Introduction

Bone marrow derived human mesenchymal stem cells (hMSC) have been assessed for their therapeutic efficacy in a range of pathologies, including cancer, Alzheimer’s disease and stroke. Trophic and immunomodulatory effects, as well as their role in cell-to-cell communication, support hMSC as promising candidates for cell therapy.¹ However, direct hMSC implantation compared to the impact of their secretions alone on therapeutic outcome, either locally at the lesion or by recruiting regenerative progenitor cells, has remained elusive. Of particular interest for potential therapeutics are extracellular vesicles (EV), which--as messengers of the hMSC secrotome--carry immunomodulatory and regulatory secretions that impact synaptic transmission, neuronal differentiation, angiogenesis and neuronal projection.² Similar to cellular therapy, in vivo tracking of EV is imperative to determining biodistribution. A novel EV labeling method with ultra-small superparamagnetic iron oxide (USPIO) contrast was used here to confirmed localization to the ischemic region. Tissue recovery metrics were assessed via ultra-high field (21.1 T) sodium (²³Na) chemical shift imaging (CSI) and ¹H relaxation-enhanced MR spectroscopy (RE-MRS), providing insight into ionic and metabolic homeostasis in response to the delivery of hMSC versus their derived EV in a cerebral ischemic stroke model.

Methods

Cell Culture and EV Processing: hMSC acquired from Tulane Center for Stem Cell Research and Regenerative Medicine were characterized based on plastic adherence and surface markers.³ EV were isolated and purified from hMSC following one week of hypoxic preconditioning; EV underwent robust characterization to ensure purity.⁴ Following EV harvest, hypoxic hMSC were exposed to 7.47-µg Fe/mL micron-sized particles of iron oxide (MPIO, Bangs Laboratory, Inc.) prior to injection. To accommodate the smaller size, EV underwent labeling with 0.5-mg/mL ultra-small super-paramagnetic iron oxide (USPIO, Sigma Aldrich) via sonication and purified utilizing a novel polyethylene glycol (PEG) enrichment method outlined in Figure 1. Both therapies were re-suspended in 50-μL saline for in vivo injection.
Animal Model: A transient middle cerebral artery occlusion model⁵ was instituted in male Sprague-Dawley rats (200-250 g) for 1 h to induce striatal ischemia, followed by immediate arterial injection of hMSC-derived therapy (cells or EV) or saline only. Neuromotor assessment was conducted concurrently with MR scanning to 21-d post-ischemia.
MR Acquisitions: Data were acquired using the 21.1-T, 900-MHz vertical magnet at the NHMFL equipped with a linear birdcage dual-tuned ²³Na/¹H radio frequency coil. Therapy administration was confirmed for both EV (1-4 h) and hMSC (1 d) with gradient recalled echo images (50x50-µm in-plane resolution). ²³Na 3D CSI was acquired at 1-mm isotropic resolution. RE-MRS evaluated metabolites in both ischemic and contralateral hemispheres. Selective bandwidth excitation pulses targeted lactate, creatine, choline and N-acetyl aspartate, while avoiding water, with spatial selectivity imparted by adiabatic selective refocusing (LASER) pulses.⁶ T₂-weighted images provided anatomical reference to the ischemic lesion and contralateral alignment.
Analysis: EV biodistribution and clearance were assessed in ParaVision 5.0 (Bruker Biospin, Inc). CSI data reconstructed in MATLAB were zero-filled to 0.5-mm isotropic resolution for volumetric and signal analysis in Amira 3D Visualization Software (Thermo Fisher Scientific, Inc). A signal threshold generated from the contralateral hemisphere was used to define the ischemic lesion. Peaks for RE-MRS data were assigned according to previous literature (NAA 2.0 ppm, Lac 1.31 ppm, Cre 3.0 ppm, Cho 3.2 ppm), referencing water at 4.7 ppm.⁷ Metabolite ratios to choline were used to track metabolic levels longitudinally. Statistical significance was calculated according to mixed-model effects with p < 0.05, and all graphs are mean ± SD.

Results

MRI confirmed presence of USPIO-labeled EV in the ischemic region immediately following MCAO and therapy administration, with complete clearance by 24 h as expected (Figure 2). hMSC remained in cerebral tissue with total clearance by 7 d.
²³Na 3D CSI demonstrated that hMSC- and EV-injected animals exhibited significant ischemic lesion reduction by 7 d (Figure 3). In contrast, control ischemic animals first exhibited significantly elevated lesion volume on day 3 followed by reduction only at 21 d. As demonstrated by ²³Na signal reductions, both hMSC and EV evoked therapeutic response, illustrated by maintenance of significantly lower sodium levels in both the ischemic and contralateral hemispheres compared to control animals. Furthermore, reduction in lactate levels closely paralleled improved sodium response in hMSC and EV groups (Figure 4). Functional recovery assessed via several neuromotor tests as well as weight (Figure 5) reinforce the therapeutic efficacy generated by hMSC-derived EV.

Discussion

For the first time, USPIO labeling of EV generated contrast in cerebral tissue immediately after administration, following a novel sonication and PEG-based enrichment protocol. EV did not exhibit changes in size distribution profile following this labeling procedure. Previous studies have demonstrated ²³Na CSI and ¹H RE-MRS are highly sensitive biomarkers of therapy efficacy and have been able to differentiate successful from ineffective therapies as early as 1 d post-injection, hence are suitable metrics to be used here to determine therapeutic efficacy of EV. Ischemic lesion volume and signal changes from ²³Na CSI as well as metabolic alterations show that hypoxic hMSC-derived EV evoke a therapeutic response similar though perhaps more muted than their cellular counterparts.

Conclusion

Efficacy of hMSC-derived extracellular vesicles labeled with USPIO localize to the ischemic legion and evoke a therapeutic response after a single bolus injection.