key: cord-0265620-v6basbeb
authors: Khare, Purva; Conway, James F.; Manickam, Devika S
title: Lipidoid nanoparticles increase ATP uptake into hypoxic brain endothelial cells
date: 2022-04-09
journal: bioRxiv
DOI: 10.1101/2022.04.07.487513
sha: eba88b64d553880784f707fa861b86bcabf71f3e
doc_id: 265620
cord_uid: v6basbeb

Lipidoid nanoparticles (LNPs) are clinically successful carriers for nucleic acid delivery to liver and muscle targets. Their ability to load and deliver small molecule drugs has not been reported yet. We propose that the delivery of adenosine triphosphate (ATP) to brain endothelial cells (BECs) lining the blood-brain barrier may increase cellular energetics of the injured BECs. We formulated and studied the physicochemical characteristics of ATP-loaded LNPs using the C12-200 ionizable cationic lipid and other helper lipids. Polyethylene glycol-dimyristoyl glycerol (PEG-DMG), one of the helper lipids, played a crucial role in maintaining colloidal stability of LNPs over time whereas the inclusion of both ATP and PEG-DMG maintained the colloidal stability of LNPs in the presence of serum proteins. ATP-LNPs formulated with PEG-DMG resulted in a 7.7- and 6.6-fold increased uptake of ATP into normoxic and hypoxic BECs, respectively. Altogether, our results demonstrate the potential of LNPs as a novel carrier for the delivery of small molecular mass actives to BECs—a CNS target. Highlights LNPs were formulated with ATP, a small molecule drug PEG-DMG plays a critical role in maintaining particle stability over tim ATP and PEG-DMG play a critical role in maintaining particle stability in 10% serum ATP-LNPs were internalized by normoxic and hypoxic brain endothelial cells (BECs) LNP delivery to BECs broadens its applicability to CNS targets Graphical Abstract

Introduction ratio of 50/10/38.5/1.5 in ethanol. The representative formulation scheme for the preparation of 1 3 9 ATP-and siRNA-loaded LNPs are shown in Table 1 and Supplementary table 1, respectively.  1  4  0 For ATP-LNPs, a 33.73 mg/mL solution of ATP was prepared in 1x PBS pH 7.4. We have 1 4 1 previously studied the effects of mixing speeds on drug loading by comparing 'slow mixing' and siRNA concentration of 50 nM using 1x PBS (pH 7.4). Particle diameters (a and b) and zeta 2 8 0 potentials (c) were measured in 1x PBS and de-ionized water, respectively, using a Malvern 2 8 1

Zetasizer Pro. Data are represented as mean ± SD of at least n=3 measurements. Negative-stain 2 8 2 electron micrographs of ATP-LNPs (d) and siRNA-LNPs (e) were acquired using a Thermofisher 2 8 3 TF20 electron microscope. Tukey's and Šídák's multiple comparisons tests were used for 2 8 4 comparative analysis using two-way ANOVA. *p<0.05, **p<0.01, ***p < 0.001, ****p < 2 8 5 0.0001 and ns: non-significant. Physicochemical characteristics of nanoparticles govern their colloidal stability and in vivo 2 8 8 biological efficacy (7, 41). Particle diameter and surface charge (zeta potential) are crucial 2 8 9 determinants of colloidal stability. Cellular uptake of ionizable cationic lipid-based LNPs is 2 9 0 largely dependent on their particle diameter homogeneity (42, 43). Colloidal particles like LNPs 2 9 1 are known to aggregate resulting in larger particle diameters and polydisperse samples over time 2 9 2 (44). We have previously demonstrated that siRNA loading further stabilized LNP particle 2 9 3 diameters against aggregation (39). PEG-DMG acts as a steric stabilizer and is known to 2 9 4 stabilize nanoparticles by hindering particle-particle aggregation resulting in monodisperse, 2 9 5 smaller particle diameters (45, 46). Therefore, we compared the colloidal stability of blank and 2 9 6 ATP-LNPs prepared in the presence or absence of PEG-DMG (+/-PEG-DMG) over a period of 2 9 7 seven days upon interim storage at 2-8 o C. We also measured LNP surface charge by measuring 2 9 8 their zeta potential following the same storage regime. Blank LNPs (+PEG-DMG) showed a particle diameter of ~78 nm that increased to ~167 nm 3 0 1 whereas blank LNPs (-PEG-DMG) had a particle diameter of ~895 nm that increased to ~2 μ m 3 0 2 over a period of seven days (Figure 1a) . ATP-LNPs (+PEG-DMG) had an initial particle 3 0 3 diameter of ~83 nm that increased to ~162 nm whereas ATP-LNPs (-PEG-DMG) showed a 3 0 4 particle diameter of ~995 nm post-preparation that increased to ~3 μ m after seven days ( Figure  3 0 5 1a). The increase in particle diameters among each group over a period of seven days was found 3 0 6 to be non-significant using two-way ANOVA. However, the particle diameters of blank and 3 0 7 ATP-LNPs prepared without PEG-DMG (-PEG-DMG) were significantly greater than those 3 0 8 prepared with PEG-DMG (+PEG-DMG) (p < 0.0001 for blank LNPs and p < 0.001 for ATP-3 0 9

LNPs) (Figure 1a) . We concluded that the inclusion of ATP did not have an additional 3 1 0 stabilizing effect on the LNP particle diameters as both ATP-loaded and blank LNPs showed 3 1 1 similar initial diameters and as well as seven days post-storage. While ATP loading or the lack of 3 1 2 it did not have an effect on particle diameters, we have previously demonstrated that the 3 1 3 inclusion of siRNA, an anionic macromolecule, resulted in a dramatic stabilizing effect (lower 3 1 4 sizes of 120 nm) as compared to blank LNPs (250 nm) (p<0.0001) (39). The observed 3 1 5 differences in siRNA-vs. ATP-loaded LNPs can be attributed to the polyionic nature of siRNA 3 1 6

vs. ATP and also due to the larger molecular mass of siRNA in comparison to the small 3 1 7 molecule, ATP (13,300 vs. 507.18 g/mol). We found a similar trend for the dispersity indices where the LNPs with PEG-DMG resulted 3 2 0 in lower dispersity indices as compared to LNPs without PEG-DMG. The dispersity index for 3 2 1 blank LNPs (+PEG-DMG) was found to be 0.13 that increased to 0.17 whereas the dispersity 3 2 2 index for blank LNPs (-PEG-DMG) was 0.56 that increased to a high value of 1.3 seven days We also studied the surface charge on the LNPs by measuring their zeta potential at a pH of freshly prepared LNPs and LNPs stored at 37 °C for 24 h. The average particle diameters of 3 5 0 blank and ATP-LNPs (+PEG-DMG) were 166 nm and 99 nm at 0 h, respectively, and 128 nm 3 5 1 observed a bimodal peak for the serum proteins similar to previous reports (52-55). We looked 3 9 6

for the presence of distinct peaks in the intensity distribution plots corresponding to those of 10% 3 9 7 FBS (bimodal) and the LNPs. If the serum proteins destabilize and result in LNP aggregation, we 3 9 8 expected the LNP peak to shift to larger diameters. If the LNPs are not able to maintain their 3 9 9

intact structure in the presence of serum proteins, it is likely that the relative peak intensity of 4 0 0

LNPs would change/decrease. Next, we examined whether ATP loading in the LNPs may affect particle stability in the 4 1 5 presence of serum proteins. ATP-LNPs (+PEG-DMG) in PBS showed a distinct peak at 146 nm 4 1 6 that shifted to 169 nm when the particles were supplemented with FBS at 0 h (Figure 3a) . Post-4 1 7 24 h, we observed a sharp peak of LNPs in FBS at 198 nm that was comparable to that of LNPs 4 1 8 in PBS with a Z avg of 126 nm (Figure 3b) . The shift in the diameter of the blank LNP (+PEG-4 1 9 DMG) counterpart was more drastic along with a shift in the LNP peak as noted post-24 h in the 4 2 0 presence of FBS when compared to ATP-LNPs (+PEG-DMG) (Figure 3a and b) . This suggests 4 2 1 that ATP loading plays a notable role in maintaining the colloidal stability of LNPs in the 4 2 2 presence of serum proteins. showed a peak at 267 nm and LNPs supplemented with FBS retained the peak albeit at a lower 4 2 8 intensity and a Z avg of 489 nm (Figure 3b ). To summarize, although blank LNPs (+PEG-DMG) 4 2 9

show a peak in the presence of FBS at 0 h, the peak was not retained at the end of 24 h 4 3 0 incubation (Figure 3a and b) . In the case of blank LNPs (-PEG-DMG), we did not observe a 4 3 1 peak at 0 h but post-24 h we noted a peak with a lower intensity. With respect to ATP-LNPs 4 3 2 (+PEG-DMG) in FBS, we observed a sharp peak at the 0 h and 24 h time points with comparable 4 3 3 Z avg values indicating the combined stabilizing effects of ATP loading and PEG-DMG in the 4 3 4 formulation (Figures 3a and b) . With ATP-LNPs (-PEG-DMG) we observed a distinct peak 4 3 5 albeit at a lower intensity immediately post-preparation as well as 24 h after incubation ( Figures  4  3  6   3a and b) . Overall, our results demonstrate that both ATP loading and PEG-DMG are crucial for 4 3 7 maintaining the serum stability of LNPs. Interestingly, ATP loading has a more pronounced During stroke, BECs suffer oxygen-glucose deprivation (OGD) due to decreased blood 4 5 4 supply that affects their mitochondrial function leading to ATP depletion. We wanted to 4 5 5 determine whether the ATP-loaded LNPs can be internalized into BECs. We used flow 4 5 6 cytometry to quantify the cellular uptake of LNPs loaded with Alexa Fluor 647 (AF647)-tagged 4 5 7 ATP. We spiked unlabeled ATP with AF647-ATP at a ratio of 100:1 prior to LNP preparation. 4 5 8

For this reason, we express cellular uptake as a -fold increase in uptake over untreated cells 4 5 9

instead of an absolute percentage (%) uptake-it should be noted that AF647-ATP was used to 4 6 0 spike the LNPs and the LNPs also contain unlabeled ATP. We incubated the cells with samples 4 6 1 containing either ATP-LNPs prepared w+/-PEG-DMG or free ATP or Lipofectamine-ATP 4 6 2 complexes. We used lipofectamine as a positive control as this cationic lipid is widely used for 4 6 3 the delivery of anionic polynucleotides such as siRNA, plasmid DNA, etc. We compared the 4 6 4 uptake of AF647-LNPs into normoxic BECs as well as into BECs exposed to OGD as an in vitro 4 6 5 model of cerebral ischemia. BECs were pre-exposed to OGD medium in a Billups-Rothenberg 4 6 6 chamber to induce ischemic damage while the normoxic cells were maintained in complete 4 6 7 growth medium in a normoxic incubator. We performed a live/dead assay to account for any cell debris or dead cells and noted that 4 7 0 about 98-99% cells were viable (Supplementary figures 1 and 2) . We next gated to eliminate 4 7 1 the autofluorescence of untreated/unstained cells (Supplementary figures 1 and 2 resulted in a 7.7-fold increase in the ATP uptake when compared to untreated cells (Figure 4a ).

This increased uptake is likely explained by their near-monodisperse, smaller particle diameters 4 7 9

(~100 nm, DI ~0.11). We noted a 3.6-fold increase in the ATP uptake in the case of 4 8 0

Lipofectamine/ATP complexes (Figure 4a) . Interestingly, ATP-LNPs (+PEG-DMG) resulted in 4 8 1 a significantly greater uptake (p<0.0001) compared to all treatment groups (Figure 4a) . We incubated hypoxic BECs with the same treatment groups. Interestingly, we observed an 4 8 4 uptake trend that was similar to BECs cultured in normoxic conditions. We observed a 2.7-fold 4 8 5 increase in the case of free ATP (Figure 4b ) that was comparable to the 2.9-fold increase comparisons test using one-way ANOVA. ****p < 0.0001. 5 0 7 5 0 8

Uptake of AF647-ATP uptake into primary human brain endothelial cells using 5 0 9 fluorescence microscopy 5 1 0

Primary human brain endothelial cells (HBMECs) were incubated with samples containing 5 5 1 1 mM ATP spiked with AF647-ATP at a ratio of 1000:1. We speculated that LNP-delivered ATP 5 1 2 will primarily localize in the cell cytosol as ionizable cationic lipids facilitate efficient 5 1 3 endosomal escape (56). We incubated HBMECs with the indicated samples for 24 h followed by 5 1 4

imaging the cells under an epifluorescent microscope (Figure 5a) . We labeled cell nuclei using 5 1 5

Hoechst to distinguish cytosolic AF647-ATP signals from the nuclei. and structural integrity, we chose to deliver ATP to offset the energy imbalance in ischemic 5 9 6

BECs. 5 9 7 5 9 8

In our previous work, we reported the effect of incorporation of PEG-DMG on the colloidal 5 9 9 stability of siRNA-LNPs (39). We also demonstrated that the incorporation of siRNA provides 6 0 0 an additional stabilizing effect to LNPs in addition to that conferred by PEG-DMG (39). Our First, we characterized particle diameters and surface charge of blank and ATP-loaded LNPs 6 0 7 prepared +/-PEG-DMG over a period of seven-days upon storage at 2-8 o C. We observed a non-6 0 8 significant increase in the particle diameters and dispersity indices of blank and ATP-loaded 6 0 9

LNPs prepared +/-PEG-DMG over a period of seven days (Figures 1a and b) . We observed a 6 1 0 significant difference in the particle diameters and dispersity indices of blank LNPs (+PEG-6 1 1 DMG) vs. blank LNPs (-PEG-DMG) (p<0.0001) and ATP-LNPs (+PEG-DMG) vs. ATP-LNPs (-6 1 2 PEG-DMG) (p<0.001) (Figures 1a and b) . These findings corroborate our earlier observation 6 1 3 that PEG-DMG is a crucial component to maintain lower particle diameters and dispersity 6 1 4 indices of both blank and ATP-LNPs. However, we did not find a significant difference in the 6 1 5 colloidal properties of blank vs. ATP-loaded LNPs (Figures 1a and b) . This implies that while studied changes in the Z avg and dispersity indices of the LNPs over pH ranging from 3 through 9 6 4 8

Naturally Derived Membrane Lipids Impact Nanoparticle-7

Mediated Delivery of CircSCMH1 Promotes Functional Recovery in Rodent and Nonhuman 7

Primate Ischemic Stroke Models

Inside out: optimization of

polyionic nature and large molecular mass of siRNA as compared to ATP. The polyionic siRNA 6 1 9likely forms multiple cooperative interactions with the cationic lipids while ATP, a small 6 2 0 molecule cargo, may not form the additional cooperative bonds that likely resulted in the 6 2 1 superior stability of siRNA-loaded LNPs. 6 2 2 6 2 3Next, we studied the surface charge (zeta potential) of the prepared LNPs over a period of 6 2 4 seven-days upon storage at 2-8 o C. We observed negative zeta potential values for all the samples 6 2 5(between -5 to -25 mV) indicative of the effects of PEG-DMG-mediated steric stabilization in 6 2 6 combination with the loading of ATP into the LNPs. We did not observe a significant difference 6 2 7in the zeta potential of LNPs over a period of seven-days for ATP-LNPs (+PEG-DMG) 6 2 8 suggestive of their stability. However, we did observe a significant difference in the zeta 6 2 9 potentials of blank LNPs (+/-PEG-DMG) (p<0.05) and ATP-LNPs (-PEG-DMG) (p<0.01) 6 3 0 ( Figure 1c) . We next studied the morphology of ATP-LNPs using TEM and compared it to the 6 3 1 morphology of siRNA-LNPs that has been extensively reported (64, 68). siRNA-LNPs show a 6 3 2 spherical morphology with a ring structure enclosing an amorphous core as previously reported 6 3 3 (64). We noted a similar morphology for ATP-LNPs (Figures 1d and e) . 6 3 4 6 3 5We characterized the colloidal stability of LNPs post-storage for 24 h at 37 °C. We did not 6 3 6 observe a significant increase in the particle diameters and zeta potential values of the ATP-6 3 7LNPs (+PEG-DMG) over a period of 24 h (Figures 2a and c) . Also, we did not note a 6 3 8 significant difference in the particle sizes of blank LNPs (+PEG-DMG) vs. blank LNPs (-PEG- indices, we did not observe significant differences among any given sample, however, we did 6 4 1The loss of mitochondrial ATP triggers the ionic, biochemical and cellular imbalances in 6 8 7 ischemic stroke (65). Protection of BECs is critical for maintaining brain homeostasis as they 6 8 8 maintain functional interactions with astrocytes and other support cells to form the BBB, the 6 8 9 neurovascular unit and the neural vascular niche (70). The BECs lining the BBB are an 6 9 0 accessible target for ischemic protection as opposed to the brain parenchyma tissue behind the 6 9 1 BBB (66). Hypoxia affects pinocytosis in BECs, causes higher cellular volumes, induces 6 9 2 structural changes in tight junction proteins and decreases the metabolic activity of BECs (71-6 9 3 74). Despite these cellular alterations, LNPs were internalized into hypoxic BECs. Overall, our 6 9 4 results suggest the potential of LNPs to deliver small molecule actives such as ATP to protect the 6 9 5 ischemic BBB. 6 9 6 6 9 7Conclusions 6 9 8For the first time, we formulated ATP, a small molecule compound using the clinically-6 9 9successful LNP platform. ATP-LNPs formulated in the presence of PEG-DMG showed optimum 7 0 0 colloidal stability at 2-8 ºC, 37 ºC and in the presence of serum proteins. Their improved 7 0 1 colloidal stability translated to their ability to increase the intracellular uptake of ATP into BECs, 7 0 2 a low pinocytic cell model. Our results demonstrate the potential of LNPs to deliver anionic 7 0 3 small molecules such as ATP to BECs lining the BBB.