Red blood cell-hitchhiking boosts delivery of nanocarriers to chosen organs by orders of magnitude

Abstract

Drug delivery by nanocarriers (NCs) has long been stymied by dominant liver uptake and limited target organ deposition, even when NCs are targeted using affinity moieties. Here we report a universal solution: red blood cell (RBC)-hitchhiking (RH), in which NCs adsorbed onto the RBCs transfer from RBCs to the first organ downstream of the intravascular injection. RH improves delivery for a wide range of NCs and even viral vectors. For example, RH injected intravenously increases liposome uptake in the first downstream organ, lungs, by ~40-fold compared with free NCs. Intra-carotid artery injection of RH NCs delivers >10% of the injected NC dose to the brain, ~10× higher than that achieved with affinity moieties. Further, RH works in mice, pigs, and ex vivo human lungs without causing RBC or end-organ toxicities. Thus, RH is a clinically translatable platform technology poised to augment drug delivery in acute lung disease, stroke, and several other diseases.

Fig. 1

Clinically translatable nanocarriers adsorb onto red blood cells. a Procedural steps of RBC hitchhiking. NCs are first adsorbed onto the RBCs ex vivo. The RBC–NC complexes are then injected via an intravascular catheter, after which the NCs transfer from the RBCs to the first downstream organ’s capillaries. b Scanning electron micrographs of PS-NPs and nanogels attached to the surface of murine RBCs. NCs were mixed with RBCs in vitro, leading to adsorption of NCs onto the RBCs. Scale bars = 1 μm. c Efficiency of radiolabeled NC adsorption onto RBCs, as defined by the % of total NC added to RBCs that pellet with RBCs. d Number of NCs adsorbed per RBC when NCs were mixed with RBCs at a ratio of 2000 NCs per RBC. e Adsorption efficiencies onto RBCs of free proteins (IgG and BSA) compared to free nanogels and to nanogels coated with each protein. f Adsorption efficiencies of unmodified and IgG-coated NCs with increasing concentrations of serum present in the buffer during adsorption. Each data point represents mean ± s.e.m (n = 3). *P < 0.05, non-paired, two-tailed t-test

Fig. 2

IV injection of optimized RH NCs massively augments lung delivery without affinity moieties. a Mice were injected IV with polysytrene nanoparticles (PS-NPs) that were covalently coated with either IgG (dark blue) or anti-PECAM antibodies (light blue) and radiolabeled with a trace amount of I-125-IgG. Separate mice were injected with one of those two PS-NPs adsorbed onto RBCs (dark and light red, respectively). Mice were sacrificed 30 min later and I-125 activity in the organs was measured in a gamma counter. Displayed is the percent injected I-125 dose (%ID) for each organ. Inset: lung-to-liver ratios, which are calculated by dividing the lung’s %ID per gram of tissue (%ID/g) by the liver’s %ID/g. b %ID in the lung of different NCs labeled and injected as in a. c Lung-to-liver (left panel) and lung-to-blood (right panel) ratios of liposomes and nanogels, the two top performing lung-directed RH NCs. d Mice were injected with nanogels that were either uncoated (bare) or covalently coated with one of the three different antibodies (random rat IgG, anti-PECAM, or anti-ICAM) (blue bars). Other mice were injected with each of those four antibody–nanogel formulations adsorbed onto RBCs (red bars). Data are plotted as %ID per organ. Each data point represents mean ± s.e.m (n = 3). *P < 0.05, non-paired, two-tailed t-test

Fig. 3

RH NCs are taken up by endothelial cells and leukocytes of target organ capillaries. a Mice were IV injected with RH rhodamine-conjugated nanogels (NGs), sacrificed 30 min later, and then the lungs were fixed and sectioned. The sections were stained with an endothelial marker (VE-cadherin, blue) and a leukocyte marker (NIMP, green). Rhodamine-NG fluorescence localized to the capillary endothelial cells with small amounts of nanogel signal colocalizing with sparse leukocytes (top row 10×, bottom row 40×). Thick white lines represent scale bars of 100 µm. b Mice were given intratracheal LPS to model ARDS prior to RH NCs injection as in a. The top left panel is a 10× image, while the top right is a 40× magnification plus digital zoom of the same region. As seen in the top left image, red nanogels are present in an overlapping distribution with both endothelial cells (blue) and leukocytes (green). The two bottom panels are 40× images of another region of the tissue, but displaying only two markers each for ease of viewing. Thick white lines represent scale bars of 100 µm. c Macrophages were plated in flow chambers and NGs (labeled red) were introduced, either free or adsorbed onto RBCs (labeled green). As seen in two separate experiments using RH NGs (top two panels), the RH NGs (red) localize to the center of the macrophages, and the RBCs (green) transiently localize on the periphery of the macrophages. In two other experiments using free NGs (bottom two panels), very little NG signal localizes with the macrophages. Scale bars represent 20 μm. In the rightmost panel, we have quantified the gain in macrophage-co-localized red fluorescence (corresponding to nanogels) during the course (10 min) of each of these experiments. Each data point represents mean ± s.e.m (n = 4 for free NG, n = 12 for RH NGs); *P < 0.05, non-paired, two-tailed t-test

Fig. 4

Optimized NCs do not cause toxicity during RH. a Mouse and human RBCs were mixed either with an agglutinating cross-species serum (dog serum) or NCs and then prepared as a “thin smear” slide. Dog serum causes RBCs to aggregate, as do bare PS-NPs, but nanogels do not. Scale bar represents 30 μm. b Round-bottom well assay of RBC aggregation, in which aggregated RBCs form a diffuse haze, while non-aggregated RBCs settle into a tight red dot. c RH nanogels, with the RBCs labeled with Cr-51, were prepared as in Fig. 2 and injected into the mice, followed by sacrifice after 30 min and organ Cr-51 signal measurement on a gamma counter. As a positive control, RBCs were intentionally aggregated with the anti-RBC antibody, Ter119, plus a cross-linking secondary antibody. Each data point represents mean ± s.e.m (n = 3). d, e RH nanogels and Ter119-aggregated RBCs were prepared as in c, but before sacrifice, the pulmonary artery pressure (PAP) was measured. Five min after injection, Ter119-aggregated RBCs (positive control) had increased the PAP, while RH nanogels had not (quantification in e). Each data point represents mean ± s.e.m (n = 4). f, g Mice were injected as in c and then had their blood oxygen measured over time (f), followed by their sacrifice at 30 min for histology (g), showing no difference between RH and RBCs-only. Scale bar represents 100 μm. Each data point represents mean ± s.e.m (n = 4). h Mice were injected with RH or free nanogels and intratracheally instilled with LPS, followed by measurement of the bronchoalveolar lavage (BAL) levels of leukocytes and protein, both measures of lung inflammation and ARDS. Each data point represents mean ± s.e.m (n = 3). *P < 0.01, non-paired, two-tailed t-test

Fig. 5

RH works in large animals and ex vivo human lungs. a Adsorption efficiency of nanogels onto various species’ RBCs. Each data point represents mean ± s.e.m (n = 3 replicates per condition). b Live pigs were injected with I-125-labeled nanogels that were either free or RH. Plotted are the lung-to-liver and lung-to-blood ratios. Each data point represents mean ± s.e.m (n = 3 pigs per condition). c Safety in pigs was assessed by showing that Cr51-labeled RBCs used in the RH experiments of b did not get stuck in the lungs. d, e Ex vivo human lungs were endovascularly cannulated and then infused through a single artery sequentially with: I-125-labeled RH nanogels; I-131-labeled free nanogels; and a green tissue dye. The percent of the injected dose of the entire lung lobe (e, left panel) and within only the well-perfused (green) zone (e, right panel, where ten 2-mL volumes of lung were measured and their mean ± s.e.m. listed on the plot). *P < 0.05, two-tailed t-test

Fig. 6

RH effectively delivers therapeutic cargo to ameliorate a model of pulmonary embolism. a Inset, schematic of nanogels (NGs) with thrombolytic enzyme reteplase (gold) conjugated to the dextran (blue) shell. The bar graph shows the degree to which reteplase-NGs, free reteplase, or saline dissolve preformed fibrin clots in vitro. Reteplase conditions contained 4 nM of reteplase, with reteplase concentration in reteplase-NGs quantified by reteplase-NG coupling efficiency. b Biodistribution of RH reteplase-NGs at 1 h after injection, showing high lung uptake. c All mice were intravenously injected with ~1–5 micron diameter fibrin clot emboli (I-125-labeled). Mice were also treated with either saline, free reteplase-NGs, or RH reteplase-NGs at 12 micrograms reteplase per mouse. The amount of I-125-labeled fibrin/fibrinogen remaining in each organ at 1 h post injection is quantified. d Degree to which initially injected emboli were cleared from the lungs. For all plots in this figure, each data point represents mean ± s.e.m (n = 4). *P < 0.001, non-paired, two-tailed t-test

Fig. 7

IA injection of RH NCs results in high NC uptake in any downstream organ. a Mice were anesthetized and had their right internal carotid artery cannulated to deliver the injectate to the brain (inset). Either free (blue) or RH (red) I-125-labeled NCs (here, nanogels (NGs)) were injected. The left panel depicts % injected dose (%ID) in the brain, while the right panel depicts brain/blood ratio. b Mice injected as in a, followed by sectioning and autoradiography of the brain. Scale bar represents 5 mm. c Mice were treated as in a, but the injectate was RH rhodamine-labeled NGs. Left panel is rhodamine signal, with more red puncta in the right (targeted) hemisphere. Right panel is rhodamine signal overlaid with DAPI signal. Scale bar represents 100 μm. d To demonstrate no gross anatomical damage done to the neurons by RH NGs, brain slices from c were H&E stained, showing no difference in morphology between the right and left hemispheres. Scale bar represents 100 μm. e Mice were treated as in a, but the arterial cannula was in the left kidney. f Mice were treated as in a, but the arterial cannula was in the common carotid artery, and the right and left sides of the “face” were harvested. Here, a hemi-“face” includes all structures on one side of the head, except for the brain. For all plots in this figure, each data point represents mean ± s.e.m (n = 3). *P < 0.05, non-paired, two-tailed t-test