Journal of Controlled Release 174 (2014) 7–14

Contents lists available at ScienceDirect

Journal of Controlled Release

j ourna l homepage: www.e lsev ie r .com/ locate / jconre l

The development of an in vitro assay to screen lipid based nanoparticlesfor siRNA delivery

Ye Zhang ⁎, Leticia Arrington, David Boardman, Jared Davis, Yan Xu, Katie DiFelice, Steve Stirdivant,Weimin Wang, Brian Budzik, Jack Bawiec, James Deng, Greg Beutner, Darla Seifried, Matthew Stanton,Marian Gindy, Anthony LeoneDepartment of RNAi Therapeutics, Merck Research Laboratories, 770 Sumneytown Pike, West Point, 19486, USA

⁎ Corresponding author. Tel.: +1 215 652 2401; fax: +E-mail address: zhangye101@gmail.com (Y. Zhang).

0168-3659/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.jconrel.2013.11.006

a b s t r a c t

a r t i c l e i n f o

Article history:Received 28 August 2013Accepted 4 November 2013Available online 12 November 2013

Keywords:siRNA deliveryLipid nanoparticlesSerum stabilityEndosome escapesiRNA release

In order to rapidly screen and select lead candidates for in vivo evaluation of lipid nanoparticles (LNPs) for sys-temic small interfering RNA (siRNA) delivery, an in vitro assay amenable to high-throughput screening (HTS)is developed. The strategy is tomimic the in vivo experience of LNPs after systemic administration, such as inter-actions with serum components, exposure to endosomal pH environments, and interactions with endosomalmembrane lipids. It is postulated that the amount of siRNA released from LNPs after going through these treat-ments can be used as a screening tool to rank order the effectiveness of siRNA delivery by lipid nanoparticlesin vivo. LNPs were incubated with 50% serum from different species (i.e. mouse, rat, or rhesus) at 37 °C. Theresulting samples were then reacted with anionic, endosomal-mimicking lipids at different pHs. The amount ofsiRNA released from LNPs was determined using spectrophotometry employing the fluorescent indicator SYBRGold. Our results indicated that the amount of siRNA liberated was highly dependent upon the species ofserum used and the pH to which it was exposed. LNPs treated with mouse serum showed higher levels ofsiRNA release, as did those subjected to endosomal pH (6.0), compared to physiological pH. Most interestingly,a good correlation between the amount of siRNA released and the in vivo efficacy was observed. In conclusion,an in vitro siRNA release assay was developed to screen and rank order LNPs for in vivo evaluation.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

Since the discovery of RNA interference (RNAi) by Fire and Mello in1998 [1] and the demonstration of synthetic small interfering RNA(siRNA) mediated RNAi by Tuschl and colleagues in 2001 [2], therapeu-tic development of siRNA drugs to silence disease genes has been pur-sued extensively by the pharmaceutical industry and academicresearchers. Similar to gene therapy, efficient delivery of siRNA to targetcells and tissues with minimal toxicity is the major challenge for thesuccessful development of siRNA based therapeutics. Various deliveryapproaches, including lipid or polymer based nanoparticles, and siRNAconjugates have been investigated [3–5]. For systemic delivery ofsiRNA, one of themost advanced delivery technologies is lipid nanopar-ticles (LNPs).

Typically, LNPs consist of 4 to 5 components: cationic lipid, neutralhelper lipid (s), polyethylene glycol (PEG) conjugated lipid, and siRNA[4,6,7]. Cationic lipids such as 3-dimethylamino-2-(Cholest-5-en-3βoxybutan-4-oxy)-1-(cis,cis-9,12-octadecadienoxy)propane (CLinDMA)[6] and 1,2-dilinoleyloxy-3-dimethylaminopropane (DLinDMA) [4] are

1 215 652 7310.

ghts reserved.

designed to have an apparent pKa of 6.5–8.0 and to produce neutralparticles in blood circulation at physiological pH. LNPs are expected tobe protonated at endosomal pH 6.0 to facilitate interactions with thenegatively charged endosomal lipids, and the subsequent release ofsiRNA into the cytoplasm [8]. Strong positive charge in general circula-tion increases the potential of LNP opsonization, reticuloendothelialsystem (RES) clearance, immune response in the form of cytokine in-duction and complement activation, as well as other toxicities [6,9,10].Therefore, a particle with pKa ~ 7 might be ideal. Neutral helper lipidssuch as cholesterol and phosphatidylcholine (PC) confer rigidity andstability to the bilayer [11–13]. The PEG lipid provides stealthness andstability to LNPs; however, it may also interfere with interactions be-tween particle and target cells, hindering endosome escape andinhibiting transfection [14,15]. Therefore, it is designed to diffuse tosome extent when interacting with serum components in circulation,to maintain the balance between stealth, stability, and cell transfection[16–19].

After intravenous administration, LNPs are exposed to various con-stituents found in whole blood, namely serum proteins, lipoproteins,and blood cells (Fig. 1).

Interactions between LNPs and these elements have been shown tochange the particle size and/or surface properties of LNPs [9,20–25]. Fol-lowing circulation, LNPs accumulate in the liver cells, either passively or

Fig. 1. Illustration of the barriers and potential LNP transformations after intravenous ad-ministration.Adapted and reproduced with permission from Journal of Medicinal Chemistry. 2010, 22,7888. Copyright 2010 American Chemical Society.

8 Y. Zhang et al. / Journal of Controlled Release 174 (2014) 7–14

through receptormediated endocytosis. Serum component Apolipopro-tein E (ApoE) has been reported as an endogenous targeting ligand forhepatocyte delivery of ionizable LNP [26]. The pHwithin the endosomesdecreases from 7.4 to 5 or 6within 10–20 min. The ability of LNP encap-sulated siRNA to escape early endosomes is critical to efficacy, failureresults in residence in late stage endosomes or lysosomes, where degra-dation occurs.

Considering the cost and labor associated with testing delivery vehi-cles in animals, many in vitro assays have been proposed and used byresearchers to screen lipid based delivery vectors for oligonucleotides.These include cell based transfection, serum stability, and/or endosomeescape propensity [8,20,24,27–31]. Most recently, a multiparametricapproach was also proposed to evaluate LNP for siRNA delivery [32].However, a discrepancy was found between in vitro and in vivo efficacydata [32,33] and none of these assays have proven good predictors forin vivo efficacy. This is mainly because the in vitro environment fails toadequately approximate the in vivo situation. For example, in culture,cells were treated with delivery vehicles either in the presence or ab-sence of 10% fetal bovine serum (FBS). This value is approximately onefifth of the physiological serum concentration (≈50% v/v). Similarly,although the serum stability assay has been conducted at a higher,more relevant concentration of 50% serum, it only evaluates LNP stabil-ity in serum. It cannot provide information on endosome escape andsiRNA release [27,34]. In the present study, we report a new in vitroassay for LNP screening which better emulates the barriers for systemicdelivery. By mimicking the in vivo environment LNPs experience aftersystemic administration, such as interactions with serum componentsand exposure to endosomal pHand lipids, greater predictive capabilitiesare garnered.

2. Materials and methods

2.1. Materials

1,2-Dioleoyl-sn-glycero-3-(phosphor-L-serine) (DOPS), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE), 1,2-dipalmitoyl-sn-glycero-3-phosphoethanolamine-N-(7-nitro-2-1,3-benzoxadiazol-4-yl) (NBD-PE) and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissaminerhodamine B sulfonyl) (N-Rh-PE) were purchased from AvantiPolar Lipids (Alabaster, AL). Cholesterol, 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC)and1,2-distearoyl-sn-glycero-3-phosphocholine(DSPC) were purchased from Northern Lipids Inc. (Burnaby, BC, Canada).Poly(ethylene glycol)2000–dimyristoylglycerol (PEG–DMG)wasmanufac-tured by NOF Corporation (Kanagawa, Japan). All cationic lipids used inthe study including CLinDMA and DLinDMA were synthesized at Merck(West Point, PA). Mouse, rat, and rhesus serum were obtained from

Bioreclamation LLC (East Meadow, NY). SYBR Gold was acquired fromInvitrogen (Carlsbad, CA). Chemically modified ApoB siRNA, targetingthe mRNA transcript for the ApoB gene (accession # NM_019287), wassynthesized at Merck (Rahway, NJ). The primary sequence (presentedin the 5′ to 3′ direction) and chemical modification pattern of the ApoBsiRNA strands are as follows:

Sense strand: iB; omeC; omeU; omeU; omeU;fluA;fluA; omeC;fluA;fluA; omeU; omeU; omeC; omeC; omeU; fluG; fluA; fluA; fluA;omeU; dTsdT-iB.Antisense strand: rA; srU; srU; someU; omeC; fluA; fluG; fluG; fluA;fluA; omeU; omeU; fluG; fluU; omeU; fluA; fluA; fluA; fluG; omeU;

someU.

Nucleotide modifications are represented as deoxy (d), 2′-fluoro(flu), 2′-o-methyl (ome), and ribo (r), with abbreviations immediatelypreceding the altered base. Phosphorothioate linkages replacing phos-phate bonds are represented by a subscript (s) between the two nucle-otides linked. Additionally, the sense strand is blocked with an invertedabasic nucleotide on the 5′ and 3′ ends (iB). All chemicals were ofreagent grade or higher quality.

2.2. Preparation of siRNA encapsulated LNPs and anionic liposomes

siRNA encapsulated LNPs were prepared as described previously byAbrams et al. [6]. LNP A contains CLinDMA, cholesterol, and PEG–DMGat a molar ratio of 68:30:2 and a nitrogen/phosphate (N/P) ratio of2.8. LNP B and the other 16 LNPs in the correlation study contain cation-ic lipid, cholesterol, DSPC and PEG–DMG at a molar ratio of 58:30:10:2and an N/P ratio of 6. Endosome mimicking anionic liposomes wasprepared by mixing DOPS:DOPC:DOPE (mol/mol/mol 1:1:2) [31] inchloroform, followed by solvent evaporation under a streamof nitrogen,and exposure to vacuum at 100 mTorr for 2 h to remove residual chlo-roform. The dried lipid film was subsequently resuspended in 1× PBS.Anionic liposomes could be stored at 4 °C for 1 month. They were son-icated for 10 min before each use. For the fluorescence labeled anionicliposomes, DOPS:DOPC:DOPE:NBD-PE and N-Rh-PE were prepared atmolar ratio of 25:25:48:1:1.

2.3. Determination of siRNA encapsulation rate of LNP by SYBR Gold assay

The fluorescent reagent SYBR Gold was employed for RNA quantita-tion to monitor the encapsulation rate of LNPs. LNPs were combinedwith this indicator and the total fluorescence produced upon excitationat 485 nmwasmonitored at 530 nm. This valuewas employed to deter-mine the amount of free siRNA. Triton X-100 was then added to disruptthe particles and assess the total siRNA in solution. The assay was per-formed using a SpectraMax M5e microplate spectrophotometer fromMolecular Devices (Sunnyvale, CA). siRNA amounts were determinedusing a siRNA standard curve. The siRNA encapsulation rate was deter-mined by the equation below

Encapsulation rate ¼ 1−free siRNA=total siRNAð Þ � 100% :

2.4. Lipid analysis by RP-UHPLC-CAD

Individual lipid concentrations were determined by Reverse PhaseUltra High-Performance Liquid Chromatography (RP-UHPLC) using aWaters ACQUITY UPLC system (Water Corporation, Milford MA) witha Corona ultra charged aerosol detector (ultraCAD) (ESA Biosciences,Inc., Chelmsford, MA). Separations were effected using an AgilentZorbax Rapid Resolution High Definition (RRHD) SB-C8 (2.1 × 50 mm,1.8 μm particle size) column with the CAD set at 80 °C. Mobile phaseA was composed of 0.1% trifluoroacetic acid (TFA) in H2O and Bcontained 0.1% TFA in MeOH. The gradient was as follows: 0 min

4858

3341

0

20

40

60

80

LNP-A LNP-B

% s

iRN

A r

elea

se f

rom

LN

P

W/ SerumW/O Serum

35

40

45

50

% s

iRN

A r

elea

se f

rom

LN

P

Mouse Rat Rhesus

48 ± 1.1

42 ± 1.7*40 ± 2.0*

A

B

Fig. 3.A: The effect of serum treatment on the siRNA release of LNP A and LNP B. Black bar:with rat serum; gray bar: without serum; B: The impact of serum species on % of siRNA re-lease from LNP A at pH 6.0. Themean ± one standard deviation of samples from triplicatemeasurements for each species is indicated in the figure. Individualmeasurement; , aver-age of triplicates. *, p b 0.01 comparing to siRNA release with mouse serum treatment.

9Y. Zhang et al. / Journal of Controlled Release 174 (2014) 7–14

(30% A), 0.25 min (30% A), 1.00 min (18% A), 2.50 min (14% A),3.25 min (0% A), 3.50 min (0% A), 3.75 min (30% A) and 4.00 min(30%A)with aflow rate of 1.4 ml/min. A standard curvewith a quadrat-ic curve fit was used to determine concentrations. Themolar percentageof each lipid was calculated based on its molecular weight.

2.5. Release of siRNA from LNP determination by fluorescence assay

LNP containing 0.6 mg/ml siRNA and serum was mixed together atequal volumes, resulting in a final serum concentration of 50%. Themixture was incubated at 37 °C for 20 min. Subsequently, anionic lipo-somes were added to the serum–LNP mixture at a desired anionic/cationic lipid mole ratio in 150 mM phosphate buffer at either pH 7.5or 6.0. The resulting combination was then incubated at 37 °C for15 min. As a control, an equal volume of 1× PBS, in place of the anionicliposome, was added to the serum–LNP mixture. The siRNA encapsula-tion rate of LNP with different treatments was obtained by the SYBRGold fluorescence assay. The % siRNA released was calculated based onthe equation below.

% siRNA release = (encapsulated rate of LNPswith anionic liposometreatment) / (encapsulated rate of LNPs with 1× PBS) × 100%.

2.6. Measurement of particle size and polydispersity of LNPs by dynamiclight scattering (DLS)

LNPs containing 3 μg siRNA were diluted to a final volume of 3 mlwith 1× PBS. Buffer 1× PBS was filtered through a Whatman Anotop0.02 μm filter to minimize background signals. The particle size andpolydispersity of the samples were measured by a dynamic light scat-tering method using a ZetaPALS instrument (Brookhaven InstrumentsCorporation, Holtsville, NY). The scattered intensity was measuredwith He–Ne laser at 25 °C with a scattering angle of 90°.

2.7. In vivo evaluation of LNP efficacy in rats

Rat studies were conducted at an AAALAC accredited MerckResearch Laboratories animal facility located at West Point, PA. Allstudy protocols were approved by the Merck West Point InstitutionalAnimal Care and Use Committee. LNPs were evaluated for in vivo effica-cy in Sprague–Dawley (Crl:CD (SD)) female rats (Charles River Labs,Wilmington, MA) by intravenous administration. Twenty-four hoursafter tail vein injection of 0.1 mg/kg LNP in a volume of 0.8 ml, ratswere sacrificed under anesthesia and livers were collected for ApoBmRNA level determination.

2.8. mRNA analysis by RT-PCR

Liver ApoB mRNA levels were determined by quantitative RT-PCRnormalizing to the geometric mean of Ppib and Gapdh. Message wasamplified from purified RNA utilizing rat ApoB, Ppib, and Gapdh

-20%

0%

20%

40%

60%

80%

0.0 1.0 2.0 3.0 4.0 5.0

DOPS-/CLinDMA+ ratio

% s

iRN

A r

elea

se f

rom

LN

P

pH 6.0

pH 7.5

Fig. 2. The effect of DOPS−/CLinDMA+ratio (0.25–4) on the siRNA release of LNPAat pH 6.0(circle) and pH 7.5 (square). Triplicate measurements were performed at a DOPS−/CLinDMA+ ratio of 1; single measurement was made at all other ratios.

commercial probe sets (Applied Biosystems Cat # Rn01499054_m1,Rn03302274_m1, and 4352338E respectively). The PCR reaction wasperformed on an ABI 7900 instrument (Azcobiotech, Inc., San Diego,CA). The ApoB mRNA level was calculated using the ddCt method, nor-malizing to the geometric mean of the housekeeping genes (dCt) thencomparing to the PBS control group (ddCt). All mRNA data is expressedas a percent relative to the PBS control dose.

2.9. PEG–DMG quantification by LC–MS/MS analysis

LNP containing0.6 mg/ml siRNA and rat serumweremixed togetherat a 1:1 (v/v) ratio. After incubation at 37 °C for 20 min, a 100 μl portionof the mixture was applied for size-exclusion separation with aSuperose-6 10/300 GL column (GE Healthcare, Pittsburgh, PA) usingan Agilent 1100 HPLC system (Santa Clara, CA). The flow rate was0.5 ml/min with 1× DPBS as the mobile phase. Eluents were collectedat 0.5 ml/min/fraction with a Foxy200 fraction collector (TeledyneIsco, Lincoln, NE) and subjected to LC–MS/MS analysis. LNP in 1×DPBS was used as a negative control.

PEG–DMGwas isolated from collected fractions byprotein precipita-tion with 4 volumes of 50% acetonitrile:50% methanol (v/v). The super-natant was further diluted with ten volumes of Milli-Q water andsamples were directly injected for quantitation using a Perkin Elmer

Fig. 4. The correlation between % siRNA release and rat in vivo efficacy of 16 LNPs.

Fig. 5. A: % of siRNA release in rat serum of LNP A with different particle sizes. Themean ± one standard deviation of samples from triplicate measurements is indicated inthe figure; B: Rat in vivo efficacy of LNP A with different particle sizes at 1 mpk. Themean ± one standard deviation of samples from four rats is indicated in the figure.♦, individual rat; , average of quadruplicates.

10 Y. Zhang et al. / Journal of Controlled Release 174 (2014) 7–14

HPLC system (Thermo Scientific,Waltham,MA) coupled to an API 4000triple quadruple mass spectrometer (Applied Biosystems, Foster City,CA). For each run, a total of 10 μl of sample was injected onto a HypersilBDS C8 column (50 × 3 mm; ID 3 μm) (Thermo Scientific). A dual elu-ent system was used: 95% H2O/5% methanol/10 mM ammonium for-mate/0.1% formic acid (A) and 40% methanol/60% n-propanol/10 mMammonium formate/0.1% formic acid (B). The flow rate was 0.4 ml/minand the gradient was as follows: 0 min (50% A), 1.0 min (0% A),2.5 min (0% A), 4.6 min (50% A), 5.0 min (50% A). PEG–DMG was nor-malized to the maximum fraction amount within each group.

2.10. Assessing the interaction between LNP and anionic liposomes by FRET

Fluorescence Resonance Energy Transfer (FRET) was used to studythe interaction between LNP and anionic liposomes. LNPs were mixedwith the anionic liposomes labeled by FRET pair of NBD-PE and Rh-PEin 150 mMphosphate buffer resulting infinal pHof 6.0. FRETwas deter-mined by kinetically monitoring the decrease in fluorescence intensityof N-Rh-PE (acceptor) at 592 nm with an excitation of 480 nm using aSpectraMax M5e microplate spectrophotometer. The concentration ofNBD-PE and Rh-PE was varied at 0.5%, 1% and 2%. A linear increase inFRET was observed as the FRET pair concentration increased from 0.5to 2% in anionic liposomes (data not shown). Therefore, in this range,any increase in the FRET pair distance caused by fusion or lipid exchangebetween LNPs and anionic liposomes could lead to a proportional de-crease in FRET. To examine LNP and anionic liposome interactions, wechose 1% fluorescence labeled anionic liposomes containing DOPS,DOPC, DOPE, NBD-PE and N-Rh-PE with a molar ratio of 25:25:48:1:1.Anionic liposome in 50% serum without LNP was used as a control.The samples were incubated at 37 °C and analyzed at 0, 5, 10, 15, 20,25 and 30 min. The samples were treated with 1% Triton X-100 for theillustration of disassembly of liposomes and maximum decrease inFRET. Energy transfer efficiency was calculated based on the formula:

E ¼ εACA

εDCD

IADIA

−1� �

where E: energy transfer efficiency; εA: extinction coefficient of accep-tor Rh-PE; CA: molar concentration of acceptor Rh-PE; εD: extinction co-efficient of donor NBD-PE; CD: molar concentration of donor NBD-PE;IAD: fluorescence intensity of acceptor Rh-PE in the presence of donorNBD-PE; IA: fluorescence intensity of acceptor Rh-PE in the absence ofdonor NBD-PE.

The distance between donor and acceptor was determined by r ¼ffiffiffiffiffiffiffiffiffiffiffi1E−16

q� R0.

Table 1Characterization data of LNP A with different particle sizes and various PEG mol%.

LNP DLS (nm) Encapsulationefficiency (%)

N/Pratio

Lipid, mol%

nm Pdl PEG Cholesterol Clin-DMA

LNP A-I(70 nm)

69.8 ± 0.3 0.05 85 2.9 1.6 37.1 61.3

LNP A-II(88 nm)

87.6 ± 0.2 0.07 83 2.9 1.5 36.4 62.1

LNP A-III(102 nm)

102.2 ± 1.1 0.07 89 2.6 1.7 38.3 60.1

LNP A-IV(1% PEG)

157.6 ± 2.2 0.10 95 2.9 0.8 39.5 59.7

LNP A-V(2% PEG)

110.9 ± 1.1 0.09 89 3.0 2.0 39.0 59.0

LNP A-VI(5.4% PEG)

67.1 ± 1.5 0.09 88 3.1 5.3 38.9 55.8

LNP A-VII(5.4% PEGspiked)

103.7 ± 1.7 0.06 87 2.9 4.9 37.4 57.7

R0: Förster distance, 56 A for NBD-PE and Rh-PE pair [35].

2.11. Determination of LNP phase by small-angle X-ray scattering (SAXS)

SAXS experiments were performed on a Bruker Nanostar SAXS in-strument as described previously [36]. LNPs in the absence or presenceof anionic liposomes were concentrated by centrifugal filtration(Amicon Ultra, RC 100 kDa membrane) by a factor of 8–10 by mass.Sampleswere scanned at room temperature for 60 min. Silver behenatewas used to calibrate the detector-to-sample distance and Datasqueezev. 2.1.5 was used to analyze the 2D X-ray data. The intensity of scatter-ingwas integrated azimuthally on the 2Ddiffraction pattern andplottedas a function of the scattering vector, q = 4πsin(θ/2)/λ, where θ is thescattering angle.

2.12. Statistical analysis

A one-way analysis of variance (ANOVA)was used to determine sta-tistical significance (p b 0.01) among the mean values of siRNA releasefrom LNPs in different species serum (Fig. 3B).

3. Results

3.1. The effect of anionic liposomes and pH on siRNA release from LNP

As indicated inMaterials andmethods Section 2.2 anionic liposomeswere used to mimic endosome membrane. To investigate the effect ofanionic liposomes and pH on % siRNA released from LNP, LNP A wasused as a model LNP. LNP A contains the lipids CLinDMA, Cholesterol,and PEG–DMG, along with ApoB siRNA with a lipid mol% of 60:38:2and a nitrogen/phosphate (N/P) ratio of 2.8. LNP A was incubated withserum for 20 min to mimic the duration of LNP in circulation. Varyingamounts of anionic liposomes (PS/PC/PE) were added to the mixtureresulting in molar ratios of the anionic lipid DOPS to the cationic lipidCLinDMA (DOPS−/CLinDMA+) of 0.25, 0.5, 1.0, 2.0 and 4.0 in 150 mMphosphate buffer at pH 6 and pH 7.5.

As shown in Fig. 2, the results indicated thatwith increasing DOPS−/CLinDMA+ molar ratios, an increase in siRNA release was observed atboth pH 6.0 and 7.5. As expected, this was exaggerated at pH 6.0 com-pared to pH 7.5. At pH 6.0, siRNA release increased steeply from 6% to

Fig. 6. A: % of siRNA release in rat serum of LNP A with different percentages of PEG. Themean ± one standard deviation of samples from triplicate measurements is indicated inthe figure; B: Rat in vivo efficacy of LNP A with different percentages of PEG at 1 mpk; C:The correlation between % siRNA release and in vivo efficacy of PEG variation LNPs. The% siRNA knockdown in rat @1 mpk is shown as the mean ± one standard deviation ofthe samples from four rats. ♦, individual measurement; , average of quadruplicates.

11Y. Zhang et al. / Journal of Controlled Release 174 (2014) 7–14

38%with increasing DOPS−/CLinDMA+ ratios from0.25 to 1.0, followedby a continued, gradual increase to 69% as the DOPS−/CLinDMA+ ratiowas further altered from 1.0 to 4.0. At pH 7.5, b5% siRNA release wasobserved at ratios below 1. Maximal siRNA release of 20% was achievedat the highest DOPS−/CLinDMA+ ratio used of 4. Based on this observa-tion, a DOPS−/CLinDMA+ ratio of 1.0 was selected for all the followingstudies.

3.2. The effect of serum treatment on siRNA release from LNP

LNP B containing DLinDMA, cholesterol, DSPC, PEG–DMG, and ApoBsiRNA at a nitrogen/phosphate (N/P) ratio of 6 was used alongwith LNPA to study serum effects. The DOPS−/cationic lipid+ ratio was kept at 1for both LNPs. In general, more siRNA was released from LNP B thanfrom LNP A. With the deprivation of serum treatment, about 30% lesssiRNA was released from both LNPs: 58% with serum treatment vs.41% without serum treatment for LNP B and 48% with serum treatmentvs. 33%without serum treatment for LNPA (Fig. 3A). To address the con-cern of the possibility that serum alone caused release enhancement,we also included a control arm for each LNP experiment. In the controlarm, instead of the anionic liposomes, only PBSwas added to LNP serumincubation mixture. A small percentage (ranging from 0 to 5%) of siRNAleakage was observed in the control arms. This low amount of siRNArelease might be due to the serum protein interaction and PEG–lipiddissociation, but the interaction was not strong enough to disrupt theintegrity of the LNP. For comparison of the efficiency of siRNA releasewith different LNPs, the data was normalized to the corresponding con-trols to account for serum treatment induced siRNA leakage (see detailsin the formula in Section 2.5). To further examine the species difference

in the impact of serum on siRNA release, various species of sera includ-ing mouse, rat and rhesus were applied in this assay. As indicated inFig. 3B, mouse serum appeared to cause the greatest extent of siRNArelease among the three species of serum (p b 0.01). Rat and rhesusserum provided similar % siRNA release.

3.3. Correlation between % siRNA release and in vivo efficacy in rat

A total of 16 LNPs were tested for % siRNA release in rat serum andin vivo efficacy in rats. All 16 LNPs contain cationic lipid, cholesterol,DSPC and PEG–DMG at a molar ratio of 58:30:10:2 and an N/P ratio of6. The 16 cationic lipids used in the study contain a variety of structures.They cover a range of different lipid tails (e.g. length, unsaturation, andasymmetry of lipid tails), various scaffolds that serve as a linkerbetween lipid tails and head groups (e.g. ether, acid labile, cyclic), andamine head groups with different pKa values. A positive correlationwith an R2 value of 0.47 was found between % siRNA release in ratserum and in vivo efficacy, as shown in Fig. 4. Among these 16 LNPs,there are one false positive (58, 45) and one false negative (17, 79).The false positive is easy to deal with considering the goal of the studyis to develop a tool to help screen large number of LNPs for in vivo stud-ies. However, the false negative suggests the potential of missing in vivohits using the assay.

3.4. The impact of particle size on siRNA release and correlation to in vivoefficacy

To assess the effect of particle size on siRNA release, three LNP Abatches, LNP A-I, LNP A-II, and LNP A-III were manufactured with thesame lipid and siRNA compositions but varying siRNA buffer solutions.All three LNPs had similar lipid mol%, N/P ratio, and encapsulation effi-ciency (Table 1).

DLS measurement indicated varying diameters of 70, 88, and102 nm for LNP A-I, LNP A-II and LNP A-III, respectively. siRNA releaseassay results in rat serum and rat efficacy data indicated a differenceamong the three LNP A (I to III). In general, with the increase in particlesize to ~100 nm, an enhancement in siRNA release and rat in vivo effica-cy was observed. Specifically, LNP A-I and LNP A-II had similar % siRNArelease of 45% and 43%, respectively in serum at pH 6.0; however, thisparameter was increased to 52% for LNP A-III (Fig. 5A). Likewise,in vivo data revealed 26% and 32% mRNA knockdown at 1 mpk forLNP A-I and LNP A-II, respectively, and an improved efficacy of 53%knockdown for LNP A-III (Fig. 5B).

3.5. Effect of PEG mol% on siRNA release and correlation to in vivo efficacy

LNPs with 1, 2 and 5.4 mol% of PEG (LNP A-IV, V and VI) were pre-pared as described in the Materials and methods section. With the in-crease in PEG content from 1 to 5.4%, a decrease in % siRNA releasewas observed in rat serum (Fig. 6A). However, this increase in PEGmol% produced a decrease in size from 158 to 67 nm (Table 1). Inorder to assure the decrease in average size was not caused by smallPEG–lipid micelles potentially formed in the preparation, especially athigh PEG–lipid mol% of 5.4, all LNPs were purified by Tangential FlowFiltration (TFF) using a 100 kDa cutoff membrane (Spectrum Laborato-ries, Inc., CA). A further CryoEM image analysis did not find anymicellesin these LNPs (data not shown). To exclude the effect of particle size onsiRNA release, LNP A-VII was prepared by spiking extra quantities ofPEG into the base LNP A-V (2% PEG) to obtain a final percentage of 5.4.DLS showed similar particle sizes (111 nm vs. 104 nm) between LNPA-V (2% PEG) and LNP A-VII (5.4% PEG spiked) (Table 1). Slightlylower siRNA release (41%) and mRNA knockdown (55%) was observedin LNP A-VII (5.4% PEG spiked), relative to LNP A-V (2% PEG), whichexhibited a siRNA release of 52% and an mRNA knockdown of 66% inrat at 1mpk (Fig. 6A and B). In general, a good correlation was observed

Fig.

7.Aprop

osed

mod

elforthe

releaseof

siRN

Afrom

LNP.Step

1:incu

bation

ofLN

Pwithserum

at37

°Cun

derp

hysiolog

icalpH

;PEG

–lip

iddiffu

sesfrom

LNP;

Step

2:LN

P-serum

mixture

treatedwithan

ioniclip

osom

esat

pH6.0;

cation

iclip

idsare

proton

ated

andelectrostatically

interact

andex

chan

gelip

idswithan

ioniclip

osom

es;S

tep3

:LNPmem

bran

edisrup

tion

andreleaseof

siRN

Afrom

LNP.

12 Y. Zhang et al. / Journal of Controlled Release 174 (2014) 7–14

between in vitro siRNA release and in vivo rat siRNA knockdown in PEGvaried LNPs with R2 = 0.97 (Fig. 6C).

4. Discussion

Themajor challenge to realizing the therapeutic potential of siRNA issafe and effective delivery to target organs and cells. LNP is one of sever-al effective means for siRNA delivery that has been advanced into theclinic by Alnylam and Tekmira (www.clinicaltrials.gov). As such, anefficient and predictive in vitro screening tool for LNP lead selectionand SAR optimization would be highly beneficial. In the present study,we have incorporated the major environmental changes that LNPsencounter after systemic administration into one assay. Serum interac-tions, acidic endosomal pH, and anionic endosomal lipids are the threemajor factors we believe to be critical to cytosolic delivery of siRNAcargo and were, therefore, applied to LNPs in this assay. The amountof siRNA freed from the LNPs after these treatments serves as ameasurefor the effectiveness of delivery, while in vivo data validates its predic-tive capability. Results have allowed the generation of a mechanistictheory, and several unique dependencies on particle characteristicsand environmental conditions.

Basically, LNPs were incubated with 50% serum from different spe-cies (i.e. mouse, rat, or rhesus) at 37 °C. The resulting samples werethen incubated with anionic liposomes, approximating the endosomalmembrane, at pH 6 and 7.5. The amount of siRNA released from LNPscan be determined using the fluorescence SYBR Gold method. Ourdata indicate all of these treatment conditions can affect the effective-ness of siRNA release from LNPs (Figs. 2 & 3A). In the absence ofserum, about 30% less siRNA was released from LNPs compared to theamount released in 50% serum (Fig. 3A). We also found that as theanionic lipid DOPS− to cationic lipid CLinDMA+ ratio increases, siRNArelease increases significantly only at pH 6.0. Based on these observa-tions, we propose a model for the release of siRNA from LNPs (Fig. 7).First, PEG–lipids gradually dissociate from LNPs through interactionswith serum proteins. This reduces the steric hindrance of LNP interac-tions with anionic membranes. Second, amines of the cationic lipids,with pKa's spanning 7.0 to 8.0, will be protonated under acidicendosomal pH (5.0 to 6.0). Opposing charges allow the protonated cat-ionic lipids to interact more effectively with the anionic endosomallipids, triggering the lamellar to hexagonal or disordered phase transi-tion in the LNP lipid bilayer. This altered geometry facilitates the disrup-tion of LNP lipid bilayer and the subsequent release of siRNA [12].

To test the hypothesis of dissociation of PEG–lipid from LNP inserum, SEC-LC–MS/MS was used to quantify the amount of PEG–lipidafter incubating LNP A with either 50% (v/v) rat serum or PBS for20 min. For the control with PBS, PEG–lipid eluted at the LNP region.For the serum treated sample, most of the PEG–lipid was dissociatedfrom the LNP and eluted around the high-density lipoprotein (HDL)region, and a small amount eluted around the low-density lipoprotein(LDL) region (Fig. 8). A similar dissociation of PEG–lipid from LNP wasalso observed with LNP B (data not shown). This result is consistentwith previous literature reports that the interaction between LNPs anda neutral lipid sink leads to the dissociation of PEG from the LNP[16,37]. To further confirm this theory, delipidated serum will be usedin future studies.

To test the second hypothesis in ourmodel, small angle X-ray scatter-ing (SAXS) was utilized to study structure changes of LNPs upon incuba-tionwith anionic lipids. FRETwas used to detect fusion or lipid exchangebetween LNPs and anionic liposomes. The anionic liposomes were la-beled with FRET pairs of NBD-PE and Rh-PE. SAXS was run on LNPswith and without anionic liposome treatment as indicated in theMaterials andmethods section 2.11. A change of structure fromweak la-mellar phase to hexagonal phasewas found for LNP B and a change of thelamellar structure to less ordered phase was found for LNP A (Fig. 9A &B). Due to the strong background of serum, SAXS was conducted onLNPs and anionic liposomes without serum treatment. We anticipated

0

20

40

60

80

100

0 5 10 15 20 25 30 35 40Time (min)

% M

ax f

or

each

sam

ple

PBS

Serum

LNP Region

LDLRegion

HDLRegion

Fig. 8. PEG–lipid elution profile post 20 min incubation of LNPAwith PBS or 50% rat serumat 37 °C. Solid line: LNPA inPBS; dotted line: LNPAwith serum treatment. LNP region, LDLregion and HDL region are defined by running each of them individually under the sameSEC separation conditions.

0

20

40

60

80

100

0 10 20 30

Time (mins)

Mea

n d

ista

nce

bet

wee

n

NB

D-P

E a

nd

Rh

-PE

(A

)

LNP A

LNP BSerum control

Fig. 10. The mean distance between NBD-PE and Rh-PE after incubation of anionic lipo-somes in serum at 37°C for 30 min in the presence or absence of LNPs at pH 6.0. ■, LNPA; ♦, LNP B; ▲, serum control (in the absence of LNP).

13Y. Zhang et al. / Journal of Controlled Release 174 (2014) 7–14

that the lipid phase changes would be enhanced with serum treatmentbased on earlier observations of PEG–lipid dissociation and more siRNArelease with serum treatment (Figs. 3A & 8). FRET data showed a kineticincrease in distance between NBD-PE and Rh-PE in the presence of LNP,suggesting a fusion or lipid exchange between LNPs and anionic lipo-somes (Fig. 10). The increase in distance happened quickly within5 min when LNPs and anionic liposomes were mixed together. After20 min, there seemed to be less change. Interestingly, the control sam-ple, anionic liposomes alone in 50% serum, also showed a slight increasein distance of the FRET pair. This is most likely due to lipid exchangebetween anionic liposomes and serum lipids.

We found that the amount of siRNA released from LNPs was alsodependent on the serum's species of origin. Mouse serum showed themost siRNA release compared to rat and rhesus serum (Fig. 3B). The dif-ferencemight be related to the varying serum lipoprotein profile in each

0.1

1

10

100

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

Inte

nsi

ty (

a.u

.)

LNP A

LNP A + anionic liposme

1

32

A

0.1

1

10

100

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

q (A-1)

q (A-1)

Inte

nsi

ty (

a.u

.)

LNP B

LNP B + anionic liposome

1

31/2

2

B

Fig. 9. SAXS profiles of LNPs and LNPs/anionic liposomemixtures at pH 6.0. Black profiles:LNPs; Gray profiles: LNPs/anionic liposomemixtures. A: the presence of diffraction peaksindexed approximately 1:2:3 indicates lamellar structure of LNP A; a disordered lamellarphasewas shown formixtures of LNP A and anionic liposomes. B: LNP B has aweak lamel-lar structure as evidenced by two broad diffraction peaks (see arrows); the diffractionpeaks indexed as 1:31/2:2 are characteristic of an organized hexagonal phase for LNP Band anionic liposome mixtures.

species. ApoE binding to LNPs has been reported in the literature toimpact liver uptake and efficacy [26,38]. However, the difference insiRNA release (Fig. 3B) could not be simply explained by the concentra-tion of ApoE in various species of serum [39–42]. The PEG–lipid dissoci-ation results in Fig. 8 suggest that lipoproteins such as HDL and LDLmight impact siRNA release. As a component of lipoproteins, we believeApoE also contributes to siRNA release by facilitating PEG–lipid dissoci-ation from LNP. Future studies with individual serum proteins or lipo-proteins in this siRNA release assay will help us understand thespecies difference.

An increase in siRNA release and rat in vivo efficacy was found withan increase in LNP particle size from 70 nm to 102 nm (Fig. 5). The sig-nificantly lower siRNA release and rat efficacy of LNP A-VI (5.4% PEG,67 nm) compared to LNP A-VII (5.4% PEG spiked, 104 nm) (Fig. 6)also suggested the positive impact of larger particle size on siRNA re-lease and in vivo efficacy. However, much larger particles are not fa-vored in terms of tissue penetration. For example, for liver delivery,LNP particle size less than 100 nm is preferred due to the size limit ofhepatic capillary fenestration [43]. Therefore, we think there is an opti-mal range for LNP particle size considering both LNP efficacy and tissuepenetration. There seemed to be contradictory reports regarding the ef-fect of particle size on LNP in vivo efficacy. Some researchers suggestedthat smaller size (b100 nm) of LNP had better efficacy since they couldpass through the fenestrations of the liver endothelium easier than larg-er size particles [19,42]. Other works indicated rapid renal clearance ofsiRNA nanoparticles between 10 nm to 100 nm through glomerularbasement membrane (GBM) deposition [44,45]. In a most recentstudy, there appeared to be no apparent correlation between LNP parti-cle size (from100 to 200 nm) and in vivo efficacy [32].We think the dis-crepancy among these researches most likely originated from theirdifferent LNP compositions and biophysical properties.

Among 16 LNPs tested for in vitro siRNA release and in vivo rat effica-cy, a good correlation with R2 = 0.47 was found. Considering the LNPtest set used in this correlation study contains a variety of cationiclipid structures, the in vitro siRNA release assay may be generalized asa screening tool to identify lead LNP for in vivo evaluation.

In conclusion, a simple in vitro screening assaymimicking the in vivoenvironment of LNPs after intravenous administration was developedby taking into account serum exposure and interaction of anionic lipo-somes at endosomal pH.

Acknowledgments

We thank the Central Pharmacology and RNAi Biology group for an-imal studies.

14 Y. Zhang et al. / Journal of Controlled Release 174 (2014) 7–14

References

[1] A. Fire, S.Q. Xu, M.K. Montgomery, S.A. Kostas, S.E. Driver, C.C. Mello, Potent and spe-cific genetic interference by double-stranded RNA in Caenorhabditis elegans, Nature391 (1998) 806–811.

[2] S.M. Elbashir, J. Harborth, W. Lendeckel, A. Yalcin, K. Weber, T. Tuschl, Duplexes of21-nucleotide RNAs mediate RNA interference in cultured mammalian cells, Nature411 (2001) 494–498.

[3] S.H. Hsu, B. Yu, X. Wang, Y. Lu, C.R. Schmidt, R.J. Lee, L.J. Lee, S.T. Jacob, K. Ghoshal,Cationic lipid nanoparticles for therapeutic delivery of siRNA and miRNA to murineliver tumor, Nanomedicine 9 (2013) 1169–1180.

[4] T.S. Zimmermann, A.C.H. Lee, A. Akinc, B. Bramlage, D. Bumcrot, M.N. Fedoruk, J.Harborth, J.A. Heyes, L.B. Jeffs, M. John, A.D. Judge, K. Lam, K. McClintock, L.V.Nechev, L.R. Palmer, T. Racie, I. Rohl, S. Seiffert, S. Shanmugam, V. Sood, J.Soutschek, I. Toudjarska, A.J. Wheat, E. Yaworski, W. Zedalis, V. Koteliansky, M.Manoharan, H.P. Vornlocher, I. MacLachlan, RNAi-mediated gene silencing innon-human primates, Nature 441 (2006) 111–114.

[5] D.B. Rozema, D.L. Lewis, D.H. Wakefield, S.C. Wong, J.J. Klein, P.L. Roesch, S.L. Bertin,T.W. Reppen, Q. Chu, A.V. Blokhin, J.E. Hagstrom, J.A. Wolff, Dynamic polyconjugatesfor targeted in vivo delivery of siRNA to hepatocytes, Proc. Natl. Acad. Sci. U. S. A. 104(2007) 12982–12987.

[6] M.T. Abrams, M.L. Koser, J. Seitzer, S.C. Williams, M.A. DiPietro, W.M. Wang, A.W.Shaw, X.Z. Mao, V. Jadhav, J.P. Davide, P.A. Burke, A.B. Sachs, S.M. Stirdivant, L.Sepp-Lorenzino, Evaluation of efficacy, biodistribution, and inflammation for a po-tent siRNA nanoparticle: effect of dexamethasone co-treatment, Mol. Ther. 18(2010) 171–180.

[7] M.G. Stanton, S.L. Colletti, Medicinal chemistry of siRNA delivery, J. Med. Chem. 53(2010) 7887–7901.

[8] S.C. Semple, A. Akinc, J.X. Chen, A.P. Sandhu, B.L. Mui, C.K. Cho, D.W.Y. Sah, D. Stebbing,E.J. Crosley, E. Yaworski, I.M. Hafez, J.R. Dorkin, J. Qin, K. Lam, K.G. Rajeev, K.F.Wong, L.B.Jeffs, L. Nechev, M.L. Eisenhardt, M. Jayaraman, M. Kazem, M.A. Maier, M. Srinivasulu,M.J. Weinstein, Q.M. Chen, R. Alvarez, S.A. Barros, S. De, S.K. Klimuk, T. Borland, V.Kosovrasti, W.L. Cantley, Y.K. Tam, M. Manoharan, M.A. Ciufolini, M.A. Tracy, A. deFougerolles, I.MacLachlan, P.R. Cullis, T.D.Madden,M.J. Hope, Rational design of cation-ic lipids for siRNA delivery, Nat. Biotechnol. 28 (2010) 172-U118.

[9] P.P. Karmali, D. Simberg, Interactions of nanoparticles with plasma proteins: impli-cation on clearance and toxicity of drug delivery systems, Expert Opin. Drug Deliv. 8(2011) 343–357.

[10] J.S. Zhang, F. Liu, L. Huang, Implications of pharmacokinetic behavior of lipoplex forits inflammatory toxicity, Adv. Drug Deliv. Rev. 57 (2005) 689–698.

[11] N.S. Templeton, D.D. Lasic, P.M. Frederik, H.H. Strey, D.D. Roberts, G.N. Pavlakis, Im-proved DNA: liposome complexes for increased systemic delivery and gene expres-sion, Nat. Biotechnol. 15 (1997) 647–652.

[12] I. Koltover, T. Salditt, J.O. Radler, C.R. Safinya, An inverted hexagonal phase of cation-ic liposome–DNA complexes related to DNA release and delivery, Science 281(1998) 78–81.

[13] S. Kitagawa, M. Kasamaki, M. Hyodo, Cationic vesicles consisting of1,2-dioleoyl-3-trimethylammonium propane (DOTAP) and phosphatidylcho-lines and their interaction with erythrocyte membrane, Chem. Pharm. Bull. 52(2004) 451–453.

[14] K. Remaut, B. Lucas, K. Braeckmans, J. Demeester, S.C. De Smedt, Pegylation of lipo-somes favours the endosomal degradation of the delivered phosphodiester oligonu-cleotides, J. Control. Release 117 (2007) 256–266.

[15] P. Harvie, F.M.P. Wong, M.B. Bally, Use of poly(ethylene glycol)–lipid conjugates toregulate the surface attributes and transfection activity of lipid–DNA particles, J.Pharm. Sci. 89 (2000) 652–663.

[16] E. Ambegia, S. Ansell, P. Cullis, J. Heyes, L. Palmer, I. MacLachlan, Stabilized plasmid–lipid particles containing PEG–diacylglycerols exhibit extended circulation lifetimesand tumor selective gene expression, Biochim. Biophys. Acta Bioenerg. 1669 (2005)155–163.

[17] F.X. Shi, L. Wasungu, A. Nomden, M.C.A. Stuart, E. Polushkin, J.B.F.N.Engberts, D. Hoekstra, Interference of poly(ethylene glycol)–lipid ana-logues with cationic-lipid-mediated delivery of oligonucleotides; role oflipid exchangeability and non-lamellar transitions, Biochem. J. 366(2002) 333–341.

[18] M.S. Webb, D. Saxon, F.M.P. Wong, H.J. Lim, Z. Wang, M.B. Bally, L.S.L. Choi, P.R.Cullis, L.D. Mayer, Comparison of different hydrophobic anchors conjugated topoly(ethylene glycol): effects on the pharmacokinetics of liposomal vincristine,Biochim. Biophys. Acta Bioenerg. 1372 (1998) 272–282.

[19] A. Akinc, M. Goldberg, J. Qin, J.R. Dorkin, C. Gamba-Vitalo, M. Maier, K.N.Jayaprakash, M. Jayaraman, K.G. Rajeev, M. Manoharan, V. Koteliansky, I.Rohl, E.S. Leshchiner, R. Langer, D.G. Anderson, Development of lipidoid–siRNA formulations for systemic delivery to the liver, Mol. Ther. 17 (2009)872–879.

[20] B.M. Tandia, M. Vandenbranden, R. Wattiez, Z. Lakhdar, J.M. Ruysschaert, A.Elouahabi, Identification of human plasma proteins that bind to cationic lipid/DNAcomplex and analysis of their effects on transfection efficiency: implications for in-travenous gene transfer, Mol. Ther. 8 (2003) 264–273.

[21] A. Chonn, S.C. Semple, P.R. Cullis, Association of bloodproteinswith large unilamellarliposomes in vivo — relation to circulation lifetimes, J. Biol. Chem. 267 (1992)18759–18765.

[22] J.P. Yang, L. Huang, Overcoming the inhibitory effect of serum on lipofection by in-creasing the charge ratio of cationic liposome to DNA, Gene Ther. 4 (1997) 950–960.

[23] O. Zelphati, L.S. Uyechi, L.G. Barron, F.C. Szoka, Effect of serum components on thephysico-chemical properties of cationic lipid/oligonucleotide complexes and ontheir interactions with cells, Biochim. Biophys. Acta Lipids Lipid Metab. 1390(1998) 119–133.

[24] S. Li, W.C. Tseng, D.B. Stolz, S.P. Wu, S.C. Watkins, L. Huang, Dynamic changes in thecharacteristics of cationic lipidic vectors after exposure to mouse serum: implica-tions for intravenous lipofection, Gene Ther. 6 (1999) 585–594.

[25] F. Sakurai, T. Nishioka, H. Saito, T. Baba, A. Okuda, O. Matsumoto, T. Taga, F.Yamashita, Y. Takakura, M. Hashida, Interaction between DNA–cationic liposomecomplexes and erythrocytes is an important factor in systemic gene transfer viathe intravenous route in mice: the role of the neutral helper lipid, Gene Ther. 8(2001) 677–686.

[26] A. Akinc, W. Querbes, S.M. De, J. Qin, M. Frank-Kamenetsky, K.N. Jayaprakash, M.Jayaraman, K.G. Rajeev, W.L. Cantley, J.R. Dorkin, J.S. Butler, L.L. Qin, T. Racie, A.Sprague, E. Fava, A. Zeigerer, M.J. Hope, M. Zerial, D.W.Y. Sah, K. Fitzgerald, M.A.Tracy, M. Manoharan, V. Koteliansky, A. de Fougerolles, M.A. Maier, Targeted deliv-ery of RNAi therapeutics with endogenous and exogenous ligand-based mecha-nisms, Mol. Ther. 18 (2010) 1357–1364.

[27] K. Buyens, M. Meyer, E. Wagner, J. Demeester, S.C. De Smedt, N.N. Sanders, Monitor-ing the disassembly of siRNA polyplexes in serum is crucial for predicting their bio-logical efficacy, J. Control. Release 141 (2010) 38–41.

[28] L.A. Carvalho, A.M. Carmona-Ribeiro, Interactions between cationic vesicles andserum proteins, Langmuir 14 (1998) 6077–6081.

[29] I.M. Hafez, N. Maurer, P.R. Cullis, On the mechanism whereby cationic lipids pro-mote intracellular delivery of polynucleic acids, Gene Ther. 8 (2001) 1188–1196.

[30] J.T. Zhang, H.H. Fan, D.A. Levorse, L.S. Crocker, Ionization behavior of amino lipids forsiRNA delivery: determination of ionization constants, SAR, and the impact of lipidpK(a) on cationic lipid–biomembrane interactions, Langmuir 27 (2011) 1907–1914.

[31] I.S. Zuhorn, V. Oberle, W.H. Visser, J. Engberts, U. Bakowsky, E. Polushkin, D.Hoekstra, Phase behavior of cationic amphiphiles and their mixtures with helperlipid influences lipoplex shape, DNA translocation, and transfection efficiency,Biophys. J. 83 (2002) 2096–2108.

[32] C.A. Alabi, K.T. Love, G. Sahay, H. Yin, K.M. Luly, R. Langer, D.G. Anderson,Multiparametric approach for the evaluation of lipid nanoparticles for siRNA deliv-ery, Proc. Natl. Acad. Sci. U. S. A. 110 (2013) 12881–12886.

[33] K.T. Love, K.P. Mahon, C.G. Levins, K.A. Whitehead, W. Querbes, J.R. Dorkin, J. Qin, W.Cantley, L.L. Qin, T. Racie, M. Frank-Kamenetsky, K.N. Yip, R. Alvarez, D.W.Y. Sah, A.de Fougerolles, K. Fitzgerald, V. Koteliansky, A. Akinc, R. Langer, D.G. Anderson,Lipid-like materials for low-dose, in vivo gene silencing, Proc. Natl. Acad. Sci. U. S. A.107 (2010) 1864–1869.

[34] Y. Zhang, T.J. Anchordoquy, The role of lipid charge density in the serum stability of cat-ionic lipid/DNA complexes, Biochim. Biophys. Acta Bioenerg. 1663 (2004) 143–157.

[35] D.E. Wolf, A.P. Winiski, A.E. Ting, K.M. Bocian, R.E. Pagano, Determination of thetransbilayer distribution of fluorescent lipid analogs by nonradiative fluorescenceresonance energy-transfer, Biochemistry 31 (1992) 2865–2873.

[36] J.T. Zhang, H.H. Fan, D.A. Levorse, L.S. Crocker, Interaction of cholesterol-conjugated ionizable amino lipids with biomembranes: lipid polymorphism,structure–activity relationship, and implications for siRNA delivery, Langmuir27 (2011) 9473–9483.

[37] J.W. Holland, C. Hui, P.R. Cullis, T.D. Madden, Poly(ethylene glycol)–lipid conjugatesregulate the calcium-induced fusion of liposomes composed of phosphatidyletha-nolamine and phosphatidylserine, Biochemistry 35 (1996) 2618–2624.

[38] J. Seki, S. Sonoke, A. Saheki, H. Fukui, T. Mayumi, Decreased binding affinity of apo-lipoprotein E plays an important role in the reduced hepatic clearance of lipidnano-sphere (LNS), J. Drug Deliv. Sci. Technol. 14 (2004) 173–179.

[39] D.M. Ney, V.A. Ziboh, B.O. Schneeman, Reduction in plasma apolipoprotein-E andHdl1levels in rats with essential fatty-acid deficiency, J. Nutr. 117 (1987) 2016–2020.

[40] G.W. Melchior, S.C. Willard, M.T. Brandon, Apo E levels and distribution in rhesusmonkeys consuming an atherogenic diet, Atherosclerosis 58 (1985) 65–80.

[41] T.G. Cole, D.C. Stockhausen, Evidence of pools of ApoE in liver not directly associatedwith lipoprotein synthetic pathway, Circulation 74 (1986) 108-108.

[42] H. Wientgen, F.E. Thorngate, S. Omerhodzic, L. Rolnitzky, J.T. Fallon, D.L. Williams,E.A. Fisher, Subphysiologic Apolipoprotein E (ApoE) plasma levels inhibit neointi-mal formation after arterial injury in ApoE-deficient mice, Arterioscler. Thromb.Vasc. Biol. 24 (2004) 1460–1465.

[43] E. Wisse, F. Jacobs, B. Topal, P. Frederik, B. De Geest, The size of endothelial fenestraein human liver sinusoids: implications for hepatocyte-directed gene transfer, GeneTher. 15 (2008) 1193–1199.

[44] C.H. Choi, J.E. Zuckerman, P. Webster, M.E. Davis, Targeting kidney mesangium bynanoparticles of defined size, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) 6656–6661.

[45] J.E. Zuckerman, C.H. Choi, H. Han, M.E. Davis, Polycation-siRNA nanoparticles candisassemble at the kidney glomerular basement membrane, Proc. Natl. Acad. Sci.U. S. A. 109 (2012) 3137–3142.