Fig. 1. TEM images of 9nm Fe₂O₃ nanoparticles before (A) and after (B) deposition of the metallic gold layer. Images were collected at 80keV acceleration and 180,000× direct magnification. Scale bars are 200nm. In (C), UV-Vis extinction spectra are shown of a suspension of 9nm Fe₂O₃ nanoparticles (blue) and the same suspension after deposition of ca. 20nm gold layer (pink). Bare Fe₂O₃ particles show typical inverse-power law type extinction properties. The addition of gold to the surface results in the appearance of a characteristic plasmon resonance peak at ca. 540nm. In (D), theoretical simulations show the relative contribution of absorption (pink) and scattering (blue) to the total extinction (black) of the nanoparticles. The calculated maximum wavelengths for absorption, extinction, and scattering are 534nm, 537nm, and 550nm, respectively and include the effect of particle size distribution as determined by TEM.

Fig. 2. Experimental setup. Cells were mounted on a microscope slide and imaged in reflected darkfield mode using a Leica DM 6000 upright microscope. A solenoid electromagnet with a cone-shaped ferrite core was attached to a programmable piezoelectric translation stage, and placed beneath the sample stage. Translation stage motion oscillated in a sinusoidal fashion with a user-definable amplitude and frequency. The magnet was powered by a power supply and amplifier delivering up to 960 W. The solenoid and motorized translation stage assembly was mechanically isolated from the microscope, which sat on a vibration isolation table.

Fig. 3. Darkfield images of a 1:1:1 mixture of A-431 cells labeled with 40nm anti-EGFR gold nanoparticles (indicated by green arrows), 50nm anti-EGFR gold/iron oxide nanoparticles (indicated by red arrows), and unlabeled cells (indicated by blue arrows) obtained using: (A) white light illumination; and (B) a 630±15nm bandpass filter. Images were acquired with a 20x darkfield/brightfield objective with a 0.5 collection NA. (C) Scattering spectra of cells labeled with 50 nm hybrid nanoparticles (red line), 40nm pure gold nanoparticles (green line) and of unlabeled cells (blue line). A fluorescent tag (AlexaFluor 488, Molecular Probes) was attached to the pure gold-antibody conjugates in order to differentiate between the two types of labeled cells.

Fig. 4. AVI file showing the magnetically induced movement of gold-iron oxide nanoparticle labeled A-431 cells captured under 20× magnification. Note no movement of unlabeled cells (which appear dim) and pure-gold labeled cells (identified by the overlaid green fluorescence signal); the magnetically labeled cell clearly responds to the oscillating magnetic field by fluctuating laterally in the horizontal direction, and also by translating in the vertical direction toward the ferrite tip which is located outside and above the field of view shown in this figure. The cell also shows some “rocking” movement which is a result of variations in the magnetic force in z-direction with movement of the solenoid. Because the cell has unevenly distributed magnetic gold nanoparticles it creates a net torque that forces cell to “rock”. Images were captured and replayed at 10 frames per second. The movie shows approximately 4 oscillations of the solenoid. [Media 1]

Fig. 5. Monochrome images of a 1:1:1 mixture of A-431 cells labeled with 40nm anti-EGFR gold nanoparticles (green arrows), 50nm anti-EGFR gold/iron oxide nanoparticles (red arrows), and unlabeled cells (blue arrows) that were magnetically actuated at 0.9Hz (A) and 1.9Hz (B) before application of a digital frequency filter. The images were obtained using a 635/15nm bandpass filter. Sections (C) and (D) show power spectra that are taken from the time-domain Fourier transform in the region containing a cell labeled with 50nm gold/iron oxide particles (red), 40nm pure gold particles (green) and an unlabeled cell (blue). Note the prominent peaks in the magnetically-labeled cells’ frequency spectra that correspond to the translation stage oscillation frequencies of 0.9Hz (C) and 1.9Hz (D). Sections (E) and (F) show the same fields of view as sections (A) and (B), respectively, after digital filtering at 0.9Hz (E) and 1.9Hz (F) uses the Hanning function implementation. Only magnetically labeled cells are visible.

Fig. 6. In (A), pixel intensity profiles are shown for the three cell types: 50nm gold/iron oxide labeled (red line), 40nm pure gold labeled (green line), and unlabeled (blue line). Profiles are drawn for the same three cells captured using white light illumination, 635/15nm bandpass illumination, as well as bandpass plus magnetic actuation and digital frequency filtering. In (B) and (C), the relative average pixel intensity from n>10 cells in each cell population and illumination/acquisition condition is compared. Asterisks and brackets in (B) and (C) indicate a statistical significant difference of the average signal values with p<10^-4.