Instant labeling of therapeutic cells for multimodality imaging

Abstract

Autologous therapeutic cells are typically harvested and transplanted in one single surgery. This makes it impossible to label them with imaging biomarkers through classical transfection techniques in a laboratory. To solve this problem, we developed a novel microfluidic device, which provides highly efficient labeling of therapeutic cells with imaging biomarkers through mechanoporation. Methods: Studies were performed with a new, custom-designed microfluidic device, which contains ridges, which compress adipose tissue-derived stem cells (ADSCs) during their device passage. Cell relaxation after compression leads to cell volume exchange for convective transfer of nanoparticles and nanoparticle uptake into the cell. ADSCs were passed through the microfluidic device doped with iron oxide nanoparticles and ¹⁸F-fluorodeoxyglucose (FDG). The cellular nanoparticle and radiotracer uptake was evaluated with DAB-Prussian blue, fluorescent microscopy, and inductively coupled plasma spectrometry (ICP). Labeled and unlabeled ADSCs were imaged in vitro as well as ex vivo in pig knee specimen with magnetic resonance imaging (MRI) and positron emission tomography (PET). T₂ relaxation times and radiotracer signal were compared between labeled and unlabeled cell transplants using Student T-test with p<0.05. Results: We report significant labeling of ADSCs with iron oxide nanoparticles and ¹⁸F-FDG within 12+/-3 minutes. Mechanoporation of ADSCs with our microfluidic device led to significant nanoparticle (> 1 pg iron per cell) and ¹⁸F-FDG uptake (61 mBq/cell), with a labeling efficiency of 95%. The labeled ADSCs could be detected with MRI and PET imaging technologies: Nanoparticle labeled ADSC demonstrated significantly shorter T₂ relaxation times (24.2±2.1 ms) compared to unlabeled cells (79.6±0.8 ms) on MRI (p<0.05) and ¹⁸F-FDG labeled ADSC showed significantly higher radiotracer uptake (614.3 ± 9.5 Bq / 1×10⁴ cells) compared to controls (0.0 ± 0.0 Bq/ 1×10⁴ cells) on gamma counting (p<0.05). After implantation of dual-labeled ADSCs into pig knee specimen, the labeled ADSCs revealed significantly shorter T₂ relaxation times (41±0.6 ms) compared to unlabeled controls (90±1.8 ms) (p<0.05). Conclusion: The labeling of therapeutic cells with our new microfluidic device does not require any chemical intervention, therefore it is broadly and immediately clinically applicable. Cellular labeling using mechanoporation can improve our understanding of in vivo biodistributions of therapeutic cells and ultimately improve long-term outcomes of therapeutic cell transplants.

Keywords: 18F-FDG; in vivo cell tracking; iron oxide nanoparticles; mechanoporation; microfluidic device.

Figure 1

Concept of instant ADSC harvest and labeling with imaging biomarkers. (A) Therapeutic cells are harvested from the pre-patellar fat pad. (B) ADSCs, adipocytes and monocytes are isolated through collagenase digestion and centrifugation. (C) The harvested cells are passaged through a novel microfluidic device, which provides instant labeling through cell compression and convective transfer of imaging biomarkers. (D) The labeled cells are seeded in scaffold. (E) The labeled cells in scaffold are implanted into cartilage defects. The engraftment of the labeled cells can be tracked in vivo with clinical imaging technologies.

Figure 2

Microfluidic device for instant cell labeling with imaging biomarkers. (A) Setup of a prototype microfluidic device with one inlet and one outlet, (B) Phase contrast image (corresponds to black box in E) of one channel of the microfluidic device with multiple ridges (C) Schematic cross-sectional view of therapeutic cells undergoing repeated compression under the ridges and expansion/relaxation after passing the ridges. The relaxation state leads to a transient increase in cell membrane permeability with convective uptake of nanoparticles and radiotracers. (D) Image of the silicon wafer used to fabricate the microfluidic devices. Image of (E) the 5-channel and (F) 1-channel microfluidic device mold for large and small cell samples.

Figure 3

Instant labeling of ADSCs with ferumoxytol nanoparticles using the novel microfluidic device. (A) Fluorescence microscopy of ADSCs before (unlabeled) and after (labeled) processing through the microfluidic device doped with FITC-conjugated ferumoxytol (10 mg Fe/ml). Labeled ADSCs demonstrate green fluorescent signal in the cytoplasm, confirming accumulation of ferumoxytol nanoparticle uptake (blue represents DAPI, red represents a phalloidin probe which selectively stains F-actin cytoskeleton and green represents FITC-conjugated ferumoxytol). (B) DAB-enhanced Prussian blue staining shows brown precipitation (arrow heads) in the cytoplasm of ADSCs, consistent with cellular accumulation of ferumoxytol nanoparticles. (C) Flow cytometry analysis of ADSCs demonstrates more than 95% labeling efficiency. (D) Iron content of ferumoxytol labeled cells, as determined by inductively coupled plasma optical emission spectroscopy shows significantly higher cellular iron uptake of ferumoxytol labeled cells compared to unlabeled controls (p < 0.0001). (E) Trypan blue assay demonstrated no significant difference in cell viability of ferumoxytol labeled cells compared to unlabeled controls. All quantitative data represent mean data ± SEM of three cell samples per experimental group; p < 0.05; unpaired t test.

Figure 4

Microfluidic device assisted cell labeling enables cell detection with PET and MR imaging. ADSCs were labeled with ferumoxytol nanoparticles and a separate set of cells was labeled with ¹⁸F-FDG, using microfluidic device-assisted mechanoporation. The cells were transferred to test tubes and scanned with clinical MRI and PET: (A) MRI shows marked hypointense (dark) T₂-signal of 1×10⁶ ferumoxytol labeled ADSCs compared to unlabeled cells. (B) Corresponding quantitative data confirmed significant shortening of T₂ relaxation times of labeled ADSCs compared to unlabeled ADSCs (p < 0.0001). (C) Viability of ferumoxytol-labeled and unlabeled cells, measured with a Cell Counting Kit-8 (CCK-8) assay was not significantly different. (D) PET imaging demonstrates high PET signal of ¹⁸F-FDG-labeled ADSCs (5×10⁴) compared to unlabeled cells. (E) ¹⁸F-FDG uptake quantification by gamma counting confirms significantly higher radioactivity in ¹⁸F-FDG-labeled ADSCs compared to unlabeled ADSCs (p < 0.0001). (F) No significant difference in viability is observed between ¹⁸F-FDG-labeled and unlabeled cells. All data are shown as mean ± SEM of three cell samples per experimental group; p < 0.05; unpaired t test.

Figure 5

PET/MR imaging of dual labeled ADSCs in cartilage defects of pig knees. ADSCs were labeled with ferumoxytol nanoparticles and ¹⁸F-FDG, using microfluidic device-assisted mechanoporation, and implanted into cartilage defects of the distal femur of pig knees. (A) MRI enables detection of a representative nanoparticle labeled ADSC implant based on a marked hypointense (dark) signal effect on a T₂-weighted scan (blue arrows). By comparison, unlabeled ADSCs are difficult to delineate from adjacent cartilage (white arrows). (B) T₂ map demonstrates T₂ relaxation times on a color scale, with marked T₂ relaxation time shortening of the ferumoxytol labeled implant (blue arrows) compared to the unlabeled implant (white arrows). (C) Corresponding quantitative data confirmed significant shortening of T₂ relaxation times of labeled implants compared to unlabeled implants (p < 0.0001). (D) PET image of the same knee joint as in (A) shows marked PET signal of the dual labeled ADSC implant (blue arrows), while the unlabeled implant shows no detectable PET signal (white arrows). (E) Fused PET/MR images show colocalization of the MRI and PET signals in the labeled implant (blue arrows). (F) Quantification of the PET signal confirms significantly higher radioactivity (15.8 MBq/ml) of ¹⁸F-FDG labeled implants compared to unlabeled implants (0.1 MBq/ml; p < 0.0001). All data are shown as mean ± SEM of four implants per group; p < 0.05; unpaired t test.