Introduction

As reported by Ling et al. [1], during the past decade, over 25,000 papers deposited on PubMed have reported on various aspects of microRNA (miRNA) genomics, biogenesis, mechanisms of action, pathway involvement, phenotypes in experimental models, and disease abnormalities. About 40 % of these publications have focused on the role of miRNAs in cancer. These studies have shown that miRNAs are dysregulated in almost all types of human cancer and specific signatures of aberrantly expressed miRNAs harbor diagnostic, prognostic, and therapeutic implications, including early diagnosis and staging, discrimination between different types of cancer, the identification of the tissue of origin of poorly differentiated tumors, prediction of chemotherapeutic response, etc. These are all challenges that are still largely unresolved and the subject of intense clinical interest [1].

The overwhelming evidence in the literature of the value of miRNAs as cancer biomarkers underscores the importance of developing tools to understand how miRNAs participate in key aspects of carcinogenesis. Here, we describe technology that provides this capability in an in vivo setting. Specifically, we have developed nanosensors based on fluorescent turn-on oligonucleotide probes (oligos) sensitive to individual microRNA-mediated RNA interference [2]. The nanosensors consist of dextran-coated aminated iron oxide magnetic nanoparticles (MNs, 25–30 nm in diameter), conjugated to sensor oligos that are perfectly complementary to endogenous miRNA species. It is well known now that the choice between messenger RNA (mRNA) degradation and translational repression is dependent on the degree of complementarity between the mRNA target and the miRNA seed region [3]. Earlier, we had hypothesized and then confirmed [2] that sensor oligonucleotides perfectly complementary around the miRNA seed region would be cleaved in a sequence-specific way by the microRNA-RNA-induced silencing complex (RISC). In this work, we showed that a perfectly complementary sensor oligonucleotide specific to miR-10b is cleaved by miR-10b-RISC, despite the fact that all of miR-10b’s known mRNA targets are bound with imperfect complementarity and regulated by translational repression and not mRNA cleavage. The sensor oligonucleotides are labeled with a fluorescent dye-quencher pair, so that upon cleavage of the oligonucleotide by the microRNA-RISC, fluorescence enhancement is observed.

We believe that the nanosensors have significant advantages over current technology. The currently established methods for microRNA detection in situ rely on PCR and northern blotting or high-affinity hybridization probes [4–14]. However, these methods are only applicable in vitro. Consequently, they do not permit longitudinal studies, in which the “evolution” of the tumor cell phenotype is monitored in an intact physiologic environment. In contrast, our approach offers the possibility for in vivo detection and monitoring. In addition, the existing methods rely on direct hybridization of the sensor oligo to the miRNA, reflecting a 1:1 ratio of fluorescent probe per miRNA. By contrast, our approach employs a powerful signal amplification strategy. Namely, consistent with our experience [15], each cell in this study is estimated to take up over 1 × 106 nanoparticles with attached sensor oligos. Each miRNA-RISC cleaves its complementary substrate oligo catalytically, leading to powerful signal amplification resulting from the cleavage of millions of synthetic substrates on the nanoparticles by the cognate miRNA-RISC complex [16, 17]. Finally, the described method, unlike approaches that rely on bioluminescent or fluorescent reporter gene expression [18–21], is composed of simple synthetic components that would one day make clinical applications more feasible. Also, unlike hybridization probes (e.g., molecular beacons and SmartFlare, Millipore) the nanosensor does not remain stably bound to the miRNA and does not interfere with cellular function [2].

In the current study, we demonstrate the feasibility of the approach by focusing on miRNA-10b. miR-10b has been implicated in epithelial to mesenchymal transition and breast cancer metastasis by multiple studies including by our own group [22, 23–25]. Consequently, it is an attractive target with preclinical and clinical significance. Still, the technology can be utilized to measure the expression of any miRNA in any solid tumor model to answer questions related to cancer epigenetics. For example, it can be applied for monitoring the expression of microRNAs implicated in cancer progression and can be utilized in preclinical drug development. In particular, since microRNAs represent highly specific and/or early predictive biomarkers of metastatic potential, the technology could be used to gather information that could aid in forecasting invasive disease and predicting response to specific therapeutic modalities, leading to the design of individualized treatments. Clinically, the described technology could one day be used as a diagnostic and/or prognostic tool and support monitoring of treatment response [26–28].

Materials and Methods

Sensor Oligonucleotide

The sensor oligonucleotide (5′-5ThioMC6D/UUU/Cy5/CACAAAUUCGGUUCUACAGGGUA/IAbRQSp-3′) was designed to be perfectly complementary to the target miRNA, miRNA-10b, which plays a key role in epithelial-mesenchymal transition [23]. The backbone of the sensor oligonucleotide was composed of natural RNA bases. The 5′-end of the sequence was modified with Cy5 (fluorescent dye) and a spacer with three nucleic acids, which had a Thio-MC6-protected thiol group for the conjugation to MNs. Iowa Black RQ (quencher) was placed at the 3′-end of the sensor oligonucleotide. The oligo was synthesized by Integrated DNA Technologies (Coralville, IA).

Synthesis of Dextran-Coated MNs

MN was synthesized following a protocol published previously [22]. Briefly, 30 ml of Dextan-T10 (0.3 g/ml, Pharmacosmos A/S, Holbaek, Denmark) was mixed with 1 ml of FeCl3·6H2O (0.65 g/ml, Sigma, Saint Louis, MO) while flushing argon gas for an hour. One milliliter of FeCl2·4H2O (0.4 g/ml, Sigma) was added to the mixture, and 15 ml of cold NH4OH (28 %, Sigma) was added dropwise to the stirred mixture. The temperature was increased to 85 °C for 1 h to start the formation of a nanoparticle dispersion and then cooled to room temperature. The magnetic nanoparticles were concentrated to 20 ml using Amicon ultra centrifugal units (MWCO 30 kDa; Millipore, Darmstadt, Germany). The resulting dextran-coated magnetic nanoparticles were cross-linked by epichlorohydrin (14 ml, 8 h, Sigma) and aminated with subsequent addition of NH4OH (28 %, 60 ml). Aminated MNs were purified by dialysis and concentrated using Amicon ultra centrifugal units.

MN properties were as follows: Iron concentration was 10.94 mg/ml as determined using Total Iron Reagent Set (Pointe Scientific, Canton, MI); each nanoparticle contained 74 amine groups as determined by the N-succinimidyl 3-[2-pyridyldithio]-propionate (SPDP) quantification method; the size of MN was 20.3 ± 0.6 nm as determined by dynamic light scattering (Zetasizer Nano ZS, Malvern Instruments Ltd., Westborough, MA).

Synthesis and Characterization of MN-miR-10b Nanosensor

MN was sequentially conjugated to the heterobifunctional linker SPDP (Thermoscientific Co., Rockford, IL) and activated oligonucleotides. Briefly, 7.5 mg of SPDP was dissolved in 25 μl of anhydrous DMSO and incubated with MN, which has a maleimide group for thioether linkage with activated oligonucleotides. To release the thiol from the 5′-ThioMC6 linker and activate the oligonucleotides for conjugation, we used 3 % Tris(2-chloroethyl)phosphate (TCEP) treatment in nuclease-free phosphate-buffered saline (PBS). Activated oligonucleotides were purified using an ammonium acetate/ethanol precipitation method. After TCEP activation and purification, activated oligonucleotides were dissolved in nuclease-free water and incubated with the SPDP-modified MN overnight. The MN-miR-10b nanosensor was purified from unconjugated sensor oligonucleotide using a magnetic column (MACS 20 μ column, Miltenyi Biotech Inc., San Diego, CA).

The MN-miR-10b nanosensor was freshly prepared before each injection, and the number of oligonucleotides per MN was estimated as 3.0 by electrophoresis as described previously [22, 29]. The final size of the MN-miR-10b nanosensor was 22.2 ± 0.3 nm.

Cells

The human breast cancer cell line, MDA-MB-231-luc-D3H2LN, was purchased from Perkin Elmer (Hopkinton, MA) and cultured in Dulbecco’s modified Eagle’s medium (Sigma, Saint Louise, MO) supplemented with 10 % fetal bovine serum (FBS) (Thermo Scientific, Waltham, MA), 1 % antibiotics (Invitrogen, Carlsbad, CA), and 2 mM L-glutamine. The human breast cancer cell line MCF-7 was purchased from ATCC and cultured in RPMI 1640 medium with 5 % FBS and 1 % antibiotics, per the supplier’s instructions (ATCC, Manassas, VA).

Fluorescence Activation by Nuclease A

To test fluorescence activation of the synthesized nanosensor, we incubated it with specific nuclease.

Five microliters of purified nanosensor solution (10 μg Fe, 400 pmol as sensor oligo) was mixed with 92 μl of PBS buffer in a 96-well plate. The plate was read one time before the addition of 3 μl nuclease A (10 mg/ml, 700 U/ml, Qiagen, Hilden, Germany). The reading was repeated according to a prefixed schedule to obtain the maximum fluorescence activation after the addition of nuclease A. The fluorescence spectra were collected using the following parameters: excitation wavelength—600 nm, emission wavelength—640–800, cutoff—630 nm, interval—2 nm, acquisition number—6.

Animal Model

The miR-10b-inhibited tumor model was generated as described in [22]. Briefly, six-week-old female old nude mice (n = 5) were implanted subcutaneously with MDA-MB-231-luc-D3H2LN cells (2 × 106 cells in the right flank). When tumors reached a diameter of ~0.5 cm, the animals were injected intravenously with a miR-10b inhibitor (MN-anti-miR-10b, nanoparticles carrying locked nucleic acid binding to miR10-b) or an inactive control (MN-scr-miR, nanoparticles carrying irrelevant locked nucleic acid) to generate the miR-10b inhibited group or active group, respectively. This treatment leads to complete inhibition of miR10-b as we have shown in our previous work [22].

For the bilateral tumor model, six-week-old female athymic nude mice (n = 5) were treated with estradiol cypionate (3 mg/kg) in cottonseed oil for 2 weeks and implanted subcutaneously in the flanks with MDA-MB-231-luc-D3H2LN (2 × 106 cells) and MCF-7 (2 × 106 cells). All animal experiments were performed in compliance with institutional guidelines and approved by the Institutional Animal Care and Use Committee (IACUC) at Massachusetts General Hospital.

Fluorescence/Bioluminescence Optical Imaging

Imaging was performed using the IVIS Spectrum imaging system (Perkin Elmer, Hopkinton, MA). Anesthetized mice were intravenously injected with MN-miR-10b nanosensor (15 mg/kg Fe; 1.7 mg/kg sensor oilgo) and scanned 1, 3, 5, 7, 9, 11, 13, 18, 24, and 40 h after injection. The acquisition conditions for fluorescence imaging were as follows: exposure time—0.5 s; binning factor—8; excitation filter range—570, 605, 640, and 675 nm; emission filter range—640, 660, 680, 700, and 720 nm; f number—2. Spectral unmixing (Living Image ver. 4.4) was employed to minimize the influence of tissue autofluorescence. Radiant efficiency was measured for the same region of interest (ROI) in all images. At the end of the fluorescence imaging session, mice were injected intraperitoneally with D-luciferin potassium salt in DPBS (200 μl of 15 mg/ml; Perkin Elmer, Hopkinton, MA). Identical imaging acquisition settings (time, 0.5~60 s; F-stop—2; binning—medium) and the same regions of interest (ROIs) were used to obtain total radiance (photons/s/cm2/sr) from the tumors.

Real-Time Quantitative Reverse Transcription-PCR

Surgically removed primary tumor tissues were homogenized to extract the miRNA-enriched fraction. Relative levels of miR-10b were determined by real-time quantitative reverse transcription-PCR (qRT-PCR; Taqman protocol) and compared to the internal housekeeping gene, SNORD44. Taqman analysis was carried out using an ABI Prism 7700 sequence detection system (Applied Biosystems, Foster City, CA). The primers (Hs-miR-10b-3 miScript Primer, Hs-SNORD44-11 miScript Primer) and assay kit (miScript PCR Starter Kit) were purchased from Qiagen. Relative expression was calculated by the ΔΔCt method.

Fluorescence Microscopy of Tissue Sections

Excised tumor tissues were embedded in Tissue-Tek OCT compound (Sakura Finetek, Torrance, CA) and snap frozen in liquid nitrogen. The tissues were cut into 7-μm sections. The slides were counterstained and mounted with Vectashield mounting medium with (4′,6-diamidino-2-phenylindole (DAPI) (Vector Laboratories, Inc. Burlingame, CA). To determine MN-miR-10b accumulation in tissue, the sections were analyzed by fluorescence microscopy using a Nikon Eclipse 50i fluorescence microscope (Nikon, Tokyo, Japan), equipped with the necessary filter sets (Chroma Technology Corporation, Bellows Falls, VT). Images were acquired using a charge-coupled device camera with near-IR sensitivity (SPOT 7.4 Slider RTKE; Diagnostic Instruments, Sterling Heights, MI). The images were analyzed using SPOT 4.0 Advance version software (Diagnostic Instruments, Sterling Heights, MI).

Statistical Analysis

Data were expressed as mean ± SD. Statistical comparisons were drawn using a two-tailed t test (SigmaStat 3.0; Systat Software, Richmond, CA). A value of p < 0.05 was considered statistically significant.

Results

General Design and Mechanism

The design of the MN-miRNA nanosensor is based on our prior publications [30, 31] [22] in which we have described the development and application of oligonucleotide-modified nanoparticles for the targeting of the RNAi mechanism in tumor cells. Specifically, the nanosensor consists of MNs (~20 nm in diameter), conjugated to sensor oligos [2] that are perfectly complementary to endogenous miRNA species (Fig. 1a). The design and testing of the prototype sensor oligos have recently been described by us [2]. Specifically, the sensor oligonucleotides are composed of RNA bases and are cleavable (nonstabilized by chemical modification) around the seed region (the conserved region within which the microRNA engages the RNA substrate). The oligonucleotide strand is labeled with a fluorescent dye-quencher pair, so that upon its cleavage by the microRNA-RISC, the dye-quencher pair separates resulting in dequenching and fluorescence enhancement (Fig. 1a). In our prior study, we demonstrated the feasibility of accurately detecting miRNA expression in a cell-free system and intact cells in vitro using the described sensor oligonucleotide design and extensively validated our hypothesis that fluorescence enhancement resulted from specific cleavage of the sensor oligo by the miR-primed RISC [2].

Fig. 1
figure 1

Nanosensor design and mechanism of action. a The nanosensor consists of magnetic nanoparticles (MNs), conjugated to sensor oligos that are perfectly complementary to endogenous miRNA species. These sensor oligonucleotides are composed of RNA bases and are cleavable (nonstabilized by chemical modification) around the seed region (the conserved region within which the microRNA engages the RNA substrate). The oligonucleotide strand is labeled with a fluorescent dye-quencher pair, so that upon its cleavage by the microRNA-RISC, the dye-quencher pair separates producing dequenched fluorescence enhancement. b In the cytosol, the nanoparticles, which carry a sensor oligo complementary to an endogenous microRNA species, bind the microRNA, leading to the recruitment to the endogenous RNA-induced silencing complex (RISC) and cleavage of the oligo at a specific position in the seed region. This cleavage results in separation between the quencher and dye located at the ends of the sensor oligo and fluorescent turn-on. The microRNA is released from the complex and is free to catalyze subsequent cleavage reactions.

Full size image

For in vivo delivery, the oligo was conjugated to the nanoparticles. Nanosensor accumulation in the tumor is driven by the innate pathways of hemodynamics and biodistribution characteristic for the nanoparticles and governed by the well-established enhanced permeability and retention (EPR) effect, as indicated by our earlier work [29, 30]. The specific mechanism of nanosensor action is described in Fig. 1b. Namely, following intravenous injection, the long-circulating nanoparticles distribute through the circulation and progressively accumulate in the tumor interstitium due to its hyperpermeable vasculature and compromised lymphatic drainage (EPR effect) [32]. Once inside the tumor interstitium, the nanoparticles are endocytosed by the tumor cells through macropinocytosis and localize to endosomes [29]. Inside endosomes, the functionalized nanoparticles rich in unsaturated amines mediate the proton sponge effect leading to endosomal rupture and nanosensor deposition in the cytosol [33]. In the cytosol, the nanoparticles, which carry a sensor oligo complementary to an endogenous microRNA species, bind the microRNA based on complementarity to its seed region, leading to the recruitment to the endogenous RISC and cleavage and degradation of the oligo. This cleavage results in separation between the quencher and dye located at the ends of the sensor oligo and fluorescent turn-on. The microRNA is released from the complex and is free to catalyze subsequent cleavage reactions [2].

The described mechanism exploits the endogenous process of RNA interference and takes advantage of powerful signal amplification. Namely, a single miRNA will cleave numerous substrate oligos, since RNAi is a “catalytic” molecular mechanism. In addition, a single nanoparticle can carry up to 40 substrate oligos, taking advantage of the phenomenon of multivalency [34, 35]. Finally, the number of nanoparticles taken up by the tumor cells is large (15 × 106 nanoparticles/cell) [15].

The MN-miRNA Nanosensor Is Functional

To assess the feasibility of the approach, we designed a nanosensor that can detect the expression of miRNA-10b. miR-10b has been implicated in epithelial to mesenchymal transition and breast cancer metastasis by multiple studies including by our own group [22, 23–25]. Consequently, the capacity to detect miR-10b expression in tumor cells can have an important preclinical and clinical impact.

To demonstrate that the nanosensor was synthesized as designed and to prove successful incorporation of the sensor oligonucleotide into the nanosensor, after purification, the nanosensor was treated with a reducing agent (TCEP). Electrophoretic analysis confirmed the successful release of the disulphide-conjugated sensor oligonucleotide from MN following TCEP treatment. No free sensor oligonucleotide was detected in the absence of TCEP, indicating that in the final nanosensor preparation, the sensor oligonucleotides were conjugated to MN (Fig. 2a).

Fig. 2
figure 2

Activation of fluorescence and the release of the sensor oligonucleotide from the MN-miR-10b nanosensor. a Electrophoresis confirmed the release of sensor oligonucleotides from MN after TCEP treatment. b Cleavage of the sensor oligo on the nanosensor by nuclease A leads to a 415 % increase in fluorescence intensity over the background.

Full size image

Having previously established that the sensor oligo can report on miRNA expression in tumor cells [2], we next set out to determine if conjugation to the nanoparticle carrier preserves the functionality of the oligo. Figure 2 illustrates the activation of the MN-miR-10b nanosensor by nuclease treatment in vitro. The nanosensor was incubated with excess RNAse A to simulate the catalytic activation of sensor oligonucleotide. As a control, the MN-miR-10b nanosensor was incubated in nuclease-free water. These results showed an over fourfold increase in fluorescence after incubation with RNAse A, indicating that conjugation to the MN carrier does not interfere with sensor oligo activation (Fig. 2b).

The MN-miRNA Nanosensor Can Report on miRNA Expression In Vivo

To determine the feasibility of detecting miRNA expression in vivo, we first generated an animal model in which miR-10b expression in the tumors was either left active or inhibited, as described in [22]. We injected the nanosensor intravenously into tumor-bearing animals and performed whole-body fluorescence reflectance imaging over a time course of 0 to 24 h using filters for the Cy5 dye on the sensor oligo. Bioluminescence imaging (BLI) for the accurate delineation of viable tumor mass was also performed.

We observed clear enhancement of the primary tumor, consistent with cleavage of the sensor oligo, separation of the Cy5 dye from the quencher, and a fluorescent turn-on (Fig. 3a). This effect was specific for miR-10b, since in the animals in which miR-10b was inhibited, the signal enhancement was significantly lower than in the animals in which miR-10b was active.

Fig. 3
figure 3

In vivo detection of miR-10b expression in a miR-10b-inhibited breast tumor model. a Fluorescence optical imaging showing significantly higher fluorescence intensity in the tumors in which miR-10b was left active over the miR-10b-inhibited tumors (i.v., 15 mg/kg Fe; 1.7 mg/kg sensor oilgo). Tumor ROIs are color-coded according to radiant efficiency. b Quantification of radiant efficiency from A indicated a peak of fluorescence activation in the miR-10b-active tumors at 18 h after nanosensor injection. c qRT-PCR of miR-10b expression in the miR-10b-active vs miR-10b-inhibited tumors demonstrated a significant difference between the two groups (data represent average ± SD; two-tailed t test; n = 3). d Fluorescence microscopy of tumor tissue from mice injected with the nanosensor. Cy5 fluorescence from the activated nanosensor was seen in the miR-10b-active tumors but not the miR-10b-inhibited tumors (blue—DAPI, nuclei; red—Cy5, activated nanosensor).

Full size image

Quantitative assessment of relative radiance efficiency revealed that the maximal difference between miR-10b-inhibited and miR-10b-active tumors was achieved 18 h after nanosensor injection (Fig. 3b, 790 ± 70 % enhancement at the 18 h time point t test, n = 3, p < 0.05), despite a lack of difference in viable tumor mass assessed by BLI. qRT-PCR confirmed these data demonstrating differential miR-10b expression between the miR-10b-inhibited and miR-10b-active tumors (Fig. 3c).

Ex vivo examination of tumor tissues confirmed the observed differential fluorescence intensity between miR-10b-inhibited and miR-10b-active tumors and confirmed that fluorescence in the miR-10b active group associated with tumor cells (Fig. 3d).

In a more clinically relevant model, we illustrated the application of the nanosensor in nude mice implanted bilaterally with the highly invasive MDA-MB-231 and the noninvasive MCF-7 breast cancer cell lines. After intravenous injection of the nanosensor, we observed significantly higher fluorescence in the MDA-MB-231 tumors than the MCF-7 tumors (Fig. 4a), corresponding to a significant difference in radiant efficiency (Fig. 4b, 63 ± 6 % enhancement at the 15-h time point, t test, n = 3, p < 0.05). This difference was most pronounced at 15–24 h after injection of the nanosensor (Fig. 4a, b) and reflected a difference in the expression levels of miR-10b between the invasive MDA-MB-231 and noninvasive MCF-7 tumors (Fig. 4c). Finally, fluorescence microscopy confirmed the differential fluorescence between the two tumors and revealed that the activated oligo resided in the cytosol of the tumor cells (Fig. 4d).

Fig. 4
figure 4

In vivo detection of miR-10b expression using the nanosensor in mice bilaterally implanted with MDA-MB-231 and MCF-7 cells. a Fluorescence optical imaging showed differential fluorescence intensity from the two tumors (i.v., 15 mg/kg Fe; 1.7 mg/kg sensor oligo). b Quantification of radiant efficiency showed a peak of differential radiant efficiency at 15–24 h after nanosensor injection. c qRT-PCR expression demonstrated significantly higher expression of miR-10b in the MDA-MB-231 tumors (data represent average ± SD; two-tailed t test; n = 3). d Fluorescence microscopy of tumor tissues from MDA-MB-231 and MCF-7 tumors. Cy5 radiant efficiency from the activated nanosensor was higher in the MDA-MB-231 tumors than in the MCF-7 tumors (blue—DAPI, nuclei; red—Cy5, activated nanosensor).

Full size image

The in vivo kinetics of the signal is expected. A 15- to 24-h range represents 1.5–2 blood half-lives of the nanosensor in mice. Therefore, at the time of maximum fluorescence intensity, about 25 % of the original injected dose would still exist in the circulation and would contribute to an increase in background signal. However, at later time points, the relative contribution of nanosensor in the tumor circulation vs inside tumor cells would be reduced due to nanosenser clearance. At that point, most of the fluorescence signal from tumor tissues is expected to be coming from activated sensor oligo inside tumor cells. However, because of a time lag in the passive delivery of the nanosensor into tumor tissues through the EPR effect, there would be a longer activation time in vivo than expected from the kinetics of the RNAi process.

After dynamic fluorescence signal quantification, the suggested mechanism of nanosensor activation is as follows. At the early time points, there is a large amount of nanosensor in the blood, but the oligo is intact (not cleaved) and therefore mostly “dark.” Consequently, the changes in fluorescence at these time points would reflect nanosensor pharmacodynamics (Suppl Figs. 1a and 2a). At 15–24 h, the nanosensor is localized in tumor tissues and continuously taken up by cells, leading to the cleavage (activation) of the oligo. However, at that time point, the cleavage products have not yet been exocytosed/cleared, resulting in a dramatic and rapid increase in fluorescence (Suppl. Figs. 1b and 2b). After 24 h, the cleaved products are secreted by the tumor cells and cleared through the circulation, leading to a loss of fluorescence (Suppl. Figs. 1b and 2b). This hypothesis is consistent with our earlier findings, indicating exocytosis of the oligonucleotide fragments after cleavage [2].

Discussion

With the goal of gaining a better understanding of the role of miRNA in cancer and of making miRNA-based diagnostics possible, we have developed technology for the profiling of tumor miRNAs in vivo. In these early feasibility studies, the nanoparticle-conjugated oligo was terminally labeled with the dye and the quencher. Consequently, the two were separated by 22 nucleotides (7.5 nm). However, it is possible to optimize the oligo by introducing internal labels and using alternative dye-quencher pairs. For example, if we have a 10-nucleotide (nt) distance (3.4 nm) between the dye and quencher, that would result in a distance of of 3.4 nm (10 × 3.4 Å/nucleotide = 34 Å = 3.4 nm). For the IRDye 700DX-QC-1 pair for example, the expected quenching efficiency at that distance will be 98.8 % (Li-COR Biosciences, Lincoln, NE).

In our studies, we demonstrated that the technology allows for profiling tumor miRNA signatures in vivo. This is important because it is only through in vivo studies that one can capture the true dynamics of cancer initiation and progression. Since the method is completely noninvasive, it will be possible to collect time-course data in genuine physiologic environments and thus gain a better understanding of individual-to-individual variation, authentic molecular processes, and long-term trends.

In a laboratory setting, the tools developed here can lead to the mapping out of a comprehensive and systematic atlas of cancer progression. In a clinical setting, the technology will facilitate early cancer detection and/or cancer risk assessment and facilitate/accelerate the process of drug discovery.

With specific relevance to clinical application, the nanocarrier component of the nanosensors is based on a clinically approved nanoparticle [36, 37]. Consequently, a derivative of the developed technology may one day become applicable in a clinical setting and as such has the potential of advancing our ability to diagnose and treat breast cancer.

A variety of scenarios could benefit from the described technology. For example, it has been established that miR-10b is upregulated in the tumors of patients who harbor metastatic disease, indicating that miR-10b could be used to diagnose the presence of metastasis or even to forecast invasive disease [25, 38, 39]. The ability to detect miR-10b upregulation in the primary tumors by noninvasive or laparoscopic methods that specifically target a suspected lesion would be highly clinically beneficial.

miR-10b is also part of a panel of miRNAs that demonstrate aberrant expression in tissues from patients with triple negative breast cancer that displays chemoresistance [40]. In this context, noninvasive or laparoscopic imaging can help guide therapeutic decisions.

These are just a few examples of how the described nanosensors could be of preclinical and clinical value. However, the list of miRNAs implicated in various aspects of tumorigenesis is poised to expand rapidly over the next decade. In addition, the knowledge about the precise roles of each of the roughly 1500 known human miRNAs is increasingly becoming more focused, as specific targets are being identified and validated in vitro. Consequently, the development of similar technologies for miRNA-based diagnostics will inevitably represent an element of precision medicine, defined by rational drug design, individualized intervention, and targeted image-guided delivery of therapy.

Conclusions

Considering that microRNAs represent highly specific and very early and/or predictive biomarkers of metastatic potential, risk assessment, prognosis, and treatment response, the described methodology can help acquire a better understanding of the epigenetic mechanisms underlying cancer emergence and progression. This knowledge could ultimately assist in forecasting invasive disease and predicting response to specific therapeutic modalities, leading to the design of individualized curative or even preventive treatments. In the long run, since magnetic nanoparticles are already in clinical use and the technology for optical imaging of the breast is gaining clinical relevance, one can envision a utility for our method in application to image-guided laparoscopic or intraoperative diagnostics.