Development of Multifunctional Fluorescent-Magnetic Nanoprobes for Selective Capturing and Multicolor Imaging of Heterogeneous Circulating Tumor Cells.

Circulating tumor cells (CTC) are highly heterogeneous in nature due to epithelial-mesenchymal transition (EMT), which is the major obstacle for CTC analysis via "liquid biopsy". This article reports the development of a new class of multifunctional fluorescent-magnetic multicolor nanoprobes for targeted capturing and accurate identification of heterogeneous CTC. A facile design approach for the synthesis and characterization of bioconjugated multifunctonal nanoprobes that exhibit excellent magnetic properties and emit very bright and photostable multicolor fluorescence at red, green, and blue under 380 nm excitation is reported. Experimental data presented show that the multifunctional multicolor nanoprobes can be used for targeted capture and multicolor fluorescence mapping of heterogeneous CTC and can distinguish targeted CTC from nontargeted cells.

Introduction

Even in the 21st century, due to tumor metastasis, cancer is the second most common cause of death in the USA.[1−4] It is now well documented that circulating tumor cells (CTC) are the main vehicles of metastatic relapse.[5−11] Because CTC are extremely rare cells (1–10 cells/mL) in blood containing millions of leukocytes and erythrocytes cells, detecting CTC without separation from blood is highly challenging in clinics.[6−15] Reported clinical data show that the CTC concentration in blood can be as low as 1 per 107 cells; thus, effective separation, enrichment, and identification steps are necessary, even for patients with advanced cancer.[1−8] Because naturally all untreated biological materials are diamagnetic, in clinics magnetic cell separation is highly popular for the separation of CTC from clinical blood samples using antibody-attached magnetic beads.[1−8] CTC separation from blood is also very important to avoid huge light scattering and autofluorescence from millions of leukocytes and erythrocytes cells.[1−8] The challenge that medical doctors are facing is that CTC are highly heterogeneous. Due to epithelial–mesenchymal transition (EMT), CTC are undetected for more than one-third of metastatic cancer patients.[5−11] This article reports for the first time the development of multicolor nanodot-conjugated magnetic nanoparticle-based multifunctional fluorescent–magnetic nanoprobes which have the capability to capture and identify the heterogeneity of CTC. In our design, highly magnetic properties of multifunctional nanoprobes have been used for the separation of epithelial, mesenchymal, and stem cell CTC from whole blood samples. On the other hand, multicolor fluorescence nanodots at the surface of multifunctional nanoprobes have been used for multicolor imaging of heterogeneous CTC selectively and simultaneously.

In the last few years, nanodots including graphene quantum dots (GQDs), carbon dots (CDs), polymer dots (PDs), and gold cluster dots (GCDs) have emerged as a new type of bright fluorescent probes for biological imaging due to very good photostability and biocompatibility with cells and tissues.[12−26] Because the size of gold clusters dots (GCDs) is comparable to the Fermi wavelength, the free electrons in GCDs generate discrete electronic transitions, which allow them to exhibit strong photoluminescence properties.[27−35] Similarly, the backbone of polymer dots (PDs), which are made from conjugated polymer structures, exhibits a very high optical cross-section, and as a result, PDs display huge fluorescence which can be tuned from visible to NIR by changing the size.[12−18] On the other hand, tunable surface functional groups in CDs exhibit huge photoluminescence which can be attributed to the presence of surface energy traps. The nanodot photoluminescence can be varied by the intrinsic inner structure and surface chemical groups.[19−26] Because organic fluorescent dyes have poor solubility in aqueous solutions and undergo photobleaching,[2−4] nanodots will be better biomolecular probes for fluorescent mapping. Although recently there have been good advances on developing different nanodots with tunable optical properties,[15−22] finding multicolor fluorescent GQDs, GCDs, CDs, or PDs at single wavelength excitation is rare. As a result, for mapping heterogeneous CTC, we used blue color fluorescence PDs, green fluorescence CDs, and red color fluorescence GCDs, where using 380 nm excitation, one can perform multicolor imaging of different subpopulations of CTC using these materials. Due to the absence of magnetic properties, nanodots will not be able to separate CTC from blood samples, and as a result, we have designed multicolor nanodot-attached magnetic nanoparticle-based fluorescent–magnetic nanoprobes for selective separation and mapping of epithelial, mesenchymal, and stem cell CTC simultaneously.

For selective capture and accurate identification of heterogeneous CTC, blue fluorescence PD-conjugated fluorescent–magnetic nanoprobes were attached to epithelial markers (anti-EpCAM or anti-HER2 antibody) which can target SK-BR-3 epithelial cancer cells. On the other hand, green fluorescence CD-conjugated fluorescent–magnetic nanoprobes were attached to mesenchymal markers (anti-twist antibody), which can capture CAL-120 breast cancer cells having high levels of mesenchymal markers, and it will be green in color in a fluorescence image. Similarly, red fluorescence GCD-conjugated fluorescent–magnetic nanoprobes were attached to CSC markers (anti-CD34 antibody), and as a result, captured CSC bone marrow CD34+ stem cells will be red color in a fluorescence image. Our reported result shows that nanodot-decorated multifunctional nanoprobes are capable of capturing and accurately identifying the subpopulations of CTC from whole blood samples.

Results and Discussion

Development and Characterization of Multifunctional Blue Fluorescent Magneto-PD Nanoprobes

Blue fluorescence polymer dot-attached magnetic nanoplatforms were synthesized using a multistep process as shown in Figure S1A,B in Supporting Information. Synthesis details are reported in Supporting Information. In the first step, blue fluorescence polymer dots (PDs) were synthesized using an amphiphilic polymer solvent evaporation technique.[16] For this purpose, amphiphilic copolymer was constructed by conjugating polyethylenimine and d,l-lactide using a ring-opening polymerization method. In the next step, for the development of polymer dots, PEI–PLA copolymer was dissolved in dichloromethane and 1%(w/v) of PVA. The mixture was kept at 35 °C in a vacuum chamber. At the end, the purified particles were characterized by high-resolution tunneling electron microscopy (TEM) and dynamic light scattering (DLS) measurement, as reported in Figure S1 and Table S1 in Supporting Information. Figure S1C shows the TEM image of the polymer dots. The inserted high-resolution image shows that the size of polymer dots is about 2–3 nm. Table S1 indicates that the average size is about 3 nm for polymer dots.

Next, the carboxylic acid-functionalized magnetic nanoparticles were prepared from ferric chloride and 1,6-hexanedioic acid using a coprecipitation method as shown in Scheme 1B. Synthesis details are reported in Supporting Information. After the process was finished, the black precipitate of Fe3O4 nanoparticles was separated from the supernatant using a neodymium magnet. As shown in Figure S1D, the high-resolution SEM image shows that the average particle size is about ∼30 nm. DLS measurement, as reported in Table S1, also indicates that the average size is about 30 nm for the magnetic nanoparticle. Inserted energy-dispersive X-ray (EDX) spectroscopy elemental mapping in Figure S1D clearly shows the presence of Fe in the developed magnetic nanoparticles.

The magnetic properties determined using a superconducting quantum interference device (SQUID) magnetometer at room temperature indicate superparamagnetic behavior with specific saturation magnetization of 39.3 emu g–1 for the amine-functionalized magnetite nanoparticles. At the end, we used EDC/NHS esterification to produce PD-coated magnetic nanoprobes, as shown in Figure S1B in Supporting Information.

For this purpose, 1-ethyl-3-(3-(dimethylamino)propyl)carbodiimide hydrochloride (EDC), N-hydroxysulfosuccinimide (NHS), and 4-(dimethylamino)pyridine (DMAP) were added to PDs and acid-functionalized Fe3O4 nanoparticles. After that, the esterified PD-attached Fe3O4 nanoparticles were separated using a neodymium magnet several times and washed with distilled water to remove the excess reactants. Figure A shows the SEM image which indicates that the size of the PD-coated magnetic nanoplatform is about ∼40 nm, which is only about 10 nm more than that of the magnetic nanoparticle. The inserted EDX elemental mapping clearly shows the presence of Fe, C, and O. We have also performed DLS measurement in the solution phase, as reported in Table 1 in Supporting Information, which indicates that the average size is about 40 nm for the PD-coated magnetic nanoplatform. Figure D shows that magneto-PD nanoprobes are highly magnetic, and as a result, we can separate them very quickly using a bar magnet. SQUID magnetometer property measurement indicates superparamagnetic behavior with specific saturation magnetization of 32.6 emu g–1 for the polymer dot-coated magnetite nanoplatforms.

Figure 1

(A) SEM image shows the morphology of magneto-PD nanoprobes. Inserted EDX elemental mapping shows the presence of Fe, C, and O in the fluorescent–magnetic nanoplatform. (B) Fluorescence image under UV light; B1: blue fluorescence from PD-coated fluorescent–magnetic nanoplatform; B2: fluorescence disappears after magnetic separation. (C) Emission spectra from magneto-PD nanoprobes under 380 nm excitation shows that the λem is around 440 nm. (D) Photograph shows that the magneto-PD nanoprobes are highly magnetic, and as a result, we can separate them by using a bar magnet.

Figure B1 shows the blue emitted fluorescence from magneto-PD nanoprobes in the presence of UV light. Figure B2 shows that no fluorescence is observed from the solution after magnetic separation, which indicates that almost all magneto-PD nanoprobes were separated by the magnet. Figure C shows the emission spectra which clearly indicate that the λmax for emission for magneto-PD nanoprobes are around 440 nm, and as a result, it shows blue color fluorescence. The photoluminescence quantum yield (QY) for magneto-PD nanoprobes was determined by counting the integrated luminescence intensities using quinine sulfate as a standard (QY 54%).[13−16] Quantum yield was calculated with respect to quinine sulfate standard using eq ,[14−25]where the nanoprobe is denoted as nd and the quinine sulfate standard is denoted as ref, Φ is the quantum yield under 380 nm excitation, A is the absorbance, I is the fluorescence intensity, and η is the refractive index. From the experimental photoluminescence and theoretical eq , we determined that the quantum yield for PDs is 0.68 under 380 nm light excitation.

Development and Characterization of Multifunctional Red Fluorescent Magneto-GCD Nanoprobes

Red fluorescence magneto-GCD nanoprobes were synthesized using a multistep process as shown in Figure S2A. Synthesis details are reported in Supporting Information. In the first step, red fluorescent gold cluster dots capped with a bidentate ligand, dihydrolipoic acid (DHLA), were synthesized by mixing sodium hydroxide, α-lipoic acid, HAuCl4·3H2O, and NaBH4 with constant stirring using the reported method.[31] A solid residue was collected in a 20 mL scintillation vial, diluted to a final volume of 5 mL with distilled water, and stored at 4 °C for future use. Figure B shows the TEM image of freshly prepared GCDs. The inserted HRTEM indicates that the GCDs are about 4 nm in size. DLS data as reported in Table S2 also indicate that the average size is about 3 nm for GCDs. Next, amine-functionalized magnetic nanoparticles were synthesized by dissolving FeCl3 in ethylene glycol, sodium acetate, and 1,6-hexadiamine, as we have reported previously.[31] The mixture was sealed in a Teflon-lined stainless steel autoclave and was heated at 230 °C for 8 h. Then the product was washed with hot water and ethanol. Figure S2B shows the SEM images of amine-functionalized magnetic nanoparticles, which indicate that the particle size is about 40 nm. The inserted EDX mapping in Figure S2B clearly shows the presence of Fe. The magnetic properties determined using the SQUID magnetometer indicate superparamagnetic behavior with specific saturation magnetization of 43.6 emu g–1 for the amine-functionalized magnetite nanoparticles. In the final step, we synthesized red fluorescence magneto-GCD nanoprobes. For the formation of fluorescent–magnetic nanoprobes, we used coupling chemistry between the CO2H group of α-lipoic acid-attached GCDs and the NH2 group of the amine-functionalized magnetic nanoparticle via amide linkages, as shown in Figure S2A. Synthetic details are described in Supporting Information. The purified particles were characterized by various spectroscopic techniques such as Fourier transform infrared spectroscopy (FTIR), TEM, and EDX analysis, as reported in Figure S2 and Figure . Figure S2C shows the FTIR data obtained from magneto-GCD nanoprobes. The reported FTIR spectrum shows a very strong amide A band which is due to the amide N–H stretching vibration. The spectrum also shows a strong amide I band which is mainly associated with the C=O stretching vibration related to the backbone conformation and an amide II band which is mainly due to the N–H bending vibration coupled with the C–N stretching vibration. We have also noted an amide III band. The high-resolution TEM data, as shown in Figure B, show that the size of the magneto-GCD nanoprobes is about 55 nm, which has been confirmed using DLS measurement in the solution phase, as reported in Table S2. EDX elemental mapping, as shown in insert of Figure B, confirms the presence of Fe and Au in the magnetic nanoplatform. SQUID magnetometer property measurement indicates superparamagnetic behavior with specific saturation magnetization of 36.6 emu g–1 for the GCD-coated magnetite nanoplatform. Figure C shows the red emitted fluorescence from GCD-coated magnetic nanoplatform in the presence of UV light. Figure D shows the emission spectra, which clearly indicate that the λmax for emission for magneto-GCD nanoprobes are around 680 nm.

Figure 2

(A) TEM image shows the morphology of the GCDs. The inserted HRTEM image indicates that the particle size is around 4 nm. It also indicates a crystalline structure for the gold dots. (B) TEM image shows the morphology of the magneto-GCD nanoprobes. Inserted EDX elemental mapping shows the presence of Fe, Au in the magneto-GCD nanoprobes. (C) Fluorescence image under UV light for the magneto-GCD nanoprobes, which clearly shows red color fluorescence at UV light excitation. (D) Emission spectra from the mixture of magneto-PD and magneto-GCD nanoprobes under 380 nm excitation show that the λem is around 440 nm due to magneto-PDs and that the λem is around 680 nm due to magneto-GCDs. We have used 83 ppm magneto-GCDs and 12 ppm magneto-PDs for the fluorescence measurement. The fluorescence intensity is in arbitrary units (a.u.).

Development and Characterization of Multifunctional Green Fluorescent Magneto-CD Nanoprobes

Green fluorescence carbon dot (CD)-attached magneto-CD nanoprobes were synthesized using a multistep process, as shown in Figure S3. Synthesis details are reported in Supporting Information. Initially, the green fluorescence carbon dots (CDs) were synthesized using a literature method.[20] In brief, o-phenylenediamine was dissolved in pure ethanol, and then the solution was transferred into a stainless steel autoclave with a Teflon liner and heated at 180 °C for 12 h. The autoclave was cooled to room temperature, and the reaction mixture was evaporated using a rotary evaporator. The orange color carbon dots were further purified with silica column chromatography using mixtures of CH2Cl2 and MeOH as eluents. Figure A shows the TEM image of freshly prepared CDs which are about 9 nm in size. Figure B shows the histogram of size distribution for carbon dots measured by DLS, which indicates that the average size is about 10 nm for the GCDs. Next, the acid-functionalized magnetic nanoparticles were prepared from ferric chloride as shown in Figure S3. For the formation of fluorescent–magnetic nanoprobes, we used EDC coupling chemistry. The purified particles were characterized by SEM and EDX analysis, as reported in Figure C, which indicate that the size of the magneto-CD nanoprobes is about 55 nm.

Figure 3

(A) TEM image shows the morphology of the CDs. The inserted HRTEM image indicates that the particle size is around 9 nm. (B) Histogram of size distribution for the carbon dots as measured by DLS. (C) SEM image shows the morphology of the magneto-CD nanoprobes. EDX elemental mapping shows the presence of Fe, C, and O. (D) Fluorescence image under UV light of the magneto-CD nanoprobes, which clearly shows green color fluorescence.

DLS measurement data reported in Table S3 indicates that the average size is about 55 nm for magneto-CD nanoprobes. EDX elemental mapping, as shown in the insert of Figure C, confirms the presence of Fe, C, and O in the magneto-CD nanoprobes. SQUID magnetometer property measurement indicates superparamagnetic behavior with specific saturation magnetization of 34.9 emu g–1 for the magneto-CD nanoprobes. Figure D shows the green emitted fluorescence from the magneto-CD nanoprobes in the presence of UV light. Figure A shows the emission spectra which clearly indicate that the λmax for emission for magneto-CD nanoprobes are around 550 nm, and as a result, they show blue color fluorescence. The photoluminescence quantum yield (QY) for magneto-CD nanoprobes was determined to be 0.23 with respect to quinine sulfate as a standard (QY 54%).

Figure 4

(A) Emission spectra from magneto-PD nanoprobes, magneto-GCD nanoprobes, and magneto-CD nanoprobes under 380 nm excitations, which exhibit three distinct fluorescence regions in the blue, green, and orange range. We used 20 ppm magneto-GCDs, 8.8 ppm magneto-CDs, and 2.9 ppm magneto-PDs for the fluorescence measurement. The fluorescence intensity is in arbitrary units (a.u.). (B) The plot demonstrates the biocompatibility of our magneto-PD nanoprobes. (C) The TEM image shows that antibody-attached GCDs are attached to the CD34+ stem cell surface. (D) The single-photon luminescence image shows that a huge amount of bone marrow CD34+ stem cells are captured by the magneto-GCD nanoprobes. (E) The single-photon fluorescence image from the supernatant indicates that almost all CD34+ stem cells are separated by the magnet. Also peripheral blood mononuclear cells and rabbit blood cells do not bind to anti-CD34 antibody-attached magneto-GCD nanoprobes. (F) Percentage of CD34+ stem cells captured by anti-CD34 antibody-attached magneto-GCD nanoprobes when whole blood was spiked with 10 cells/mL CD34+ stem cells and 106 cells/mL peripheral blood mononuclear cells (PBMC). (G) Percentage of HER2+ cancer cells captured by anti-CD34 antibody-attached magneto-GCD nanoprobes. Our results clearly show that anti-CD34 antibody-attached magneto-GCD nanoprobes are highly selective to capture CD34+ stem cells, and as a result, they do not bind to HER2(+) SK-BR-3 cancer cells. (H) Percentage of cells captured by anti-CD34 antibody-attached magneto-GCD nanoprobes when (i) whole blood was spiked with 10 cells/mL CD34+ stem cells and 106 cells/mL PBMC, (ii) whole blood was spiked with 10 cells/mL CD34+ stem cells and 105 cells/mL HaCaT normal cells, (iii) whole blood was spiked with 4.5 × 105 cells/mL SK-BR-3 cells and 106 cells/mL PBMC, and (iv) whole blood was spiked with 4.5 × 105 cells/mL HaCaT normal cells and 106 cells/mL PBMC. All the reported data clearly show that anti-CD34 antibody-conjugated GCD-coated magnetic nanoplatforms are highly selective for the capture of CD34+ stem cells. (I) Percentage of HER2-positive cells captured by anti-HER2 antibody-attached magneto-PD nanoprobes when (i) whole blood was spiked with 10 cells/mL HER2-positive SK-BR-3 cells and 106 cells/mL PBMC, (ii) whole blood was spiked with 105 cells/mL CD34+ stem cells and 106 cells/mL PBMC, (iii) whole blood was spiked with 105 cells/mL HaCaT cells and 106 cells/mL PBMC. All the reported data clearly show that anti-HER2 antibody-conjugated magneto-PD nanoprobes are highly selective for the capture of HER2-positive SK-BR-3 cells.

Developing Antibody-Conjugated Nanoprobes and Determining Their Biocompatibility and Photostability

Figure A shows that our PD-based, GCD-based, and CD-based magnetic nanoprobes exhibit distinct fluorescence in blue, green, and red color range, respectively, when excited under 380 nm light. As a result, we can use them to target epithelial, mesenchymal, and stem cells selectively and simultaneously. For targeted capture and imaging of SK-BR-3 epithelial cancer cells, blue fluorescence magneto-PD nanoprobes were attached to epithelial markers (anti-HER2 antibody). To accomplish this, initially magneto-PD nanoprobes were coated with amine-modified polyethylene glycol (NH2-PEG). After PEGylation, anti-HER2 antibody was conjugated with amine-functionalized PEG-coated magneto-PD nanoprobes using our reported method.[9,10,24] Similarly, to capture CAL-120 breast cancer cells having high levels of twist mesenchymal markers, green fluorescence magneto-CD nanoprobes were conjugated with anti-twist antibody. Also to target bone marrow CD34+ stem cells, red fluorescence magneto-GCD nanoprobes were conjugated with anti-CD34 antibody.

Because biocompatibility is very important for imaging, first we determined the biocompatibility of the antibody-attached fluorescent–magnetic nanoprobes. For this purpose, different epithelial, mesenchymal, and CSC cells as well as normal skin HaCaT cells (7.8 × 104 cells/mL) were incubated separately with magneto-PD nanoprobes for 24 h. After that, the cell viability was measured using the MTT test. Figure B clearly shows that even after 24 h of incubation, more than 98% cell viability was observed. We performed the same experiment for magneto-GCD nanoprobes and magneto-CD nanoprobes. We have not observed cytotoxicity from any of our developed fluorescent–magnetic nanoprobes reported here. All the cytotoxicity results clearly show very good biocompatibility for our newly developed fluorescent–magnetic nanoprobes. Next, to understand the photostability of the multifunctional fluorescent–magnetic nanoprobes, we performed time-dependent intensity-change experiments upon exposure to 380 nm light for 1 h. As shown in Figure B, the luminescence signals from fluorescent–magnetic nanoprobes remain almost unchanged (decrease maximum 6%), even after 1 h of illumination. Our reported photostability data clearly show very good photostability of the multifunctional fluorescent–magnetic nanoprobes developed by us.

Targeted Separation and Mapping of Epithelial, Mesenchymal, and Stem Cell CTC from Whole Blood Samples

Next, to find out whether the magnetic nanoplatform can be used for capturing SK-BR-3 epithelial cancer cells, CAL-120 mesenchymal cancer cells, and CD34+ stem cells selectively and simultaneously from whole blood samples, 10 cells/mL tumor cells and 106 cells/mL peripheral blood mononuclear cells (PBMC) were spiked into 15 mL suspensions of citrated whole rabbit blood purchased from Colorado Serum Company. Because in the actual spiked blood samples from patients the CTC coexist with several million peripheral blood mononuclear cells, we spiked 106 cells/mL PBMC with cancer cells in the spiked whole blood kit. The amounts of HER2, twist, or CD34+ present in different cells were measured using the enzyme-linked immunosorbent assay (ELISA). Using ELISA, we found that HER2, twist, or CD34+ are absent in whole rabbit blood or PBMC. For the control experiment, citrated whole rabbit blood was spiked with HaCaT normal skin cells. Using ELISA, we found that HER2, EpCAM, twist, or CD34+ are absent in HaCaT cells. We maintained the concentration of each cell type in the mixture so that, after mixing, the epithelial, mesenchymal, or stem cell CTC concentration is 10 cells/mL in the spiked whole blood sample.

After 90 min of gentle shaking of 10 cells/mL tumor cells, 106 cells/mL PBMC, and 15 mL suspensions of citrated whole rabbit blood mixture, we used the spiked blood for the targeted capturing and imaging experiment. In the next step, anti-CD34 antibody-attached magneto-GCD nanoprobes at different concentrations were mixed with spiked whole blood containing 10 cells/mL tumor cells and 106 cells/mL PBMC for 30 min at room temperature before performing the magnetic separation experiment. After that, targeted cells bound to magneto-GCD nanoprobes were separated using a bar magnet. At the end, targeted CTC captured by magneto-GCD nanoprobes and the amount of CTC in the supernatant after magnetic separation were characterized using an enzyme-linked ELISA kit and fluorescence mapping analysis as shown in Figure . ELISA experimental data as reported in Figure F show that the CD34+ stem cell capture efficiency by anti-CD34 antibody-conjugated magnetic nanoplatforms is more than 98%. Because red fluorescent GCDs are decorated on anti-CD34 antibody-conjugated magneto-GCD nanoprobes, which bind to bone marrow CD34+ stem cells, we used single photon imaging to visualize the capture of bone marrow CD34+ stem cells. Figure D shows the red luminescence image of bone marrow CD34+ stem cells, demonstrating that anti-CD34 antibody-conjugated magneto-GCD nanoprobes can be used for very bright red emission imaging of cancer cells. Figure E shows that the anti-CD34 antibody-conjugated magneto-GCD nanoprobes do not bind to peripheral blood mononuclear cells or rabbit blood cells due to the lack of antigen–antibody interaction, and as a result, we have not observed any luminescence image from the supernatant after magnetic separation. The TEM image reported in Figure C shows that GCDs are captured by stem cells.

All the above-reported results show that anti-CD34 antibody conjugated the marrow CD34+ stem cells and that anti-CD34 antibody-conjugated magneto-GCD nanoprobes can be used to separate and map bone marrow CD34+ stem cells from whole blood samples. To determine the selectivity of the cell capture and mapping for CD34+ stem cells from spiked blood using anti-CD34 antibody-conjugated magneto-GCD nanoprobes, we performed the cell capture and fluorescence mapping experiment using CD34(−) SK-BR-3 breast tumor cells and HaCaT normal skin cells. For this purpose, we used spiked blood containing 4.5 × 105 cells/mL SK-BR-3 tumor cells and spiked blood containing 4.5 × 105 cells/mL HaCaT normal cells separately. Figure H shows that the anti-CD34 antibody-conjugated magneto-GCD nanoprobes do not bind to CD34(−) SK-BR-3 breast tumor cells or HaCaT normal skin cells. As a result, cell capture efficiency was less than 1%. Similarly, we also performed a capture efficiency experiment with anti-HER2 antibody-attached magneto-PD nanoprobes for CD34+ stem cell-spiked blood. As shown in Figure G, our experimental data clearly show that anti-HER2 antibody-attached magneto-PD nanoprobes do not bind to CD34+ stem cells, and as result, capture efficiency was less than 1%. On the other hand, as shown in Figures I and 5A, the capture efficiency is more than 98% for HER2(+) SK-BR-3 cells by anti-HER2 antibody-attached magneto-PD nanoprobes. All the above-reported experimental data clearly show that anti-CD34 antibody-conjugated magneto-GCD nanoprobes are highly selective for capturing and mapping of CD34+ stem cells. Because blue fluorescence PDs are decorated on anti-HER2 antibody-conjugated magneto-PD nanoprobes, which bind to SK-BR-3 breast cancer epithelial cells, as shown in Figure B, we observed very bright blue emission from the SK-BR-3 cancer cells. Figure C shows that the anti-HER2 antibody-attached magneto-PD nanoprobes do not bind to blood cells due to the lack of antigen–antibody interaction, and as a result, we have not observed any luminescence from the supernatant after magnetic separation. All the above-reported results show that anti-HER2 antibody-attached magneto-PD nanoprobes can be used to separate and map epithelial SK-BR-3 cells from whole blood samples.

Figure 5

(A) Percentage of HER2(+) SK-BR-3 epithelial cells that are captured by anti-HER2 antibody-attached magneto-PD nanoprobes when whole blood was spiked with 10 cells/mL SK-BR-3 epithelial cells and 106 cells/mL PBMC. (B) The single-photon luminescence image shows that a huge amount of HER2(+) SK-BR-3 epithelial cells are captured by magneto-PD nanoprobes. The blue color of the observed fluorescence is due to the presence of magneto-fluorescent PD nanoprobes on the cancer cell surface. (C) The single-photon fluorescence image from the supernatant shows no observable fluorescence image, which indicates that blood cells do not bind to anti-HER2 antibody-attached magneto-PD nanoprobes and also that all HER2(+) SK-BR-3 epithelial cells are captured by the magneto-PD nanoprobes. Parts D and E demonstrate the capture of stem cells and mesenchymal cells simultaneously using nanoprobes. (D) ELISA data show the percentage of CD34(+) stem cells and twist(+) CAL-120 mesenchymal cells that are captured simultaneously by anti-CD34 antibody-attached magneto-GCD nanoprobes and antitwist antibody-attached magneto-CD nanoprobes. (E) The fluorescence image shows that antibody-conjugated multicolor fluorescent magneto-nanoprobes are capable of capturing stem and mesenchymal cells simultaneously from the spiked blood. Parts F–H demonstrate the capture of epithelial, stem, and mesenchymal cells simultaneously using nanoplatforms. (F) ELISA data show the percentage of HER2(+) epithelial cells, CD34(+) stem cells, and twist(+) CAL-120 mesenchymal cells captured simultaneously by anti-HER2 antibody-attached magneto-PD nanoprobes, anti-CD34 antibody-attached magneto-GCD nanoprobes, and antitwist antibody-attached magneto-CD nanoprobes. (G) The fluorescence image from the supernatant shows that about all epithelial, stem, and mesenchymal cells are separated by the magnet. (H) The fluorescence image shows that multicolor nanodot-decorated antibody-conjugated nanoprobes are capable of capturing epithelial, stem, and mesenchymal cells simultaneously from spiked blood.

Next, to demonstrate that the versatile multicolor fluorescent– magneto nanoprobes can be used for the capture of mesenchymal and stem cells CTC simultaneously, we performed experiments with mesenchymal and stem cell-spiked blood samples. For this purpose, at first we used whole blood spiked with 10 cells/mL CD34+stem cells, 10 cells/mL twist(+) CAL-120 mesenchymal cells and 106 cells/mL PBMC. For capturing and mapping mesenchymal and stem cells CTC simultaneously, we added anti-HER2 antibody-attached magneto-PD nanoprobes and anti-CD34 antibody-attached magneto-GCD nanoprobes to a 15 mL spiked blood sample. After 30 min of shaking, we captured CTC with a bar magnet. ELISA data, as shown in Figure D, indicate that our multicolor fluorescent–magnetic nanoprobes have the capability to capture multiple types of mesenchymal and stem cell CTC from spiked blood samples and that capturing efficiency can be about 97%.

The multicolor fluorescence image, as shown in Figure E, shows that nanodot-decorated antibody-conjugated multicolor fluorescent– magneto nanoprobes are capable of capturing stem and mesenchymal cells simultaneously from spiked blood. As shown in Figure E, red color fluorescence cells are CD34(+) stem cells and due to the presence of anti-CD34 antibody-attached magneto-GCD nanoprobes on the stem cell surface. Similarly, due to the presence of antitwist antibody-attached CD-coated magneto-CD nanoprobes on the mesenchymal cell surface, green color fluorescence cells are CAL-120 mesenchymal cells. Next, for capturing and mapping epithelial, mesenchymal, and stem cells simultaneously, we added anti-HER2 antibody-attached magneto PD nanoprobes, anti-CD34 antibody-attached magneto-GCD nanoprobes, and antitwist antibody-attached magneto-CD nanoprobes to 15 mL spiked blood samples. For this purpose, we spiked the whole blood sample with 10 cells/mL HER2(+) epithelial cells, 10 cells/mL CD34+stem cells, 10 cells/mL twist(+) CAL-120 mesenchymal cells, and 106 cells/mL PBMC. After 30 min of shaking, we captured CTC subpopulations with a bar magnet. Figure F shows the ELISA data, which clearly indicate that our bioconjugated multicolor fluorescent–magnetic nanoprobes have the capability to capture epithelial, mesenchymal, and stem cell CTC from spiked blood samples and that the capturing efficiency from spiked blood samples can be about 97%.

Figure H shows the multicolor fluorescence image, which indicates that nanodot-decorated antibody-conjugated multicolor fluorescent–magnetic nanoprobes are capable of capturing epithelial, stem, and mesenchymal cells simultaneously from spiked blood. As shown in Figure H, blue color fluorescence cells are HER2(+) SK-BR-3 cells and due to the presence of anti-HER2 antibody-attached magneto-PD nanoprobes on the stem cell surface. On the other hand, red color fluorescence cells are CD34(+) stem cells and due to the presence of anti-CD34 antibody-attached magneto-GCD nanoprobes on the stem cell surface. Similarly, due to the presence of antitwist antibody-attached magneto-CD nanoprobes on the mesenchymal cell surface, green color fluorescence cells are CAL-120 mesenchymal cells. The fluorescence image from supernatant, as shown in Figure G, indicates that almost all epithelial, stem, and mesenchymal cells are captured by the magnet. Also, peripheral blood mononuclear cells and rabbit blood cells do not bind antibody-attached nanodot-coated multicolor fluorescent–magnetic nanoprobes, and as a result, we have not observed any fluorescence image from the supernatant. All the above experimental data clearly show that different antibody-attached multicolor fluorescent–magnetic nanoprobes can be used for capturing epithelial, stem, and mesenchymal cells simultaneously from spiked blood and that they are highly selective for capturing targeted tumor cells from spiked blood.

Conclusions

We have reported the design of bioconjugated multifunctional nanoprobes that exhibit excellent magnetic and multicolor fluorescent properties with targeted capturing and mapping capability for epithelial, mesenchymal, and stem cells simultaneously. We have shown new means of capturing and analyzing epithelial, mesenchymal, and stem cell CTC from spiked blood using multicolor nanodot-attached antibody-coated nanoprobes, which are capable for the characterization of CTC heterogeneity found in clinical samples. We have demonstrated that our nanoprobes are capable of selectively and simultaneously detecting different subpopulation CTC containing SK-BR-3 epithelial, CAL-120 mesenchymal, and bone marrow CD34+ stem cells in spiked whole blood. Our reported data show that the nanoprobes are highly selective for capturing targeted tumor cells and that the capture efficiency can be as high as 97% for epithelial, mesenchymal, and stem cells simultaneously. Reported data demonstrate that multicolor fluorescence imaging can be used for mapping epithelial, mesenchymal, and stem cell CTC simultaneously, which indicates that nanoprobes are capable of characterizing circulating tumor cell heterogeneity by accurately identifying the multiple subpopulations of CTC from blood samples. Although the sensitivity of ELISA for CTC detection is comparable with the reported nanodot-based assay, ELISA had to be coupled to magnetic beads for enrichment of CTC from blood samples, because the concentration of CTC in blood can be as low as 1 per 107 cells whereas in the nanoprobes developed by us, the magnetic nanoparticles enable enrichment, separation, and detection via fluorescence imaging. Because nanoprobes exhibit narrow emission bands, they can be used for simultaneous separation and detection of multiple CTC together, which has been demonstrated here. We anticipate that the nanoprobe design reported here will allow EMT profiling in CTC from clinical samples after proper engineering design. Although we have performed CTC detection on spiked 15 mL whole blood samples containing 10 cells/mL CTC, in clinical settings only 1–10 CTC/mL are present in cancer patient blood. CTC need to be detected in 7.5 mL of whole blood; thus, a better design is necessary to enhance the sensitivity.