other targets, such as the RNA aptamers to tenascin-C (5 nM), Trypanosoma cruzi (172 nM), and prostate-specific membrane antigen aptamer (2–11 nM).(38)
One of the interesting features of our aptamer is that it is able to bind to human cancer cells expressing a high level as well as a low–medium level of EpCAM. For example, DY647-EpDT3 binds to both gastric carcinoma Kato III cells, which have 893 100 EpCAM protein molecules per cell, and breast carcinoma MDA-MB-231 cells, which have 1700 EpCAM protein molecules per cell(30) (Fig. 4). In other words, the protein level of EpCAM in MDA-MB-231 cells resembles a 99.8% knockdown of EpCAM in Kato III by siRNA. Our aptamer was shown to have a slightly different aptamer dissociation constant (K’d) towards different human cancer cell lines when analyzed under physiological temperature (37°C) and magnesium concentration (2.5 mM) (Fig. 3). These results appear to be consistent with previous reports of aptamer binding to different cell lines. For example, there is a 2.5- to 6-fold difference in KD for aptamer GL56 against glioma cell lines U87MG and SB-G.(39) Interestingly, the EpCAM aptamer can detect both cell lines (Figs 3P,4A), despite the fact that MDA-MB-231 expresses only 0.2% EpCAM on its cell surface compared to Kato III. Importantly, in the flow cytometric analysis with EpCAM aptamer (Fig. 3P), MDA-MB-231 has only a slight, but definitive, right-shift in mean fluorescence intensity from the background fluorescence in the EpCAM-negative HEK293T cells. That is, MDA-MB-231 displayed a much weaker mean fluorescence intensity than Kato III. Therefore, these data correspond well with the known level of EpCAM in the cell lines, suggesting that the EpCAM aptamer does bind specifically to the EpCAM on cancer cells but not to other molecules present on the human cell surface.
For clinical applications, the high affinity of a tumor-targeting ligand may not necessarily be advantageous. This is particularly true for EpCAM, as in addition to being expressed at high levels in most human adenocarcinomas, EpCAM is also expressed at low levels in a number of normal epithelial cells, including gastrointestinal tract, bile ducts, and pancreas.(40) Therefore, a high affinity EpCAM ligand might bind to EpCAM-positive tumors as well as normal epithelial cells. Indeed, recent clinical trials showed that high-affinity humanized mAbs to EpCAM, 362W94 (K’d = 0.19 nM) and ING-1 (K’d = 0.16 nM) caused acute pancreatitis in patients,(41–43) whereas an antibody to EpCAM with moderate high affinity, adecatumumab (K’d = 91 nM), did not display such dose-limiting toxicity and was well tolerated in patients.(37) Importantly, our EpCAM aptamers do not bind to human cells that do not express EpCAM (Figs 3M–Q,4C). Thus, the DY647-EpDT3 with a moderate affinity (K’d = approximately 55 nM) could serve as a selective tumor-targeting agent for cancer nanomedicine.
Having determined the specificity and selectivity of our EpCAM aptamer, we sought to determine if this molecule is internalized after binding to its cell surface target. It is critically important for a cancer-targeting ligand to be actively internalized into the cancer cells as this will enable the delivery and release of the payload, such as chemotherapy drugs, toxins, and/or therapeutic radioisotopes, inside cancer cells, thus overcoming multidrug resistance and minimizing the collateral killing of normal cells. Furthermore, the intracellular entrapment of a molecular imaging ligand will afford higher tumor to blood or tumor to normal tissue ratio, enabling sensitive detection of cancer in vivo. One of the disadvantages associated with aptamers, and nucleic acids in general, is that these molecules cannot be directly taken up into the cell due to the repulsion between the negatively charged nucleic acids and the cell membrane. In general, nucleic acid aptamers are not capable of being specifically internalized inside living cells, with a few exceptions being a DNA anti-PTK7 aptamer and RNA aptamers against PSMA, CD4, and HIV gp120.(44–46) The length of the final truncation product of EpDT3 is 19 nt, with a molecular mass of approximately 5.9 kDa. In this study, we showed that EpDT3 aptamer is efficiently internalized following binding to the cell surface EpCAM (Fig. 4). We obtained evidence suggesting that our aptamer is internalized in an energy-dependent manner, as incubating cells at 4°C or pretreating cells with 0.4% sodium azide, which is an inhibitor of cellular respiration and thus inhibits active transport, significantly attenuated aptamer internalization (data not shown). We then subjected cells to pretreatment of hypertonic sucrose and potassium depletion before incubation with the EpCAM aptamer (Fig. 4D). Hypertonic shock inhibits several internalization pathways in addition to clathrin-mediated endocytosis, whereas potassium depletion is a more selective inhibitor to lipid raft/caveolae-mediated endocytosis.(32) Both of these treatments blocked the internalization of transferrin (Fig. S2), which is known to be internalized through receptor-mediated or clathrin-mediated endocytosis.(31) Therefore, it is plausible that, after binding to cell surface EpCAM, the EpCAM aptamer is actively internalized through receptor-mediated endocytosis.
To conclude, we have developed the first RNA aptamer against a cancer stem cell marker that specifically binds to the cell surface EpCAM followed by active internalization. This small and chemically synthesized cancer stem cell ligand will facilitate the development of novel targeted cancer nanomedicine and molecular imaging agents.
We thank Michael Famulok and Anthony Keefe for advice, and H. Feng for generating the EpCAM expression construct. This work was supported by the Australia–India Strategic Research Fund grant ST010013 and Victorian Cancer Agency grant PTCP-02 to WD.
None of the authors have any conflict of interest.