Influence of Polyethylene Glycol Lipid Desorption Rates on Pharmacokinetics and Pharmacodynamics of siRNA Lipid Nanoparticles.

Lipid nanoparticles (LNPs) encapsulating short interfering RNAs that target hepatic genes are advancing through clinical trials, and early results indicate the excellent gene silencing observed in rodents and nonhuman primates also translates to humans. This success has motivated research to identify ways to further advance this delivery platform. Here, we characterize the polyethylene glycol lipid (PEG-lipid) components, which are required to control the self-assembly process during formation of lipid particles, but can negatively affect delivery to hepatocytes and hepatic gene silencing in vivo. The rate of transfer from LNPs to plasma lipoproteins in vivo is measured for three PEG-lipids with dialkyl chains 14, 16, and 18 carbons long. We show that 1.5 mol % PEG-lipid represents a threshold concentration at which the chain length exerts a minimal effect on hepatic gene silencing but can still modify LNPs pharmacokinetics and biodistribution. Increasing the concentration to 2.5 and 3.5 mol % substantially compromises hepatocyte gene knockdown for PEG-lipids with distearyl (C18) chains but has little impact for shorter dimyristyl (C14) chains. These data are discussed with respect to RNA delivery and the different rates at which the steric barrier disassociates from LNPs in vivo.Molecular Therapy-Nucleic Acids (2013) 2, e139; doi:10.1038/mtna.2013.66; published online 17 December 2013.

Introduction

A new generation of highly potent, well-tolerated lipid nanoparticles (LNPs) for systemic delivery of short interfering RNA (siRNA) therapeutics has recently moved from research into human clinical testing with promising results.[1,2] RNA interference (RNAi) is a biological mechanism to control gene expression through silencing of mRNA transcripts. The discovery that RNAi can be induced by synthetic siRNAs designed to inhibit specific gene targets initiated extensive research into the development of therapeutics to silence the expression of previously “undruggable” proteins responsible for disease.[3] When siRNA is introduced into cells, specifically into cytoplasmic compartments, it may get loaded into RNA-induced silencing complex and direct sequence-specific cleavage of target mRNA. However, intracellular delivery of these large molecular weight, polyanionic compounds is a major challenge as they do not passively diffuse across plasma membranes. Several years ago it was discovered that siRNAs encapsulated in fusogenic, dialkyl amino LNPs potently inhibit hepatic gene expression in vivo, producing profound reductions in target protein levels for several weeks after a single dose.[4]

Rational molecular design has been applied to enhance the potency and safety of the amino lipid components,[5,6,7] and it was recently discovered that endogenous apolipoprotein E (Apo E) specifically targets these delivery systems to hepatocytes in vivo where they undergo Apo E–dependent, receptor mediated endocytosis.[8] The most advanced dialkyl amino LNP formulations are composed of four lipids. Two of these, cholesterol and saturated phosphatidylcholine, are considered structural components that do not appear to play a direct role in the mechanism of action. Amino lipid and polyethylene glycol lipid (PEG-lipid) components, on the other hand, play more active roles in both the formulation process and the mechanism of intracellular delivery. In previous studies, we defined the structure-activity relationship for dialkyl amino lipids with respect to hepatic gene silencing. Applying these molecular principles resulted in the synthesis of a highly active, novel amino lipid dilinoleylmethyl-4-dimethylaminobutyrate (DLin-MC3-DMA), which is referred to as MC3 and is the amino lipid used in the studies described here.[5] LNPs containing MC3 and encapsulating siRNA therapeutics are currently in Phase II clinical trials for hypercholesterolemia and transthyretin amyloidosis.[1,2] More recently, we have developed novel, next-generation lipids that combine the excellent potency of the most advanced lipids currently available with biodegradable functionality.[9]

In this work, we turn our attention to the PEG-lipid component of siRNA LNPs. PEG-lipids are typically incorporated into lipid-based drug delivery systems to provide a hydrophilic steric barrier that inhibits the binding of plasma proteins, including opsonins, which mark LNPs for removal by the mononuclear phagocytic system. The resulting increase in circulation lifetime enables drug delivery vehicles to take full advantage of enhanced permeability and retention effects, resulting in their accumulation at sites of disease.[10] However, PEG-lipid is incorporated into the current generation of siRNA LNPs for quite a different reason. Its primary role is to facilitate the self-assembly of LNPs by providing a steric barrier at the surface of nascent particles formed when siRNA is rapidly mixed in acidic ethanol solutions with positively charged amino lipids. PEG steric hindrance prevents inter-particle fusion and promotes the formation of a homogeneous population of siRNA LNPs with diameters <100 nm.[11,12]

Yet the presence of a PEG shield on siRNA LNPs in vivo inhibits hepatic gene silencing activity. The protection it provides against interactions with plasma proteins appears to also interfere with Apo E-mediated endocytosis into hepatocytes.[8] Moreover, PEG-lipids reduce the close apposition and fusion required for the release of siRNA into the cytoplasm from endocytic vesicles.[13] Consequently, to maximize hepatic gene silencing in vivo, the concentration of PEG-lipid is reduced to the minimum required to support self-assembly and the alkyl chains are kept relatively short to promote shedding of the steric barrier following intravenous (i.v.) administration.

Given the important role of PEG-lipids in siRNA LNP delivery systems there is surprisingly little information available concerning their interactions with plasma after injection. Specifically, little is known about the kinetics of PEG release from LNPs in vivo or the identity of the PEG-lipid acceptors. Therefore, we conducted this study to characterize three PEG-lipids with saturated alkyl chain lengths of 14, 16, and 18 carbons. Dual labeled LNPs containing 14C-MC3 and 3H-PEG-lipid were employed to directly measure desorption rates for each chain length after separating LNPs from plasma lipoproteins. The desorption kinetics of the steric barrier were then correlated to LNP pharmacokinetics (PK), biodistribution (BD) and the ability to silence the hepatocyte-expressed factor VII gene.

Results

Dose and clearance of LNPs from blood

The objective of the first study was to determine the lipid dose at which to conduct this analysis. The current generation of siRNA LNPs is so potent that a typical therapeutic dose represents <1 mg/kg of total lipid injected i.v., of which only 3% by weight is PEG-lipid.[5] Taking into account the specific activities of 14C- and 3H-labeled lipids available, we estimated an siRNA dose administered at 0.3 mg/kg (3.3 mg/kg total lipid) would be sufficient to provide a limit of detection equivalent to ~1% of the injected dose per animal. Although this lipid dose is low for typical liposome drug delivery systems, it is still tenfold higher than that required for a therapeutic effect using siRNA LNPs in mouse models. Therefore, to ensure the data obtained in these studies is relevant to therapeutic applications, an experiment was conducted to compare the PK of LNPs at 0.03, 0.3, and 1.0 mg/kg siRNA.

LNPs encapsulating FVII siRNA were trace labeled with tritiatedcholesterylhexadecylether (3H-CHE), an ether-linked cholesterol ester analogue. This lipid is nonexchangeable and nonmetabolizable in mice[14] therefore 3H remains associated with particles in blood and tissues for at least 24 hours. LNPs were administered at total lipid doses of 0.33, 3.3, and 11.1 mg/kg (corresponding to 0.03, 0.3, and 1 mg/kg siRNA, respectively), and then blood, liver, and spleen samples were removed and measured for radioactivity over 24 hours. Expressed as % injected dose, all three concentrations show similar profiles for LNP clearance from the blood compartment and subsequent accumulation in liver and spleen tissues (). Less than 5% of the injected dose remains in circulation after 4 hours and <1% after 24 hours (). The rates of accumulation by liver and spleen are also similar for the three doses, with 60–80% of the injected dose recovered in the liver () and ~10% in the spleen (). These data, together with the corresponding PK values in , show the proposed siRNA dose of 0.3 mg/kg (total lipid dose of 3 mg/kg) falls within a range where dose proportionality is observed (linear area under the curve (AUC) versus lipid dose plots inserted into ,) with no indication that saturation of the mononuclear phagocytic system has been reached. Therefore, PK data obtained at this dose are expected to extrapolate to the therapeutic dose range.

Characterizing the 14C-MC3 to 3H-CHE ratio in vivo

In order to measure desorption rates of 3H-PEG-lipids from LNPs, we intended to use 14C-MC3 as the reference lipid that remains associated with LNPs in the circulation. To confirm its suitability, a dual label study was conducted, which showed the ratio of 14C-MC3 to 3H-CHE, a lipid known to stay associated with LNPs in the circulation, did not change over a 2-hour period in vivo (). During this time, 90% of the injected LNP dose is removed from blood, therefore the disposition of PEG-lipids in the circulation was determined over a 0- to 2-hour time course.

Desorption of PEG-lipids with different chain lengths from LNPs in vivo

The most common PEG-lipids employed in the current generation of siRNA LNPs contain two saturated, dimyristyl chains (PEG-C14).[7] We have shown in vitro that PEG-C14 transfers from LNPs to lipoproteins at a rate of ~50%/hour in mouse plasma at an LNP-to-plasma ratio equivalent to administering LNPs in vivo at a total lipid dose of 20 mg/kg, which is sevenfold higher than the dose employed here (manuscript in preparation). In vivo, a gradual increase in the 3H/14C ratio measured in whole blood and plasma occurs over the first hour indicating that 14C-MC3–labeled LNPs are leaving the circulation faster (80% cleared in the first hour, ), than 3H-PEG-C14. These data are consistent with 3H-PEG-C14 transferring to lipoproteins, which remain in circulation longer than LNPs, resulting in an increase in 3H/14C ratio. Specific loss of 3H-PEG-lipid from LNPs becomes more apparent when plasma, isolated from blood drawn at each time point, is separated from lipoproteins by size exclusion chromatography enabling the 3H/14C ratio of the recovered LNPs to be measured directly (). The ratio decreases rapidly, with ~80% of the initial 3H-PEG-C14 leaving LNPs within 2 hours.

The hydrophobic interactions responsible for associating PEG-lipids with LNP membranes are proportional to their alkyl chain length.[14,15] Therefore, PEG-C16 and PEG-C18 lipids should remain associated with LNPs in vivo longer than PEG-C14, as more energy is required for the longer chains to leave the membrane. Analysis of LNPs containing these lipids confirms this, with desorption estimated at 45, 1.3, and 0.2%/hour for C14, C16, and C18 PEG-lipids, respectively ().

The effect of PEG-lipid chain length and concentration on LNP pharmacodynamics

In the next series of experiments, we determine the effect of PEG chain length on murine hepatic gene silencing using FVII siRNA LNPs. Activity was measured over a range of doses which enables the determination of the dose required to reduce plasma FVII protein concentration by 50% (ED50), for each formulation (). We have shown previously that this represents the most accurate method to compare the potency of different formulations.[5,7] There is no statistically significant difference (P > 0.05) in the ED50 values for LNPs containing 1.5 mole % PEG-C14, -C16, or -C18 (, ED50 range 0.02–0.04 mg siRNA/kg). Increasing the concentration of PEG-C14 from 1.5 to 2.5 and 3.5 mol % results in ED50 values for FVII silencing of 0.02, 0.03, and 0.06 respectively (), differences which are also not statistically significant. However, LNPs containing 2.5 and 3.5% PEG-C18 are inactive over the concentration range employed ().

The effect of PEG-lipid chain length on LNP PK and BD

Steric barriers provided by PEG-lipid are most commonly employed to reduce the rate at which drug delivery systems distribute from the blood compartment and to redirect their BD away from the liver.[10] Consequently, both PK and BD of LNPs can be modified by changing the anchor chain length of PEG-lipids.[14,16] This effect is shown for siRNA LNPs containing PEG-C14, -C16, and -C18 where both 3H-PEG-lipid and 14C-MC3 are tracked in blood, liver, and spleen over 24 hours ( and ). Using 14C-MC3 to track LNPs, the most rapid clearance of LNPs from blood is observed for PEG-C14 (t1/2 = 0.64 hours), followed by PEG-C16 and -C18 LNPs with t1/2 = 2.18 and 4.03 hours, respectively ( and ). The rate of 14C-MC3 clearance from the blood compartment is not significantly different (P > 0.05) from that obtained for the same LNPs labeled with 3H-CHE (see and ), indicating that the MC3 lipid remains associated with LNPs while in circulation. This is also shown in , where 14C-MC3 remains associated with LNPs over 2 hours in circulation, during which time 90% of the particles are removed. In contrast, 3H-PEG-C14 exhibits a longer t1/2 and greater blood AUC than the 14C-MC3 ( and ), consistent with the movement of this lipid into the lipoprotein pool, where it remains in circulation for longer.

Distribution out of the blood compartment for the three formulations correlates with concomitant accumulation in liver tissue (). PEG-C14 LNPs accumulate rapidly, reaching a maximum 55% of the injected dose at 4 hours. Note, however, that the total amount of PEG-C14 LNPs removed by liver tissue at this dose is actually 70%, determined using the stable lipid marker 3H-CHE (). We speculate that it appears lower when measured using 14C-MC3 as trace label because the ester between the 14C-label and the alkyl chains is slowly hydrolyzed. This would also account for the lower AUC(0–24) for 14C-MC3 in the liver, which is ~25% lower than that obtained using the nonexchangeable and nonmetabolizable 3H-CHE label ( and ). An even lower liver accumulation is observed for the 3H-PEG-C14, consistent with the rapid dissociation of this molecule from circulating LNPs as discussed above (). Peak accumulation occurs after 2 hours at just 30% of the injected dose (). Interestingly, there is also a steady decline in 3H-PEG-C14 recovered between 2 and 24 hours (), indicating that when this lipid is taken up by the liver it is either metabolized or excreted over time.

Both PEG-C16 and -C18 accumulate more slowly in the liver and only reach a maximum of 35 and 25% of the injected dose respectively ( and ). Moreover, the elimination of 3H-PEG-lipid from the liver over 24 hours also appears to depend on chain length, with a large decrease in PEG-C14 levels between 2 and 24 hours compared to PEG-C16 and PEG-C18. These lipids have identical structures including ether linkages between the alkyl chains and headgroup that are resistant to hydrolysis.[17] Therefore, these data suggest that the rate of PEG-lipid elimination from liver tissue may depend on its hydrophobicity and ability to interact with biological membranes. When uptake is expressed in grams lipid/gram of tissue, the spleen removes equivalent amounts of LNPs as liver. However, the spleen is smaller and therefore distribution to this organ typically accounts for <10% of the injected dose. Although standard errors are larger for the spleen data, the trends with respect to PEG-lipid chain length and accumulation are similar to those observed in liver (,).

Discussion

The steric barrier provided by PEG-lipid is necessary to control particle size during self-assembly of siRNA LNPs but has the negative attribute of inhibiting hepatic gene silencing in vivo.[7,8] In this study, we evaluate three PEG-lipids with dialkyl chains of 14, 16, and 18 carbons and correlate transfer rates with their effect on PK and activity in vivo. LNP delivery systems containing ionizable dialkylamino lipids require endogenous Apo E in order to access hepatocytes through Apo E–dependent, receptor mediated endocytosis. Therefore, one explanation for inhibition of activity is that PEG interferes with Apo E binding to LNPs.[8] However, it is also possible that a PEG shield prevents bound Apo E from undergoing the lipid-induced structural transition required for binding to hepatocyte receptors or the PEG barrier interferes with receptor recognition of Apo E on the LNP surface.[18,19] In this study, we characterize how a short chain PEG-lipid, with covalently attached PEG of molecular weight 2000, helps to mitigate this problem by rapidly exchanging from the LNP surface to plasma lipoproteins following i.v. administration. We evaluate three PEG-lipids with dialkyl chains of 14, 16, and 18 carbons and correlate transfer rates with their effect on PK and activity in vivo.

The spontaneous transfer of lipid from a membrane monolayer into the aqueous phase proceeds through a transition-state complex where the molecule is attached to the membrane by the tip of its hydrophobic tail.[20] The activation energy required to enter this state is very high and largely determined by the number and length of hydrophobic chains within the molecule. Lipids that achieve this transition state can rapidly diffuse through the aqueous space down a concentration gradient to acceptor sites such as other membranes and lipoproteins. The large energy barrier associated with entering the activated state is why the majority of membrane phospholipids, which have two acyl chain lengths >16 carbons, form stable bilayers and exhibit extremely low rates of transfer, with typical half-times of hundreds of hours in the absence protein facilitators.[20] The rate of lipid desorption, however, increases exponentially with decreasing chain length and is further enhanced by large hydrophilic headgroups, such as PEG, that contribute to lower the energy required to adopt transition-state complexes.[13,19]

By isolating plasma at various time intervals after injection and separating LNPs from lipoproteins, we determined the rate of loss for each PEG-lipid from LNPs in vivo to be >45%/hour, 1.3%/hour, and 0.2%/hour for C14, C16, and C18, respectively. Formulation optimization studies have consistently demonstrated that to form stable siRNA LNPs with diameters <100 nm the minimum amount of PEG2000-lipid required is ~1.5 mol %.[5,7] At this concentration, FVII siRNA LNPs containing PEG-C14, -C16, and -C18 exhibit similar siRNA ED50's of ~0.03 mg/kg. When the PEG-lipid content is raised to 2.5 mol % there is little or no effect on the ED50 of PEG-C14 LNPs, whereas PEG-C18 LNPs completely lose activity in the dose range tested. This impact on activity resulting from such a small increase in mol percent highlights the effectiveness of the steric barrier provided by PEG-lipids.

The rate of PEG-C18 desorption from LNPs is 0.2% per hour, which is negligible relative to their circulation half-life of under 4 hours and consistent with the general assumption that PEG-C18 steric barriers can be considered permanently associated to circulating lipid-based delivery systems. Both theoretical and experimental data indicate that the concentration of PEG2000 required to completely cover both sides of model membrane bilayer surface is ~2–3 mol % of total lipid.[21,22] Below 1 mol % there is sufficient space between neighboring PEG clouds, or mushroom configurations, to allow large molecules access to the membrane surface. As the PEG concentration increases to 3 mol % the mushrooms are squeezed together to cover the whole surface, forming a cohesive PEG barrier. Between 3–10 mol % the increased PEG density causes the polymers to extend away from the membrane surface and adopt a brush configuration.[21] Recent data indicate that for siRNA LNPs all the PEG-lipid is located in the outer lipid monolayer,[23] therefore 1.5 mol % of total lipid would approximate 3 mol % of the outer monolayer lipid. Consequently, the abrupt change in siRNA LNP activity corresponds reasonably well to crossing the threshold between incomplete and complete coverage of the surface by PEG. It is also consistent with 1.5 mol % being the minimum PEG2000 concentration necessary to prevent particle-particle fusion during self-assembly.

The initial rate of desorption for PEG-C14 from LNPs in vivo is ~2% per minute, this slows down at later time points presumably because of back exchange from lipoproteins (). However, the rate of transfer to lipoproteins is fast enough to significantly mitigate the negative effect of the PEG shield on FVII knockdown at 2.5 and 3.5%. PK and BD data for siRNA LNPs indicate that the mechanism of LNP removal from the circulation does not exhibit the same PEG-lipid threshold effect as seen for hepatic gene silencing. At 1.5 mol % PEG-lipid the circulation half-life of 14C-MC3–labeled LNPs (estimated over 24 hours) for PEG-C14, -C16, and -C18 LNPs is 0.64, 2.18, and 4.03 hours respectively. There is also a marked decrease in liver and spleen uptake with increasing chain length. These data indicate that both 1.5 mol % PEG-C16 and -C18 steric barriers reduce the association of opsonins with LNPs in the circulation. The reason for the different PEG effects for opsonin induced clearance and Apo E-mediated uptake are not known at this time, but may reflect the molecular weights of the various proteins involved and their ability to penetrate a low density PEG shield.

Despite the differences in liver uptake between LNPs containing 1.5 mol % PEG-C14 and PEG-C18 they exhibit similar gene silencing activities, indicating that delivery to the liver alone is not a limiting factor for activity under these dosing conditions. As might be expected, a comparison of 14C-MC3 to 3H-PEG-lipid in liver shows that the amount of 3H-label taken up reflects the rates of transfer to lipoproteins and erythrocytes while LNPs are in the circulation (,). However, it is interesting to note that by 24 hours, >50% of the 3H-PEG-C14 accumulated in the liver at 2 hours has been eliminated whereas the concentration of 3H-PEG-C18 remains constant. The 3H-label is located in the alkyl chains, which are joined to the headgroup by metabolically stable ether linkers.[17] Consequently, the differences in liver processing between the short and long chain PEG-lipids may also be related to lipid transfer rates and perhaps reflect PEG-C14 excretion into bile rather than metabolism. The % injected dose that distributes to the spleen is much less than the liver as it is a much smaller organ but the chain length trends are similar.

In summary, we demonstrate that the presence of a 1.5 mol % PEG2000-lipid steric barrier does not significantly affect in vivo hepatic gene silencing for siRNA LNPs. However, increasing the PEG-lipid concentration by just 1 mol % can significantly reduce activity unless a short chain PEG-C14 is used, which rapidly desorbs from LNPs in the circulation and transfers to lipoproteins and erythrocytes. Moreover, different combinations of PEG-lipid chain length and concentration can be employed to modify the PK and BD of siRNA LNPs without greatly impacting hepatic gene silencing.

Materials and methods

General. Distearoylphosphatidylcholine and cholesterol were purchased from Avanti Polar Lipids (Alabaster, AL) and Sigma (St Louis, MO), respectively. The syntheses of MC3[5] and PEG-lipids[24] have been described previously. FVII siRNA was synthesized by Alnylam and characterized by electrospray mass spectrometry and anion exchange HPLC, the sequence has been presented elsewhere.[25]

Radiolabeled lipids. The PEG-lipid precursors containing one unsaturation in each of the lipid chains (myrystoleyl, palmetoleyl and oleyl chains, C14, C16, and C18 respectively) were synthesized using a similar procedure to that used for the synthesis of corresponding saturated analogs as reported.[24] 3H-PEG-lipids (410 Ci/mol) were synthesized at MoravekBiochemicals (Brea, CA) by catalytic reduction of the double bonds present in the lipid chains using tritium gas over Pd/C. Tritiatedcholesterylhexadecylether (3H-CHE, 48,000 Ci/mol) was purchased from Perkin Elmer (Waltham, MA) and 14C-MC3 (58 Ci/mol) was synthesized as described elsewhere.[6]

Preparation of LNP formulations. MC3, distearoylphosphatidylcholine, cholesterol and PEG-lipid were solubilized in ethanol at a molar ratio of 50:10:38.5:1.5. For LNPs containing 2.5 and 3.5% PEG-lipids, the cholesterol content was lowered to 37.5 and 36.5%, respectively. Radiolabeled lipids were included in the ethanolic lipid solution at a 3H to 14C dpm ratio of ~3 and their mass is taken into account in the target lipid mole ratios indicated above. The 3H-CHE content was adjusted to 0.1, 0.02, and 0.006 mol % with respect to the total lipid for LNPs dosed at 0.33, 3.33, and 11.1 mg/kg lipid respectively. LNP specific activities were calculated to allow a minimum detection of ~1% of the injected dose per animal.

LNPs were prepared at a total lipid to siRNA weight ratio of ~10:1 as described elsewhere.[6] Briefly, the siRNA was diluted to 0.2 mg/ml in 10 mmol/l citrate buffer, pH 4. Syringe pumps were used to mix the ethanolic lipid solution with the siRNA aqueous solution at a 1:5 (vol/vol) ratio using pump rates of 6.8 and 34 ml/minute, respectively. The two syringes containing the siRNA and lipid solutions were connected to a union connector (IDEX Health & Science #P-728) with PEEK HPLC tubing (0.02 inch ID for siRNA solution and 0.01 inch ID for lipid solution). A length of PEEK HPLC tubing (0.04 inch ID) was connected to the outlet of the union connector and led to a collection tube. The ethanol was then removed and the external buffer replaced with phosphate-buffered saline (155 mmol/l NaCl, 3 mmol/l Na2HPO4, 1 mmol/l KH2PO4, pH 7.5) by dialysis. Finally, the LNPs were filtered through a 0.2 μm pore sterile filter. LNP particle size was 70–90 nm diameter as determined by quasi-elastic light scattering using a Nicomp370 submicron particle sizer (Santa Barbara, CA). Lipid concentrations were calculated based on the total cholesterol measured using the cholesterol E enzymatic assay kit from Wako Chemicals (Richmond, VA). SiRNA content was determined by A260 after solubilizing an aliquot in a chloroform:methanol:aqueous mixture (final volume ratio of 1:2.1:1). Removal of free siRNA was performed using VivaPureDMiniH columns (Sartorius Sedim Biotech, Goettingen, Germany). Encapsulation efficiency was determined from the ratio of siRNA before and after removal of free siRNA content, normalized to lipid content.

PK and BD of radiolabeled LNP. Radiolabeled LNPs were administered intravenously via the lateral tail vein at a volume of 10 ml/kg in 6- to 8-week-old CD-1 mice. At various times, the mice were anesthetized with Ketamine/Xylazine and blood withdrawn by cardiac puncture and collected in microtainer tubes with EDTA (Becton-Dickinson, Franklin Lakes, NJ). A portion of the blood was centrifuged at 500g for 10 minutes to isolate plasma. Liver and spleen tissue were also processed by transferring the pre-weighed whole spleen or a piece of the liver (60–80 mg) into Fastprep tubes and homogenized with phosphate-buffered saline (0.75–1 ml) using a Fastprep-24 (MP Biomedical, Santa Ana, CA). Aliquots (0.1–0.2 ml) of the homogenate, blood, and plasma were transferred to 7 ml glass scintillation vials and subjected to a digestion and decolorization protocol as follows. An aliquot (0.5 ml) of Solvable (Perkin Elmer, Waltham, MA) was added to the vial and the sample digested for 2 hours at 60 °C, after which the samples were cooled before adding 30% hydrogen peroxide (0.2 ml), 200 mmol/l EDTA (0.05 ml), and 10 N hydrochloric acid (0.025 ml). The mixture was decolorized overnight prior to adding 5 ml of PicoFluorPlus scintillation fluid (Perkin Elmer). Radioactivity was measured using a Beckman Coulter LS6500 liquid scintillation counter (Mississauga, Ontario, Canada). The percent recovery in blood and plasma was calculated based on a volume of 70 and 38.5 ml/kg animal weight, respectively. Liver and spleen associated radioactivity are expressed as percent injected dose per total organ weight. Encapsulated siRNA was not followed in this study, but previous reports have shown that similar formulations do not leak nucleic acid in the circulation.[7,26]

PK and statistical analysis. AUC0–24 was calculated using the trapezoidal rule. The half-life for LNP and/or label distribution from the blood compartment into tissues was calculated by fitting the concentration versus time curve to first order exponential function Ct = C0e(−kt), where Ct and C0 are the radiolabel concentrations at time t and zero. The distribution constant k derived from the plot was used to calculate t1/2 using the formula t1/2 = 0.693/k. Statistical significance (P ≤ 0.05) between data sets was determined using the Excel TTEST function.

Size exclusion chromatography to determine the rate of 3H-PEG-lipid transfer from LNPs to lipoproteins The separation of LNPs from plasma lipoproteins using size exclusion chromatography has been described previously.[27] Loss of 3H-PEG-lipid from 14C-MC3–labeled LNP in plasma was determined using Sepharose CL-4B (Sigma) and 1.5 cm diameter × 30 cm columns. Plasma samples (0.2 ml) were eluted with phosphate-buffered saline and 20 × 2 ml fractions collected then mixed with 4.8 ml of PicoFluorPlus scintillation fluid and counted for radioactivity.

In vivo efficacy studies characterizing FVII gene silencing. Six- to eight-week-old female C57Bl/6 mice were obtained from Charles River Laboratories (Wilmington, MA). LNPs were diluted to the required concentrations in sterile phosphate-buffered saline before use and administered intravenously via the lateral tail vein at a volume of 10 ml/kg. After 24 hours, animals were anesthetized with Ketamine/Xylazine and blood collected by cardiac puncture. Samples were processed to serum using microtainer serum separator tubes (Becton-Dickinson) and stored at −70 °C for later analysis of FVII protein content using the colorimetric Biophen VII assay kit (Aniara, West Chester, OH). All procedures were approved by the University of British Columbia institutional animal care committee and performed in accordance with guidelines established by the Canadian Council of Animal Care.

The determination of FVII ED50 values has been described in detail elsewhere.[7] Briefly, formulations were administered over a range of siRNA doses chosen to include data points within 10–90% residual FVII activity (typically three to four points). ED50 values represent the siRNA dose in mg/kg at which the serum concentration of FVII has been reduced by 50% compared to the saline-control treated animals and is derived from a linear interpolation of the FVII activity profile as shown in .

Table 1

Pharmacokinetic parameters for mice administered siRNA LNPs containing 1.5 mol % PEG-C14

Table 2

Pharmacokinetic parameters for mice administered siRNA LNPs containing 1.5 mol % C14, C16, and C18 PEG-lipids