Vesicles built from expired blood: Ohio State moves EV loading into a microfluidic chip

Researchers at The Ohio State University reported in Advanced Healthcare Materials that extracellular vesicles assembled from red blood cell lipids by microfluidics could evade immune cells and target cancer cells ^[2]. The reason to pay attention is not the targeting, which is a familiar trick, but the fact that the team stopped harvesting vesicles and started building them.

The group's own account of how it got there is the useful part. They began with the idea of using natural extracellular vesicles produced by red blood cells, ran into limits on scaling up production and on cargo-loading flexibility, and moved to engineering methods instead ^[3]. That is the standard failure mode of the whole EV field: the biology is attractive, the yield and the loading are not. Assembling the particles from purified lipids inverts the problem. Senior author Eduardo Reategui, a professor of chemical and biomolecular engineering, said the lipid composition of the engineered vesicles closely matches that of natural red blood cell vesicles, retaining their biocompatibility ^[4]. He also declined the usual claim of superiority, saying only that the process offers more controllability over the vesicle's composition ^[5].

Controllability shows up most concretely in loading. Because the microfluidic process incorporates therapeutic cargo as the vesicles form, there is no separate post-hoc loading step ^[6], and the cargo range reported spans genetic material, proteins and whole viruses used in gene therapy ^[7]. That includes an adeno-associated virus, the established vector for many gene therapies; the team encapsulated a therapeutic AAV inside an engineered vesicle and reported that the gene therapy still worked and was delivered into cells ^[1]. Reategui said wrapping an AAV in a vesicle tagged with a CD47 peptide could reduce the chance of triggering an immune response ^[15], which is the live commercial problem with AAV, not a theoretical one.

The surface engineering is conventional and works as advertised in the paper. Attaching a CD47 peptide to the outer surface protected the carriers from being taken up by macrophages ^[8]. Adding PD-L1-recognition molecules, including anti-PD-L1 nanobodies from co-author Blaise Kimmel's lab, produced preferential uptake in PD-L1-positive breast cancer tumors ^[9]. In mice, the engineered vesicles stayed in circulation and distributed across multiple organs in patterns similar to natural extracellular vesicles, with notable accumulation in the lungs ^[10]. Lung accumulation is the usual fate of injected nanoparticles, so this reads as a biodistribution result to be managed rather than solved.

The supply chain is the quietly sensible piece. The lipids come from expired red blood cells sourced from the lab of co-author Andre Palmer, which routinely purifies hemoglobin from them for red blood cell substitutes; Palmer notes the units cannot be transfused into patients and would otherwise be discarded ^[11]. A feedstock that is already waste, already collected and already handled under blood-bank rules is a better starting point than a bioreactor of donor cells.

Two things to watch. First, the comparison the researchers themselves invoked: they said these engineered vesicles could function similarly to CAR T-cell therapies made from a patient's own T cells ^[13], and Palmer framed an autologous version using a patient's own red blood cell lipids ^[14]. Second, the numbers. The announcement reports no vesicle yield per unit of blood, no encapsulation efficiency, no dose, and no survival or tumor-response data ^[12]. Until those appear, the manufacturing claim is a design argument rather than a demonstrated cost curve.