The enhanced binding of multiple ligands to a particular cellular target is a common approach in nature to fine-tune molecular and cellular recognitions with increased specificity.1-11
This kind of multivalent binding strategy can offer unique advantages for developing selective and highly sensitive nanoparticle-based diagnostics and effective therapeutics. Nanoparticles with multiple targeting ligands offer the advantage of a surface-mediated multivalent affinity, resulting from multiple interactions between the high local concentration of binding ligands on the nanoparticle's surface and epitopes on the corresponding target. In particular, the conjugation of multiple targeting ligands to iron oxide nanoparticles (IONP) has allowed the creation of multivalent magnetic relaxation nanosensors (MRnS) for the detection of molecular targets and events, such as DNA, RNA, proteins, enzymatic activity, small molecule, viruses, enzymatic and metabolic activity.12-16
Detection is achieved by changes in the solution's water relaxation times (ΔT2), as these nanosensors self-assemble upon interaction with the specific target. However, all these cases shared a common target characteristic; the target was smaller than or had roughly the same size (in the case of a virus) with the nanosensor. Recently, the detection of a bacterium, a much larger target compared to the nanosensor, was reported.17
In that report, it was found that the use of a multivalent entity (bacterium) as a biological target compensated for the size difference between the nanoprobe and the target, promoting nanoparticle assembly on the bacterial surface with concomitant target-concentration-dependent changes in ΔT2. It was speculated that these differences may have been attributed to the ratio of nanoparticles interacting per target, hinting that at low bacterial concentrations more nanoparticles self-assembled on the surface of the bacterium (hence higher ΔT2), whereas at high bacterial concentrations fewer nanoparticles interacted per target to result in lower ΔT2.
Therefore, we hypothesized whether the nanoparticles' valency may affect the MRnS detection limit, allowing the engineering of ultrasensitive probes to accommodate a particular cellular concentration range. We reasoned that nanoparticles with low amounts of ligand conjugated on their surface (low valency) would assemble on the cell's surface, resulting in prominent shifts in the ΔT2 (). As the cell concentration increases, the low valency nanoparticles would switch to a quasi-dispersed state, due to their limited interaction with target moieties on discrete cells, thus causing smaller changes in the ΔT2 at high cell concentrations (). This mechanism would be in line with the reported MRnS-mediated detection of bacteria (Mycobacterium avium
– MAP) using magnetic nanoparticles conjugated with anti-MAP polyclonal antibodies, where prominent ΔT2 was observed at low MAP amounts and low ΔT2 was recorded at high MAP concentrations. 17
In contrast, we reasoned that high valency nanoparticles should not cluster in the presence of cells at low concentrations. Instead, the multiple ligands present on the high valency nanoparticle can interact with multiple receptors on the cell surface (). Therefore, the interaction between high valency nanoparticles and cells at low concentration would result in a less pronounced ΔT2, because a fewer number of nanoparticles may simultaneously interact with multiple receptors in a given cell. Alternatively, as the cell concentration increases, the probability of high valency nanoparticle binding to surface receptors in multiple cells increases. This should facilitate the binding of multiple ligands on the same nanoparticle with multiple receptors on different cells, causing extensive clustering of the nanoparticles and an increase in ΔT2 as the number of cells increases. ().
Scheme 1 The iron oxide nanoparticles's valency facilitates distinct magnetic relaxation sensing trends, which are modulated by the nanoparticle – target interactions. A) At low target concentrations, nanoparticles with low valency readily assemble on (more ...)
Hence, the lack of a comprehensive study addressing how multivalency affects the nanoparticles' cell surface assembly (e.g. bacterium or cell) and the corresponding MRnS response prompted us to utilize MRnS with engineered valency towards the detection of cancer cells in blood samples. We selected cancer cells as a model target, because it has been recently reported that viable tumor-derived epithelial cells found in circulation (circulating tumor cells) are a novel class of cancer biomarkers, participating in the initiation of the process of metastasis. 18-20
Current molecular diagnostic techniques cannot detect low concentrations of cells in complex media (i.e. blood),20
because of the matrix's optical properties and complexity. Furthermore, commonly used cytological and immunocytochemical techniques require the cell's isolation, purification, fixation and quantification using fluorescent probes and microscopic examination, which prevent propagation of the cells for further analyses.21,22
All of the above highlight the importance of developing nondestructive methods for the sensitive detection of circulating tumor cells in blood samples. For our studies, we decided to use a small molecule as opposed to an antibody, because its small size could guarantee higher nanoparticle multivalency and its stability would lead to the development of robust field-deployable nanosensors that are not susceptible to thermal denaturation. As a model small molecule ligand, we chose folic acid (folate), which is the canonical affinity ligand of the folate receptor (FR) that is overexpressed in some tumor cells.23
Although circulating tumor cells more prevalently overexpress other cell surface markers, such as EpCAM,19
recent studies indicate that the folate receptor (FR) is overexpressed in numerous cancers, including ovarian, testicular, breast and lung. 23
Furthermore, as the expression of FR is upregulated during metastasis, this receptor is a good marker for diagnosis, targeted imaging and drug delivery. As our cellular target, we used A549 lung cancer cells, as lung cancer is among the most common sites of origin of metastatic cancer through migration of lung cancer cells via the circulation,24
and this cell line is known to overexpress the folate receptor.23
Therefore, we investigated if the folate nanoparticle preparations, and particularly the high-folate ones, can be used for the quantification and sensitive detection of single FR-expressing A549 cells in complex media, via magnetic relaxation. Results showed that low valency nanoparticles at low target concentrations induced high ΔT2, whereas at high cell concentrations the nanoparticles switched to a dispersed-like state with lower ΔT2 values. Corresponding dynamic light scattering (DLS) analysis and cell associated fluorescence studies confirmed these results. On the other hand, high valency nanoparticles from a disperse-like state switched to a clustered state, when we increased the cell concentration. Similar observations were made when we used low and high valency nanoparticles for the detection of bacterial cells, indicating that the interaction was modulated by the nanoparticle valency and not the targeted cell. Furthermore, results show that the high valency nanoparticles performed better than their low valency counterpart as they achieve a high sensitivity, faster detection kinetics and more efficient magnetic isolation of cells for subsequent analyses. Hence, considering the need for circulating cancer cell diagnostic modalities, herein we demonstrate that multivalent iron oxide nanosensors, carrying a small affinity ligand, can facilitate (i) fast detection of tumor cells in blood via magnetic relaxation, (ii) the magnetic isolation and propagation of these cells for further analyses, and (iii) the cells' colorimetric identification via the nanoparticles' intrinsic peroxidase activity. The same approach can be expanded to the detection of other cells in circulation, as well as pathogens.